Operational Oceanography in the 21st Century
Andreas Schiller€•Â€Gary B. Brassington Editors
Operational Oceanography in the 21st Century
1 3
Editors Dr. Andreas Schiller Centre for Australian Weather and Climate Research CSIRO GPO Box 1538 Hobart 7001, Tasmania Australia
[email protected] Dr. Gary B. Brassington Centre for Australian Weather and Climate Research Bureau of Meteorology PO Box 1289 Melbourne 3001, Victoria Australia
[email protected] ISBN 978-94-007-0331-5â•…â•…â•…â•… e-ISBN 978-94-007-0332-2 DOI 10.1007/978-94-007-0332-2 Springer Dordrecht Heidelberg London New York Library of Congress Control Number: 2011925930 All Rights Reserved for Chapters 21 and 22 © Springer Science+Business Media B.V. 2011 No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work. Cover illustration: Daily mean sea surface height anomaly for the 28th July 2009 as estimated by the operational BLUElink ocean prediction system. Red colors indicate positive height anomalies whilst blue colors represent negative height anomalies. Anomalies represent the estimated sea surface height relative to the model’s dynamic topography of a mean from a multi-year integration forced by reanalysis winds. Shown in the image is the east Indian Ocean with anticyclonic and cyclonic eddies in the mid-latitudes and the South Equatorial Current in the tropics that derives volume flux from the western warm pool in the Pacific Ocean via the Indonesian Throughflow. Image produced by Dr. Justin Freeman. Cover design: deblik, Berlin Printed on acid-free paper Springer is part of Springer Science+Business Media (www.springer.com)
Preface
In the mid 1990s the research community and operational agencies saw an emerging opportunity for near-realtime ocean forecasts similar to those produced in Numerical Weather Prediction: combining numerical models and observations via data assimilation in order to provide ocean prediction products on various space and time scales. This development was facilitated through an international framework provided by the Global Ocean Data Assimilation Experiment (GODAE). GODAE aimed at advancing ocean data assimilation by synthesizing satellite and in-situ observations with state-of-the art models of global ocean circulation. In the past few years ocean forecasting has matured to a stage where many nations have implemented global and basin-scale ocean analyses and short-term forecast systems that provide routine products to the oceanographic community serving a variety of applications in areas such as marine environmental monitoring and management, ocean climate, defense and industry applications. The authors within this book provide an up to date description of the major components of ocean analyses and forecasting systems. The chapters cover a wide range of topics including, but not limited to, scientific advances and challenges in ocean forecasting, the associated descriptions of the forecasting systems and end user applications. This integrated view of ocean forecasting is the end result of an International Summer School for Observing, Assimilating and Forecasting the Ocean held in Perth, Australia, in January 2010. The flow diagram (Fig. 1) captures the main functional components and sources of inputs implemented under GODAE and required by any ocean forecasting system. These are: the data and product servers, the assimilation centres and the users of the outputs. It captures many of the interactions required to ensure or enhance the quality of the systems and their outputs (Bell et al. 2009). The measurement network and data assembly and processing centres provide the main inputs to the assimilation centres (top centre and right of Fig. 1). In this book Le Traon (2011), Josey (2011), Ravichandran (2011) and Oke and O’Kane (2011) provide concise overviews of the in situ and satellite components of the current global observing system and discuss the continuing work required to sustain and optimise it. The GODAE-sponsored Global High Resolution Sea Surface Temperature (GHRSST) project has resulted in a coordinated network of centres disseminating v
vi
Preface
Fig. 1↜渀 Functional components of operational ocean forecasting systems developed during GODAE
SST data in real-time in a common format to agreed standards from a wide range of microwave and infra-red instruments on polar orbiting and geostationary satellites. Cummings (2011) summarizes the substantial achievements during GODAE in the quality control of observational data and the joint use of in situ and satellite data. Dombrowsky (2011) provides an overview of progress in the capabilities of ocean prediction systems and data and product servers (see left of middle row of Fig. 1). Brassington (2011) examines key properties of the real-time system and their impact on operational system design. They provide an overview of the underpinning concepts and technologies which enable the observed data to be discovered, visualised, downloaded, intercompared and analysed all over the world. Progress in ocean data assimilation (the central item in Fig. 1) is described in a number of papers (Zaron 2011; Moore 2011; Brasseur 2011). The tables and descriptions in Dombrowsky (2011) and Zhu (2011) provide a useful overview of the present modelling and assimilation components of the major systems involved in coastal and basin-scale ocean forecasting. Most centres now operate systems with 1/10° or finer horizontal grid spacing; have a global capability; make use of community ocean models (e.g. HYCOM, MOM4 or NEMO; see Barnier et al. 2011, Chassignet 2011; Hurlburt et al. 2011 and Matear 2011); and assimilate in situ profile data, altimeter data and some form of surface temperature data. Martin (2011) illustrates the skill of the high-resolution systems in forecasting sea surface currents and sea surface temperature.
Preface
vii
Product assessments and interactions with research users (lower right area of Fig. 1) have been key activities since the inception of operational ocean forecasting systems. Hernandez (2011) describes the procedures developed to intercompare forecasts produced by different centres and illustrates the insights these can give into the performance of the systems. Alves et al., (2011) describe some examples of how systems developed for ocean state estimation have been used for climate variability/seasonal forecasting research and how intercomparisons of results from the systems are being used to assess the consistency and uncertainty of the state estimates. Oke and O’Kane (2011) summarise results gathered by observing system design research and outline the exciting prospects for future work. They illustrate the complementarity of the SST, altimeter and profile data for mesoscale prediction, and present statistics on the dependence of the accuracy of 7-day forecasts, real-time analyses and delayed mode analyses on the availability of altimeter data. Wilkin et al. (2011) summarise the wide-ranging coastal applications in ocean forecasting. Finally Matear and Jones (2011) outline a number of categories of potential ecological and biogeochemical applications and discuss the challenges this area poses to the fidelity of the physical models and assimilation schemes and to the measurement technologies. The lower left part of Fig. 1 depicts the information flows to application centres (also known as downstream services) and users. Barras (2011) and King et al. (2011) describe the legal framework and use of ocean forecasting outputs in monitoring and prediction of marine pollution (such as oil spills) and the value of GODAE forecasts for safety and effectiveness of operations at Sea. Woodham (2011) provides examples of the wide variety of information and tactical decision aids generated using GODAE products to assist Naval operations. Ivey (2011) summarises the current operational use of upper ocean heat content information to forecast the intensity of tropical cyclones and current research in this area. Fundamentals and applications of sea-level variability, surface waves and tsunamis are discussed by Pattiaratchi (2011) and Greenslade and Tolman (2011). Huckerby (2011) and Mann (2011) provide an introduction to and overview about the emerging field of ocean renewable energy and the corresponding need for ocean state information to determine the available energy resources as well as the impact of ocean renewable energy on the physical environment. The editors gratefully acknowledge the students and lecturers listed below who actively contributed to the success of the summer school as well as the first round of reviews of the draft manuscripts. Primary support for this summer school was provided by the National Oceanographic and Atmospheric Administration (NOAA), USA, the Bureau of Meteorology, Australia, and the Commonwealth Scientific and Industrial Research Organisation (CSIRO), Australia. This support is gratefully acknowledged. The editors would like to thank the speakers for their contributions during the summer school and for providing their manuscripts within a tight time frame. We also express our appreciation to the GODAE OceanView Science Team who contributed in numerous ways to the success of the summer school and this book. We
viii
Preface
thank Charitha Pattiaratchi, Diana Greenslade, Tim Pugh, Roger Proctor, Bernard Barnier, Clothilde Langlais, Fabrice Hernandez, Marie Drevillon and Andy Taylor for preparing and conducting an excellent set of student exercises. We thank all the attendees (see list in Appendix) for participating actively in the lectures and in the lecture review process. Finally, our thanks go to Val Jemmeson, Nick D’Adamo and Charitha Pattiaratchi who spent considerable time with the logistics of the summer school. A special thank goes to Denise McMullen for her help in coordinating the editorial process of the manuscripts. Australia 28 May 2010
Andreas Schiller Gary B. Brassington
List of Lecturers and Students
Lecturers Dr. Oscar Alves╇ CAWCR, Australia, e-mail:
[email protected] Dr. Bernard Barnier╇ HMG, France,
[email protected] Ms. Kathryn Barras╇ Minter Ellison, Australia e-mail:
[email protected] Dr. Pierre Brasseur╇ HMG, France, e-mail:
[email protected] Dr. Gary B. Brassington╇ CAWCR, Australia, e-mail:
[email protected] Prof. Eric P. Chassignet╇ FSU, USA, e-mail:
[email protected] Dr. James A. Cummings╇ NRL, USA, e-mail:
[email protected] Dr. Eric Dombrowsky╇ Mercator, France e-mail:
[email protected] Dr. Maria Drevillon╇ Mercator, France, e-mail:
[email protected] Dr. Diana Greenslade╇ CAWCR, Australia, e-mail:
[email protected] Dr. Fabrice Hernandez╇ Mercator, France e-mail:
[email protected] Dr. John Huckerby╇ New Zealand, e-mail:
[email protected] Dr. Harley E. Hurlburt╇ NRL, USA, e-mail:
[email protected] Dr. Gregory N. Ivey╇ UWA, Australian, e-mail:
[email protected] Dr. Simon A. Josey╇ NOC, United Kingdom, e-mail:
[email protected] Dr. Brian King╇ APASA, Australia, e-mail:
[email protected] Dr. Clothilde Langlais╇ CSIRO, Australia, e-mail:
[email protected] Dr. Pierre Yves Le Traon╇ IFREMER, France e-mail:
[email protected] Dr. Laurence D. Mann╇ Carnegie, Australia, e-mail:
[email protected] Dr. Matthew Martin╇ UK MetOffice, United Kingdom e-mail:
[email protected] Dr. Richard J. Matear╇ CAWCR, Australia, e-mail:
[email protected] Prof. Andrew M. Moore╇ UCSC, USA, e-mail:
[email protected] Prof. Charitha Pattiaratchi╇ UWA, Australia, e-mail:
[email protected] Dr. Roger Proctor╇ UTAS, Australia, e-mail:
[email protected] Mr. Tim Pugh╇ CAWCR, Australia, e-mail:
[email protected] ix
x
List of Lecturers and Students
Dr. Muthalagu Ravichandran╇ INCOIS, India, e-mail:
[email protected] Dr. Andreas Schiller╇ CSIRO, Australia, e-mail:
[email protected] Dr. Hendrik Tolman╇ NOAA, USA, e-mail:
[email protected] Mr. Geoff Wake╇ Woodside, Australia, e-mail:
[email protected] Assoc. Prof. John L. Wilkin╇ Rutgers University, USA e-mail:
[email protected] Comm. Robert Woodham╇ RAN, Australia, e-mail:
[email protected] Dr. Edward D. Zaron╇ Portland, USA, e-mail:
[email protected] Dr. Jiang Zhu╇ IAP, China, e-mail:
[email protected] Students Amjadali Amanda╇ Australia, e-mail:
[email protected] Andutta Fernando╇ Australia, e-mail:
[email protected] Backeberg Bjorn╇ South Africa/Germany, e-mail:
[email protected] Ban Natalie╇ Australia, e-mail:
[email protected] Ban Stephen╇ Australia, e-mail:
[email protected] Bean Richard╇ Australia, e-mail:
[email protected] Beck Elise╇ USA, e-mail:
[email protected] Bluteau Cynthia╇ Australia, e-mail:
[email protected] Brushett Ben╇ Australia, e-mail:
[email protected] Cheah Wee╇ Australia, e-mail:
[email protected] Choi Byoung╇ Korea, e-mail:
[email protected] Choukroun Severine╇ Australia, e-mail:
[email protected] Desportes Charles╇ France, e-mail:
[email protected] Divakaran Prasanth╇ Australia, e-mail:
[email protected] Downes Stephanie╇ USA, e-mail:
[email protected] Duchez Aurelle╇ France, e-mail:
[email protected] Durrant Tom╇ Australia, e-mail:
[email protected] Exarchou Eleftheria╇ Germany, e-mail:
[email protected] Fernandez Mariana╇ Uraguay, e-mail:
[email protected] Ford David╇ England, e-mail:
[email protected] Furner Rachel╇ England, e-mail:
[email protected] Garvey Michael╇ Australia, e-mail:
[email protected] Gasparin Florent╇ France/New Cal., e-mail:
[email protected] Geard Simon╇ Australia Hanson Christine╇ Australia, e-mail:
[email protected] Hertzel Yasha╇ Australia, e-mail:
[email protected] He Zhongjie╇ China, e-mail:
[email protected] Ishizaki Shiro╇ Japan, e-mail:
[email protected] Joseph Sudheer╇ India, e-mail:
[email protected] Law Chune Stephan╇ France, e-mail:
[email protected] Lesser Giles╇ Australia, e-mail:
[email protected] List of Lecturers and Students
Luz-Clara Moira╇ Argentina, e-mail:
[email protected] Macdonald Helen╇ Australia, e-mail:
[email protected] Meinvielle Marion╇ France, e-mail:
[email protected] Monteiro Igor╇ Brazil, e-mail:
[email protected] Morales Ruben╇ Mexico, e-mail:
[email protected] Mulet Sandrine╇ France, e-mail:
[email protected] O’Callaghan Joanne╇ New Zealand, e-mail:
[email protected] O’Loughlin Julian╇ Australia, e-mail:
[email protected] Prakya Shreeram╇ India, e-mail:
[email protected] Prandi Pierre╇ France, e-mail:
[email protected] Rahaman Hasibur╇ India, e-mail:
[email protected] Rayson Matthew╇ Australia, e-mail:
[email protected] Rozman Polona╇ Germany, e-mail:
[email protected] Rousseaux Cecile╇ Australia, e-mail:
[email protected] Shimizu Kenji╇ Australia, e-mail:
[email protected] Shu Yeqiang╇ China, e-mail:
[email protected] Stevenson Kate╇ Australia, e-mail:
[email protected] Subramanian Aneesh╇ USA, e-mail:
[email protected] Summons Nicholas╇ Australia, e-mail:
[email protected] Swart Neil╇ South Africa, e-mail:
[email protected] Swart Sebastian╇ South Africa, e-mail:
[email protected] Taebi Sohelia╇ Australia, e-mail:
[email protected] Tanajura Clemente╇ Brazil, e-mail:
[email protected] Taylor Andy╇ Australia, e-mail:
[email protected] Teixeira Carlos╇ Australia, e-mail:
[email protected] Tonbol Kareem╇ Egypt, e-mail:
[email protected] Usui Norihisa╇ Japan, e-mail:
[email protected] Volkov Denis╇ Canada, e-mail:
[email protected] Wakamatsu Tsuyoshi╇ Canada, e-mail:
[email protected] Wedd Robin╇ Australia, e-mail:
[email protected] Welhena Thisara╇ Australia, e-mail:
[email protected] Weller Evan╇ Australia, e-mail:
[email protected] Wood Julie╇ Australia, e-mail:
[email protected] Xie Jiping╇ China, e-mail:
[email protected] Zheng Fei╇ China, e-mail:
[email protected] Zhou Wei╇ China, e-mail:
[email protected] xi
Contents
Part Iâ•… Introduction 1â•…Ocean Forecasting in the 21st Century ����������������������������������尓��������������� â•…â•… 3 Andreas Schiller Part IIâ•… Oceanographic Observing System 2â•…Satellites and Operational Oceanography ����������������������������������尓���������� ╅╇ 29 Pierre-Yves Le Traon 3â•…In-Situ Ocean Observing System ����������������������������������尓������������������������� â•… ╇ 55 Muthalagu Ravichandran 4â•…Ocean Data Quality Control ����������������������������������尓�������������������������������� ╇╅ 91 James A. Cummings 5â•…Observing System Design and Assessment ����������������������������������尓��������� â•… 123 Peter R. Oke and Terence J. O’Kane Part IIIâ•… Atmospheric Forcing and Waves 6â•…Air-Sea Fluxes of Heat, Freshwater and Momentum �������������������������� â•… 155 Simon A. Josey 7â•…Coastal Tide Gauge Observations: Dynamic Processes Present in the Fremantle Record ����������������������������������尓������������������������� â•… 185 Charitha Pattiaratchi 8â•…Surface Waves ����������������������������������尓������������������������������������尓������������������� â•… 203 Diana Greenslade and Hendrik Tolman
xiii
xiv
Contents
╇ 9╇Tides and Internal Waves on the Continental Shelf ���������������������������� ╅ 225 Gregory N. Ivey Part IV╅ Modelling 10╅Eddying vs. Laminar Ocean Circulation Models and Their Applications ����������������������������������尓������������������������������������尓��������������������� ╅ 239 ╇ Bernard Barnier, Thierry Penduff and Clothilde Langlais 11╅Isopycnic and Hybrid Ocean Modeling in the Context of GODAE ���� ╅ 263 ╇ Eric P. Chassignet 12╅Marine Biogeochemical Modelling and Data Assimilation ��������������� ╅ 295 ╇ Richard J. Matear and E. Jones Part V╅ Data Assimilation 13╅Introduction to Ocean Data Assimilation ����������������������������������尓��������� ╅ 321 ╇ Edward D. Zaron 14╅Adjoint Data Assimilation Methods ����������������������������������尓������������������ ╅ 351 ╇ Andrew M. Moore 15╅Ensemble-Based Data Assimilation Methods ����������������������������������尓��� ╅ 381 ╇ Pierre Brasseur Part VI╅ Systems 16╅Overview Global Operational Oceanography Systems ��������������������� ╅ 397 ╇ Eric Dombrowsky 17╅Overview of Regional and Coastal Systems ����������������������������������尓������ ╅ 413 ╇ Jiang Zhu 18╅System Design for Operational Ocean Forecasting ��������������������������� ╅ 441 ╇ Gary B. Brassington 19╅Integrating Coastal Models and Observations for Studies of Ocean Dynamics, Observing Systems and Forecasting ���������������� ╅ 487 ╇John L. Wilkin, Weifeng G. Zhang, Bronwyn E. Cahill and Robert C. Chant 20╅Seasonal and Decadal Prediction ����������������������������������尓����������������������� ╅ 513 ╇ Oscar Alves, Debra Hudson, Magdalena Balmaseda and Li Shi
Contents
xv
Part VIIâ•… Evaluation 21â•…Dynamical Evaluation of Ocean Models Using the Gulf Stream as an Example ����������������������������������尓������������������������������������尓���� â•… 545 ╇Harley E. Hurlburt, E. Joseph Metzger, James G. Richman, Eric P. Chassignet, Yann Drillet, Matthew W. Hecht, Olivier Le Galloudec, Jay F. Shriver, Xiaobiao Xu and Luis Zamudio 22â•…Ocean Forecasting Systems: Product Evaluation and Skill �������������� â•… 611 ╇ Matthew Martin 23â•…Performance of Ocean Forecasting Systems— Intercomparison Projects ����������������������������������尓����������������������������������� â•… 633 ╇ Fabrice Hernandez Part VIIIâ•… Applications, Policies and Legal Frameworks 24â•…Defence Applications of Operational Oceanography ������������������������� â•… 659 ╇ Robert Woodham 25â•…Applications for Metocean Forecast Data—Maritime Transport, Safety and Pollution ����������������������������������尓������������������������� â•… 681 ╇ Brian King, Ben Brushett, Trevor Gilbert and Charles Lemckert 26â•…Marine Energy: Resources, Technologies, Research and Policies ���� â•… 695 ╇ John Huckerby 27â•…Application of Ocean Observations & Analysis: The CETO Wave Energy Project ����������������������������������尓������������������������������������尓������ â•… 721 ╇ Laurence D. Mann 28â•…International Marine Environmental Law (Oil Pollution) ��������������� â•… 731 ╇ Kathryn Barras Index ����������������������������������尓������������������������������������尓������������������������������������尓����� â•… 741
Part I
Introduction
Chapter 1
Ocean Forecasting in the 21st Century From the Early Days to Tomorrow’s Challenges Andreas Schiller
Abstract╇ This article provides a brief introduction to the history of oceanography with a focus on elements that laid the scientific foundation of ocean forecasting, i.e. ocean observations, ocean general circulation models and data assimilation tools. It then describes the scientific achievements of the first phase of internationally coordinated efforts in the development of global and basin-scale operational ocean forecasting systems during the Global Ocean Data Assimilation Experiment (1997– 2008). This is followed by a description of the challenges in ocean forecasting in the twenty-first century and a summary and conclusion. This article represents an introduction to the modelling, data assimilation and observing system topics discussed in more detail in the subsequent chapters of this book.
1.1â•…Brief History of Oceanography The focus of this article and this book is on the two branches of oceanography that deal with • Physical oceanography, or marine physics, that studies the ocean’s physical attributes including temperature-salinity structure, mixing, waves, internal waves, surface tides, internal tides, and currents, acoustical and optical oceanography; • and, to some extent, biogeochemical oceanography which involves the scientific study of the chemical, physical, geological, and biological processes and reactions that govern the composition of the natural environment, and the cycles of
Centre for Australian Weather and Climate Research—A partnership between CSIRO and the Bureau of Meteorology; CSIRO Wealth from Oceans National Research Flagship, Hobart, Tasmania, Australia. A. Schiller () CSIRO Marine and Atmospheric Research, Castray Esplanade, GPO Box 1538, Hobart 7001, Tasmania, Australia e-mail:
[email protected] A. Schiller, G. B. Brassington (eds.), Operational Oceanography in the 21st Century, DOI 10.1007/978-94-007-0332-2_1, ©Â€Springer Science+Business Media B.V. 2011
3
4
A. Schiller
matter and energy that transport the Earth’s chemical components in time and space. Biogeochemical oceanography focuses on chemical cycles which are either driven by or have an impact on biological activity such as carbon, nitrogen, and phosphorus cycles. We first describe briefly the general history that laid the foundation of ocean forecasting. The focus here is not on a comprehensive description of the whole science of oceanography but to focus on those components that are underpinning today’s ocean forecasting systems, in particular the development of an ocean observing system and hydrodynamic numerical modelling. Man first began to acquire knowledge of the waves, tides and currents of the seas and oceans in pre-historic times. During The Age of Discovery (approximately late 1400s to early 1700s) exploration of the oceans was primarily for cartography and mainly limited to its surfaces, although depth soundings were taken by lead line. During the beginning of the scientific voyages (late 1700s to twentieth century) in 1769 Benjamin Franklin published one of the earliest maps of the Gulf Stream (Fig.€1.1).
Fig. 1.1↜渀 Map of the Gulf Stream created by Benjamin Franklin. The Gulf Stream is depicted as the dark gray swath that runs along the east coast of what is now the United States. (Franklin 1769, Courtesy NOAA Photo Library)
1â•… Ocean Forecasting in the 21st Century
5
One of the most famous voyages of discovery of this time began in 1768 when HMS Endeavour left Portsmouth, England, under the command of Captain James Cook. Over 10 years Cook led three world-encircling expeditions and mapped many countries, including Australia, New Zealand and the Hawaiian Islands. He was an expert seaman, navigator and scientist who made observations wherever he went. James Rennell and John Purdy wrote the first scientific textbooks about currents in the Atlantic and Indian oceans during the late eighteenth and at the beginning of the nineteenth century (e.g. Rennell and Purdy 1832). The steep slope beyond the continental shelves was not discovered until 1849. Matthew Fontaine Maury’s Physical Geography of the Sea (Fig.€1.2) was the first textbook of oceanography based on his work as superintendant of the Depot of Charts and Instruments of the Navy Department in Washington D.C. (Maury 1855).
Fig. 1.2↜渀 Matthew Maury: “The Physical Geography of the Sea,” which is credited as “the first textbook of modern oceanography.” (Maury 1855)
6
A. Schiller
Fig. 1.3↜渀 Ocean surface currents around Australia from Black and Hall’s Atlas of the World published by A. & C. Black, Edinburgh (1865)
The first comprehensive maps that showed the surface circulation of the global oceans with reasonable accuracy were published by A. and C. Black in 1865 (Fig.€1.3). In 1871, under the recommendations of the Royal Society of London, the British government sponsored an expedition to explore the world’s oceans and conduct scientific investigations. Modern oceanography began with the Challenger Expedition between 1872 and 1876, when Charles Wyville Thompson and Sir John Murray launched the Challenger Expedition. It was the first expedition organized specifically to gather data on a wide range of ocean features, including ocean temperatures, seawater chemistry, currents, marine life, and the geology of the seafloor. They took water samples and temperature measurements, recorded currents and barometric pressures and collected bottom samples. The results of this expedition were published in 50 volumes covering biological, physical and geological aspects (Thompson et€al. 1880–1895). In 1893 Norwegian scientist Fridtjof Nansen allowed his ship Fram to be frozen in the Arctic ice. As a result he was able to collect valuable oceanographic, magnetic, and meteorological information in the Arctic. The rest of his career was equally as distinguished including the invention of a water-sampling bottle that permitted
1â•… Ocean Forecasting in the 21st Century
7
isolation of water samples from various depths to measure temperature, salinity and other parameters. Other European and American nations also sent out scientific expeditions (as did private individuals and institutions). The first purpose-built oceanographic ship, the Albatros was built in 1882. The four-month 1910 North Atlantic expedition headed by Sir John Murray and Johan Hjort was at that time the most ambitious research oceanographic and marine zoological project ever, and led to the classic book The Depths of the Ocean (Murray and Hjort 1912). At the beginning of the Age of Modern Oceanography (1900s to mid twentieth century) the first acoustic measurement of sea depth was made in 1914. Between 1925 and 1927 the Meteor expedition surveyed the Mid-Atlantic Ridge and gathered 70,000 ocean depth measurements using an echo sounder. Virtually all civilian ocean research ceased in 1939 with the outbreak of World War II, when scientific resources were mobilised. However, many advances were made in instrumentation, and our understanding of the ocean was greatly improved. For example, there were major advances in predicting wave conditions (important for amphibious invasions). Mapping features of ocean basins was greatly expanded to improve the ability to detect submarines. In 1942, Sverdrup et€al. (1942) published The Ocean which was a major landmark in oceanography. The nineteenth and twentieth century also saw major progress towards quantitative descriptions of observed phenomena. Examples of key areas of progress are (some of which are tightly linked to progress in meteorology): • the rotation of the Earth and associated impact on ocean currents (Coriolis 1835); • the effect of winds on the ocean-atmosphere interface (Ekman 1905); and • the development of vorticity theories and theorems for the ocean as an extension to Newton’s law in a rotating fluid (Ertel 1942; Sverdrup 1947). This enhanced capability to describe the ocean within a mathematical framework allowed the development of numerical models. Consequently, from the 1970s onwards there has been increased emphasis on the application of computers for oceanography to allow numerical simulations and predictions of the state of the ocean. The Mid-Ocean Dynamics Experiment (MODE) was one of the first large-scale and extensively instrumented field experiments carried out by physical oceanographers. Conducted in two phases between November 1971 and July 1973, the experiment explored the role of mesoscale eddy motions in the dynamics of general oceanic circulation (mesoscale eddies are at the centre of attention in today’s largescale ocean forecasting systems). The 1970s and 1980s also saw the development and first applications of socalled inverse methods to oceanographic data (e.g. Wunsch 1978). These methods can be interpreted as simple data assimilation tools that paved the way for the development of more complex data assimilation and model initialization tools used nowadays in ocean forecasting systems and often derived from numerical weather prediction applications.
8
A. Schiller
The Tropical-Ocean-Global-Atmosphere (TOGA) Program began in 1985 and was a ten-year research effort to investigate the global atmospheric response to the coupled ocean-atmosphere forcing from the tropical regions. It was among the first large-scale programs that addressed the predictability of the coupled tropical oceans and global atmosphere by drawing on observations and by recognizing the key role of models for understanding tropical air-sea interactions as a prerequisite for launching successful climate predictions into the future. In the 1980s the TAO/TRITON oceanographic buoy array was established in the Pacific to allow monitoring and ultimately prediction of El Niño events (http:// www.pmel.noaa.gov/tao/proj_over/taohis.html). Enhancements to the in situ and satellite observing system together with the first evolving model lead to the first successful ENSO prediction (Zebiak and Cane 1987). 1990 saw the start of the World Ocean Circulation Experiment (WOCE) which continued until 2002. WOCE was a component of the international World Climate Research Program, and aimed to establish the role of the world ocean in the Earth’s climate system. The WOCE field phase ran between 1990 and 1998 (Fig.€1.4), and was followed by an analysis and modelling phase that ran until 2002. The results are summarised in “Ocean Circulation and Climate: Observing and Modelling the Global Ocean” (Siedler et€al. 2001). Before the 1980s, when satellites became more commonly available, oceanographers were “data poor”. Since then, significant technological and scientific advances in satellite remote sensing provide near-real time measurements of sea surface height anomalies, SST and ocean colour. These key observations have, for the first time, enabled ocean forecasting applications (Fu and Cazenave 2001). The realisation of the network of 3,000 Argo profiling floats freely reporting temperature and salinity profiles to 2,000€m depth in a timely fashion has transformed the in situ ocean measurement network in the new millennia (Fig.€1.5). This allows, for the first time, continuous monitoring of the temperature, salinity, and velocity of the upper ocean, with all data being relayed and made publicly available within hours after collection. Based on significant advances in supercomputing technologies, the 1990s saw the emergence of the first large-scale eddy-resolving models (Semtner and Chervin 1992) and the first ocean-atmosphere coupled climate change projections (see, e.g. IPCC First Assessment Report 1990). More detailed accounts of the history of oceanography can be found in the published literature and, e.g. at http://core.ecu.edu/geology/woods/HISTOCEA.htm.
1.2â•…The Achievements of GODAE (1997–2008) As described in the previous paragraphs, over the last 20 years the global ocean observing system (↜in situ and remote sensing) has been progressively implemented and led to a revolution in the amount of data available for research and forecasting applications. The ocean observing system, primarily designed to serve climate
Fig. 1.4↜渀 WOCE Hydrographic Program One Time Survey (1990–1998) (http://woce.nodc.noaa.gov/wdiu/diu_summaries/whp/figures/whpot.htm)
1â•… Ocean Forecasting in the 21st Century 9
10
A. Schiller
60°N 30°N 0°N 30°S 60°S 60°E
120°E
180°
120°W
60°W
0°
Fig. 1.5↜渀 Status of global Argo float array, December 2009 (http://www.argo.ucsd.edu/)
research, is used as a backbone for most operational oceanography applications. Although significant progress has been made, sustaining the global ocean observing system remains a challenging task (Clark et€al. 2009). This recent progress in the global ocean observing system was complemented by advances in supercomputing technology, allowing the development and operational implementation of eddyresolving (~10€km) basin-scale ocean circulation models. The Global Ocean Data Assimilation Experiment (GODAE) was set up in 1997 with the aims of (i) demonstrating the feasibility and utility of global ocean monitoring and forecasting on the daily to weekly time scale and on eddy-resolving spatial scales and (ii) to assist in building the infrastructure for global operational oceanography (Smith and Lefebvre 1997; GODAE Strategic Plan 2000; Bell et€al. 2009). From its inception in 1997 to its conclusion in 2008, GODAE has had a major impact on the development of global operational oceanography capability. Global modelling and data assimilation systems have been progressively developed, implemented and inter-compared (Dombrowsky et€al. 2009; Cummings et€al. 2009; Hernandez et€al. 2009). In-situ and remote sensing data are now routinely assimilated in global and regional ocean models to provide an integrated description of the ocean state. Observation, analysis and forecast products are readily accessible through major data and product servers (Blower et€al. 2009). There has been increased attention to the development of products and services and the demonstration of their utility for applications such as marine environment monitoring, weather forecasting, seasonal and climate prediction, ocean research, maritime safety and pollution forecasting, national security, the oil and gas industry, fisheries management and coastal and shelf-sea forecasting (Davidson et€al. 2009; Hackett et€al. 2009; Jacobs et€al. 2009). GODAE as an experiment ended in 2008 having achieved most of its goals. It has been demonstrated that global ocean data assimilation is feasible and GODAE has made important contributions to the establishment of an effective and efficient infrastructure for global operational oceanography that includes the required observing systems, data assembly and processing centres, modelling and data assimilation centres and data and product servers.
1â•… Ocean Forecasting in the 21st Century
11
1.3â•…Key Future Research Priorities in Ocean Forecasting Although there are still major challenges to face (e.g. completing and sustaining the global ocean observing system being an obvious one), global operational oceanography now needs to transition from a demonstration to a permanent and sustained capability. Operational1 data and products are needed for most applications as well as for climate research. This is critical for applications which cannot develop without operational services. In parallel, continuous improvements of operational oceanography systems are needed to satisfy new requirements (e.g. for coastal zone and ecosystem monitoring and forecasting, climate monitoring).
1.3.1 The Challenges for the Next Decade Most national forecasting centres have or are now transitioning towards operational or pre-operational status. Ocean forecasting systems are also evolving to satisfy new requirements just mentioned and must benefit from scientific advances in ocean modelling and data assimilation. International collaboration and coordination of both operational and research activities related to ocean analysis and forecasting must continue during this sustained operational phase. The challenges and expectations are very demanding and can only be achieved through international collaboration. These main challenges and opportunities for the next decade are summarised below. During the last decade new pressing societal issues to which ocean analysis and forecasting can make substantial contributions have evolved. They are now quite diverse and are not limited to open ocean forecasts (although open ocean forecasts will continue to serve major applications areas). The most important are: • The use of data assimilation to provide integrated descriptions of the global ocean state (reanalyses) and to characterise and detect climate change in the ocean; • The application of ocean prediction techniques to the prediction of climate change (so-called decadal prediction); • The assessment and characterisations of specific sources of uncertainty in downscaling of climate and climate-change scenario simulations and predictions in studies of the impact of climate change in coastal regions (e.g. extreme events, flooding, ecosystems); • The development of improved atmospheric and climate forecasts (near coasts, hurricanes/tropical cyclones, monsoons, seasonal); • Real-time forecasting in near-shore / coastal waters (physics, biogeochemistry and ecosystems) and coupling between open ocean and coastal areas; 1╇ Following the GODAE Strategic Plan (2000), “operational” is used here “whenever the processing is done in a routine and regular way, with a pre-determined systematic approach and constant monitoring of performance. With this terminology, regular re-analyses may be considered as operational systems, as may be organized analyses and assessment of climate data”.
12
A. Schiller
• Ecosystem modelling and the development of ecosystem based management of marine resources (influence of physical transports and processes on marine life, modelling up to high trophic levels); • Marine environment monitoring in support of policies (e.g. European Marine Strategy). Continuous improvement of operational oceanography systems and the development of new capability is needed to address these new societal needs. This demands state-of-the art research leadership and calls for dedicated cooperation with international research programs such as CLIVAR, GEOTRACES, SOLAS and IMBER2. In the following paragraphs we address some of the main research topics that operational oceanography faces: high resolution physical modelling, downscaling, biogeochemical and ecosystem modelling, ocean-wave-atmosphere coupling, data assimilation and coupled data assimilation, error estimates, long-term reanalyses and use of new observations. Major developments in the decade will see a maturing of eddy-resolving dataassimilating models, and a stronger integration into coupled numerical weather prediction and climate modelling.
1.3.2 Ocean Modelling The science about turbulent closure schemes is now fairly mature, but there may still be surprises associated with subtle aspects of vertical mixing in the deep ocean that may have important consequences on long time scales. Vertical mixing is also critically important for biogeochemical cycles, because it controls the return of nutrients to the surface euphotic zone, and therefore the magnitude of primary production. Another area where there still is room for improvement concerns the exchanges of heat, momentum and freshwater across the ocean surface. Accuracy, resolution, and extent (in time ahead) of wind forecasts are the primary limiting factors for sea-state and surge forecasting. Likewise, sea surface heat exchange is clearly a determining factor in forecasting ocean mixed-layer depth and ice formation. In both cases, the need for dynamically coupled ocean-wave-iceatmosphere models is an essential element to improve atmospheric forcing. Coastal ocean modelling and forecasting is a major challenge for the scientific community because of the specific and rich dynamics of those regions, and because of the various couplings with the lower atmosphere and exchanges with the near-shore and offshore regions. These issues, needs and challenges have led to the development of a wide range of models of various types. Phenomena of interest include coastal current interactions, coastal meso-scale, tides and storm surges, 2╇ CLIVAR╛=╛World Climate Research Program (WCRP) project that addresses Climate Variability and Predictability. IMBER╛=╛Integrated Marine Biogeochemistry and Ecosystem Research; GEOTRACES╛=╛International study of the global marine biogeochemical cycles of trace elements and their isotopes; SOLAS╛=╛Surface Ocean Lower Atmosphere Study.
1â•… Ocean Forecasting in the 21st Century
13 14
60 N
13 58 N 12 56 N 11 52 N 10 50 N 9
48 N
8
44 N
12 W 10 W 8 W
6W
4W 2W
0E
7 °C
Fig. 1.6↜渀 Application of the 2-way grid-refinement software AGRIF to the Bay of Biscay, tested in the framework of the MERSEA project (Cailleau et al. 2008). The large-scale model is a 1/3° (Mercator grid) North Atlantic configuration of the NEMO Ocean general circulation model. The fine-scale model is regional configuration of NEMO at a resolution of 1/15° (Mercator grid). Both models are run simultaneously and interactively for years of simulation on either vector or massively parallel super computers. The computational surcharge induced by the 2-way coupling of the grids is very small (just a few percent). The regional model benefits from the smooth and regular behaviour of the large-scale model at its open boundaries. On longer time-scales, the largescale model benefits from the local improvements brought by the high resolution to the representation of the dynamics in the Bay of Biscay, especially the slope current. The above figure displays a sea surface temperature snapshot on 22 March 1996. One shall notice the fine-scale and the intense eddy field of the fine-grid model, but also the continuity at the limit between the two grids
tsunamis, shoreline change, coastal upwelling, river plumes and regions of freshwater influence, atmosphere-driven processes, surface waves, and sea ice (Fig.€1.6). Coastal ocean systems can have very high spatial gradients in both the vertical and horizontal, especially near river mouths, requiring the use in models of sophisticated mixing schemes, and high order numerics. The key constraints on the accuracy of these models now lie with the specification of input data (bathymetry, bottom roughness, lateral and surface forcing). In these shallow systems, and especially along exposed shorelines, wave-current interactions play an important role. Measuring and predicting exchanges between the underlying sediment and the water
14
A. Schiller
column is critical for coastal biogeochemistry, and is still a key challenge. Sediment models attempt to represent the effects of re-suspension and deposition of particulate material, and their interaction with the circulation, on suspended concentrations (turbidity, important for optical properties and hence primary production) and bed thickness and composition (geomorphology). Models of these processes are still under active development. With respect to biological processes, we are faced with the general problem of biogeochemical and ecosystem modelling (used here synonymously); namely, choosing the right level of abstraction and approximation in describing and predicting the structure and function of a complex system with many nested levels of complexity.
1.3.3 Initialisation and Forecasting There are still some significant challenges in the data-assimilation techniques themselves, and one can expect to see significant improvement there. The assimilation of observations into present-day ocean models is still far from being optimal. Improved estimates of the state of the physical ocean, marine ecosystems and oceanatmosphere interactions will rely upon new cross-cutting research directions in terms of both methods and operational implementations. In meteorology (the history of which predates the evolution of ocean forecasting), the implementation of data assimilation methodology has followed a progressive pace starting with optimal interpolation, followed by sequential approaches and today most larger NWP centres are increasingly investing in 4D-VAR variational approaches with a noticeable increase in interest in ensemble approaches. Operational oceanography is today at the stage of applying sequential approaches but variational methodologies are on the verge of being used, at least for seasonal forecast applications. Because of the specificities of oceanography (e.g. the mesoscale non-linearities) it is still unclear whether 4D-VAR is fully applicable (Luong et€ al. 1998) and further research must be undertaken in this direction. A promising way might be the hybridisation between variational and sequential approaches thus combining advantages of both methodologies (Robert et€al. 2006). However, 4D-VAR systems have not been comprehensively tested for highly non-linear applications. For instance, as we move to higher resolution and longer predictive time scales, the assumptions that underpin VAR systems (e.g. linearity in tangent-linear models) become less valid. The development of data assimilation into physical coastal ocean models has lagged behind its development in basin-scale models, and is still in its infancy. Current methods need to be tested and enhanced for coastal applications. Data assimilation in coastal models has a vital role to play, not only as a tool to provide short-term forecasts, but more importantly for the rigour it brings to the analysis of model error, and to the design of observing systems (see the CSSWG White Paper, De Mey et€al. 2007 for a detailed account).
1â•… Ocean Forecasting in the 21st Century
15
Biogeochemical modelling and data assimilation are much less mature than physical modelling. Consequently, there is a strong need for both on-going development and validation of biogeochemical and, ultimately, ecosystem models. The impact of the physical models on the accuracy of the ecosystem models is of particular importance (e.g. Berline et€al. 2007). High horizontal and vertical resolution physical models are required to resolve the physical features that are critical to the ecosystem. Errors in physical models are problematic and can render outputs from ecosystem models meaningless. Vertical velocities are a particular example as they are critical for nutrient transport. In coastal areas correct representation of optical depth is also critical for primary production (Fig.€1.7). This requires accurate suspended sediment concentrations. These requirements for accuracy present a challenge for physical models. A major trend in environmental research in the coming decade will see the development of the next generation weather, climate, and Earth system monitoring, assessment, data-assimilation, and prediction systems (Shapiro et€al. 2008). These
Fig. 1.7↜渀 MODIS image of ocean colour off Australian NW shelf. The figure illustrates the complex processes acting in the coastal zone due to blending of different time/ space scales (e.g. ocean-shelf topographic interaction). Forecasting systems operating in such complex environments require sophisticated multi-scale (nested) models and scale-sensitive observing systems for accurate initialization (Courtesy: CSIRO Marine and Atmospheric Research)
16
A. Schiller
systems will no longer focus on individual components of the Earth system (such as the oceans) but aim at treating the complex physical and biogeochemical components as one system. Coupled data assimilation means that observations in one medium impact the state of the other medium. In 4D-VAR fully coupled assimilation means simultaneous minimization of the cost function of the component models, e.g. atmosphere and ocean. An example of a less complex system is coupled ocean-atmosphere modelling. Ultimately truly coupled physical-biogeochemical initialization systems need to be developed, whereby the ocean, sea-ice, land surface, and atmosphere are initialized in unison. Consequently, a key challenge in data assimilation over the next decade will be the development of data assimilation techniques for Earth system modelling that are fit-for-purpose for a wide range of applications, including ocean-atmosphere weather forecasting, seasonal-to-decadal and climate change prediction.
1.3.4 The Global Ocean Observing System Over the last 10 years, a global ocean observing system (↜in situ and remote sensing) has been progressively implemented. The system, primarily designed to serve climate research, is used as a backbone for most operational oceanography applications. Although significant progress has been made (e.g. Argo and Jason are outstanding successes), sustaining the global ocean observing system remains a challenging task (Freeland et€al. 2010; Wilson et€al. 2010). There is also a pressing need to develop further regional and coastal components and, as discussed above, to extend the measurement capabilities to biogeochemical parameters. This endeavour is clearly beyond the scope of ocean analysis and forecasting teams and involves major international programs or intergovernmental organizations (e.g. WMO and IOC through JCOMM, GOOS and GCOS, GEOSS, CEOS) and research programs (e.g. WCRP, IGBP and SOLAS)3. Nowadays, use is made of observations from satellites, autonomous floats, onshore devices (radar, tide gauges etc.), off-shore moorings, aircraft, AUVs (Autonomous Underwater Vehicles), VOS (Voluntary Observing Ships) and more. Especially in the coastal zone more and better observational data, extending over longer periods, are essential if modelling accuracy and capabilities are to be enhanced (Malone et€al. 2010). International collaboration is an obvious and valuable means of achieving this goal. While international funding supports some satellite programs (although most of these are still regarded as non-operational), synergistic in situ monitoring presently relies on national funding. Examples are the Argo profiling 3╇ WMO╛=╛World Meteorological Organisation; IOC╛=╛Intergovernmental Oceanographic Commission; JCOMM╛=╛Joint Committee for Oceanography and Marine Meteorology; GOOS╛=╛Global Ocean Observing System; GCOS╛=╛Global Climate Observing System; GEOSS╛=╛Global Earth Observation System of Systems; CEOS╛=╛Committee on Earth Observation Satellites; WCRP╛=╛World Climate Research Program; IGBP╛=╛International Geosphere-Biosphere Program.
1â•… Ocean Forecasting in the 21st Century
17
Fig. 1.8↜渀 Liverpool Bay Coastal Observatory in the Irish sea, indicating simultaneous multiparameter measurements and satellite AVHRR sea surface temperatures. (Courtesy Roger Proctor, Proudman Oceanographic Laboratory, UK)
floats, the TAO/TRITON array in the Pacific (USA and Japan), the PIRATA array in the Atlantic (France, USA and Brazil) and the IndOOS array in the Indian Ocean (India, USA and Japan). These basin-scale observing systems are subject to international coordination whereas design and implementation of coastal ocean observing systems are largely the responsibility of individual national efforts (Fig.€1.8). Despite the limited progress in implementing ocean biogeochemical observing systems there is an increasing user pull for enhanced ocean forecasting capability that includes information about physics, biogeochemistry and ultimately ecosystem components. The biogeochemical and physical systems interact on a variety of processes and scales. Most notable is the impact of biology and associated attenuation depth of light on solar shortwave penetration and thus mixed-layer depth, and the corollary, the impact of suspended material on light scattering and penetration and biological production. Consequently, joint assimilation of physical and ecosystem observations likely will benefit both components though the challenges involved are manifold.
1.3.5 Observing System Design and Adaptive Sampling Ocean analysis and forecasting systems are an appropriate and powerful means to assess the impact of the observing system, to identify gaps and to improve the efficiency/effectiveness of the observing system. An enhanced focus on observing system design and adaptive sampling in data assimilating systems will allow as-
18
A. Schiller
sessments of individual components of the observing system and provide scientific guidance for improved design and implementation of the ocean observing system. OSEs (Observing System Evaluation) assess the impact of existing individual components of the observing system on forecast skills, whereas OSSEs (Observing System Simulation Experiments) are tools for planning new observing systems. OSEs undertaken during GODAE demonstrate that global and regional forecast systems strongly depend on the availability of high resolution altimeter data (e.g. Pascual et€al. 2006). Significant degradation of the performance of these forecasting systems (e.g. forecast skill) and applications (e.g. offshore industry in the Gulf of Mexico) was thus observed when the number of available altimeters was reduced from three to two due to the unavailability of ENVISAT data. OSSEs in the Indian Ocean have provided an estimate of the respective contribution of Argo, XBT and moorings to the observing system in the Indian Ocean (e.g. Sakov and Oke 2008). These are extremely valuable tools to develop an improved understanding of the ocean and to help the design of global and regional observing systems. While OSEs and OSSEs provide an integrated, but methodology-dependent, performance assessment of an observational array, recently proposed approaches based on the representer matrix spectrum (e.g. Hénaff et€al. 2008) focus on the capacity of a given array to detect model errors. This can be achieved independently of any data assimilation method, e.g. from stochastic modelling, or as part of an Ensemble Kalman Filter. An evolving method for optimising observing arrays is adaptive sampling (e.g. Wilkin et€al. 2005). The key idea of adaptive sampling is that the initial estimate or observation can detect correlations in the environment, providing information about the number of future observing platforms needed or to specify the frequency and spatial distribution required for future sampling certain features in the environment (e.g. eddies, fronts etc.). Thus, adaptive sampling can save costs compared to dense, non-adaptive sampling, and, simultaneously, provide high-resolution information where needed.
1.4â•…Scientific Objectives of GODAE OceanView The GODAE OceanView Science Team (GOVST) was established in 2008, with the mission to define, monitor, and promote actions aimed at coordinating and integrating research associated with multi-scale and multidisciplinary ocean analysis and forecasting systems, thus enhancing the value of GODAE OceanView outputs for research and applications. Over the next decade, the science team will provide international coordination and leadership in: • The consolidation and improvement of global and regional analysis and physical forecasting systems. • The progressive development and scientific testing of the next generation of ocean analysis and forecasting systems, covering biogeochemical and ecosys-
1â•… Ocean Forecasting in the 21st Century
19
tems as well as physical oceanography, and extending from the open ocean into the shelf sea and coastal waters. • The exploitation of this capability in other applications (weather forecasting, seasonal and decadal prediction, climate change detection and its coastal impacts). • The assessment of the contribution of the various components of the observing system and scientific guidance for improved design and implementation of the ocean observing system. Members as representatives of national ocean forecasting systems of GODAE OceanView adhere to the same principles of free, open and timely exchange of data and products, sharing of scientific results and experience developing applications which were important factors in the success of GODAE. The societal benefits from these systems will only be realised through joint work with other teams of experts. Potential benefits include improvements in the day-to-day management of coastal waters, the management of marine ecosystems, weather prediction from hours to decades ahead, and the expected impacts of climate change on the oceans and coastal waters. The GOVST develops linkages with other groups and reports on its progresses, achievements and recommendations. As GODAE prototype systems transition to operational systems, international collaboration on product standardization and interoperability between systems must be maintained and developed. The WMO/IOC Joint Technical Commission for Oceanography and Marine Meteorology (JCOMM) provides an appropriate intergovernmental mechanism for the coordinating role and has recently established an Expert Team on Operational Oceanographic Forecasting Systems (ET-OOFS) within its Services Program Area for this purpose. GODAE OceanView informally reports to JCOMM and has strong links with JCOMM ETOOFS. GODAE OceanView coordinates the development of new capabilities, in cooperation with other relevant international research programs, through a number of task teams. The initial list of GODAE OceanView Task Teams includes: • Intercomparison and Validation Task Team: The team pursues activities developed during GODAE. It coordinates and promotes the development of scientific validation and intercomparison of operational oceanography systems. Activities include the definition of metrics to assess the quality of analyses and forecasts (e.g. forecast skills) both for physical and biogeochemical parameters and the setting up of specific global and regional intercomparison experiments. Metrics related to specific applications are also defined. The team liaises with the JCOMM ET-OOFS team for operational implementation. • Observing System Evaluation Task Team: One of the aims of GODAE OceanView is to formulate more specific requirements for observations on the basis of improved understanding of data utility. The team is jointly formed by GODAE OceanView and the GOOS Ocean Observations Panel for Climate (OOPC). Through the task team, GODAE OceanView
20
•
•
A. Schiller
and OOPC partners get organized at the international level to provide consistent and scientifically justified responses to agencies and organizations in charge of sustaining the global and regional ocean observing systems used for ocean monitoring and forecasting at short-range, seasonal and decadal time-scales. This activity requires harmonized protocols for observation impact assessment (e.g. OSEs and OSSEs), tools for routine production of appropriate diagnostics using NWP-derived methods, common sets of metrics for intercomparison of results, and objective methodologies which can be used to provide recommendations to the appropriate agencies and organizations. In the longer term consideration will need to be given to an evaluation strategy for identifying observing system requirements for different, possibly user-specific, applications. Coastal Ocean and Shelf Seas Task Team: This task team deals with scientific issues in support of multidisciplinary analysis and forecasting of the coastal transition zone and shelf/open ocean exchanges in relation with the larger-scale efforts. The specific objectives include: (1) discuss and promote the uses of GODAE OceanView products and results for coastal ocean forecasting systems and for coastal applications in a wider community; (2) discuss and foster integration of the varied routine sources of information in coastal ocean forecasting systems: large-scale forecasts, satellite observations, coastal observatories, etc.; discuss and support the development of coastal observing systems in terms of science and technology; (3) discuss the key physical and biogeochemical processes which have the greatest impact on modeling and forecasting quality and their utility for applications; this includes validation and forecast verification; (4) discuss and promote state-of-the-art methodology such as two-way coupling, unstructured-grid modeling, downscaling, data assimilation and array design. Marine Ecosystem Monitoring and Prediction Task Team: The integration of new models and assimilation components for ocean biogeochemistry and marine ecosystem monitoring and prediction will be required to bridge the gap between the current status and new applications in areas such as fisheries management, marine pollution and carbon cycle monitoring. The Task Team has been set up with the goal to define, promote and coordinate actions between developers of operational systems and ecosystem modelling experts, in tight connection with IMBER. The objectives of the task team are (1) to design appropriate ecosystem modelling and assimilation strategies that will be compatible with the functionalities of operational systems; (2) to develop numerical experiments aimed at improving, assessing and demonstrating the value of operational products for marine ecosystem monitoring and prediction; (3) to expand the concept of the “GODAE metrics” to biogeochemical variables and to coordinate intercomparison exercises across international groups to assess implementation progress and performances; (4) to identify the essential sets of physical and biogeochemical observations required to constrain the coupled models and to formulate relevant recommendations to further develop the global ocean observing system; (5) to promote and organise educational activities (summer schools, training workshops, etc.) aimed at sharing experience between young
1â•… Ocean Forecasting in the 21st Century
21
scientists, operational oceanographers and marine ecosystem experts. In addition to the link with IMBER, the task team articulates its activities with other relevant international programs such as GEOTRACES and SOLAS.
1.5â•…Summary and Conclusions Over the past 40 years, numerical modelling has developed rapidly in scope (from hydrodynamics to ecology) and resolution (from one-dimensional, 102 elements to three-dimensional, 108 elements) exploiting the contemporaneous development of computing power. Although we have made significant progress with the implementation of the global ocean observing system, concurrent development in observational capabilities has not been achieved yet in areas demanding high spatial resolution such as coastal domains (despite exciting advances in areas such as in remote sensing and sensor technologies). Nowadays, diverse applications involving ocean forecasting systems range from short term prediction of the three-dimensional circulation and density fields, waves, tides and storm surges to coupled ocean-atmosphere-land scenario forecasting of the effects of global climate change on terrestrial, fluvial and ecology over millennia. The accuracy of model simulations depends on the availability and suitability (accuracy, resolution and duration) of both observational and linked meteorological, oceanic and hydrological model data to set-up, force, and assess calculations. Modelling is at a stage where major and sustained investments are required in infrastructure and organisation: e.g. access to supercomputers, software maintenance and data exchange (Shapiro et€al. 2008). Many research approaches developed under GODAE are just at their beginning and will require ongoing international research collaboration and coordination. There are still many challenges related to the development of services and links with end users (which are beyond the scope of this chapter). On the scientific side, many of the fundamental modelling issues that were evoked in the book edited by Chassignet and Verron (1998) are still unresolved. They represent new challenges and require step changes to our current efforts. An incomplete list of scientific challenges follows: • Ocean modeling (for a more comprehensive list about ocean modelling issues see Griffies et€al. 2010): − Mesoscale eddying models can exhibit numerical diapycnal diffusion far larger than is observed. Spurious diapycnal mixing originating from numerical advection remains an issue, with consequences of variable and/or eddyresolving resolutions and dynamical meshes largely unexplored. Reducing the level of spurious diapycnal mixing in models facilitates collaborative efforts to incorporate mixing theories into simulations, which in turn helps to focus observational efforts to measure mixing and determine its impact on ocean circulation. Progress has been made to rectify this problem through improve-
22
A. Schiller
−
−
−
−
−
ments to tracer advection schemes, but further work is needed to quantify these advances. Largely unexplored areas of research involve the local scaling of viscosity and diffusivity coefficients. Lateral viscous friction remains the default approach for closing the momentum equation in ocean models. However, large levels of lateral viscous dissipation used by models do not mimic energy dissipation in the real ocean. The ocean floor should be represented continuously across finely resolve mesh regions to faithfully simulate topographically influenced flows. This property is routinely achieved with terrain following vertical coordinates, yet optimal strategies for unstructured mesh models remain under investigation. Large-scale ocean-waves-atmosphere coupling remains an area of active research. While wind-induced surface waves contribute primarily to mixing through generation of internal waves at the ocean surface, geostrophic motions may also sustain wave induced interior mixing. In addition, tidal waves can affect the whole water column. Submesoscale fronts and related instabilities are ubiquitous, and those active in the upper ocean provide a relatively rapid restratification mechanism that should be parameterized in ocean simulations, even those resolving the mesoscale. The coupling between physical, biogeochemical and ecosystem models in terms of consistency of scales, processes resolved and consistent parameterisations requires further research.
• Observing systems: − The exploration of impact of new types of observations on forecasting systems (e.g. remotely sensed sea surface salinity, high resolution wide swath altimetry) requires dedicated efforts and resources. − In collaboration with international programs such as IMBER and SOLAS research is under way on the implementation of real-time biogeochemical and ecosystem ocean observing systems, e.g. cost-effective sensor-technologies. − An enhanced focus on observing system design and its analogue of adaptive sampling will allow assessments of individual components of the observing system and provide scientific guidance for improved design and implementation of the ocean observing system. • Data assimilation: − Development of data assimilation tools such as coupled atmosphere-ocean initialisation techniques that are fit-for-purpose for a wide range of applications, including short-range, seasonal-to-decadal and climate change prediction (in collaboration with WMO programs) is work-in-progress. − Efficient data assimilation techniques for biogeochemical and ecosystem modules of ocean circulation models are being developed that are fit for operational purposes.
1â•… Ocean Forecasting in the 21st Century
23
− Another research focus is the representation of model and data errors using ensemble methods based on various forecasting systems thus delivering more accurate background error estimates. − Multi-scale data assimilation and joint estimation of interior and open boundary solutions in nested systems remain largely unresolved. • Coastal ocean: − Users increasingly demand an extension of the critical path of routinelyavailable global information (satellite and in-situ observations, nowcasts and forecasts) to coastal and littoral applications. − A prerequisite for an enhanced user uptake of coastal ocean forecasts are enhancements to existing systems and development of new coastal ocean forecasting systems that downscale (and upscale, i.e. two-way coupling) the global basin-wide model estimates as part of the local data assimilation problem, resolving the rich scale interactions, tides and high frequencies, and experimenting novel approaches such as coupled modelling and unstructured grid modeling. − These forecasting tools will need to be able to contribute to the objective design of observing systems for the coastal ocean, such as new satellite sensors, coastal observatories, etc.; use of such observations in the local forecasting system and upscaling of the information to the basin-scale systems. Consequently, ocean forecasting in the twenty-first century still faces many challenges with time scales ranging from weather to climate. It is inherently an international issue, requiring broad collaboration to span the global oceans; it is beyond the capability of any one country. Over the past decade, GODAE through its International GODAE Steering Team (IGST) has coordinated and facilitated the development of global and regional ocean forecasting systems and has made excellent progress. GODAE as an experiment has ended in 2008. The next decade will spawn new research activities in ocean forecasting under the auspices of the GODAE OceanView Science Team that will build on the success of GODAE. GODAE OceanView will promote the development of ocean modelling and assimilation in a consistent framework to optimize mutual progress and benefit. It will promote the associated utilization of improved ocean analyses and forecasts and will provide a means to assess the relative contributions of and requirements for observing systems, and their respective priorities. The GODAE OceanView programme will result in the long-term international collaboration and cooperation that is required for the next, sustained, phase of operational oceanography in the twenty-first century. The grand vision and key research challenge is to develop coupled initialisation systems of numerical weather prediction and eddy-resolving ocean models. These systems will contribute to and benefit from recent progress in Earth systems modelling. With increasing computing resources the next decade is also likely to see an even stronger emphasis on “seamless” integrations across time and space scales,
24
A. Schiller
covering global, regional and coastal/near-shore ocean prediction systems and addressing an increasing number of user applications. Acknowledgements╇ This paper was written with inputs from the former members of the GODAE International Science Team, and, more recently, the members of the GODAE OceanView Science Team and their Patrons groups. The author would like to particularly thank Pierres-Yves Le Traon, Mike Bell, Eric Dombrowsky, Kirsten Wilmer-Becker, Pierre Brasseur, Pierre De Mey, Roger Proctor, Jacques Verron, Peter Oke and John Parslow for their contributions through many discussions on issues of relevance to this paper.
References Bell MJ, Lefèbvre M, Le Traon P-Y, Smith N, Wilmer-Becker K (2009) GODAE: the global ocean data assimilation experiment. Oceanog Mag 22(3):14–21 (Special issue on the Revolution of Global Ocean Forecasting—GODAE: 10 years of achievement) Berline L, Brankart JM, Brasseur P, Ourmières Y, Verron J (2007) Improving the physics of a coupled physical–biogeochemical model of the North Atlantic through data assimilation: impact on the ecosystem. J Mar Syst 64(1–4):153–172 Black A, Hall S (1865) Black’s general atlas of the world. A&C Black, Edinburgh Blower JD, Blanc F, Clancy M, Cornillon P, Donlon C, Hacker P, Haines K, Hankin SC, Loubrieu T, Pouliquen S, Price M, Pugh TF, Srinivasan A (2009) Serving GODAE data and products to the ocean community. Oceanogr Mag 22(3):70–79 (Special issue on the Revolution of Global Ocean Forecasting—GODAE: 10 years of achievement) Cailleau S, Fedorenko V, Barnier B, Blayo E, Debreu L (2008) Comparison of different numerical methods used to handle the open boundary of a regional ocean circulation model of the Bay of Biscay. Ocean Model 25(1–2):1–16. doi:10.1016/j.ocemod.2008.05.009 Chassignet EP, Verron J (1998) Ocean modeling and parameterization. In: Chassignet EP, Verron J (eds) Proceedings of the NATO advanced study Institute on ocean modeling and parameterization, Kluwer Acadamic, Dordrecht, p€451. Les Houches, France, 20–30 Jan 1998 (NATO ASI Series C, 516) Clark C, In Situ Observing System Authors, Wilson S, Satellite Observing System Authors (2009) An overview of global observing systems relevant to GODAE. Oceanogr Mag 22(3):22–33 (Special issue on the Revolution of Global Ocean Forecasting—GODAE: 10 years of achievement) Coriolis G (1835) Memoire sur le equations du movement relative des systems de corps. J de l’Ecole Royale Polytechnique 15:142 Cummings J, Bertino L, Brasseur P, Fukumori I, Kamachi M, Martin MJ, Mogensen K, Oke P, Testut CE, Verron J, Weaver A (2009) Ocean Data Asimilation Systems for GODAE. Oceanogr Mag 22(3):96–109 (Special issue on the Revolution of Global Ocean Forecasting—GODAE: 10 years of achievement) Davidson FJM, Allen A, Brassington GB, Breivik Ø, Daniel P, Kamachi M, Sato S, King B, Lefevre F, Sutton M, Kaneko H (2009) Applications of GODAE ocean current forecasts to search and rescue and ship routing. Oceanogr Mag 22(3):176–181 (Special issue on the Revolution of Global Ocean Forecasting—GODAE: 10 years of achievement) De Mey P, Craig P, Kindle J, Ishikawa Y, Proctor R, Thompson K, Zhu J (2007) Towards the assessment and demonstration of the value of GODAE results for coastal and shelf seas and forecasting systems. GODAE White Paper, GODAE Coastal and Shelf Seas Working Group (CSSWG), 2nd ed, p€79 Dombrowsky E, Bertino L, Brassington GB, Chassignet EP, Davidson F, Hurlburt HE, Kamachi M, Lee T, Martin MJ, Mei S, Tonani M (2009) GODAE systems in operation. Oceanogr
1â•… Ocean Forecasting in the 21st Century
25
Mag 22(3):80–95 (Special Issue on the Revolution of Global Ocean Forecasting—GODAE: 10 years of achievement) Ekman VW (1905) On the influence of the earth’s rotation on ocean currents. Arkiv för Matematik. Astronomi och Fysik 2(11):52 Ertel H (1942) Ein neuer hydrodynamischer Erhaltungssatz. Naturwissenschaften 30:543–544 Franklin B (1786) A letter from Dr. Benjamin Franklin, to Mr. Alphonsus le Roy, member of several academies at Paris. Containing sundry maritime observations. At sea, on board the London packet, Capt. Truxton, August 1785. Transactions of the American Philosophical Society, held at Philadelphia, for Promoting Useful Knowledge II: 294–329. Includes chart and diagrams. Held by NOAA Central Library, Silver Spring, MD Freeland H et€al (2010) Argo—a decade of progress. In: Hall J, Harrison DE & Stammer D (eds) Proceedings of OceanObs’09: sustained Ocean Observations and Information for Society, vol€2, Venice, Italy, 21–25 Sept 2009. ESA Publication WPP-306 Fu LL, Cazenave A (2001) Satellite altimetry and earth sciences. A handbook of techniques and applications. Academic, San Diego Griffies S et€al (2010) Problems and prospects in large-scale ocean circulation models. In: Hall J, Harrison DE & Stammer D (eds) Proceedings of OceanObs’09: sustained Ocean observations and information for society, vol€2, Venice, Italy, 21–25 Sept 2009. ESA Publication WPP-306 Hackett B, Comerma E, Daniel P, Ichikawa H (2009) Marine oil pollution prediction. Oceanogr Mag 22(3):168–175 (Special Issue on the Revolution of Global Ocean Forecasting—GODAE: 10 years of achievement) Hernandez F, Bertino L, Brassington G, Chassignet E, Cummings J, Davidson F, Drévillon M, Garric G, Kamachi M, Lellouche J-M, Mahdon R, Martin MJ, Ratsimandresy A, Regnier C (2009) Validation and intercomparison studies with GODAE. Oceanogr Mag 22(3):128–143 (Special Issue on the Revolution of Global Ocean Forecasting—GODAE: 10 years of achievement) € International GODAE Steering Team (2000) The Global Ocean Data Assimilation Experiment Strategic Plan, GODAE Report No. 6 IPCC First Assessment Report (1990) Scientific Assessment of Climate Change—Report of Working Group I, 1, Houghton JT, Jenkins GJ, Ephraums JJ (eds) Cambridge University Press, UK, p€365 Jacobs GA, Woodham R, Jourdan D, Braithwaite J (2009) GODAWE applications useful to Navies throughout the World. Oceanogr Mag 22(3):182–189 (Special Issue on the Revolution of Global Ocean Forecasting—GODAE: 10 years of achievement) Le Hénaff M, De Mey P, Marsaleix P (2008) Assessment of observational networks with the Representer Matrix Spectra method—application to a 3-D coastal model of the Bay of Biscay. Ocean Dyn 59(1):3–20 (Special Issue, 2007 GODAE Coastal and Shelf Seas Workshop, Liverpool, UK) Luong B, Blum J, Verron J (1998) A variational method for the resolution of a data assimilation problem in oceanography. Inverse Probl 14:979–997 Malone T, DiGiacomo P, Muelbert J, Parslow J, Sweijd N, Yanagi T, Yap H, Blanke B (2010) Building a global system of systems for the coastal ocean. In: Hall J, Harrison DE, Stammer D (eds) Proceedings of OceanObs’09: sustained ocean observations and information for society, vol€2, Venice, Italy, 21–25 Sept 2009. ESA Publication WPP-306 Maury MF (1855) Physical geography of the sea. Harper & Brothers, New York Murray JS, Hjort J (1912) The depths of the ocean: a general account of the modern science of oceanography based largely on the scientific researches of The Norwegian Steamer Michael Sars in The North Atlantic. Macmillan, London Pascual A, Faugere Y, Larnicol G, Le Traon P-Y (2006) Improved description of the ocean mesoscale by combining four satellite altimeters. Geophys Res Lett 33:L02611. doi:10.1029/2005GL024633 Rennell J, Purdy J (1832) An investigation of the currents of the Atlantic Ocean, and of those which prevail between the Indian Ocean and the Atlantic. In: Purdy J (ed). Nabu Press, London
26
A. Schiller
Robert C, Blayo E, Verron J (2006) Comparison of reduced-order sequential, variational and hybrid data assimilation methods in the context of a tropical Pacific ocean model. Ocean Dyn 56(5–6):624–633 Sakov P, Oke PR (2008) Objective array design: application to the tropical Indian Ocean. J Atmos Ocean Technol 25:794–807 Semtner AJ, Chervin RM (1992) Ocean general circulation from a global eddy resolving model. J Geophys Res 97:5493–5550 Shapiro M, Shukla J, Hoskins B, Church J, Trenberth K, Béland M, Brasseur G, Wallace M, McBean G, Caughey J, Rogers D, Brunet G, Barrie L, Henderson-Sellers A, Burridge D, Nakazawa T, Miller M, Bougeault P, Anthes R, Toth Z, Palmer T (2008) The socioeconomic and environmental benefits of a revolution in weather, climate and earth-system prediction: a weather, climate and earth-system prediction project for the 21st century. Group on earth observations, Tudor Rose, Geneva, pp€136–138 Siedler G, Church J, Gould J (eds) (2001) Ocean circulation and climate: observing and modelling the Global Ocean. Academic Press, San Diego Smith N, Lefebvre M (1997) Monitoring the oceans in the 2000s: an integrated approach The Global Ocean Data Assimilation Experiment (GODAE). International Symposium, Biarritz Sverdrup HU, Johnson MW, Fleming RH (1942) The oceans: their physics, chemistry, and general biology. Prentice-Hall, Englewood Cliffs, p€1087 Sverdrup HU (1947) Wind-driven currents in a Baroclinic Ocean; with application to the equatorial currents of the Eastern Pacific. Proc Natl Acad Sci U S A 33:318–326 Thompson Sir Wyville, Sir John Murray, George S. Nares, and Frank Tourle Thompson (1880– 1895) Report on the scientific results of the voyage of H.M.S. Challenger during the years 1873–76 under the command of Captain George S. Nares, R.N., F.R.S. and the late Captain Frank Tourle Thomson, R.N./prepared under the superintendence of the late Sir C. Wyville Thompson, and now of John Murray; published by order of Her majesty’s Government. H.M. Stationery Office Wilkin JL, Arango HG, Haidvogel DB, Lichtenwalner CS, Glenn SM, Hedstrom KS (2005) A regional ocean modeling system for the long-term ecosystem observatory. J Geophys Res 110, C06S91. doi:10.1029/2003JC002218 Wilson S, et€al (2010) Ocean surface topography Constellation: the next 15 years in satellite altimetry. In: Hall J, Harrison DE, Stammer D (eds) Proceedings of OceanObs’09: sustained ocean observations and information for society, vol€ 2, Venice, 21–25 Sept 2009. ESA Publication WPP-306 Wunsch (1978) The North Atlantic general circulation west of 50°W determined by inverse methods. Rev GeophysSpace Phys 6(4):583–620 Zebiak SE, Cane MA (1987) A model of El Nino—southern oscillation. Mon Weath Rev 115:2262– 2278
Part II
Oceanographic Observing System
Chapter 2
Satellites and Operational Oceanography Pierre-Yves Le Traon
Abstract╇ The chapter starts with an overview of satellite oceanography, its role and use for operational oceanography. Main principles of satellite oceanography techniques are then summarized. We then describe key techniques of radar altimetry, sea surface temperature, ocean colour satellite measurements. This includes measurement principles, data processing issues and the use of these data for operational oceanography. SAR, scatterometry, sea ice and sea surface salinity measurements are also briefly described. Main prospects are given in the conclusion.
2.1â•…Introduction There are very strong links between satellite oceanography and operational oceanography. The development of operational oceanography has been mainly driven by the development of satellite oceanography capabilities. The ability to observe the global ocean in near real time at high space and time resolution is indeed a prerequisite to the development of global operational oceanography and its applications. The first ocean parameter to be globally monitored from space was the sea surface temperature on board meteorological satellites in the late 1970s. It is, however, the advent of satellite altimetry in the late 1980s that led the development of ocean data assimilation and global operational oceanography. In addition to providing all weather observations, sea level from satellite altimetry is an integral of the ocean interior and provides a strong constraint on the 4D ocean state estimation. The satellite altimetry community was also keen to develop further the use of altimetry and this required an integrated approach merging satellite and in-situ observations with models. GODAE demonstration was thus phased with the Jason-1 and ENVISAT altimeter missions (Smith and Lefebvre 1997). Satellite oceanography is now a major component of operational oceanography. Data are usually assimilated in ocean models but they can also be used directly for P.-Y. Le Traon () IFREMER, Centre de Brest, Technopôle Brest Iroise BP70, 29280 Plouzané, France e-mail:
[email protected] A. Schiller, G. B. Brassington (eds.), Operational Oceanography in the 21st Century, DOI 10.1007/978-94-007-0332-2_2, ©Â€Springer Science+Business Media B.V. 2011
29
30
P.-Y. Le Traon
applications. An overview of satellite oceanography will be given here focusing on the most relevant issues for operational oceanography. The chapter is organized a follows. Section€2.2 provides an overview of satellite oceanography, its role and use for operational oceanography. Main operational oceanography requirements are summarized. The complementary role of in-situ observations is also emphasized. Main principles of satellite oceanography and general data processing issues are described in Sect.€2.3. We then detail key techniques of radar altimetry and gravimetry, sea surface temperature, ocean colour satellite measurements in Sects.€2.4, 2.5 and 2.6. This includes measurement principles, data processing issues and the use of these data for operational oceanography. SAR, scatterometry, sea ice and the new sea surface salinity measurements are briefly described in Sect.€2.7. Main prospects are given in the conclusion.
2.2╅Role of Satellites for Operational Oceanography 2.2.1 T he Global Ocean Observing System and Operational Oceanography Operational oceanography critically depends on the near real time availability of high quality in-situ and remote sensing data with a sufficiently dense space and time sampling. The quantity, quality and availability of data sets directly impact the quality of ocean analyses and forecasts and associated services. Observations are required to constrain ocean models through data assimilation and also to validate them. Products derived from the data themselves can also be directly used for applications (e.g. in the case of a parameter observed from space at high resolution). This requires an adequate and sustained global ocean observing system. Climate and operational oceanography applications share the same backbone system (GOOS, GCOS, JCOMM). Operational oceanography has, however, specific requirements for high resolution measurements. Operational oceanography requirements have been presented in the GODAE strategic plan and in Le Traon et€ al. (2001). They have been refined and detailed in Clark and Wilson (2009) and Oke et€al. (2009).
2.2.2 The Unique Contribution of Satellite Observations Satellites provide long-term, continuous, global, high space and time resolution data for key ocean parameters: sea level and ocean circulation, sea surface temperature (SST), ocean colour, sea ice, waves and winds. These are the core variables observations required to constrain global, regional and coastal ocean monitoring
2â•… Satellites and Operational Oceanography
31
and forecasting systems. They are also needed to validate them. Only satellite measurements can, in particular, provide observations at high space and time resolution to partly resolve the mesoscale variability and coastal variability. Satellite data can also be directly used for applications (e.g. SAR for sea ice and oil pollution monitoring, ocean colour for water quality monitoring). Sea surface salinity is a new and important parameter that could be operationally monitored from space; the demonstration is underway with the European Space Agency SMOS mission (and later on with the NASA/CONAE Aquarius mission).
2.2.3 Main Requirements The main requirement for operational oceanography is to have a long-term, continuous and near real time access to the core operational satellite observations of sea level, SST, ocean colour, sea ice, wave and winds. For a given parameter, this generally requires several satellites flying simultaneously to get sufficient space and time resolution. The main requirements can be summarized as follows (e.g. Le Traon et€al. 2006; Clark and Wilson 2009): • In addition to meteorological satellites, a high precision (AATSR-class) SST satellite is needed to give the highest absolute SST accuracy. A microwave mission is also needed to provide an all weather global coverage. • At least three or four altimeters are required to observe the mesoscale circulation. This is also useful for significant wave height measurements. A long-term series of a high accuracy altimeter system (Jason satellites) is needed to serve a reference for the other missions and for the monitoring of climate signals. • Ocean colour is increasingly important, in particular, in coastal areas. At least two satellites are required. • Two scatterometers are required to globally monitor at high spatial resolution the wind field. • Two SAR satellites are required for waves, sea-ice characteristics and oil slick monitoring. These minimum requirements have been only partly met over the past ten years. Long-term continuity and transition from research to operational mode remains a major challenge (e.g. Clark and Wilson 2009). Specific requirements for altimetry, SST and ocean colour are discussed in the following sections.
2.2.4 Role of In-Situ Data Satellite observations need to be complemented by in-situ observations. First, in-situ data are needed to calibrate satellite observations. Most algorithms used
32
P.-Y. Le Traon
to transform satellite observations (e.g brightness temperatures) into geophysical quantities are partly based on in-situ/satellite match up data bases. In-situ data are then used to validate satellite observations and to monitor the long term stability of satellite observations. The stability of the different altimeter missions is, for example, commonly assessed by comparing the altimeter sea surface height measurements with those from tide gauges (Mitchum 2000). Other examples includes the validation of altimeter velocity products with drifter data (e.g. Pascual et€al. 2009), the systematic validation of satellite SST with in-situ SST from drifting buoys and the use of dedicated ship mounted radiometers to quantify the accuracy of satellite SST (Donlon et€al. 2008). The comparison of in-situ and satellite data can also provide useful indication on the quality of in-situ data (e.g. Guinehut et€al. 2008). The comparison of in-situ and satellite data is also useful to check the consistency between the different data sets before they are assimilated in an ocean model (e.g. Guinehut et€al. 2006). In-situ data are also (and mainly) mandatory to complement satellite observations and to provide measurements of the ocean interior. Only the joint use of high resolution satellite data with precise (but sparse) in-situ observations of the ocean interior has the potential to provide a high resolution description and forecast of the ocean state.
2.2.5 Data Processing Issues Satellite data processing includes different steps: level 0 and level 1 (from telemetry to calibrated sensor measurements), level 2 (from sensor measurements to geophysical variables), level 3 (space/time composites of level 2 data) and level 4 (merging of different sensors, data assimilation). Processing from level 0 to level 2 is generally carried out as part of the satellite ground segments. Assembly of level 2 data from different sensors, intercalibration of level 2 products, and higher level data processing is usually done by specific data processing centers or thematic assembly centers. The role of these data processing centers is to provide modelling and data assimilation centers with the real time and delayed mode data sets required for validation and data assimilation. This also includes uncertainty estimates that are critical to an effective use of data in modelling and data assimilation systems. Links with data assimilation centers are needed, in particular, to organize feedback on the quality control performed at the level of data assimilation centers (e.g. comparing an observation with a model forecast), on the impact of data sets and data products in the assimilation systems and on new or future requirements. High level data products (level 3 and 4) are also needed for applications (e.g. a merged altimeter surface current product for marine safety or offshore applications) and can be used to validate data assimilation systems (e.g. statistical versus dynamical interpolation) and complement products derived through modelling and data as-
2â•… Satellites and Operational Oceanography
33
similation systems. It is important, however, to be fully aware of limitations of high level satellite products (e.g. gridded SST or sea level data sets) when using them.
2.2.6 Use of Satellite Data for Assimilation into Ocean Models This is discussed at length in other chapters. Three important issues are emphasized here: 1. There can be large differences in data quality between real time and delayed mode (reprocessed) data sets. Depending on applications, trade-offs between time delay and accuracy often need to be considered. 2. Error characterisation is mandatory for data assimilation and a proper characterisation of error covariance can be quite complex for satellite observations. Data error covariance should always be tested and checked as part of the data assimilation systems. 3. It is much better in theory and for advanced assimilation schemes to use raw data (level 2 or in some cases level 1 when the model can provide data needed for level 1 processing). The data error structure is generally more easily defined. The model and the assimilation scheme should also do a better high level processing (e.g. a model forecast should provide a better background than climatology or persistence). However, in practice, this is not always true. Some data high level processing (e.g. correcting biases or large scale errors, intercalibration) is often needed as it cannot be easily done within the assimilation systems.
2.3╅Overview of Satellite Oceanography Techniques 2.3.1 Passive/Active Techniques and Choice of Frequencies There are two main types of satellite techniques to observe the ocean1. Passive techniques measure the natural radiation emitted from the sea or from reflected solar radiation. Active or radar techniques send a signal and measure the signal received after its reflection at the sea surface. In both cases, the propagation of the signal through the atmosphere, the emission from the atmosphere itself must be taken into account to isolate the sea surface signal. The intensity and frequency distribution of the radiation that is emitted or reflected from the ocean surface allows the inference of its properties. The polarization of the radiation is also often used in microwave remote sensing. 1╇ Gravimetry satellites (e.g. GRACE, GOCE) which measure the earth gravity field and its variations do not enter into these two categories.
34
P.-Y. Le Traon
Satellite systems operate at different frequencies depending on the signal to be derived. Visible (400–700€nm) and infra-red (0.7–20€μm) frequencies are used for ocean colour and SST measurements. Passive (radiometry) microwave systems (1– 30€cm) are used for SST in cloud situations, wind, sea ice and sea surface salinity retrievals. Radars operate in the microwave bands and provide measurements of sea surface height, wind speed and direction, wave spectra, sea ice cover and types and surface roughness. Radar pulses are emitted obliquely (15°–60°) (SAR, scatterometer) or vertically (altimetry). The choice of frequencies is limited by other usages (e.g. radio, cellular phones, military and civilian radars, satellite communications). Those are particularly important at microwave frequencies in the range 1–10€GHz which puts strong pressures on the frequencies used for earth remote sensing. The atmosphere also greatly affects the transmission of radiation between the ocean surface and the satellite sensors. The presence of fixed concentrations of atmospheric gases (e.g. O2, CO2, O3) and of water vapor means that only a limited number of windows exist in the visible, infra-red and microwave for ocean remote sensing. Even at these frequencies, the propagation effects through the atmosphere must be taken into account and corrected for. Propagation effects through the ionosphere must also be taken into account. Clouds are a strong limitation of visible and infrared measurements. There are also technological constraints for the choice of frequencies. The resolution of a given sensor is generally related to the ratio between the observed wavelength (↜λ) and the antenna diameter (D). For antenna diameters of a few meters, typical resolution around 1€ GHz (wavelength of 30€ cm) is about 100€ km while at 30€GHz (wavelength of 1€cm), resolution is about 10€km. Radar altimeters use pulse limited techniques (that are much less sensitive to mispointing errors). Their footprint size is related to the pulse duration and is much smaller than for a beam limited sensor. Synthetic Aperture Radar uses the motion of the satellite to generate very long antenna (e.g. 20€km for ASAR) and thus to provide very high resolution measurements (up to a few meters).
2.3.2 Satellite Orbits and Measurement Characteristics Orbits for ocean satellites are geostationary, polar or inclined orbits. A geostationary orbit is one in which the satellite is always in the same position with respect to the rotating Earth. The satellite orbits at an elevation of approximately 36,000€km because that produces an orbital period equal to the period of rotation of the Earth. By orbiting at the same rate, in the same direction as Earth, the satellite appears stationary. Geostationary satellites provide a large field of view (up to 120°) at very high frequency enabling coverage of weather events. Because of the high altitude, spatial resolution is of a few km while it is of 1€km or less for polar orbiting satellites. Because a geostationary orbit must be in the same plane as the Earth’s rotation, that is the equatorial plane, it provides distorted images of the polar regions. Five or six geostationary meteorological satellites can provide a global coverage of the earth (for latitudes below 60°).
2â•… Satellites and Operational Oceanography
35
Polar-orbiting satellites provide a more global view of Earth by passing from pole to pole, observing a different portion of the Earth with each orbit due to the Earth’s own rotation. Orbiting at an altitude of 700–800€km these satellites have an orbital period of approximately 90€ min. These satellites usually operate in a sun-synchronous orbit. The satellite passes the equator and any given latitude at the same local solar time each day. Inclined orbits have an inclination between 0° (equatorial orbit) and 90° (polar orbit). They are used, in particular, to observe tropical regions (e.g. TMI on TRMM mission). High accuracy altimeter satellites such as TOPEX/Poseidon and Jason use higher altitude and non synchronous orbits to reduce atmospheric drag and (mainly) to avoid aliasing of the main tidal signals. Depending on instrument types (along-track, imaging or swath), frequencies and antennas (see above), the sampling pattern of a given satellite will be different. In addition, in the visible and infrared frequencies, cloud cover can strongly reduce the effective sampling.
2.3.3 Radiation Laws and Emissivity 2.3.3.1â•…Radiation from a Blackbody Planck’s law describes the rate of energy emitted by a blackbody as a function of frequency or wavelength. A blackbody absorbs all the radiation it receives and emits radiation at a maximum rate for its given temperature. Planck’s law gives the intensity of radiation Lλ emitted by unit surface area into a fixed direction (solid angle) from the blackbody as a function of wavelength (or frequency). The Planck Law can be expressed through the following equation: Lλ = 2hc2 /λ5 [exp(hc/λkT ) − 1]
where T is temperature, c the speed of light (2.99â•›·â•›10−8€m€s−1), h the Planck’s constant (6.63â•›·â•›10−34€ Jâ•›⋅â•›s), k the Boltzmann’s constant (1.38â•›·â•›10−23 J€ K−1) and Lλ the spectral radiance per unit of wavelength and solid angle in W€m−3€sr−1. The Planck law gives a distribution that peaks at a certain wavelength; the peak shifts to shorter wavelengths for higher temperatures. The Wien displacement law and the Stefan-Boltzmann law are two other useful radiation laws that can be derived from the Planck law. The Wien law gives the wavelength of the peak of the radiation distribution (↜λmaxâ•›= 3€╛·â•›107/T) while the Stefan-Boltzmann law gives the total energy E being emitted at all wavelengths by the blackbody (Eâ•›=╛╛·â•›T4). Thus, the Wien law explains the shift of the peak to shorter wavelengths as the temperature increases, while the Stefan-Boltzmann law explains the growth in the height of the curve as the temperature increases. Notice that this growth is very abrupt, since it varies as the fourth power of the temperature. The Rayleigh-Jeans approximation (↜Lλâ•›=â•›2kcT/λ4) holds for wavelengths much greater than the wavelength of the peak in the black body radiation form. This approximation is valid over the microwave band.
36
P.-Y. Le Traon
2.3.3.2â•…Graybodies and Emissivity Most bodies radiate less efficiently than a blackbody. The emissivity e is defined as the ratio of graybody radiance to the blackbody. It has a non dimensional unit and is comprised between 0 and 1. The emissivity generally depends on wavelength (↜λ) and polarization and has a directional dependence. e can be considered as a physical surface property and is a key quantity for ocean remote sensing. A blackbody absorbs all the energy it receives. A graybody absorbs only part of it and the remaining part is reflected and/or transmitted. The absorptivity is equal to the emissivity as a surface in equilibrium must absorb and emit energy at the same rate (Kirchoff’s law). Similarly the reflectivity is equal to 1â•›−â•›e. The brightness temperature (BT) is defined as BTâ•›=â•›eâ•›·â•›T where T is the (physical) temperature. In the microwave band, it is proportional to the radiation Lλ. 2.3.3.3â•…Retrieval of Geophysical Parameters for Microwave Radiometers The brightness temperature is an integrated measurement that includes all surface and atmosphere emitted power. Depending on frequency, it is more sensitive to a given parameter. Physical retrieval algorithms for geophysical parameters, such as the sea surface temperature, sea surface wind speed, sea ice or sea surface salinity are derived from a radiative transfer model (RTM), which computes the brightness temperatures that are measured by the satellite as a function of these variables. The RTM is based on a model for the sea surface emissivity and a model of microwave absorption in the Earth’s atmosphere. The ocean sea surface emissivity (or reflectivity see above) depends on the dielectric constant ε (which is a function of frequency, water temperature and salinity), small scale sea surface roughness, foam as well as viewing geometry and polarization. The retrieval of a given parameter is possible through the inversion of a set of brightness temperatures measured at different frequencies and/ or at different incidence angles. Inversion methods minimize the difference between measured and simulated (through a RTM) brightness temperatures. Statistical or empirical inversions are also often used given uncertainties in RTMs. They use a regression formalism (e.g. parametric, neural network) to find the best relation between brightness temperatures and the geophysical parameter to be retrieved.
2.4╅Altimetry 2.4.1 Overview Satellite altimetry is the most essential observing system required for global operational oceanography. It provides global, real time, all-weather sea level measurements (SSH) with high space and time resolution. Sea level is directly related to ocean circulation through the geostrophic approximation (see Sect.€2.4.5). Sea level is also an integral of the ocean interior and is a strong constraint for inferring the
2â•… Satellites and Operational Oceanography
37
4D ocean circulation through data assimilation. Altimeters also measure significant wave height, which is essential for operational wave forecasting. High resolution from multiple altimeters is required to adequately represent ocean eddies and associated currents (the “ocean weather”) in models. Only altimetry can constrain the 4D mesoscale circulation in ocean models which is required for most operational oceanography applications.
2.4.2 Measurement Principles An altimeter is active radar that sends a microwave pulse towards the ocean surface. Precise clock on board measures the return time of the pulse from which the distance or range (d) between the satellite and the sea surface is derived (dâ•›=â•›t/2c). The range precision is of a few centimeters for a distance of 800–1,300€km. The altimeter also measures the backscatter power (related to surface roughness and wind) and significant wave height. An altimeter mission generally includes a bifrequency altimeter radar (usually in Ku and C or S Band) (for ionospheric corrections), a microwave radiometer (for water vapor correction) and a tracking system for precise orbit determination (Laser, GPS, Doris) that provides the orbit altitude relative to a given earth ellipsoid. Altimeter missions provide along-track measurements every 7€km along repetitive tracks (e.g. every 10 days for the TOPEX/Poseidon and Jason series and 35 days for ERS and ENVISAT). The distance between tracks is inversely proportional to the repeat time period (e.g. about 315€km at the equator for TOPEX/Poseidon and 90€km for ERS/ENVISAT). The main measurement for an altimeter radar is the sea surface height (SSH) relative to a given earth ellipsoid. The SSH is derived as the difference between the orbit altitude and the range measurement. SSH precision depends on orbit and range errors. Altimeter range measurements are affected by a large number of errors (propagation effects in the troposphere and ionosphere, electromagnetic bias, errors due to inaccurate ocean and terrestrial tide models, inverse barometer effect, residual geoid errors). Some of these errors can be corrected with dedicated instrumentation (e.g. dual frequency altimeter, radiometer). For a comprehensive description of altimeter measurement principles, the reader is referred to Chelton et€al. (2001).
2.4.3 Geoid and Repeat-Track Analysis The sea surface height SSH(x,t) measured by altimetry can be described by: SSH(x, t) = N(x) + η(x, t) + ε(x, t)
N is the geoid, the dynamic topography and are measurement errors. The quantity of interest for oceanographer is the dynamic topography (see next section).
38
P.-Y. Le Traon
Present geoids are not generally accurate enough to estimate globally the absolute dynamic topography except at long wavelengths. The variable part of the dynamic topography ′(↜╛−â•›) (or SLA for sea level anomaly) is, however, easily extracted using the so-called repeat track method. For a given track, ′ is obtained by removing the mean profile over several cycles, which contains the geoid N and the mean dynamic topography : SLA(x, t) = SSH(x, t) − < SSH(x) > t = η(x, t) − < η(x) > t + ε (x, t)
To get the absolute signal, one has thus to use a climatology or to use existing geoids together with altimeter Mean Sea Surface (MSS) (or both). One can also rely on a model mean. Gravimetric missions (CHAMP, GRACE) are now providing much more accurate geoids, GOCE should almost “solve” the problem. Even with GOCE, however, repeat-track analysis will still be needed because of the small scales of geoid (below 50–100€km) will not be precisely known. GOCE will be used with an altimetric MSS to derive t that can then be added to ′.
2.4.4 High Level Data Processing Issues and Products The SSALTO/DUACS system is the main multi-mission altimeter data center used today for operational oceanography. It aims to provide directly usable, high quality near real time and delayed mode (for reanalyses and research users) altimeter products to the main operational oceanography and climate centers in Europe and worldwide. Main processing steps are product homogenization, data editing, orbit error correction, reduction of long wavelength errors, production of along track and maps of sea level anomalies. Major progress has been made in higher level processing issues such as orbit error reduction (e.g. Le Traon and Ogor 1998), intercalibration and merging of altimeter missions (e.g. Le Traon et€al. 1998; Ducet et€al. 2000; Pascual et€al. 2006). The SSALTO/DUACS weekly production moved to a daily production in 2007 to improve timeliness of data sets and products. A new real time product was also developed for specific real time mesoscale applications. The mean dynamic topography (MDT) is an essential reference surface for altimetry. Added to the sea level anomalies, it provides the absolute sea level and ocean circulation (see previous section). After a preliminary MDT computed in 2003, a new MDT, called RIO-05, was computed in 2005. It is based on the combination of GRACE data, drifting buoy velocities, in-situ T,S profiles and altimeter measurements. The MDT was tested and is now used by several GODAE modelling and forecasting centers. It has a positive impact on the ocean analysis quality and forecast skill. An updated version was recently delivered (CNES-CLS09). Major improvement is expected soon with the use of data from the GOCE mission.
2â•… Satellites and Operational Oceanography
39
2.4.5 Sea Level Measurement Content Satellite altimetry provides measurements of the dynamic topography (i.e. sea level relative to the geoid). Assuming geostrophy and hydrostatic balance, one has:
1 ∂P fv = ρ0 ∂x 1 ∂P −f u = ρ0 ∂y
(2.1) f = 2 sin θ
(2.2)
∂p = −ρg ∂z
(2.3)
with u, v zonal and meridional currents, P pressure and f the Coriolis parameter. At the surface Pâ•›=╛╛·â•›gâ•›·â•› (↜╛=â•›sea surface topography relative to the geoid), thus there is a direct relationship between the dynamic topography and the surface (geostrophic) current: ∂η fv = g ∂x (2.4) ∂η −f u = g ∂y Taking the derivative of (2.3), one gets the thermal wind equation. It means that density horizontal variations are associated with vertical shear (baroclinic motions):
f
∂v −g ∂ρ = ∂z ρ0 ∂x
(2.5)
The integration of (2.5) from z0 to z1 yields: g z1 1 ∂ρ v(z) = v(z0 ) − dz f z0 ρ0 ∂x
g ∂ηs or v(z) = v(z0 ) + with ηs (z0 , z1 ) = − f ∂x
(2.6)
z1
z0
ρ dz ρ0
s is the steric height. s is generally defined as s (bottom, surface). At the surface, one has: g ∂ηs g ∂η v(z0 ) + f ∂x = f ∂x Pz0 ⇒ η = ηs + ρ0 g 1 ∂Pz0 with v(z0 ) = f ρ0 ∂x
(2.7)
(2.8)
P.-Y. Le Traon
40
The dynamic topography (measured by altimetry) is thus the sum of a steric height term (integral of density anomalies which is generally referred to the baroclinic component) and a bottom pressure term (barotropic component). Sea level is thus more than a « surface » measurement. It corresponds to a signal over the full depth of the ocean and provides a strong constraint for inferring (together with in-situ measurements) the 4D ocean structure through data assimilation.
2.4.6 Operational Oceanography Requirements Le Traon et€al. (2006) have defined the main priorities for altimeter missions in the context of the European GMES (Global Monitoring for Environment and Security) Marine Core Service. Their Tables€2.1 and 2.2 give the requirements for different applications of altimetry and characteristics of altimeter missions. The main operational oceanography requirements for satellite altimetry can be summarized as follows: 1. Need to maintain a long time series of a high accuracy altimeter system (Jason series) to serve a reference mission and for climate applications. It requires one class A altimeter with an overlap between successive missions of at least 6 months. 2. The main requirement for medium to high resolution altimetry would be to fly three class B altimeters in addition to the Jason series (class A). Most operational oceanography applications (e.g. marine security, pollution monitoring) require high resolution surface currents that cannot be adequately reproduced without a high resolution altimeter system. Recent studies (e.g. Pascual et€al. 2006) show Table 2.1↜渀 User requirements for different applications of altimetry Application area Accuracya Spatial resolution (cm) (km) 1.╇Climate applications and 1 300–500 reference mission 2.╇Ocean nowcasting/forecasting for 3 50–100 mesoscale applications 3 10 3.╇ Coastal/local a For the given resolution b Limited by feasibility Table 2.2↜渀 Altimeter mission characteristics Class Orbit Mission characteristics A
Non-sun synchronous
B
Polar
High accuracy for climate applications and to reference other missions Medium-class accuracy
Revisit time (days) 10–20
Priority
7–15
High
1
Lowb
High
Revisit interval (days) 10–20
Track separation at the equator (km) 150–300
20–35
80–150
2â•… Satellites and Operational Oceanography
41
that, at least three, but preferably four, altimeter missions are needed for monitoring the mesoscale circulation. This is particularly needed for real time nowcasting and forecasting. Pascual et€al. (2009) showed that four altimeters in real time provide similar results as two altimeters in delayed mode. Such a scenario would also provide an improved operational reliability. Moreover, it would enhance the spatial and temporal sampling for monitoring and forecasting significant wave height. In parallel, there is a need to develop and test innovative instrumentation (e.g. wide swath altimetry with the NASA SWOT mission) to better answer existing and future operational oceanography requirements for high to very high resolution (e.g. mesoscale/submesoscale and coastal dynamics). There is also a need to improve nadir altimetry technology (resolution, noise) and to develop smaller and cheaper instruments that could be embarked on a constellation of small satellites. The use of the Ka band (35€Ghz) allows, in particular, a major reduction in the size and weight of the altimeter. It will be tested for the first time with the CNES/ISRO SARAL satellite scheduled for launch in late 2011.
2.5╅Sea Surface Temperature 2.5.1 S ea Surface Temperature Measurements and Operational Oceanography Sea surface temperature (SST) is a key variable for operational oceanography and for assimilation into ocean dynamical models. SST is strongly related to air-sea interaction processes and provides a means to correct for errors in forcing fields (heat fluxes, wind). It also characterizes the mesoscale variability of the upper ocean (eddies, frontal structures) at very high resolution (a few km). SST data are often directly used for operational oceanography applications. They provide useful indices (e.g. climate changes, upwelling, thresholds). SST data can also be used to derive high resolution velocity fields (e.g. Bowen et€al. 2002). Accurate, stable, well resolved maps of SST are essential for climate monitoring and climate change detection. They are also central for Numerical Weather Prediction for which the role of high resolution SST measurements has been recently evidenced (e.g. Chelton 2005).
2.5.2 Measurement Principles Infrared radiometers operate at wavebands around 3.7, 10.5 and 11.5€ μm where the atmosphere is almost transparent. The brightness temperature measured from infrared radiometers differs from the actual temperature of the observed surface because of non-unit emissivity and the effect of the atmosphere. Emissivity at IR
42
P.-Y. Le Traon
frequencies is between 0.98 and 0.99 (close to a black body). Atmospheric correction is based on multispectral approach, when the differences between brightness temperatures measured at different wavelengths are used to estimate the contribution of the atmosphere to the signal. At 10€μm, the solar irradiance reaching the top of the atmosphere is about 1/300 of the sea surface emittance. At 3.7€μm, the incoming solar irradiance is the same order as the surface emittance. As a result, this wavelength can be used during nighttime only. Different algorithms are thus used for nighttime and daytime. There is no IR way of measuring SST below cloud. The first priority is thus to detect cloud through a variety of methods. For cloud detection, the thermal and near-infrared waveband thresholds are used, as well as different spatial coherency tests. Consequences of poor cloud detection are low biases in SST climatic averages and “false hits” of cloud that can hide frontal and other dynamical structures. Geostationary infra-red sensors can see whenever the cloud breaks. Microwave sensors operate at several frequencies. Retrieval of SST is done at 7 and/or 11€GHz. Higher frequency channels (19–37€GHz) are used to precisely estimate the attenuation due to oxygen, water vapor, and clouds. The polarization ratio (horizontal versus vertical) of the measurements is used to correct for sea surface roughness effects. The great advantage of microwave measurements compared to infra-red ones is that SST can be retrieved even through non-precipitating clouds, which is very beneficial in terms of geographical coverage.
2.5.3 SST Infra-Red and Microwave Sensors Infra-red radiometers such as the Advanced Very High Resolution Radiometer (AVHRR) on board operational meteorological polar orbiting satellites offer a good horizontal resolution (1€km) and potentially a global coverage, with the important exception of cloudy areas. However, their accuracy (0.4–0.5€ K derived from the difference between collocated satellite and buoy measurements) is limited by the radiometric quality of the AVHRR instrument and the correction of atmospheric effects. Geostationary satellites (e.g. GOES and MSG series) are carrying radiometers with similar infrared window channels as the AVHRR instrument. Their horizontal resolution is coarser (3–5€km), but their great contribution comes from their high temporal sampling. Pre-operational demonstrators for advanced measurement of SST suitable for climate studies include the Along Track Scanning Radiometer ((A) ATSR) series of instruments that have improved on board calibration, and make use of dual views at nadir and 55° incidence angle. The along track scanning measurement provides an improved atmospheric correction leading to an accuracy of better than 0.2€K (O’Carroll et€al. 2008). The main drawback of these instruments is their limited coverage, due to a much narrower swath than the AVHRR instruments. Several microwave radiometers have also been developed and flown over the last 10 years (e.g. AMSR, TMI). The horizontal resolution of these products is around 25€km and their accuracy around 0.6–0.7€K.
2â•… Satellites and Operational Oceanography
43
2.5.4 Key Developments in SST Data Processing During the past ten years, a concerted effort to understand satellite and in situ SST observations has taken place leading to a revolution in the way we approach the provision of SST data to the user community. GODAE, recognizing the importance of high resolution SST data sets for ocean forecasting, initiated the GODAE High Resolution SST Pilot Project (GHRSST-PP) to capitalize on these developments and develop a set of dedicated products and services. There have been key developments in data processing of SST data sets over the last 10 years. As a result, new or improved products are now available. A full description of the GHRSST-PP is provided in Donlon et€ al. (2009). Data processing issues are summarized in Le Traon et€al. (2009). A satellite measures the so-called skin temperature, i.e. at a depth from a few tens of microns (infra-red) up to a few mm only (microwave). Diurnal warming changes the SST over a layer of 1–10€m. The effect can be particularly large in regions of low wind speed and high solar radiation. GHRSST has defined the foundation SST as the temperature of the water column free of diurnal temperature variability. A key issue in SST data processing is to correct satellite SST measurements for skin and diurnal warming effects to provide precise estimations of the foundation SST. Night and day SST data from different satellites can then be merged through an optimal interpolation or a data assimilation system. Several new analyzed high resolution SST products have been produced, in particular, in the framework of GHRSST-PP. These high resolution data sets are estimated by optimal interpolation methods merging SST satellite measurements from both infrared and microwave sensors. The pre-processing consists mainly in a screening and quality control of the retrieved observations from each single datasets and in constructing a coherent merged multi-sensor set of the most relevant and accurate observations (level 3). The merging of these observations requires a method for bias estimate and correction (relative to a chosen reference, currently AATSR). The gap free SST foundation field is finally computed from the merged set of selected observations using an objective analysis method. The guess is either climatology or a previous map.
2.5.5 Operational Oceanography Requirements Table€2.3 from Le Traon et€al. (2006) summarises weather, climate and operational oceanography requirements for sea surface temperature. In order to meet the key requirements for SST no single sensor is adequate. To remedy this, GHRSST-PP has established an internationally accepted approach to blending SST data from different sources that complement each other (see previous section). For this to work effectively, there must be an assemblage of four distinct types of satellite SST missions in place at any time, as defined in Table€2.4 (from Le Traon et€al. 2006).
P.-Y. Le Traon
44 Table 2.3↜渀 User requirements for SST provision Application area Temperature accuracy (K) 0.2–0.5 1.╇ Weather prediction 2.╇ Climate monitoring 0.1 3.╇ Ocean forecasting 0.2
Spatial resolution (km) 10–50 20–50 1–10
Revisit time
Priority
6–12€h 8€day 6–12€h
High High High
Table 2.4↜渀 Minimum assemblage of missions required to meet the need for operational SST SST mission type Radiometer Nadir Swath width Coverage/ wavebands resolution revisit 3 thermal IR ~2,500€km Day and night ~1€km A.╇Two polar orbiting (3.7, 11, meteorological satellites global 12€μm), 1 with infra-red radiometers. coverage Generates the basic global by each near-IR, 1 Vis coverage satellite ~1€km ~500€km Earth coverB.╇Polar orbiting dual-view 3 thermal IR age in ~4 (3.7, 11, radiometer. SST accuracy days 12€μm), 1 approaching 0.1€K, used as near-IR, 1 reference standard for other Vis, each types with dual view Requires chanC.╇Polar orbiting microwave ~1,500€km ~50€km Earth covernels at ~7 and radiometer optimised for (25€km age in 2 ~11€GHz SST retrieval. Coarse resodays pixels) lution coverage of cloudy regions Sample Earth disk 2–4€km D.╇ Infra-red radiometers on 3 thermal IR interval from (3.7, 11, geostationary platforms. τf , Po = min (Pb , Pd )
(4.7)
where pb is the composite background error probability, pd is the composite dataderived error probability, pg and pr are the global and regional forecast background error probabilities, pc and px are the climate and cross validation error probabilities, τf is the forecast error threshold probability, and po is the overall probability the observation contains a random error. The forecast error probability threshold for the system is typically set to 0.99 (3 standard deviations). The algorithm first determines if the observation is consistent with the model background fields by taking the minimum error probability of the global and re-
110
J. A. Cummings
gional forecasts. If the minimum background error probability is less than the prescribed forecast error tolerance limit, then the algorithm returns it as the overall probability of error for the observation. However, if the minimum model background error probability exceeds the forecast error threshold, then it is compared against the data-derived error defined as the minimum of the cross validation and climatology error probabilities. The overall observation error probability is returned as the minimum of the composite background and composite data-derived errors. In this way, cross validation and climate backgrounds determine data quality only if the observation is not consistent with the forecast. Experience has shown that requiring observations to always be consistent with climate backgrounds results in spurious rejection of valid observations during extreme events. Once the overall probability of error for an observation has been determined, output from the various specific observing system quality control tests are simply added to the error probability using unique integer-valued flags. The quality control flags have three levels of severity: (1) information-only ( development of operational oceanography and servicing in Australia. J Res Pract Inf Techol 39:151–164 Cardinali C, Pezzulli S, Andersson E (2004) Influence-matrix diagnostic of a data assimilation system. Q J R Meterologic Soc 130:2767–2786 Chambers DP, Tapley DB, Stewart RH (1999) Anomalous warming in the Indian Ocean coincident with El Niño. J Geophys Res 104:3035–3047 Chapnik B, Desroziers G, Rabier F, Talagrand O (2006) Diagnosis and tuning of observational error statistics in a quasi operational data assimilation setting. Q J R Meteorologic Soc 132:543–565 CLIVAR–GOOS Indian Ocean Panel et€al (2006) Understanding the role of the Indian Ocean in the climate system—implementation plan for sustained observations. WCRP Informal Rep. 5/2006, ICOP Publ. Series 100, GOOS Rep. 152, p€76 Corazza M, Kalnay E, Patil D, Yang S-C, Morss R, Cai M, Szunyogh I, Hunt B, Yorke J (2003) Use of the breeding technique to estimate the structure of the analysis errors of the day. Nonlinear Process Geophys 10:233–243 Evensen G (2003) The ensemble Kalman filter: theoretical formulation and practical implementation. Ocean Dyn 53:343–367 Feng M, Meyers GA, Wijffels SE (2001) Interannual upper ocean variability in the tropical Indian Ocean. Geophys Res Lett 28:4151–4154 Fujii Y, Tsujino H, Usui N, Nakano H, Kamachi M (2008) Application of singular vector analysis to the Kuroshio large meander. J Geophys Res 113. doi:10.1029/2007JC004476 Gallagher K, Sambridge M, Drijkoningen G (1991) Genetic algorithms: an evolution from MonteCarlo methods for strongly non-linear geophysical optimization problems. Geophys Res Lett 18:2177–2180 Gelaro R, Buizza R, Palmer TN, Klinker E (1998) Sensitivity analysis of forecast errors and the construction of optimal perturbations using singular vectors. J Atmos Sci 55:1012–1037 Gelaro R, Langland RH, Rohaly GD, Rosmond TE (1999) As assessment of the singular-vector approach to targeted observing using the FASTEX dataset. Q J R Meteorologic Soc 125:3299– 3327 Guinehut S, Le Traon P-Y, Larnicol G, Phillips S (2004) Combining argo and remote-sensing data to estimate the ocean three-dimensional temperature fields: a first approach based on simulated observations. J Mar Sys 46:85–98 Hackert EC, Miller RN, Busalacchi AJ (1998) An optimized design for a moored instrument array in the tropical Atlantic Ocean. J Geophys Res 103:7491–7509
150
P. R. Oke and T. J. O’Kane
Heimbach P et€al (2010) Observational requirements for global-scale ocean climate analysis: lessons from ocean state estimation. In: Hall J, Harrison DE, Stammar D (eds) Proceedings of OceanObs’09: sustained ocean observations and information for society, vol€2. ESA Publication WPP-306, Venice, Italy, 21–25 Sept 2009 (submitted) Hernandez F, Le Traon P-Y, Barth N (1995) Optimizing a drifter cast strategy with a genetic algorithm. J Atmos Ocean Technol 12:330–345 Holland WR, Malanotte-Rizzoli P (1989) Assimilation of altimeter data into an ocean circulation model: space versus time resolution studies. J Phys Oceanogr 19:1507–1534 Houtekamer P, Derome J (1995) Methods for ensemble prediction. Mon Weather Rev 123:2181– 2196 Khare SP, Anderson JL (2006) An examination of ensemble filters based adaptive observation methodologies. Tellus 58A:179–195 Kuo TH, Zou X, Huang W (1998) The impact of global positioning system data on the prediction of an extratropical cyclone: an observing system simulation experiment. Dyn Atmos Oceans 27:439–470 Kurapov AL, Egbert GD, Allen JS, Miller RN (2009) Representer-based analyses in the coastal upwelling system. Dyn Atmos Oceans 48:198–218 Langland RH (2005) Issues in targeted observations. Q J R Meteorologic Soc 131:3409–3425 Langland RH, Baker NL (2004) Estimation of observation impact using the NRL atmospheric variational data assimilation adjoint system. Tellus 56A:189–201 Masumoto Y, Meyers GA (1998) Forced Rossby waves in the southern tropical Indian Ocean. J Geophys Res 103:27589–27602 McPhaden MJ et€ al (1998) The tropical ocean global atmosphere (TOGA) observing system: a decade of progress. J Geophys Res 103:14169–14240 Miller RN (1990) Tropical data assimilation experiments with simulated data: the impact of the tropical ocean, global atmosphere thermal array for the ocean. J Geophys Res 95:11461–11482 Moore AM, Farrell F (1993) Rapid perturbation growth on spatially and temporally varying oceanic flows determined using an adjoint method: application to the Gulf Stream. J Phys Oceanogr 23:1682–1702 Moore AM, Arango HG, Di Lorenzo E, Miller AJ, Cornuelle BD (2009) An adjoint sensitivity analysis of the southern California current circulation and ecosystem. J Phys Oceanogr 39:702–720 Murtugudde R, McCreary JP, Busalacchi AJ (2000) Oceanic processes associated with anomalous events in the Indian Ocean with relevance to 1997–1998. J Geophys Res 105:3295–3306 O’Kane TJ, Frederiksen JS (2008a) Statistical dynamical subgrid-scale parameterizations for geophysical flows. Phys Scr 2008(T132):014033. doi:10.1088/0031-8949/2008/T132/014033 O’Kane TJ, Naughton M, Xiao Y (2008) AGREPS: the Australian global and regional ensemble prediction system. ANZIAM J 50:C308–C321 Oke PR, Schiller A (2007) Impact of argo, SST and altimeter data on an eddy-resolving ocean reanalysis. Geophys Res Lett 34. doi:10.1029/2007GL031549 Oke PR, Schiller A, Griffin DA, Brassington GB (2005) Ensemble data assimilation for an eddyresolving ocean model of the Australian region. Q J R Meteorologic Soc 131:3301–3311 Oke PR, Brassington GB, Griffin DA, Schiller A (2008) The Bluelink ocean data assimilation system (BODAS). Ocean Model 21:46–70 Oke PR, Balmaseda M, Benkiran M, Cummings JA, Dombrowsky E, Fujii Y, Guinehut S, Larnicol G, Le Traon P-Y, Martin MJ (2009) Observing system evaluations using GODAE systems. Oceanography 22(3):144–153 Oke PR, Balmaseda M, Benkiran M, Cummings JA, Dombrowsky E, Fujii Y, Guinehut S, Larnicol G, Le Traon P-Y, Martin MJ (2010) Observational requirements of GODAE Systems. In: Hall J, Harrison DE, Stammar D (eds) Proceedings of OceanObs’09: sustained ocean observations and information for society, vol€2, ESA Publication WPP-306, Venice, Italy, 21–25 Sept 2009 Palmer TN, Gelaro R, Barkmeijer J, Buizza R (1998) Singular vectors, metrics, and adaptive observations. J Atmos Sci 55:633–653
5â•… Observing System Design and Assessment
151
Rabier F, Courtier P, Pailleuz J, Hollingsworth A (1996) Sensitivity of forecast errors to initial conditions. Q J R Meteorologic Soc 122:121–150 Rabier F, Gauthier P, Cardinali C, Langland R, Tsyrulnikov M, Lorenc A, Steinle P, Gelaro R, Koizumi K (2008) An update on THORPEX-related research in data assimilation and observing strategies. Nonlinear Process Geophys 15:81–94 Rao SA, Behera SK (2005) Subsurface influence on SST in the tropical Indian Ocean: structure and interannual variability. Dyn Atmos Oceans 39:103–135 Sakov P, Oke PR (2008) Objective array design: application to the tropical Indian Ocean. J Atmos Ocean Technol 25:794–807 Schiller A, Wijffels SE, Meyers GA (2004) Design requirements for an Argo float array in the Indian Ocean inferred from observing system simulation experiments. J Atmos Ocean Technol 21:1598–1620 Schott FA, McCreary JP (2001) The monsoon circulation of the Indian Ocean. Prog Oceanogr 51:1–123 Schouten WP, de Ruijter M, van Leeuwen PJ, Dijkstra HA (2002) An oceanic teleconnection between the equatorial and southern Indian Ocean. Geophys Res Lett 29:1812. doi:10.1029/2001GL014542 Snyder C, Joly A (1998) Development of perturbations within a growing baroclinic wave. Q J R Meteorologic Soc 124:1961–1983 Tippett MK, Anderson JL, Bishop CH, Hamill TM, Whitaker JS (2003) Ensemble square root filters. Mon Weather Rev 131:1485–1490 Toth Z, Kalnay E (1997) Ensemble forecasting at NCEP and the breeding method. Mon Weather Rev 125:3297–3319 Tracton M, Kalnay E (1993) Operational ensemble prediction at national meteorological center: practical aspects. Weather Forecast 8:379–398 Tremolet Y (2008) Computation of observation sensitivity and observation impact in incremental variational data assimilation. Tellus 60:964–978 Vecchi GA, Harrison MJ (2007) An observing system simulation experiment for the Indian Ocean. J Climate 20:3300–3319 Vidard A, Anderson DLT, Balmaseda M (2007) Impact of ocean observation systems on ocean analysis and seasonal forecasts. Mon Weather Rev 135:409–429 Wang X, Bishop CH (2003) A comparison of breeding and ensemble transform Kalman filter ensemble forecast schemes. J Atmos Sci 60:1140–1158 Wei M, Frederiksen JS (2004) Error growth and dynamical vectors during southern hemisphere blocking. Nonlinear Process Geophys 11:99–118 Wei M, Toth Z, Wobus R, Zhu Y, Bishop CH, Wang X (2006) Ensemble transform Kalman filter-based ensemble perturbations in an operational global prediction system at NCEP. Tellus 58A:28–44 Wijffels SE, Meyers GA (2004) An intersection of oceanic waveguides: variability in the Indonesian throughflow region. J Phys Oceanogr 34:1232–1253
Part III
Atmospheric Forcing and Waves
Chapter 6
Air-Sea Fluxes of Heat, Freshwater and Momentum Simon A. Josey
Abstract╇ An overview of the air-sea fluxes of heat, freshwater and momentum is presented with the emphasis being on methods used to determine these fluxes and the role they play within the wider climate system. The equations used to determine the various heat flux components and the wind stress (which is equivalent to the momentum flux) are described in detail, together with the main spatial characteristics of the resulting global fields. This is followed by an overview of currently available flux datasets, including in situ, remotely sensed, atmospheric reanalysis and hybrid products. Methods for evaluation of these datasets are explored, including recent developments in the use of air-sea flux reference sites to discriminate between the different fields. Several topics that place surface fluxes in the context of global climate are then discussed including the ocean heat budget closure problem, climate change related trends in surface fluxes and impacts of extreme heat fluxes at high latitudes. Finally, some outstanding challenges are presented including the need for a better understanding of ocean-atmosphere interaction in the Southern Ocean and the potential for use of the integrated surface density flux to estimate variability in the Atlantic meridional overturning circulation.
6.1╅Introduction The exchanges of heat, freshwater and momentum between the oceans and the atmosphere play a pivotal role in the global climate system. In the tropics, there is a net input of heat to the ocean which is subsequently transported to mid-high latitudes and released back to the atmosphere, modifying the climate over land downstream (e.g. Rhines et€al. 2008). At several high latitude sites, intense winter heat loss (together with the effects of net evaporation and brine rejection associated with ice formation) drives deep convection and dense water formation, supplying the deep limb of the global overturning circulation. The wind stress on the ocean, which S. A. Josey () National Oceanography Centre, Southampton, UK e-mail:
[email protected] A. Schiller, G. B. Brassington (eds.), Operational Oceanography in the 21st Century, DOI 10.1007/978-94-007-0332-2_6, ©Â€Springer Science+Business Media B.V. 2011
155
156
S. A. Josey
is equivalent to the momentum exchange, is the other major driver of the circulation, and regional wind forcing also plays a key role in dense water formation through preconditioning of water masses as a result of upwelling. The freshwater flux (evaporation-precipitation) has a major impact on the ocean surface salinity field which to a large extent reflects the pattern of surface net evaporation. Despite their major role in the climate system, our level of knowledge regarding many aspects of ocean-atmosphere interaction remains at a basic level. Attempts to develop global datasets of these fluxes have been severely hampered by the lack of observations in many regions. The primary source of data has historically been merchant ship meteorological reports which tend to follow the main shipping routes leaving large areas of the ocean, particularly the Southern Ocean, extremely undersampled. This situation has improved for some flux related variables (sea surface temperature, wind speed) with the advent of satellite observations but these are only available for the past two decades and do not as yet provide reliable estimates of all terms in the surface heat budget. Anthropogenic climate change is widely expected to lead to changes in the fluxes of heat and freshwater as a result of global warming and strengthening of the hydrological cycle. There is compelling evidence that an increase in global ocean heat content has already happened (e.g. Levitus et€al. 2009) and this implies an increase in the global mean net ocean heat gain. However, the expected change is small, only about 0.5€W€m−2. This signal is too small to be detectable given the accuracy of currently available heat flux datasets and this situation is unlikely to change in the near future. A strengthening of the hydrological cycle will influence the ocean-atmosphere exchange of freshwater and potentially leave an imprint in ocean salinity. Due to problems with obtaining reliable precipitation measurements, the level of uncertainty in freshwater flux datasets is greater than that for heat flux and it is again difficult to detect anthropogenic climate change in this variable. However, there is some evidence that changes in the hydrological cycle have modified ocean salinity as this acts as an integrator of variations in the surface freshwater exchange (Stott et€al. 2008). In this paper, I provide a short overview of the current state of ocean-atmosphere interaction research. A thorough review of all aspects of air-sea exchanges was carried out by the Working Group on Air-Sea Fluxes in the late 1990s (WGASF 2000) and this remains a major resource which the interested reader is recommended to consult. A further valuable point of reference from the perspective of the ocean observing system is the Plenary White Paper on air-sea fluxes prepared for Ocean Obs’09 (Gulev et€al. 2009). Progress in understanding ocean-atmosphere interaction in the face of a fundamental sampling problem and uncertainty over significant elements of the underlying physics has been the result of dedicated efforts by a wide international research community. I have attempted to summarise some of the key results here from a personal perspective which stems from my own research developing and analysing in situ observation based fields and more recently studying the wider role of fluxes using coupled models. I began my research career studying a very different class of surface flux, the effects of the infalling flux of primordial gas (primarily neutral hydrogen) onto the discs of spiral galaxies (Josey and Tayler 1991; Josey and Arimoto 1992). This presents a very different set of research prob-
6â•… Air-Sea Fluxes of Heat, Freshwater and Momentum
157
lems but provides an interesting alternative perspective on the effects that surface exchanges have on a system. I count myself lucky to have worked initially in this field and subsequently on the equally fascinating, and arguably more important, role of surface fluxes in the global climate system. Following this introduction, an outline of the formulae used to estimate surface fluxes is given in Sect.€6.2 and an overview of the different flux datasets in Sect.€6.3. Flux evaluation methods are then considered in Sect.€6.4. Several issues related to the role of surface fluxes in the global climate system are discussed in Sect.€6.5, while the final Sect.€6.6 highlights several outstanding issues and potential future applications of the air-sea exchanges, particularly as regards estimates of variability in the ocean overturning circulation.
6.2╅Surface Flux Theory 6.2.1 Flux Components and Spatial Variation The net air-sea heat flux is the sum of four components: two turbulent heat flux terms (the latent and sensible heat fluxes) and two radiative terms (the shortwave and longwave fluxes). These are shown schematically in Fig.€6.1 together with their global mean values from a globally balanced air-sea heat flux dataset (Grist and Josey 2003).
Fig. 6.1↜渀 Schematic representation of the different components of the air-sea heat exchange with global annual mean values of the key terms from a balanced flux dataset. (Grist and Josey 2003)
158 80
S. A. Josey NOC1.1a Latent Heat Flux
40
200
80
100
40
0
0
– 100
– 40
– 200 80
NOC1.1a Longwave Flux
40
80 40 0 – 40
0 – 100 – 200
100
40
– 200 NOC1.1a Net Heat Flux
100
– 40 80
– 100
– 40
200
0
200
0
0
NOC1.1a Sensible Heat Flux
0 – 40
NOC1.1a Shortwave Flux
200 100 0 – 100 – 200
200 100 0 – 100 – 200
Fig. 6.2↜渀 Climatological annual mean fields of the different heat flux components and the net heat flux. (Source: National Oceanography Centre 1.1a (NOC1.1a) flux climatology, units W€m−2, Grist and Josey 2003)
Climatological annual mean fields of the different components and net heat flux are shown in Fig.€6.2. The sign convention is for positive fluxes to represent heat gain by the ocean. For the turbulent heat flux components (i.e. the sensible and latent terms), the areas of strongest loss are over the Gulf Stream and Kuroshio with latent heat losses of order 200€W€m−2. Enhanced latent heat loss is also seen in the South-East Indian Ocean where the trade winds are particularly strong. The sensible heat flux is typically much smaller in magnitude than the latent term, the strongest losses occur in regions where very cold air is advected over the ocean from neighbouring land masses particularly the Labrador and Norwegian Seas. The global variation in the net longwave flux is relatively small, typical values ranging from 30–70€W€ m−2. However, within this range there is a degree of structure which reflects the balance between the sea-air temperature difference, the cloud cover and the amount of water vapour. The most noticeable feature is a band of reduced longwave loss under the Inter-Tropical Convergence Zone (ITCZ). In contrast, the shortwave field has a primarily meridional variation determined by the mean solar elevation with peak values of order 200€W€m−2. The main departures from this variation occur under regions of increased cloud cover such as the ITCZ. Finally, the net heat flux field is seen to be dominated by the contributions from the
6â•… Air-Sea Fluxes of Heat, Freshwater and Momentum
159
shortwave and latent heat fluxes with shortwave driven ocean heat gain in the Tropics and latent heat driven ocean heat loss over the western boundary current regions. The processes controlling these exchange terms and methods for their estimation are discussed below, for a more detailed review see WGASF (2000).
6.2.2 Turbulent Flux Bulk Formulae The latent and sensible heat fluxes are proportional to the products of the near surface wind speed with the sea-air humidity and sea-air temperature difference respectively. However, the detailed form of these relationships remains poorly known under certain conditions, in particular at high wind speeds and this provides a significant source of uncertainty in estimates of these fluxes. The sensible and latent heat fluxes, QH and QE, are generally determined using the following bulk formulae:
QH = ρcp Ch u(Ts − (Ta +γ z))
(6.1)
QE = ρLCe u(qs − qa )
(6.2)
where ρ is the density of air; cp , the specific heat capacity of air at constant pressure; L, the latent heat of vaporisation; Ch and Ce, the stability and height dependent transfer coefficients for sensible and latent heat respectively; u, the wind speed; Ts, the sea surface temperature; Ta, the surface air temperature with a correction for the adiabatic lapse rate, , z, the height at which the air temperature was measured; qs, 98% of the saturation specific humidity at the sea surface temperature to allow for the salinity of sea water, and qa, the atmospheric specific humidity. A major amount of research has been devoted over the past few decades to accurately determining values for the transfer coefficients and their functional dependence on wind speed and near surface stability by means of direct flux measurements, in particular through the eddy correlation method. This work has lead to the development of the COARE flux algorithm (Fairall et€al. 2003) which has greatly reduced uncertainty in the values of the transfer coefficients although questions still remain in several areas, particularly the high wind speed regime and inclusion of the effects of sea spray.
6.2.3 Radiative Flux Parameterisations The shortwave flux is primarily a function of solar elevation and cloud amount with an additional dependence on ocean albedo. The longwave (infrared) flux is the difference between large downwelling and upwelling terms from the ocean and atmosphere respectively and depends on sea surface temperature, air temperature and humidity in addition to cloud amount. The longwave and shortwave flux components have been determined using a wide range of empirical formulae over the years
160
S. A. Josey
(e.g. Clark et€al. 1974; Bignami et€al. 1995; Josey et€al. 2003). The performance of several bulk formula parameterisations for the net longwave flux has been assessed by comparison with radiometer measurements made at sea during a number of cruises (Josey et€al. 1997). More recently Josey et€al. (2003) carried out a detailed evaluation of both the Clark et€al. (1974) and Bignami et€al. (1995) formulae using measurements made on a long meridional research cruise from 20–63°N at 20°W in the North Atlantic. This analysis made use of recent advances in understanding of various biases in the pyrgeometer instrument used to measure the longwave flux (Pascal and Josey 2000). Neither formula was found to be capable of providing reliable estimates of the atmospheric longwave flux over the full range of latitudes. The Clark formula overestimated the cruise mean measured longwave flux of 341.1€W€m−2 by 11.7€W€m−2, while Bignami underestimated by 12.1€W€ m−2. Josey et€ al. (2003) developed an alternative formula which expresses the combined effects of cloud cover and other relevant parameters on the atmospheric longwave in terms of an adjustment to the measured air temperature. The net longwave flux, QL, across the ocean-atmosphere interface is given by:
QL = QLS − (1 − αL )QLA
(6.3)
4 QLA = σSB TEff
(6.4)
where QLS is the emitted longwave radiation from the sea surface, QLA is the downwelling longwave radiation from the atmosphere, and the coefficient (1−αL ), where αL is the longwave reflectivity, takes account of the component of the downwelling radiation reflected from the sea surface. They characterise the downwelling longwave radiation by an effective blackbody temperature, TEff, such that,
where σSB is the Stefan-Boltzmann constant (5.67â•›×â•›10−8€W€m−2 K−4). Given that the observed variable is Ta instead of TEff, they write TEff as the sum of Ta and a temperature adjustment, ∆Ta, which includes the effects of cloud cover, atmospheric humidity and other, as yet unknown, variables on the downwelling longwave, such that,
QLA = σSB (Ta + Ta )4
(6.5)
∆Ta is thus the difference between the measured air temperature and the effective temperature of a blackbody which emits a radiative flux equivalent to the atmospheric longwave. The problem of obtaining a reliable estimate for QLA then becomes one of parameterising the dependence of ∆Ta on cloud cover, vapour pressure and any other relevant variables. The air temperature is adjusted by the amount necessary to obtain the effective temperature of a blackbody with a radiative flux equivalent to that from the atmosphere. A simple parameterisation of the temperature adjustment solely in terms of the total cloud amount leads to a net longwave flux formula which has an improved mean bias error with respect to the cruise measurements of −1.3€W€m−2. The new formula still exhibits significant biases under certain situations, in particular overcast, low cloud base conditions at high latitudes. However, by modify-
6â•… Air-Sea Fluxes of Heat, Freshwater and Momentum
400 Estimated Longwave, W m– 2
Fig. 6.3↜渀 Comparison of the atmospheric component of the net longwave flux estimated using Eq.€(6.6) with measurements made on a research cruise in the North Atlantic. (Modified version of figure from Josey et€al. (2003), copyright American Geophysical Union)
161
350
300
250 250
300 350 Measured Longwave, W m– 2
400
ing this formula to include a dependence on the dew point depression, good agreement between the measured and estimated mean longwave over the full range of observations can be obtained and the mean bias error reduced to 0.2€W€m−2 (see Fig.€6.3). The resulting formula for the net longwave flux is as follows:
QL = εσSB Ts4 − (1 − αL )σSB {Ta + an2 + bn + c + 0.84(D + 4.01)}4 (6.6)
where is the emissivity of the sea surface, taken to be 0.98, αLâ•›=â•›0.045 and n is the fractional cloud cover. The terms a, b and c are empirical constants and D is the dew point depression, Dâ•›=â•›TDewâ•›−â•›Ta, where TDew is the dewpoint temperature (i.e. the temperature at which it becomes saturated) of the air in the surface layer. The new formula was tested using independent measurements made on two more recent cruises and found to perform well, agreeing to within 2€W€m−2 in the mean, at midhigh latitudes. In contrast, to the formulae for the sensible, latent and longwave fluxes which may be used with individual ship meteorological reports, widely-used formulae for the net shortwave flux typically provide monthly mean values. In particular, the following formula of Reed (1977) provides the monthly mean net shortwave flux,
QSW = (1 − α)Qc [1 − 0.62 n + 0.0019 θ N ]
(6.7)
where α is the albedo, Qc is the clear-sky solar radiation, n is the monthly mean fractional cloud cover and θ N is the monthly mean local noon solar elevation. Gilman and Garrett (1994) note that under conditions of low cloud cover, the Reed formula estimate of the mean incoming shortwave can become greater than the clear-sky value if θ N is sufficiently large. and suggest that the incoming shortwave be constrained to be less than or equal to Qc .
162
S. A. Josey
Finally, the net heat flux, QNet, is given by the sum of the four individual components, (6.8)
QNet = QE + QH + QL + QSW
where QE is the latent heat flux; QH, the sensible heat flux; QL, the longwave flux and QSW, the shortwave flux.
6.2.4 Wind Stress Estimates of the zonal, τx , and meridional, τy , components of the sea surface wind stress are typically obtained using the following equations, τx = ρCD ux (ux2 + uy2 )1/2
(6.9)
τy = ρCD uy (ux2 + uy2 )1/2
where ux and uy are the zonal and meridional components of the wind speed respectively, and CD is the drag coefficient which depends upon the height of the wind measurement and the atmospheric stability as well as wave characteristics (e.g. Smith 1988; Taylor and Yelland 2001). Climatological analyses of the wind stress using these formulae with ship meteorological reports have been carried out in a number of studies (e.g. Hellerman and Rosenstein 1983; Harrison 1989; Josey et€al. 2002). More recently various satellite products have become available which avoid the sampling issues inherent with ship observations but are restricted to the past decade or so, for example microwave scatterometer measurements made by QuikSCAT (http://winds.jpl.nasa.gov/). The climatological annual mean wind stress field from the NOC1.1 flux dataset is shown in Fig.€6.4. The figure reveals patterns associated with subtropical and subpolar gyres, the ITCZ and the band of intense westerly wind stress in the Southern
Latitude
NOC1.1 Wind Stress - Annual Mean (N m– 2) 80 60 40 20 0 – 20 – 40 – 60 – 80
0.2 0.15 0.1 0.05
80
70
60
50 Longitude
40
30
0
Fig. 6.4↜渀 Climatological annual mean wind stress, source NOC1.1 climatology, units N€m−2, Josey et€al. 2002. Colours show the magnitude of the wind stress vectors. (Modified version of figure in Josey et€al. (2002), copyright American Meteorological Society)
6â•… Air-Sea Fluxes of Heat, Freshwater and Momentum
163
Ocean. The curvature of the wind stress field is a measure of local upwelling and downwelling and its integral at a given latitude provides a measure of the strength of the wind driven circulation via the Sverdrup transport (for further discussion of these fields with reference to the NOC1.1 climatology see Josey et€al. 2002).
6.2.5 Freshwater Flux The air-sea freshwater flux is simply the difference between evaporation lost from the ocean surface and precipitation gained by the ocean from the atmosphere, often written E-P (i.e. evaporation-precipitation). It is linked to the net heat flux as the evaporation term corresponds to the latent heat flux component of the net heat exchange discussed above. Estimates of the evaporation are available from ship-based flux datasets, atmospheric model reanalyses and satellite measurements. Various precipitation products are available from satellites (Gulev et€al. 2009), for example as a result of the Global Precipitation Climatology Project Version 2 (GPCPv2, Adler et€al. 2003). However, there are significant regional differences between the various products and as a consequence precipitation is the least well determined surface exchange field. Atmospheric model reanalyses also provide precipitation but here care must be taken as unphysical trends have been observed in some areas, particularly for the European Centre for Medium Range Weather Forecasting (ECMWF) reanalysis in the Tropics. Precipitation is difficult to measure directly at sea (Weller et€al. 2008) but may be estimated from present weather codes in voluntary observing ship meteorological reports (via limited historical calibration against island station rain measurements) and was included in the NOC1.1 flux dataset (Josey et€al. 1999). However, further work is needed before this method can be reliably used for climate studies.
6.2.6 Density Flux The combined impact of the net heat flux and evaporation on the buoyancy of water in the sea surface layer may be expressed in terms of the density flux. The total density flux, Fρ , into the ocean surface is given by the following equation, QNet E−P (6.10) Fρ = −ρ α − βS ρcP (1 − S/1000) where ρ is the density of water at the sea surface; cP, the specific heat capacity of water; S, the sea surface salinity and α and β, the thermal expansion and haline contraction coefficients which are defined as follows,
α=−
1 ∂ρ ; ρ ∂T
β=
1 ∂ρ ρ ∂S
(6.11)
164
S. A. Josey
The density flux is frequently split into thermal, FT, and haline, FS, contributions defined as follows, where,
(6.12)
Fρ = FT + FS
FT = −α
QNet ; cP
FS = ρβS
E−P (1 − S/1000)
(6.13)
Heat loss from the ocean (QNet╛╛P) then result in positive values for FT and FS respectively and an increase in the density of the near surface layer. The thermal term usually dominates the density flux with the haline term playing only a minor role (e.g. Josey 2003; Grist et€al. 2007) except at high latitudes.
6.3â•…Air-Sea Flux Datasets The three primary sources of information regarding air-sea fluxes are surface meteorology reports (mainly from Voluntary Observing Ships), satellite observations and atmospheric model reanalyses which assimilate various data types. All three sources have been employed with the bulk formulae (Eqs.€6.1 and 6.2) to estimate the latent and sensible heat fluxes given a knowledge of the surface meteorology. The radiative fluxes have been determined either from empirical formulae, of the type described in the previous section, or from radiative transfer models. Many air-sea flux datasets have been developed over the past four decades. For example, the pioneering effort of Bunker (1976) relied on merchant ship meteorological reports, while in recent years satellite observations and output from numerical weather prediction models have been combined in new hybrid products (e.g. Yu and Weller 2007). The first flux datasets comprised climatological monthly fields of ether the full set or a subset of the heat, momentum and freshwater fluxes typically based on observations spanning many decades. In the 1990s, several analysis efforts continued to focus on producing climatological fields and addressing specific scientific problems—principally achieving closure of the global ocean heat budget—but in addition provided the individual monthly fields on which the climatologies were based (da Silva et€al. 1994; Josey et€al. 1998). In recent years, climatological fields have taken a back seat and several new flux products contain fields at daily timescales as well as monthly. This tendency has been driven, in part, by the high time resolution possible with the atmospheric reanalyses and the need to include high frequency variability in forcing fields for ocean model runs. A full survey of the wide range of methods used to produce flux datasets and the details of the underlying observing system is beyond the scope of the current paper. Instead an overview of the main classes of flux datasets is presented and the interested reader is referred to WGASF (2000) and Gulev et€al. (2009) for further details.
6â•… Air-Sea Fluxes of Heat, Freshwater and Momentum
165
6.3.1 In Situ Observation Based Fields The only source of information regarding air-sea fluxes for many years were routine merchant ship meteorological reports collected under the Voluntary Observing Ships (VOS) programme and collated to form the Comprehensive Ocean-Atmosphere Dataset (COADS, Woodruff et€ al. 1987) which has now become International COADS (Worley et€al. 2005, 2009). Estimates of the various surface heat flux components were obtained either from individual surface meteorology reports, or from monthly averaged values of the key variable such as wind speed (although this has the potential to lead to biases as a result of neglected correlations between the different variables, Josey et€al. 1995) using formulae of the type discussed in Sect.€6.2. The resulting flux estimates are then combined using various averaging and interpolation techniques to form gridded fields. Two widely used flux products developed using this approach have been the UWM/COADS dataset of da Silva et€al. (1994) and the National Oceanography Centre 1.1 (NOC1.1) flux dataset (Josey et€al. 1998, 1999—formerly termed the Southampton Oceanography Centre (SOC) flux climatology), recently revised using optimal interpolation (NOC2, Berry and Kent 2009) to include error estimates (Kent and Berry 2005). The major problem with ship based flux datasets is the uneven distribution of meteorological reports, which are heavily concentrated along the major shipping routes, leading to significant undersampling of the required fields in many regions—including much of the Southern Hemisphere (for example see Fig.€6.2 of Josey et€al. 1999). This is likely to have played a major role in the ocean heat budget closure problem which has affected to a certain extent all flux datasets produced to date and is manifest as a 20–30€W€m−2 global mean net ocean heat gain while in reality the budget should be closed to of order 1€W€m−2 at decadal and longer timescales. We will return to this issue in Sect.€ 6.5.1 but note here that several flux datasets have achieved closure by applying inverse analysis techniques with hydrographic observations of ocean heat transport as constraints (e.g. the NOC1.1a fields described in Grist and Josey (2003) which are an adjusted, globally balanced version of the original NOC1.1 climatology). A further issue with ship based fluxes is the diverse range of instrumentation types used for making the routine meteorological measurements (e.g. air temperature, specific humidity) under the VOS programme. Each sensor type has its own error characteristics that need to be determined in order to correct for biases prior to determining the fluxes (e.g. Josey et€al. 1999). A recent development, targeted at reducing these errors is the VOS Climate Project (VOSCLIM) originally suggested by Taylor et€al. (2001). One of the goals of this project is to provide a high-quality VOS data subset that can be used to better calibrate the VOS fleet as a whole. A further initiative, the Shipboard Automated Meteorological and Oceanographic System (SAMOS, Smith et€al. 2010), seeks to collect high quality meteorological and flux measurements from research ships and provide these as a resource which may be used for better determination of biases in both the VOS measurements and other flux products (e.g. the reanalyses). SAMOS has focused on data obtained from the
166
S. A. Josey
US research ships but provides an example which, if applied internationally, would create an even more valuable resource.
6.3.2 Remotely Sensed fluxes Remote sensing is now capable of providing observations of some of the key airsea flux terms and has the major advantage over ship based estimates of essentially complete global coverage. However, satellite estimates suffer because it is not yet possible to reliably measure near surface air temperature and humidity directly from space. Indirect techniques must be used instead and this leads to a major source of uncertainty in the turbulent heat flux terms which are critically dependent on the sea-air temperature and humidity difference. Estimates of the radiative flux terms are available from various sources, most recently from the Moderate Resolution Imaging Spectro-radiometer (MODIS, e.g. Pinker et€al. 2009) and have been combined with indirect estimates of the turbulent fluxes to form net heat flux products; a recent example is the Hamburg Ocean-Atmosphere Parameters and Fluxes from Satellite Data version 3 (HOAPS3, Andersson et€ al. 2010). However, significant uncertainties remain in such net heat flux fields because of problems with determining the latent and sensible heat flux. In contrast, to the net heat flux, the wind stress is now well determined as a result of QuikSCAT although there are concerns as to whether this will remain the case in the near future given the likely imminent demise of this mission. Precipitation has also been determined using various techniques including infrared measurements of cloud top brightness temperature, which acts as a proxy for rain rate, and passive microwave measurements. Such estimates have been combined under the Global Precipitation Climatology project (GPCP) to form best estimates of the rainfall (CPCPv2, Adler et€ al. 2003). However, validation of these fields over the ocean is challenging due to the lack of high quality measurements from rain sensors and the difficulty with making this measurement (e.g. Weller et€al. 2008). As a consequence, major uncertainty remains in the precipitation fields with knock-on effects for attempts to estimate the air-sea freshwater flux (E-P).
6.3.3 Atmospheric Model Reanalyses Numerical weather prediction models assimilate a wide range of observations including surface meteorological reports, radiosonde profiles and remote sensing measurements. In recent decades, these models have had the potential to provide the complete set of air-sea flux fields at high (6€hourly) resolution with full spatial coverage. However, they are of course dependent on the model physics which, although constrained to some extent by the assimilated observations, has the potential to produce large biases, particularly in the radiative flux fields and precipitation (e.g. Trenberth et€al. 2009). Fixed versions of the models run over multidecadal pe-
6â•… Air-Sea Fluxes of Heat, Freshwater and Momentum
167
riods are commonly referred to as atmospheric reanalyses—the two major products being those from the National Center for Environmental Prediction and the National Center for Atmospheric Research (NCEP/NCAR) and ECMWF. For the reanalyses, the turbulent flux terms are again estimated from the model surface meteorology fields while the shortwave and longwave flux are output from the radiative transfer component of the atmospheric model. To date, available reanalyses have been on a relatively coarse grid on the order of 1.5–2°. However, higher resolution reanalyses are anticipated in the near-term which will, for the first time, assimilate radiance measurements from satellites. There are hopes that these new products will contain smaller biases than those currently available (Trenberth et€al. 2009).
6.3.4 Other Flux Products In addition to the three primary classes of flux dataset described above, flux fields are available from several other types of products. The leading example here is the Objectively Analyzed air-sea Fluxes (OAFLUX) dataset (Yu and Weller 2007) which blends reanalysis and satellite surface meteorology fields prior to estimation of the fluxes, but still suffers from being unable to close the global ocean heat budget. A further product, combining reanalysis and satellite measurements, is the Common Ocean Reference Experiment (CORE) flux dataset (Large and Yeager 2009) which has been designed to provide forcing fields for ocean models. This requires closure of the ocean heat budget and this has been achieved via adjustments to several of the underlying fields which, although plausible, are not the result of comprehensive analysis. Thus this product must be regarded as a possible solution to the closure problem rather than necessarily being the correct solution. The climatological annual mean net air-sea heat flux field for the mid-latitude North Atlantic from four different flux products (including OAFLUX) is illustrated in Fig.€6.5. The same broad scale pattern is observed for each dataset with strong heat loss over the Gulf Stream and a transition towards ocean heat gain from west to east. The NCEP/NCAR fields tend to have stronger heat loss than the other 3 datasets considered and this is partly due to use of a transfer coefficient scheme which results in high values that are not supported by observational analyses. NOC1.1, NOC2 and OAFLUX all show similar results for the location of the zero net heat flux line which extends from south-west to north-east across the basin. Surface fluxes are also available from various ocean synthesis efforts, that is ocean models with data assimilation such as the Estimating the Circulation and Climate of the Ocean (ECCO) model. These are typically forced by NCEP or ECMWF reanalysis fields which are then adjusted as a result of the assimilation process. For the ECCO model, in some regions, comparisons against independent measurements suggest the resulting fields may be an improvement over the original forcing data (Stammer et€al. 2004). However, there remains a high degree of divergence between the different ocean model syntheses, and although this method holds some promise, it is not yet at the stage where it can provide reliable estimates of the surface
168
S. A. Josey NCEP/NCAR
NOC1.1
150
150
60
60
100
100 55
55
50
50 50
50 0
45
– 50
40
0
45
– 50
40
– 100 35
35
b
a
30 – 80 – 60 – 40 – 20
– 100
0
20
40
NOC2
– 150 30 – 80 – 60 – 40 – 20
– 150 0
20
40
OAFLUX
150
150
60
60 100
100 55
55
50
50 50
50 0 45 – 50
40
– 50
40
– 100
– 100 35
35 30
0
45
c
– 80 – 60 – 40 – 20
– 150 30 0
20
40
d
– 80 – 60 – 40 – 20
– 150 0
20
40
Fig. 6.5↜渀 Annual mean net air-sea heat flux from (a) NCEP/NCAR, (b) NOC1.1, (c) NOC2 and (d) OAFLUX for the common period 1984–2004, units W€m−2. Blue colours indicate ocean heat loss to the atmosphere, red indicate ocean heat gain
exchanges. Finally, the so-called residual method obtains the net surface heat flux as the residual of top of the atmosphere heating, measured by satellites, and the atmospheric heat divergence obtained from reanalysis (e.g. Trenberth and Caron 2001). This method has the potential to provide a valuable complementary estimate of the net heat exchange (but not of course the individual components). However,
6â•… Air-Sea Fluxes of Heat, Freshwater and Momentum
169
it is dependent on the accuracy of the atmospheric reanalysis which as noted above requires improvement. Each of these classes of flux product has its own advantages and disadvantages and it is not possible to recommend a best flux product; rather, the choice of flux dataset must be guided by the scientific issue which is to be addressed.
6.4â•…Methodology for Evaluating Surface Fluxes The discussion above has provided some indication of the diverse range of air-sea flux datasets that are now available for the community to use. All of these are limited in some manner by spatially and temporally dependent biases and it is therefore vital that each new flux dataset is properly evaluated against a range of independent measures in order to quantify these biases and understand their causes. Historically, this has not been the case, partly because of a lack of reference data. This issue has been recognised for some time, in particular Josey and Smith (2006) developed a methodology for evaluation of air-sea heat, freshwater and momentum flux datasets in response to a recommendation of the CLIVAR Global Synthesis and Observation Panel (GSOP). The panel recognised the need for such guidelines in order to facilitate consistent evaluation and intercomparison of the many new flux datasets being developed (particularly those from ocean reanalyses). The methodology makes use of both research quality data from flux buoys and research vessels (local evaluation) and large scale constraints (regional and global evaluations). For clarification of terminology, Josey and Smith (2006) defined two main classes of flux dataset. The first consists of the large scale ‘gridded flux datasets’ (typically at spatial resolutions of order 1° and timescales from 6€hourly to monthly) produced from in situ, model or remote sensing sources, or some combination thereof. The second class of datasets was termed ‘research quality data’, most of which are in-situ point measurements (for example radiative fluxes and meteorological variables from research buoys/vessels) at high temporal resolution (typically available as averages on timescales of order minutes). In summary, their key evaluation points are as follows: a. Local evaluation of time averaged fluxes and meteorological variables at specific grid locations with corresponding research quality data from surface flux reference moorings and vessels. b. Regional evaluation of either gridded flux product ocean transports or, preferably, area averaged fluxes with corresponding research quality data from hydrographic sections. c. Global evaluation of gridded flux product area weighted mean fluxes through closure of the appropriate property budget within observational constraints. They noted several difficulties in implementing this method including the lack of a central archive of heat and freshwater transports required for point b. This remains a problem at present and the creation of such an archive would be highly
170
S. A. Josey
desirable for flux evaluation studies. Despite these problems this methodology has been adopted to some extent in recent studies particularly for the OAFlux and CORE products (Yu and Weller 2007; Large and Yeager 2009). Evaluations of flux products in specific air-sea interaction regimes using flux reference buoys are becoming more common practice as the global distribution of such buoys increases, fostered through the OceanSITES programme (Send et€al. 2009). A recent example is an evaluation of the new satellite based J-OFURO2 flux dataset using two moorings in the Kuroshio region of the north-west Pacific Ocean (Tomita et€al. 2010).
6.5â•…Surface Fluxes in the Global Climate System 6.5.1 T he Implied Ocean Heat Transport and the Closure Problem The excess of heat gain over heat loss in the Tropics, as revealed in the net heat flux spatial field (Fig.€6.2), requires that the oceans transport energy away from the equator and towards the poles. Evidence for this latitudinal variation is provided by direct estimates of the ocean heat transport from hydrographic sections, which were collected in significant numbers for the first time as part of the World Ocean Circulation Experiment (WOCE); this variation is illustrated by the crosses in Fig.€6.6. In addition to the direct estimates of the heat transport, indirect estimates, Hϕ , may be obtained by integrating the net heat flux, QN, across successive latitude bands from a reference latitude ϕo which has a known value of the heat transport, Ho, from hydrography,
Hϕ = Ho −
ϕo λ2
QN dλdϕ
(6.14)
ϕ λ1
where λ1 and λ2 are the longitude limits at the western and eastern continental boundaries respectively of a given latitude band. The general form for this equation includes a term that accounts for heat storage by the ocean. However, as heat storage is relatively small at multi-decadal timescales, the storage term may be set equal to zero for calculating the implied climatological transport. Taking this approach, the implied ocean heat transports obtained with a range of surface flux datasets for the Atlantic, Pacific and Global Oceans are shown in Fig.€6.6. These reveal a peak in the transport values at about 20°N although the details differ between the datasets. In some cases, the hydrography can be used to indicate problems with the surface forcing fields, for example the ECMWF product diverges from hydrography in the southern hemisphere. It should, however,
6â•… Air-Sea Fluxes of Heat, Freshwater and Momentum
171
Fig. 6.6↜渀 Climatologically implied ocean heat transport derived by integrating the net heat surface flux southwards from 65°N. Key: ECMWF—Dash-dot red; Large and Yeager (2009)—Dashed blue; NCEP—Dashed magenta; NOC1.1a—Solid black; Trenberth residual—Dashed black; UWM/COADS—Solid grey. The crosses with error bars represent direct hydrographic estimates of the heat transport—updated version of Fig.€6.9. (In Grist and Josey (2003), copyright American Meteorological Society)
be noted that all of the flux products shown have been adjusted either directly or indirectly to achieve global closure and this to some extent ensures agreement with the hydrography. In the case of the reanalyses, the values for the transfer coefficients in the turbulent flux formulae are higher than can be supported by observations (e.g. Renfrew et€al. 2002). NOC1.1a and UWM/COADS have been made to agree with at least some of the hydrographic values using the technique of inverse analysis, first applied by Isemer et al. (1989). Most recently, the Large and Yeager (2009) fields have been modified using various plausible adjustments as noted earlier. Without such adjustments, the implied ocean heat transport would diverge rapidly from the hydrographic values and this is a manifestation of the more
172
S. A. Josey
general ocean heat budget problem i.e. the inability to close the global ocean heat budget at decadal timescales to within the 1€W€m−2 required to avoid unrealistically large warming signals. The budget closure problem has been recognised for many years and, despite various advances in our understanding of air-sea interaction, it remains a major issue for both ship based (e.g. NOC1.1 and NOC2) and remote sensing/reanalysis hybrid products (OAFLUX) all of which have global mean net heat flux values in the range 20–30€W€m−2. Progress towards resolution of this problem has been limited and it is likely to be the result of the combination of various small biases which amount to 3–5€W€m−2 in the global mean. These are likely to include (i) sampling issues revolving around the gross deficit of information on air-sea exchange in the Southern Hemisphere, (ii) missing physics in the high and low wind speed regime applications of the turbulent bulk flux formulae, (iii) a potential fair weather bias i.e. avoidance of high wind regions in merchant ship reports which will affect both in situ climatologies (directly) and reanalyses (indirectly as they rely on surface observations in the data assimilation), (iv) residual biases in ship meteorological reports which have yet to be determined, (v) uncertainty in the empirical formulae used to estimate the radiative fluxes (in situ based fields) and problems with representation of clouds (reanalyses). Only by a careful examination of each of these issues will progress be made towards obtaining an accurate picture of the global ocean-atmosphere heat exchange field. At a time when it is possible to calculate the climate change related signal in the global mean net heat flux to be of order 0.5€W€m−2 from observed variations in ocean heat content, it remains a major problem that it is not possible to reliably close the global mean ocean heat budget to better than 20€W€m−2.
6.5.2 Climate Change Related Trends in Surface Fluxes Both observation and model based analyses of changes in the surface air-sea heat flux associated with increasing global ocean heat content have revealed that the anthropogenic climate signal is small compared to natural variability (Pierce et€al. 2006; Levitus et€al. 2009). Changes in the net surface heat flux over the past 50 years at global and basin scales are expected to be about 0.5€W€m−2 with corresponding changes in the individual heat flux components of less than 2€W€m−2. Lozier et€al. (2008) have examined the spatial pattern of heat-content change in the North Atlantic using historical hydrographic station data from the National Oceanic Data Center World Ocean Database from 1950 to 2000. They find that the total heat gained by the North Atlantic Ocean is equivalent to a basin wide increase in the flux of heat across the ocean surface of 0.4€W€m−2. However, they note that it is not possible to say whether this gain is due to anthropogenic warming because natural variability may be masking this signal. An example of the total net heat flux variability since 1949 from a region in the mid-latitude North Atlantic is given in Fig.€6.7. The figure shows a time series of
Net Heat Flux (W m– 2)
6â•… Air-Sea Fluxes of Heat, Freshwater and Momentum
173
100 0
–100 1950
1960
1970
1980 Year
1990
2000
2010
Fig. 6.7↜渀 Monthly mean net air-sea heat flux anomaly for the box (40–55°N, 20–40°W) from NCEP/NCAR (↜red), NOC1.1 (↜green), NOC2 (↜blue) and OAFLUX (↜black), units W€m−2
the monthly net heat flux anomaly (i.e. with seasonal cycle removed) averaged over an example box (40–55°N, 20–40°W) in the mid-latitude North Atlantic for each of the four flux datasets. Strong month to month variability is evident in the figure with box averaged anomalies often exceeding 50€W€m−2. Similar variations are observed in each of the datasets for the periods in which they overlap. To some extent this is to be expected as, despite major differences in analysis methods, observations from Voluntary Observing Ships are a primary source of data for each of the flux products considered. The advent of Argo float data has enabled the study of the role of surface heat flux variability in causing interannual variability in ocean heat content in the North Atlantic in recent years (e.g. Hadfield et€al. 2007; Wells et€al. 2009). At decadal timescales, the relative roles of ocean heat transport and surface heat flux variations in North Atlantic temperature variability have been examined from an ocean model perspective by Marsh et€al. (2008) and Grist et€al. (2010). An intensification of the hydrological cycle is also expected as a result of anthropogenic climate change (e.g. IPCC 2007) with regional impacts on E-P as spatial patterns and the relative intensity of the evaporation and precipitation shift. It is worth noting that changes in evaporation imply a corresponding change in the latent heat flux, the two being related by the following simple equation, QE = ρ0 LE
where ρ0 is the fresh water density as a function of temperature. Thus, analysis of changes to the evaporation rate using observational datasets also need to take into account the implied change in latent heat flux and use the value obtained as a check on whether the changes in E are physically plausible. This is particularly important as spurious trends in E have the potential arise from time dependent biases in the wind speed.
6.5.3 Relationship to Major Modes of Atmospheric Variability It is now well recognised that atmospheric variability on a range of timescales may be characterised to a certain extent by various spatial patterns or modes typically expressed in terms of pressure on a given level. These modes have been determined primarily using statistical techniques, such as principal component analysis (Barnston and Livezey 1987) but have also been indexed in some cases via their expres-
174
S. A. Josey
sion in the surface pressure fields as the difference in pressure anomaly (i.e. actual value-long term mean) between two points (e.g. Hurrell 1995). The leading mode in the Atlantic is the North Atlantic Oscillation (NAO), characterised by variations in the pressure difference between the Azores High and Iceland Low. The NAO has been the subject of numerous studies documenting its influence on a range of oceanic, land and atmospheric physical processes, as well as its influence on ecosystems (see the comprehensive review of Hurrell et€al. 2003). Likewise, in the Tropical Pacific the El Nino-Southern Oscillation (ENSO) east-west pattern associated with variations in the strength of the Walker Cell has profound consequences for the ocean and neighbouring land masses. It too has been the subject of intensive research over many decades and received significant attention prior to the discovery of the NAO (Philander 1990). More recently, a north-south variation in the pressure difference between the Southern Ocean and Antarctic landmass has been dubbed the Southern Annular Mode (SAM). Attention here has focused on the strengthening of the SAM index over the past several decades and consequences of the associated southwards displacement of the main westerly wind belt over the Southern Ocean (e.g. Ciasto and Thompson 2008; Böning et€al. 2008). Mode-associated variations in the surface pressure gradient naturally lead to changes in the strength and direction of the wind field, and the source region for the air mass advected over a particular region of ocean (and thus its temperature and humidity characteristics). As discussed previously (Sect.€6.2.2, Eqs.€6.1 and 6.2), the wind speed and near surface air temperature and humidity are the primary variables which establish the strength of the latent and sensible heat loss, hence the leading modes of variability have a clear signature in the surface heat flux (e.g. Josey et€al. 2001 for the NAO). The air temperature and humidity also impact the longwave flux (Eq. 6.6), and the change in air mass characteristics can also lead to variations in cloud amount, thus the modes may also impact on both radiative flux terms. As an example of mode impacts on the wind speed and net surface heat flux, these fields are shown for the two leading modes of variability in the North Atlantic, the aforementioned NAO, and the second mode which is widely termed the East Atlantic Pattern (EAP), in Fig.€6.8. The NAO exhibits the well known north-south dipole in sea level pressure which results in stronger than normal winds from the north-west over the Labrador Sea and heat flux anomalies of up to −80€W€m−2 in this region for a unit positive value of the NAO index. Other notable features include enhanced flow of air from the south-east over the Gulf Steam which additional analysis shows to be anomalously warm, reducing the heat loss in this region. The EAP is characterised by a monopole structure in sea level pressure with lower than normal values in the East Atlantic at about 50–55°N. This gives rise to anomalously strong northerly winds in the mid-high latitude western Atlantic and strong heat loss at 45–50°N. Other features may be identified in both plots, and in general these are consistent with the increase in wind speed and change in air temperature expected from the anomalous wind direction. Note that in addition to the leading mode, there may be a further 3 or 4 modes which can be identified as being of importance for understanding the atmospheric variability and its impacts depending on the region considered.
6â•… Air-Sea Fluxes of Heat, Freshwater and Momentum
80
Sep-Mar NAO
80 60
70
40
60
80
175 Sep-Mar EAP
80 60
70
40
60
20
20 50
50 0 40
0 40 – 20
– 20 30
– 40
20
10 – 100 – 80 – 60 – 40 – 20
– 60
0
20
40
– 80
30
– 40
20
10 – 100 – 80 – 60 – 40 – 20
– 60
0
20
40
– 80
Fig. 6.8↜渀 Composites of the NCEP/NCAR reanalysis net heat flux (↜coloured field, units W€m−2), sea level pressure (contours, intervals 1€mb, negative values solid, zero and positive values dashed) and wind speed (↜arrows) on winter-centred Climate Prediction Center NAO and EAP values for the period 1958–2006
In addition to the net heat flux, the main modes of variability also have a direct impact on the freshwater flux as the change in the latent heat flux has an equivalent signature in the evaporation field, and variations in the evaporation result in modified precipitation downstream. For example, such variations in E-P associated with the NAO and EAP have been identified by Josey and Marsh (2005) and linked by them to changes in ocean surface salinity. These authors find that much of the multidecadal freshening in the eastern subpolar gyre region of the North Atlantic from the 1960s through to the 1990s can be attributed to a change in the strength of the East Atlantic Pattern (see also Myers et€al. 2007 for an extension of this work to the Labrador Sea). Variations in full depth ocean salinity are more difficult to relate to changes in the surface exchanges and this implies a leading role for advective effects (Boyer et€al. 2007). The combined effects of heat and freshwater flux anomalies lead to mode-related changes in the surface density flux field (via Eq.€6.10). Such changes have their greatest impact in dense water formation regions, for example at high latitudes in the North Atlantic. Here changes in the surface buoyancy loss associated with the NAO have lead to a multidecadal variation in the location of the dominant site for deep water formation from the Greenland Sea to the Labrador Sea as the NAO shifted from a primarily negative state in the 1960s to a positive state in the 1990s (Dickson et al. 1996, 2008). Finally, as regards mode impacts on the surface exchanges, changes in
176
S. A. Josey
the wind field have a direct impact on the wind stress (via Eq. 6.9) and thus the wind driven response of the ocean. See for example Josey et€al. (2002) who include an analysis of variations in the Ekman transport and wind driven upwelling associated with the NAO as part of a wider study of the wind stress forcing of the ocean. The brief discussion of mode impacts on high latitude buoyancy loss in the previous section, opens up a wider area, which will be only briefly touched on here, namely the dominant processes controlling dense water formation. Recent work has focused on both the wind- driven preconditioning for such events in the Nordic Seas (Gamiz-Fortis and Sutton 2007) and the role of heat loss (Grist et€al. 2007, 2008). Gamiz-Fortis and Sutton (2007) find that doming of isopycnals in response to wind stress curl anomalies and the consequent increase in surface density due to upwelling play a role in dense water formation. Grist et€al. (2007, 2008) have studied the impacts of heat flux extrema on Nordic Seas dense water formation and transport through the Denmark Strait in a range of coupled models. They find that heat flux extrema alone are sufficient to trigger new dense water production and find a consistent response across the model considered in terms of the response at the Denmark Strait. An increase in heat loss from −80 to −250 W€m−2 results in a strengthening of the dense water transport through the Strait of 1–2€Sv depending on the model considered. Other processes are also expected to play a significant role in the dense water formation, for example exchanges of water with fresher coastal boundary currents which are strongly influenced by Arctic outflows (for a full overview of this complex region see Dickson et€al. 2008).
6.6â•…Unresolved Issues and Conclusion There are many unresolved issues and areas for future improvement in the field of ocean-atmosphere interaction, including those surrounding the global heat budget closure problem, two particular examples follow.
6.6.1 The Southern Ocean Sampling Problem Observations that can be used to provide surface latent/sensible heat flux estimates are extremely sparse at high latitudes resulting in large uncertainties in the various flux products in the Southern Ocean. A primary factor here is the lack of the combined set of observations (wind speed, air temperature, surface humidity, sea surface temperature) necessary to estimate these flux terms. This is illustrated in Fig.€6.9 which shows all available surface meteorological reports from the COADS dataset with sufficient information to estimate the latent heat flux over the 5 year period from 2000–2004 in January and July. The situation is most severe in winter when we have essentially no information on this key field for assimilation into reanalyses or generation of in situ flux datasets.
Fig. 6.9↜渀 All available surface meteorological reports from the COADS dataset with sufficient information to estimate the latent heat flux over the 5 year period from 2000–2004 in July (↜left panel) and January (↜right panel)
6â•… Air-Sea Fluxes of Heat, Freshwater and Momentum 177
178
S. A. Josey 80
60
40
20
0
– 20
– 40
– 60
– 80
Fig. 6.10↜渀 Annual mean net heat flux (units W€m−2) from the ECMWF reanalysis for the period 1979–1993
There is a tendency to think of heat exchange in the Southern Ocean as being relatively uniform in a zonal sense when, at least according to the available reanalysis datasets, there is quite a significant amount of zonally asymmetric structure in the surface forcing. For example, the ECMWF annual mean net heat flux (Fig.€6.10) shows heat loss in the SE Pacific at 50–60°S of −20€W€m−2, which contrasts with a heat gain of 10–40€W€m−2 at the same latitudes in the Atlantic and Indian sectors of the Southern Ocean. How do we go about determining whether this zonal asymmetry is real with existing/future observing systems?
6.6.2 E stimating Meridional Overturning Circulation (MOC) Variability from Surface Fluxes The surface fluxes of heat and freshwater each act to modify the density of the ocean surface layer via their impact on temperature and salinity. Cooling of the ocean surface and net freshwater loss serve to increase the density as they result in
6â•… Air-Sea Fluxes of Heat, Freshwater and Momentum
179
a reduction in temperature and increase of salinity (the converse holds for ocean warming and freshwater gain). The combined effect of the heat and freshwater exchanges can be expressed in terms of the surface density flux (also referred to as the buoyancy flux). Variations in the density flux at high latitudes have potentially significant implications for European climate as they modify the amount of dense water formed in deep convection regions (Grist et€al. 2007, 2008) and consequently the overturning circulation of the North Atlantic. The impact of the air-sea density flux on the amount of water formed in different density classes can be determined using water mass transformation theory (Walin 1982) and these techniques have been employed in many model studies (e.g. Marsh et€al. 2005). A modification of this method has been recently used to estimate surface forced variability in the North Atlantic overturning circulation (Grist et€al. 2009; Josey et€al. 2009). The method has been shown to provide useful estimates of the MOC variability in the range 35–65°N with the HadCM3 coupled climate model and has been applied using NCEP/NCAR reanalysis flux fields to estimate surface forced variability in the mid-high latitude North Atlantic for the past 45 years. The variability of the MOC at latitude 55°N obtained using this technique is shown in Fig.€6.11. The figure reveals a tendency for an anomalously high overturning circulation, by about 1–2€Sv, from the late 1970s to the late 1990s. This period coincides with the prolonged positive phase of the North Atlantic Oscillation and may indicate that surface forcing associated with this mode plays a significant role in determining the strength of the circulation at this latitude. From 2000 onwards, there is some indication of weakening of the transport which probably reflects natural variability. Further work is planned to refine the method which has the potential to provide valuable complementary information on circulation variability at mid-high latitudes to that obtained from the Rapid mooring array at 26°N.
SFOC Anomaly (Sv)
4
55 N
2 0 –2 –4
1970
1980 Year
1990
2000
Fig. 6.11↜渀 Reconstruction of the maximum surface forced North Atlantic overturning circulation anomaly (units Sverdrup, 1€ Svâ•›=â•›106€ m3s−1) at 55°N using density fluxes determined from the NCEP/NCAR reanalysis. Details of the method are given in Josey et€al. (2009), the different lines are estimates based on surface flux fields integrated over 6 years (↜dash-dot line), 10 years (↜solid line) and 15 years (↜dashed line)
180
S. A. Josey
In conclusion, the main aim of this paper has been to provide an overview of the air-sea fluxes of heat, freshwater and momentum focusing on methods used to determine these fluxes and their role in the wider climate system. The intention being to provide a firm basis for future studies which seek to evaluate the importance of air-sea fluxes for operational oceanography. This is a rapidly developing field as highlighted by the other papers in this volume and at present the relative importance of surface fluxes as opposed to other processes in obtaining short range (i.e. up to 1 week) ocean forecasts is a matter of debate and will depend on the region and particular timescales being considered. It is to be expected that surface fluxes will prove key to obtaining reliable forecasts of, for example, ocean mixed-layer depth or density structure. Significant progress in this area is likely over the next few years and will benefit from evaluations of the accuracy of surface flux datasets (in particular from numerical weather prediction models) being carried out in a wider climate context beyond operational oceanography. Developments in the observing network, in particular the advent of Argo and the increasing number of surface flux reference sites, will enable such evaluations. An exciting recent development has been the deployment, for the first time, in March 2010 of a surface flux buoy in the Southern Ocean (http://imos.org.au/sofs.html). Such deployments in regions previously unsampled with high quality surface flux instrumentation promise major advances in our understanding of air-sea interaction processes and a better picture of how transfers across the ocean-atmosphere interface influence the climate system. Acknowledgements╇ The research summarised here is the result of efforts by a very broad community and I would like to thank the many people with whom I’ve discussed ocean-atmosphere interaction over the years. In particular, I would like to express my gratitude to Peter Taylor for guiding my thinking through much of my research career and to the UK Natural Environment Research Council for funding much of my research activity. In addition, I am grateful for many helpful comments on the manuscript by the anonymous reviewer and by members of the GODAE Summer School, in particular Cynthia Bluteau and Stephanie Downes.
References Adler RF, Huffman GJ, Chang A, Ferraro R, Xie P, Janowiak J, Rudolf B, Schneider U, Curtis S, Bolvin D, Gruber A, Susskind J, Arkin P (2003) The Version 2 Global Precipitation Climatology Project (GPCP) Monthly Precipitation Analysis (1979-Present). J Hydrometeorol 4:1147–1167 Andersson A, Fennig K, Klepp C, Bakan S, Graßl H, Schulz J (2010) The Hamburg ocean atmosphere parameters and fluxes from satellite data–HOAPS-3. Earth Syst Sci Data Discuss 3:143–194. doi:10.5194/essdd-3-143-2010 Barnston AG, Livezey RE (1987) Classification, seasonality and persistence of low-frequency atmospheric circulation patterns. Mon Wea Rev 115:1083–1126 Berry DI, Kent EC (2009) A new air-sea interaction gridded dataset from ICOADS with uncertainty estimates. Bull Am Meteor Soc 90:645–656. doi:10.1175/2008BAMS2639.1 Bignami F, Marullo S, Santoleri R, Schiano ME (1995) Longwave radiation budget in the Mediterranean Sea. J Geophys Res 100(C2):2501–2514
6â•… Air-Sea Fluxes of Heat, Freshwater and Momentum
181
Böning CW, Dispert A, Visbeck M, Rintoul SR, Schwarzkopf FU (2008) The response of the Antarctic circumpolar current to recent climate change. Nat Geosci 1:864–869 Boyer T, Levitus S, Antonov J, Locarnini R, Mishonov A, Garcia H, Josey SA (2007) Changes in freshwater content in the North Atlantic Ocean 1955–2006, Geophys Res Lett 34(16):L16603. doi:10.1029/2007GL030126 Bunker AF (1976) Computations of surface energy flux and annual air-sea interaction cycles of the North Atlantic Ocean. Mon Wea Rev 104:1122–1140 Ciasto LM, Thompson DWJ (2008) Observations of large-scale ocean-atmosphere interaction in the Southern Hemisphere. J Clim 21(6):1244–1259 Clark NE, Eber L, Laurs RM, Renner JA, Saur JFT (1974) Heat exchange between ocean and atmosphere in the eastern North Pacific for 1961–71. NOAA Tech Rep NMFS SSRF-682, U.S. Department of Commerce, Washington, p€108 da Silva AM, Young CC, Levitus S (1994) Anomalies of directly observed quantities, vol€2. Atlas of Surface Marine Data, NOAA Atlas NESDIS 2, p€419 Dickson R, Lazier J, Meincke J, Rhines P (1996) Long-term coordinated changes in the convective activity of the North Atlantic. In: Willebrand DAJ (ed) Decadal climate variability: dynamics and predictability. Springer, Berlin, p€211–262 Dickson R, Hansen B, Rhines P (eds) (2008) Arctic-Subarctic Ocean Fluxes (ASOF). Springer, Dordrecht Fairall CW, Bradley EF, Hare JE, Grachev AA, Edson JB (2003) Bulk parameterization of air-sea fluxes: updates and verification for the COARE algorithm. J Clim 16:571–591 Gamiz-Fortis SR, Sutton RT (2007) Quasi-periodic fluctuations in the Greenland-Iceland-Norwegian Seas region in a coupled climate model. Ocean Dyn 57:541–557 Gilman C, Garrett C (1994) Heat flux parameterizations for the Mediterranean Sea: the role of atmospheric aerosols and constraints from the water budget. J Geophys Res 99:5119–5134 Grist JP, Josey SA (2003) Inverse analysis adjustment of the SOC air–sea flux climatology using ocean heat transport constraints. J Clim 16:3274–3295 Grist JP, Josey SA, Sinha B (2007) Impact on the ocean of extreme Greenland sea heat loss in the HadCM3 coupled ocean-atmosphere model. J Geophys Res 112:C04014. doi:10.1029/2006JC003629 Grist JP, Josey SA, Sinha B, Blaker AT (2008) Response of the Denmark strait overflow to nordic seas heat loss. J Geophys Res 113:C09019. doi:10.1029/2007JC004625 Grist JP, Marsh RA, Josey SA (2009) On the relationship between the North Atlantic meridional overturning circulation and the surface-forced overturning stream function. J Clim 22(19):4989–5002. doi:10.1175/2009JCLI2574.1 Grist JP, Josey SA, Marsh R, Good SA, Coward AC, deCuevas BA, Alderson SG, New AL, Madec G (2010) The roles of surface heat flux and ocean heat transport convergence in determining Atlantic Ocean temperature variability. Ocean Dyn 60(4):771–790. doi: 10.1007/s10236-010-0292-4 Gulev S, Josey SA, Bourassa M, Breivik L-A, Cronin MF, Fairall C, Gille S, Kent EC, Lee CM, McPhaden MJ, Monteiro PMS, Schuster U, Smith SR, Trenberth KE, Wallace D, Woodruff SD (2010) Surface energy and CO2 fluxes in the Global Ocean-Atmosphere-Ice System. Plenary White Paper in Proceedings of the “OceanObs’09: Sustained Ocean Observations and Information for Society” Conference. ESA Publication WPP-306, Venice, Italy, 21–25 Sept 2009 Hadfield RE, Wells NC, Josey SA, JJ-M Hirschi (2007) On the accuracy of North Atlantic temperature and heat storage fields from Argo. J Geophys Res 112:C01009. doi:10.1029/2006JC003825 Harrison DE (1989) On climatological monthly mean wind stress and wind stress curl fields over the World Ocean. J Clim 2(1):57–70 Hellerman S, Rosenstein M (1983) Normal monthly wind stress over the World Ocean with error estimates. J Phys Oceanogr 13:1093–1104 Hurrell JW (1995) Decadal trends in the North Atlantic oscillation regional temperatures and precipitation. Science 269:676–679 Hurrell JW, Kushnir Y, Visbeck M, Ottersen G (2003) An overview of the North Atlantic oscillation. In: Hurrell JW, Kushnir Y, Ottersen G, Visbeck M (Eds) The North Atlantic oscillation:
182
S. A. Josey
climate significance and environmental impact. Geophysical Monograph Series. American Geophysical Union, Washington D.C., p€134 IPCC (2007) Climate change 2007: the physical science basis. Contribution of Working Group I to the 4th assessment report of the inter-governmental panel on climate change. Cambridge University Press, Cambridge Isemer H-J, Willebrand J, Hasse L (1989) Fine adjustment of large scale air-sea energy flux parameterizations by direct estimates of ocean heat transport. J Clim 2:1173–1184 Josey SA (2003) Changes in the heat and freshwater forcing of the eastern Mediterranean and their influence on deep water formation. J Geophys Res 108(C7):3237. doi:10.1029/2003JC001778 Josey SA, Arimoto N (1992) The colour gradient in M31: evidence for disc formation by biased infall? Astron Astrophys 255:105 Josey SA, Tayler RJ (1991) The oxygen yield and infall history of the solar neighbourhood. Mon Not R Astron Soc 251:474 Josey SA, Marsh R (2005) Surface freshwater flux variability and recent freshening of the North Atlantic in the eastern Subpolar Gyre. J Geophys Res 110:C05008. doi:10.1029/2004JC002521 Josey SA, Smith SR (2006) Guidelines for evaluation of Air-Sea heat, freshwater and momentum flux datasets, CLIVAR Global Synthesis and Observations Panel (GSOP) White Paper, July 2006, pp€14. http://www.clivar.org/organization/gsop/docs/gsopfg.pdf Josey SA, Kent EC, Taylor PK (1995) Seasonal variations between sampling and classical mean turbulent heat flux estimates in the eastern North Atlantic. Annal Geophys 13:1054–1064 Josey SA, Oakley D, Pascal RW (1997) On estimating the atmospheric longwave flux at the ocean surface from ship meteorological reports. J Geophys Res 102(C13):27,961–27,972 Josey SA, Kent EC, Taylor PK (1998) The Southampton Oceanography Centre (SOC) oceanatmosphere heat, momentum and freshwater flux atlas. Southampton Oceanography Centre Report No. 6, Southampton, UK, p€30 Josey SA, Kent EC, Taylor PK (1999) New insights into the ocean heat budget closure problem from analysis of the SOC air–sea flux climatology. J Clim 12:2856–2880 Josey SA, Kent EC, Sinha B (2001) Can a state of the art atmospheric general circulation model reproduce recent NAO related variability at the Air-Sea interface? Geophys Res Lett 28(24):4543–4546 Josey SA, Kent EC, Taylor PK (2002) On the wind stress forcing of the ocean in the SOC climatology: comparisons with the NCEP/NCAR, ECMWF, UWM/COADS and Hellerman and Rosenstein datasets. J Phys Oceanogr 32(7):1993–2019 Josey SA, Pascal RW, Taylor PK, Yelland MJ (2003) A new formula for determining the atmospheric longwave flux at the ocean surface at mid-high latitudes. J Geophys Res 108(C4). doi:10.1029/2002JC001418 Josey SA, Grist JP, Marsh RA (2009) Estimates of meridional overturning circulation variability in the North Atlantic from surface density flux fields. J Geophys Res—Oceans. 114:C09022. doi:10.1029/2008JC005230 Kent EC, Berry DI (2005) Quantifying random measurement errors in voluntary observing ships’ meteorological observations. Int J Climatol 25(7):843–856. doi:10.1002/joc.1167 Large W, Yeager S (2009) The global climatology of an interannually varying air-sea flux data set. Clim Dynamics. doi:10.1007/s00382-008-0441-3 Levitus S, Antonov JI, Boyer TP, Locarnini RA, Garcia HE, Mishonov VA (2009) Global ocean heat content 1955–2008 in light of recently revealed instrumentation problems. Geophys Res Lett 36:L07608. doi:10.1029/2008GL037155 Lozier MS, Leadbetter S, Williams RG, Roussenov V, Reed MSC, Moore NJ (2008) The spatial pattern and mechanisms of heat-content change in the North Atlantic. Science 319(5864):800– 803. doi:10.1126/science.1146436 Marsh R, Josey SA, Nurser AJG, de Cuevas BA, Coward AC (2005) Water mass transformation in the North Atlantic over 1985–2002 simulated in an eddy-permitting model. Ocean Sci 1:127–144 Marsh R, Josey S, de Cuevas B, Redbourn L, Quartly G (2008) Mechanisms for recent warming of the North Atlantic: insights with an eddy-permitting model. J Geophys Res 113:C04031
6â•… Air-Sea Fluxes of Heat, Freshwater and Momentum
183
Myers P, Josey S, Wheler B, Kulan N (2007) Interdecadal variability in labrador sea precipitation minus evaporation and salinity. Prog Oceanogr 73(3–4):341–357 Pascal RW, Josey SA (2000) Accurate radiometric measurement of the atmospheric longwave flux at the sea surface. J Atmos Oceanic Technol 17(9):1271–1282 Philander SGH (1990) El Nino, La Nina at the Southern Oscillation. Academic Press, San Diego Pierce DW, Barnett TP, AchutaRao KM, Gleckler PJ, Gregory JM, Washington WM (2006) Anthropogenic warming of the oceans: observations and model results. J Clim 19(10):1873–1900 Pinker RT, Wang H, Grodsky1 SA (2009) How good are ocean buoy observations of radiative fluxes? Geophys Res Lett 36:L10811. doi:10.1029/2009GL037840 Reed RK (1977) On estimating insolation over the ocean. J Phys Oceanogr 7:482–485 Renfrew IA, Moore GWK, Guest PS, Bumke K (2002) A comparison of surface-layer and surface turbulent-flux observations over the Labrador Sea with ECMWF analyses and NCEP reanalyses. J Phys Oceanogr 32:383–400 Rhines PB, Hakkinen S, Josey SA (2008) Is oceanic heat transport significant in the climate system? In: Dickson R, Hansen B, Rhines P (eds) Arctic-Subarctic Ocean Fluxes. Springer, Berlin, p.€87–110 Send U, Weller R, Wallace D, Chavez F, Lampitt R, Dickey T, Honda M, Nittis K, Lukas R, McPhaden M, Feely R (2009) OceanSITES. Community White Paper, Oceanobs’09 Smith SD (1988) Coefficients for sea surface wind stress, heat flux and wind profiles as a function of wind speed and temperature. J Geophys Res 93:15,467–15,474 Smith S et al (2010) The data management system for the shipboard automated meteorological and oceanographic system (SAMOS) initiative. Community White Paper in proceedings of the “OceanObs’09: Sustained Ocean Observations and Information for Society” Conference. ESA Publication WPP-306, Venice, Italy, 21–25 Sept 2009 Stammer D, Ueyoshi K, Köhl A, Large WB, Josey S, Wunsch C (2004) Estimating air-sea fluxes of heat, freshwater and momentum through global ocean data assimilation. J Geophys Res 109:C05023. doi:10.1029/2003JC002082 Stott PA, Sutton RT, Smith DM (2008) Detection and attribution of Atlantic salinity changes. Geophys Res Lett 35:L21702. doi:10.1029/2008GL035874 Taylor PK, Yelland MJ (2001) The dependence of sea surface roughness on the height and steepness of the waves. J Phys Oceanog 31:572–590 Taylor PK, Bradley EF, Fairall CW, Legler L, Schulz J, Weller RA, White GH (2001) Surface fluxes and surface reference sites. In: Koblinsky CJ, Smith NR (eds) Observing the Oceans in the 21st Century. GODAE Project Office/Bureau of Meteorology, Melbourne, p€177–197 Tomita H, Kubota M, Cronin MF, Iwasaki S, Konda M, Ichikawa H (2010) An assessment of surface heat fluxes from J-OFURO2 at the KEO/JKEO sites. J Geophys Res-Oceans 115:13 Trenberth KE, Caron JM (2001) Estimates of meridional atmosphere and ocean heat transports. J Clim 14:3433–3443 Trenberth KE, Dole R, Xue Y, Onogi K, Dee D, Balmaseda M, Bosilovich M, Schubert S, Large W (2009) Atmospheric reanalyses: a major resource for ocean product development and modeling. Community White Paper, Oceanobs’09 Walin G (1982) On the relation between sea-surface heat flow and thermal circulation in the ocean. Tellus 34:187–195 Weller RA, Bradley EF, Edson JB, Fairall CW, Brooks I, Yelland MJ, Pascal RW (2008) Sensors for physical fluxes at the sea surface: energy, heat, water, salt. Ocean Sci 4:247–263 Wells NC, Josey SA, Hadfield RE (2009) Towards closure of regional heat budgets in the North Atlantic using Argo floats and surface flux datasets Ocean Sci 59–72. SRef-ID:1812-0792/ os/2009-5-59 WGASF (2000) Intercomparison and validation of ocean-atmosphere energy flux fields—Final report of the Joint WCRP/SCOR Working Group on Air–Sea Fluxes(WGASF) In: Taylor PK (ed) WCRP-112, WMO/TD-1036, World Climate Research Programme, Geneva. p€306 Woodruff SD, Slutz RJ, Jenne RL, Steurer PM (1987) A comprehensive ocean-atmosphere data set. Bull Am Meteor Soc 68:1239–1250
184
S. A. Josey
Worley SJ, Woodruff SD, Reynolds RW, Lubker SJ, Lott N (2005) ICOADS release 2.1 data and products. Int J Climatol 25:823–842 Worley SJ, Woodruff SD, Lubker SJ, Ji Z, Freeman JE, Kent EC, Brohan P, Berry DI, Smith SR, Wilkinson C, Reynolds RW (2009) The role of ICOADS in the sustained ocean Observing System. Community White Paper, Oceanobs’09 Yu L, Weller RA (2007) Objectively analyzed air-sea flux fields for the global ice-free oceans (1981–2005). Bull Am Meteor Soc 88:527–539
Chapter 7
Coastal Tide Gauge Observations: Dynamic Processes Present in the Fremantle Record Charitha Pattiaratchi
Abstract╇ Coastal sea level variability occurs over timescales ranging from hours to centuries. Globally, the astronomical forces of the Sun and the Moon are the dominant forcing which results in the tidal variability with periods of 12 and 24€h. In many regions, the effects of the tides dominate the water level variability – however, in regions where the tidal effects are small other processes also become important in determining the local water level. In this paper, sea level data from Fremantle (tidal range ~0.5€m), which has one of the longest time series records in the southern hemisphere, and other sea level recoding stations from Western Australia are presented to highlight the different processes ranging from seiches, tsunamis, tides, storm surges, continental shelf waves, annual and inter-annual variability. As the contribution from each of these processes is of the same order of magnitude – the study of sea level variability in the region is very interesting and reveals both local and remote forcing.
7.1╅Introduction Coastal regions experience rise and fall of sea level which vary at timescales of hours, days, weeks, months, annually and so on, governed by the astronomical tides, meteorological conditions, local bathymetry and a host of other factors. An overview of these processes may be found in Pugh (1987, 2004) and Boon (2004). Globally, the astronomical forces of the Sun and the Moon are the dominant forcing which results in the tidal variability with periods of 12 and 24€h. In many regions, the effects of these tides dominate the water level variability; however, in regions where the tidal effects are small other processes become important in determining the local water level. In this paper, sea level data from Fremantle (Fig.€7.1) which C. Pattiaratchi () School of Environmental Systems Engineering & UWA Oceans Institute, The University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia e-mail:
[email protected] A. Schiller, G. B. Brassington (eds.), Operational Oceanography in the 21st Century, DOI 10.1007/978-94-007-0332-2_7, ©Â€Springer Science+Business Media B.V. 2011
185
186
C. Pattiaratchi
Fig. 7.1↜渀 Location of tide gauges used in the present study and the track of the tropical cyclone Frank
has one of the longest time series records in the southern hemisphere, are presented to highlight the different processes ranging from seiches, tsunamis, tides, storm surges, continental shelf waves, annual and inter-annual variability (Table€7.1). It should be noted that there are some processes which are not inherent in the Fremantle record but may be present in other tide gauge records are not included in this paper. These include storm surges (due to local changes in atmospheric pressure and winds), sea level changes due to ocean eddy interactions with the coast and wave set-up. In Fremantle, it is difficult to separate the surge effects due to local and remote forcing (Eliot and Pattiaratchi 2010) and thus are included in the section ‘continental shelf waves’. The auto spectrum of water levels recorded at Fremantle over three years indicated several peaks, ranging from hours to seasonal timescales Table 7.1↜渀 Decomposition of processes observable at Freemantle tide gauge and their approximate amplitudes Process Duration Scale (m) Reference Wave action 2–20€s ~5 Lemm et€al. (1999) Wave set-up 5–30€min ~0.3 Bode and Hardy (1997) Seiches 30–90€min ~0.2 Ilich (2006) ~0.2 Reid (1990) Pressure surge 1–3€h ~0.2 Pugh (1987) Wind set-up 3–6€h ~0.8 Easton (1970) Tidal conditions 12–24€h * Pattiaratchi et€al. (1997) Sea breezes 24€h Pressure systems (cycle) 1–10 days ~0.8 Hamon (1966) Continental shelf waves 3–10 days ~0.6 Fandry et€al. (1984) Pattiaratchi and Buchan (1991) Oceanic currents Seasonal ~0.3 Nodal tide 18.6 years ~0.15 Pugh (1987) Climate variability Decades ~0.2 Pariwono et€al. (1986) Climate change 103+ years ~10 Wyrwoll et€al. (1995)
7â•… Coastal Tide Gauge Observations
187
Weather Band
109
Tides
Spectral Density (m2 / Hz)
108
Seiches
107
Ocean Currents
106 105
20d 20m
5d 104 103
1 year
24h
102
160m 12h
10–8
10–7
10–6 10–5 Frequency (Hz)
60m 10–4
10–3
Fig. 7.2↜渀 Spectra of water levels at Fremantle showing the different scales of variability
reflecting these processes (Fig.€7.2). The contribution from each of these processes, which includes both direct and remote forcing to the total sea level variability is of the same order of magnitude and thus is equally important. Sea level variability is important for a range of activities including navigation, coastal stability and coastal planning. The significance of coastal sea level change for coastal management has been recognised, effective for both gradual change and intermittent fluctuations (Komar and Enfield 1987; Allan et€al. 2003). In order to interpret historic patterns of coastal management and predict possible future needs, it is necessary to document both short and long-term trends and fluctuations of sea level.
7.1.1 The Study Region Fremantle is located along the western-coast of Australia at latitude 32°S (Fig.€7.1). Weather systems impacting on the region are dominated by anti-cyclonic high-pressure systems with periodic tropic and mid-latitude depressions and local seasonal sea-breezes (Eliot and Clarke 1986). Anticyclones move to the east and pass the coast every 3–10 days (Gentilli 1972). The peak occurrence of mid-latitude depressions is in July and the strongest winds in the system are the north-westerlies (Gentilli 1972; Lemm et€al. 1999). Tropical cyclones track down from the Northwest
188
C. Pattiaratchi
coast infrequently during late summer and can have significant impact on the coastline (Eliot and Clarke 1986). The seasonal movement of the high-pressure systems results in a strong seasonality in the wind regime. During the summer southerly winds prevail whilst in winter there is no dominant wind direction although the strongest winds are north-westerly during the passage of frontal systems. Sea breezes, which are stronger during the summer dominate the coastal region with offshore (westward) winds in the morning and strong (up to 15€ms−1) shore parallel sea breezes commencing around noon and weakening during the night (Pattiaratchi et€al. 1997; Masselink and Pattiaratchi 2001). During winter, the region is subject to the passage of mid-latitude depressions and associated frontal systems, and ~30 storm wave events are experienced (Lemm et€al. 1999). During the passage of a frontal system, the region is subject to strong winds (up to 25–30€ms–1) from the north through west, which rapidly change direction towards west through southwest then progressively more southerly over 12–16€h. South to south-westerly winds gradually weakens over two to three days, and calm, cloud-free conditions prevail for another three to five days prior to the passage of another frontal system.
7.2╅Data Data presented here was recorded at the long-term tide station located at Fremantle and maintained by the WA Department for Planning and Infrastructure. The sampling intervals vary between 2€min and 1€h. In addition, monthly mean data from the same gauge was obtained from the Permanent Service for Mean Sea Level, located at Proudman Oceanographic Laboratory, Liverpool, UK (www.pol.ac.uk/psmsl/).
7.3â•…Seiches A free oscillation in an enclosed or semi-enclosed body of water, similar to the oscillation of a pendulum where the oscillation continues after the initial force has stopped, is defined as a seiche (Miles 1974). Several factors cause the initial displacement of water from a level surface, and the restoring force is gravity, which tends to maintain a level surface. Once formed, the oscillations are characteristic only of the system’s geometry (length and depth) and may persist for many cycles before decaying under the influence of friction or energy leakage. A simplest model of a continental shelf seiche is a standing wave with an antinode at the shoreline and a node at the shelf edge. The period of the seiche is given by four times the travel time from the coast to the shelf-edge. For a mean water depth h, shallow water wave theory gives (Pugh 1987):
Tn =
4L 1 √ (2n − 1) gh
(7.1)
7â•… Coastal Tide Gauge Observations
189
Water Level (cm)
Wind Stress (Pa)
Here, n is the number of nodes (nâ•›=â•›1 is the fundamental mode and is also the most common; L is the width of the continental shelf; g is acceleration due to gravity and h is the mean water depth. In the auto-spectrum (Fig.€7.2), 3 seiche periods were identified at Fremantle: 2.8€h, 1€h and 20€min. The seiches have amplitudes between 10 and 40€cm and contained 40–70% energy relative to the main 24€h diurnal tidal oscillation. Ilich (2006) found that the maximum amplitudes of the 2.8€h and 20€min seiches to be ~45€cm and ~12€cm respectively although the 45€cm seiche amplitude could be due to the superposition of all three seiches. Width of the continental shelf off Fremantle is ~50€km whilst the mean shelf depth is ~50€m which yields from Eq.€(7.1), a period of ~2.5€h which is close to the observed value of 2.8€h. Ilich (2006) found that changes in direction of the wind stress initiate seiching (Fig.€7.3). In particular: (a) strong wind events onshore or offshore components initiate seiching at the 1€h and to a lesser extent, the 20€min seiche; (b) strong southerly (shore-parallel) events rarely cause excitation; (c) sea breeze patterns occurring for more than two days decrease spectral energy of the entire spectrum. Continental shelf seiches are also generated by tsunamis (Pattiaratchi and Wijeratne 2009) and are discussed in Sect.€7.4.
0.5 0 –0.5 a 320
325
E (-ve) / W (+ve)
330
335
150 100 50
b
0 320 –7
Frequency (In(Hz))
N (-ve) / S (+ve)
–7.5 –8
325 330 Energy plot of a spectrum of frequencies over time series WL from FBH 2001
335
25mins
1hr
–8.5 –9 –9.5
2.8hr
c
–10 320
325
Time (day of the year)
330
335
Fig. 7.3↜渀 Time series of a wind stress; b water level; and, c time-frequency diagram for a 15 day period in November 2001 showing that onshore winds (-ve easterly) initiates seiching. (From Ilich 2006)
190
C. Pattiaratchi
7.4â•…Tsunamis
Residual Water Level (cm)
The Indian Ocean region experienced its most devastating natural disaster through the action of a tsunami, resulting from an earthquake off the coast of Sumatra, on 26th of December 2004. This was followed by tsunamis in March 2005, June 2006 and July 2007 and tide gauges in Western Australia recorded sea level oscillations related to all 4 tsunamis but did not result in large scale property damage (Pattiaratchi and Wijeratne 2009). The tide gauge data along the west coast indicated that the tsunami waves incident at Geraldton (0720), Carnarvon (0740) and Fremantle (0740). The initial waves all indicated an increase in the water level, corresponding to leading elevation waves, and the heights along the west coast ranged from 0.33€m at Fremantle to 1.650€m at Geraldton (Fig.€7.4 and Table€7.2). Examination of the residual time series, maximum wave heights, and the elapsed time between the initial and maximum waves indicated that: (1) the maximum wave heights recorded at Carnarvon, Geraldton, and Fremantle (Table€7.2), all exceeded the mean spring tidal range at these locations; (2) at Geraldton, although initial oscillations due to the tsunami waves were observed at 0720 UTC, there was a lag of
Carnarvon 200
No Data
Geraldton
100 0 26
Fremantle
26.5
27 27.5 28 Local Time (Days in December 2004)
28.5
29
Fig. 7.4↜渀 Time series of residual sea level from coastal stations located along the west coast of Australia. The dashed line shows the time of the earthquake. (Note: local time is +8€h UTC)
Table 7.2↜渀 Characteristics of the 26 December 2004 tsunami as recorded by tide gauges Station Initial wave Maximum wave Arrival time/date Wave heighta Elapsed time Wave height (UTC) (number) Carnarvon 07:40 26/12/04 0.38€m 15€h 20€m (25) 1.14€m 15€h 15€m (19) 1.65€m Geraldton 07:20 26/12/04 0.13€m Fremantle 07:40 26/12/04 0.33€m 7€h 20€m (9) 0.60€m a Maximum wave height is listed as the trough to crest height
7â•… Coastal Tide Gauge Observations
191
five hours before the highest water level (2.6€m relative to datum) was reached at 1210 GMT, which coincided with the tidal high water (Fig.€7.4). However, the highest waves (trough to crest) were recorded ~10€h later and were associated with a wave group (see Fig.€7.4 below). The water levels recorded at Geraldton during this event were the highest and lowest levels recorded at this station, which has been in continuous operation for more than 40 years; (3) The residual time series indicated the arrival of a group of waves with higher wave heights at Geraldton some 13–15€h after the arrival of the initial wave (Table€7.2) suggests a reflected wave from the island of Madagascar or the Mascarene ridge (Pattiaratchi and Wijeratne 2009); and, (4) the tsunami set-up seiching along the continental shelf with periods of 4 and 2.7€h at Geraldton and Fremantle, respectively (Fig.€7.4). These periods were the same as those excited by the meteorological effects (Sect.€7.3).
7.5â•…Tides Periodic movements which are directly related in amplitude and phase to some periodic geophysical force are defined a tides and astronomic tides are the most widely recognised phenomena affecting water levels (Pugh 1987). These tides are the harmonic fluctuations of water level developed through the gravitational attraction from astronomic bodies (mainly the sun and moon). In majority of the world’s coastlines there are two tidal cycles per day (i.e. two high and low waters per day and these are termed semi-diurnal (twice-daily) tides with a tidal period of 12.24€h. In a few locations (e.g. Gulf of Mexico, Gulf of Thailand), there is only one high and low waters per day and are known as diurnal (daily) tides and have periods ~24€h. Spring tides are periods of increased tidal range and occur when the Earth, the Sun, and the Moon are along the same axis such that the gravitational forces of the Moon and the Sun both contribute to the tides. Spring tides occur immediately after full moon and the new moon. Neap tides periods of low tidal range which occur when the gravitational forces of the Moon and the Sun are perpendicular to one another (with respect to the Earth). Neap tides occur during 1st and 3rd quarter of the moons. As the tide generating forces are related to the periodic gravitational forces of the Moon and the Sun, there are specific periods which may be identified from the equilibrium tide (Pugh 1987). For example, these include the 12.24€h and the 24€h for the main semi-diurnal and diurnal periods; the lunar month (29.5 days); the period between two successive full or new moons; the annual cycle due to the changes in the earth’s orbit around the Sun. In the longer term, changes in the orbits of the moon and the sun provide 4.45 year and 18.6 year variations in the tides and these are discussed in Sect.€7.9. The dynamic theory of tides which governs the tidal characteristics of the ocean basins considers the configuration of the ocean basins (width, length, and depth),
192
C. Pattiaratchi
frictional forces, Coriolis force, convergence and resonance, and many other variables (Boon 2004). As a result, tides are considered as a series of Amphidromic systems consisting of rotating (Kelvin) waves which rotate around a point where the amplitude of the tide is zero, defined as an Amphidromic point. Due to the influence of the Coriolis force, the Amphidromic systems rotate clockwise in the southern hemisphere and counterclockwise in the northern hemisphere. Close to an Amphidromic point the tidal range is zero and the range increases away from the point (Boon 2004). The tides at Fremantle, which generally are representative of the tides experienced along south-western Australia, are classified as diurnal (Ranasinghe and Pattiaratchi 2000). This is due to the location of a semi-diurnal amphidromic system close to the coast and the diurnal amphidromic system located off the coast of South Africa. The four largest tidal constituents (Pugh 1987) are associated with diurnal and semi-diurnal effects of the sun and moon (Fig.€7.5; Table€7.3). Along south-western Australia, the tide’s diurnal component has a range of 0.6€m, and the semidiurnal tide has a range of only 0.2€m. The semidiurnal tidal range is related to the lunar cycle, with the maximum tidal range occurring close
Sea Level (m)
2 1.5 1 0.5
a
0 1996
1996.1
2
1996.2
1996.3
Solstice
4
1996.4
1996.5
1996.6
Equnox
1996.7
1996.8
Solstice
1996.9
1997
Equnox
Period (hours)
8
Semi-diurnal
16
Diurnal
32 64 128 1996
b
1996.1
1996.2
1996.3
1996.4 1996.5 Time (year)
1996.6
1996.7
1996.8
1996.9
1997
Fig. 7.5↜渀 a Time series of water level record at Fremantle for 1996 and b Morlet wavelet analysis of Fremantle tide record showing the alignment of diurnal and semi-diurnal energy during the equinox and out of phase during the solstice Table 7.3↜渀 Principal tidal constituents for Fremantle Constituent Amplitude Period 0.165€m 23.93€h K1 O1 0.118€m 25.82€h M2 0.052€m 12.42€h 0.047€m 12.00€h S2
Description Principal Lunar Diurnal Principal Solar Diurnal Principal Lunar Semi-diurnal Principal Solar Semi-diurnal
7â•… Coastal Tide Gauge Observations
193
to the full and new moons, and minimum tidal ranges occurring close to the lunar cycle’s first and last quarter’s—the spring–neap cycle (see above). Diurnal tides are related to the declination angle of the moon’s orbital plane. Therefore, the terminology of spring and neap tides is inaccurate in diurnal systems and are defined as tropic and equatorial tides. For tropic tides (analogous to spring tides for semidiurnal systems) the tidal range is a maximum when the declination of the moon is a maximum north or south of the equator. For equatorial tides (analogous to neap tides for semi-diurnal systems) the moon is directly above the equator resulting in a low tidal range. The diurnal and semidiurnal tides oscillate at a frequency of 13.63 and 14.77 days, respectively. This phase difference, of 1.14 days, between the two tidal signals modulates the resultant tide over an annual cycle, causing the diurnal and semidiurnal tides that are in phase during the solstice (resulting in a maximum tropic tidal range) and out of phase at the equinox (resulting in a minimum tropic tidal range). This process is illustrated in Figs.€7.5 and 7.6. This means the highest tropic tidal range does not always correspond with the full/new moon cycle with the daily tidal range varying biannually, with solstice tidal peaks (December–January and June–July) producing a tidal range that is about 20% higher than during equinoctial troughs (February–March and September–October).
Fig. 7.6↜渀 a Diurnal and semi-diurnal components of the tide at Fremantle from day 273 (October 1) to day 365 (December 31) in 2001; b water level from the summation of the diurnal and semidiurnal constituents. The moon phases are shown at the top of the Figure. (From O’Callaghan et€al. 2010)
194
C. Pattiaratchi
During the solstice, when the diurnal and semidiurnal tides are in phase, the maximum tidal range corresponds with the full/new moon cycle; during the equinox, the maximum tidal range does not correspond with the full/new moon cycle. Mixed tides occur during equatorial tides closest to the equinox, with two high and low waters commonly observed over a tidal cycle. Hence, in a diurnal tidal system, such as along south-west Australia, definitions such as spring and neap tides do not always relate to phases of the moon, as is the case for semidiurnal tides. Another consequence of the diurnal tides is the seasonal change in the times of high/low water. During the summer, along the south-west Australian coast, low water generally occurs between 4 a.m. and 12 p.m., depending on the phase of the moon, with high water in the evening. As summer progresses, the low water occurs earlier; as winter starts, the low water occurs later at night, becoming progressively earlier in the evening (with high water occurring in the morning).
7.6â•…Coastal-Trapped Waves The power spectra of sea level (Fig.€7.2) indicates a broad peak in energy in the ‘weather’ band (5–20 days) and these are generally due to atmospheric effects. Closer examination and comparison of the tidal residuals with local meteorological data revealed that a number of significant tidal residuals that were not fully explained by local synoptic conditions but was a combination of locally generated and remotely generated signals, the former through local changes in atmospheric pressure and local wind. The remote signal is characteristic of a long period coastally trapped shelf wave, travelling anti-clockwise relative to the Australian coast. A coastally trapped wave is defined as a wave that travels parallel to the coast, with maximum amplitude at the coast and decreasing offshore. Examples of these waves include continental shelf waves (CSWs) and internal Kelvin waves (Le Blond and Mysak 1978), which are governed through vorticity conservation (Huyer 1990). Coastally trapped waves need a shallowing interface and may develop a range of modes according to the shelf structure (Tang and Grimshaw 1995). They travel with the coast to the left (right) in the southern (northern) hemisphere. Along the Australian coast, shelf waves propagate anti-clockwise relative to the landmass. The governing equations (neglecting advection and friction) are (Huyer 1990):
∂u dη = −g + fv ∂t dx
(7.2)
∂v dη = −g − fu ∂t dy
(7.3)
where, u and v are the velocities in the x (east) and y (north) directions; η is the displacement of the sea surface and f is the Coriolis parameter. The solutions for
7â•… Coastal Tide Gauge Observations
195
Eqs.€7.2 and 7.3 (together with the continuity equation and appropriate boundary conditions), along a boundary oriented east-west are given by (Huyer 1990):
√ gh
η = ηo e−fy/
cos (kx − ωt)
(7.4)
where, ηo is the maximum amplitude at the shoreline, h is the water depth, k and are the wave number and frequency, respectively. This is an equation of a Kelvin wave, propagating along the coastal boundary, with the wave signal reducing in amplitude exponentially with distance offshore. Continental shelf waves (CSWs) depend on only the cross-shelf bathymetry profile and the vertical density profile controls the structure of an internal Kelvin wave (Huyer 1990). The alongshore component of wind stress usually generates CSWs, which are active along the Western Australian coast, first reported by Hamon (1966). Provis and Radok (1979) demonstrated that these waves propagate anti-clockwise along the south coast of the Australian continent over a maximum distance of 4000€km at speeds of 5–7€ms−1 (see also Eliot and Pattiaratchi 2010). Along the west Australian coastline, the continental shelf waves are generated through the passage of mid-latitude low-pressure systems and tropical cyclones. The continental shelf waves can be identified from the sea level records by low-pass filtering (i.e. removal of the tidal component). An example is shown on Fig.€7.7 for tidal records from Geraldton, Fremantle and Albany (Fig.€7.1). Several CSWs with amplitudes ranging from 0.1 to 0.5€m can be identified. For example, between days 290 and 295, an increase of ~0.5€m in the sub-tidal water level was observed at Geraldton. The same variation in water level signal was seen at Fremantle and Albany,
Fig. 7.7↜渀 Low-frequency water levels at a Geraldton, b Fremantle, and c Albany for days 275–365 in 2001 showing the presence of continental shelf waves. (From O’Callaghan et€al. 2007)
196
C. Pattiaratchi
and could be attributed to the passage of a CSW. The correlation coefficients between sub-tidal water levels at these three locations were all greater than 0.8, despite observations being several hundred kilometers apart. The propagation time of the CSW between Geraldton and Fremantle was 23€h, and between Fremantle and Albany it was 17€h, yielding a mean propagation speed of ~500€km day–1 (~6€ms−1). The period of the continental shelf wave range between 3–10 days and corresponds to the passage of synoptic systems from west to east across the west Australian coastline. Tropical cyclones are intense low pressure systems which form over warm ocean waters at low latitudes and are associated with strong winds, torrential rain and storm surges (in coastal areas). They may cause extensive damage as a result of strong winds and flooding (caused by either heavy rainfall and/or coastal storm surges). The impacts of tropical cyclones on the North-West region of Australia are well known with several severe cyclones impacting this region over the past few years. The most noticeable impacts of these cyclones are normally restricted to the region of impact of the cyclone, and hence the direct effect of cyclones on southwestern Australia is rare. Fandry et€al. (1984) identified 1 to 2€m amplitude peaks in sea level propagating southwards with speeds ranging between 400–600€km€day−1. These were associated with tropical cyclones travelling southward and were attributed to a resonance phenomenon when speeds of the southward component of the cyclone speeds were close to the southward propagating continental shelf wave. Sea level records at Fremantle indicate remote forcing due to tropical cyclones. Comparison between the low frequency component of sea level records along the west and south coasts of Western Australia with the occurrence of tropical cyclones in the North-West shelf region has revealed that every tropical cyclone, irrespective of its severity and path, generated a southward propagating sea level signal or a continental shelf wave (Eliot and Pattiaratchi 2010). The wave can be identified in the coastal sea level records, initially as a decrease in water level, 1–2 days after the passage of the cyclone and has a period of about 10 days. As an example, water level record at Fremantle for the period 1–19 December 1995 is shown on Fig.€7.8. Tropical cyclone Frank was declared as a category 1 cyclone on 7 December and
Fig. 7.8↜渀 Sea level record at Fremantle (↜thin black line) during December 1995 showing the lowfrequency water level variation (↜thick-line) induced by Tropical Cyclone Frank
7â•… Coastal Tide Gauge Observations
197
developed into a category 4 cyclone by 11 December and crossed the coastline near Carnarvon on 12 December. The evidence of the continental shelf wave becomes evident on 8 December when the water level starts to decrease and reaches a minimum level on 10 December and a maximum peak on 14 December. The wave height (trough to crest) was 0.55€m, higher than the tidal range during this time (Fig.€7.8).
7.7â•…Seasonal Changes Mean sea level varies in an annual cycle averaging 0.22€m with water levels reaching a maximum in May–June and minimum October–November (Fig.€ 7.9). This variation is attributed to changes in the strength of the major ocean current in the region, the Leeuwin Current (Thompson 1984; Pattiaratchi and Buchan 1991; Feng et€al. 2004). The Leeuwin Current is a shallow (
λ π
(8.22)
or the water depth needs to be greater than about a third of the wavelength for the deep water approximation to apply. Typical swell waves in the ocean with periods of about 8€sec, have wavelengths of about 100€m, so these will be considered deep water waves right up until a depth of about 30€m, i.e. the waves will only start to feel the bottom when they are in water of less than 30€m depth. With the typical coarse spatial resolutions of global wave forecasting models (see Sect.€8.5) there are very few grid points that are in depths of 30€m or less, so it is often a reasonable approach to run global systems with deep water physics only. Now considering the shallow water approximation, we see that yâ•›=â•›tanh x is very close to the yâ•›=â•›x line for values of x less than about 0.5, in fact tanh (0.45) ≈ 0.422. Again, if we think that this is a tolerable approximation, then we can say that our shallow water approximation holds when kH < 0.45, or
H < 0.07λ
(8.23)
In other words, for the waves to behave as purely shallow water waves, the water depth needs to be less than 7% of the wavelength. Our 100€m long swell waves will thus only become purely shallow water waves when the water depth is less than 7€m. On top of this, wavelengths become shorter in shallow water, moving the shallow water limit for swell with deep water wavelengths of 100€m to even shallower water. An important point to note here is that the definitions for “deep water” and “shallow water” are actually defined as relationships between the wave and the water depth, rather than as an absolute value of the water depth, so there is no specific depth at which the water can be called either “deep” or “shallow”. For example, the wavelength of a tsunami is related to the width of the rupture of the earthquake that generated it. This is typically of order 100€km wide. Therefore, tsunamis will act as shallow water waves when the water depth is less than 7% of 100€km which is 7000€m. Almost all of the global ocean is shallower than this, so this is why tsunamis are considered to be shallow water waves.
210
D. Greenslade and H. Tolman
8.3.1 Phase Velocity and Group Velocity Some interesting features of wave propagation can be easily derived from the deep and shallow water approximations to the dispersion relation. The phase speed of a wave is simply the speed of propagation of the wave crest. The definition of the period (↜T) of a wave is the time taken for successive crests of a wave to pass a fixed point, thus a wave will move a distance in time T and so the phase speed (↜cp) is
cp =
λ ω = T k
(8.24)
For a disturbance represented by a number of different sinusoidal waves, the group velocity describes the velocity at which the energy of the group of waves is propagating. This can be shown to be (e.g. Holthuijsen 2007; Young 1999)
cg =
dω dk
(8.25)
So we see that in deep water, from Eq.€(8.18):
cp =
g 1 and cg = k 2
g k
(8.26)
while in shallow water (Eq.€8.20):
cp =
gH and cg = gH
(8.27)
Equation€(8.26) says that in deep water, the individual waves are propagating at twice the speed of the energy that they are carrying. This is an intriguing concept and it can be seen quite easily in nature. If you throw a small stone in a puddle, providing the puddle is deep enough, you will see a group of ripples propagating outwards obeying the deep water dispersion relation. As the ripples propagate away from the disturbance, you will see that individual waves appear at the back of the group, move forwards through the group and then disappear as they get to the front of the group. Equation€(8.26) also shows that the speed of propagation of the waves is related to the wavenumber, so waves of different wavelengths will propagate at different speeds. For a disturbance composed of waves of a number of different frequencies (or wavelengths), as they propagate away from the area of disturbance the longer waves will travel faster than the shorter waves and thus the wave energy will disperse. This is where the term dispersion relation comes from. Equation€(8.27) says that in shallow water, the individual waves propagate at the same speed as the wave energy and this speed is dependent only on the water depth. Thus waves of all wavelengths will travel at the same speed and shallow water waves are therefore non-dispersive.
8â•… Surface Waves
211
In addition to these interesting features of wave propagation, further useful properties of the motion can be derived from Eq.€(8.14). For example, it can be shown that the trajectories of the fluid particles (defined by u and w) describe circles in deep water and ellipses in shallow water. These are often referred to as the orbital velocities of the waves. The derivation is not shown here, but details can be found in Young (1999), Holthuijsen (2007) or Kundu (1990).
8.4â•…Basic Definitions The analysis above is mainly concerned with the very simple situation where we consider just one sinusoidal wave component. We have seen that it is possible to derive some readily seen characteristics of the ocean surface with the various assumptions, however, it is clear that this is not a valid description of the actual ocean surface. A more appropriate description is that the sea-surface is characterised as the superposition of a large number of sinusoidal components, with each of these sinusoidal components behaving as described in the previous section. Figure 8.4 shows an example with five sinusoidal components. Each of these components has a different frequency and a different amplitude and they sum together to produce the more complex sea-surface elevation depicted at the bottom. This is again just in one dimension but it can easily be extended to two dimensions by considering a range of different wave directions as well. Thus, the sea-surface elevation in general can be described by
η(t) =
N i=1
(8.28)
ai sin (ωi t + φi )
where ai , ωi and φi represent the amplitude, frequency and phase of the ith wave component, respectively.
+ + + +
=
Fig. 8.4↜渀 Representation of a 1-D ocean surface as a sum of 5 sinusoidal components
212
D. Greenslade and H. Tolman
8.4.1 The Wave Spectrum Consider the variance of the sea-surface elevation. This is, by definition, the mean of the square of the surface elevation, and so, assuming the mean of η is zero:
variance = σ 2 =
N 1 2 a 2N i=1 i
(8.29)
We can also consider how this variance is distributed over the different frequencies present in the wave fields, i.e., over the frequency interval fi . This gives us the variance density spectrum:
ai2 2fi
(8.30)
ai2 f →0 2f
(8.31)
F (fi ) =
which becomes, in the limit F (f ) = lim
or
2
σ =
∞
F (f ) df
(8.32)
0
This is the frequency spectrum. It can be generalised to the directional case as
2
σ =
2π ∞ 0
F (f , θ) df dθ
(8.33)
0
So to summarise, the directional frequency spectrum F (f , θ ) can be used to describe the variability of the sea-surface elevation. Note that there is no phase information in this description, so the actual surface elevation as depicted in Fig.€8.4 could not be reconstructed from the spectrum, but instead, it describes the distribution of the energy in the wave field according to wave frequency and direction. The wave spectrum is a very useful construct and is the prognostic variable for current state-of-the-art wave models. A couple of examples of directional wave spectra are shown in Fig.€8.5. The top panel of this figure shows both a full directional wave spectrum and its directionally integrated one-dimensional equivalent. This depicts a relatively simple sea-state in which most of the wave energy is propagating towards the west, with a fairly large spread around this direction. The peak energy occurs at a frequency of around 0.15€Hz, i.e. most of the energy is being carried by waves with period of about 6.7€sec (this is the peak period, Tp). For the spectrum shown in the bottom
8â•… Surface Waves
213 1.2
Spectral Density (m2 sec)
1.0
N
W
0.8
0.6
0.4
E
0.05 0.1
0.2
0.2 0.3
0.0 0.0
0.4 S
0.1
0.2 0.3 0.4 Frequency (Hz)
N
N-W
N-E
0.4
W
E
S-W
S-E
S
Fig. 8.5↜渀 Examples of directional wave spectra
0.5
214
D. Greenslade and H. Tolman
panel, there are a number of different components to the sea state, with wave energy clearly propagating in a number of different directions. You can imagine that the sea-state described by this wave spectrum would look quite complicated and very different to the wave field represented by the spectrum in the top panel.
8.4.2 Significant Wave Height Significant Wave Height (↜Hs) is another very important concept that is used frequently to describe the sea state. The idea of wave height for a simple sinusoidal wave is trivial—the wave height is defined to be twice the amplitude, so for each of the 5 wave components depicted in Fig.€8.4, it is straightforward to determine the wave height. But what is the wave height of the resulting wave field? Hs has come to be used to describe a number of different “wave heights” that can be derived from a wave field. These are all typically very close in value, but given their different methods of derivation, there are some subtle differences of which it is important to be aware. The original definition is that based on visual observations. Someone out on a boat in the open ocean can observe the waves and estimate what the “average” wave height is. Clearly this will be a subjective estimate and different observers may well produce different wave height estimates. This is called the Significant Wave Height. A second definition is that obtained through direct observations of the sea-surface elevation. In this case, the Significant Wave Height is defined to be the average of the one-third highest waves in a sample, where a “wave” is defined through the upward or downward crossing definition (see, for example Holthuijsen (2007) for definitions of these). In this case, the resulting wave height should more accurately be referred to as H1/3, but Significant Wave Height is more often used. It has been shown that the visually observed wave height is closely correlated to this definition of wave height (Jardine 1979). It implies that an observer only sees the higher waves, and automatically ignores smaller waves riding on the dominant waves. Hs can also be derived from the wave spectrum. Using the definition that it is the mean value of the highest one-third of the waves in a given record, and assuming that the wave heights (or more specifically the crest heights) are Rayleigh-distributed, then H1/3 can be shown to be equal to (Holthuijsen 2007): √ 4.004 . . . m0 (8.34) where m0 is the zeroth-order moment of the wave spectrum given by
m0 =
2π ∞ 0
F (f , θ) df dθ
(8.35)
0
This is equivalent to the volume enclosed by the two-dimensional spectrum (the one-dimensional version would be the area under the curve of the one-dimensional
8â•… Surface Waves
215
spectrum). The value of 4.004… is typically rounded to 4 and so the spectrallyderived definition of H1/3, which more formally should be referred to as Hm0 can be written as 2π ∞ (8.36) Hm0 = 4 F (f , θ) df dθ 0
0
Again, this is almost always referred to as Hs. In order to determine this from a modelled wave spectrum, the integral needs to be expressed as a sum over the discrete frequency and directional range of the modelled spectrum. Given that the model has a limited range of frequencies that it can resolve, a high-frequency tail is usually included, with a slope of f −n, where n is usually 4 or 5, so it is straightforward to determine the area under this part of the spectrum and it can be added to the Hs. (See the one-dimensional spectrum in Fig.€8.5—the spectral values stop abruptly at the highest frequency that the model is able to resolve). The Significant Wave Height is a statistical measure for the wave height. Clearly, individual waves can be both lower and higher. It can be shown that in a simple spectrum describing a single coherent wave system, the probability distribution of the height of individual waves closely follows the Rayleigh distribution (e.g., Holthuijsen 2007). This distribution implies that 1 in 100 waves is expected to be as large as 1.51Hm0 , and 1 in 1000 waves is expected to be as large as 1.86Hm0 . Higher waves rapidly become less likely, which is why waves higher than approximately 2.0Hm0 are typically called “freak” or “rogue” waves. We have seen here that there are a number of different ways of describing the “wave height” of a particular wave field and these are typically all referred to as Significant Wave Height, or Hs. Clearly, this one value used for describing the seastate is a gross simplification. It would be reasonable to use this to describe a simple sea-state in which there is only one dominant component to the wave field, but consider the two sea-states in Fig.€8.5. The Hs is similar in each panel (↜Hsâ•›=â•›1.36€m in the top panel compared to Hsâ•›=â•›1.03€m in the bottom panel) even though the seastates depicted by the spectra are very different. Simply using Hs to describe a seastate means that you lose a lot of information about the structure of the wave field. This is similar to giving a weather forecast with a simple maximum temperature value. It doesn’t tell you whether you need to take your umbrella or not!
8.5â•…Operational Wave Modelling 8.5.1 Background and Basics This section focuses on operational wave modelling in the context of wave forecasting. As mentioned previously, most current state of the art wave forecast models
216
D. Greenslade and H. Tolman
are phase-averaged third generation models, which have the wave spectrum as their prognostic variable. The most common models in usage at international forecasting centres are WAM (WAMDIG 1988; Komen et€al. 1994) and WAVEWATCH III® (Tolman et€al. 2002, 2009). These are computationally efficient models that can be used for large scale global forecasting. Also the SWAN model (Booij et€al. 1999; Ris et€al. 1999) is extensively used, but more for near-shore engineering applications. A review of the state of the art of operational (and research) wave modelling can be found in Cavaleri et€al. (2007). The basis of virtually all wind wave models used in operational forecasting is some form of the balance equation for the wave energy spectrum F (f , θ ) as discussed in Sect.€8.4.1. In its most simple form, it is given as
∂F + ∇.(cg F ) = Sin + Snl + Sds + Sbot ∂t
(8.37)
where the left hand side represents the effects of linear propagation, and the right hand side represents sources and sinks for spectral wave energy. Propagation, in its simplest form, only considers wave components in the spectrum to propagate along great circles, until the wave energy gets absorbed at the coast (either as part of the propagation algorithm, or due to the dissipation source terms). More advanced versions of this equation, as used in prevalent models, also consider refraction (changing of wave direction due to interaction with the bottom in shallow water) and shoaling (changing of wave height and length due to changing water depths), and some consider similar effects due to the presence of mean currents. So far, all operational wave models consider linear propagation only. Many operational models now address the effects of unresolved islands and reefs as sub-grid obstructions. Traditionally, three source terms have been considered; Sin describing the input of wave energy due to the action of the wind, Snl describing the effects of nonlinear interactions between waves, and Sds describing the loss of wave energy due to wave breaking or “whitecapping”. Many early models for shallow water applications added a wave-bottom interaction source term, Sbot, which was typically concerned with wave energy loss due to friction in the bottom boundary layer. Of these source terms the nonlinear interactions have a special relevance. Effects of nonlinear interactions occur as source terms in this equation, because the propagation description in the equation is strictly linear. Furthermore, the interactions are essential for wave growth, and not for propagation. They represent the lowest order process known to effectively lengthen waves during growth, and they have been shown to stabilize the spectral shape at frequencies higher than the spectral peak (e.g., Komen et€al. 1994). Nonlinear interactions consider resonant exchanges of energy, action and momentum between four interacting wave components, governed by a six-dimensional integration over spectral space. The SWAMP study in the 1980’s (SWAMP group 1985) identified the explicit computations of these interactions as essential for practical wave models. The development of the Discrete Interaction Approximation (DIA) (Hasselmann et€al. 1985) made this economically feasible. Models that explicitly compute nonlinear four-wave interactions are identified as third-generation wave models.
8â•… Surface Waves
217
Present operational wave models address source terms in a much more detailed fashion. Wind input is turning into wind-wave interactions, and can include feedback of energy and momentum to the atmosphere (“negative input”). Further to this, wave breaking is seen as impacting on atmospheric turbulence, and hence influencing atmospheric stresses and wave growth. Nonlinear interactions now regularly include both four-wave interactions in deep water and three-wave (triad) interactions in shallow water. Wave dissipation now regularly addresses traditional whitecapping in the deep ocean, and separate mechanisms for depth-induced (“surf”) breaking, and much slower dissipation mechanisms that influence swell travelling across basins with decay time scales of days to weeks. Many additional wave-bottom interactions are also considered in shallow water. Most prevalent are bottom friction source terms, but other processes such as wave-sediment interactions associated with bottom friction, percolation and scattering of waves due to bottom irregularities have been proposed and are available in some wave models. Of special recent interest is the interaction of waves with muddy bottoms, which both adds a source term and may modify the dispersion relation and hence wave propagation. Source terms for other processes such as wave-ice interactions and effects of rain on waves have been proposed, but are presently not used in any practical wave models.
8.5.2 Operational Centres Many operational weather forecast centres run operational wind wave models. This is not done by accident. During the 1974 Safety of Life at Sea (SOLAS) conference, international agreement was reached to consider wind waves as part of the weather, explicitly giving weather forecast centres the responsibility to do wave forecasting for the public. The first numerical wave predictions, however, far precede this date, and in the U.S.A. can be traced back to 1956 (see historical overview in Tolman et€al. 2002). Many of the larger weather forecast centres such as the European Centre for Medium Range Weather Forecasts (ECMWF1, Europe), The National Centers for Environmental Prediction (NCEP2, USA) and the Bureau of Meteorology (Bureau3, Australia) produce wave forecasts for up to 10 days ahead, on 6–12€ h forecast cycles. Most of these centres use a global wave model, with one or more higherresolution nested regional models for areas of special interest. For example, the configuration of WAM at the Bureau (as at end of 2009) is shown in Fig.€8.6. The highest resolution model (blue boundary) is run at 0.125° resolution in latitude and longitude, and is nested inside a model at 0.5° spatial resolution (red boundary) which is in turn nested inside the global model at 1°. Typically, the higher resoWeb site at http://www.ecmwf.int. Wave data at http://polar.ncep.noaa.gov/waves. 3╇ Wave data at http://www.bom.gov.au/marine/waves.shtml. 1╇ 2╇
218
D. Greenslade and H. Tolman 20 E
40 E
60 E
80 E
100 E
120 E
140 E
160 E
160 W
140 W
120 W
100 W
80 W
60 W
40 W
20 W
70 N
70 N
50 N
50 N
30 N
30 N
10 N
10 N
10 S
10 S
30 S
30 S
50 S
50 S
70 S
70 S
20 E
40 E
60 E
80 E
100 E
120 E
140 E
160 E
160 W
140 W
120 W
100 W
80 W
60 W
40 W
20 W
75 N 60 N
45 N 30 N
15 N
EQ 15 S 135 E 150 E 165 E
180 165 W 150 W 135 W 120 W 105 W 90 W 30 x 30 15 x 10 10 x 10 8x4 4x4
75 W 60 W
Fig. 8.6↜渀 Examples of configurations of some operational wave model systems. Top panel shows the Bureau and bottom panel shows NCEP
lution models obtain data from the lower resolution models without feeding any information back, but full two-way nesting of such models is now used at NCEP (Tolman 2008). Configuration of the NCEP system (as at end of 2009) is also shown in Fig.€8.6. This incorporates a range of different spatial resolutions ranging from global at 0.5° down to the highest resolution models at 4 arc minutes (1/15th of a degree) around the coastlines. The spatial resolutions of the wave models are typically dictated by the resolutions of the atmospheric models from which the wave models obtain their wind forcing and additionally, by the availability of computing resources. In an operational forecasting environment, a major consideration is the
8â•… Surface Waves
219
time taken for the model to complete a forecast and the speed with which the results can be disseminated. Some centres also run specialised wave models for specific conditions; for example NCEP run wave models specifically for hurricanes, with specialized forcing from hurricane weather models. Finally, several centres run wind wave ensembles, to provide probabilistic information on the expected reliability of the forecast. While such ensembles have been generated for up to a decade, they have not been scrutinized as much as corresponding atmospheric ensembles, and may not have reached the same level of maturity. Further details of operational wave forecast systems can generally be found at the websites for the forecast centres as given in the footnotes. In addition to differences in the spatial resolutions of the models, there is considerable variety in other aspects of the operational implementations of wave forecast systems at each forecasting centre. For example, the wind forcing used to force the wave model will typically be provided by the centre’s Numerical Weather Prediction (NWP) model, and these can vary considerably in detail. Whether the wave model incorporates data assimilation or not can also contribute to differences in the forecasts. The most widely used data source that is assimilated in wave models is Hs from satellite altimeters. This can significantly improve the skill of wave forecasts (Greenslade and Young 2005), particularly in cases where the surface winds are known to have deficiencies. One limitation to the assimilation of Hs data is that it can not provide any direct information on the observed wave spectrum, so a number of assumptions need to be made in adjusting the modelled spectrum (Greenslade 2001). This issue can be somewhat overcome by incorporating the assimilation of wave spectra from Synthetic Aperture Radar (SAR), such as is performed at the ECMWF (ECMWF 2008). In situ wave buoys could also provide wave spectra for assimilation. However, the limitation of these is that compared to satellite data, they are very sparsely distributed and they tend to be located near the coast, for logistical reasons. The fact that they are typically not used in wave data assimilation schemes means that they can be used as a valuable independent data source for model verification. Many of the operational forecast centres share their model results through a wave model intercomparison study supported by the Joint Commission for Oceanography and Marine Meteorology (JCOMM) (Bidlot et€al. 2007). Model forecasts are also compared to observations from in situ buoys around the globe. This project provides a mechanism for benchmarking and the quality assurance of wave forecast products. The results are available each month to all participants and published on the web.4 An example of the intercomparison at one location is shown in Fig.€8.7. This shows 24-hour forecasts of Hs and Tp at buoy 44005 (located 78 nautical miles off the coast of New Hampshire, in the northwest Atlantic) for the month of November 2009. In the top panel, it can be seen that all wave models are able to forecast the Hs reasonably well, with the synoptic scale variability being captured very well. There is some spread around the observed Hs, and for this example, most of the models have overpredicted the peak Hs occurring around the 15th of November. The Tp is also quite well captured this month, particularly the dominance of long waves (high 4╇
Web site at http://www/jcomm.info
Fig. 8.7↜渀 An example of results from the wave intercomparison activity
220 D. Greenslade and H. Tolman
8â•… Surface Waves
221
wave period) during the middle of the month and the trend towards shorter period waves at the end of the month. The high variability in Tp seen in both the observations and the models from the 3rd to the 13th suggests that there were a number of different wave systems present during this period. There are also a number of summary results from this intercomparison activity produced each month. An example is shown in Fig.€8.8. This shows the root-mean-
Fig. 8.8↜渀 Summary statistics for one month from the wave model intercomparison project. Each coloured line represents forecasts from a different operational centre. Top panel: Hs, middle panel, u10 and bottom panel, Tp. The x-axis in each panel represents forecast period, in days
222
D. Greenslade and H. Tolman
square (rms) error amongst the forecast models averaged over all buoy data available for the three parameters, Hs, Tp and u10 (wind speed at 10€m above the surface). The error is defined as the difference between the modelled and observed parameters. The rms error can be seen as a measure of the skill of a model. Figure€8.8 shows that the rms error for a 24-hour forecast (1 day) is approximately 0.5€m, although it varies from about 0.4€m to about 0.7€m. Errors of wave models normalized with mean conditions for the better models are of the order of 15% for hindcasts and short term forecasts (results not shown). Another feature obvious from this figure is the growth in error with forecast period. It can also be seen that there is a strong correlation between the rms error in the surface winds and the rms error of the wave forecasts, i.e. those centres that have accurate surface winds also have high skill for the wave forecasts. With continuously improving weather models at all centres, differences in wave models and in the selection of numerical and physical options in these models is becoming more and more apparent and important. After a decade of relatively small changes to wind wave modelling approaches, this has recently lead to an increased interest in improved physics approaches in the corresponding wave models.
8.5.3 Outlook As mentioned above, a renewed interest in wave model development has surfaced in the last few years. This is particularly clear with the recently started National Oceanographic Partnership Program (NOPP) project which aims to provide the next generation of source term formulations for operational wind wave models. Literally all source terms in the wave models will be addressed in this study, with a focus on deep water and continental shelf physics. A greater focus of operational centres on coastal wave modelling is emerging, Partly due to increased requirements from users of the service and also due to the increasing ability of wave models to address this, given advances in computing power. With this, alternate modelling approaches such as curvilinear and unstructured grids are becoming more prevalent and more important. Furthermore, the mode of operation of many forecast centres is slowly changing. Traditionally, operational centres have focused on isolated topical forecast problems such as weather and waves. More and more, such centres are moving toward an integrated earth-system modelling approach, where the links between models are seen as essential to improve the quality of the individual models. Wind waves literally are the interface between the atmosphere and the ocean. In a systems design approach, a wind wave model could become an advanced boundary layer module for an integrated atmosphere-ocean modelling system. At ECMWF, a first step into this direction was made more than a decade ago, when their wind wave model started providing real time surface roughness information (including wave-induced roughness) to the weather model. At NCEP, coupled atmosphere-ocean models are used for climate and hurricane prediction. Experimental versions of the hurricane
8â•… Surface Waves
223
model now include a three-way-coupled system, consisting of full weather model (HWRF), a full ocean model (HYCOM) and a full wave model (WAVEWATCH III). A similar system is under development at the Bureau. In such a model the wind waves play a key role; they modify surface roughness and therefore stresses; they may temporarily store momentum extracted from the atmosphere, and release this to the ocean in a geographically distant location; spray generated by waves influences (and links) momentum, heat and mass fluxes between the ocean and atmosphere. Indeed, the most complete estimates of spray production are directly related to the wave spectrum, and hence require a full wave model. Another important forecast problem in which wind waves become important is coastal inundation, where many coastal inundation problems are directly linked to momentum produced by incoming swell rather than by wind pushing up water in a tradition storm surge situation. Several decades of experience with wave-driven coastal circulation and inundation problems can be found in the civil engineering literature, but these experiences have not yet been used in operational forecasting procedures.
References Bidlot J-R, Li JG, Wittmann P, Fauchon M, Chen H, Lefevre J-M, Bruns T, Greenslade DJM, Ardhuin F, Kohno N, Park S, Gomez M (2007) Inter-Comparison of Operational Wave Forecasting Systems. Proceedings of the 10th international workshop on wave hindcasting and forecasting, Oahu, Hawaii, USA, Nov 2007 Booij N, Ris RC, Holthuijsen LH (1999) A third-generation wave model for coastal regions 1. Model description and validation. J Geophys Res 104:7649–7666 Cavaleri L, Alves JHGM, Ardhuin F, Babanin AV, Banner ML, Belibassakis K, Benoit M, Donelan MA, Groeneweg J, Herbers THC, Hwang P, Janssen PAEM, Janssen T, Lavrenov IV, Magne R, Monbaliu J, Onorato M, Polnikov V, Resio DT, Rogers WE, Sheremet A, McKee Smith J, Tolman HL, Van Vledder G, Wolf J, Young IR (2007) Wave modeling—The state of the art. Prog Oceanogr 75:603–674 ECMWF (2008) IFS Documentation—CY33r1, Part VII: ECMWF Wave model. http://www. ecmwf.int/research/ifsdocs/CY33r1/WAVES/IFSPart7.pdf Greenslade DJM (2001) The assimilation of ERS-2 significant wave height data in the Australian region. J Mar Sys 28:141–160 Greenslade DJM, Young IR (2005) The impact of inhomogenous background errors on a global wave data assimilation system. J Atmos Oc Sci 10(2). doi:10.1080/17417530500089666 Hasselmann SK, Hasselmann JH, Allender, BarnettTP (1985) Computation and parameterization of the nonlinear energy transfer in a gravity wave spectrum. Part II: Parameterizations of the nonlinear energy transfer for application in wave models. J Phys Oceanogr 15:1378–1391 Holthuijsen LH (2007) Waves in oceanic and coastal waters. Cambridge University Press, Cambridge Jardine TP (1979) The reliability of visually observed wave heights. Coast Eng 3:33–38 Komen GJ, Cavaleri L, Donelan M, Hasselmann K, Hasselmann S, Janssen PAEM (1994) Dynamics and modelling of ocean waves. Cambridge University Press, Cambridge, p€532 Kundu PK (1990) Fluid mechanics. Academic Press Inc., San Diego Ris RC, Holthuijsen LH, Booij N (1999) A third-generation wave model for coastal regions 2. Verification. J Geophys Res 104:7667–7681 SWAMP Group (1985) Ocean wave modeling Plenum Press, London, p€256 Tolman HL (2008) A mosaic aproach to wind wave modeling. Ocean Model 25:35–47
224
D. Greenslade and H. Tolman
Tolman HL (2009) User manual and system documentation of WAVEWATCH III version 3.14. NOAA/NWS/NCEP/MMAB Technical Note 276. http://polar.ncep.noaa.gov/mmab/papers/ tn276/MMAB_276.pdf Tolman HL, Balasubramaniyan B, Burroughs LD, Chalikov DV, Chao YY, Chen HS, Gerald VM (2002) Development and implementation of wind generated ocean surface wave models at NCEP. Weather Forecast 17:311–333 WAMDIG (1988) The WAM model—A third generation ocean wave prediction model. J Phys Oceanogr 18:1775–1810 Young IR (1999) Wind generated ocean waves. Elsevier Science Ltd, Amsterdam
Chapter 9
Tides and Internal Waves on the Continental Shelf Gregory N. Ivey
Abstract╇ We review recent laboratory experiments, field observations and numerical modeling of internal waves produced by tidal motions, with specific focus on the Australian North West Shelf. Distinct regimes are observed depending upon both the characteristics of the ambient density stratification, the topography, and the intensity of the tidal forcing. The character of the near boundary flow in the region where waves are generated is very important in determining the internal wave response. When cyclones are present, the intense mixing over the water column can suppress the formation of tidally generated internal wave motions for many days.
9.1â•…Introduction Internal waves are ubiquitous in the ocean and can be generated by turbulent stirring (e.g. Munroe and Sutherland 2008) or by mean motion, such as tidal flows over topography (e.g. Baines and Fang 1985). The action of the tide sweeping stratified water over oceanic topography leads to the generation of internal waves of tidal origin (internal tides) which can, in turn, play an important role in deep ocean mixing and large-scale ocean circulation (e.g. Munk and Wunsch 1998; Wunsch and Ferrari 2004) and is the focus of the present paper. Freely propagating internal waves with frequency ω propagate energy in the direction of the group velocity vector at an angle θ to the horizontal given by the dispersion relation
ω2 = N 2 sin2 θ + f 2 cos2 θ ≈ N 2 sin2 θ
(9.1)
where the simplification is valid providing the Coriolis parameter f is small compared to the buoyancy frequency N. An important parameter in the tidal generation G. N. Ivey () School of Environmental Systems Engineering and UWA Oceans Institute, The University of Western Australia, M015 35 Stirling Highway, Crawley, WA 6009, Australia e-mail:
[email protected] A. Schiller, G. B. Brassington (eds.), Operational Oceanography in the 21st Century, DOI 10.1007/978-94-007-0332-2_9, ©Â€Springer Science+Business Media B.V. 2011
225
226
G. N. Ivey
of internal waves is thus the topographic steepness parameter γ = S/α , where S = hs /ls is the average slope of the topography ( hs and ls are characteristic vertical and horizontal lengthscales) and wave slope is defined as α = tan θ. Note that by this definition γ is an overall parameter and the usual definition of a critical point (e.g. Gostiaux and Dauxois 2007; Zhang et€al. 2008) is when the local bottom slope matches the wave ray slope. In addition to the forcing frequency ω , as the tide is characterized by a tidal velocity U0 , other parameters of importance in internal tide generation are (e.g. Garrett and Kunze 2007) the topographic Froude number F r = U0 /N hs and the tidal excursion parameter U0 /ωls . For subcritical topography (↜γ < 1), and in the limits when U0 /ωls 1 and hs /H 1, linear internal tides are generated (Balmforth et€al. 2002; Bell 1975; Legg and Huijts 2006). As the topography approaches criticality (↜γâ•›=â•›1), the internal tide manifests itself as a beam-like structure, emanating from the critical point on the topography (Gostiaux and Dauxois 2007; Griffiths and Grimshaw 2007), while for U0 /ωls > 1 the response is dominated by higher harmonic frequencies (e.g. Bell 1975). Internal wave motions are commonly observed near continental slopes (Holloway et€al. 2001; Lien and Gregg 2001), seamounts (Lueck and Mudge 1997; Toole et€al. 1997), mid-ocean ridges (Ray and Mitchum 1997) and near continental shelf regions such as the Australian North West Shelf (NWS). The NWS has strong tidal forcing and is home to many vigorous internal wave motions which can play an important role in the energy budget and hence turbulent stirring of the Shelf waters (e.g. Holloway et€al. 2001; Van Gastel et€al. 2009). The NWS region lies in the parameter space γ < 2, U0 /ωls 1, and F r 1. Field measurements are sparse and only provide information at point locations. Therefore it is often difficult to identify the physical generation mechanism of the internal tide and the subsequent internal wave propagation and dissipation in a given region. This paper therefore reviews recent laboratory, field observations and numerical modeling of internal tide generation with specific focus on the NWS. As the NWS is also prone to cyclones during the summer season, we conclude with an examination of cyclone influences on the generation of internal tides.
9.2â•…Laboratory Models Most laboratory studies have focused on the process of internal tide generation in a continuously stratified fluid where, near critical points (↜γâ•›=â•›1), the internal tide manifests itself as a beam-like structure emanating locally parallel to the bottom from the critical point on the topography (e.g. Gostiaux and Dauxois 2007; Peacock et€al. 2008; Zhang et€al. 2008). Two recent studies (Lim et€al. 2008, 2010), have examined the generation process by both varying γ and the intensity of forcing relative to turbulent stirring. Characterizing the turbulence in the near bottom boundary layer with an eddy viscosity K, the effect of forcing can be characterized by a local Reynolds number defined as Re = U02 /(NK) (Legg and Klymak 2008). The upper range of barotropic forcing, and hence Re, examined in these studies is consider-
9â•… Tides and Internal Waves on the Continental Shelf
227
$VLQZW K
KV
K
KV /V
a
7ZROD\HUVWUDWLILFDWLRQ $VLQZW
KV /V
b
&RQWLQXRXVVWUDWLILFDWLRQ
Fig. 9.1↜渀 Configuration of laboratory experiments by Lim et€al. (2008, 2010). The vertically oscillating plunger at the left end generates an oscillating barotropic flow over the slope/shelf topography at the other end of the tank. Experiments were done with a two-layer density stratification and b continuous density stratification
ably larger than previous experimental studies and the experimental configuration is shown in Fig.€9.1, where both idealized two layer and continuously stratified versions of the stratification were used. In the two layer experiments, Lim et€al. (2008) documented differing responses √ delineated by two parameters: a Froude number F r = U0 / g hE and the layer depth ratio on the shelf β = h1 /h1 + h2S (see Fig.€9.1). Their classification scheme is shown in Fig.€9.2. If the upper layer depth on the shelf was thin (i.e. β < 0.5 ), linear internal waves of depression moved onto the shelf. Only if the lower layer depth on the shelf was thin (i.e. β > 0.5 ) were strongly non-linear waves observed with both surges and distinct bores present. In this latter category, the Froude number also became important, and with strong tidal forcing when F r → 1 there was no internal wave response observed at all on the shelf. In the continuous stratification case, three types of basic flow response were observed by Lim et€ al. (2010) over the parameter range of the experiments ( 0.3 < γ < 2.2, 1 < Re < 480 ): beams, bolus structures, and finally no waves. The presence of both a critical point in the domain and a stable boundary layer with flow parallel to the boundary were found to be fundamental criteria for beam generation. With the oscillating mean flow locally parallel to the boundary, for locally critical conditions, the movement of fluid in the bottom boundary was along the direction
228
G. N. Ivey
5HJLPH$ ZDYHRIGHSUHVVLRQ KV
5HJLPH' QRZDYHV KV!!
)U
5HJLPH% LQWHUQDOVXUJH KV
$XVWUDOLDQ1:6 1RUWK5DQNLQUHJLRQ
5HJLPH& LQWHUQDOEROXV KV!
5HJLPH'QRZDYHV
Fig. 9.2↜渀 Two √ layer regime classification scheme (from Lim et€ al. 2008). The Froude number F r = U0 / g hE where the equivalent depth hE = h1 h2S /(h1 + h2S ), β = h1 /h1 + h2S and the depth scales are defined in Fig.€9.1a
of the wave characteristic slope (see Eq.€(9.1)) tangent to the local bottom slope. If there was no critical slope present in the domain, then there was no internal beam generation. The region of beam generation was shown to occur within a finite length of the slope ( 0.75 < s/scrit < 1.3 ), and this length was approximately twice the local near-bottom fluid excursion. The velocities along the wave characteristic was shown to be elevated, consistent with previous field studies (e.g. Holloway et€al. 2001; Lien and Gregg 2001) and laboratory (e.g. Peacock et€al. 2008; Zhang et€al. 2008) observations. Increasingly energetic conditions led to the generation of a bolus (e.g. Venayagamoorthy and Fringer 2007) causing much over-turning and stirring as it propagated upslope and dissipated. Lim et€al. (2010) found the overall flow behaviour varies with both Re and γ, but the two non-dimensional parameters can be combined to define a single generation parameter G 1/2 U02 ω2 Re U02 ω = G= ≈ (9.2) γ NKS N 2 − ω2 N 2 KS where the simplification is valid since ω2 N 2 in the laboratory.
9â•… Tides and Internal Waves on the Continental Shelf
229
VPDOOHUIRUFLQJ IUHTXHQF\ VWURQJHU VWUDWLILFDWLRQ
,QWHUQDO%HDPV *
γ
,QWHUQDO%ROXV *
ODUJHUIRUFLQJ IUHTXHQF\ ZHDNHU VWUDWLILFDWLRQ
1R:DYHV
ORZHUIRUFLQJ
5H
KLJKHUIRUFLQJ
Fig. 9.3↜渀 Regime diagram for continuous stratification (see also Lim et€al. 2010). The parameter γ = S/α and Re = U02 /N K, and G is defined in Eq.€(9.2)
The general trend is that as the forcing increases the response changes from linear beams to nonlinear bolus features and finally to no waves at all. A summary of the observed behaviour is shown in Fig.€9.3. Beams required a critical slope to be present and were observed in the regime G╛╛BYSâ•›>â•›ECS in general. The lowest hindcast skill is in BYS during the summer. This region- and season-dependence is also supported by the skills of persistence. This fact indicates that to some extend the model hindcast skills are influenced by underlying dynamical processes in different regions and seasons and reflect the seasonal and regional short-term SST predictabilities. The results from two hindcast cases are shown in Figs.€17.11 and 17.12. During the period from June 21 to 27, SST over ECS increases sharply as indicated by the northward movement of the 24°C isoline near the coast. A cooling event and its
Fig. 17.11↜渀 An example of SST hindcast in June, 2006. The scale of the color bar is °C
17â•… Overview of Regional and Coastal Systems
433
Fig. 17.12↜渀 The same as Fig.€17.10, but for another example in September, 2006
hindcast are shown in Fig.€17.11. The 28°C isoline moves southward during September 8 to 14 and the water with SST less than 28°C covers all the shelf area on September 14. The model hindcasts capture the two events successfully while the persistence cannot.
17.4.3 Hindcast Error Distributions We further explore the predictability of SST in the marginal seas around China. It is beyond the scope of this paper to investigate all aspects of predictability. We only focus on identifying where the SST forecast errors grow fast in summer and winter. Figure€17.13 shows the spatial distribution of the averaged RMSE of a 6-day hindcasts in winter and summer. The large errors in winter mainly locate in the Kuroshio path in the ECS, in the Luzon Strait, off the Vietnam coast along 110E and in the Taiwan Strait. This explains that the hindcast skills are lower in ECS than that in SCS and BYS as shown in Fig.€17.9. In the summer, the large errors mainly locate at coast and shelf regions and northeast of Taiwan. And thus cause the low hindcast skill in BYS during the summer. Because the mixed layer is very shallow in BYS (about 8€m according to Chu et€al. 1997) and a strong stratification exist beneath the mixed layer in summer, the SST is very sensitive to the atmospheric forcing and has large day-to-day changes there. The atmospheric forcing errors could be one of the main factors that cause the large hindcast errors in BYS during summer via the
434
J. Zhu
Fig. 17.13↜渀 The spatial distributions of the averaged RMSE of 6-day hindcasts. The scale of the color bar is °C. The errors in the water shallower than 40€m are marked out due to relative large errors in the validation data
thermal effect. The errors in wind have strong impact on the coast SST hindcast due to the upwelling and downwelling processes. Another reason may be the lack of tides in the model. As mentioned before, the tidal mixing has large impact on summer SST distribution in BYS. Apart from the errors in the atmospheric forcing and the ocean mixing, the errors in horizontal advections also have large impacts on the hindcast skills. Since the horizontal advection is determined by the inner product of the surface current vector and the SST spatial gradient. Figure€17.14 shows the spatial distribution of absolute
Fig. 17.14↜渀 The spatial distributions of the temporally averaged absolute values of the local SST spatial gradients over the winter and the summer of 2006. The SST analysis is used to calculate the gradient. The scale of color bar is 10−4°C/m
17â•… Overview of Regional and Coastal Systems
435
values of the temporally averaged local SST spatial gradients over the winter and the summer of 2006. Instead in winter a strong gradient appears in the shelf between Hainan Island and Taiwan, Kuroshio path associated with the shelf break in ECS, Changjiang River estuary, and the area off the southwest coast of Korea. In summer the gradient is weaker than that in winter and shows its strength mainly around the Korea coast. The front patterns agree well with the previous studies based on multiyear remote sensing SST data (Hickox et€al. 2000; Wang et€al. 2001). The strong SST gradients agree well with the large hindcast error distributions in Fig.€17.12 especially in the Kuroshio path and in the shelf between Hainan Island and Taiwan. However, off the east coast of Vietnam and the Luzon Strait large hindcast errors exist but the SST gradient is only moderate. Considering the strong current there (e.g., Fig.€17.2 of Li et€al. 2010), the horizontal advection errors may also cause a large hindcast error.
17.5╅Summary and Outlook During the GODAE period, several regional operational and preoperational systems have been developed in the Asia-Oceania region. These systems have demonstrated their usefulness via providing routine service for public, government and commerce users or by successful forecasting/hindcasting high impact (socially, economically and scientifically) events. All these regional systems have strong connections with the GODAE products. The Argo and GHRSST datasets are essential inputs for initialization of these forecast systems. The large scale GODAE products are also used to provide side boundary conditions (e.g., MOVE-NP). A 15 year ocean reanalysis from BLUElink has proved to be useful for engineering design. For example the modelling of internal waves through downscaling assisted the planning of the successful search of HMAS Sydney which was sunk during World War II. The BLUElink operational forecasts have also demonstrated to have good skills at forecasting a wide range of phenomena including: extreme coastal sea level, anomalous currents impacting offshore oil and gas operations, anomalous heat content over the North West Shelf influencing continental rainfall and many other processes. Most of these systems have been developed in GODAE related projects. For example, the 3D ocean forecasting system DMI BSHcmod has been continuously developed in the projects like in the MERSEA and ECOOP projects. The MERSEA Baltic Sea forecasting system is based on the same model. It is still a challenging task to further develop the existing systems from science perspective (De Mey et€al. 2009). Establishing more observation networks, increasing model resolution, adding sea ice model, using more advanced data assimilation and coupling with atmospheric models are among the near future activities. For example, in the near future, JMA will introduce an assimilation scheme of sea ice
436
J. Zhu
concentration to MOVE/MRI.COM-WNP, which would yield to some improvements not only in the sea ice extent but also in the ocean state of the subarctic region, especially in the Okhotsk Sea. JMA is also planning to develop a coastal ocean modeling/assimilation system using a high-resolution model with a horizontal resolution of a few kilometers that is intended for a possible operational use in JMA’s forecasting and warning systems for the coastal region of Japan. BLUElink through a follow-on research project will also introduce an upgraded reanalysis and operational prediction system (mid-2010) and introduce a new coupled regional oceanatmosphere system for tropical cyclone prediction. The global prediction system will enhance the eddy-resolving region to include the Indian Ocean and South Pacific and will also include advances in data assimilation, initialization scheme and atmospheric fluxes. Another challenge is how to further apply the achievements in Asia-Oceania operational activities comes from high-level decision-makers. The Indian Ocean is relatively less covered by these regional systems. GOOS-CLIVAR’s effort in establishing the Indian Ocean Observing system (IndOOS) will improve the situation. Research Moored Array for African-Asian-Australian Monsoon Analysis and Prediction, a new observational Multi-national network designed for Indian Ocean, a subset of IndOOS, similar to TAO (Pacific) and PIRATA (Atlantic), aims to address outstanding scientific questions related to Indian Ocean variability and the monsoons. 22 out of 46 moored buoys are already occupied (McPhaden et€al. 2009). On the other hand, Southeast Asian countries also urgently requires such a service for storm-surge forecast, coastal engineering and disaster prevention etc, and some existing systems such as YEOS system are ready to be extended to cover the entire NW Pacific coastal/shelf seas. There are encouraging signs of more Asia-Oceania countries plan to develop operational systems. A study on pre-operational oceanographic system will be funded by the Ministry of Land Transport and Ocean Affairs of Korea from next year. They will start to produce data products of coastal and environmental forecasts for the coastal waters around Korea. From a series of SST hindcast experiments in China marginal sea, we found that several following-up works are necessary. The initial conditions provided by data assimilation seem to have room to further reduce the analysis misfit to data. The causes of the visible initial shocks after the assimilation should be further investigated. Apart from the generation of gravity waves, other reasons should also be thought. Counillon and Bertino (2009) found a data assimilation set-up that produces little noise that is dampened within two days, when the model is pulled strongly towards observations. Part of it is caused by density perturbations in the isopycnal layers, or artificial caballing. Because the model used by them is also HYCOM, their results are very suggestive. The tide-induced mixing has strong impact on the thermal field in BYS. A higher resolution model setup than the present one, with tides is now running and will be used to perform new hindcast experiments. Acknowledgements╇ Part of this lecture note comes from Zhu et€al. (2008) to which co-authors: Toshiyuki Awaji, Gary B. Brassington, Norihisa Usuii, Naoki Hirose, Young Ho Kim, Qinzheng Liu, Jun She, Yasumasa Miyazawa, Tatsuro Watanabe and M. Ravichandran have contributed greatly.
17â•… Overview of Regional and Coastal Systems
437
References Arakawa A (1972) Design of the UCLA general circulation model. Numerical simulation weather and climate, technical report. No. 7. Department of Meteorology, University of California, Los Angeles, p€116 Awaji T, Masuda S, Ishikawa Y, Sugiura N, Toyoda T, Nakamura T (2003) State estimation of the North Pacific Ocean by a four-dimensional variational data assimilation experiment. J Oceanogr 59:931–943 Bleck R (2002) An oceanic general circulation model framed in hybrid isopycnic-Cartesian coordinates. Ocean Model 4:55–88 Bloom SC, Takacs LL, da Silva AM, Ledvina D (1996) Data assimilation using increment analysis updates. Mon Weather Rev 124:1256–1271 Brassington GB, Pugh T, Spillman C, Schulz E, Beggs H, Schiller A, Oke PR (2007) BLUElink> Development of operational oceanography and servicing in Australia. J Res Pract Inf Technol 39:151–164 Chassignet EP, Smith LT, Halliwell GR, Bleck R (2003) North Atlantic simulations with the Hybrid Coordinate Ocean Model (HYCOM): impact of the vertical coordinate choice, reference pressure, and thermobaricity. J Phys Oceanogr 33:2504–2526 Chen D, Busalacchi AJ, Rothstein LM (1994) The roles of vertical mixing, solar radiation, and wind stress in a model simulation of the sea surface temperature seasonal cycle in the tropical Pacific Ocean. J Geophys Res 99:20345–20359 China Ocean Press (1992) Marine Atlas of Bohai Sea, Yellow Sea, and East China Sea, hydrology. China Ocean Press, Beijing, p€524 Chu PC, Fralick CR Jr, Haeger SD, Carron MJ (1997) A parametric model for the Yellow Sea thermal variability. J Geophys Res 102(C5):10499–10507 Counillon F, Bertino L (2009) Ensemble optimal interpolation: multivariate properties in the Gulf of Mexico. Tellus-A 61:296–308 De Mey P, Craig P, Kindle J, Ishikawa Y, Proctor R, Thompson K, Zhu J, CSSWG (2009) Applications in coastal modeling and forecasting. Oceanography 22(3):198–205 Dick S, Kleine E, Müller-Navarra S, Klein H, Komo H (2001) The Operational Circulation Model of BSH (BSHcmod). Model description and validation. Bericte des Bundesamtes für Seeschiffahrt und Hydrographie. Nr. 29/2001. Hamburg, Germany, p€48 Fujii Y, Kamachi M (2003) Three-dimensional analysis of temperature and salinity in the equatorial Pacific using a variational method with vertical coupled temperature-salinity empirical orthogonal function modes. J Geophys Res 108(C9):3297. doi:10.1029/2002JC001745 Gent PR, McWilliams JC (1990) Isopycnal mixing in ocean circulation models. J Phys Oceanogr 20:150–155 Griffies S, Hallberg RW (2000) Biharmonic friction with a Smagorinsky-like viscosity for use in large-scale Eddy-permitting ocean models. Mon Weather Rev 128:2935–2946 Griffies SM (1998) The Gent–McWilliams skew flux. J Phys Oceanogr 28:831–841 Griffies SM, Harrison MJ, Pacanowski RC, Rosati A (2004) A technical guide to MOM4 GFDL Ocean Group technical report No. 5, NOAA/Geophysical Fluid Dynamics Laboratory, p€339 Guo X, Hukuda H, Miyazawa Y, Yamagata T (2003) A triply nested ocean model simulating the Kuroshio—roles of horizontal resolution on JEBAR-. J Phys Oceanogr 33:146–169 Hasumi H (2000) CCSR Ocean Component Model (COCO) Version 2.1. CCSR Report No.13 Hasumi H (2006) CCSR ocean component model (COCO) version 4.0. Report No. 25. Center for Climate System Research, The University of Tokyo Hickox R, Belkin I, Cornillon P, Shan Z (2000) Climatology and seasonal variability of ocean fronts in the East China, Yellow and Bohai Seas from satellite SST data. Geophys Res Lett 27(18):2945–2948 Hirose N, Kawamura H, Lee HJ, Yoon JH (2007) Sequential forecasting of the surface and subsurface conditions in the Japan Sea. J Oceanogr 63:467–481
438
J. Zhu
Hunke EC, Ducowicz JK (2002) The elastic–viscous–plastic model for sea ice dynamics. Mon Weather Rev 130:1848–1865 Ichikawa H, Beardsley RC (2002) The current system in the Yellow and East China Seas. J Oceanogr 58:77–92 In T, Ishikawa Y, Shima S, Nakayama T, Kobayashi T, Togawa T, Awaji T (2008) A triple nesting approach toward the improved nowcast/forecast of coastal circulation off Shimokita Peninsula (to be submitted) Ishikawa Y, Awaji T, Toyoda T, In T, Nishina K, Nakayama T, Shima S, Masuda S (2009) Highresolution synthetic monitoring by a 4-dimensional variational data assimilation system in the northwestern North Pacific. J Mar Syst 78(2):237–248 Ishizaki H, Motoi T (1999) Reevaluation of the Takano-Oonishi scheme for momentum advection on bottom relief in ocean models. J Atmos Ocean Technol 16:1994–2010 Kalnay E et€al (1996) The NCEP/NCAR 40-year reanalysis project. Bull Am Meteorologic Soc 77:437–471 Kim YH, Chang KI, Park JJ, Park SK, Lee SH, Kim YG, Jung KT, Kim K (2009) Comparison between a reanalyzed product by the 3-dimensional variational assimilation technique and observations in the Ulleung Basin of the East/Japan Sea. J Mar Syst 78:249–264 Larsen J, Høyer JL, She J (2007) Validation of a hybrid optimal interpolation and Kalman filter scheme for sea surface temperature assimilation. J Mar Syst 65:122–133 Leonard A (1979) A stable and accurate convective modeling procedure based on quadratic upstream interpolation. Comput Methods Appl Mech Eng 19:59–98 Leonard BP, MacVean MK, Lock AP (1993) Positivity-preserving numerical schemes for multidimensional advection. NASA Technical Memorandum 106055, ICOMP-93-05 Levitus S, Burgett R, Boyer TP (1994) World Ocean Atlas, vol€3, salinity. NOAA Atlas NESDIS, 3, United States Department Of Commerce, Washington Li XC, Zhu J, Xiao YG, Wang RW (2010) A model-based observation thinning scheme for assimilation of high resolution SST in the shelf and coastal seas around China. J Atmos Ocean Technol 27:1044–1058 Marchesiello P, McWilliams JC, Shchepetkin A (2001) Open boundary conditions for long-term integration of regional oceanic models. Ocean Model 3:1–20 McPhaden MJ, Mayers G, Ando K, Masumoto Y, Murty VSN, Ravichandran M, Syamsudin F, Vialard J, Yu L, Yu W (2009) RAMA: the research moored array for African-Asian-Australian monsoon analysis and prediction. Bull Am Metrologic Soc. 90: 459–480 Mellor GL, Blumberg A (2004) Wave breaking and ocean surface layer thermal response. J Phys Oceanogr 34:693–698 Mellor G, Kantha L (1989) An ice-ocean coupled model. J Geophys Res 94:10937–10954 Mellor GL, Hakkinen S, Ezer T, Patchen R (2002) A generalization of a sigma coordinate ocean model and an intercomparison of model vertical grids. In: Pinardi N, Woods JD (eds) Ocean forecasting: conceptual basis and applications. Springer, New York, pp€55–72 Miyazawa Y, Kagimoto T, Guo X, Sakuma H (2008a) The Kuroshio large meander formation in 2004 analyzed by an eddy-resolving ocean forecast system. J Geophys Res 113:C10015. doi: 10.1029/2007JC004226 Miyazawa Y, Komatsu K, Setou T (2008b) Nowcast skill of the JCOPE2 ocean forecast system in the Kuroshio-Oyashio mixed water region (in Japanese with English abstract and figure captions). J Mar Meteorol Soc (Umi to Sora) 84(2):85–91 Moon I-J (2005) Impact of a coupled ocean wave–tide–circulation system on coastal modeling. Ocean Model 8:203–236 Noh Y, Kang YJ, Matsuura T, Iizuka S (2005) Effect of the Prandtl number in the parameterization of vertical mixing in an OGCM of the tropical Pacific. Geophys Res Lett 32:L23609. doi: 10.1029/2005GL024540 Uppala SM, Kallberg PW et€al (2005) The ERA-40 re-analysis. Q J R Meteorol Soc 131:2961–3012 Oey LY, Chen P (1992) A Nested-Gris Ocean model—with application to the simulation of Meanders and Eddies in the Norwegian Coastal Current. J Goephys Res 97:20063–20086
17â•… Overview of Regional and Coastal Systems
439
Oke PR, Schiller A, Griffin DA, Brassington GB (2005) Ensemble data assimilation for an eddyresolving ocean model of the Australian region. Q J R Meteorol Soc 131:3301–3311 Oke PR, Brassington GB, Griffin DA, Schiller A (2008) The Bluelink Ocean Data Assimilation System (BODAS). Ocean Model 21:46–70 Pacanowski RC, Griffies SM (1999) MOM3.0 manual. WWW Page, http://www.gfdl.noaa. gov/~smg/MOM/web/guide_parent/guide_parent.html Rio MH, Hernandez F (2004) A mean dynamic topography computed over the world ocean from altimetry, in situ measurements, and a geoid model. J Geophys Res 109:C12032. doi:10.1029/2003JC002226 Schiller A, Oke PR, Brassington GB, Entel M, Fiedler R, Griffin DA, Mansbridge J (2008) Eddyresolving ocean circulation in the Asian-Australian region inferred from an ocean reanalysis effort. Prog Oceanogr 76:334–365 Toyoda T, Awaji T, Ishikawa Y, Nakamura T (2004) Preconditioning of winter mixed layer in the formation of North Pacific eastern subtropical mode water. Geophys Res Lett 31:L17206. doi: 10.1029/2004GL020677 Tsujino H, Usui N, Nakano H (2006) Dynamics of Kuroshio path variations in a high-resolution general circulation model. J Geophys Res 111:C11001. doi:10.1029/2005JC003118 Usui N, Tsujino H, Fujii Y, Kamachi M (2006) Short-range prediction experiments of the Kuroshio path variabilities south of Japan. Ocean Dyn 56:1616–7341 Wan L, Zhu J, Bertino L, Wang H (2008) Initial ensemble generation and validation for ocean data assimilation using HYCOM in the Pacific. Ocean Dyn 58:81–99. doi:10.1007/s10236008-0133-x Wang D, Liu Y, Qi Y, Shi P (2001) Seasonal variability of thermal fronts in the northern South China Sea from satellite data. Geophys Res Lett 28(20):3963–3966 Weaver A, Courtier P (2001) Correlation modeling on the sphere using a generalizing diffusion equation. Q J R Meteorol Soc 127:1815–1846 Xiao Y, Zhu J (2007) Numerical simulation of circulations in coastal and shelf sea around China using a hybrid coordinate ocean model. Technical report (in Chinese) Xie S, Hafner J, Tanimoto Y, Liu WT, Tokinaga H, Xu H (2002) Bathymetric effect on the winter sea surface temperature and climate of the Yellow and East China Seas. Geophys Res Lett 29(24):2228. doi:10.1029/2002GL015884 Xie JP, Zhu J, Yan L (2008) Assessment and inter-comparison of five high resolution sea surface temperature products in the shelf and coastal seas around China. Cont Shelf Res 28:1286–1293 Zhu J, Awaji T, Brassington GB, Usuii N, Hirose N, Kim YH, Liu Q, She J, Miyazawa Y, Watanabe T, Ravichandran M (2008) Asia and oceania applications. Proceedings of the final GODAE symposium. Available from the GODAE website, pp€359–372
Chapter 18
System Design for Operational Ocean Forecasting Gary B. Brassington
Abstract╇ The scientific and technical advances in ocean modelling, ocean data assimilation and the ocean observing systems over the past decade have made the grand challenge of ocean forecasting an achievable goal with the implementation of the first generation systems (Dombrowsky et€al. 2009). Implementation of these components into a truly operational forecasting system introduces a number of unique constraints that can lead to reduced performance. These practical constraints, such us the limitations in the coverage and quality of critical components of the ocean observing systems in real-time as well as the constraints of completing forecast integrations within a fixed schedule are unavoidable components for any forecast system and require additional strategies to achieve robustness and maximise performance. We begin by defining commonly used terms such as operational and forecasting in this context. We then review the design choices that can be taken with each component of an ocean prediction system when implemented as an operational system to achieve the most reliable performance.
18.1╅Introduction Operational ocean forecasting systems have been established over the past decade by several agencies and institutions (Dombrowsky et€al. 2009). Hurlburt et€al. 2009 provides an appraisal of key developments over this period. These systems employ a wide variety of techniques (Kamachi et al. 2004; Cummings 2005; Brasseur et€al. 2005; Martin et€al. 2007; Oke et€al. 2005, 2008) largely due to the maturing state of the science. None of these techniques are theoretically optimal as defined by the use of a 4D variational scheme (Lorenc 2003) or an ensemble Kalman Filter (Evensen 2003). However, the computational cost of eddy resolving models which preclude the use of 4DVar and EnKF approaches, together with the poor knowledge of the G. B. Brassington () Centre for Australian Weather and Climate Research, Bureau of Meteorology, Melbourne, Australia e-mail:
[email protected] A. Schiller, G. B. Brassington (eds.), Operational Oceanography in the 21st Century, DOI 10.1007/978-94-007-0332-2_18, ©Â€Springer Science+Business Media B.V. 2011
441
442
G. B. Brassington
background error covariances that apply in the ocean has led to a wide variety of sub-optimal approaches being employed. Guiding principles for good design can be found in many quotations of which we cite three. The first of these is referred to as the Law of the instrument and is attributed to Abraham Maslow, “When the only tool you have is a hammer, it is tempting to treat everything as if it were a nail”. The law of the instrument is a warning to new scientists and engineers that need to work on improving existing systems that many of the design choices are based on the known methods and techniques at that time. All design choices are constrained by those methods and should be regularly questioned and reviewed. The second quotation is a warning against reductionism and attributed to Albert Einstein, “Make things as simple as possible, but not simpler”. All components of the ocean prediction system contain assumptions that reduce the problem into simpler elements that offer advantages e.g., methods of solution. All assumptions that reduce the parameter space of the system are true under defined conditions e.g., Boussinesq, hydrostatic and incompressible assumptions. A thorough knowledge of these assumptions and the conditions under which they hold is critical when re-applying methods or systems for new applications. Alternatively, all the advantages of new efficient method are of no use if they do not solve the target problem to within a required precision. The third and final quote is the antithesis of the previous quote and again is attributed to Albert Einstein, “Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius—and a lot of courage—to move in the opposite direction”. This quote is particularly relevant with present systems as the trend is toward higher model resolutions, more complex data assimilation methods, ensemble forecasting and coupled physical models. It serves to pause and justify before automatically introducing greater system complexity. This trend is scaling with the improvement in computing system performance and is likely to continue. A good analogy for operational ocean forecasting design today is that of the chronometer invented by John Harrison (Sobel 1995). Take a visit to the museum in Greenwich, London and you will see an incredible piece of design/art called H1 (see Fig.€18.1a). This was designed by John Harrison to solve the Longitude problem by
Fig. 18.1↜渀 a The H1 clock and b chronometer designed by John Harrison to solve the longitude problem
18â•… System Design for Operational Ocean Forecasting
443
producing a clock that could perform accurately at sea and claim the significant monetary prize. Anyone cannot help but admire the quality of the design and the achievement of its clock. However, this particular clock was abandoned by John Harrison after 17 years of development as he realised how he could improve to eventually arrive at a pocket sized device called the chronometer (see Fig.€18.1b). Operational oceanography today is analogous to the H1 where it functions as it was designed to, contains many novel and elegant solutions but remains far from where it will be over the coming decades in terms of its techniques and importantly its reliable performance. In this paper, we begin by offering a definition for commonly used terms related to ocean forecasting specifically identify properties unique to operational forecasting. We then provide a short overview of applications for ocean forecasting and common servicing requirements influencing design. Section€18.4 introduces the system elements of an ocean forecasting system which is followed by an expanded discussion on each of these elements with particular emphasis on the properties of each component that influence the system design. This includes, Sect.€18.5 real-time observing system, Sect.€18.6 real-time forcing system, Sect.€18.7 modelling, Sect.€18.8 data assimilation, Sect.€18.9 initiatlization, Sect.€18.10 forecasting cycle, Sect.€18.11 system performance. Throughout we have highlighted aspects of an operational system that require design choices to be made and are of a general interest to system design. By way of demonstration, examples are drawn from specific systems with the cautionary note that these may or may not be general practice. A majority of the examples are drawn from the BLUElink Ocean Model, Analysis and Prediction System (OceanMAPS) which is noted throughout. We then end with a short conclusion.
18.2╅Definitions The initial development of all forecasting systems is performed under hindcast conditions (see Table€18.1). In many respects hindcasts frequently attempt to mimic the forecast environment however, many of the conditions that occur in real-time are difficult to reproduce and are not necessarily normally distributed e.g., drop outs in satellite products (see Fig.€18.2). Alternatively, it is often desirable to determine the statistical performance of a system operating under ideal conditions which sets the upper bound in performance. In practice, this level of performance only ocTable 18.1↜渀 Definitions of terms frequently used in reference to assimilated ocean model states Forecasting terminology Hind-analysis Best estimation using optimal methods and maximum information Hindcast Behind real-time simulation of forecasts i.e., model initialisation from a hindanalysis and model projection Hindcasts are typically performed under ideal conditions and represent an upper bound in forecast performance Nowcast Estimation of the state and circulation at real-time that can be used as a persistence forecast Forecast Prediction of the state and circulation beyond real-time
G. B. Brassington
444
d_sst a_sst
Number
3 .105
2 .105
1.105
0 01/04 2009
11/04
21/04
01/05
11/05 21/05 Date (dd/mm)
31/05
10/06
20/06
Fig. 18.2↜渀 Observation retrievals from AMSR-E (-) ascending and (…) descending swaths obtained at the Bureau of Meteorology between 4th January 2009 and 30 June 2009
curs when the forecast conditions approach the ideal. In the design of a forecast system the performance of the system under less than ideal conditions is of equal importance. This frequently introduces additional strategies to minimise impact to achieve the highest lower bound. For this reason it is critical to use terminology of forecast and hindcast systems appropriately and to define the conditions of the system accurately. The term operational is frequently used with a wide variety of working definitions but interestingly also has a specific philosophical heritage (see http://plato. stanford.edu/entries/operationalism/). A useful working definition was outlined during the develoment of EuroGOOS (Prandle and Flemming 1998). The term as it applies to operational forecasting is summarised here in Table€18.2 as relating Table 18.2↜渀 Definitions for the meaning of operational as they apply to world meteorological agencies Operational Real-time System and products targeting nowcast and forecasts Routine Performs to a regular schedule Robust Technological: High-end computing and communications with designed failovers and fit-for purpose scheduling Scientific: Detect and mitigate changes in system state to ensure minimum impact to performance Consistent Consistently achieving the designed performance
18â•… System Design for Operational Ocean Forecasting
445
to: real-time services delivered routinely and robustly. Many operational centres measure success against the delivery of services 24/7. A considerable amount of resources are expended in order to achieve the level of servicing of 99.99% up time typical of a WMO agency. Consistency of the quality of the services is also critical to design choices.
18.3â•…Applications Prior to designing any system it is important to define the applications to be targeted for the system and to define the service requirements that need to be met. This is critical to both the design of the observing system and forecasting systems. However, operational oceanography has not rigorously followed this idealised approach. Operational oceanography was initiated as an experiment, Global Ocean Data Assimilation Experiment (Smith and Lefebvre 1997) motivated by the opportunity presented by the new and expanding global ocean observing system particularly with the introduction of satellite altimetry. The many sectors that could potentially benefit from ocean forecasting services were more or less known at that time. However, the specific applications and the forecast skill requirements were not known. There are several properties of the applications that will influence the design of ocean forecast systems and the impact of those services are summarised in Table€18.3. These include the type of application, its social or economic value, the sophistication of the user community and the service requirements. A subset of the potential applications are represented in Fig.€18.3. Figure€18.3a, b represents an upwelling event that took place off the Bonney coast in South Australia on the 10th February 2008. Upwelling frequently impacts local marine ecosystems bringing nutrient rich water into the photic zone resulting in a chlorophyll bloom that is observable by ocean colour (Fig.€ 18.3b) on the 30th March 2008. Upwelling can also have a stabilising effect on the local atmospheric boundary layer reducing the transfer of momentum to the surface. Upwelling can occur very rapidly and therefore can be absent from atmospheric forecasts that persist SST boundary conditions. This specific event resulted in a forecast failure where strong winds were forecast but local observers experienced weak winds. This lead to a complaint by local tourist operators who had cancelled ocean cruises. Upwelling events can also be associated with sea fog which is difficult to observe using either infrared radiation (e.g., Advanced Very High Resolution Radiometer (AVHRR)) due to the presence of fog or microwave (e.g., Advanced Microwave Scanning Radiometer— EOS (AMSR-E)) due to the coarse resolution (~25€km/pixel) and interference near coastlines. A dynamical forecast is required to generate the cool SST’s in response to the wind which can be observed however, the precision of the forecast SST’s is difficult to validate. Marine accident and emergency services from ships (Fig.€ 18.3c) and from oil wells (Fig.€ 18.3f) including the airborne and ship-based salvage operations (Fig.€18.3g) are an obvious application of ocean forecasting services. However, the
446
G. B. Brassington
Table 18.3↜渀 Properties of the applications and user communities that impact the design choices in ocean forecasting systems Applications Types Ad hoc time and space (e.g., Search and rescue, Marine accident and emergencies, defense) Planning and management (e.g., fisheries by-catch, marine park management) Engineering/industrial (e.g., offshore oil and gas, ship routing, renewable energy) Global and continuous (e.g., weather, wave, ecosystem forecasting) Coastal shelf (e.g., ports management, bilge discharge, coastal surge) Public good (e.g., recreational fishing, diving, swimming, sailing) Social and economic Life, safety or security threatening value Property damage Marine health Economic value and energy User community Are the users structured and coordinated? sophistication Are the service needs well-defined? Are the impacts of ocean services understood? Capacity to interpret ocean products and add value Capacity to monitor and assess impacts Capacity to engage in a relationship and provide usable feedback Service requirements Hindcasts, Short-, medium-, long-range forecasts Performance thresholds Sensitivity to error Sensitivity to extremes Observational requirements Timeliness and frequency of forecast products
requirements for skilful Lagrangian trajectories has been difficult to achieve. The present and future global ocean observing system over the next decade is unlikely to be sufficient to meet the needs of these applications (Hackett et€al. 2009; Davidson et al. 2009; Rixen et€al. 2009; Brassington et€al. 2010a). A characteristic of these events is that they occur infrequently at ad hoc locations and are localised making them suitable for short term, intense observing deployments through the use of gliders, AUV’s and drifting buoys etc. An atmospheric feature that is common to Australia, Brazil and the United States on their respective eastern coastlines is the formation of rapidly intensifying extratropical cyclones (see Fig.€18.3d). These storms are sometimes referred to as bombs due to their severity and impacts. The event in June 2007 made famous by the grounding of the cargo ship Pasha Bulka also resulted in loss of life due to flooding in Newcastle. On the east coast of Australia these storms form when a cut-off low of cold dry air moves over a warm moist marine boundary leading to vertical convection and a positive feedback in the atmosphere of convergence of the up-
Fig. 18.3↜渀 A collage of applications that require real-time forecast services a forecast SST’s for a coastal upwelling event of South Austrlai. b The ocean colour response for the same vent. c Oil washed on shore off Queensland due to a leak from the Pacific Adventurer. d Modelled 10€m winds of an East coast cyclone off the NSW coast. e The modelled SST conditions associated with the event. f Oil discharge from the Montara oil well. g Ship based salvage operations for the same event. h Surge along the Derwent river in Tasmania and i the forecast sea level for this event
18â•… System Design for Operational Ocean Forecasting 447
448
G. B. Brassington
per layer potential vorticity (McInnes et€al. 1992). The ocean heat content along these coastlines is highly variable due to the turbulent western boundary current that transports warm/fresh water from the tropics to higher latitudes. The modelled SST shown in Fig.€18.3e exhibits a temperature front in the same position as the storm. The warm SST’s were maintained by a warm-core anticyclonic eddy (Brassington 2010; Brassington et€al. 2010b). An ocean forecast system can provide forecasts of the heat content conditions with potential for use in coupled forecasts. High sea level along the coast is typically associated with the coincidence of tides and storm surge. Forecasting systems are typically based on so-called “stormsurge” models local to the event to estimate risk in combination with tides and sea level pressure. Simulations of non-tidal sea level in ocean forecasting systems can also be impacted by other oceanographic effects of remote coastally trapped waves and impinging warm boundary currents. For example, a high sea level event in the Derwent river (see Fig.€18.3h) resulted from a local storm and a large amplitude coastal trapped wave propagating from South Australia. A characteristic of the coastal trapped wave is the high sea level in the Bass Strait. (see Fig.€18.3i). Regional forecasters did not issue a warning due to their use of traditional methods of computing sea level which did not account for the remote contribution. Ocean forecast systems have the potential to provide total sea level forecasts. In each of these applications the oceanographic conditions play an important role for which accurate forecasts can provide valuable information. Detailed analysis of these and other similar cases can identify the relevant oceanographic variables and the sensitivity to error to derive the requirements in terms of performance. In these examples, SST, heat content, surface currents and sea level are directly relevant which accounts for four of the five prognostic variables in a hydrostatic ocean general circulation model. Though it is important to note that their forecast are dependent upon the knowledge of all prognostic variables. National agencies and institutions are regularly engaged with local users and have opportunities to acquire this information. The JCOMM Expert-Team on Operational Ocean Forecast Systems (ET-OOFS) is tasked with providing international coordination to generalise this information into observational and service requirements.
18.4╅System Elements All operational ocean forecasting systems available today follow a similar sequential and cyclic structure which involves handling of the latest observational data, performing a model-data fusion, performing a model forecast to generate data products including ocean state estimates, performance diagnostics and error estimates. This sequential procedure is repeated on a regular schedule or performed in an ad hoc basis e.g., triggered by a specific event. The system diagram for the BLUElink OceanMAPS system is shown in Fig.€ 18.4. This includes retrieval and archival storage of observations, surface fluxes, model and data assimilation dependent data files.
ODAS data retrieval system
Profile data archival system
OGCM data retrieval system
Bureau operational NWP products
Satellite data retrieval system
Bureau operational SST products
Database/Archive
Satellite data archival system
Profile data retrieval system
Surface flux retrieval system
449
Servers and GFS file system
18â•… System Design for Operational Ocean Forecasting
ODAS
Super computing
OGCM
Super computing ODAS product archival system
Data servicing system
OGCM product archival system
Fig. 18.4↜渀 A schematic diagram of the system elements for an operational ocean forecasting system. (Based on the BLUElink OceanMAPS, Brassington et€al. 2007)
A Supervisor Monitor Scheduler (SMS) developed at the European Centre for Medium-Range Weather Forecasts (ECMWF; see http://www.ecmwf.int/products/ data/software/sms.html) or equivalent software, is implemented at operations centre, to control the job flow monitoring the successful completion of dependent system components. The data and file handling is performed on servers whilst the large memory and computationally intensive tasks for data assimilation and model integration are submitted to high-end super-computing systems. The performance of the computing environment and the level of optimisation that can be achieved with the software is critical to design of ocean forecasting systems. Eddy-resolving ocean forecast systems are at the high-end of supercomputing application both for the prognostic model and the data assimilation inversion. The total wall clock time and the computing resources available in an operations centre are limited and managed among several other forecast systems. The efficiency of the software and the consistency of the completion times for different components has important impacts on design. For example, the computational cost of a data assimilation system will scale with the size of the inversion problem. Targeting a reduction in cost may compromise the number of observations processed through super-observations or thinning, require the implementation of localisation or limit the specification of the
450
G. B. Brassington
background error covariance. Similarly the cost of ocean model design scales with the number of grid points/cells and the timestep constraint for numerical stability. Targeting a specific cost limit will compromise the horizontal/vertical resolution within the model or the area of high resolution. The system described above accounts for the majority of the science and technical design for an ocean forecasting service. However there are several steps to the provision of a quality service to end users. These include infrastructure for robust data product dissemination, forecaster guidance as well as support services for specifying user requirements and evaluating impacts. These important steps are not discussed further here.
18.5╅Real-Time Observing System The global ocean is now observed by a growing number of instruments and platforms that each have specific properties, some common and some unique, that will impact the design in operational ocean forecasting. These properties are summarised in Table€ 18.4 and include the timeliness, coverage, expected errors and quality. The relative immaturity of ocean instrumentation and infrastructure leads to more frequent system failures in practice compared with numerical weather prediction. System failures are frequently random and unpredictable though the sensitivity of the forecasting system to failures in the observing system are measurable. Strategies to minimise the impact need to be considered in the system design. For a more detailed discussion on aspects of the ocean observing system refer to Le Traon (2011) and Ravichandran (2011).
18.5.1 In Situ—Profiles The ocean state is routinely profiled in real-time by Conductivity-TemperatureDepth (CTD) sensors from traditional platforms such as ships and moorings and Table 18.4↜渀 Properties of the real-time ocean observing system that result in unique design choices in ocean forecasting systems Real-time bserving system Timeliness How close to real-time are the observations received? Are delayed products available with higher quality? Coverage What is the minimum/maximum coverage? How homogeneous is the coverage? Observation error estimation Instrument error Representation error Quality control Does the product include quality flags? Valid tests for the observation error model Non-normal behaviour Instrument failures, communication and system failures
18â•… System Design for Operational Ocean Forecasting
451
relatively new platforms such as autonomous Argo floats and gliders, from volunteer ships. In addition eXpendable Bathy-Thermograph (XBT) are operated from volunteer ships and reported in real-time. The sampling by in situ measurements has significantly increased over the past decade and coverage has increased in regions that have been historically poorly sampled such as the Indian Ocean and Southern Ocean. The Argo array is now the dominant source of in situ sampling having largely achieved its target density of one float per 3°â•›×â•›3° over the global ocean withâ•›>â•›3000 autonomous floats. Each float profiles the ocean water column from ~2000€m to the surface every 10 days reporting in real-time at the surface via ARGOS or Iridium (Roemmich et€al. 2010). A user guide to the range of Argo data products available and the server access points are given online (http://www.argo.ucsd.edu/Argo_Date_Guide. html). Observations are retrieved by a network of Data Assembly Centres (DAC’s) which are responsible for performing an automatic quality control procedure and distributing the observations to both the WMO Global Telecommunication System (GTS) and the two Global DAC’s (GDAC’s). The DAC’s also perform an objective quality control in delayed mode. Profiles that pass the automatic quality control are reported in real-time to the GTS without quality control information in TESAC format. A fast mode product is available from GDAC servers within 3 days in a format that contains the quality control flags and native observations on pressure coordinates. Other important CTD profiles are obtained from the mooring arrays, TAO/TRITON (Pacific; McPhaden et€al. 2001), PIRATA (Atlantic; Bourles et€al. 2008) and recently RAMA (Indian; McPhaden et€al. 2009). These moored arrays report in real-time, multiple times per day and are reported onto the GTS. Increasingly gliders are being used to adaptively sample the ocean however, the data acquisitions are as yet not coordinated internationally in the same way as Argo and lack a common real-time quality control procedure, integration with the GTS and other DAC/GDAC product delivery. XBT’s have been maintained along specific ship routes and sampling is constrained by the frequency of the volunteer ships that occupy the route (Goni et€al. 2010). XBT’s provide high vertical resolution profiles of temperature and depth at regular spacing along the ship route. Profiles, subsampled in the vertical, are reported on the GTS without quality control flags. The profiles are subsequently subjectively quality controlled by a number of centres using a common set of procedures (Bailey et€al. 1994). As an example the number of profiles that were retrieved at the Bureau of Meteorology each day from the GTS and the two Argo GDAC’s (Coriolis and USGODAE) between the 15th January 2010 and 1st March 2010 are shown in Fig.€18.5. The GTS reports consistently ~1200 profiles per day although the most recent retrievals show an increase in the number of observations due to shallow coastal observations in the USA. The GDACS’s report on average 300 profiles per day corresponding to the expected number of Argo floats surfacing each day. The number of profiles retrieved from each GDAC do not correlate and are clearly not simply a mirror site. Coriolis also frequently has bursts of profiles which largely contain old profiles that a DAC has subjectively QC’d. The best daily observations available in near real-time are obtained by sorting amongst the three sources. Ideally the three sources should contain a maximum of three duplicates for the same profile that must
452
G. B. Brassington
,QVLWX3URILOH
PPW JRGDH FRULROLV JWV
1XPEHU
Fig. 18.5↜渀 The number of ocean profiles retrieved daily between 15 January 2010 and 1st March 2010 from the GTS (↜purple), Coriolis (↜blue) and USGODAE (↜green) and the number of duplicate free profiles (↜red)
be reduced to one. The best profile is determined to be the one that has both the most complete set of observations and the maximum set of quality control tests applied. The number of profiles obtained for each day from the duplicate checking procedure is shown in Fig.€18.5 in red. The best daily observations provides consistently ~1200 profiles per day. The decline in profiles near real-time shows the impact of timeliness of the profiles with a small percentage of the total profiles obtained several days behind real-time. An algorithm developed at the Bureau of Meteorology (Brassington et€al. 2007) to select the best profiles replaces profiles obtained from the GTS with more complete profile information, particularly quality control, obtained from the GDAC’s. A typical example of the timeliness, volume and source of the profiles obtained from that system is shown in Fig.€18.6 for the 13th September 2009. Within the first two days of real-time, the number of profiles is dominated by those obtained from the GTS. Within the first day GTS profiles are being replaced by profiles from the GDAC’s. In the 3rd and subsequent days profiles from the GDAC’s continue to replace those obtained from the GTS. The number of profiles replaced declines as the time behind real-time increases.
18.5.2 Satellite SST Sea surface temperature is the most frequently observed ocean state variable by satellites with multiple sensors and multiple orbits. Microwave sun-synchronous and IR Geostationary platforms provide higher coverage whilst the IR polar orbiting missions provide the highest resolution and accuracy in cloud free conditions. There
18â•… System Design for Operational Ocean Forecasting
453
Fig. 18.6↜渀 The profiles received in real-time on 13th September 2009 that represent the best profile available from previous retrievals and the source of the profile
are several known limitations to the use of observed SST for ocean forecasting with diurnal warming and skin effects. Specific algorithms are required to perform quality control relevant to the foundation temperature (refer to the online definitions maintained by the Global High Resolution Sea Surface Temperature (GHRSST) science team, http://www.ghrsst.org/SST-Definitions.html). Foundation temperature specifically refers to the near surface temperature of the ocean excluding diurnal skin effects. In practice, observations are withheld from the analysis as being impacted by diurnal skin effects based on the time of day and the magnitude of the 10€m winds as a proxy for near surface mixing (Donlon et€al. 2002). The algorithms do not attempt to correct the temperature values for any diurnal effects, therefore the day time temperatures will include a small residual bias. Night-time observations of SST are also affected by a cool skin effect however this is a relatively small perturbation compared with daytime biases. Algorithms use a smaller constraint on atmospheric winds resulting in greater coverage. Therefore the night-time foundation SST’s represent a more robust estimate and offer greater coverage compared with day-time products. The majority of ocean forecasting systems at present do not explicitly represent the diurnal skin layer which requires a vertical resolution 24€h and the period of one earth orbit 24€h. Both descending and ascending swaths show reduced coverage over the inter-tropical convergence zones and monsoons, though the position of these changes with season. In the high latitudes, SST coverage is near 100% up to the ice edge where the atmospheric conditions are of high winds and dry air. Foundation SST from AMSR-E must remove all pixels
Fig. 18.7↜渀 Percentage of days observed by AMSR-E for Austral seasons and ascending(asc)/ desending(desc) orbits a summer, desc, b summer, asc, c winter (desc) and d winter (desc)
18â•… System Design for Operational Ocean Forecasting
455
contaminated by precipitation up to a chosen threshold. In some applications, where maximum coverage is essential a higher threshold can be used. However, for ocean forecasting (foundation temperature) a more conservative approach is important. The so-called Level-2 Pre-processed (L2P) product (refer to http://www.ghrsst.org/ L2P-Observations.html) provides all of the necessary fields to diagnose and select the threshold for ocean forecast applications. The NOAA AVHRR series has been sustained as an operational platform with wide-swath infrared sensors and multiple satellites in sun-synchronous orbits. NAVOCEANO provide a merged, foundation temperature, swath L2P product available in near real-time. The resolution ~1€km is greater than that of current and near future ocean forecast systems. This permits the construction of super-observations (e.g., Lorenc 1981; Purser et€ al. 2000) that have reduced representation error increasing the weighting in the analysis. The higher resolution also provides observations over the continental shelf and Gulf regions compared with microwave sensors. An observation error for the foundation temperature can be constructed to account for residual diurnal signals based on the time from nearest local night-time as well as an age penalty for time from the analysis time (Andreu-Burillo et€al. 2009).
18.5.3 Satellite Altimetry Remotely sensed satellite altimetry observes a broad spectrum of dynamical processes including: tides, wind-waves and swell and steric anomalies. Steric anomalies relate to the changes in height from the vertical integral of specific volume anomalies from the background. Vertically coherent specific volume anomalies are prominent in ocean eddies where they can have relatively warm and/or fresh cores relative to the surrounding ocean state leading to positive height anomalies or relatively cool and/or salty cores leading to negative height anomalies. Analyses of merged altimetry have revealed that 50% of the variability of the world ocean is accounted for by eddies with height anomalies of 5–25€cm and diameters 100– 200€km (Chelton et€al. 2007). The speed of propagation for the majority of eddies is found to range from 2.5 to 12.5€cm/s with a westward propagation ±10° (Chelton et€ al. 2007). In regions where the geostrophic turbulence is more active such as near western boundary currents the eddy propagation speeds can transiently exceed 40€cm/s (Brassington 2010) and can develop height anomalies in excess of 25€cm and diameters in excess of 200€km (see Fig.€18.8). Recovering sea surface height anomalies from satellite altimetry requires precise estimation of a large number of corrections (Chelton 2001). For example the ssha is recovered from Jason1 by the following equation (Desai et€al. 2003).
ssha = (orbit − (range_ku + iono + dry + wet + ssb)) − (mss + setide + otide + pole + invbar) + bias
(18.1)
where range_ku refers to the range delay for the Ku-band and iono, dry, ssb refer to range corrections for the ionosphere, dry/wet troposphere and sea state bias.
456
G. B. Brassington
Fig. 18.8↜渀 a An example of on d 98.55 ay of altimetry passes from Envisat, Jason1 and Jason2 for the 1st January 2010, in the Australian region. b ±2 days of altimetry passes about the 1st January 2010 overlaying the corresponding background sea level anomaly in the Tasman Sea from OceanMAPS. c same as b but for ±5 days
The terms mss, setide, otide, pole and invbar refer to geophysical effects of mean sea surface, solid Earth tide, ocean and load tide, pole tide, and inverse barometer response. Bias is a correction term resulting from calibration of the orbits. The mean sea surface or geoid is estimated by the time mean of orbit tracks repeated for several years to a precision of 1€km. It is for this reason that the repeat missions Jason1 and Jason2 to TOPEX-Poseidon are put into the same orbits (Robinson 2006). Ocean tidal harmonics are known and can be estmated to high precision with inverse methods (Le Provost 2001). The errors attributed to the TOPEX-Poseidon, Jason class missions is 3€cm, ERS, Envisat and Sentinel missions is 6€ cm and GFO is 10€ cm (Robinson 2006). The precision that can be achieved by the merger of the Jason class mission and ERS missions is 5€ cm (Ducet et€al. 2000). Future altimetry missions from the HY-2 series from China, SARAL for the Ka-band altimeter (Altika) from an Indian consortia and Cryosat have as yet unknown errors but are able to obtain improved errors through calibration against the Jason series. All altimeters launched to date have been nadir-viewing instruments. The spatial and temporal scales that are resolved by these missions are then determined by the spatial and temporal coverage offered by the satellite orbit. It is essential for many of the corrections that a non-sunsynchronous, repeat orbit pattern be used. The repeating polar orbits used are a trade-off between the period between repeat orbits, the equator separation between adjacent passes and the latitudinal range (inclination). The Jason series have a repeat orbit of 9.92 days and a pass separation of
18â•… System Design for Operational Ocean Forecasting
457
156.6€km (254 passes/cycle) and a latitudinal range of ±66.15°. The ERS/Envisat/ Sentinel series use a retrograde sunsynchronous repeat orbit period of 35 days, a pass separation of 79.9€km (501 passes/cycle) and a latitudinal range of ±81.45° (inclination 98.55°). The combination of multiple satellite missions is critical to providing improved temporal and spatial coverage to support SLA analyses and operational ocean forecasting (Ducet et€al. 2000; Pascual et€al. 2009). At present we have Jason1 and Jason2 in a tandem orbit and Envisat delivering near real-time products. An example of the passes obtained for a single day (1st January 2010) in the Australian region from these three missions is shown in Fig.€18.8a. The coverage per day is sparse compared with the spatial scales of the error covariances used in ocean forecasting which scales with the order of eddies, 100€km (Oke et€al. 2005, 2008; Martin et€al. 2007; Brasseur et€al. 2005; Cummings 2005). A larger observation window is employed in all operational systems in order to increase the spatial coverage and improve the quality of the least squares analysis. Examples of the coverage or a 5 day window and 11 day window overlayed on a background of SLA from the OceanMAPS system for the Tasman Sea is shown in Fig.€18.8b, c. A 5 day window shows gaps in coverage that are comparable or larger than the spatial scale of the ocean eddies. An 11 day window provides full coverage from Jason1 and Jason2 and partial coverage from Envisat and offers spatial coverage that is comparable to the scales of the ocean eddies (see Fig.€18.8c). The average altimetry coverage in the Australian region has been estimated for 1°â•›×â•›1° bins for single and multiple missions available in near real-time (see Fig.€ 18.9). The along-track observations have been normalised by thinning to a sampling rate of ~1 observation per 50€km which corresponds to a skip of 8 for Jason1 and Jason2 (i.e., 8â•›×â•›5.78€ kmâ•›~â•›46€ km) and a skip of 6 for Envisat, (i.e., 6â•›×â•›7.53€kmâ•›~â•›45€km). The thinning can be interpreted as the scale that might be used to construct so-called super-observations (e.g., Lorenc 1981; Purser et€ al. 2000). This is a formal method for compacting observations to reduce the redundancy of the raw observations relative to the target scales which in this case is chosen to be 1°â•›×â•›1° bins. Super-obs have a number of beneficial properties in practice including: increasing the homogeneity of coverage, reducing the observation space (i.e., computational cost) improve the condition of the matrix inversion in an analysis (see Daley 1991, p.€111). The average coverage obtained by the multi-satellite missions is a function of the orbit properties described earlier. In practice the coverage is also impacted by periods of communication failures and satellite manoeuvres or equipment failovers. This is evident in the coverage for Envisat (see Fig.€18.9b) which is impacted by the loss of satellite passes during maintenance between the 12th and 27th November 2009 (approximately half a repeat orbit period). The average coverage obtained from Jason1, Jason2 and Envisat (see Fig.€18.9d) over the open ocean ranges between 0.2 and 0.7 observations per 1°â•›×â•›1° bin per day with the mean coverage ~0.44. The coverage in the coastal regions is reduced in all cases and is effected by the quality control of observations and the proportion of the 1°â•›×â•›1° bin that is seawater. The average coverage of Jason1 (see Fig.€18.9a) does not exceed ~0.45. The
458
G. B. Brassington
Fig. 18.9↜渀 The average SLA observations per 1°â•›×â•›1° bin per day over the period 1 January 2009 to 1 March 2010 obtained from a Jason1. b ENVISAT. c Jason1 and Jason2 and d Jason1 and Jason2 and ENVISAT. The along-track observations have been normalised to ~1 observation/50€km
tandem mission of Jason1 and Jason2 (see Fig.€18.9c) show the overall improvement in the spatial distribution of coverage compared with Jason1. The normalized frequency distribution of SLA observation coverage corresponding to each Fig.€18.9a–d is plotted in Fig.€18.10. Due to the relatively coarse orbit sampling of Jason1, the 23% of 1°â•›×â•›1° bins are not sampled at all. With the introduction of the tandem missions Jason1 and Jason2 the number of 1°â•›×â•›1° bins that are not sampled drops to ~8%. The Envisat mission samples virtually all of the bins. The mode of each distribution curve is (0.15; 0.2; 0.35; 0.5) obs. per bin per day for Envisat; Jason1 (ignoring the zero peak); Jason1 and Jason2; Jason1, Jason2 and Envisat respectively. The number of obs. per day for all bins never exceed 0.73 for the three altimeters. The distribution shows that 50% of bins in the Australian region have a coverage of better than (0.15; 0.15; 0.3; 0.45) obs. per bin per day for Envisat; Jason1; Jason1 and Jason2; Jason1, Jason2 and Envisat respectively. Sea level anomaly products are processed in two to three modes dependent on the satellite which vary in quality and timeliness. The quality is determined by the
18â•… System Design for Operational Ocean Forecasting
459
Frequency distribution of altimetry observations
0.45
Jas1 Env
0.4
Jas1, Jas2, Env Jas1, Jas2
0.35
Frequency
0.3 0.25 0.2 0.15 0.1 0.05 0
0
0.1
0.2
0.3 0.4 0.5 Number of obs / 1°x 1° bin / day
0.6
0.7
0.8
Fig. 18.10↜渀 Normalized frequency distribution of altimetry observations per 1°â•›×â•›1° bin per day for the Australian region and satellite combinations shown in Fig.€18.9
quality to which the Geophysical Data Record (GDR) is estimated as well as the precision of other correction terms. Precise orbit positions are determined some time after real-time (e.g., 60 days) and are only relevant to hindcasting. Interim GDR (IGDR) target a faster orbit determination that is less accurate but can be delivered within 2–3 days (Jason series) and 3–5 days (Envisat). For the Jason series additional on-board satellite instrumentation allow an Operational GDR (OGDR) product to be delivered within 24€h of real-time. Due to instrument failure on Jason-1 the OGDR was unavailable but has been restored on the AVISO server. Following the launch of Jason-2 this product is now also available. A summary of events related to operational satellite altimetry can be found online from AVISO (http://www.aviso.oceanobs.com/no_cache/en/data/operational-news/index.html). In summary, the complete orbit of the Jason1 and Jason2 IGDR product is available between 3 and 12 days behind real-time, the complete orbit of Envisat IGDR product is available between 5 and 40 days behind real-time and Jason2 OGDR product is available 1–10 days behind real-time. Due to the reduced quality of the IGDR and OGDR products as well as the timeliness of the products it has been determined that the analysis performance from four real-time altimeters is equivalent to two delayed mode altimeters (Pascual et€al. 2009).
460
G. B. Brassington
18.6â•…Real-Time Forcing System The ocean is a forced dissipative system, where the forcing is largely through the flux of mass, heat and momentum through the air-sea interface. Atmospheric fluxes are available from operational numerical weather prediction systems that are mature and provide robust and consistent performance. However, the performance of atmospheric fluxes is relatively low compared with the state variables due to the limited direct observations of fluxes and errors in boundary conditions. The properties that influence the selection of atmospheric flux products and flux parameterizations for ocean prediction is summarized in Table€18.5. The oceans relatively large inertia, thermal inertia compared with the atmosphere mean that on short timescales air-sea fluxes are a relatively small perturbation to the ocean state at the surface and decays with depth. Even under extreme conditions, such as tropical cyclones, the surface temperature in the cold wake has been observed to be between 1°C and 6°C (Price 1981) and that the majority of the temperature change is due to entrainment and mixing of the ocean water masses in response to the momentum fluxes rather than changes due to surface heat flux. The momentum flux local to the atmospheric winds is largely transferred into the gravity waves which radiate from the source region. Local momentum transfer from high winds occurs through Langmuir circulation (McWilliams et€al. 1997), wave breaking (Melville 1996) and wave dissipation which persist during the wind event and is a function of wave age (Drennan et€al. 2003). A large fraction of the energy radiates away and dissipates through small scale turbulence and topographic interactions in locations remote from the winds. In the coastal region, the reduced volume of seawater is more sensitive to atmospheric fluxes. Storm-surge and coastal trapped waves (e.g., coastal Kelvin waves) are a result of horizontal mass flux into the coast as an Ekman response to the applied wind stress and lower atmospheric pressure (see Fig.€18.3h, i). Coastal upwelling of more dense, often cool and nutrient water masses are a response to mass flux away from the coast from an applied wind stress in the opposing direction (see Table 18.5↜渀 Properties of the atmospheric flux products that impact the ocean forecasting system Real-time forcing system Real-time surface fluxes Robust, well-defined and consistent performance Period, resolution and region of forecast systems Global, regional, sub-regional Forecast skill curve Boundary conditions, persisted SST, surface roughness Land-sea-ice masks Atmospheric boundary layer, cloud and radiation physics Observational constraints (e.g., scatterometry) Hindcast fluxes Performance during data assimilation Flux parameterisation Fixed boundary condition flux products Forecast atmospheric state with dynamic ocean boundary conditions Coupled air-sea or air-wave-sea Ocean dynamics Sensitivity of the ocean state to surface fluxes Sensitivity of ocean forecast error to surface flux errors
18â•… System Design for Operational Ocean Forecasting
461
Fig.€18.3a, b). The coastal region also has less heat capacity due to its reduced depth and is more sensitive to diurnal warming. The coast is also a region where atmospheric precipitation collects over land basins and can discharge from river mouth as a less dense freshwater plume. All of these processes have timescales comparable to those of the atmospheric weather and can produce observable changes to the ocean state and circulation of the coastal region. The skill of coastal ocean state forecasts is therefore sensitive to the skill of the atmospheric fluxes. The atmospheric fluxes for ocean forecasting systems are obtained from operational numerical weather prediction systems (e.g., GASP; Seaman et€al. 1995). A typical configuration for NWP is to perform an analysis every 6€h and a forecast every 12€ h. During ocean hindcasting, 24€ h of analysis fluxes can be composed of four 6€ h analyses. Common averaging periods for surface fluxes are 3€ h and 6€h. Atmospheric forecasting is typically composed of a suite of global and multiply nested regional prediction systems. In general, the horizontal resolution of the atmospheric models are coarser than the comparable ocean model and require regridding. One of the key discrepancies between models of differing resolutions is the mismatch in land-sea mask. A comparison of the land-sea masks from GASP (0.75°) and Ocean Forecast Australia Model (OFAM; Schiller et€al. 2008) (0.1°) is shown in Fig.€18.11. There are specific areas that show where some area correspond to New Land (Sea mask in the source and land in the target) or New Sea.
Fig. 18.11↜渀 A comparison of the land-sea masks of GASP (Seaman et al.) and OFAM (Schiller et al.). The four combinations both land (↜brown), both sea (↜blue), GASP land/OFAM sea (↜yellow) and GASP sea/OFAM land (↜red) ignoring ice masks in the Australian region
462
G. B. Brassington
In general, the magnitude of atmospheric fluxes across the land-sea boundaries is discontinuous largely due to the change in surface roughness, albedo and heat capacity. The magnitude of discontinuities varies with each variable and with the time of day. As the coastal ocean state is sensitive to atmospheric fluxes the fluxes over land need to be explicitly removed. Replacing the fluxes into New Sea locations is commonly performed by a Laplacian operation with the boundary conditions of the fluxes over sea points as this is computationally inexpensive. This method does not apriori preserve the alignment of winds or other properties with the coastline. There are many software packages that perform regridding including many of the earth system couplers (e.g., OASIS, Redler et€ al. 2010) however it is important to test these schemes and not assume that they will satisfy the requirements. An important property for regridding is to conserve the total integral of the field from the source grid to the target grid. The OASIS coupler has implemented the Spherical Coordinate Remapping and Interpolation Package (SCRIP; Jones 1999) as a regridding option. Another simple approach is to use an integral variable where the control volume integrals Eq.€18.1a are summed to form the discrete integral variable Eq.€18.1b, (Leonard 1995) which is exact at each cell interface and implicitly conserves the fluxes on the source grid. xi+.5 ¯ (18.1a) φi = φdx, i ∈ [1, I] xi−.5
ψj =
0 ψj −1 + x φ¯ j
j =0 j ∈ [1, I ]
(18.1b)
Regridding the original cell volume to a finer grid resolution, ∆X with an index kâ•›∈â•›[1,K] with the constraint that I∆xâ•›=â•›K∆X, can be performed by constructing an equivalent integral variable Ψk through interpolation of the integral variable as, 0 k=0 (18.2a) k = interp(ψ) k ∈ [1, K] The cell average values are then recovered as,
φ¯k =
k − k−1 , k ∈ [1, K] . X
(18.2b)
The exact integrals of the discrete integral variable are advantageous when ∆x is chosen to be an integer multiple of ∆X (i.e., ∆xâ•›=â•›n∆X) such that a subset of Ψk are equivalent to j. This formulation has been expressed for a uniform grid in Cartesian coordinates but this method is readily extended to non-uniform grids and other orthogonal curvilinear coordinate systems. It is also noteworthy that a centre point value is a second-order accurate estimate of a averaged over the cell volume, (Sanderson and Brassington 1998) so that the above method can be applied to most ocean general circulation models. An example of longwave radiation flux for the region surrounding Tasmania (Fig.€18.12a) for GASP. This field shows significant variability in the small scales
18â•… System Design for Operational Ocean Forecasting
463
Fig. 18.12↜渀 Longwave radiation heat flux from the GASP forecast a native resolution and b regridded to target resolution
as some of the physics is computed by a 1D radiation scheme but also shows front systems that have a step-like structure in the coarse resolution model. Regridding of coarse resolution information to finer resolution needs an algorithm that de-aliases. The integral variable can be used to de-aliase by performing interpolation for a subset of j such that jâ•›∈â•›[0:m:J] where J is an integer multiple of m. De-aliasing in multiple dimensions can be achieved by iteratively applying the integral variable interpolation in each dimension. The regridded longwave radiation onto the OFAM grid (Fig.€18.12b) is performed by applying the integral variable method for grid refinements of nâ•›=â•›~2 and a dealiasing parameter of mâ•›=â•›2 successively alternating in each dimension. The target grid resolutions are ∆xâ•›=â•›0.4°, 0.2° and 0.1°. Accurate direct observations of fluxes and flux budgets are sparse in time and space. The scatterometer provide an instantaneous estimate of stress which in practice has a limited weighting and impact to atmospheric analyses and forecasts. The monitoring of SST from multiple satellites and sensors together with the Argo and mooring arrays provide a basis for diagnosing errors in flux parameterisations. Atmospheric forecast errors grow rapidly and are constrained through data assimilation commonly on a 6€h cycle. Numerical weather prediction systems provide three alternative strategies for computing fluxes for ocean forecasting: (a) prescribed fluxes, (b) re-estimate the fluxes and (c) coupling. Numerical weather prediction systems currently persist SST analyses, may or may not have a dynamic surface roughness from a wave model and assume the ocean currents are negligible which will lead to a deterioration in skill in the surface fluxes in the forecasts. The next level of sophistication is to use the prescribed atmospheric state variables and replace the ocean boundary conditions by the forecast conditions using a bulk formula method (e.g., Large et€al. 1997). Two specific flaws to this approach include (a) the near-surface atmospheric state variables in the forecast have been forecasted using boundary layer turbulence models based on the original boundary conditions and (b) the ocean boundary conditions for SST may be less accurate or have greater bias than persisted SST. This is presently the case for the BLUElink OceanMAPS system compared with the RAMSSA (Beggs et€al. 2006). In part this is because the background errors from
464
G. B. Brassington
a forecast model are more difficult to define and the analysis for OceanMAPS is multi-variate and by definition will not fit the same SST observations as a univariate analysis. As the ocean forecasting systems continue to mature the performance gap is expected to close. This is also expected to be critical to achieve before more complex solutions of earth system coupling will yield the performance gains in operations (Brassington 2009).
18.7â•…Modelling The governing equations for the ocean are an extension of the Navier-Stokes for a thin layer on a rotating planet. The ocean state equation is an empirical formulae dependent on temperature, salinity and pressure. There are a number of assumptions that can be introduced to simplify the governing equations that exploit the properties of the ocean such as incompressible, hydrostatic which are either convenient for analytical, numerical or data analysis. Software designed to solve these governing equations are referred to as ocean general circulation models (OGCMs). A summary of the design choices in OGCM’s is summarised in Table€18.6. The prevalence of
Table 18.6↜渀 Properties of ocean modelling that result in unique design choices in ocean forecasting systems Ocean modelling Selection of model code Compressible/incompressible Hydrostatic/Nonhydrostatic, Non-Boussinesq/Non-Boussinesq Vertical coordinate system Community models NEMO, HYCOM, ROMS, MOM, … Non-eddy, eddy permitting eddies are ubiquitous in global ocean and eddy resolving Focii, horizontal mesh 0.1 a minimum Geostrophic turbulent closure and submesoscale High order, conservative advection schemes Coastal and bathymetric Vertical/Horizontal control Bathymetry products Practical bathymetry tuning Explicit tides or parameterised (more an assimilation challenge) Boundary conditions Open boundaries, radiation conditions Nonhydrostatic/Hydrostatic (Lattice-Boltzmann methods) Nesting 3:1, alignment of grids, common interfacial bathymetry Explicit/Implicit Numerical methods A-grid, B-grid, C-grid (Arakawa) and computational Order accuracy of methods performance Numerical stability Parallelism and scalability Turbulent parameterizations Surface and bottom boundary layers Tidal mixing Diapycnal mixing
18â•… System Design for Operational Ocean Forecasting
465
community ocean models mean that the first design choice is to select a community model. Community models have already made several design choices on the governing equations as well as to make some choices optional. Many ocean models could be categorised by their primary applications, climate modelling, coastal modelling however many community models aspire to be applicable to multi-scale modelling. It is important to be aware of these design choices and their potential impact to performance and range of applications. Starting from the position that a community model (e.g., Modular Ocean Model version 4, MOM4; Griffies et€al. 2003) has been selected the first step to implementation is to compile and configure the environment for the software on the system architecture. It is then important to optimise the performance and diagnose the scaling. This is a specialist area that can be architecture and compiler specific and is not discussed further. The next step in development is to define the model grid making use of the latest bathymetry products (e.g., Smith and Sandwell 1997). The target resolution in ocean forecasting is for eddy-resolving which is approximately development of operational oceanography and servicing in Australia. J Res Pract Inf Technol 39:151–164 Brassington GB, Hines A, Dombrowsky E, Ishizaki S, Bub F, Ignaszewski M (2010a) Short- to medium range ocean forecasts: delivery and observational requirements. In: Hall J, Harrison DE, Stammer D (eds) Proceedings of OceanObs’09: sustained ocean observations and information for society vol 1, Venice, Italy, 21–25 September 2009. ESA Publication WPP-306. doi: 10.5270/OceanObs09.pp08 Brassington GB, Summons N, Lumpkin R (2010b) Observed and simulated Lagrangian and eddy characteristics of the East Australian current and Tasman sea. Deep Sea Res Res Part II. doi: 10.1016/j.dsr2.2010.10.001 Chelton DB, Ries JC, Haines BJ, Fu L-L, Callahan PS (2001) Satellite altimetery. In: Fu L-L, Cazenave A (eds) Satellite altimetry and earth sciences. Academic Press, San Diego, pp€1–131 Chelton DB, Schlax MG, Samelson RM, de Szoeke RA (2007) Global observations of large oceanic eddies. Geophys Res Lett 34:L15606. doi:10.1029/2007GL030812 Chua B, Bennett AF (2001) An inverse ocean modeling system. Ocean Model 3:137–165 Cummings JA (2005) Operational multivariate ocean data assimilation, Q J R Meteorol Soc 131:3583–3604 Daley R (1991) Atmospheric data analysis. Cambridge University Press, New York, p€457 Davidson F, Allen A, Brassington GB, Breivik O, Daniel P, Kamachi M, Sato S, King B, Lefevre F, Sutton M, Kaneko H (2009) Application of GODAE ocean current forecasts to search and rescue and ship routing. Oceanography 22(3):176–181 Dee DP (2005) Bias and data assimilation. Q J R Meteorol Soc 131:3323–3343 Desai SD, Haines BJ, Case K (2003) Near real time sea surface height anomaly products for Jason-1 and Topex/Poseidon user manual. NASA, JPL D-26281, p€13 De Szoeke RA, Samelson R (2002) The duality between the Boussinesq and Non-Boussinesq hydrostatic equations of motion. J Phys Oceanogr 12:2194–2203 Dombrowsky E, Bertino L, Brassington GB, Chassignet EP, Davidson F, Hurlburt HE, Kamachi M, Lee T, Martin MJ, Mei S, Tonani M (2009) GODAE systems in operation. Oceanography 22(3):80–95 Donlon CJ, Minnett P, Gentemann C, Nightingale TJ, Barton IJ, Ward B, Murray J (2002) Towards improved validation of satellite sea surface skin temperature measurements for climate research. J Climate 15(4):353–369 Drennan WM, Graber HC, Hauser D, Quentin C (2003) On the wave age dependence of wind stress over pure wind seas. J Geophys Res 108(C3):8062. doi:10.1029/2000JC000715 Ducet N, Le Traon P-Y, Reverdin G (2000) Global high resolution mapping of ocean circulation from TOPEX/Poseidon and ERS-1/2. J Geophys Res 105(19):19477–19498 Ducowicz JK (1997) Steric sea level in the Los Alamos POP code—Non-Boussinesq effects, numerical methods in atmospheric and oceanic modeling. In: Lin C, Laprise R, Richie H (eds) The Andre Robert memorial volume, Canadian meteorological and oceanographic society, NRC Research Press, Ottawa, p€533–546
484
G. B. Brassington
Evensen G (2003) The Ensemble Kalman Filter theory and practical implementation. Ocean Dyn 118:1–23 Fan Y, Ginis I, Hara T (2009) The effect of wind–wave–current interaction on air–sea momentum fluxes and ocean response in tropical cyclones. J Phys Oceanogr 39:1019–1034 Freeland H, Roemmich D, Garzoli S, LeTraon P, Ravichandran M, Riser S, Thierry V, Wijffels S, Belbeoch M, Gould J, Grant F, Ignaszewski M, King B, Klein B, Mork K, Owens B, Pouliquen S, Sterl A, Suga T, Suk M, Sutton P, Troisi A, Vélez-Belchi P, Xu J (2010) Argo—a decade of progress. In: Hall J, Harrison DE, Stammer D (eds) Proceedings of OceanObs’09: sustained ocean observations and information for society vol 2, Venice, Italy, 21–25 September 2009. ESA Publication WPP-306. doi: 10.5270/OceanObs09.cwp.32 Garrett C, Müller P (2008) Supplement to “extreme events”. Bull Am Meteorol Soc 89:ES45ES56. doi:10.1175/2008BAMS2566.2 (by Chris Garrett and Peter Müller Bull Am Meteorol Soc 89:1733) Gaspari G, Cohn SE (1999) Construction of correlation functions in two and three dimensions. Q J R Meteorol Soc 125:723–757 Goni G, Meyers G, Ridgeway K, Behringer D, Roemmich D, Willis J, Baringer M, Ichi I, Wijffels S, Reverdin G, Rossby T (2010) Ship of opportunity program. OceanObs’09 ESA Special Publication (in press) Greatbatch RJ (1994) A note on the representation of steric sea level in models that conserve volume rather than mass. J Geophys Res 99:12767–12771 Griffies SM, Harrison MJ, Pacanowski RC, Rosati A (2003) A technical guide to Mom4 Gfdl ocean group technical Report No. 5, NOAA/Geophysical Fluid Dynamics Laboratory Version prepared on 23 Dec 2003 Hackett B, Comerma E, Daniel P, Ichikawa H (2009) Marine oil pollution prediction. Oceanography 22(3):168–175 Hernandez F, Bertino L, Brassington GB, Chassignet E, Cummings J, Davidson F, Drevillon M, Garric G, Kamachi M, Lellouche J-M, Mahdon R, Martin MJ, Ratsimandresy A, Regnier C (2009) Validation and intercomparison studies within GODAE. Oceanography 22(3):128–143 Hurlburt HE, Brassington GB, Drillet Y, Kamachi M, Benkiran M, Bourdalle-Badie R, Chassignet EP, Jacobs GA, Le Galloudec O, Lellouche JM, Metzger EJ, Oke PR, Pugh TF, Schiller A, Smedsted OM, Tranchant B, Tsujino H, Usui N, Wallcraft AJ (2009) High resolution global and basin-scale ocean analyses and forecasts. Oceanography 22(3):110–127 Johnson GC, McTaggart KE (2010) Equatorial Pacific 13°C Water eddies in the eastern subtropical South Pacific Ocean. J Phys Oceanogr 40:226–236 Jones PW (1999) First- and second-order conservative remapping schemes for grids in spherical coordinates. Mon Weather Rev 127:2204–2210 Kamachi M, Kuragano T, Ichikawaj H, Nakamura H, Nishina A, Isobe A, Ambe D, Arais M, Gohda N, Sugimoto S, Yoshita K, Sakura T, Ubold F (2004) Operational data assimilation system for the Kuroshio South of Japan: reanalysis and validation. J Oceanogr 60:303–312 Kepert JD (2009) Covariance localisation and balance in an Ensemble Kalman filter. Q J R Meteorol Soc 135:1157–1176 Large WG, Danabasoglu G, Doney SC, McWilliams JC (1997) Sensitivity to surface forcing and boundary layer mixing in a global ocean model: annual-mean climatology. J Phys Oceanogr 27:2418–2447 Leonard BP, Lock AP, Macvean MK (1995) The nirvana scheme applied to one-dimensional advection. Int J Numerical Methods Heat Fluid Flow 5:341–377 Le Provost C (2001) Ocean tides. In: Fu L-L, Cazenave A (eds) Satellite altimetry and earth sciences. Academic Press, San Diego, pp€267–303 Le Traon P-Y (2011) Satellites and operational oceanography. In: Schiller A, Brassington GB (eds) Operational oceanography in the 21st century. doi:10.1007/978-94-007-0332-2-18, Springer, Dordrecht, pp 29–54 Lorenc AC (1981) A global three-dimensional multivariate statistical interpolation scheme. Mon Weather Rev 109:701–721
18â•… System Design for Operational Ocean Forecasting
485
Lorenc AC (2003) The potential of the ensemble Kalman filter for NWP—A comparison with 4DVar. Q J R Meteorol Soc 129:3183–3203 Martin MJ, Hines A, Bell MJ (2007) Data assimilation in the FOAM operational short-range ocean forecasting system: a description of the scheme and its impact. Q J R Meteorol Soc 133:981– 995 McDougall TJ, Greatbatch RJ, Lu Y (2002) On the conservation equations in oceanography: how accurate are Boussinesq ocean models? J Phys Oceanogr 32:1574–1584 McInnes KM, Leslie LM, McBride JL (1992) Numerical simulation of cut-off lows on the Australian east coast: sensitivity to sea surface temperature. Int J Climatol 12:1–13 McPhaden MJ, Delcroix T, Hanawa K, Kuroda Y, Meyers G, Picaut J, Swenson M (2001) The El Niño/Southern Oscillation (ENSO) observing system. In: Koblinski C, Smith N (eds) Observing the ocean in the 21st century. Australian Bureau of Meteorology, Melbourne, pp€231–246 McPhaden MJ, Meyers G, Ando K, Masumoto Y, Murty VSN, Ravichandran M, Syamsudin F, Vialard J, Yu L, Yu W (2009) RAMA: the research moored array for African-Asian-Australian monsoon analysis and prediction. Bull Am Meteorol Soc 90:459–480 McWilliams JC, Sullivan PP, Moeng C-H (1997) Langmuir turbulence in the ocean. J Fluid Mech 334:1–30 Melville WK (1996) The role of wave breaking in air- sea interaction. Ann Rev Fluid Mech 28:279–321 Moore A (2011) Adjoint applications. In: Schiller A, Brassington GB (eds) Operational oceanography in the 21st century. doi:10.1007/978-94-007-0332-2-18, Springer, Dordrecht, pp 351–379 Murphy AH (1988) Skill scores based on the mean square error and their relationships to the correlation coefficient. Mon Weather Rev 116:2417–2424 Murray RJ (1996) Explicit generation of orthogonal grids for ocean models. J Comput Phys 126:251–273 Oke PR, Schiller A, Griffin DA, Brassington GB (2005) Ensemble data assimilation for an eddyresolving ocean model of the Australian region. Q J R Meteorol Soc 131:3301–3311 Oke PR, Brassington GB, Griffin DA, Schiller A (2008) The Bluelink ocean data assimilation system (BODAS). Ocean Model 21:46–70 Ourmieres Y, Brankart L, Berline L, Brasseur P, Verron J (2006) Incremental analysis update implementation into a sequential ocean data assimilation system. J Atmos Ocean Technol 23:1729–1744 Pascual A, Boone C, Larnicol G, Le Traon PY (2009) On the quality of real-time altimeter gridded fields: comparison with in situ data. J Atmos Ocean Technol 26:556–569 Price JF (1981) Upper ocean response to a hurricane. J Phys Ocean 11:153–175 Prandle D, Flemming NC (eds) (1998) The science base of EuroGOOS. EuroGOOS Publication No. 6, 1998, EG97.14, unpaginated Purser RJ, Parrish D, Masutani M (2000) Meteorological observational data compression; an alternative to conventional “super-obbing.” NCEP Office Note 430, p€13. Available online at http:// www.emc.ncep.noaa.gov/officenotes/FullTOC.html Ravichandran M (2011) In situ ocean observing system. In: Schiller A, Brassington GB (eds) Operational oceanography in the 21st century. doi:10.1007/978-94-007-0332-2-18, Springer, Dordrecht, pp 55–90 Redler R, Valcke S, Ritzdorf H (2010) OASIS4—a coupling software for next generation earth system modelling. Geosci Model Dev 3:87–104 Reinaud JN, Dritschel DG (2002) The merger of vertically offset quai-geostrophic vortices. J Fluid Mech 469:287–315 Rixen M, Book JW, Orlic M (2009) Coastal processes: challenges for monitoring and prediction. J Mar Syst 78(1):S1–S2. ISSN 0924-7963, doi:10.1016/j.jmarsys.2009.01.006, Nov 2009 Robinson I (2006) Satellite measurements for operational ocean models. In: Chassignet EP, Verron J (eds) Ocean weather forecasting: an integrated view of oceanography. Springer, Netherlands, pp€147–189
486
G. B. Brassington
Sanderson B, Brassington G (1998) Accuracy in the context of a control-volume model. Atmosphere-Ocean 36:355–384 Sandery PA, Brassington GB, Freeman J (2010) Adaptive nonlinear dynamical initialization. J Geophys Res. doi: 10.1029./2010JC006260 Schiller A, Oke PR, Brassington GB, Entel M, Fiedler RAS, Griffin DA, Mansbridge JV, Meyers GA, Ridgway KR, Smith NR (2008) Eddy-resolving ocean circulation in the Asia-Australian region inferred from an ocean reanalysis effort. Prog Oceanogr 76:334–365 Seaman R, Bourke W, Steinle P, Hart T, Embery G, Naughton M, Rikus L (1995) Evolution of the Bureau of Meteorology’s global assimilation and prediction system, Part 1: analyses and initialization. Aust Met Mag 44:1–18 Smith N, Lefebvre M (1997) The Global Ocean Data Assimilation Experiment (GODAE). Monitoring the oceans in the 2000s: an integrated approach. International Symposium, Biarritz, 15–17 Oct 1997 Smith WHF, Sandwell DT (1997) Global seafloor topography from satellite altimetry and ship depth soundings. Science 277:1956–1962 Sobel D (1995) Longitude: the true story of a Lone Genius who solved the greatest scientific problem of his time. Walker & Company, New York, p€216 Spiegel EA, Veronis G (1960) On the Boussinesq approximation for a compressible fluid. Astrophys J 131:442–447 Stanski HR, Wilson LJ, Burrows WR (1989) Survey of common verification methods in meteorology. World Weather Watch Tech. Report No.8, WMO/TD No.358, WMO, Geneva, p€114 Stammer D (1997) Global characteristics of ocean variability from regional TOPEX/POSEIDON altimeter measurements. J Phys Oceanogr 27:1743–1769 Taylor KE (2001) Summarizing multiple aspects of model performance in a single diagram. J Geophys Res 106(D7):7183–7192 Thompson KR, Wright DG, Lu Y, Demirov E (2006) A simple method for reducing seasonal bias and drift in eddy resolving ocean models. Ocean Model 13:109–125 Zaron E (2011) Basics of data assimilation and inverse methods. In: Schiller A, Brassington GB (eds) Operational oceanography in the 21st century. doi:10.1007/978-94-007-0332-2-18, Springer, Dordrecht, pp 321–350 Zeytounian RKh (2003) Joseph Boussinesq and his approximation: a contempory view. C R Mec 331:575–586
Chapter 19
Integrating Coastal Models and Observations for Studies of Ocean Dynamics, Observing Systems and Forecasting John L. Wilkin, Weifeng G. Zhang, Bronwyn E. Cahill and Robert C. Chant
Abstract╇ In coastal oceanography, simulation models are used to a variety of ends. Idealized studies may address particular dynamical processes or features of coastline and bathymetry; reproducing the circulation in a geographical region can compliment studies of ecosystems and geomorphology; and models may be employed to simulate observing systems and to forecast oceanic conditions for practical operational needs. Frequently, the interplay between multiple forcing mechanisms, geographic detail, stratification, and nonlinear dynamics, is significant, and this demands that ocean models for coastal applications are capable of representing a comprehensive suite of dynamical processes. Drawing on a series of recent modelbased studies of the inner to mid-shelf region of the Middle Atlantic Bight (MAB) we illustrate, by example, these methodologies and the breadth of dynamical processes that influence coastal ocean circulation. We demonstrate that the recent introduction of variational methods into coastal ocean simulation is a development that greatly enhances our ability to integrate models with data from the evolving coastal ocean observatories for the purposes of improved ocean prediction, adaptive sampling and observing system design.
19.1â•…Introduction The discharge of rivers to continental shelf seas represents an important mechanism by which human activities in urban watersheds impact the neighbouring marine environment. Biogeochemical, sediment, and ecosystem processes that determine the ultimate fate of nutrients and pollutants delivered into the coastal ocean by river sources depends on the pathways and time scales of dispersal of these buoyant discharges. How coastal models, in conjunction with observations, can be used to study these circulation processes is illustrated here by example, by reviewing results J. L. Wilkin () Institute of Marine and Coastal Sciences, Rutgers, The State University of New Jersey, New Brunswick, NJ, USA e-mail:
[email protected] A. Schiller, G. B. Brassington (eds.), Operational Oceanography in the 21st Century, DOI 10.1007/978-94-007-0332-2_19, ©Â€Springer Science+Business Media B.V. 2011
487
488
J. L. Wilkin et al.
from a recent series of model-based studies of the Hudson River outflow into New York Bight. On many coasts, the flux of freshwater from rivers or groundwater first enters an estuary where it mixes with more salty waters of oceanic origin before reaching the adjacent shelf sea. The salinity of the estuary discharge can be sufficiently low that horizontal buoyancy gradients are a significant force influencing the plume circulation. A classical view of the ensuing dynamics is that the buoyancy force balances the Coriolis force, and the outflow turns to the right (in the northern hemisphere) and forms a narrow coastal current a few internal Rossby radii in width trapped against the coast. If the front that defines the outer extent of the low salinity water reaches the sea floor then the plume becomes bottom-attached and details of the coastal bathymetry strongly influence the plume trajectory. Alternatively, if the low salinity discharge is confined to a relatively thin surface layer the plume is described as surface-advected (Yankovsky and Chapman 1997) and may be more responsive to local wind forcing. Whether a plume falls into the surface-advected or bottom-trapped regime, or transitions from one regime to the other, depends on river discharge, bathymetry, and mixing within the surface and bottom boundary layers. It is often the case that the freshwater transport of the coastal current is less than the freshwater flux out of the estuary, particularly during episodes of elevated river discharge, and this leads to the formation of a pronounced low salinity bulge near the estuary outflow. The across shelf scale of the bulge can be several times the width of the coastal current, especially for a surface-advected plume. The low buoyancy of the bulge evolves an anti-cyclonic circulation that significantly prolongs the duration that water discharged from the estuary is retained in the vicinity of the estuary mouth. Laboratory rotating tank experiments have shown that the coastal current can receive as little as one third of the estuary outflow (Avicola and Huq 2003), or in extreme circumstances the recirculation can pinch off from the coastal current and for a period of time direct all flow into the bulge (Horner-Devine et€al. 2006). Numerical model studies show that the ratio of coastal current transport to estuary discharge decreases as the flow becomes increasingly non-linear as characterized by the Rossby number, i.e. the ratio of inertial to rotational forces (Fong and Geyer 2002; Nof and Pichevin 2001). Thus river flow rate, vertical turbulent mixing within the estuary and on the shelf, bathymetric detail, stratification, non-linear dynamics, and wind forcing are all factors that influence river plume dispersal characteristics. Shelf-wide alongshelf mean currents established by regional winds (Fong and Geyer 2002) or by upstream or offshore remote forcing further influence the circulation (Zhang et€al. 2009a). Consequently, ocean models that seek to simulate interactions between river discharges and the adjacent inner shelf must be quite comprehensive in the suite of dynamical processes that they represent. In this article we demonstrate the capabilities of one such model, the Regional Ocean Modelling System (ROMS; www.myroms.org), by summarizing results from a sequence of studies of the Hudson River’s discharge into the coastal ocean based on efforts during the Lagrangian Transport and Transformation Experiment (LaTTE) (Chant et€al. 2008). The Hudson River watershed is highly industrialized,
19â•… Integrating Coastal Models and Observations
489
and the LaTTE field program included observations—following the river plume associated with the spring freshest in the years 2004, 2005 and 2006—of phytoplankton and zooplankton assemblages, and natural and human-source nutrients, organic matter, and metal contaminants. An emphasis of the project was to investigate how the plume’s physical structure influenced biogeochemical processes. Key processes in this regard include mixing that dilutes salinity and influences certain chemical reactions, light levels that affect photochemistry, and residence times and transport pathways that can impact rates of bioaccumulation and modify where regions of net export of particulate suspended matter might occur. Dynamical and computational features of ROMS that are pertinent to the LaTTE simulations (and coastal processes in general) are described in Sect.€19.2, and revisited in Sect.€19.4 in a discussion of aspects of New York Bight (NYB) regional dynamics worthy of further analysis. Section€19.3 describes modelling approaches we have taken to address specific scientific objectives. Section€ 19.3.1 considers forward simulations initialized from climatology and forced with observed river flows and an atmospheric forecast model used for short-term forecasting for adaptive sampling during the LaTTE field experiments, and idealized studies of how the plume responds to the wind. Multi-year simulations to examine long-term transport and dispersal pathways, and the mean dynamics of the circulation, are presented in Sect.€ 19.3.2. Section€ 19.3.3 describes a reanalysis of the 2006 LaTTE season using Incremental Strong Constraint 4-Dimensional Variational Data Assimilation (IS4DVAR) to adjust initial conditions to each daily forecast cycle, and gives a brief overview of how variational methods might also be employed to assist observing system operation. In Sect.€ 19.5 it is summarized how the studies described here collectively illustrate how coastal models are being increasingly integrated with the growing network of regional coastal ocean observing systems to better understand coastal ocean processes, and improve ocean predictions.
19.2â•…Regional Ocean Modelling System 19.2.1 Dynamical and Numerical Core ROMS solves the hydrostatic, Boussinesq, Reynolds-averaged Navier-Stokes equations in terrain-following vertical coordinates. It employs a split-explicit formulation whereby the 2-dimensional continuity and barotropic momentum equations are advanced using a much smaller time step than the 3-dimensional baroclinic momentum and tracer equations. The ROMS computational kernel is described elsewhere (Shchepetkin and McWilliams 2005, 2009a, b) and will not be detailed here, but we do note several aspects of the kernel that are particularly attractive for coastal ocean simulation. These include a formulation of the barotropic mode equations that accounts for the non-uniform density field so as to reduce aliasing and coupling errors associated with the split-explicit method (Higdon and de Szoeke 1997) in terrain-following coordinates. Temporal-weighted averaging of the barotropic mode prevents alias-
490
J. L. Wilkin et al.
ing of unresolved signals into the slow baroclinic mode while accurately representing barotropic motions resolved by the baroclinic time step (e.g. tides and coastaltrapped waves). Several features of the kernel substantially reduce pressure-gradient force truncation error that has been a long-standing problem in terrain following coordinate ocean models. A finite-volume, finite-time-step discretization for the tracer equations improves integral conservation and constancy preservation properties associated with the variable free surface, which is important in coastal applications where the free surface displacement represents a significant fraction of the water depth. A positive-definite MPDATA (multidimensional positive definite advection transport algorithm) advection scheme (Smolarkiewicz 1984) is available, which is attractive for biological tracers and sediment concentration. A monotonized, highorder vertical advection scheme for sinking of sediments and biological particulate matter integrates depositional flux over multiple grid cells so it is not constrained by the CFL criterion (Warner et€al. 2008a). Interested readers are referred to Shchepetkin and McWilliams (2009b) for a thorough review of the choices of algorithmic elements that make ROMS particularly accurate and efficient for high-resolution simulations in which advection is strong, and currents, fronts and eddies are approximately geostrophic—characteristics of mesoscale processes in the coastal ocean and adjacent deep sea.
19.2.2 Vertical Turbulence Closure ROMS provides users with several options for the calculation of the vertical eddy viscosity for momentum and eddy diffusivity for tracers. In the majority of recent ROMS coastal applications the choice of vertical turbulence-closure formulation has been either (1) a k-profile parameterization (KPP) for both surface and bottomboundary layers (Large et€al. 1994; Durski et€al. 2004), (2) Mellor-Yamada level 2.5 (MY25) (Mellor and Yamada 1982), or (3) the generic length-scale (GLS) method (Umlauf and Burchard 2003) which encompasses a suite of closure and stability function options. The KPP scheme specifies turbulent mixing coefficients in the boundary layers based on Monin-Obukhov similarity theory, and in the interior principally as a function of the local gradient Richardson number (Large et€al. 1994; Wijesekera et€al. 2003). The KPP method is diagnostic in the sense it does not solve a time evolving (prognostic) equation for any of the elements of the turbulent closure, whereas the MY25 and GLS schemes are of the general class of closures where two prognostic equations are solved—one for turbulent kinetic energy and the other related to turbulence length scale. Warner et€ al. (2005) describe the implementation of the GLS formulation in ROMS, and contrast the performance of the various GLS sub-options (representing different treatments of the turbulent length scale) and the historically widely used MY25 scheme. They find that the differing schemes lead to differences in the vertical eddy mixing profiles, but the net impact on profiles of model state variables
19â•… Integrating Coastal Models and Observations
491
(velocities and tracers) is relatively minor. Wijesekera et€al. (2003) reach similar conclusions, but note that results for KPP tend to be less similar to GLS and MY25, which are quite alike. Warner et€al. (2005) found that suspended sediment concentrations in their sediment transport model are much more sensitive to the choice of closure than is salinity in estuarine mixing simulations. In the LaTTE simulations we use the GLS k-kl closure option, which is essentially an implementation of MY25 within the GLS conceptual framework.
19.2.3 Forcing 19.2.3.1â•…Air-Sea Fluxes Air-land-sea contrasts, orography, upwelling, fog, and tidal mixing over variable bathymetry in the coastal ocean can all contribute to creating wind and air temperature conditions at sea level that have much shorter time and length scales than typically occur further offshore or in the open ocean. Accordingly, coastal ocean simulations benefit from the availability of spatially and temporally well-resolved meteorological forcing and accurate parameterization of air-sea momentum and heat fluxes. Surface atmospheric forcing in the LaTTE simulations made use of two sets of marine boundary layer products derived from atmospheric models. The short time scale simulations (Sect.€19.3.1) and IS4DVAR reanalysis (Sect.€19.3.3) used marine boundary conditions (downward long-wave radiation, net shortwave radiation, 10-m wind, 2-m air temperature, pressure and humidity) at 3-hourly intervals from the North American Mesoscale model (NAM; Janjic 2004)—a 12€ km resolution 72-h forecast system operated by the National Centers for Environmental Prediction (NCEP). The multiyear simulations (Sect.€19.3.2) used marine boundary layer conditions taken from the North American Regional Reanalysis (Mesinger 2006)—a 25-km resolution 6-hourly interval data assimilative reanalysis product. Air-sea fluxes of momentum and heat were computed using standard bulk formulae (Fairall et€al. 2003) from the atmospheric model based marine boundary layer conditions in conjunction with the sea surface temperature from ROMS. 19.2.3.2â•…River Inflows and Open Boundary Conditions In coastal Regions of Freshwater Influence (ROFI) (Hill 1998), lateral buoyancy input from rivers produces density gradients that are principally horizontal, which leads to relatively weak vertical stability compared to the vertical stratification generated from comparable surface air-sea buoyancy fluxes. Density stratification in ROFI subsequently arises from the baroclinic adjustment of these density gradients, and destratification and restratification can occur rapidly in response to changing rates of vertical mixing associated with wind forcing and tides (which may have significant spring-neap variability in intensity).
492
J. L. Wilkin et al.
On some coasts, groundwater discharge directly to the coastal ocean or freshwater input from numerous small streams and rivers can be significant, but in the NYB terrestrial buoyancy input is overwhelmingly from large rivers, and predominantly from the Hudson. For river input to the LaTTE model we used daily average observations of river discharge from U.S. Geological Survey gauging stations on the Hudson and Delaware rivers, modified to include ungauged portions of the watershed following Chant et€al. (2008). At the open boundaries to the LaTTE model domain, simple Orlanski-type radiation conditions were applied to tracers (temperature and salt) and 3-D velocity. Our emphasis here on the buoyancy driven circulation associated with the Hudson River plume allows this simplification with its implicit neglect of the influence of remote sources of freshwater and heat. Open boundary sea level and depth-averaged velocity variability was set using the Chapman (1985) and Flather (1976) schemes to radiate surface gravity waves while also imposing tidal harmonic velocity variability derived from a regional tide model (Mukai et€al. 2002). In the long multiyear simulations (Sect.€19.5), the boundary depth averaged velocity was augmented with the estimate of mean southwestward current on the shelf derived by Lentz (2008) based on long-term current-meter observations and momentum balance arguments.
19.2.4 Sub-Models for Interdisciplinary Studies ROMS incorporates a set of sub-models for interdisciplinary applications that are integrated with the dynamical kernel. Among these are several ecosystem models formulated in terms of Eulerian functional groups wherein 3-D tracers representing nutrients, plankton, zooplankton, detritus, etc., expressed in terms of some common currency (usually equivalent nitrogen concentration), are advected and mixed according to the same transport equations as the dynamic tracers. Haidvogel et€al. (2008) give an overview of examples of these models, which range in complexity from a four component nitrogen-based (NPZD) model (Powell et€al. 2006; Moore et€al. 2009) to a carbon based bio-optical model (EcoSim) (Bissett et€al. 1999; Cahill et€al. 2008) with a spectrally resolved light field and more than 60 state variables representing four phytoplankton, five pigments, five elements, bacteria, dissolved organic matter, and detritus. A Community Sediment Transport Model (CSTM; Warner et€al. 2008a) and wave model (SWAN, Surface Waves in the Nearshore; Booij et€al. 1999) are integrated with ROMS for studies of sediment dynamics and circulation in nearshore environments; wave radiation stresses are included in the momentum equations and wave-current interaction that enhances bottom stress is included in the bottom boundary layer dynamics. A user-defined set of non-cohesive sediment classes is tracked, with differential erosion and deposition of the various size classes contributing to the evolution of a multi-level sediment bed with varying layer thickness, porosity, and mass, which allows computation of bed morphology and stratigraphy. The application of the ROMS/ SWAN/CSTM to studies of sediment morphology, sorting and transport in an idealized tidal inlet and Massachusetts Bay are presented by Warner et€al. (2008a).
19â•… Integrating Coastal Models and Observations
493
19.3╅ROMS Simulations of the New York Bight Region for LaTTE 19.3.1 Dispersal of the Plume During High River Discharge The ROMS model domain for LaTTE (Fig.€ 19.1) extends from south of Delaware Bay to eastern Long Island, and from the New Jersey and New York coasts to roughly the 70-m isobath. The model has 30 vertical layers and horizontal grid resolution is 1€km. In spring 2005 and 2006 the model was used to forecast circulation in the NYB in support of LaTTE field observation programs (Foti 2007). Figure€19.2 shows vis-
Fig. 19.1↜渀 The model domain (↜black line) and locations of observations used in the 4DVAR data assimilation (Sect.€19.3.4). Bathymetry of the New York Bight is in greyscale; black dash lines are model isobaths in metres; yellow star in the location of Ambrose Tower; green squares indicate the five HF radar stations
494
J. L. Wilkin et al.
Fig. 19.2↜渀 Left: Visible imagery from Ocean Colour Monitor (OCM) instrument aboard Indian IRS-P4 satellite, and MODIS instrument aboard NASA Terra satellite showing turbid waters associated with the Hudson River discharge, and vectors of surface current from HF radar (CODAR), on two days during the spring 2005 LaTTE experiment. Right: Modelled surface salinity and currents at the corresponding times
ible satellite imagery of the Hudson River plume as it enters the NYB on two days in 2005 overlaid with vectors showing surface current observed by HF-radar, and the modelled velocity and surface salinity and corresponding time—surface salinity being a proxy for the signature of the river source waters. A recirculating bulge of low salinity water is being over-run by a renewed ebb tide discharge of Hudson River estuary waters. Figure€19.3 compares satellite observed absorption at wavelength 488€nm from Oceansat-1 (a proxy for relative chlorophyll abundance and the presence ζ of river source water) with the modelled equivalent freshwater thickness δfw = −h (So − S(z))/So dz, where S is salinity, h is the water depth, and z = ζ is the sea surface. If it were possible to locally “unmix” the water column into two layers of salinities zero and So, the thickness of the fresh water layer would be δfw . This depicts the horizontal extent of freshwater dispersal more faithfully than sea surface salinity. Here we use a reference salinity Soâ•›=â•›32. Figures€ 19.2 and 19.3, and further model-data comparisons in Zhang et€ al. (2009a), indicate that fundamental features of the river plume circulation such as the across and along-shelf length scales, the extent of the freshwater bulge, veloc-
19â•… Integrating Coastal Models and Observations
495
Fig. 19.3↜渀 Top row: Modelled equivalent freshwater thickness in meters (↜left) and satellite observed absorption at wavelength 488€nm from Oceansat-1 (↜right) showing the patterns of influence of Hudson River source waters. Bottom: Observed and modelled salinity along the northernmost west-east transect indicated in the top right panel
ity patterns, and the transport pathway from the harbor to the coastal current, are similar in model and observations. Figure€19.4 shows the time evolution of simulated equivalent freshwater thickness during the spring freshet of 2005. From 1 to 7 April the river discharge exceeded 2,500€m3/s, or more than four times the annual mean, and peaked at 6,500€m3/s on 4 April. Initially, southward downwelling favourable winds drive the river plume rapidly southward along the New Jersey coast, but this flow is abruptly arrested on 4 April with the onset of northward upwelling favourable winds. This causes the river flow during peak discharge to form a large low-salinity recirculating bulge located predominantly on the northern side of the Hudson Shelf Valley. From 10 to 15 April a period of weak and variable winds associated with the sea breeze phenomenon enable the bulge to partially drain into a New Jersey coastal current. The return of upwelling winds on April 17 drives more low salinity water eastward and detaches the bulge
496
J. L. Wilkin et al.
Fig. 19.4↜渀 Modelled equivalent freshwater thickness in meters during the spring freshet of 2005 and winds observed at Ambrose Tower in the New York Bight apex
from the estuary discharge that previously fed it. In the week that follows, sustained winds further disperse the plume as the river discharge drops and the freshet ends. The influence of wind direction and strength on Hudson River plume dispersal has been considered in some detail (Choi and Wilkin 2007) using the same model but for idealized winds and freshet river discharge. Figure€19.5 contrasts the plume behaviour commencing from the same initial conditions (Fig.€19.5a) in response to winds from differing directions (Fig.€19.5d–g) sustained for 3 days. The sensitivity described for the April 2005 simulations is confirmed. Southward winds, and to a lesser extent eastward winds, favour New Jersey coastal current formation. Northward winds eliminate the buoyancy-driven coastal current, disperse the bulge eastward and drive flow along the Long Island coast. Westward winds hamper the discharge from the Hudson River estuary, leading to a build up of low salinity water in New York Harbor. In the absence of wind forcing, the low salinity bulge continues to grow in volume in agreement with the modelling and tank experiments noted in Sect.€19.1. In the LaTTE region then, winds play a crucial role in determining the fate of material transported by the Hudson River to the inner shelf. Choi and Wilkin (2007) also considered the influence of river discharge magnitude on the relative contribution of buoyancy and wind forcing to the momentum balance of the river plume. They found that relatively modest wind speeds of order 5€m/s are sufficient to overwhelm buoyancy forcing during typical non-freshet conditions. It follows then that relatively short timescale variability in river discharge and weather conditions could lead to different dispersal patterns for the freshet in any
19â•… Integrating Coastal Models and Observations
497
Fig. 19.5↜渀 Surface salinity of the Hudson River plume showing sensitivity of plume trajectory to wind during a high discharge event (3,000€m3/s)
given year, and this was indeed found to be the case in the three LaTTE field seasons (Chant et€al. 2008). In 2004, river waters were first transported southward in a modest coastal current, and then dispersed eastward in the surface Ekman layer associated with strong upwelling winds; 2005 was characterized by strong bulge formation and sea breeze activity as described above; while in 2006 unusually large river discharge fed a coastal current that flooded the New Jersey inner shelf with low salinity water, but this flow subsequently detached from the coast leading to significant across-shelf transport in the region south of the Hudson Shelf valley.
19.3.2 Shelf-Wide Transport and Dispersal Pathways The preceding studies revealed that while some processes act to trap river plume water near the apex of the NYB (i.e. the recirculating bulge, and coastal current flow reversals) others disperse it widely (i.e. fast coastal currents and offshore winddriven Ekman transport). Therefore the duration that river source waters dwell in the vicinity of the coastline can be quite variable, and questions arise as to where these waters eventually go. To examine the ultimate fate of Hudson River source waters on time scales much longer than the spring freshet, we conducted multi-year simulations using the same
498
J. L. Wilkin et al.
model configuration but with modified open boundary inflow/outflow transport conditions and meteorology forcing from NARR. The open boundary conditions were adapted to acknowledge that on inter-annual timescales the mid and outer New Jersey shelf is flushed by a southwestward alongshelf mean flow. An analysis of long term current meter observations and the mean momentum balance (Lentz 2008) indicates the depth-averaged along-shelf current is roughly proportional to water depth; this provides a convenient relationship upon which to base the time mean boundary transports to which we add the tidally varying currents. The modelled mean circulation for 2005–2006 (Zhang et€al. 2009a) is shown in Fig.€19.6. Buoyancy input from the Hudson River dominates flow in the apex of the NYB by driving the anticyclonic recirculation (a local maximum in sea surface height, SSH) associated with the low salinity bulge. This feature is sustained in the annual mean because it is the consequence not only of the spring freshet but also
Fig. 19.6↜渀 Mean SSH (sea surface height) contours (a, top), and velocity at sea surface (b, centre) and 20-m depth (c, bottom) over the 2-year period 2005–2006
19â•… Integrating Coastal Models and Observations
499
of high discharge events that can occur throughout the year. In the 3 years of the LaTTE program, the peak discharge actually occurred in July 2006 following heavy rains across all of New York State. Transport is eastward along the Long Island coast, but this current ultimately detaches from the coast and reverses in the face of the mean flow that enters from the eastern open boundary. On the mid to outer shelf the flow is to the southwest, largely parallel to isobaths, and deflected by the Hudson Shelf Valley as evidenced by the currents at 20€ m (Fig.€19.6c). The influence of the valley extends throughout the water column and affects SSH. In the very apex of NYB the flow at 20€m is toward New York Harbor, indicating that the HSV serves as a conduit for shoreward flow that is vertically mixed and entrained into the estuary outflow and bulge recirculation. Away from the coast the surface currents (Fig.€19.6b) are dominated by southward wind-driven Ekman flow. A New Jersey coastal current is not readily apparent in the annual mean. Zhang et€al. (2009a) show it is prominent in spring and fall, moderate in winter, but overwhelmed by upwelling winds in the summer. To avoid the ambiguity of reference salinity in lengthy simulations and to distinguish the Hudson River from other freshwater sources, Zhang et€ al. (2009a) introduce a passive tracer with unit concentration in the modelled Hudson River source and follow it to obtain an unambiguous measure of the dispersal pathways. Figure€19.7 shows the flux of Hudson River source water identified by its tracer signature across a set of arcs centred on the Harbor entrance. The qualitative features noted above are again evident. The New Jersey coastal current is clearly very tightly trapped against the coast, which partly explains why it is not conspicuous in Fig.€19.6a, b. Figure€19.7 quantifies the volume transports across sectors of the
Fig. 19.7↜渀 Left: Two-year averaged, vertically integrated freshwater flux (↜thick black lines) across arcs of radius 20, 40, 60, 80, 100, and 120€km (numbered 1–6) centred at the entrance to New York Harbor (↜star). Right: Freshwater transport (m3/s) across the segments of the arcs on either side of the Hudson Shelf Valley (↜gray dashed–dotted line), and across the valley itself
500
J. L. Wilkin et al.
arcs split at the HSV. In this 2-year mean, we see that river discharge is entirely to the shelf north of the HSV but that the majority of this flow subsequently crosses the valley within the general region of the recirculating bulge. Once south of the valley, the outflow is partitioned between the coastal current and a weaker but much broader across-shelf pathway guided by the south flank of the HSV. The latter current feature has been noted from HF radar surface current observations (Castelao et€al. 2008). Despite initially entering the coastal ocean along the New York coast, the Hudson River discharge is thus ultimately dispersed to the mid and outer shelf on the south side of the Hudson Shelf Valley. Biogeochemical observations during LaTTE (Moline et€al. 2008) support the notion that the coastal current is typically supplied with biogeochemically processed water that has circulated around the bulge’s perimeter rather than newly discharged water from the estuary. In an example of the type of controlled dynamical analysis one can conduct with a model, Zhang et€al. (2009a) separately withdrew individual forcing processes to examine the effect of each on the circulation. Their results are shown in Fig.€19.8, which should be compared to Fig.€19.6a, b for the full physics solution.
Fig. 19.8↜渀 Mean SSH (sea surface height) contours (↜left) and surface currents and magnitude (↜right) over the 2-year period 2005–2006 for three simulations with changes to the full physics configuration shown in Fig.€19.6. Top row: Outer shelf boundary forcing removed. Middle row: Wind stress removed. Bottom row: Bathymetry of Hudson Shelf Valley filled in
19â•… Integrating Coastal Models and Observations
501
Without the remotely forced along-shelf mean flow the bulge recirculation remains, but the across-shelf surface flow is more eastward being the result solely of Ekman transport and not combined with geostrophic southward flow. In the absence of wind forcing the bulge is more intense, in accordance with the results of Fong and Geyer (2002) who found that along-shore transport driven by wind arrests continuous growth of bulge recirculation. As in the full physics case, part of this recirculation feeds flow on the south side of the HSV, but without winds the downstream flow is largely at mid-shelf parallel to the coast and does not disperse to the outer shelf. Zhang et€al. (2009a) explored whether the Hudson Shelf Valley impacts circulation by simply removing the valley from the model bathymetry. Figure€19.8 shows that in the No Valley case the SSH signature of the bulge is substantially weakened, and surface velocity shows far more of the estuary outflow enters the NJ coastal current. In an extension of their passive tracer approach for following Hudson River waters, Zhang et€al. (2010a) employ the concept of ‘mean age’ (Deleersnijder et€al. 2001) to determine the transit time from river source to shelf ocean. If we denote the equation governing the transport of a passive tracer with concentration C by ∂C + ∇ · (uC) = ∇ · (K · ∇C) ∂t
then an ‘age concentration’ tracer α can be introduced satisfying ∂α + ∇ · (uα) = ∇ · (K · ∇α) + C ∂t
where the last term on the right causes α to increase in proportion to the concentration of river source water present. The concentration of the tracers in the river source are Câ•›=â•›1 and αâ•›=â•›0. The ‘mean age’ (Deleersnijder et€al. 2001) is given by a(x, t) = α(x, t)/C(x, t) and describes the duration it has been on average since the waters at a given position and time (x, t) entered the domain at the river source. Figure€19.9 illustrates how mean age evolves in a simulation where the river tracer release commenced on 13 March. It takes some 4–5 days for river water to reach the bulge circulation, and water on the southwest side of the bulge is clearly older than water to the north. On March 18 an increase in river discharge a few days previously introduces a surge of younger water that forms a sharp gradient in mean age across the western edge of bulge. In 7 days none of the river water has escaped the bulge. In regions the passive tracer has not reached the mean age is undefined. Zhang et€ al. (2010a) show that mean age patterns in the 2005 LaTTE period mimic an age proxy determined from a ratio of satellite observed water leaving radiance that expresses the relative concentration of CDOM (Coloured Dissolved Organic Matter) to phytoplankton. CDOM is the dominant optical constituent in river source waters and has high absorption at 490€nm but it subsequently photodegrades whereas phytoplankton concentration (with chlorophyll-a spectral peak at 670€nm) increases as the plume ages, so the CDOM decrease and phytoplankton
Fig. 19.9↜渀 Modelled mean age (colour scale in days) for a simulation commencing on 13-March
502 J. L. Wilkin et al.
19â•… Integrating Coastal Models and Observations
503
increase produces a spectral shift in the remote sensing reflectance. Zhang et€ al. (2010a) found a robust empirical relationship between simulated age and observed reflectance ratio that has promise for estimating river water age in the NYB—a property of relevance to rates of biogeochemical transformation of river source organic matter and pollutants (Moline et€al. 2008).
19.3.3 Data Assimilation and Observing System Design The NYB is among the most densely observed coastal oceans in the world, having been the target of pioneering deployments of new observing instruments including a cabled observatory (Glenn and Schofield 2003), surface current measuring highfrequency radar (CODAR) (Kohut et€al. 2006) and autonomous underwater vehicles (gliders) (Schofield et€al. 2007). To these systems and regular satellite imagery, LaTTE added moorings, surface drifters, and towed undulating CTD instruments deployed from the research vessels Cape Hatteras and Oceanus. These data and the sustained operation of much of the instrumentation make the NYB an attractive location to explore the integration of observation and modelling capabilities through advanced data assimilation. The locations of LaTTE 2006 in situ observations are shown in Fig.€19.1. CODAR coverage was near complete from Long Island to Delaware Bay and out to the 40€m isobath, with some gaps in the apex of NYB. There were satellite SST data from approximately four passes each day, cloudiness permitting. Here we use data assimilation (DA) for state estimation; namely, to obtain an analysis for initializing subsequent forecasts so as to enhance short-term forecast skill. This approach is common practise in Numerical Weather Prediction (NWP). We use a 4-dimensional (time-dependent) variational (4DVAR) method for DA, which is one among many possible approaches but again one that draws on experience in advanced NWP. We use the so-called Incremental Strong Constraint (IS4DVAR) formulation (Courtier et€ al. 1994) whose implementation in ROMS is described in detail elsewhere (Broquet et€al. 2009; Powell et€al. 2008; Zhang et€al. 2010b). IS4DVAR minimizes a cost function expressing the mismatch between observations and the model state at each observation location and time, summed over an analysis interval. Our implementation uses a 3-day interval—short enough for the linearization assumption of the incremental formulation to hold, but long enough for the model physics (embodied in the adjoint and tangent linear models) to exert the strong constraint interconnection (covariance) of model state variables. The control variables of the DA are the initial conditions to each 3-day analysis, with the intervals overlapped so as to generate initial conditions each day to launch a new 72-h forecast. IS4DVAR does not explicitly allow for model error as would, for example, representer-based or weak constraint 4DVAR (Bennett 2002; Courtier 1997). Errors in model physics, numerics, meteorological forcing and boundary conditions are incorporated into the model background error covariance. The observations are assigned error variances appropriate to the observation source.
504
J. L. Wilkin et al.
Fig. 19.10↜渀 Added skill introduced by data assimilation for analysis and forecast periods for individual forecast variables. Results are ensemble average of 60 forecast cycles. Vertical bars on symbols indicate 95% confidence intervals. Vertical dashed lines denote the boundary between analysis window and forecast window
Our reanalysis was conducted after the data were gathered, but we describe a DA and forecast system that could have operated in real-time because glider and vessel data are telemetered to shore. Lessons learned from this study on practical issues of data timeliness, quality control, and configuration of the IS4DVAR algorithm on a broad, shallow shelf, with significant tides have been incorporated in the Experimental System for Predicting Shelf and Slope Optics (ESPreSSO1) that currently runs operationally for the Mid-Atlantic Bight and encompasses the LaTTE domain. The value that DA adds to the forecast system can be evaluated by considering how well observations are forecast prior to their assimilation on later analysis cycles. We quantify this with a DA skill metric S = 1 − (RMS after DA /RMS before DA )
where RMS is the root-mean-square of model-observation mismatch weighted by observational error. For 60 days of simulation spanning LaTTE 2006 we have multiple 1-day, 2-day, etc. forecasts that may be combined into ensemble estimates for increasing forecast window. Figure€ 19.10 shows the skill for different variables 1╇
ESPreSSO results may be viewed at www.myroms.org/applications/espresso.
19â•… Integrating Coastal Models and Observations
505
when all available data are assimilated (black lines), and when selected data categories are withdrawn from the analysis step (coloured lines). Forecast times less than zero are in the analysis interval, and show the ability of the system to match observations and model prior to launching the forecast. As forecast time proceeds the skill declines, but note that Sâ•›=â•›0 does not say the model has no utility at all, merely that assimilation no longer adds any advantage to the model predictive skill. For temperature, DA adds skill to the forecast out to some 10–15 days, for salinity 5–10 days, and for velocity about 2–3 days. The more rapid decline in skill for velocity compared to tracers reflects the shorter autocorrelation timescales for velocity and that it is inherently less predictable. Not surprisingly, withdrawing data diminishes skill for that variable, i.e. without HF-radar data the velocity skill falls, and without satellite SST the temperature skill falls. However, there can be a modest increase in skill for other variables, e.g. salinity forecast skill is slightly higher when SST are not assimilated. We interpret this as the DA system not needing to reconcile glider and satellite temperatures and having rather more freedom to adjust initial salinity to improve the salinity analysis; recall that all the variables are dynamically linked through the strong constraint of the adjoint and tangent linear models. Overall, skill is best when all data are included, and therefore diversity in the data sources is to be preferred. Details of the ROMS IS4DVAR configuration for LaTTE with respect to background error covariance and the pre-processing of observations are discussed by Zhang et€ al. (2010b), who also examine surface versus subsurface skill, and the influence of errors in surface forcing on system performance. A further application of variational methods in ocean modelling is adjoint sensitivity analysis, which allows some inference of observation locations that are likely to have greater impact on the DA analysis. Studies using adjoint sensitivity in coastal oceanography are still relatively few compared to meteorology and mesoscale and gyre-scale oceanography, but Moore et€al. (2009) examine how upwelling, eddy kinetic energy and baroclinic instability in the California Current are affected by surface forcing on seasonal timescales. Here we present some results due to Zhang et€al. (2009b) who use the adjoint of the LaTTE model to reveal the spatial and temporal distribution of ocean model state variables that are “dynamically upstream” to features of coastal circulation. A characteristic of New Jersey coastal ocean dynamics is that significant SST variability is driven by along-shore winds (Chant 2001; Münchow and Chant 2000). Zhang et€al. (2009b) considered this process by introducing a scalar function that expresses SST anomaly variance averaged over a localized area adjacent to the coast t2 1 2 J = (Ts − T s ) dAdt, 2(t2 − t1 )A t1 A where Ts is SST and T s is its temporal mean; this definition considers temperature anomaly within an area A during a set time interval. Here, the time period is chosen to be the last three hours of the simulation time window. Defining J in quadratic form prevents the cancellation of positive and negative anomalies.
506
J. L. Wilkin et al.
Temperature, salinity and velocity outside region A affect J through transport (advection and diffusion) and dynamics (baroclinic pressure gradients, stratification, turbulent mixing). Denoting the 4-dimensional ocean state (↜T, S, u, v, ζ ) by a vector Φ, it can be shown that ∂J /∂—representing the dependence of J on the ocean state—is the solution of the ROMS adjoint model integrated backward in time and forced by ∂J /∂T computed from the forward model. See Zhang et€al. (2009b) for details. Although J is a scalar, ∂J /∂ has the same dimension as Φ, i.e. the entire ocean state through time, which emphasizes that all the surrounding ocean can potentially project on to SST variance in A. This adjoint sensitivity concept can be grasped, qualitatively, from an example: Fig.€19.11 maps the sensitivity of J to surface temperature, i.e. ∂J /∂T at zâ•›=â•›0, over the 3 days that precede the interval t1 to t2 over which J is defined, for the cases of downwelling and upwelling winds. The sequence proceeds backwards in time from day 3 to day 0. We have already demonstrated that southward (downwelling) winds favour coastal current formation, and for this case (Fig.€19.11, top row) the adjoint sensitivity advances from region A (delineated by the black box) back along the trajectory of the coastal current to New York Harbor. In the upwelling wind case (Fig.€ 19.11, bottom row), surface temperatures in preceding times have very little impact of SST variance in A. This is because the coastal current is not dynamically upstream in this situation; rather, surface temperatures depend more on source waters drawn from below the surface. The final panel on the right shows ∂J /∂T at tâ•›=â•›0 along a vertical cross-section slightly south of region A, and confirms that J is sensitive to remote subsurface temperatures during upwelling. While these results have a ready qualitative interpretation, adjoint sensitivity quantifies the dependence and immediately indicates where “upstream” is. Zhang et€al. (2009b) further quantify the relative importance of other state variables by contrasting the magnitude of ∂J /∂T with ∂J /∂S, ∂J /∂u, etc. One can immediately see the potential for this information to assist observing system operation. By identifying the timing and location of ocean conditions having significant influence on the subsequent evolution of specific circulation features (characterized by some chosen J), adjoint sensitivity indicates where, when and what observations are likely to have greater impact in a 4DVAR assimilation system. In a companion paper, Zhang et€al. (2010c) extend this approach using socalled representers, also based on variational methods, to examine the information content of a set of observations such as might be gathered routinely on a repeat transect occupied by an autonomous vehicle, or by a sustained cabled observatory.
19.4╅Processes and Dynamics for Further Study 19.4.1 Air-Sea and Wave-Current Interaction The results described above all utilize essentially the same model configuration options emphasized in Sect.€19.2, but the LaTTE program identified roles for some
Fig. 19.11↜渀 Sensitivity of J to surface temperature at different times during the 3-day period. Top row: Southward down-welling winds. Bottom row: Northward upwelling winds. Panel at right shows sensitivity at day 0 (upwelling case) on a vertical section. (See the text for discussion)
19â•… Integrating Coastal Models and Observations 507
508
J. L. Wilkin et al.
dynamical processes that were not incorporated in the model physics employed here that are worthy of incorporation in future model-based studies. In the NYB, sea-land-breeze system (SLBS) activity can be pronounced during spring (Hunter et€ al. 2007, 2010) when ocean temperatures are still cool but the land is warming. Since this is precisely the time of year when river discharge peaks with the spring freshet, atmosphere-ocean interactions fundamental to SLBS dynamics are likely important to achieving realistic simulations of the plume circulation. Furthermore, mid-summer SLBS activity further south on the Jersey Shore is influenced by SST changes associated with wind-driven coastal upwelling (Bowers 2004). Full synchronous coupling of ROMS with an atmospheric forecast model has the potential to improve both ocean and atmosphere forecasts when SLBS conditions occur, and this capability has been added to ROMS by coupling to the COAMPS (Coupled Ocean Atmosphere Prediction System) (Warner et€al. 2008b) and WRF (Weather Research and Forecasting) models. Surface wind waves mediate air-sea interaction by modifying drag and hence net momentum exchange, plus surface wave radiation stress, Stokes drift and wavecurrent interaction processes in the bottom boundary layer drag are important in the ocean momentum balance itself. It was noted in Sect.€19.2.4 that these dynamical processes are now incorporated in ROMS, including the option to synchronously couple with the SWAN wave model. Studies of the Hudson plume that employed higher resolution than the 1€km grid used here and placed greater emphasis on processes in shallow waters near the coast (inside the 15-m isobath) or at the leading edge of the plume, may demonstrate that inclusion of these dynamics are important to faithful simulation of the plume evolution.
19.4.2 Ecosystem-Optics and Heating Interaction Like most coastal ocean models, ROMS assumes constant absorption coefficients for shortwave radiation (Paulson and Simpson 1977) leading to a vertical exponential decay in internal solar heating. But optical properties of coastal waters can be far from spatially uniform, and observations during LaTTE exhibited distinct regions of turbid water associated with the river plume, motivating Cahill et€al. (2008) to use the EcoSim model (Sect.€19.2.4) to examine coupling between shortwave radiation attenuation, buoyancy and photosynthesis. The solar heating parameterization was modified to make shortwave absorption dependent on the concentration of river source freshwater as a proxy for increased attenuation in the plume. The feedback between solar heating and vertical stratification was sufficient to modify the buoyancy driven circulation and mixed layer depth. This in turn raised concentrations of chlorophyll, detritus and coloured dissolved organic matter (CDOM) in the upper water column increasing attenuation of photosynthetically active radiation (PAR) and further impacting phytoplankton growth. Simulations with full ecosystem-absorption-heating feedback (i.e. spectrally resolved 3-dimensional radiative absorption determined by optically active con-
19â•… Integrating Coastal Models and Observations
509
stituents in the water column) have shown differences in simulated temperature can be as much as 2°C warmer at the surface, and correspondingly cooler some 10€m deeper, in the Hudson River plume. The associated changes in plume trajectory and ecosystem dynamics alter net export of particulate matter to mid shelf waters. Incorporating these optical properties into the 4-dimensional ocean state is a natural future step to enhance data assimilation in coastal ocean models.
19.5â•…Summary We have described a series of model-based studies of circulation in the New York Bight region that utilize data from a sustained coastal ocean observing system complemented by extensive in situ observations from the LaTTE project. Observations are used to evaluate the performance of traditional forward simulations where the model formulation is treated as an initial and boundary value problem. Circulation on the New Jersey inner shelf, and especially within the NYB, is strongly locally driven and direct forward simulations with ROMS are quite skilful—a result we attribute to the model being comprehensive and accurate in the suite of dynamical processes it represents and the numerical algorithms it employs, suitably configured in terms of bathymetric and coastline detail, and driven by meteorological, hydrological and tidal forcing with sufficient resolution and accuracy. Using forward model simulations we have seen that the NYB circulation is particularly responsive to wind forcing, how buoyancy dynamics contribute to the retention of river source waters in the NYB apex through formation of a persistent anti-cyclonic recirculation, and that the model can be used to quantify this residence time by incorporating an age tracer. Long simulations reveal the pathways by which Hudson River borne material is ultimately dispersed across the New Jersey shelf. Moving beyond traditional forward simulations, we have illustrated how coastal models are now being increasingly integrated with the growing network of regional coastal ocean observing systems. The creation of variational complements to the ROMS nonlinear forward model (i.e. the ROMS adjoint and tangent linear models) has enabled the implementation of 4-dimensional variational data assimilation in coastal ocean analysis with an attendant improvement in forecast skill. Variationalbased methods have further capabilities beyond data assimilation, through helping inform adaptive sampling strategies and observing system design targeted at improving predictive skill.
References Avicola G, Huq P (2003) The role of outflow geometry in the formation of the recirculating bulge region in coastal buoyant outflows. J Mar Res 61:411–434 Bennett AF (2002) Inverse modeling of the ocean and atmosphere. Cambridge University Press, Cambridge, p€234
510
J. L. Wilkin et al.
Bissett WP, Walsh JJ, Dieterle DA, Carder KL (1999) Carbon cycling in the upper waters of the Sargasso sea: I. Numerical simulation of differential carbon and nitrogen fluxes. Deep Sea Res Part I Oceanogr Res Pap 46:205–269 Booij N, Ris RC, Holthuijsen LH (1999) A third-generation wave model for coastal regions. Part I: model description and validation. J Geophys Res 104(C4):7649–7666 Bowers L (2004) The effect of sea surface temperature on sea breeze dynamics along the coast of New Jersey. M.S. thesis, Rutgers University, New Brunswick Broquet G, Edwards CA, Moore AM, Powell BS, Veneziani M, Doyle JD (2009) Application of 4D-variational data assimilation to the California current system. Dyn Atmos Oceans 48:69–92 Cahill B, Schofield O, Chant R, Wilkin J, Hunter E, Glenn S, Bissett P (2008) Dynamics of turbid buoyant plumes and the feedbacks on near-shore biogeochemistry and physics. Geophys Res Lett 35, L10605. doi:10.1029/2008GL033595 Castelao RM, Schofield O, Glenn S, Chant RJ, Kohut J (2008) Cross-shelf transport of fresh water on the New Jersey shelf. J Geophys Res 113, C07017. doi:10.1029/2007JC004241 Chant RJ (2001) Evolution of near-inertial waves during an upwelling event on the New Jersey inner shelf. J Phys Oceanogr 31:746–764 Chant RJ, Wilkin J, Zhang W, Choi B-J, Hunter E, Castelao R, Glenn S, Jurisa J, Schofield O, Houghton R, Kohut J, Frazer TK, Moline MA (2008) Dispersal of the Hudson River plume in the New York Bight: synthesis of observational and numerical studies during LaTTE. Oceanography 21(4):148–161 Chapman DC (1985) Numerical treatment of cross-shelf open boundaries in a barotropic ocean model. J Phys Oceanogr 15:1060–1075 Choi B-J, Wilkin JL (2007) The effect of wind on the dispersal of the Hudson River plume. J Phys Oceanogr 37:1878–1897 Courtier P, Thépaut J-N, Hollingsworth A (1994) A strategy for operational implementation of 4DVAR using an incremental approach. Q J R Meteorol Soc 120:1367–1388 Courtier P (1997) Dual formulation of four-dimensional variational assimilation. Q J R Meteorol Soc 123:2449–2461 Deleersnijder E, Campin J-M, Delhez EJM (2001) The concept of age in marine modelling: I. Theory and preliminary model results. J Mar Syst 28:229–267 Durski S, Glenn SM, Haidvogel D (2004) Vertical mixing schemes in the coastal ocean: comparison of the level 2.5 Mellor-Yamada scheme with an enhanced version of the K-profile parameterization. J Geophys Res 109, C01015. doi:10.1029/2002JC001702 Fairall CW, Bradley EF, Hare JE, Grachev AA, Edson J (2003) Bulk parameterization of air–sea fluxes: updates and verification for the COARE algorithm. J Climate 16:571–591 Flather RA (1976) A tidal model of the northwest European continental shelf. Memoires Soc R Sci Liege Ser 6(10):141–164 Fong DA, Geyer WR (2002) The alongshore transport of freshwater in a surface-trapped river plume. J Phys Oceanogr 32:957–972 Foti G (2007) The Hudson River plume: utilizing an ocean model and field observations to predict and analyze physical processes that affect the freshwater transport. M.S. Thesis, Rutgers University, New Brunswick Glenn SM, Schofield O (2003) Observing the oceans from the COOLroom: our history, experience, and opinions. Oceanography 16:37–52 Haidvogel D, Arango H, Budgell W, Cornuelle B, Curchitser E, Di Lorenzo E, Fennel K, Geyer WR, Hermann A, Lanerolle L, Levin J, McWilliams JC, Miller A, Moore AM, Powell TM, Shchepetkin AF, Sherwood C, Signell R, Warner JC, Wilkin J (2008) Ocean forecasting in terrain-following coordinates: formulation and skill assessment of the regional ocean modeling system. J Comput Phys 227:3595–3624 Higdon RL, de Szoeke RA (1997) Barotropic-baroclinic time splitting for ocean circulation modeling. J Comput Phys 135:31–53 Hill AE (1998) Buoyancy effects in coastal and shelf seas. In: Robinson AR, Brink KH (eds) The sea. The global coastal ocean, vol€10. Harvard University Press, London, pp€21–62
19â•… Integrating Coastal Models and Observations
511
Horner-Devine AR, Fong DA, Monismith SG, Maxworthy T (2006) Laboratory experiments simulating a coastal river outflow. J Fluid Mech 555:203–232 Hunter E, Chant R, Bowers L, Glenn S, Kohut J (2007) Spatial and temporal variability of diurnal wind forcing in the coastal ocean. Geophys Res Lett 34, L03607. doi:10.1029/2006GL028945 Hunter E, Chant R, Wilkin J, Kohut J (2010) High-frequency forcing and sub-tidal response of the Hudson River plume. J Geophys Res 115, C07012. doi:10.1029/2009JC005620 Janjic ZL (2004) The NCEP WRF core. 20th Conference on Weather Analysis and Forecasting/16th Conference on Numerical Weather Prediction, Seattle. Am Meteorol Soc. http://ams.confex. com/ams/84Annual/techprogram/paper_70036.htm Kohut JT, Roarty HJ, Glenn SM (2006) Characterizing observed environmental variability with HF doppler radar surface current mappers and acoustic doppler current profilers: environmental variability in the coastal ocean. IEEE J Ocean Eng 31:876–884 Large WG, McWilliams JC, Doney SC (1994) A review and model with a nonlocal boundary layer parameterization. Rev Geophys 32:363–403 Lentz SJ (2008) Observations and a model of the mean circulation over the Middle Atlantic Bight continental shelf. J Phys Oceanogr 38:1203–1221 Mellor GL, Yamada T (1982) Development of a turbulence closure model for geophysical fluid problems. Rev Geophys Sp Phys 20:851–875 Mesinger F, DiMego G, Kalnay E, Mitchell K, Shafran P, Ebisuzaki W, Jovic D, Woollen J, Rogers E, Berbery E, Ek M, Fan Y, Grumbine R, Higgins W, Li H, Lin Y, Manikin G, Parrish D, Shi W (2006) North American regional reanalysis. B Am Meteorol Soc 87:343–360 Moline MA, Frazer TK, Chant R, Glenn S, Jacoby CA, Reinfelder JR, Yost J, Zhou M, Schofield O (2008) Biological responses in a dynamic Buoyant River plume. Oceanography 21(4):70–89 Moore AM, Arango HG, Di Lorenzo E, Miller AJ, Cornuelle BD (2009) An adjoint sensitivity analysis of the southern California Current circulation and ecosystem. J Phys Oceanogr 39:702–720 Mukai AY, Westerink JJ, Luettich RA, Mark D (2002) Eastcoast 2001, a tidal constituent database for the western North Atlantic, Gulf of Mexico and Caribbean Sea. Tech. Rep. ERDC/CHL TR-02-24, p€196 Münchow A, Chant RJ (2000) Kinematics of inner shelf motion during the summer stratified season off New Jersey. J Phys Oceanogr 30:247–268 Nof D, Pichevin T (2001) The ballooning of outflows. J Phys Oceanogr 31:3045–3058 Paulson CA, Simpson JJ (1977) Irradiance measurements in the upper ocean. J Phys Oceanogr 7:952–956 Powell TM, Lewis CVW, Curchister EN, Haidvogel DB, Hermann AJ, Dobbins EL (2006) Results from a three-dimensional, nested biological-physical model of the California Current system and comparisons with statistics from satellite imagery. J Geophys Res 111, C07018. doi:10.1029/2004JC002506 Powell BS, Arango HG, Moore AM, Di Lorenzo E, Milliff RF, Foley D (2008) 4DVAR data assimilation in the Intra-Americas sea with the regional ocean modeling system (ROMS). Ocean Model 25:173–188 Schofield O, Bosch J, Glenn SM, Kirkpatrick G, Kerfoot J, Moline MA, Oliver M, Bissett P (2007) Bio-optics in integrated ocean observing networks: potential for studying harmful algal blooms. In: Babin M, Roesler C, Cullen JJ (eds) Real time coastal observing systems for ecosystem dynamics and harmful algal blooms. UNESCO, Valencia, pp€85–108 Shchepetkin A, McWilliams J (2005) The regional oceanic modeling system (ROMS): a split-explicit, free-surface, topography-following-coordinate oceanic model. Ocean Model 9:347–404 Shchepetkin A, McWilliams J (2009a) Computational kernel algorithms for fine-scale, multi-process, long-term oceanic simulations. In: Temam R, Tribbia J (Guest eds) Computational methods for the ocean and the atmosphere. In: Ciarlet PG (ed) Handbook of numerical analysis, vol€14. Elsevier, Amsterdam, pp€119–182. doi:10.1016/S1570-8659(08)01202-0 Shchepetkin A, McWilliams J (2009b) Correction and commentary for ocean forecasting in terrain-following coordinates: formulation and skill assessment of the regional ocean modeling
512
J. L. Wilkin et al.
system. J Comp Phys 228:8985–9000 (by Haidvogel et al., J Comp Phys 227:3595–3624). doi:10.1016/j.jcp.2009.09.002 Smolarkiewicz PK (1984) A fully multidimensional positive-definite advection transport algorithm with small implicit diffusion. J Comput Phys 54:325–362 Umlauf L, Burchard H (2003) A generic length-scale equation for geophysical turbulence models. J Mar Res 61:235–265 Warner J, Sherwood C, Arango H, Signell R (2005) Performance of four turbulence closure models implemented using a generic length scale method. Ocean Model 8:81–113 Warner JC, Sherwood CR, Signell RP, Harris CK, Arango HG (2008a) Development of a threedimensional, regional, coupled wave, current, and sediment-transport model. Comput Geosci 34:1284–1306. doi:10.1016/j.cageo.2008.02.012 Warner JC, Perlin N, Skyllingstad ED (2008b) Using the Model Coupling Toolkit to couple earth system models. Environ Model Softw 23:1240–1249 Wijesekera HW, Allen JS, Newberger PA (2003) Modeling study of turbulent mixing over the continental shelf: comparison of turbulent closure schemes. J Geophys Res 108(C3):3103 Yankovsky AE, Chapman DC (1997) A simple theory for the fate of buoyant coastal discharges. J Phys Oceanogr 27:1386–1401 Zhang W, Wilkin J, Chant R (2009a) Modeling the pathways and mean dynamics of river plume dispersal in New York Bight. J Phys Oceanogr 39:1167–1183. doi:10.1175/2008JPO4082.1 Zhang W, Wilkin J, Levin J, Arango H (2009b) An adjoint sensitivity study of buoyancy- and wind-driven circulation on the New Jersey inner shelf. J Phys Oceanogr 39:1652–1668. doi:10.1175/2009JPO4050.1 Zhang W, Wilkin J, Schofield O (2010a) Simulation of water age and residence time in the New York Bight. J Phys Oceanogr. doi:10.1175/2009JPO4249.1 Zhang W, Wilkin J, Arango H (2010b) Towards an integrated observation and modeling system in the New York Bight using variational methods. Part I: 4DVAR data assimilation. Ocean Model 35:119–133. doi:10.1016/j.ocemod.2010.08.003 Zhang W, Wilkin J, Levin J (2010c) Towards an integrated observation and modeling system in the New York Bight using variational methods. Part II: representer-based observing system design. Ocean Model 35:134–145. doi:10.1016/j.ocemod.2010.06.006
Chapter 20
Seasonal and Decadal Prediction Oscar Alves, Debra Hudson, Magdalena Balmaseda and Li Shi
Abstract╇ Dynamical seasonal prediction has grown rapidly over the last decade or so. At present, a number of operational centres issue routine seasonal forecasts produced with coupled ocean-atmosphere models. These require real-time knowledge of the state of the global ocean since the potential for climate predictability at seasonal time scales resides mostly in information provided by the ocean initial conditions, in particular the upper thermal structure. The primary aim of the coupled model is to predict sea surface temperature variability and how this variability impacts regional climate through large scale teleconnections. This paper reviews recent advances in dynamical seasonal prediction using coupled ocean-atmosphere models. It discusses the sources of predictability at seasonal time scales, the probabilistic nature of seasonal forecasts, the ensemble methods used to deal with it, and the current levels of skill. The ocean initialisation receives special focus, with a discussion on initialisation strategies, ocean data assimilation methods, and the role of the observing system in seasonal forecast skill. Assimilation of observations into an ocean model forced by prescribed atmospheric fluxes is the most common practice for initialisation of the ocean component of a coupled model. Assimilation of ocean data reduces the uncertainty in the ocean estimation arising from the uncertainty in the forcing fluxes and from model errors. Although data assimilation also usually improves the skill of seasonal forecasts, its impact is often overshadowed by errors in the coupled models. The paper also briefly discusses decadal prediction, for which there is growing demand, particularly in the context of climate change adaptation. Although decadal prediction is still in its infancy, recent development shows promising results, highlighting the role of ocean initial conditions. The initialisation of the ocean for decadal predictions is a major challenge for the next decade.
O. Alves () Bureau of Meteorology, Centre for Australian Weather and Climate Research (CAWCR), GPO Box 1289, Melbourne, VIC 3001, Australia e-mail:
[email protected] A. Schiller, G. B. Brassington (eds.), Operational Oceanography in the 21st Century, DOI 10.1007/978-94-007-0332-2_20, ©Â€Springer Science+Business Media B.V. 2011
513
514
O. Alves et al.
20.1╅Introduction Dynamical seasonal prediction has grown rapidly over the last decade or so. At present, multiple operational centres routinely issue seasonal forecasts produced with coupled ocean-atmosphere models (e.g., Fig.€ 20.1). The basis of dynamical seasonal prediction resides in variability driven by slow-processes in the climate system, particularly the ocean. The El Nino Southern Oscillation (ENSO) is the most prominent mode of climate variability on seasonal to interannual timescales and is the major source of predictability. The success of dynamical seasonal prediction is therefore often related to the ability to initialise and forecast ENSO, as well as capturing its teleconnections to regional climates. This paper focuses on dynamical seasonal prediction with coupled ocean-atmosphere models. Early efforts with dynamical prediction used atmosphere-only general circulation models, but today most operational centres use fully coupled ocean-atmosphere general circulation models. The starting point for dynamical seasonal prediction is specifying the initial state of the climate system. Seasonal prediction is generally viewed as an ocean initial Model Forecasts of ENSO from Apr 2010 3
Dynamical Model: NASA GMAO NCEP CFS JMA SCRIPPS LDEO AUS/POAMA ECMWF UKMO KMA SNU ESSIC ICM ECHAM/MOM COLA ANOM MetFRANCE JPN-FRCGC COLA CCSM3
2.5 2
NINO3.4 SST Anomaly(ºC)
1.5 1 0.5 0 –0.5
Statistical Model: CPC MRKOV CDC LIM CPC CA CPC CCA CSU CLIPR UBC NNET UCLA-TCD
–1 –1.5 –2 OBS FORECAST –2.5 JFM Mar MAM AMJ MJJ
JJA
JAS
ASO SON OND NDJ DJF
2010
Fig. 20.1↜渀 Sample forecasts of El Nino produced by international models and assembled by the IRI. (http://iri.columbia.edu/climate/ENSO/currentinfo/SST_table.html)
20â•… Seasonal and Decadal Prediction
515
condition problem, but there are also benefits from realistic atmosphere (e.g., Hudson et€al. 2010) and land (e.g., Koster et€al. 2010) initial conditions. Data assimilation can improve forecasts by correcting the model state and/or variability, but it can also create problems such as initialisation shock. Recent studies examining the impact of ocean and atmosphere initialisation on seasonal forecast skill concluded that the most skilful initialisation scheme is that which makes the most use of the observed data, even though initial imbalances in the coupled state are generated (Balmaseda and Anderson 2009; Hudson et€al. 2010). To date, initialisation of the ocean and atmosphere is done separately, although there are emerging attempts at approaching initialisation as a coupled ocean-atmosphere problem, where the component models are well-balanced. This is not trivial, particularly given the different time scales upon which the atmosphere and ocean operate. Seasonal prediction is inherently uncertain and needs to be addressed in a probabilistic framework. Dynamical seasonal prediction aims to address these uncertainties and the chaotic nature of the atmosphere by producing an ensemble of forecasts. Perturbations to the initial state or model formulation generate forecasts that diverge, producing a range of possible future outcomes from which probabilistic forecasts can be produced. Ideally, generation of the ensemble should take into account uncertainties in the initial conditions (e.g., Vialard et€ al. 2005), as well as uncertainties associated with imperfect models (e.g., Murphy et€al. 2004; Berner et€al. 2008). New ocean assimilation schemes represent the uncertainty in the ocean state by producing an ensemble of ocean initial conditions (Balmaseda et€al. 2008; Yin et€al. 2011). Coupled models are far from perfect and drift with forecast lead time towards the biased coupled model climate. A common approach is to remove the drift aposteriori (e.g., Stockdale 1997). A set of retrospective forecasts (or hindcasts) is produced to provide an estimate of how the model climatology changes with lead time, and this is then used for a-posteriori calibration of the forecast results. Ideally the hindcasts should span as long a period as possible, but in practice most centres only produce hindcasts over a 15–30 year period. The hindcasts are also needed for skill assessment of the seasonal forecast system. Implicit in the production of a set of retrospective forecasts is the need for ocean initial conditions spanning the chosen hindcast period, equivalent to an ocean “reanalysis” of the historical data stream. The interannual variability represented by the ocean reanalysis (particularly due to changes in the ocean observing system) will have an impact on both forecast calibration and the assessment of skill. This paper provides a review of dynamical seasonal prediction, with a focus on the initialisation of seasonal forecasts. Section€20.2 describes the primary drivers of seasonal prediction skill, Sect.€20.3 summarises current levels of skill and Sect.€20.4 provides some background behind ensemble prediction.€ Sects. 20.5 and 20.6 focus on data assimilation and initialisation and in particular the role of ocean observations. Section€20.7 provides an example of seasonal prediction in the Australian context. Section€20.8 introduces decadal prediction, which relies heavily on the ocean initialisation. Finally, a summary is provided in Sect.€20.9. Four recent review papers provide additional, more detailed reading on the topic of this chapter:
516
O. Alves et al.
one documenting the current status of seasonal prediction and our understanding of seasonal to interannual climate variability (produced for the Copenhagen World Climate Conference 3; Stockdale et€al. 2010), two focussing on the initialisation of seasonal and decadal forecasts and the role of ocean observations (Balmaseda et€al. 2010a, b) and one reviewing the status of decadal prediction (Hurrell et€al. 2010).
20.2â•…Predictability: What is the Source of Seasonal Prediction Skill? Predictability is a feature of the climate system and cannot be changed or improved by forecast methodologies—it represents the theoretical upper limit of our prediction skill. This maximum level of predictability has not yet been achieved in seasonal forecasting: forecast skill is limited by model error, imperfect initialisation and the fact that not all the interactions in the climate system are currently fully resolved i.e. there may be sources of predictability that are unaccounted for (Kirtman and Pirani 2009). An understanding of climate variability and its key drivers offers insight into the processes providing predictability, as well as into how model shortcomings may be limiting forecast skill. Climate variability occurs on all timescales. Atmospheric processes tend to vary over short timescales (less than a few days) and are a source of unpredictable noise for seasonal prediction. Processes operating over longer timescales, primarily those associated with the ocean, form the basis of seasonal predictability. Apart from the ocean, other potential sources of seasonal predictability include: the longer timescales of variability of the coupled ocean-atmosphere system, sea-ice, soil conditions, snow cover and the state of the stratosphere (Stockdale et€al. 2010). ENSO is the most prominent mode of climate variability on seasonal to interannual timescales and is the major source of predictability. Although mainly associated with coupled ocean-atmosphere variations in the tropical Pacific (Walker 1923, 1924; Bjerknes 1969), the effects of ENSO can be felt globally, with teleconnections to regional temperature and precipitation in many countries (e.g., Rasmusson and Carpenter 1983; Ropelewski and Halpert 1987). For example, El Nino events are typically associated with above average rainfall in Peru and Ecuador, northern Argentina, East Africa and California, and dryer than normal conditions over Australia, southern Africa and parts of the Amazon basin. Figure€20.2 shows the sea surface temperature (SST) anomaly during December 1997, near the peak of the El Nino. This was the largest El Nino of the century, with SST anomalies peaking over 4°C in the eastern Pacific. For reviews of our understanding of ENSO and the mechanisms involved, see, for example, Neelin et€al. (1998); Philander (2004) and Chang et€al. (2006). The first successful prediction of ENSO with a simple coupled ocean-atmosphere dynamical model was produced by Zebiak and Cane (1987). Since then, increasingly complex and comprehensive coupled ocean-atmosphere models have been developed and dynamical prediction of ENSO is now commonplace in major operational centres.
20â•… Seasonal and Decadal Prediction
517
Fig. 20.2↜渀 Sea surface temperature anomalies during December 1997
Low-frequency coupled ocean-atmosphere variations in the Indian and Atlantic Oceans, although less dominant than the Pacific, can also drive temperature and precipitation anomalies on seasonal timescales across the globe (e.g., Goddard and Graham 1999; Folland et€al. 2001; Rodwell and Folland 2002; Saji and Yamagata 2003; Kushnir et€ al. 2006; Ummenhofer et€ al. 2009). The Indian Ocean Dipole (IOD) has been identified as a low frequency coupled mode of variability in the tropical Indian Ocean (Saji et€al. 1999; Webster et€al. 1999). In Fig.€20.2 an IOD event can be seen in the Indian Ocean, with negative SST anomalies in the east off the Java-Sumatra coast and positive anomalies in the west. IOD events, like the one in Fig.€20.2, are often triggered by easterly wind anomalies as a result of the atmospheric response to the development of El Nino. The IOD is much less predictable (practically and theoretically) than ENSO (e.g., Luo et€al. 2007; Wajsowicz 2007; Zhao and Hendon 2009), largely due to weaker surface-subsurface ocean coupling, strong interactions with the Australian-Asian monsoon and intraseasonal oscillations causing chaotic forcings in both the ocean and atmosphere (Zhao and Hendon 2009). Although the IOD is a measure of the difference between the western and eastern parts of the equatorial Indian Ocean, these two components are not always related and the skill of each component can be different. The lack of skill of the IOD is mainly due to a lack of skill in predicting the SST in the eastern component of the IOD. Other modes of atmospheric variability (not necessarily related to oceanic forcing) that may provide predictive skill on seasonal timescales, include the Northern Annular and Southern Annular modes (NAM and SAM), the Pacific North American (PNA) pattern and the North Atlantic Oscillation (NAO) (Stockdale et€al. 2010). The land surface is a potential source of seasonal predictability, primarily associated with soil moisture memory in the earth-atmosphere system (e.g., Fennessy and Shukla 1999; Koster and Suarez 2003; Seneviratne et€al. 2006; Koster et€al. 2004, 2010), although anomalous snow cover/amount may also be important (e.g., Fletcher et€al. 2009). The coordinated approach of the Global Land-Atmosphere Coupling Experiment (GLACE; Koster et€al. 2006, 2010), using a variety of state-of-the-art seasonal forecasting systems, has significantly improved our understanding of the role of land surface processes in seasonal prediction.
518
O. Alves et al.
There have also been suggestions that the stratosphere could make a contribution to seasonal prediction skill in the troposphere, particularly in the Northern Hemisphere (e.g., Baldwin and Dunkerton 2001; Ineson and Scaife 2008; Bell et€al. 2009; Cagnazzo and Manzini 2009). However, most contemporary seasonal prediction models have a poorly resolved stratosphere and do not give a realistic representation of stratospheric circulation (Maycock et€al. 2009). A recent study by Marshall and Scaife (2009) suggests that improving the resolution of the Quasi-Biennial Oscillation (QBO), a dominant mode of variability in the tropical stratosphere, could improve seasonal prediction of QBO-induced surface anomalies over Europe.
20.3â•…Forecast Skill As mentioned in Sect.€ 20.2, ENSO is the most predictable large-scale phenomenon on seasonal to interannual timescales, and is the major source of predictability. Successful predictions with a coupled seasonal forecast model are, therefore, often related to a model’s ability to reproduce the slow coupled dynamics of ENSO and accurately forecast its amplitude, spatial pattern and detailed temporal evolution (Wang et€al. 2008a). The skill of forecasting ENSO varies depending on the season, as well as the phase and intensity of ENSO. For example, there is usually greater skill at predicting ENSO events compared to neutral events, and predicting the growth phases of warm and cold events compared to the corresponding decaying phases (e.g., Jin et€ al. 2008). In terms of season, many seasonal forecast systems experience a decline in skill during the boreal spring, often referred to as the “spring predictability barrier”. At this time of year, SST anomalies are particularly variable and although dynamical forecast models may have reduced skill, their advantage over persistence forecasts is at a maximum (e.g., van Oldenborgh et€al. 2005; Jin et€al. 2008; Wang et€al. 2008a). Large multi-model projects, such as DEMETER (Palmer et€al. 2004), ENSEMBLES (Weisheimer et€al. 2009) and APCC/CliPAS (Wang et€al. 2008a), have provided a basis for intercomparing the skill and errors from coupled models, benchmarking seasonal prediction skill and assessing progress. Weisheimer et€al. (2009) report that results from the European ENSEMBLES project (using 5 European coupled models) have shown a significant reduction in the systematic SST errors (SST drift over the Pacific as the forecast progresses) compared to the previous generation project, DEMETER. For the NINO3 region (5°S–5°N; 150°W–90°W) the SST drift in DEMETER varied between +2°C and −7°C for up to 6 months lead, whereas the drift from the ENSEMBLES models was less than ±1.5°C (Weisheimer et€al. 2009). They conclude that since DEMETER, the coupled models have improved significantly in terms of their physical parameterisations, resolution and initialisation. They also show that although probabilistic skill scores suggested increases in SST prediction skill in the 4–6 month forecast range in the ENSEMBLES multi-model ensemble (MME) compared to the DEMETER MME, the increases were not statistically significant, suggesting that substantially better models (perhaps with a higher resolution than
20â•… Seasonal and Decadal Prediction
519
available now) are required to improve upon the current skill of forecasting tropical Pacific SSTs. As an example of current skill levels, the anomaly correlation skill in predicting NINO3.4 SST anomalies (an area average over 5°N–5°S, 170°–120°W) from an ensemble of 10 coupled seasonal forecast models (for hindcasts performed over 1980–2001) is 0.86 after 6 months of the forecast (Jin et€al. 2008). This level of skill from the MME is greater than from any single model, but at this lead time all models have skill greater than persistence and many of the models have anomaly correlation skills exceeding 0.8 (Jin et€al. 2008). Skill in predicting Indian Ocean SST anomalies is lower than over the Pacific. This is clear from Fig.€20.3 which shows the anomaly correlation skill of predicting SST anomalies at 6 months lead time from the POAMA (Predictive Ocean Atmosphere Model for Australia) seasonal forecast model. This is very typical of most seasonal forecast models. Prediction of the IOD is currently limited to about one season, with a strong boreal winter-spring predictability barrier (partly because the IOD is not well defined prior to June) (e.g., Luo et€al. 2007; Wajsowicz 2007; Zhao and Hendon 2009). In terms of tropical Atlantic SST anomalies, current seasonal prediction models show very little skill beyond one or two months of the forecast and skill is often no better than persistence (e.g., Stockdale et€al. 2006, 2011). The forecast skill of regional surface air temperature and precipitation anomalies is strongly dependent on season and region. Skill is highest in the tropics and decreases towards middle and high latitudes, and is usually higher for temperature than precipitation (e.g., Wang et€al. 2008a; Doblas-Reyes et€al. 2009). At 1-month lead there is very little skill in predicting seasonal mean temperature and precipitation anomalies over land in extra-tropical regions (e.g., Wang et€al. 2008a; DoblasReyes et€al. 2009). Those extra-tropical land regions that do exhibit some skill (e.g., southern Africa and the southern United States for precipitation in DJF) are usually a result of the models capturing the atmospheric teleconnections from ENSO. Consequently, model bias and drift in the simulation of ENSO may degrade global teleconnections to regional rainfall and temperature. For example, most models exhibit a cold bias in the central equatorial Pacific and a westward drift of maximum SST variability away from the eastern Pacific with increasing lead time (e.g., Jin et€al. 2008). In the POAMA seasonal forecast model, after about a season, these biases hinder the model’s ability to discern between different types of ENSO events (e.g., classical east Pacific versus central Pacific events) and the teleconnection between ENSO and Australian climate is adversely affected (Hendon et€al. 2009; Lim et€al. 2009).
20.4╅Ensemble Prediction: Representing Uncertainty There is considerable uncertainty inherent in seasonal predictions, some natural and some due to deficiencies in the forecasting systems. Figure€20.4 shows 90 forecasts of the onset of the 1997/1998 El Nino. Each forecast was produced using the POAMA-1 model (Alves et€al. 2003). The ensemble was generated by making
Fig. 20.3↜渀 SST anomaly correlation at 6 month lead time from POAMA-1.5 forecasts (↜left) and persistence (↜right). (From Wang et€al. 2008b)
520 O. Alves et al.
20â•… Seasonal and Decadal Prediction
521
Fig. 20.4↜渀 Forecasts NINO3.4 SST anomaly during the onset of the 1997/1998 El Nino. A 90-member ensemble, where each ensemble member is generated by applying a 0.001C random perturbation to the initial SST. (From Shi et€al. 2009)
0.001°C changes to the initial SST. These changes are physically insignificant, but because the climate system, in particular the atmosphere, is chaotic, the ensemble members can spread rapidly with time. The plot shows that while all of the forecast were for El Nino conditions, they range from a very weak El Nino with NINO3.4 SST anomalies of around 0.5°C to very strong El Nino conditions with NINO3.4 anomalies greater than 2.5°C by August. The spread in the forecasts indicates the stochastic component of the climate model, i.e. natural uncertainty and therefore the limits to predictability. In a seasonal forecast system the ensemble spread should be commensurable to the uncertainty arising from natural stochastic processes, but this is not always the case due to errors in the forecast system. For practical reasons, the uncertainty is classified into that arising from an imperfect initial state (inital conditions uncertainty) and that arising from imperfect models (model data sampling uncertainty, model parametric uncertainty, model structural uncertainty). In dynamical seasonal prediction, ensembles are used to quantify the uncertainty (e.g., Stephenson 2008; Doblas-Reyes et€al. 2009). Uncertainties in the initial conditions are taken into account by generating an ensemble from slightly different atmospheric and/or ocean
522
O. Alves et al.
analyses, where the differences are intended to reflect the uncertainty in these conditions (e.g., Vialard et€ al. 2005). Uncertainties in model formulation have been addressed using ensembles based on stochastic physics (Jin et€ al. 2007; Berner et€al. 2008), perturbed parameter (Murphy et€al. 2004; Stainforth et€al. 2005; Collins et€al. 2006) and multi-model approaches (Palmer et€al. 2004; Weisheimer et€al. 2009). Doblas-Reyes et€ al. (2009) assessed the relative merits of these three approaches using sets of seasonal and decadal hindcasts (done under the auspices of the European ENSEMBLES project; see van der Linden and Mitchell 2009). In general, they concluded that the three methods had comparable overall skill (the multi-model was slightly better for lead times up to 4 months, and the perturbed physics slightly better at longer leads). The perturbed physics and stochastic parameter methods are promising methods of sampling model uncertainty within a single model system. Probabilistic forecasts are produced from dynamical seasonal forecasting systems by using the aforementioned ensemble of forecasts. The forecasts follow different evolutions because they are produced from perturbed initial conditions or model formulations. After the first week, the ensemble spread is large and the forecast needs to be delivered and assessed in a probabilistic fashion. Good reviews of probability forecasting in a seasonal context, including basic concepts, recalibration and verification, are provided by Stephenson (2008) and Mason and Stephenson (2008). The distribution of the ensemble members should indicate uncertainty in the forecast: if the forecasts from the ensemble members differ widely, the inferred probability distribution is also wide and the forecast is uncertain, whereas if the ensemble members are in close agreement it might suggest less uncertainty. However, in practice, forecasts from dynamical seasonal forecast models tend to be overconfident, i.e. their spread is too narrow to match the range of observed outcomes, and there is often little relationship between ensemble spread and the error in the forecast. The prime reason for this is believed to be model error (Vialard et€al. 2005; Stockdale et€al. 2010). Multi-model approaches, where ensembles from different state-of-the-art models are combined, thereby implicitly averaging out some of the model errors, generally produce more skilful forecasts than do the results from a single model (Palmer et€al. 2004; Wang et€al. 2008a; Weisheimer et€al. 2009). A counter example of the limitation of the multi-model approach is provided by Balmaseda et€al. (2010b), showing that for a given SST index, the skill of a single model can be superior to that of the multi-model product. But this is not yet the case for useful atmospheric variables such as precipitation, where reliable seasonal forecasts benefit from the multi-model approach. Multi-model forecast systems are becoming increasingly apparent in operational seasonal forecasting. For example, the APEC Climate Center (APCC) produces real-time operational climate predictions based on a well-validated multi-model multi-institute ensemble system (http://www.apcc21.org) and ECMWF has collaborated with France and the United Kingdom to produce an operational multi-model seasonal forecast system known as EUROSIP (http://www.ecmwf.int/products/forecasts/seasonal/ documentation/eurosip/).
20â•… Seasonal and Decadal Prediction
523
20.5â•…Data Assimilation and Initialization Dynamical seasonal prediction is essentially an initial value problem, where predictive skill comes from information contained in the initial states of the coupled system: ocean, atmosphere, land and sea-ice. Most of the skill comes from the initial conditions of the upper ocean, particularly those associated with large scale patterns of variability such as ENSO and the IOD. Assimilation of ocean observations for ocean initialisation in seasonal forecasts has become a common practice, with several institutions around the world producing routine ocean re-analyses to initialise their operational seasonal forecasts. Table€20.1, from Balmaseda et€al. (2009), provides a summary of the ocean analyses used for initialisation of operational or quasi-operational seasonal forecast systems. In all these systems, the initialisation of the ocean and atmosphere is done separately, aiming at generating the best analyses of the atmosphere and ocean through comprehensive data assimilation schemes. The simplest way to initialise the tropical ocean is to run an ocean model forced with atmospheric fluxes and with a strong relaxation of the model SST to observations. Inter-annual variability in the tropical ocean is to a large extent driven by variability in the surface wind field. This technique would be satisfactory if errors in the forcing fields and ocean model were small. However, surface flux products and ocean models are both known to have significant errors. Assimilation of ocean observations is then used to constrain the estimation of the ocean state. In ocean assimilation, ocean sub-surface observations are ingested into an ocean model forced by prescribed atmospheric fluxes. The emphasis is on the initialisation of the upper ocean thermal structure, particularly in the tropics, where SST anomalies have a strong influence on the atmospheric circulation. Most of the initialisation systems use observed subsurface temperature (from XBT’s, TAO/TRITON/ PIRATA and Argo). Some of the more recent systems also use salinity (mainly from Table 20.1↜渀 Summary of different ocean assimilation systems used in the initialisation of operational and quasi-operational seasonal forecasts. (Based on Balmaseda et€al. 2009) MRI-JMA http://ds.data.jma.go.jp/tcc/tcc/products/elnino/index.html Multi-variate 3-dimension Variational (3D-VAR). Usui et€al. 2006 ORA-S3 (ECMWF System 3) http://www.ecmwf.int/products/forecasts/d/charts/ocean/real_time/ Multivariate Optimum Interpolation (OI). Balmaseda et€al. 2008 POAMA – PEODAS (CAWCR, Melbourne) http://poama.bom.gov.au/research/assim/index.htm Multivariate Ensemble OI. Yin et al. 2011 GODAS (NCEP) http://www.cpc.ncep.noaa.gov/products/GODAS/ 3D-VAR. Behringer 2007 MERCATOR (Meteo France) http://bulletin.mercator-ocean.fr/html/welcome_en.jsp Multivariate reduced order Kalman filter. Pham et€al. 1998 MO (MetOffice) http://www.metoffice.gov.uk/research/seasonal/ Multivariate OI. Martin et€al. 2007 GMAO ODAS-1 http://gmao.gsfc.nasa.gov/research/oceanassim/ODA_vis.php GMAO Seasonal Forecasts: http://gmao.gsfc.nasa.gov/cgi-bin/products/climateforecasts/index.cgi OI and Ensemble Kalman Filter Keppenne et€al. 2008
524
O. Alves et al.
Argo), and altimeter derived sea-level anomalies. The latter usually needs the prescription of an external Mean Dynamic Topography, which can be a problem, and is usually taken from a model integration rather than observations. In the longer term it is hoped that it can be derived indirectly from gravity missions such as GRACE and GOCE. Several studies have demonstrated the benefit of assimilating ocean data on the prediction of ENSO (e.g., Alves et€al. 2004; Dommenget et€al. 2004; Cazes-Boezio et€al. 2008; Stockdale et€al. 2011). The benefits are less clear in other areas, such as the equatorial Atlantic, where model errors are large. Balmaseda and Anderson (2009) evaluated three different initialisation strategies, each of which used different observational information. They showed that the ocean initialisation has a significant impact on the mean state, variability and skill of coupled forecasts at the seasonal time scale. They also showed that, using their model, the initialisation strategy that makes the most comprehensive use of the available observations leads to the best skill. Since ocean assimilation is important for seasonal prediction, an interesting question is: how accurate are ocean analyses from ocean assimilation systems? Figure€20.5 shows the composite El Nino evolution of heat content along the equator in the Pacific and Indian Oceans. The composite plots consist of 30 months spanning each El Nino event from −9 months (year prior to warm event), 12 months (warm event), to +9 months (year after warm event), and these are denoted as Year −1, Year 0 and Year +1 respectively. The selection criteria for El Niño/La Niña events is defined as the monthly Niño3 SST anomaly reaches or exceeds ±0.5°C for at least 5 consecutive months over the period 1982 to 2006. Composites from two state of the art international analyses are shown to illustrate how they differ and give an indication of the level of error in the analysis. The assimilation systems used to generate each analysis are quite different and so are the forcing fields used to drive the ocean model during the re-analysis phase. The composite El Nino evolution shows El Nino peaking at the end of the year with maximum heat content anomalies in eastern Pacific. At the same time there are heat content anomalies in western Pacific, forming a strong gradient between east and west Pacific, which is driven by anomalous westerly winds (not shown). Normally during the peak of El Nino there are also easterly winds in the Indian Ocean, which leads to an east-west pattern that is the reverse of the pattern in the Pacific. The composites also show the evolution of positive heat content anomalies from the western Pacific at the beginning of the year where El Nino develops, towards the eastern Pacific through the action of equatorial Kelvin waves. There is considerable agreement between the two re-analyses, likely to be due to a reasonable observing network, particularly in the Pacific with the TOGA-TAO array and this decade with Argo. The same is not true of salt content. Figure€20.6 compares the evolution of salt content along the equator for the same El Nino composite. One re-analysis shows significant salt anomalies throughout the equatorial Pacific during El Nino, while the other shows weaker anomalies. For example, the first re-analysis shows strong freshening in the central/west Pacific at the peak of El Nino of just over 0.1ppt,
–1.2
–0.9 –0.6
–0.3
0
0.3
0.6
0.9
1.2
1.5
–1.5
–1.2
–0.9 –0.6
–0.3
0
0.3
0.6
0.9
1.2
1.5
40E 60E 80E 100E 120E 140E 160E 180 160W 140W120W 100W 80W
May
Jul
Sep
Nov
Jan
Mar
May
Jul
Sep
Nov
Jan
Mar
May
Jul
Sep
Composite T300 Anom (5S – 5N) for EINino GODAS
Fig. 20.5↜渀 Evolution of a composite El Nino in two different state of the art ocean re-analyses. Each plot shows the evolution of heat content anomalies (upper 300€m) along the equator for a composite El Nino. The plot covers the period from April of the year prior to El Nino developing, to September of the year after El Nino develops. The same El Nino events were included in both composites
–1.5
40E 60E 80E 100E 120E 140E 160E 180 160W 140W 120W 100W 80W
May
Jul
Sep
Nov
Jan
Mar
May
Jul
Sep
Nov
Jan
Mar
May
Jul
Year +1 Year 0 Year –1
Sep
Composite T300 Anom (5S – 5N) for EINino PEODAS
20â•… Seasonal and Decadal Prediction 525
Fig. 20.6↜渀 As for Fig.€20.5 except for salt content of the upper 300€m
–0.16 –0.14 –0.12 –0.1 –0.08 –0.06 –0.04 –0.02 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
–0.16 –0.14 –0.12 –0.1 –0.08 –0.06 –0.04 –0.02 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
80E 100E 120E 140E 160E 180 160W 140W 120W 100W 80W
40E 60E
May
Jul
Sep
Nov
Jan
Mar
May
Jul
Sep
Nov
Jan
Mar
May
Jul
Sep
Composite S300 Anom (5S – 5N) for EINino GODAS
40E 60E 80E 100E 120E 140E 160E 180 160W 140W 120W 100W 80W
May
Jul
Sep
Nov
Jan
Mar
May
Jul
Sep
Nov
Jan
Mar
May
Jul
Year +1 Year 0 Year –1
Sep
Composite S300 Anom (5S – 5N) for EINino PEODAS
526 O. Alves et al.
20â•… Seasonal and Decadal Prediction
527
presumably due to eastward advection of fresh water associated with the anomalous westerlies. However, the second re-analysis does not show such strong anomalies, generally less than 0.04ppt. This clearly indicates that, at least at present, there are significant differences in how state-of-the-art ocean re-analyses represent interannual variability of salinity. It has been shown (Balmaseda and Weaver 2006) that in the absence of salinity data, the assimilation of temperature observations can increase the uncertainty in the salinity field. The salinity field can influence the seasonal forecasts by influencing the barrier layer, which acts as a reservoir of warm water (above 28°C), and can be instrumental for the development of El Niño when propagated eastward by westerly winds (Fujii et€al. 2011). Interestingly both re-analyses show similar salt content patterns in the Indian Ocean. This is probably due to the lack of salinity and temperature data in the Indian Ocean, at least before Argo. Without much temperature and salinity data, the re-analyses are simply ocean simulations driven by surface forcing, which is likely to lead to similar patterns. There are three main ways of evaluating ocean analyses produced by data assimilation systems: (1) how well the analysis fits the assimilated observations, (2) how well the analysis fits independent observations and (3) whether the analysis leads to improved forecasts. Way (3) may not be a reliable method because if the models have significant errors a better initial state could potentially lead to a worse forecast. Way (1) is also not entirely satisfactory since it simply reflects how well the analysis fits the observations, which is mostly a function of the background and observation error variances. Way (2) is the most desirable way, but it can be difficult since usually all temperature and salinity observations are used in the assimilation. To date no assimilation system utilises ocean current data. This is one source of independent data, and Fig.€20.7 illustrates the use of ocean current data to evaluate different re-analyses. Figure€ 20.7 shows the correlation between re-analyses and pseudo-observed ocean surface currents (the currents are derived from altimeter data: OSCAR; Bonjean and Lagerloef 2002). Three re-analyses are assessed against the OSCAR current data. Figure€ 20.7a uses the PEODAS re-analysis (Yin et€ al. 2011) which is representative of a current generation ocean re-analysis. It makes dynamically balanced corrections to the currents based on the temperature and salinity corrections. The current corrections are based on the cross-covariances derived from a time evolving ensemble, see Yin et€al. (2011) for more details. Figure€20.7b shows the correlation from a control re-analysis, i.e. the same as PEODAS except no observations are assimilated. This is essentially an ocean model forced with reanalysis surface fluxes and will do a reasonable job of representing the inter-annual variability, at least as far as it is represented in the forcing fields. Figure€20.7c uses a re-analysis from an older generation ocean assimilation system, in this case from the POAMA-1 seasonal prediction system (Alves et€al. 2003). Typical of this generation, only temperature observations (and not salinity) are assimilated. However, corrections to currents are made based on the temperature corrections by assuming geostrophic balance, as in Burgers et€al. (2002). In all three re-analyses no altimeter data are assimilated. The figures show that the PEODAS re-analysis produces the best correlation with the observed data in both the tropical Pacific and Indian
528
O. Alves et al.
Fig. 20.7↜渀 Correlations between the zonal surface velocity from OSCAR and a PEODAS. b Control, and c POAMA-1. Note the non-linear correlation scale. (From Yin et€al. 2011)
20â•… Seasonal and Decadal Prediction
529
Oceans. Interestingly the last generation POAMA-1 system produces the worst comparison to observations, even worse than the control which uses no data. This is likely for two reasons. Firstly, salinity data are not assimilated in POAMA-1, which can lead to incorrect density profiles since density corrections are only based on temperature, which in turn can lead to wrong current increments when using the geostrophic relation. Secondly, the geostrophic relation may not be appropriate, especially for the surface current which has a significant Ekman component. While the control re-analysis does not use any observations, it does maintain a surface current that is in dynamical balance with the surface forcing and the pressure fields. These results illustrate the progress over the last decade that has led to the current state of the art in ocean data assimilation. Ensemble based data assimilation schemes, such as Ensemble Kalman Filters, provide an ensemble of analyses. The spread of the ensemble members represents the uncertainty in the estimated ocean state and the standard deviation of the ensemble spread about the ensemble mean can be considered a measure of the analyses error. Ensemble spread from the PEODAS ocean assimilation scheme (Yin et€al. 2011) is shown in Fig.€20.8. The highest spread in SST (Fig.€20.8a) occurs in the eastern equatorial Pacific and along the western boundary currents, as one might expect as these are the regions of highest variability. The highest spread in surface salinity (Fig.€ 20.8b) occurs in regions of highest rainfall, such as along the Inter-Tropical Convergence Zone, the South Pacific Convergence Zone and the high rainfall regions of the West Pacific warm pool. Figure€20.8c shows the temperature ensemble spread at depth along the equator. Maximum spread occurs along the thermocline, the region of maximum temperature variability. Maximum salinity spread (Fig.€20.8d) occurs at the surface.
20.6╅The Impact of Ocean Observations The ocean observing system has undergone major changes over the last couple of decades. In the early 1990s the TOGA-TAO array in the tropical Pacific was introduced. This allowed the heat content of the equatorial upper ocean to be monitored on a daily basis. In the early 1990s sea level measurements from satellite altimeters became routine, although not all operational ocean data assimilation systems ingest altimeter data. During the 2000s Argo floats were introduced, and this was perhaps the biggest revolution in ocean observations for climate. Large areas of the ocean that were previously unobserved were now covered with autonomous Argo floats. Figure€ 20.9a shows the temperature observation density pre Argo in the Indian Ocean. Observations were mainly taken along the main shipping lanes as part of the Ship of Opportunity Program (SOOP). Large gaps remained throughout the Indian Ocean. During the Argo period (Fig.€20.9c) the temperature distribution changed dramatically, with almost every grid square experiencing at least one observation. Perhaps the biggest impact of Argo is that it also measures salinity. For the first time there were enough salinity profiles to perform assimilation of salinity data.
Fig. 20.8↜渀 Spread of the ensemble (before assimilation) over the re-analysis period showing fields of a SST (°C). b Temperature section along the equator (°C). c Sea surface salinity (psu) and d Salinity section along the equator (psu). (From Yin et€al. 2011). Ensemble spread is calculated relative to a central analysis. (See Yin et€al. 2011, for full details)
530 O. Alves et al.
Fig. 20.9↜渀 The density of ocean sub-surface observations per 1â•›×â•›1 degree square per year. a and c are Temperature and b and d are Salinity. a and b are pre Argo and c and d are during the Argo period
20â•… Seasonal and Decadal Prediction 531
532
O. Alves et al.
25
TAO/TRITON
ARGO
20
%
15 10 5 0 –5
NINO12 NINO3 NINO34 NINO4 NINOW
WTIO
Fig. 20.10↜渀 The impact of the TAO/TRITON and Argo data on seasonal forecast skill. Bars show the relative increase in root mean square errors of the 1–7 month forecasts of monthly SST anomalies resulting from withholding TAO/TRITON and Argo data in the initialization of JMA seasonal forecasts for different ocean areas. (From Fuji et€al. 2008, where areas are defined)
Figure€20.9b shows the salinity observation density before Argo and Fig.€20.9d during Argo. The change is dramatic. Before Argo most of the Indian Ocean was unobserved. During Argo the salinity observation density is similar to that for temperature. The importance of salinity observations is discussed in Fujii et€al. (2011). The results of Usui et€al. (2006) indicate that only when salinity observations are assimilated is it possible to represent the strong meridional salinity gradient in the Western Equatorial Pacific, with low salinity waters north of the equator. Results also show that without the balance relationship between temperature and salinity it is not possible to represent the high salinity of the South Pacific Tropical Water, leading to the erosion of the vertical stratification and eventual degradation of the barrier layer. The seasonal forecast skill can also be used to evaluate the ocean observing system. Fujii et€al. (2011), evaluate the impact of the TAO/TRITON array and Argo float data on the JMA seasonal forecasting system by conducting data retention experiments. Their results (Fig.€20.10) show that TAO/TRITON data improves the forecast of SST in the eastern equatorial Pacific (NINO3, NINO4), and that Argo floats are essential observations for the prediction of the SST in tropical Pacific and Indian Oceans. Similar results have been obtained with the European Centre for Medium-range Weather Forecasts (ECMWF) seasonal forecasting system (Balmaseda et€al. 2007, 2009).
20.7â•…Seasonal Prediction in Australia The Australian Bureau of Meteorology has produced seasonal outlooks since the late 1980s. Currently a seasonal rainfall and temperature outlook for Australia is produced operationally based on statistical links between tropical SSTs and local climate (Chambers and Drosdowsky 1999). However, it is felt that statistical ap-
20â•… Seasonal and Decadal Prediction
533
Fig. 20.11↜渀 POAMA monthly GBR Index (area average SST anomalies for the red box shown in the map insert) for December 2009 to May 2010 in the official outlook issued on 1 December 2009, with the distribution by quartiles of the ensemble composed of the last 30 daily forecasts. Overlaid is the ensemble mean (↜black). The shading indicates upper and lower climatological terciles from the POAMA v1.5 hindcasts. (http://www.bom.gov.au/oceanography/oceantemp/GBR_SST.shtml)
proaches have essentially reached the limits of their predictive ability, particularly as climate change is invalidating the assumptions of stationary that is fundamental to statistical approaches. The Bureau, in collaboration with CSIRO, has been developing successive versions of a dynamical coupled modelling system called POAMA (Predictive Ocean Atmosphere Model for Australia; http://poama.bom. gov.au). The first version was implemented in Bureau operations in 2002 and generated forecasts of ENSO-related SST indices. The POAMA system was upgraded in 2007 with version 1.5 and the operational products were extended to include forecasts of SST in the equatorial Indian Ocean (Zhao and Hendon 2009). More recently the products have been extended to give warnings of potential bleaching of coral in the Great Barrier Reef in the season ahead (e.g., Fig.€20.11; Spillman and Alves 2009). POAMA-1.5 has been shown to have high skill in prediction not only of ENSO and the IOD, but also the “flavour of ENSO”, i.e. classical versus Modoki modes (Hendon et€al. 2009; Lim et€al. 2009). POAMA can skilfully predict tropical SST anomalies associated with ENSO two to three seasons in advance (Wang et€al. 2008b) and can depict the teleconnection to Australian rainfall (Lim et€al. 2009). POAMA can predict the peak phase of the occurrence of the IOD in austral spring (SON) with about four months lead time (Zhao and Hendon 2009). The most skilful season for POAMA in predicting rainfall over Australia is during spring (SON), when the relationship between ENSO and Australian rainfall is strong. Fig.€20.12 shows that the skill (proportion correct) of predicting above median rainfall is high over south-eastern Australia and better than climatology over most of
Fig. 20.12↜渀 Proportion of ensemble members correctly predicting above median rainfall with (a) POAMA at LT0 (b) POAMA at LT3 and (c) with the current operational statistical model (NCC model). The contour interval is 10%. The proportion correct greater than 60% is shaded. (From Lim et€al. 2009)
534 O. Alves et al.
20â•… Seasonal and Decadal Prediction
535
the country at lead time 0 (LT0, i.e. forecasts initialised at the start of September and verified in SON, over the period 1980–2006) (Lim et€al. 2009). This region of skill is where the teleconnection between rainfall and tropical SST is strong (Lim et€al. 2009). However, operational regional rainfall and temperature forecasts at the Bureau are still based on the statistical system rather than POAMA at this point in time. Experimental rainfall products, such as probabilities of above median rainfall, from POAMA have been shown to be more skilful than those based on the statistical system based on skill measures such as the ROC score or hit rates (e.g., Fig.€20.12), but the forecast reliability is low, i.e. the forecasts are too emphatic (over-confident) often showing probabilities in excess of 90%. Work is in progress to address this reliability issue so that POAMA rainfall can form the basis for the Bureau’s seasonal climate outlooks, including a pragmatic statistical correction and recalibration in the short term and investigating methods to increase ensemble spread in the long term. A new version, POAMA-2 has been developed with improved physics and a new ocean data assimilation system, the POAMA Ensemble Ocean Data Assimilation System (PEODAS), mentioned in Sect.€20.5. A comprehensive set of hindcasts are currently being generated and the system is due to be implemented operationally towards the end of 2010. Preliminary results show a significant increase in SST skill in the Pacific Ocean in POAMA-2 compared to POAMA-1.5. Development of the POAMA-3 system is also underway, which includes a new coupled model based on the UKMO Unified Atmospheric model and the GFDL MOM4, to be run at a higher resolution than the current system. The ocean data assimilation system is also being extended to include the atmosphere and land surface, which will result in a multivariate ensemble coupled assimilation system. The new ocean data assimilation system, PEODAS (Yin et€al. 2011), is a major new development in POAMA. The system is based on multivariate ensemble optimum interpolation (Oke et€al. 2005) where the background error covariance is calculated from an ensemble of ocean states. However, unlike Oke et€al. (2005) which uses a static ensemble, PEODAS uses a time evolving ensemble to calculate a time dependent multivariate error covariance matrix. An ensemble is run in parallel to the main analyses by perturbing the ocean model forcing about the main analysis run, using a method developed by Alves and Robert (2005). An ocean reanalysis has been conducted from 1977 to 2007, assimilating temperature and salinity observations from the ENACT/ENSEMBLE project. During the assimilation, temperature and salinity were relaxed to monthly climatology through the water column with an e-folding time scale of 2 years. The model SST was strongly nudged to the SST product from the NCEP reanalysis with a 1-day time scale. In Sect.€20.5 it was shown that the PEODAS ocean reanalysis is an improvement with respect to the previous POAMA version. Preliminary results also suggest that these improvements lead to better forecast skill of SST at seasonal time scales. For each reanalysis a set of hindcasts starting each month from 1980 to 2001 were produced. For the PEODAS reanalysis a 10-member ensemble was generated using the main PEODAS reanalysis. For the old POAMA reanalysis a 10-member ensemble was also generated, however this time by using the same ocean initial conditions
536 Fig. 20.13↜渀 NINO3.4 SST anomaly correlation skill as a function of leadtime (months). Red—POAMA-2 initialised from PEODAS, Black—POAMA-1.5 initialised using the old POAMA assimilation, black dash—persistence
O. Alves et al. Nino3.4 Corr ALL 82-06
1 0.95 0.9 0.85 0.8 0.75 0.7 0.65 0.6 0.55
0
1
2
3
4
5
6
7
8
9
10
11
(since perturbed states were not available) and taking atmospheric initial conditions six hours apart. Figure€20.13 shows the NINO3.4 forecast skill with lead time for forecasts from each set of reanalysis and based on the 10-member ensemble means. Forecasts using PEODAS initial conditions show significantly more skill than those using the old POAMA assimilation initial conditions. While the old reanalysis had a similar fit to observed temperature as the new reanalysis, the old reanalysis showed a considerably worse fit for salinity and zonal current. This result can be taken as an indication that, for the assimilation to improve forecast skill, it is important to keep the dynamical and physical balance among variables, and therefore all variables, not just those directly constrained by observations, should show consistent improvement.
20.8â•…Decadal Prediction Decadal climate prediction is very much in its infancy, but has the potential to provide information enabling better adaptation to climate change. Anthropogenic climate change signals are strongly modulated by natural climate variability, particu-
20â•… Seasonal and Decadal Prediction
537
larly variability driven by slow processes in the ocean on decadal time-scales (Hurrell et€al. 2010). There is growing evidence that, like seasonal prediction, decadal prediction is an initial-value problem, with recent results from the ENSEMBLES project (Smith et€ al. 2007; van der Linden and Mitchell 2009) showing that initialised decadal forecasts have the potential to provide improved information compared with traditional climate change projections. Decadal predictability originates primarily from changes in radiative forcing, including anthropogenic greenhouse gases and aerosols, and long-lived variations in the ocean. Examples of the latter include variations associated with the Pacific Decadal Oscillation (PDO; e.g., Mantua et€al. 1997), the Inter-decadal Pacific Oscillation (IPO; e.g., Power et€al. 1999) and the Atlantic Multidecadal Oscillation (AMO; e.g., Knight et€al. 2005). The ability to predict these long term climate variations depends therefore, in part, on accurate ocean initial conditions. However, compared to seasonal prediction, decadal prediction relies on the less well observed deeper ocean. Recent improvements in the ocean observing system, in particular the advent of Argo data, offers potential for increased skill of decadal forecasts (Balmaseda et€al. 2010a). The Argo data (available since 2003) are likely to be critical, for example, for making skilful predictions of the Atlantic Meridional Overturning Circulation (MOC) (Balmaseda et€al. 2010a). But, a major challenge for decadal prediction is how to evaluate the hindcasts and forecasts, particularly in view of sparse historical ocean observations (Balmaseda et€al. 2010a; Hurrell et€al. 2010). In addition, as a result of our short observational record, the mechanisms of decadal variations are not well understood and the representation of this variability differs considerably among models (Hurrell et€al. 2010). This means that the theoretical upper limit of our prediction skill on the decadal time scale is also not well established (Hurrell et€al. 2010). Another challenge facing decadal prediction is how to initialise the forecasts. Current systems (Smith et€al. 2007; Keenlyside et€al. 2008; Pohlmann et€al. 2009) use anomaly initialisation, rather than full initialisation, such that models are initialised with observed anomalies added to the model climate. This method is a way of dealing with model bias and reducing initialisation shock. However, the best approach for initialising decadal forecasts remains unclear (Hurrell et€al. 2010).
20.9â•…Summary Today’s sophisticated operational seasonal forecast systems rely on a number of interrelated components: data assimilation and initialisation, a coupled ocean-atmosphere general circulation model, ensemble generation and forecast calibration. The ocean plays a key role in each component. Predictive skill in seasonal forecasting comes from the initial state of the coupled system, particularly the upper ocean. Correctly initialising the important modes of seasonal and interannual variability, such as ENSO and the IOD, is vital. Real-time estimates of the ocean initial state have improved dramatically over the last two decades with improvements to the ocean observing network, especially from the TAO/TRITON array and Argo floats.
538
O. Alves et al.
However, seasonal forecasting requires an ocean reanalysis going back in time in order to initialise the retrospective forecasts required for skill assessment of the forecast system and calibration of the forecasts. The non-stationarity of the ocean observing system poses huge challenges for the initialisation and verification of seasonal, as well as decadal, hindcasts and forecasts. Results have shown that the method of ocean initialisation has a significant impact on the mean state, variability and skill of the forecasts (Balmaseda and Anderson 2009). Because of deficiencies in the coupled model, the aim of producing the best initial state, closest to observed, may not produce the best forecasts. There may be long-term effects of model spinup or initialisation shock when using observed initial conditions. Recent research suggests that the initialisation scheme that makes the most use of the observed data will produce the most skilful forecasts, even though initial imbalances in the coupled state are generated (Balmaseda and Anderson 2009). Clearly, however, the impact of the initialisation scheme is very dependent on the quality of the coupled model. Current research is addressing the prospect of “coupled assimilation”, where data assimilation for the atmosphere and ocean are done by the coupled model, leading to a well-balanced initial state. Seasonal prediction is a complex and challenging field of research and application. This paper addresses dynamical seasonal prediction using coupled oceanatmosphere models, with particular focus on data assimilation and initialisation. The delivery, value and use of seasonal forecasts have not been discussed. It is the latter that will continue to drive future advances in coupled models, data assimilation, ensemble techniques and the ocean observing system. Acknowlegements╇ The authors would like to acknowledge Eun-Pa Lim, Claire Spillman, Guomin Wang and Yonghong Yin for providing some of the figures used in this paper.
References Alves O, Robert C (2005) Tropical Pacific Ocean model error covariances from Monte Carlo simulations. Quart J Roy Meteor Soc 131:3643–3658 Alves O, Wang O, Zhong A, Smith N, Tseitkin F, Warren G, Schiller A, Godfrey JS, Meyers G (2003) POAMA: bureau of meteorology operational coupled model forecast system. National Drought Forum, Brisbane (15–16 April) Alves O, Balmaseda M, Anderson D, Stockdale T (2004) Sensitivity of dynamical seasonal forecasts to ocean initial conditions. Quart J Roy Meteor Soc 130:647–668 Baldwin MP, Dunkerton TJ (2001) Stratospheric harbingers of anomalous weather regimes. Science 294:581. doi:10.1126/science.1063315 Balmaseda MA, Weaver A (2006) Temperature, salinity, and sea-level changes: climate variability from ocean reanalyses. Paper presented at the CLIVAR/GODAE meeting on ocean synthesis evaluation, 31 August–1 September 2006, ECMWF, Reading, UK. http://www.clivar.org/organization/gsop/synthesis/groups/Items3_4.ppt. Accessed 26 May 2009 Balmaseda MA, Anderson D (2009) Impact of initialisation strategies and observations on seasonal forecast skill. Geophys Res Lett 36:L01701. doi:10.1029/2008GL035561 Balmaseda MA, Anderson DLT, Vidard A (2007) Impact of Argo on analyses of the global ocean. Geophys Res Lett 34:L16605. doi:10.1029/2007GL030452
20â•… Seasonal and Decadal Prediction
539
Balmaseda MA, Vidard A, Anderson D (2008) The ECMWF ORA-S3 ocean analysis system. Mon Wea Rev 136:3018–3034 Balmaseda MA, Alves O, Arribas A, Awaji T, Behringer D, Ferry N, Fujii Y, Lee T, Rienecker M, Rosati T, Stammer D (2009) Ocean initialisation for seasonal forecasts. Oceanography 22:154–159 Balmaseda MA, Fujii Y, Alves O et€al (2010a) Initialisation for Seasonal and Decadal Forecasts. In: Hall J, Harrison DE,Stammer D (eds) Proceedings of OceanObs’09: sustained ocean observations and information for society, vol€2, ESA Publication WPP-306, Venice, 21–25 September 2009 Balmaseda MA, Fujii Y, Alves O et€al (2010b) Role of the ocean observing system in an end-toend seasonal forecasting system. Plenary paper, OceanObs’09, Venice, 21–25 September 2009. http://www.oceanobs09.net/plenary/index.php Behringer DW (2007) The Global Ocean Data Assimilation System at NCEP. 11th symposium on integrated observing and assimilation systems for atmosphere, oceans, and land surface, AMS 87th Annual Meeting, San Antonio, pp€12 Bell CJ, Gray LJ, Charlton-Perez AJ, Scaife AA (2009) Stratospheric communication of ENSO teleconnections to european winter. J Clim 22:4083–4096 Berner J, Doblas-Reyes FJ, Palmer TN, Shutts G, Weisheimer A (2008) Impact of a quasi-stochastic cellular automaton backscatter scheme on the systematic error and seasonal prediction skill of a global climate model. Philos Trans R Soc A 366:2561–2579 Bjerknes J (1969) Atmospheric teleconnections from the equatorial Pacific. Mon Wea Rev 97:163–172 Bonjean F, Lagerloef GSE (2002) Diagnostic model and analysis of surface currents in the tropical Pacific Ocean. J Phys Oceanogr 32:2938–2954 Burgers G, Balmaseda MA, Vossepoel FC, van Oldenborgh GJ, van Leeuwen PJ (2002) Balanced Ocean-Data Assimilation near the Equator. J Phys Oceanogr 32:2509–2519 Cagnazzo C, Manzini E (2009) Impact of the stratosphere on the winter tropospheric teleconnections between ENSO and the North Atlantic and European region. J Clim 22:1223–1238 Cazes-Boezio G, Menemenlis D, Mechoso CR (2008) Impact of ECCO ocean-state estimates on the initialisation of seasonal climate forecasts. J Clim 21:1929–1947 Chambers LE, Drosdowsky W (1999) Australian seasonal rainfall prediction using near global sea surface temperatures. AMOS Bull 12(3):51–55 Chang P, Yamagata T, Schopf P, Behera SK, Carton J, Kessler WS, Meyers G, Qu T, Schott F, Shetye S, Xie S-P (2006) Climate fluctuations of tropical coupled systems—the role of ocean dynamics. J Clim 19:5122–5174 Collins M, Booth BBB, Harris GR, Murphy JM, Sexton DMH, Webb MJ (2006) Towards quantifying uncertainty in transient climate change. Clim Dyn 27:127–147 Doblas-Reyes FJ, Weisheimer A, Deque M et€al (2009) Addressing model uncertainty in seasonal and annual dynamical ensemble forecasts. Quart J R Meteor Soc 135:1538–1559 Dommenget D, Stammer D (2004) Assessing ENSO simulations and predictions using adjoint ocean state estimation. J Clim 17:4301–4315 Fennessy MJ, Shukla J (1999) Impact of initial soil wetness on seasonal atmospheric prediction. J Clim 12(11):3167–3180 Fletcher CG, Hardiman SC, Kushner PJ, Cohen J (2009) The dynamical response to snow cover perturbations in a large ensemble of atmospheric GCM integrations. J Clim 22:1208–1222 Folland CK, Colman AW, Rowell DP, Davey MK (2001) Predictability of Northeast Brazil rainfall and real-time forecast skill. J Clim 14:1937–1958 (1987–1998) Fujii Y, Matsumoto S, Kamachi M, Ishizaki S (2011) Estimation of the equatorial Pacific salinity field using ocean data assimilation system. Adv Geosci (In Press) Goddard L, Graham NE (1999) The importance of the Indian Ocean for simulating rainfall anomalies over eastern and southern Africa. J Geophys Res 104:19099–19116 Hendon HH, Lim E, Wang G, Alves O, Hudson D (2009) Prospects for predicting two flavors of El Niño. Geophys Res Lett. doi:10.1029/2009GL040100
540
O. Alves et al.
Hudson D, Alves O, Hendon HH, Wang G (2010) The impact of atmospheric initialisation on seasonal prediction of tropical Pacific SST. Clim Dyn. doi:10.1007/s00382-010-0763-9 Hurrell J, Delworth TL, Danabasoglu G et€al (2010) Decadal climate prediction: opportunities and challenges. In: Hall J, Harrison DE, Stammer D (eds) Proceedings of OceanObs’09: sustained ocean observations and information for society, vol€ 2. ESA Publication WPP-306, Venice, 21–25 September 2009 Ineson S, Scaife AA (2008) The role of the stratosphere in the European climate response to El Nino. Nat Geosci 2:32–36 Jin F-F, Lin L, Timmermann A, Zhao J (2007) Ensemblemean dynamics of the ENSO recharge oscillator under statedependent stochastic forcing. Geophys Res Lett 34:L03807. doi:10.1029/2006GL027372 Jin EK, Kinter JL III, Wang B et€ al (2008) Current status of ENSO prediction skill in coupled ocean-atmosphere models. Clim Dyn 31:647–664. doi:10.1007/s00382-008-0397-3 Keenlyside N, Latif M, Jungclaus J, Kornblueh L, Roeckner E (2008) Advancing decadal-scale climate prediction in the North Atlantic Sector. Nature 453:84–88 Keppenne CL, Rienecker MM, Jacob JP, Kovach R (2008) Error covariance modeling in the GMAO ocean ensemble kalman filter. Mon Wea Rev 136:2964–2982. doi:10.1175/2007MWR2243.1 Kirtman BP, Pirani A (2009) The state of the art of seasonal prediction: outcomes and recommendations from the first world climate research program workshop on seasonal prediction. Bull Am Meteor Soc 90:455–458 Knight JR, Allan RJ, Folland CK et€al (2005) A signature of persistent natural thermohaline circulation cycles in observed climate. Geophys Res Lett 32:L20708. doi:1029/2005GL024233 Koster RD, Suarez MJ (2003) Impact of land surface initialisation on seasonal precipitation and temperature prediction. J Hydrometeor 4:408–423 Koster RD, Suarez MJ, Liu P et€al (2004) Realistic initialisation of land surface states: impacts on subseasonal forecast skill. J Hydrometeor 5:1049–1063 Koster RD,Guo Z, Dirmeyer PA et€al (2006) GLACE: The global land-atmosphere coupling experiment. Part I: overview. J Hydrometeor 7:590–610 Koster RD, Mahanama SPP, Yamada TJ et€ al (2010) Contribution of land surface initialisation to subseasonal forecast skill: first results from a multi-model experiment. Geophys Res Lett 37:L02402. doi:10.1029/2009GL041677 Kushnir Y, Robinson WA, Chang P, Robertson AW (2006) The physical basis for predicting Atlantic sector seasonal-to-interannual climate variability. J Clim 19:5949–5970 Lim E-P, Hendon HH, Hudson H, Wang G, Alves O (2009) Dynamical forecasts of inter-El Niño variations of tropical SST and Australian spring rainfall. Mon Wea Rev 137:3796–3810 Luo JJ, Masson S, Behera S, Yamagata T (2007) Experimental forecasts of the Indian ocean dipole using a coupled OAGCM. J Clim 20:2178–2190 Mantua NM, Hare SR, Zhang Y, Wallace JM, Francis RC (1997) A Pacific interdecadal climate oscillation with impacts on salmon production. Bull Am Meteor Soc 78:1069–1079 Marshall AG, Scaife AA (2009) Impact of the QBO on surface winter climate. J Geophys Res 114:D18110. doi:10.1029/2009JD011737 Mason SJ, Stephenson D (2008) How do we know whether seasonal climate forecasts are any good? In: Troccoli A, Harrison M, Anderson DLT, Mason SJ (eds) Seasonal climate: forecasting and managing risk. NATO Science Series. Springer, Dordrecht, pp€467 Martin MJ, Hines A, Bell MJ (2007) Data assimilation in the FOAM operational short-range ocean forecasting system: a description of the scheme and its impact. Quart J R Meteor Soc 133:981–995 Maycock AC, Keeley SPE, Charlton-Perez AJ, Doblas-Reyes FJ (2009) Stratospheric circulation in seasonal forecasting models: implications for seasonal prediction. Clim Dyn. doi:10.1007/ s00382-009-0665-x Murphy JM, Sexton DMH, Barnett DN, Jones GS, Webb MJ,Collins M, Stainforth DA (2004) Quantification of modelling uncertainties in a large ensemble of climate change simulations. Nature 430:768–772
20â•… Seasonal and Decadal Prediction
541
Neelin D, Battisti DS, Hirst AC, Jin F-F, Wakata Y, Yamagata T, Zebiak S (1998) ENSO theory. J Geophys Res 103:14261–14290 Oke PR, Schiller A, Griffin DA, Brassington GB (2005) Ensemble data assimilation for an eddyresolving ocean model of the Australian region. Quart J R Meteor Soc 131:3301–3311 Oldenborgh GJ van, Balmaseda MA, Ferranti L, Stockdale TN, Anderson DLT (2005) Did the ECMWF seasonal forecast model outperform a statistical model over the last 15 years? J Clim 18:2960–2969 Palmer TN, Alessandri A, Andersen U et€al (2004) Development of a European multimodel ensemble system for seasonal-to-interannual prediction (DEMETER). Bull Am Meteor Soc 85:853–872 Pham DT, Verron J, Roubaud MC (1998) A singular evolutive extended Kalman filter for data assimilation in oceanography. J Mar Syst 16:323–340 Philander SG (2004) Our affair with El nino. Princeton University Press, Princeton, pp€275 Pohlmann H, Jungclaus J, Marotzke J, Köhl A, Stammer D (2009) Improving predictability through the initialization of a coupled climate model with global oceanic reanalysis. J Clim 22:3926–3938 Power S, Casey T, Folland C, Colman A, Mehta V (1999) Inter-decadal modulation of the impact of ENSO on Australia. Clim Dyn 15:319–324 Rasmusson EM, Carpenter TH (1983) The relationship between eastern equatorial Pacific SSTs and rainfall over India and Sri Lanka. Mon Wea Rev 111:517–528 Rodwell MJ, Folland CK (2002) Atlantic air-sea interaction and seasonal predictability. Quart J R Meteor Soc 128:1413–1443 Ropelewski CF, Halpert MS (1987) Global and Regional Scale Precipitation Patterns Associated with the El Niño/Southern Oscillation. Mon Wea Rev 115:1606–1626 Saji NH, Yamagata T (2003) Possible impacts of Indian Ocean Dipole mode events on global climate. Clim Res 25:151–169 Saji, NH, Goswami BN, Vinayachandran PN, Yamagata T (1999) A dipole mode in the tropical Indian Ocean. Nature 401:360–363 Seneviratne SI, Koster RD, Guo Z et€al (2006) Soil moisture memory in agcm simulations: analysis of global land-atmosphere coupling experiment (GLACE) data. J Hydrometeor 7:1090–1112 Shi L, Alves O, Hendon HH, Wang G, Anderson D (2009) The role of stochastic forcing in ensemble forecasts of the 1997/98 El Niño. J Clim 22:2526–2540 Smith D, Cusack S, Colman A, Folland C, Harris G, Murphy J (2007) Improved surface temperature prediction for the coming decade from a global circulation model. Science 317:796–799 Spillman CM, Alves O (2009) Dynamical seasonal prediction of summer sea surface temperatures in the Great Barrier Reef. Coral Reefs. doi:10.1007/s00338-008-0438-8 Stainforth DA, Aina T, Christensen C, Collins M, Faull N, Frame DJ, Kettleborough JA, Knight S, Martin A, Murphy JM, Piani C, Sexton D, Smith LA, Spicer RA, Thorpe AJ, Allen MR (2005) Uncertainty in predictions of the climate response to rising levels of greenhouse gases. Nature 433:403–406 Stephenson D (2008) An Introduction to Probability Forecasting. In: Troccoli A, Harrison M, Anderson DLT and Mason SJ (eds) Seasonal climate: forecasting and managing risk. NATO Science Series. Springer, Dordrecht, pp€467 Stockdale TN (1997) Coupled ocean–atmosphere forecasts in the presence of climate drift. Mon Wea Rev 125:809–818 Stockdale TN, Balmaseda MA, Vidard A (2006) Tropical Atlantic SST prediction with coupled ocean-atmosphere GCMS. J Clim 19:6047–6061 Stockdale TN, Alves O, Boer G et€al (2010) Understanding and predicting seasonal to interannual climate variability—the producer perspective. White Paper for WCC3. Draft. http://www. wcc3.org/sessions.php?session_list=WS-3 Stockdale TN, Anderson DLT, Balmaseda MA, Doblas-Reyes F, Ferranti L, Mogensen K, Palmer TN, Molteni F, Vitart F (2011). ECMWF Seasonal forecast system 3 and its prediction of sea surface temperature. Clim Dyn (In Press)
542
O. Alves et al.
Ummenhofer CC, England MH, McIntosh PC, Meyers GA, Pook MJ, Risbey JS, Gupta AS, Taschetto AS (2009) What causes southeast Australia’s worst droughts? Geophys Res Lett. doi:10.1029/2008GL036801 Usui N, Ishizaki S, Fujii Y, Tsujino H, Yasuda T, Kamachi M (2006) Meteorological research institute multivariate ocean variational estimation (MOVE) system: some early results. Adv Space Res 37:806–822 van der Linden P, Mitchell JFB (eds) (2009) ENSEMBLES: Climate change and its impacts: summary of research and results from the ENSEMBLES project. Met Office Hadley Centre, Exeter, pp€160 Vialard J, Vitart F, Balmaseda M, Stockdale T, Anderson D (2005) An ensemble generation method for seasonal forecasting with an ocean-atmosphere coupled model. Mon Wea Rev 133:441–453 Wajsowicz RC (2007) Seasonal-to-interannual forecasting of tropical Indian Ocean sea surface temperature anomalies: potential predictability and barriers. J Clim 20:3320–3343 Walker G (1923) Correlation in seasonal variations of weather VIII. A preliminary study of world weather. Mem Indian Meteorol Dept 24(4):75–131 Walker GT (1924) Correlation in seasonal variations of weather IX. Mem Indian Meteorol Dept 24(9):275–332 Wang B, Lee J-Y, Kang I-S, et€al (2008a) Advance and prospectus of seasonal prediction: assessment of the APCC/CliPAS 14-model ensemble retrospective seasonal prediction (1980–2004). Clim Dyn. doi:10.1007/s00382-008-0460-0 Wang G, Alves O, Hudson D, Hendon H, Liu G, Tseitkin F (2008b) SST skill assessment from the new POAMA-1.5 System. BMRC Res Lett 8:2–6 (Bureau of Meteorology, Australia) Webster PJ, Moore AM, Loschnigg JP, Leben RR (1999) Coupled ocean–atmosphere dynamics in the Indian Ocean during 1997–1998. Nature 401:356–360 Weisheimer A, Doblas-Reyes FJ, Palmer TN et€al (2009) ENSEMBLES: a new multi-model ensemble for seasonal-to-annual predictions—Skill and progress beyond DEMETER in forecasting tropical Pacific SSTs. Geophys Res Lett 36(21):L21711 Yin Y, Alves O, Oke PR (2011) An ensemble ocean data assimilation system for seasonal prediction. Mon Wea Rev. doi:10.1175/2010MWR3419.1 Zebiak SE, Cane MA (1987) A model El nino-southern oscillation. Mon Wea Rev 115:2262–2278 Zhao M, Hendon HH (2009) Representation and prediction of the Indian Ocean dipole in the POAMA seasonal forecast model. Quart J R Meteor Soc 135(639):337–352
Part VII
Evaluation
Chapter 21
Dynamical Evaluation of Ocean Models Using the Gulf Stream as an Example Harley E. Hurlburt, E. Joseph Metzger, James G. Richman, Eric P. Chassignet, Yann Drillet, Matthew W. Hecht, Olivier Le Galloudec, Jay F. Shriver, Xiaobiao Xu and Luis Zamudio
Abstract╇ The Gulf Stream is the focus of an effort aimed at dynamical understanding and evaluation of current systems simulated by eddy-resolving Ocean General Circulation Models (OGCMs), including examples with and without data assimilation and results from four OGCMs (HYCOM, MICOM, NEMO, and POP), the first two including Lagrangian isopycnal coordinates in the vertical and the last two using fixed depths. The Gulf Stream has been challenging to simulate and understand. While different non-assimilative models have at times simulated a realistic Gulf Stream pathway, the simulations are very sensitive to small changes, such as subgrid-scale parameterizations and parameter values. Thus it is difficult to obtain consistent results and serious flaws are often simulated upstream and downstream of Gulf Stream separation from the coast at Cape Hatteras. In realistic simulations, steering by a key abyssal current and a Gulf Stream feedback mechanism constrain the latitude of the Gulf Stream near 68.5°W. Additionally, the Gulf Stream follows a constant absolute vorticity (CAV) trajectory from Cape Hatteras to ~70°W, but without the latitudinal constraint near 68.5°W, the pathway typically develops a northern or southern bias. A shallow bias in the southward abyssal flow of the Atlantic Meridional Overturning Circulation (AMOC) creates a serious problem in many simulations because it results in abyssal currents along isobaths too shallow to feed into the key abyssal current or other abyssal currents that provide a similar pathway constraint. Pathways with a southern bias are driven by a combination of abyssal currents crossing under the Gulf Stream near the separation point and the increased opportunity for strong flow instabilities along the more southern route. The associated eddy-driven mean abyssal currents constrain the mean pathway to the east. Due to sloping topography, flow instabilities are inhibited along the more northern routes west of ~69°W, especially for pathways with a northern bias. The northern bias occurs when the abyssal current steering constraint needed for a realistic pathway is missing or too weak and the simulation succumbs to the demands of linear dynamics for an overshoot pathway. Both the wind forcing and the upper ocean branch of H. E. Hurlburt () Oceanography Division, Naval Research Laboratory Stennis Space Center, Mississippi, MS, USA e-mail:
[email protected] A. Schiller, G. B. Brassington (eds.), Operational Oceanography in the 21st Century, DOI 10.1007/978-94-007-0332-2_21, © Springer Science+Business Media B.V. (outside the USA) 2011
545
546
H. E. Hurlburt et al.
the AMOC contribute to those demands. Simulations with a northern pathway bias were all forced by a wind product particularly conducive to that result and they have a strong or typical AMOC transport with a shallow bias in the southward flow. Simulations forced by the same wind product (or other wind products) that have a weak AMOC with a shallow bias in the southward limb exhibit Gulf Stream pathways with a southern bias. Data assimilation has a very positive impact on the model dynamics by increasing the strength of a previously weak AMOC and by increasing the depth range of the deep southward branch. The increased depth range of the southward branch generates more realistic abyssal currents along the continental slope. This result in combination with vortex stretching and compression generated by the data-assimilative approximation to meanders in the Gulf Stream and related eddies in the upper ocean yield a model response that simulates the Gulf Streamrelevant abyssal current features seen in historical in situ observations, including the key abyssal current near 68.5°W, a current not observed in the assimilated data set or corresponding simulations without data assimilation. In addition, the model maintains these abyssal currents in a mean of 48 14-day forecasts, but does not maintain the strength of the Gulf Stream east of the western boundary.
21.1â•…Introduction Ocean models run with atmospheric forcing but without ocean data assimilation are useful in studies of ocean model dynamics and simulation skill. Models that give realistic simulations with accurate dynamics, when run without data assimilation, are essential for eddy-resolving ocean prediction because of the multiple roles that ocean models must play in ocean nowcasting and forecasting, including dynamical interpolation during data assimilation, representing sparsely observed subsurface ocean features from the mixed layer depth to abyssal currents, converting atmospheric forcing into ocean responses, imposing topographic and geometric constraints, performing ocean forecasts, providing boundary and initial conditions to nested regional and coastal models, and providing forecast surface temperature to coupled atmosphere and sea ice models. A wide range of ocean dynamics contribute to these different roles. Here we focus on evaluating and understanding the dynamics of mid-latitude ocean currents simulated by state-of-the-art, eddy-resolving ocean general circulation models (OGCMs), using the Gulf Stream as an example. Dynamical understanding and evaluation of current systems simulated by OGCMs has been a challenge because of the complexity of the models and the current systems, a topic discussed in recent reviews by Chassignet and Marshall (2008) and Hecht and Smith (2008) in relation to the Gulf Stream and North Atlantic. In some regions greater progress has been made. Tsujino et€al. (2006) investigated the dynamics of large amplitude Kuroshio meanders south of Japan. Usui et€al. (2006) used the same model to make Kuroshio forecasts from a data-assimilative initial state, typically demonstrating 40 to 60-day forecast skill south of Japan. Usui et€al. (2008a, b) also used the model in dynamical studies of a 1993–2004 data-assimilative hindcast. Hurlburt et€al. (2008b) examined OGCM dynamics and their relation
21â•… Dynamical Evaluation of Ocean Models Using the Gulf Stream as an Example
547
to the underlying topography in studying mean Kuroshio meanders east of Japan and mean currents in the southern half of the Japan/East Sea. The simulations were consistent with observations and with dynamics found in purely hydrodynamic models with lower vertical resolution and vertically-compressed but otherwise realistic topography confined to the lowest layer. Consistent with observations (Gordon et€al. 2002), the same Japan/East Sea OGCM simulation modeled the dynamics of intrathermocline eddy formation in that region, as discussed in Hogan and Hurlburt (2006). These are dynamics that could not be simulated by the purely hydrodynamic model. Hurlburt et€al. (2008b) also investigated OGCM dynamics in simulating the Southland Current system east of South Island, New Zealand, where the topography of the Campbell Plateau and the Chatham Rise intrude well into the stratified ocean so that the design of the low vertical resolution model did not apply. In that case an alternative approach was used to investigate the dynamics. Recent observational evidence was sufficient to provide strong support for the results of the study. In dynamical evaluation of the Gulf Stream simulations by eddy-resolving global and basin-scale OGCMs, we adopt an augmented version of the approach used by Hurlburt et€al. (2008b) for OGCM simulations of the Kuroshio and Japan/East Sea. Thus we build from an explanation of Gulf Stream separation from the western boundary and its pathway to the east in Hurlburt and Hogan (2008). This explanation was derived using results from a 5-layer hydrodynamic isopycnal model with vertically-compressed but otherwise realistic topography confined to the lowest layer. It was tested versus observational evidence and theory, parts of the latter contributing directly to the explanation. In Sect.€21.2 we discuss the explanation and related 5-layer model results, theory, and observational evidence. In Sect.€21.3 we evaluate Gulf Stream dynamics in eddy-resolving OGCM simulations by the HYbrid Coordinate Ocean Model (HYCOM) (Bleck 2002), the Miami Isopycnic Coordinate Ocean Model (MICOM) (Bleck and Smith 1990), the Nucleus for European Modelling of the Ocean (NEMO) (Madec 2008), as used in the French Mercator ocean prediction effort, and the Parallel Ocean Program (POP) (Smith et€al. 2000). Both simulations with a realistic Gulf Stream and those with a variety of unrealistic features are assessed and specific deficiencies are identified. In Sect.€21.4 we assess the impacts of data assimilation on variables relevant to Gulf Stream dynamics that are sparsely observed, in some cases not observed at all in real time. Are realistic model dynamics maintained in data-assimilative models? Are unrealistic dynamics improved? What are the impacts of dynamics on Gulf Stream forecast skill?
21.2â•…Dynamics of Gulf Stream Boundary Separation and Its Pathway to the East 21.2.1 Linear Model Simulation of the Gulf Stream As an initial step, we examine a linear equivalent barotropic solution with the same wind forcing and upper ocean transport for the Atlantic meridional overturning cir-
548
H. E. Hurlburt et al.
culation (AMOC) as the nonlinear solutions discussed in Sect.€ 21.2. The model boundary is located at the shelf break and the resolution is comparable to that used in nonlinear solutions discussed later in this chapter. The spun up mean solution has a Sverdrup (1947) interior, Munk (1950) western boundary currents and is consistent with the Godfrey (1989) island rule, except that, unlike Munk (1950), the solution is obtained by running a numerical model with horizontal friction applied everywhere. Figure€21.1 depicts the mass transport streamfunction from a 1/16º l.5 layer linear reduced-gravity simulation (with the lower layer infinitely deep and at rest) forced by the smoothed Hellerman and Rosenstein (1983) wind stress climatology plus the northward upper ocean flow of a 14€Sv AMOC. In comparison to the overlaid mean IR northwall pathway that lies along the northern edge of the Gulf Stream, the linear solution gives two unrealistic pathways, a broad one centered near the observed separation latitude (35.5ºN) that extends eastward and a second one with nearly the same transport extending northward along the western boundary. The eastward pathway is wind-driven (~22€Sv) and the northward pathway has a 14€Sv AMOC component plus an 8€Sv wind-driven component, but both pathways contribute to a situation where ~31€Sv out of 44€Sv (~70%) separate from the western boundary north of the observed separation latitude. From Fig.€21.1 it is easy to appreciate the challenge of simulating an accurate nonlinear Gulf Stream pathway 45 N 40 N 35 N 30 N 25 N 20 N 15 N 10 N 90 W
80 W
70 W
60 W
50 W
40 W
30 W
20 W
10 W
0
Fig. 21.1↜渀 Mean transport streamfunction (Ψ) from a 1/16°, 1.5-layer linear reduced-gravity simulation forced by the smoothed Hellerman and Rosenstein (1983) wind stress climatology and the northward upper ocean flow (14€Sv) of the Atlantic meridional overturning circulation (AMOC), forcing used for all of the simulations in Sect.€21.2. The contour interval is 2€Sv. A 15-year mean (1982–1996) Gulf Stream IR northwall pathwayâ•›±1σ by Cornillon and Sirkes (unpublished) is overlaid. This pathway has 0.1° longitudinal resolution and is based on an average of 674 data points per 0.1° increment between 76° and 55°W. An earlier analysis of this frontal pathway and its variability (based on data from 1982–1989) is discussed in Lee and Cornillon (1996). The streamfunction shown here covers the 9–47°N model domain used by all the nonlinear simulations discussed in Sect.€21.2. (From Hurlburt and Hogan 2008, as adapted from Townsend et€al. 2000)
21â•… Dynamical Evaluation of Ocean Models Using the Gulf Stream as an Example
549
in an ocean model. See Townsend et€al. (2000) for linear solutions from 11 different wind stress climatologies.
21.2.2 I mpacts of the Eddy-Driven Abyssal Circulation and the Deep Western Boundary Current (DWBC) on Gulf Stream Boundary Separation and Its Pathway to the East It has been a popular theory, proposed by Thompson and Schmitz (1989), that the DWBC affects Gulf Stream separation from the western boundary as it passes underneath. To investigate this hypothesis Hurlburt and Hogan (2008) used a nonlinear 5-layer hydrodynamic isopycnal model covering the same domain shown in Fig.€21.1. They also used monthly climatological wind forcing and included a 14€Sv AMOC, the latter via inflow and outflow ports in the northern and southern boundaries. Figure€21.2 depicts the mean sea surface height (SSH) from six simulations.
DWBC 44 N 42 N
– 64 – 48 – 32 – 16 0
16 32 48
1/16º Cb =.002
(cm)
NO DWBC
1/16º Cb =.002
40 N 38 N 36 N 34 N 44 N 42 N
a
b
c
d
1/32º Cb =.002
1/32º Cb =.002
40 N 38 N 36 N 34 N 44 N 42 N
1/32º Cb =.02
1/32º Cb =.02
40 N 38 N 36 N 34 N
e
80 W 75 W
70 W
65 W
60 W
55 W
50 W
45 W
40 W
f
75 W
70 W
65 W
60 W
55 W
50 W
45 W 40 W
Fig. 21.2↜渀 Mean SSH from six 5-layer Atlantic Ocean simulations (9–47°N) zoomed into the Gulf Stream region between Cape Hatteras and the Grand Banks. The simulations depicted in a, c and e include a DWBC while those in b, d and f do not. a and b Depict results from 1/16° simulations. c–f From corresponding 1/32° simulations. a–d With a coefficient of quadratic bottom friction, Cbâ•›=â•›0.002. d and f with a 10× increase to Cbâ•›=â•›0.02. The northward upper ocean flow of the AMOC is included in all six simulations. The Laplacian coefficient of isopycnal eddy viscosity is Aâ•›=â•›20 (10) m2/s for the 1/16º (1/32°) simulations. The SSH contour interval is 8€cm. The mean Gulf Stream IR northwall pathway ±1σ by Cornillon and Sirkes is overlaid on each panel. For more information about the simulations used in Sect.€21.2, see Hurlburt and Hogan (2008). (From Hurlburt and Hogan 2008)
550
H. E. Hurlburt et al.
The northward upper ocean component of the AMOC resides in the top 4 layers and is always included, while the DWBC residing in the abyssal layer is included in the simulations in the left column of Fig.€21.2 and turned off in the simulations in the right column. Since the model is purely hydrodynamic, the DWBC can be turned off without altering the watermass characteristics. In the three rows of Fig.€21.2 the model resolution is varied in tandem with the horizontal friction and in the bottom row the bottom friction is increased 10-fold to damp the eddy-driven abyssal circulation. East of 68ºW all of the simulations give similar, generally-realistic Gulf Stream pathways, except near 50ºW, where the simulations with a DWBC exhibit two mean pathways (inner and outer meanders) at the location of the Gulf Stream transition to the North Atlantic Current as it rounds the southern tip of the Grand Banks, a phenomenon discussed dynamically in Hurlburt and Hogan (2008). All three of the simulations with a DWBC and one of the simulations without it exhibit a realistic mean Gulf Stream pathway west of 68ºW, but the other two simulations without a DWBC exhibit pathways that overshoot the observed separation latitude in accord with the constraint of linear theory on the flow. These results indicate an abyssal current impact on the pathway west of 68ºW. To investigate the impacts of abyssal currents on the Gulf Stream pathway, we use a two-layer theory for abyssal current steering of upper ocean current pathways (Hurlburt and Thompson 1980; Hurlburt et€al. 1996, 2008b). In a two-layer model with no diapycnal mixing, the continuity equation for layer 1 is
hlt + v1 · ∇h1 + h1 ∇ · v1 = 0,
(21.1)
where h1 is upper layer thickness, t is the time derivative and vi is the velocity in layer i. The geostrophic component of the advective term in (21.1) can be related to the geostrophic velocity (vig) in layer 2 by
v1g · ∇h1 = v2g · ∇h1 ,
(21.2)
k × f (v1g − v2g ) = −g ∇h1 ,
(21.3)
because from geostrophy,
v1gâ•›–â•›v2g is parallel to contours of h1. In (21.3) k is a unit vector in the vertical, fâ•›=â•›2ωsinθ is the Coriolis parameter, ω is the Earth’s rotation rate, θ is latitude, g′â•›=â•›g(↜ρ2â•›−â•›ρ1)/ρ2 is the reduced gravity due to buoyancy, g is the gravitational acceleration of the Earth, and ρi is the water density in layer i. Since geostrophy is typically a very good approximation outside the equatorial wave guide and normally near-surface currents are much stronger than abyssal currents, then usually |v1|››|v2|, making h1 a good measure of v1 under these conditions. From the preceding we see that abyssal currents can advect upper layer thickness gradients and therefore the pathways of upper ocean currents. Abyssal current advection of upper ocean current pathways is strengthened when strong abyssal currents intersect upper ocean currents at nearly right angles, but often the end result of this advection is near barotropy because the advection is reduced as v1 and v2 become more nearly parallel (or antiparallel). This theory has proven useful in understanding the dynamics of ocean models with higher vertical resolution, when all of the following conditions are satisfied:
21â•… Dynamical Evaluation of Ocean Models Using the Gulf Stream as an Example
551
(a) the flow is nearly geostrophically balanced, (b) the barotropic and first baroclinic modes are dominant, and (c) the topography does not intrude significantly into the stratified ocean. Additionally, the interpretation in terms of near-surface currents applies when |vnear sfc|››|vabyssal|. Note the theory does not apply at low latitudes because of (a) and (b), but should be useful in large parts of the stratified ocean, even where current systems are relatively weak, as seen in the well-stratified southern half of the Japan/East Sea (Hurlburt et€al. 2008b). While abyssal currents driven by any means can steer upper ocean current pathways, baroclinic or mixed barotropicbaroclinic instability is an important source of abyssal currents because baroclinic instability is very effective in transferring energy from the upper to abyssal ocean. These eddy-driven abyssal currents are constrained to follow the geostrophic contours of the topography and in turn can steer the pathways of upper ocean currents, including their mean pathways. This upper ocean—topographic coupling via flow instabilities requires that the physics of baroclinic instability be very well resolved in order to obtain sufficient downward transfer of energy. As a result, this type of coupling is a key criterion in distinguishing between eddy-resolving and eddypermitting ocean simulations, in regions where it occurs (Hurlburt et€ al. 2008b). Results from this model and ocean models discussed in Sect.€21.3 indicate that the upper ocean—topographic coupling requires the first baroclinic Rossby radius of deformation be resolved by at least 6 grid intervals and even higher resolution is required for realistic eastward penetration of inertial jets. This coupling also highlights the need for eddy-resolving ocean models in ocean prediction systems and in climate prediction models, as discussed in Hurlburt et€al. (2008a, 2009). Based on the preceding discussion, we look in Fig.€21.3 for abyssal currents west of ~68ºW that may advect the simulated Gulf Stream pathways in Fig.€21.2. We start with the simulation shown in Figs.€21.2c and 21.3c because it has 1/32º resolution, the standard bottom friction, and a DWBC. In that simulation abyssal currents pass under the Gulf Stream near 68.5ºW, 72ºW, and the western boundary, all generally southward. The abyssal currents near 68.5ºW and 72ºW cross under at large angles and could clearly advect the Gulf Stream pathway, but the abyssal current adjacent to the western boundary is nearly antiparallel as it crosses under the Gulf Stream, a point noted by Pickart (1994) based on observations, and thus has a weak steering effect on the Gulf Stream pathway. The corresponding simulation without a DWBC (Figs.€ 21.2d and 21.3d) has nearly the same Gulf Stream pathway with an even stronger abyssal current crossing under it near 68.5ºW. The two other simulations without a DWBC have only a weak mean abyssal current crossing under it at this longitude (4€cm/s). None of the simulations without a DWBC have an abyssal current crossing under near 72ºW, while all of the simulations with a DWBC have one fed by two branches from the north side. The 1/32º simulations with a DWBC and standard (Figs.€21.2c and 21.3c) or high bottom friction (Figs.€21.2e and 21.3e) have nearly the same Gulf Stream pathway between the western boundary and 68ºW, but in the simulation with high bottom friction the abyssal currents crossing under the Gulf Stream near 72ºW are extremely weak. Thus, the abyssal current crossing under the Gulf Stream near 68.5ºW is clearly the one that is essential for the model’s simula-
e
b
a
75 W
70 W
65 W
60 W
55 W
50W
45 W
40 W
f
d
b
75 W
70 W
65 W
60 W
55 W
50 W
45 W
40 W
Fig. 21.3↜渀 Same simulations as Fig.€21.2 but depicting mean abyssal currents (arrows) overlaid on isotachs (in cm/s). The DWBC is most easily seen paralleling the northern model boundary north of 41°N between 65 and 51°W in panels (a, c, e). In the simulations with no DWBC (panels b, d, f) that current is not present. (From Hurlburt and Hogan 2008)
80 W
34 N
36 N
38 N
40 N
42 N
44 N
34 N
36 N
38 N
40 N
42 N
44 N
34 N
36 N
38 N
40 N
42 N
44 N
552 H. E. Hurlburt et al.
21â•… Dynamical Evaluation of Ocean Models Using the Gulf Stream as an Example
553
tion of a realistic Gulf Stream pathway between the western boundary and 68ºW. Further, the DWBC is not necessary for simulation of a realistic Gulf Stream pathway, but it augments the key abyssal current sufficiently for that to occur in the two simulations with the weaker eddy-driven abyssal circulations. The 1/32º simulation with standard bottom friction and a DWBC (Fig.€21.3c) is used in a zoom of the mean abyssal currents with the addition of topographic contours (Fig.€21.4a). The plotted contours are for the vertically-uncompressed (real) topography to facilitate comparisons between model and observed abyssal currents in relation to topographic features. Figure€21.4b depicts mean abyssal currents and uncompressed topography from a corresponding 1/8º eddy-permitting simulation over a larger region with the zoom region in Fig.€21.4a marked with a box. It should be noted that eddy-resolving and eddy-permitting OGCMs with higher vertical resolution and thermodynamics are typically characterized by their equatorial resolution, whereas the simulations in Sect.€21.2 are characterized by mid-latitude resolution. Thus, the corresponding equatorial resolution of the simulations in Fig.€ 21.4a, b would be 1/24 and 1/6º, respectively. Unlike the 1/32º simulation (Fig.€21.4a), the abyssal circulation in the 1/8º model is dominated by the DWBC, which crosses under the observed location of the Gulf Stream near 72ºW, and the eddy-driven abyssal circulation is extremely weak (Fig.€21.4b). In particular, the 1/8º model does not simulate the key abyssal current near 68.5ºW. The DWBC augments this current in two of the simulations (Fig.€21.3a, e) because the DWBC and the eddy-driven abyssal circulation interact and become intertwined in the eddy-resolving simulations. The surface circulation in the 1/8º model is basically a wiggly version of the linear solution (Hurlburt and Hogan 2000, their Fig.€ 4a), who also present numerous model-data comparisons for the 1/16º simulation in Figs.€21.2a and 21.3a and the 1/32º simulation in Figs.€21.2c and 21.3c. In addition to the abyssal current adjacent to the western boundary, abyssal currents are seen crossing under the Gulf Stream via three different pathways centered over different isobaths between the western boundary and 68ºW. North of the Gulf Stream these pathways are centered over the 4,200, 3,700 and 3,100€m isobaths, the first crossing under near 68.5ºW, the other two crossing under in a confluence near 72ºW. All three abyssal currents cross isobaths to deeper depths while passing under the Gulf Stream. They do this to conserve potential vorticity in relation to the downward north to south slope of the base of the thermocline in accord with the theory of Hogg and Stommel (1985). The two currents over deeper isobaths retroflect toward the east and then take a variety of simple to complex pathways into the ocean interior (complex even in the mean, e.g. Fig.€21.3c). Ultimately all of these pathways emerge from the interior as a single strong abyssal current along a gentle escarpment. That current exits Fig.€ 21.4a near 72ºW and rejoins the DWBC along the continental slope near 33ºN. In contrast, the branch centered over the 3,100€m isobath north of the stream continues along the continental slope south of the stream (centered above the 3,700€m isobath). Each cross-under pathway is influenced by specific features of the topography and each also flows along one side of an associated eddy-driven abyssal gyre centered directly beneath the Gulf Stream. These gyres are located in regions where the slopes of the topography and the base of the thermocline are
554 41 N
H. E. Hurlburt et al.
a 1/32°
40 N
39 N
38 N
37 N
36 N
35 N
34 N 76 W 44 N
74 W
72 W
70 W
68 W
66 W
b
42 N 1/8° 40 N 38 N 36 N 34 N 32 N 30 N
80 W
75 W 5 cm/s
70 W 0
1
2
65 W 3
4
60 W 5
6
7
55 W 8
9
10
50 W 11
12
45 W
(cm/s)
Fig. 21.4↜渀 a Zoom of Fig.€21.3c with (full amplitude, uncompressed) depth contours (in m) overlaid to facilitate geographical co-location in the model-data comparisons. b Same as a but covering a larger region from a corresponding 1/8° simulation overlaid with a box outlining the region covered by a. In the 1/8º simulation Aâ•›=â•›100€m2/s and Cbâ•›=â•›0.002. (From Hurlburt and Hogan 2008)
21â•… Dynamical Evaluation of Ocean Models Using the Gulf Stream as an Example
555
matched closely enough to create regions of quite uniform potential vorticity for abyssal currents, as shown in Hurlburt and Hogan (2008). The shallowest and westernmost gyre is anticyclonic, while the two associated with eastward retroflections into the interior are cyclonic, all three in accord with the sign of the relative vorticity generated due to topographic constraints on the pathways of the associated abyssal currents as they cross under the Gulf Stream (shown in Hurlburt and Hogan 2008).
21.2.3 O bservational Evidence of Abyssal Currents in the Gulf Stream Region Figure€21.5 (bottom) (from Johns et€al. 1995) presents observational evidence for the key abyssal current crossing under the Gulf Stream near 68.5ºW, including current speeds similar to the model, currents crossing isobaths to deeper depths beneath the Gulf Stream, and a closed cyclonic circulation. Additionally, the currents above the shallowest isobaths within the observational array flow along isobaths that would feed into the retroflecting abyssal current that crosses under the Gulf Stream near 72ºW. Figure€21.6 (from Pickart and Watts 1990) provides a composite of historical abyssal current measurements 100–300€ m above the bottom. It provides striking evidence of the complete cyclonic abyssal gyre centered near 37ºN, 71ºW with current speeds similar to the model. Another salient observation is the ~12.5€cm/s west-southwestward current near 34.5ºN, 71.1ºW that corroborates the strong abyssal current along the gentle escarpment in Fig.€21.4a (10.5€cm/s at the same location in the model). Like the model (Fig.€ 21.4a), the observation-based abyssal current schematic of Schmitz and McCartney (1993, their Fig.€ 12a) depicts a retroflecting abyssal current pathway that later rejoins the DWBC, in addition to a pathway that continues along the continental slope. These two pathways are also consistent with Range and Fixing of Sound (RAFOS) float trajectories at 3,500€m depth discussed in Bower and Hunt (2000). RAFOS floats that crossed under the Gulf Stream west of ~71ºW continued generally southward along a deeper isobath of the continental slope, while floats crossing under east of ~71ºW retroflected into the interior, most of them taking complex eddying trajectories, but of the six retroflecting trajectories shown in Bower and Hunt (2000, their Fig. 7), the one that crossed under at the location of the key abyssal current (near 69ºW) (their Fig. 7j) took an eddying trajectory en route to a small amplitude double retroflection, first to the east (at 36.7ºN, 70.1ºW) and then to the west (at 36.0ºN, 68.4ºW) before rapidly following a nearly straight-line trajectory along the gentle escarpment, an overall trajectory in good agreement with the model mean in Fig.€21.4a and one that provides additional evidence for the strong eddy-driven abyssal current along the gentle escarpment (seen in the southern part of Fig.€21.4a). This abyssal current (also seen in Fig.€21.6) is completely absent in the 1/8º eddy-permitting simulation (Fig.€21.4b), as are the observed cyclonic abyssal gyre centered near 37ºN, 71ºW (Fig.€21.6) and the abyssal current observed crossing under the Gulf Stream between 68º and 69ºW (Fig.€21.5).
40° N
1000 m 2000
400 m 39° N
I1
H2
38° N
HC
G2
3500
Latitude
I2
H3
3000
H4
4000
G3
4100
37° N
I3 M13 H5
I4
4200 4300
I5
H6
CS (2852 m)
4400
HF
36° N
4500 4600
4700 4800 4900 5000
50 cm/s
5100
40° N
N
1000 m 2000
3500 m 39° N
I1
H2
38° N
G2
Latitude
3500 4000
I3
H4
M13 I4
G3
4100
37° N
I2
H3
3000
H5
4200
I5
4300
H6
4400
36° N
4500 4600
N
4700 4800 4900 5000
35° N 72° W
5 cm/s
5100
70° W
68° W
66° W
Longitude
Fig. 21.5↜渀 Mean current meter velocities at 400€ m (↜top) and 3,500€ m (↜bottom) over the entire deployment, June 1988–August 1990. All of the vectors represent 26-month means except at sites H5 and M13, which are approximately 1-year means. (From Johns et€al. 1995)
21â•… Dynamical Evaluation of Ocean Models Using the Gulf Stream as an Example 77°
75°
73°
71°
69°
67°
557 65°
40°
38°
36° CURRENTS 100 – 300 m OFF BOTTOM 0
10 cm/s
34°
Fig. 21.6↜渀 Mean current meter velocities 100–300€m above the bottom from historical measurements collected in the middle Atlantic Bight. The record lengths of the measurements vary from 4 months to 2 years, and the box associated with each vector represents the uncertainty of the mean, typically 1–2€cm/s. (From Pickart and Watts 1990)
21.2.4 G ulf Stream Separation and Pathway Dynamics, Part I: Abyssal Current Impact An eddy-driven abyssal current, the local topographic configuration, and a Gulf Stream feedback mechanism constrain the latitude of the Gulf Stream near 68.5ºW. To help illustrate the steps explaining this statement, Fig.€ 21.7 depicts the mean depth of the base of the model thermocline overlaid with the same mean abyssal currents and topographic contours as Fig.€21.4a. The results are from the same 1/32º simulation with a DWBC shown in Figs.€21.2c, 21.3c, and 21.4a. The steps in the explanation are (1) an eddy-driven abyssal current, possibly augmented by the DWBC, approaches from the northeast and advects the Gulf Stream pathway southward, i.e. prevents the overshoot pathway seen in Figs.€21.2b, f. (2) To conserve potential vorticity, the abyssal current crosses to deeper depths while passing under the Gulf Stream (Hogg and Stommel 1985), a feedback mechanism that allows the Gulf Stream to help determine its own latitude. (3) Due to the topographic configuration, the passage to deeper depths requires curvature toward the east and generation of positive relative vorticity. (4) Once the abyssal current becomes parallel to the Gulf Stream, further southward advection of the Gulf Stream
558
H. E. Hurlburt et al.
1
LQP
1
1 1
1
1
1 :
:
:
:
:
:
FPV
Fig. 21.7↜渀 Same as Fig.€21.4a but with isotachs (↜in color) replaced by the mean depth at the base of the model thermocline (↜in m) from the same simulation (↜in color), i.e. the mean depth of the interface between layer 4 and layer 5 (the abyssal layer) from the 1/32º simulation depicted in Figs.€21.2c and 21.3c. (From Hurlburt and Hogan 2008)
pathway is halted. (5) The local latitude of the Gulf Stream is determined by the northernmost latitude where the abyssal current can become parallel to the Gulf Stream. (6) Due to constraints of the local topographic configuration on this process, the resulting local Gulf Stream latitude is not very sensitive to the strength of the abyssal current, once it is sufficient to perform the advective role. However, the results of these dynamics would be sensitive to the location of abyssal currents in relation to the isobaths, the accuracy of the model in representing key topographic features, and the depth change in the base of the thermocline across the Gulf Stream. Essentially the same explanation can be applied to the effects of the abyssal current crossing under the Gulf Stream near 72ºW (when present and sufficiently strong) and to abyssal currents that develop either cyclonic or anticyclonic curvature and become either parallel or antiparallel to the Gulf Stream while crossing underneath. However, the response to the abyssal current near 72ºW is minimal as evidenced in Figs.€21.2 and 21.3 and an impact is visible only in the 1/32º simulation with a DWBC and standard bottom friction (Cbâ•›=â•›0.002) (Fig.€21.2c). In Fig.€21.2c there is a straightening of the Gulf Stream pathway over ~73−70ºW not seen in the other figure panels. This phenomenon is also evident in the overlaid mean Gulf Stream IR northwall frontal pathway and in the Gulf Stream pathway as depicted by
21â•… Dynamical Evaluation of Ocean Models Using the Gulf Stream as an Example
559
the 12ºC isotherm at 400€m depth, the latter shown in Watts et€al. (1995). An explanation for the slight impact of this abyssal current on this Gulf Stream simulation is discussed in the next subsection. Additionally, it should be noted that the scale of the eddy-driven mean abyssal gyres beneath the Gulf Stream is similar to the width of the stream (Fig.€21.7) and related to regions of nearly uniform potential vorticity beneath the stream (Hurlburt and Hogan 2008), where slopes of topography and the base of the thermocline are quite well matched. These gyres are not related to mean meanders in the Gulf Stream. In contrast, the Kuroshio exhibits two mean northward meanders just east of where the Kuroshio separates from the coast of Japan, meanders that are dynamically related to eddy-driven mean abyssal gyres, as discussed in Hurlburt et€al. (1996, 2008b).
21.2.5 G ulf Stream Boundary Separation as an Inertial Jet Following a Constant Absolute Vorticity (CAV) Trajectory Constraint of the Gulf Stream latitude near 68.5ºW is not a sufficient explanation of the Gulf Stream pathway between the western boundary and 69ºW. Further, the abyssal current crossing under the Gulf Stream near 72ºW demonstrated little effect on the pathway. Thus, there must be another essential contribution to Gulf Stream pathway dynamics over that longitude range. Using along-track data from four satellite altimeters, Fig.€21.8 depicts only a narrow band of high SSH variability along the Gulf Stream west of 69ºW, indicating a relatively stable pathway segment in that region. Thus, we test the relevance of a particular type of theoretical inertial jet pathway, namely a CAV trajectory (Rossby 1940; Haltiner and Martin 1957; Reid 1972; Hurlburt and Thompson 1980, 1982). In a nonlinear 1.5 layer reduced-gravity model, a CAV trajectory requires a frictionless steady free jet with the streamline at the core of the current following contours of constant SSH and layer thickness. The latter requires geostrophic balance so that conservation of potential vorticity becomes conservation of absolute vorticity along a streamline at the core of the current. Accordingly, the simulations in Fig.€21.2 were tested to see if (a) the mean path of the current core in the top layer of the model (black line in Fig.€ 21.9) overlaid an SSH contour (yellow-green line in Fig.€21.9) and (b) there was a narrow band of high SSH variability along the current core between the western boundary and 69ºW (plotted in color in Fig.€21.9). Following Reid (1972) and Hurlburt and Thompson (1980, 1982), the CAV trajectories were calculated from
cos α = cos αo + 1/2y2 /r 2 − y/γo ,
(21.4)
which is an integrated form of the differential equation that assumes the velocity at the core of the current, υc, is a constant and where râ•›=â•›(υc/β)½, β is the variation of
560
H. E. Hurlburt et al.
45 N
40 N
35 N
30 N 80 W
70 W
75 W 0
65 W 0.1
60 W
55 W
0.2 meters
50 W 0.3
45 W
40 W
35 W
0.4
Fig. 21.8↜渀 Along-track SSH variability from quasi-contemporaneous satellite altimeter data in 4 different orbits overlaid on topographic contours (depth in m), Jason-1 over the period 15 Jan. 2002–18 Oct. 2007, GFO over 15 July 1999–12 Dec. 2007, Envisat over 24 Sept. 2002–29 Oct. 2007, and Topex in an interleaved orbit over 16 Sept. 2002–8 Oct. 2005. The tracks are overlaid in the following order from top to bottom: (1) Envisat, (2) GFO, (3) Jason-1, and (4) Topex interleaved. (Provided by Gregg Jacobs, NRL). (From Hurlburt and Hogan 2008)
the Coriolis parameter with latitude, α is the angle of the current with respect to the positive x-axis on a β-plane, y is the distance of the trajectory from the x-axis, γ is the trajectory radius of curvature, and the subscript o indicates values at the origin of the trajectory calculation (here at an inflection point where γo → ∞). The amplitude (b) (here the northernmost point) of the trajectory in relation to the inflection points can be calculated from
b = 2r sin 1/2αo .
(21.5)
In order for the Gulf Stream to separate from the western boundary as a free jet following a CAV trajectory, the CAV trajectory must be initialized with a trajectory inflection (γo → ∞) located at the separation point. Since the angle of separation (↜αo) is north of due east, the CAV trajectory must subsequently develop curvature that is concave toward the south. If the simulation exhibits curvature to the north after separation, then it does not separate from the western boundary as a free jet, even through it may have one or more segments downstream that follow a CAV trajectory. The calculated CAV trajectories are overlaid as red curves on Fig.€21.9. Details of the CAV trajectory calculations can be found in Table€2 of Hurlburt and Hogan (2008). The speed at the core of the current (υc) near separation from the western boundary is 1.6–1.7€m/s in the 1/16º simulations and 1.9–2.0€m/s in the 1/32º
21â•… Dynamical Evaluation of Ocean Models Using the Gulf Stream as an Example DWBC
41 N
0
2
4 6
8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
(in cm)
1/16º 40 N Cb =.002
561
NO DWBC
1/16° Cb=.002 Cb
39 N 38 N 37 N 36 N 35 N 40 N
a
b
1/32° Cb=.002
1/32° Cb=.002
c
dd
39 N 38 N 37 N 36 N 35 N 40 N
1/32° Cb=.02
1/32° Cb=.02
39 N 38 N 37 N 36 N 35 N
e
76 W
74 W
72 W
70 W
68 W
66 W
f
76 W
74 W
72 W
70 W
68 W
66 W
Fig. 21.9↜渀 CAV trajectory analysis for Gulf Stream pathways simulated by the six simulations illustrated in Fig.€21.2. The pathway of the maximum velocity at the core of the current (↜black line), the closest SSH contour (↜yellow-green line), the corresponding CAV trajectory (↜red line with a dot at the inflection point), the observed IR northwall frontal pathway ± std. dev. (↜violet lines), and the simulated SSH variability are overlaid on each panel. Due to the hierarchy of the overlaid lines (↜light violet, red, black, yellow-green from top to bottom), lines on the bottom tend to be obscured where close agreement occurs. That is particularly the case for the yellow-green SSH contour west of ~68°W, where the core of the current overlaying a single SSH contour is a prerequisite for the existence of a CAV trajectory. The SSH contour closest to the pathway of the velocity maximum is skewed toward the north side of the model Gulf Stream as depicted in SSH and is a −24€cm. b −16€cm. c −28€cm, and d–f −24€cm. See the corresponding panels in Fig.€21.2. Near the western boundary the Gulf Stream axis from Topex/Poseidon altimetry (Lee 1997) diverges from the IR frontal pathway in accord with the model simulations of panels a, c, d, and e (see Hurlburt and Hogan 2000, their Fig.€7). (From Hurlburt and Hogan 2008)
562
H. E. Hurlburt et al.
simulations, in line with observations of 1.6–2.1€m/s reported in Halkin and Rossby (1985), Joyce et€al. (1986), Johns et€al. (1995), Schmitz (1996), and Rossby et€al. (2005). A model mean υc over 75–70ºW was used in the CAV trajectory calculations. The angle of separation is 53±3º north of due east for the simulations with a realistic pathway and the inflection points used to initialize the CAV trajectory calculations are marked by red dots on the trajectories. Between the western boundary and ~70ºW, the four simulations with a realistic Gulf Stream pathway demonstrate close agreement between the model pathway, as represented by υc, and the corresponding CAV trajectory. However, the two simulations with pathways that overshoot the latitude of the observed Gulf Stream pathway exhibit curvature to the north immediately after separation and an inflection point (red dot) located northeast of separation from the western boundary. That means they do not separate from the western boundary as a free jet, but instead indicate a strong influence from the constraints of linear dynamics (Fig.€21.1). Thus, CAV trajectory dynamics alone are not sufficient to explain the Gulf Stream pathway between the western boundary and 69ºW. However, they do explain the small impact of the abyssal current crossing under the Gulf Stream near 72ºW (Fig.€21.4a), because the abyssal current and the CAV trajectory give nearly the same Gulf Stream latitude at that location (Fig.€21.9c).
21.2.6 G ulf Stream Separation and Pathway Dynamics, Part II: Role of CAV Trajectories In the simulations with a realistic Gulf Stream, the mean pathway closely follows a CAV trajectory between its separation from the western boundary and ~70ºW. The CAV trajectory depends on (1) the angle of boundary current separation (with respect to latitude), as largely determined by the angle of the shelf break prior to separation, (2) the speed at the core of the current, and (3) an inflection point located where boundary current separation occurs.
21.2.7 G ulf Stream Separation and Pathway Dynamics, Part III: The Cooperative Interaction of Abyssal Currents and CAV Trajectories Neither abyssal currents nor CAV trajectories alone are sufficient to explain Gulf Stream separation from the western boundary and its pathway to the east. Abyssal current constraint of the Gulf Stream latitude near 68.5ºW, in conjunction with the topographic configuration and a Gulf Stream feedback mechanism, is not a sufficient explanation of the Gulf Stream pathway between the western boundary and 68ºW. Gulf Stream simulations with realistic speeds at the core of the current are not sufficiently inertial (a) to overcome the linear solution demand for an overshoot
21â•… Dynamical Evaluation of Ocean Models Using the Gulf Stream as an Example
563
pathway and (b) to obtain realistic separation without assistance from the abyssal current near 68.5ºW. Thus a CAV trajectory and the constraint on the latitude of the Gulf Stream near 68.5ºW work together in simulation of a realistic Gulf Stream pathway between the western boundary and 68ºW. The eddy-driven abyssal circulation is sufficient to obtain the key abyssal current, which was not simulated without it. The DWBC is not necessary, but did augment the key abyssal current and did assist the eddy-driven abyssal circulation in effecting realistic Gulf Stream separation, when the latter was not strong enough by itself. The impact of the DWBC on Gulf Stream separation was resolution dependent, required at 1/16º, but not at 1/32º resolution. Finally, the dynamical explanation is robust. As long as the speed at the core of the current was consistent with observations and the key abyssal current was sufficiently strong, the simulated Gulf Stream separation and its pathway to the east were in close agreement with observations despite differences in model resolution, bottom friction, strength of the abyssal circulation, and the presence or absence of a DWBC. Further, the explanation is consistent with a wide range of key observational evidence in the upper and abyssal ocean, including a 15-year mean Gulf Stream IR northwall pathway, the speed at the core of the current near Gulf Stream separation, the pattern of sea surface height variability from satellite altimetry, and mean abyssal currents. Hurlburt and Hogan (2000) present a large number of additional model-data comparisons for the simulations depicted in Fig.€21.2a, c.
21.3â•…Dynamical Evaluation of Gulf Stream Simulations by Eddy-Resolving Global and Basin-Scale OGCMs Significant success has been achieved in simulating the Gulf Stream pathway in eddy-resolving basin-scale OGCMs with thermodynamics and higher vertical resolution (20–50 layers or levels) than the 5 layers used in the hydrodynamic model discussed in Sect.€21.2. However, the OGCM simulations have been very sensitive to changes, such as subgrid scale parameterizations and parameter values. Thus, it has been difficult to obtain consistent results and many simulations have exhibited serious flaws (Paiva et€al. 1999; Smith et€al. 2000; Bryan et€al. 2007; Chassignet and Marshall 2008; Hecht and Smith 2008; Hecht et€al. 2008). In this section we perform a dynamical evaluation of eddy-resolving global and basin-scale OGCM simulations of Gulf Stream separation and its pathway to the east. The immediate goals are to better identify and understand the sources of success and failure, and in Sect.€21.4 the impacts of data assimilation. So far, eddy-resolving global and basinscale ocean prediction systems have demonstrated only 10–15€day forecast skill in the Gulf Stream region based on anomaly correlation >0.6 versus 30 days or more in some regions (Smedstad et€al. 2003; Shriver et€al. 2007; Hurlburt et€al. 2008a; Chassignet et€al. 2009; Hurlburt et€al. 2009). Future goals are improved and more consistently realistic simulations of the Gulf Stream, increased ability to nowcast and forecast it on time scales up to a month, improved climate prediction in the Gulf
564
H. E. Hurlburt et al.
Stream region, and increased efforts to understand OGCM dynamics and dynamically evaluate their simulations in other regions. A set of eddy-resolving global and basin-scale simulations from HYCOM, MICOM, NEMO, and POP is used in the evaluation (see Table€21.1). The resolution and model domain range from 1/10° Atlantic to 1/25° global. In addition to simulations with a realistic Gulf Stream pathway and dynamics consistent with observations, simulations with several types of flaws are evaluated, including (a) a realistic pathway with unrealistic dynamics, (b) overshoot pathways, (c) premature separation south of Cape Hatteras (the observed location), (d) pathways that separate at Cape Hatteras but have a pathway segment that is too far south east of the separation point, (e) pathways that bifurcate at or after separation, and (f) pathways impacted by unrealistic behavior upstream of the separation point, such as excessive variability or persistent large seaward loops east of the observed mean pathway. All four of the models used here have simulated a variety of Gulf Stream pathways, as illustrated here and in the references cited above. To streamline the evaluation for the purpose of this discussion, we focus on the following: (1) To evaluate the mean path, mean SSH from the model is overlaid by the 15-year mean Gulf Stream IR northwall pathwayâ•›±1 (standard deviation) by Cornillon and Sirkes (unpublished). This frontal pathway has 0.1º longitudinal resolution and lies along the northern edge of the Gulf Stream. (2) SSH variability is used to look for a narrow band of high variability west of 69ºW and, combined with abyssal eddy kinetic energy (EKE), it is used to identify regions of baroclinic instability. Thus these fields help identify the dynamics of Gulf Stream pathway segments and source regions for eddy-driven mean abyssal currents. (3) Mean speed at the core of the current is used to assess whether or not the simulated Gulf Stream inertial jet is consistent with observations near the western boundary. (4) The DWBC (a term used to identify mean abyssal currents that are clearly part of the AMOC) and eddy-driven mean abyssal currents are used to assess their impact in steering the Gulf Stream pathway and related upper ocean features. Depending on their strength and location in relation to the isobaths, abyssal currents have the potential to improve or increase the errors in the simulated pathways. (5) Both the strength and depth structure of the AMOC can affect the Gulf Stream pathway. Increasing the strength can make the simulated Gulf Stream more inertial, but can also increase the tendency for an overshoot pathway based on linear dynamics. The depth structure of the AMOC influences the depths of the isobaths followed by the DWBC and interactions between the DWBC and the eddy-driven abyssal circulation. (6) The basin-wide linear solution response to the mean wind stress forcing yields the constraints of linear dynamics on the strength and pathways of wind-driven currents in the Gulf Stream region. CAV trajectories were not calculated because there is sufficient proxy information to assess this from the mean pathway, the mean core speed near separation, and the characteristic narrow band of SSH variability along the Gulf Stream west of ~69ºW. In the dynamical evaluation we focus on a segment of the Gulf Stream that extends from 30ºN, 80ºW, upstream of the observed separation latitude near 35.5ºN, 74.5ºW, to about 68ºW, and characterizations of accuracy refer to pathway segments
14.1 18.0
Global HYCOM Global HYCOM
32 coordinate surfaces 32 coordinate surfaces
Twin of 1/12° 14.1 but no tides d Twin of 1/12° 9.7 with tides Near twin of 1/25° 4.0
Near twin of 1/12° global 18.0
c
Comments
Global HYCOM 4.0 1/25° 32 coordinate surfaces Near twin of 1/12° 18.0 Section€21.4 Simulations and hindcasts Global HYCOM 5.8 1/12° 32 coordinate surfaces 2004–2006 No data assimilation e 60.5 1/12° 32 coordinate surfaces 2004–2006 Global HYCOM Cooper-Haines Global HYCOM 19.0 1/12° 32 coordinate surfaces 6/2007–5/2008 No data assimilation e Global HYCOM 74.2 1/12° 32 coordinate surfaces 6/2007–5/2008 MODAS synthetics a MICOM Miami Isopycnic Coordinate Ocean Model, isopycnal coordinates on a C-grid, NEMO Nucleus for European Modelling of the Ocean, z-levels with terrain-following coordinates in shallow water on a C-grid, POP Parallel Ocean Program, z-levels on a B-grid; HYCOM HYbrid Coordinate Ocean Model, hybrid isopycnal/pressure levels/terrain-following in shallow water on a C-grid; b Resolution for each prognostic variable; c Twin of 1/12° global HYCOM18.0 except for the model domain and relaxation to temperature (T) and salinity (S) climatology in buffer zones within 3° of the model boundaries at 28°S and 80°N, global HYCOM experiments are from the GLBa series and all HYCOM experiments use topogrophy based on DBDB2 by D.S.K. (see http://www.7320. nrlssc.navy.mil/DBDB2_WWW); d Includes external and internal tides from 8 tidal constituents (Arbic et€al. 2010); e Downward projection method for the SSH updates, i.e. Cooper and Haines (1996) or synthetic T&S profiles using the Modular Ocean Data Assimilation System (MODAS) (Fox et€al. 2002). In both cases the Navy Coupled Ocean Data Assimilation (NCODA) system (Cummings 2005) was then used to assimilate all the data
1/12° 1/12°
2004–2007 4–6 and 9–10 3, 5–10
12–15 2004–2007
32 coordinate surfaces 32 coordinate surfaces
9.4 9.7
Global HYCOM Global HYCOM
1/12° 1/12°
1982–1983 2004–2006 2004–2006 1998–2000 3–6 and 11–13
Years used
Table 21.1↜渀 Description of OGCM simulations and hindcasts used for dynamical analysis Experiment Horizontal Vertical Resolution Ocean Modela numbera Resolutionb Section€21.3 Simulations Atlantic MICOM 1.0 1/12° 20 coordinate surfaces T46 1/12° 50 levels Atlantic NEMO Global NEMO T103 1/12° 50 levels Atlantic POP 14x 1/10° 40 levels Atlantic HYCOM 1.8 1/12° 32 coordinate surfaces
21â•… Dynamical Evaluation of Ocean Models Using the Gulf Stream as an Example 565
566
H. E. Hurlburt et al.
and other features within this region, even though a larger region may be depicted in some figures. In Sect.€21.3.1 we present the mass transport streamfunction from linear simulations forced by wind stress products used in forcing the OGCM simulations discussed later. In Sect.€21.3.2 we discuss four simulations with a realistic Gulf Stream pathway and quite realistic Gulf Stream dynamics. In the remaining subsections we discuss simulations with different types of flaws, outlined earlier, including a simulation with a realistic Gulf Stream pathway but unrealistic separation dynamics. In each case one to four examples are used to help illustrate the range of simulated results and dynamics. None of the simulations in Sect.€21.3 include ocean data assimilation. Additionally, simulations in Sects.€21.2 and 21.3.1 are characterized by mid-latitude resolution (in º), whereas OGCMs in Sects.€21.3 and 21.4 are characterized by equatorial resolution, making 1/16º resolution in Sects.€21.2 and 21.3.1 approximately the same as 1/12º resolution for OGCMs in Sects.€21.3 and 21.4, ~7€km at mid-latitudes.
21.3.1 L inear model Gulf Stream Simulations from Wind Stress Products Used for OGCMs in Sects.€21.3 and 21.4 Linear barotropic solutions were obtained for the wind stress forcing products used by OGCM simulations discussed in Sects.€21.3 and 21.4. The solutions were obtained with the same model used in Sect.€21.2.1, but here excluding a contribution from the AMOC. Also, the model was run in barotropic, flat bottom mode rather than reduced gravity mode, which yields the same mean transport streamfunction. Figure€21.10 depicts the Atlantic mass transport streamfunction from 1/16º linear barotropic simulations forced by several different wind products, but only covering the latitude range of interest here. The wind stress products used to obtain the results in Fig.€21.10 are (a) an interim European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis mean over the years 2004–2006, (b) mean operational ECMWF over 2004–2006, (c) a 1978–2002 climatology derived using an ECMWF 40-year reanalysis (ERA-40) (Kallberg et€ al. 2004) and (d) a 2003–2008 climatology derived from the Navy Operational Global Atmospheric Prediction System (NOGAPS) (Rosmond et€al. 2002). In both (c) and (d) the wind stress was calculated from 10€m winds using a bulk formula from Kara et€al. (2005) and with the 10€m wind speeds corrected using a monthly QuikSCAT scatterometer climatology (Kara et€al. 2009). For (e) an ERA-15 (Gibson et€al. 1999) climatology was used and in (f) the ECMWF TOGA global surface analysis 1985-early 2001, based on operational Fig. 21.10↜渀 Mean transport streamfunction (ψ) from 1/16º linear barotropic flat bottom simulations forced by monthly mean wind stress from a An interim ECMWF reanalysis over 2004–2006. b Operational ECMWF over 2004–2006. c A 1978–2002 climatology from ECMWF ERA-40 with wind speed corrected by a QuikSCAT scatterometer climatology. d NOGAPS over 2003– 2008 also with the QuikSCAT correction. e An ECMWF ERA-15 climatology, and f An ECMWF TOGA global surface analysis over 1998–2000. The contour interval is 1 Sv
21â•… Dynamical Evaluation of Ocean Models Using the Gulf Stream as an Example 45 N 40 N
8
20
12
16
a
4 12
8
0
40 N 16
35 N
0
4
8
12
12
20 24
b
4
0
8
4
16
4
4
45 N
4
4
8
12
8
0
16
35 N
45 N
12
0
567
12
6
16
1
0
8 12
8
4
40 N
45 N
28
35 N
20
2
16 4
0 2
c
24
8
12
16
4
0 12
8
12
0
12
4
8
4 12 4
16
24
0 8
e
8
12
4
8
16 20
35 N
45 N
12
0
40 N
0
24
d
8
45 N
20
8
35 N
16
428
220
16
8
4
40 N
4
12
4
40 N
4
8
12
8
0
12
4
20
30 N f 80 W
16
35 N
16
70 W
60 W
12
50 W
8
40 W
30 W
20 W
10 W
0
568
H. E. Hurlburt et al.
ECMWF products, was used (Smith et€al. 2000; Bryan et€al. 2007). In the latter 10€m winds were converted to surface stresses using the neutral drag coefficient of Large and Pond (1981). Note that 5 out of the 6 wind stress products are linked to ECMWF and one to NOGAPS. Although a temporal mean of the interannual wind products was used to force the linear simulations, otherwise the wind products listed above were used in forcing the simulations listed in Table€21.1: (a) 1/12° Atlantic NEMO, (b) 1/12° global NEMO, (c) all of the HYCOM simulations except as noted, (d) 1/12° global HYCOM 19.0 and 74.2, (e) 1/12° Atlantic MICOM and 1/12° global HYCOM 5.8 and 60.5, and (f) 1/10° Atlantic POP. The resulting streamfunctions are all generally similar in the Gulf Stream region and quite different from that simulated using the smoothed Hellerman and Rosenstein (1983) wind stress climatology (Fig.€21.1). They separate from the western boundary with transports ranging from 20 to 27€Sv versus 30€Sv from smoothed Hellerman-Rosenstein. In all, a large majority of the streamfunction contours separate from the western boundary north of the observed Gulf Stream separation latitude (35.5ºN) and at least 50% separate between 35º and 40ºN and trend eastnortheastward after separation. The two wind stress products with the QuikSCATcorrected wind speeds give the strongest transports in Fig.€21.10 ((c) 26€Sv from ERA-40/QuikSCAT and (d) 27€ Sv from NOGAPS/QuikSCAT). It is significant that almost all of the streamfunction contours driven by these two products leave the western boundary north of the observed Gulf Stream separation latitude. In the case of smoothed Hellerman-Rosenstein, 17€Sv leave the western boundary north of 35.5ºN, suggesting an even stronger tendency for the wind stress products used in Fig.€21.10c, d to drive an overshoot pathway in the OGCM simulations.
21.3.2 Simulations with a Realistic Gulf Stream Pathway Figure€21.11 presents mean SSH and Fig.€21.12 SSH variability from four simulations with a realistic Gulf Stream pathway in the segment of interest between separation from the western boundary and 68ºW. These are 1/12º Atlantic MICOM (Figs.€21.11a, 21.12a), 1/12º global NEMO (Figs.€21.11b, 21.12b), 1/12º Atlantic HYCOM (Figs.€21.11c, 21.12c), and 1/25º global HYCOM (Figs.€21.11d, 21.12d). In the simulations, the mean IR northwall frontal pathway (in red) closely follows the northern edge of the simulated Gulf Stream over the segment of interest and the simulated pathway is generally realistic within the plot domain. In 1/12º Atlantic
Fig. 21.11↜渀 Mean SSH from simulations with a realistic Gulf Stream pathway: a 1/12º Atlantic MICOM over 1982–1983. b 1/12º global NEMO over 2004–2006. c 1/12º Atlantic HYCOM-1.8, years 3–6. d 1/25º global HYCOM-4.0, years 5–8 (see Table€21.1). The contour interval is 5€cm, a contour interval for SSH used throughout Sects.€21.3 and 21.4. The mean Gulf Stream IR northwall frontal pathway€±1 by Cornillon and Sirkes is overlaid in red on each panel and in red or black on many other panels. Sep vâ•›=â•›mean speed at the Gulf Stream core near separation from the western boundary. Sep v is also given on other mean SSH and near surface current figure panels
21â•… Dynamical Evaluation of Ocean Models Using the Gulf Stream as an Example
1
569
6HSY PV
1
1
1 1
$WO0,&20\UV
D 6HSY PV
1
1
1 1
E
*OREDO1(027\UV
6HSY PV
1
1
1 1
F
$WO+0.6 is only 10 days. The abyssal currents are generated via vortex stretching and compression when the assimilated SSH updates from altimeter data plus the mean SSH are projected downward. The data assimilation approximates the observed variations in the ocean features, such as current pathways and eddies, and in response the model dynamics interpolate and extrapolate the updates. The results indicate that in the process a more realistic AMOC generates more realistic abyssal currents along the continental slope and the representation of real variations in the Gulf Stream and related eddies in the upper ocean produces a model response that simulates flow instabilities well enough to generate realistic eddy-driven mean abyssal currents and maintain them in 14-day forecasts. Acknowledgements╇ The project US Global Ocean Data Assimilation Experiment (GODAE): Global Ocean Prediction with the HYbrid Coordinate Ocean Model (HYCOM), funded under the National Ocean Partnership Program (NOPP); the 6.1 project Global Remote Littoral Forcing via Deep Water Pathways, funded by the Office of Naval Research (ONR) under program element 601153N; and grants of computer time from the US Defense Department High Performance Computing Modernization Program. Alan Wallcraft is in charge of developing and maintaining
606
H. E. Hurlburt et al.
the standard version of HYCOM and Ole Martin Smedstad the data assimilative experiments. The European high-resolution global ocean model was developed in France by Mercator Océan with the financial support of the European MERSEA integrated project for the development, validation, and exploitation of the system and from the Région Midi Pyrénées, which financed a dedicated computer for this project. The mean Gulf Stream northwall pathway, based on satellite infrared imagery, is an unpublished analysis performed by Peter Cornillon (University of Rhode Island) and Ziv Sirkes (deceased) for the ONR project Data Assimilation and Model Evaluation Experiment–North Atlantic Basin (DAMEE–NAB).
References Arbic BK, Wallcraft AJ, Metzger EJ (2010) Concurrent simulation of the eddying general circulation and tides in a global ocean model. Ocean Model 32:175–187 Bane JM Jr, Dewar WK (1988) Gulf Stream bimodality and variability downstream of the Charleston bump. J Geophys Res 93(C6):6695–6710 Barnier B, Madec G, Penduff T, Molines JM, Treguier AM, Le Sommer J, Beckmann A, Biastoch A, Böning C, Dengg J, Derval C, Durand E, Gulev S, Remy E, Talandier C, Theeten S, Maltrud M, McClean J, De Cuevas B (2006) Impact of partial steps and momentum advection schemes in a global ocean circulation model at eddy-permitting resolution. Ocean Dyn 56:543–567. doi:10.1007/s10236-006-0082-1 Barron CN, Smedstad LF, Dastugue JM, Smedstad OM (2007) Evaluation of ocean models using observed and simulated drifter trajectories: impact of sea surface height on synthetic profiles for data assimilation. J Geophys Res 112:C07019. doi:10.1029/2006JC002982 Bleck R (2002) An oceanic general circulation model framed in hybrid isopycnic-cartesian coordinates. Ocean Model 37:55–88 Bleck R, Smith L (1990) A wind-driven isopycnic coordinate model of the north and equatorial Atlantic Ocean. 1. Model development and supporting experiments. J Geophys Res 95:3273– 3285 Bower AS, Hunt HD (2000) Lagrangian observations of the deep western boundary current in the North Atlantic Ocean. Part II. The Gulf stream—deep western boundary current crossover. J Phys Oceanogr 30:784–804 Bryan FO, Hecht MW, Smith RD (2007) Resolution convergence and sensitivity studies with North Atlantic circulation models. Part I: the western boundary current system. Ocean Model 16:141–159 Chassignet EP, Marshall DP (2008) Gulf Stream separation in numerical ocean models. In: Hecht M, Hasumi H (eds) Ocean modeling in an eddying regime, geophysical monograph 177. American Geophysical Union, Washington Chassignet EP, Hurlburt HE, Metzger EJ, Smedstad OM, Cummings JA, Halliwell GR, Bleck R, Baraille R, Wallcraft AJ, Lozano C, Tolman HL, Srinivasan A, Hankin S, Cornillon P, Weisberg R, Barth A, He R, Werner F, Wilkin J (2009) US GODAE: global ocean prediction with the HYbrid Coordinate Ocean Model (HYCOM). Oceanography 22:64–75 Cooper M, Haines KA (1996) Altimetric assimilation with water property conservation. J Geophys Res 24:1059–1077 Cummings JA (2005) Operational multivariate ocean data assimilation. Quart J R Meteor Soc 131:3583–3604 Fox DN, Teague WJ, Barron CN, Carnes MR, Lee CM (2002) The modular ocean data analysis system (MODAS). J Atmos Ocean Technol 19:240–252 Gibson JK, Kallberg P, Uppala S, Hernandez A, Nomura A, Serrano E (1999) ERA ECMWF reanalysis project report series 1. ERA-15 description, version 2. European Centre for MediumRange Weather Forecasts, Reading
21â•… Dynamical Evaluation of Ocean Models Using the Gulf Stream as an Example
607
Glenn SM, Ebbesmeyer CC (1994a) The structure and propagation of a Gulf Stream frontal eddy along the North Carolina shelf break. J Geophys Res 99(C3):5029–5046 Glenn SM, Ebbesmeyer CC (1994b) Observations of Gulf Stream frontal eddies in the vicinity of Cape Hatteras. J Geophys Res 99(C3):5047–5055 Godfrey JS (1989) A Sverdrup model of the depth-integrated flow for the world ocean allowing for island circulations. Geophys Astrophys Fluid Dyn 45:89–112 Gordon AL, Giulivi CF, Lee CM, Furey HH, Bower A, Talley L (2002) Japan/East Sea thermocline eddies. J Phys Oceanogr 32:1960–1974 Halkin D, Rossby HT (1985) The structure and transport of the Gulf Stream at 73°W. J Phys Oceanogr 15:1439–1452 Haltiner GJ, Martin FL (1957) Dynamical and Physical Meteorology. McGraw-Hill, New York Hecht MW, Smith RD (2008) Towards a physical understanding of the North Atlantic: a review of model studies in an eddying regime. In: Hecht M, Hasumi H (eds) Ocean modeling in an eddying regime, geophysical monograph 177. American Geophysical Union, Washington Hecht MW, Petersen MR, Wingate BA, Hunke E, Maltrud ME (2008) Lateral mixing in the eddying regime and a new broad-ranging formulation. In: Hecht M, Hasumi H (eds) Ocean modeling in an eddying regime, geophysical monograph 177. American Geophysical Union, Washington Hellerman S, Rosenstein M (1983) Normal monthly wind stress over the world ocean with error estimates. J Phys Oceanogr 13:1093–1104 Hogan PJ, Hurlburt HE (2006) Why do intrathermocline eddies form in the Japan/East Sea? A modeling perspective. Oceanography 19:134–143 Hogg NG, Stommel H (1985) On the relation between the deep circulation and the Gulf Stream. Deep-Sea Res 32:1181–1193 Hurlburt HE (1986) Dynamic transfer of simulated altimeter data into subsurface information by a numerical ocean model. J Geophys Res 91(C2):2372–2400 Hurlburt HE, Hogan PJ (2000) Impact of 1/8° to 1/64° resolution on Gulf Stream model-data comparisons in basin-scale subtropical Atlantic Ocean models. Dyn Atmos Ocean 32:283–329 Hurlburt HE, Hogan PJ (2008) The Gulf Stream pathway and the impacts of the eddy-driven abyssal circulation and the Deep Western Boundary Current. Dyn Atmos Ocean 45:71–101 Hurlburt HE, Thompson JD (1980) A numerical study of Loop Current intrusions and eddy shedding. J Phys Oceanogr 10:1611–1651 Hurlburt HE, Thompson JD (1982) The dynamics of the Loop Current and shed eddies in a numerical model of the Gulf of Mexico. In: Nihoul JCJ (ed) Hydrodynamics of semi-enclosed seas. Elsevier, Amsterdam Hurlburt HE, Wallcraft AJ, Schmitz WJ Jr, Hogan PJ, Metzger EJ (1996) Dynamics of the Kuroshio/Oyashio current system using eddy-resolving models of the North Pacific Ocean. J Geophys Res 101(C1):941–976 Hurlburt HE, Chassignet EP, Cummings JA, Kara AB, Metzger EJ, Shriver JF, Smedstad OM, Wallcraft AJ, Barron CN (2008a) Eddy-resolving global ocean prediction. In: Hecht M, Hasumi H (eds) Ocean modeling in an eddying regime, geophysical monograph 177. American Geophysical Union, Washington Hurlburt HE, Metzger EJ, Hogan PJ, Tilburg CE, Shriver JF (2008b) Steering of upper ocean currents and fronts by the topographically constrained abyssal circulation. Dyn Atmos Ocean 45:102–134. doi:10.1016/j.dynatmoce.2008.06.003 Hurlburt HE, Brassington GB, Drillet Y, Kamachi M, Benkiran M, Bourdallé-Badie R, Chassignet EP, Jacobs GA, Le Galloudec O, Lellouche JM, Metzger EJ, Oke PR, Pugh TF, Schiller A, Smedstad OM, Tranchant B, Tsujino H, Usui N, Wallcraft AJ (2009) High-resolution global and basin-scale ocean analyses and forecasts. Oceanography 22:110–127 Johns WE, Shay TJ, Bane JM, Watts DR (1995) Gulf Stream structure, transport, and recirculation near 68°W. J Geophys Res 100:817–838 Joyce TM, Wunsch C, Pierce SD (1986) Synoptic Gulf Stream velocity profiles through simultaneous inversion of hydrographic and acoustic Doppler data. J Geophys Res 91:7573–7585 Kallberg P, Simmons A, Uppala S, Fuentes M (2004) ERA-40 project report series: 17. The ERA40 archive. ECMWF. Reading
608
H. E. Hurlburt et al.
Kara AB, Hurlburt HE, Wallcraft AJ (2005) Stability-dependent exchange coefficients for air-sea fluxes. J Atmos Ocean Technol 22:1080–1094 Kara AB, Wallcraft AJ, Martin PJ, Pauley RL (2009) Optimizing surface winds using QuikSCAT measurements in the Mediterranean Sea during 2000–2006. J Mar Sys 78:119–131 Large WG and Pond S (1981) Open ocean momentum flux measurements in moderate to strong winds. J Phys Oceanogr 11(3):324–336 Lee H (1997) A Gulf Stream synthetic geoid for the TOPEX altimeter. MS Thesis Rutgers University, New Brunswick Lee T, Cornillon P (1996) Propagation of Gulf Stream meanders between 74° and 70°W. J Phys Oceanogr 26:205–224 Legeckis R, Brown CW, Chang PS (2002) Geostationary satellites reveal motions of ocean surface fronts. J Mar Sys 37:3–15 Legeckis RV (1979) Satellite observations of the influence of bottom topography on the seaward deflection of the Gulf Stream off Charleston, South Carolina. J Phys Oceanogr 9:483–497 Madec G (2008) NEMO ocean engine. Report 27 ISSN No 1288–1619. Institute Pierre-Simon Laplace (IPSL), France Munk WH (1950) On the wind-driven ocean circulation. J Met 7:79–93 Paiva AM, Hargrove JT, Chassignet EP, Bleck R (1999) Turbulent behavior of a fine mesh (1/12°) numerical simulation of the North Atlantic. J Mar Sys 21:307–320 Pickart RS (1994) Interaction of the Gulf Stream and Deep Western Boundary Current where they cross. J Geophys Res 99:25155–25164 Pickart RS, Watts DR (1990) Deep Western Boundary Current variability at Cape Hatteras. J Mar Res 48:765–791 Reid RO (1972) A simple dynamic model of the Loop Current. In: Capurro LRA, Reid JL (eds) Contributions on the Physical Oceanography of the Gulf of Mexico. Gulf Publishing Co, Houston Rosmond TE, Teixeira J, Peng M, Hogan TF, Pauley R (2002) Navy operational global atmospheric prediction system (NOGAPS): forcing for ocean models. Oceanography 15(1):99–108 Rossby CG (1940) Planetary flow patterns in the atmosphere. Quart J R Meteor Soc 66:68–87 Rossby T, Flagg CN, Donohue K (2005) Interannual variations in upper-ocean transport by the Gulf Stream and adjacent waters between New Jersey and Bermuda. J Mar Res 63:203–226 Schmitz WJ Jr (1996) On the world ocean circulation: Volume I. Some global features/North Atlantic circulation. Technical Report WHOI-96–03 Woods Hole Oceanographic Institution, Woods Hole Schmitz WJ Jr, McCartney MS (1993) On the North Atlantic circulation. Rev Geophys 31:29–49 Shriver JF, Hurlburt HE, Smedstad OM, Wallcraft AJ, Rhodes RC (2007) 1/32° real-time global ocean prediction and value-added over 1/16° resolution. J Mar Sys 65:3–26 Smedstad OM, Hurlburt HE, Metzger EJ, Rhodes RC, Shriver JF, Wallcraft AJ, Kara AB (2003) An operational eddy resolving 1/16° global ocean nowcast/forecast system. J Mar Sys 40– 41:341–361 Smith RD, Maltrud ME, Bryan FO, Hecht MW (2000) Numerical simulation of the North Atlantic Ocean at 1/10°. J Phys Oceanogr 30:1532–1561 Sverdrup HU (1947) Wind-driven currents in a baroclinic ocean—with application to the equatorial currents of the eastern Pacific. Proc Natl Acad Sci U S A 33:318–326 Thompson JD, Schmitz WJ Jr (1989) A regional primitive-equation model of the Gulf Stream: design and initial experiments. J Phys Oceanogr 19:791–814 Townsend TL, Hurlburt HE, Hogan PJ (2000) Modeled Sverdrup flow in the North Atlantic from 11 different wind stress climatologies. Dyn Atmos Ocean 32:373–417 Tsujino H, Usui N, Nakano H (2006) Dynamics of Kuroshio path variations in a high-resolution general circulation model. J Geophys Res. doi:10.1029/2005JC003118 Usui N, Tsujino H, Fujii Y (2006) Short-range prediction experiments of the Kuroshio path variabilities south of Japan. Ocean Dyn 56:607–623 Usui N, Tsujino H, Fujii Y, Kamachi M (2008a) Generation of a trigger meander for the 2004 Kuroshio large meander. J Geophys Res. doi:10.1029/2007JC004266
21â•… Dynamical Evaluation of Ocean Models Using the Gulf Stream as an Example
609
Usui N, Tsujino H, Nakano H, Fujii Y (2008b) Formation process of the Kuroshio large meander in 2004. J Geophys Res. doi:10.1029/2007JC004675 Watts DR, Tracey KL, Bane JM, Shay TJ (1995) Gulf Stream path and thermocline structure near 74°W and 68°W. J Geophys Res 100:18291–18312 Xie L, Liu X, Pietrafesa LJ (2007) Effect of bathymetric curvature on Gulf Stream instability in the vicinity of the Charleston Bump. J Phys Oceanogr 37(3):452–475
Chapter 22
Ocean Forecasting Systems: Product Evaluation and Skill Matthew Martin
Abstract╇ The evaluation of output from ocean forecasting systems is important in order to inform users how much confidence can be placed in the products, and helps identify areas for improvement in the systems. An overview of the statistical methods which can be used to perform the evaluation is provided. Examples of some commonly used methods from various GODAE systems are given, including evaluation of large-scale model performance, the use of output from data assimilation systems, the use of independent data, and comparison of forecasts with analyses.
22.1â•…Introduction The aim of ocean forecasting systems is to provide information about the past, present and future state of the ocean to a range of users. There are a wide range of applications including defence, ship routing, oil spill prediction, weather forecasting, climate monitoring and scientific research. In order for the outputs produced by ocean forecasting systems to be useful for these applications, the ability of the systems to represent the real world must be assessed. This will inform the users of where and when the products can be used, and with how much confidence. It also aids development of the forecasting systems themselves, highlighting areas where improvements can be made. The state variables from ocean forecasting systems are the sea surface height and the three-dimensional temperature, salinity and currents. For those systems which include sea-ice models, the sea-ice concentration, velocity and thickness are also produced. Other diagnostic quantities such as mixed-layer depth and transports are also of interest to users of model output. The applications which use this ocean information cover a wide range of time and space scales from large-scale climate monitoring and seasonal forecasting applications, which look at evolution over months with basin-wide and global coverage, through to the analysis and prediction M. Martin () Met Office, Fitzroy Road, Exeter, UK e-mail:
[email protected] A. Schiller, G. B. Brassington (eds.), Operational Oceanography in the 21st Century, DOI 10.1007/978-94-007-0332-2_22, © British Crown 2011, the Met Office, UK
611
612
M. Martin
of mixed-layer depth over a diurnal cycle. This range of spatial and temporal scales must therefore be taken into account when assessing the products. The amount of information available from ocean forecasting systems is immense—the model state vector usually contains at least of the order of 107 variables at any particular time (and for some ocean forecasting systems significantly more than this). It is usually impossible for users of the output to access all of this data, and so some post-processing is often performed to synthesise this information. This may involve interpolation or averaging in space and time, and may also involve production of other diagnostic information which is of more relevance to a particular user. The impact of the post-processing on the accuracy of the data which are provided must be assessed, usually by assessing the post-processed fields directly. In order to evaluate products from ocean forecasting systems and relate them to the real world, observations are required. These could take the form of climatologies, analyses of satellite data, or raw observed values. In all cases, the accuracy of the observations needs to be assessed and taken into account when performing the comparison between model and observations. It is also important to use observations which have been quality controlled—comparison with “bad” observations can lead to confusing results. A number of aspects affect the quality of the output from ocean forecasting systems. The most obvious is the quality of the model used to produce the forecast, including its horizontal and vertical resolution, and the parametrisations which are used. The surface forcing fields used to drive the model (or in the case of a coupled model, the quality of the atmospheric model) also have a significant impact on the quality of the ocean forecast. For regional models, lateral boundaries can play a significant role. The data assimilation scheme used to initialise the model has a large impact on the accuracy of the forecast—the type and number of observations being used in the assimilation, the assimilation scheme itself, and the quality control of observations all have an impact on the accuracy of the analysis and the subsequent forecast. A review of some statistical concepts which are required to assess model output is given in the next section. A summary of the main issues with the observations available for use in the evaluation of model products is then given. This is followed by some specific examples of product evaluation taken from various GODAE systems. An overall summary is then given.
22.2╅Statistical Concepts A number of statistical measures are required to thoroughly assess the output of ocean forecasting systems. Three different concepts are described here, aimed at determining the accuracy of the analysis and forecast, the ability of the model to represent the patterns of the observations, sometimes termed association (Murphy et€al. 1995), and the skill of the forecast. Representing this information in a concise
22â•… Ocean Forecasting Systems: Product Evaluation and Skill
613
way can be done through some well-known summary diagrams which are briefly described. We assume that there is a verification data set consisting of a set of N observa¯ The model values at the same time and tions, yi , i = 1, . . . , N , with mean value y. location as these observations are denoted xi , i = 1, . . . , N and have mean value x. ¯ The mean difference is a measure of the ability of the model to represent the mean observed state:
MD =
N 1 (xi − yi ). N i=1
(22.1)
22.2.1 Accuracy The accuracy of the forecast is usually assessed using the root mean square differences between the model and observed values: N 1 (22.2) RMSD = (xi − yi )2 . N i=1
22.2.2 Pattern The ability of the model to reproduce the pattern in the observations can be measured using the correlation coefficient:
R=
N 1 (xi − x)(y ¯ i − y) ¯ N i=1
σx σy
,
(22.3)
where σx and σy are the standard deviations of the model and observations respectively. The correlation coefficient provides information about whether the patterns in the model are similar to the patterns of the observations, but not about the amplitude of variation in the two fields. It reaches a value of 1 when the two fields have the same centred pattern of variation, a value of −1 when the two fields vary in the opposite sense to each other, and a value of zero when no correlation exists between the two fields. The square of the correlation coefficient, R 2 , is also a useful quantity as it provides information on the fraction of the variance explained. When the dominant source of variability in a field is a large scale signal, for instance the seasonal cycle, most ocean models would easily reproduce the signal, resulting in high values of R. However, ocean forecasting systems produce infor-
614
M. Martin
mation at smaller temporal and spatial scales. To assess these, it is instructive to calculate the anomaly correlation coefficient:
ACC =
N
i=1 N
i=1
(xi − Ci )(yi − Ci )
(xi − Ci )
2
N
i=1
,
(22.4)
2
(yi − Ci )
which provides information about the ability of the model forecast to reproduce the observational information when the seasonally varying climate signal, denoted C, has been removed.
22.2.3 Skill Determining the skill of a model forecast is dependent on the application and it is not possible to define one skill score that is universally appropriate. A number of scores have been suggested in the literature, some examples of which are given below. The skill of a forecast can be defined as the accuracy of the forecast relative to the accuracy of a reference field such as a climatology or persistence (Murphy 1995). A simple way of measuring this is given by:
SS1 = 1 −
MSD , MSDref
(22.5)
which measures the relative accuracy of the forecast to some reference, where MSD indicates Mean Square Difference (the square of Eq.€ 22.2) and the subscript ref indicates that the model value in Eq.€22.2 has been replaced by a climatology or persistence estimate. A value of 1 implies that the forecast has perfect skill while a value of zero implies no skill. In the above skill score, no account is taken of correlations or bias. Taylor (2001) suggests the following score which is based on the correlation coefficient and the model and observed variances:
SS2 =
4(1 + R)
(σˆ x + 1/σˆ x )2 (1 + Ro )
,
(22.6)
where Ro is the maximum correlation attainable (Ro = 1) and σˆ x = σx /σy is the normalised standard deviation. Another skill score which uses the correlation, variances, and also includes the biases in model and observations, as suggested by Metzger et€al. (2008), is given by:
SS3 = R 2 − [R − (σy /σx )]2 − [(y¯ − x)/σ ¯ x ]2 .
(22.7)
22â•… Ocean Forecasting Systems: Product Evaluation and Skill
615
This skill score is equivalent to that defined in Eq.€22.5 when the reference used is the mean of the observations (Murphy 1988). This way of decomposing the score can be useful because the various contributions can be assessed—the correlation, conditional bias and unconditional bias. For probabilistic forecasting systems, a wide range of skill scores are often used such as the Brier skill score (Brier 1950), or the Relative Operating Characteristics (ROC) score which is used to determine the relationship between the number of events which were correctly forecast to the number of false alarms. These skill scores are widely used in seasonal prediction systems and ensemble weather forecasting systems, but few short-range ocean forecasting systems currently produce ensemble forecasts. These skill scores are not covered in depth here—see Atger (1999) and references therein for further information.
22.2.4 Summary Diagrams In order to characterise the differences between the model and observations it is important to take into account the correspondence in both the patterns and the variances of the two fields. We define the centred pattern RMSD as: N 1 (22.8) CRMSD = [(xi − x) ¯ − (yi − y)] ¯ 2. N i=1 Taylor (2001) noticed that a simple relationship exists between the correlation coefficient, the centred pattern RMS difference, and the variances of the fields in question. The relationship is given by:
CRMSD2 = σx2 + σy2 − 2σx σy R,
(22.9)
which takes the same form as the law of cosines (c2 = a 2 + b2 − 2ab cos (γ )) . This relationship can be used to plot the information about R, CRMSD and the variances in the model and observations as a point on a single diagram. In order to make it possible to compare fields with different units, the statistics can be nondimensionalised by normalising each variable in Eq. (22.8) by the standard deviation in the observed field, which leaves the correlation coefficient unchanged. A schematic Taylor diagram is shown in Fig.€22.1. If the model exactly reproduced the observations, it would lie at the point indicated by the black circle. The distance between this black circle and the actual model point (the blue diamond in this example) represents the CRMSD and the dotted arcs on the diagram represent lines of constant CRMSD. The correlation coefficient is represented on the outer arc of the diagram with increasing correlation with the angle from the y-axis. The normalised standard deviation is represented as the distance to the origin, with a ratio of one denoted by the dashed arc (if the point is closer to the origin the model has lower
616
M. Martin
Fig. 22.1↜渀 Schematic description of a Taylor diagram
UUH
&R
LRQ ODW
6WDQGDUGGHYLDWLRQQRUPDOLVHGZUW2EV
9DULDQFHUDWLR PRGHOREV
&RUUHODWLRQ FRHIILFLHQW &506'
variance than the observations). The power of the Taylor diagram lies in the ability to plot numerous model runs on a single diagram and to compare these various aspects of the models’ performance. One drawback of the Taylor diagram is that the mean error of the models is not accounted for. The so-called Target diagram (Jolliff et€al. 2009) can be used to represent complementary information about the statistical performance of models. In this case, the relationship between the total mean square difference and the unbiased MSD and bias, RMSD2 = MD2 + CRMSD2, is plotted on a diagram where the x-axis represents CRMSD and the y-axis represents the bias. Since CRMSD is a positive quantity by definition, the negative x-axis can be utilised to include information about the standard deviation difference by multiplying the CRMSD by the sign of the standard deviation difference.
22.3â•…Observations Various observation types are available for use in validating and verifying ocean forecasting systems and are detailed elsewhere. Some general points about the use of these data in evaluating model output are outlined here. For satellite data, a number of levels of processing are performed to produce observations of the quantities which are output by ocean models. For instance, sea surface temperature (SST) data undergoes various levels of data processing from the level 1 brightness temperatures measured by the satellites, through the level 2 conversion to sea surface temperature at the native resolution, the level 3 re-gridding of the data, through to the level 4 objective analyses. Each level
22â•… Ocean Forecasting Systems: Product Evaluation and Skill
617
of processing affects the accuracy and representativeness of the data so it is important to be clear what the observations are representing before using them for evaluation. It is important to recognise that the observations used in model evaluation are not themselves perfect representations of the true state of the ocean. Measurement techniques will introduce some error into the observations. The observations are also usually made at a specific location whereas the model represents an area-average value. This means that the model cannot represent all processes affecting the observations. These errors of representativity, as well as increasing the apparent error in the observations, can lead to correlated errors which will affect the interpretation of the resulting statistics. All of these errors should therefore be taken into account when assessing the results of any model-observation comparison. As well as the errors described in the previous paragraph, observations often report erroneous values. This can happen for a number of reasons such as misreporting of location, corruption of the observation during transmission, or instrument error. One or two bad observations can significantly impact the results of any validation/verification, so it is important that a thorough check on the quality of the data is performed prior to the evaluation. This quality control can be performed in a number of ways, but usually consists of a comparison between the observations and some reference field either from a model forecast, or from observed climatology (see for example Ingleby and Huddleston 2007).
22.4╅Evaluating Ocean Analyses and Forecasts The usual process for developing a new ocean forecasting system, or significant upgrades to an existing system, involves a number of stages. Scientific developments will be tested individually to ensure that they are producing the expected change in the system. Once a number of developments are available, they will be put together into a new version of the system and this must then be thoroughly evaluated during the validation phase. This validation is usually done by means of the evaluation of a set of hindcasts of the system, where the system is run over a multi-annual period in the past. This tests that the overall changes to the system produce the expected improvements. Once the validation has been carried out, the system can be implemented operationally. At this stage it is important to continuously assess and monitor the accuracy of the system using a verification system. The results of both the validation and verification are useful for providing information to users of the system about the expected accuracy. User-specific evaluations can also be carried out to assess the suitability of the system for a given application. A number of examples of evaluating ocean forecasting systems are given below taken from various sources (e.g. Ferry et€al. 2007; Oke et€al. 2008; Metzger et€al. 2008, 2009; Storkey et€al. 2010), providing illustrations of some commonly used methods. The advantages and shortcomings of each method are outlined.
618
M. Martin
22.4.1 Evaluating the Large-Scale Mean and Variability It is important to check that the average properties of the ocean forecasting systems are providing a good representation of the ocean climate. This is usually done by comparing multi-annual averages to climatologies generated from observational data-sets. One example of this is a comparison between a mean dynamic topography (MDT, such as that of Rio et€al. 2005; or Maximenko and Niiler 2005), with the model’s average sea surface height field. This provides a useful guide as to the ability of the model to represent the large scale ocean circulation (see for example Metzger et€al. 2008). Temperature and salinity can also be assessed using a suitable climatological data-set. In Fig.€22.2, the annual mean temperature anomalies from the World Ocean
– 1000
depth (m)
depth (m)
– 1000 – 2000 – 3000
– 60
– 40
– 20
0
20
latitude (deg)
depth (m)
depth (m)
– 3000 – 4000
– 20
0
20
latitude (deg)
– 2000 – 3000 – 4000
– 50
0
50
– 50
d
latitude (deg)
0
latitude (deg)
50
depth (m)
– 1000
depth (m)
– 1000 – 2000 – 3000 – 4000
0
– 40
– 1000
– 2000
e
– 60
b
– 1000
c
– 3000 – 4000
– 4000
a
– 2000
– 2000 – 3000 – 4000
100
200
longitude (deg)
300
f
0
100
200
300
longitude (deg)
Fig. 22.2↜渀 Annual mean temperature anomalies from WOA05 climatology for 2008 from FOAM both without (a, c, e) and with (b, d, f) data assimilation. a, b Show a cross-section in the Indian Ocean along 90°E. c, d Show a cross-section in the Atlantic Ocean along 30°W. e, f Show a crosssection along the equator
22â•… Ocean Forecasting Systems: Product Evaluation and Skill
619
Atlas 2005 (Locarnini et€al. 2006) are shown as cross-sections from two hindcasts of the ¼° resolution global FOAM system, one without assimilation and one with. This shows that the data assimilation is able to reduce the drifts of the model away from climatology. One has to be careful when performing these comparisons that any inter-annual signal is not contaminating the results. For example in Fig.€22.2f, there is a clear La Nina signal, where the model is representing the true deviation from climatology. The variability in the model and observations can also be assessed. For instance, sea surface height (SSH) can be used to measure the amount of mesoscale activity. This can be estimated from observations provided by satellite altimeters, and also from ocean models. Figure€22.3 shows an example of this from the GLORYS reanalysis produced by Mercator using the ¼° resolution NEMO model with data assimilation. The standard deviations of the data are shown next to the standard deviation of the model fields from a 6 year period. Here, the data used to calculate the observed variability are being assimilated in the reanalysis product so this test is only useful to check that the assimilation of data is working correctly. The model analyses are reproducing the observed variability very well, including the western 80°N
Latitude
40°N
0°
40°S
80°S
0°
100°E
a
160°W
60°W
Longitude 80°N
0.3 0.28 0.26 0.24 0.22 0.2 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02
Latitude
40°N
Fig. 22.3↜渀 Standard deviation of SSH for the period 2002–2007 from a Aviso data and b from the GLORYS reanalysis product
0°
40°S
80°S
b
0
0°
100°E
160°W Longitude
60°W
620
M. Martin
boundary currents which are difficult areas to accurately represent the mesoscale variability with ¼° resolution. The only regions where the model variability is significantly different to the observed are in the Zapiola Rise region and in parts of the South Pacific. Both the average and variability comparisons described above are useful as a first-order check on the ability of the model to represent the large-scale ocean features, and can give confidence that the model is behaving as expected. However, they do not give information about the accuracy or skill of the model and so are of limited use to most users. More detailed investigations are required for this, and are described below.
22.4.2 Data Assimilation Statistics In the data assimilation process, the observation operator h is used to interpolate the model forecast field xf to the location in time and space of the observations, y. This enables calculation of the innovations, d = [y − h(xf )]. Once the data assimilation has been performed it is also possible to calculate the equivalent using the analysis field to produce the residuals, r = [y − h(xa )]. The reduction in the errors between the analysis and the forecast can be used as an a posteriori check that the data assimilation process is working as expected, and is fitting the observations to within their error (see for example Cummings 2005). The increments generated through the data assimilation process also provide an important source of information. The time-average of these increments can indicate areas of significant model bias. However, it is not always obvious how to diagnose the source of these biases. For validation and verification of the model forecast, it is the innovation statistics that are of most interest, as they provide a pseudo-independent check on its accuracy. The observations being used for this comparison have not previously been assimilated so from that point of view are independent. However, previous observations of the same type will have been assimilated on previous data assimilation cycles so they cannot be viewed as completely independent. An example of the innovation statistics from a 2-year reanalysis using the ¼° global FOAM system (Storkey et€al. 2010) is shown in Fig.€22.4. This includes the mean and RMS of the innovations for SSH and for temperature. The mean errors show that the system is able to represent the global average observed SSH and temperature well, although a small positive temperature bias exists below the top 50€m, with a cold bias above this depth for most of the period. The RMS of the innovations provides a measure of the overall accuracy of the system both as a function of time, and of depth (for temperature). The maximum of the global temperature errors is located within the top 200€m with much smaller errors below this depth. These timeseries plots also illustrate the stability of the system, with the SSH being relatively stable, whereas the temperature RMS errors appear to have a seasonal cycle with smaller errors in Northern hemisphere winter.
22â•… Ocean Forecasting Systems: Product Evaluation and Skill
621
0.15
Sea surface height error (m)
0.10
0.05
0.00
– 0.05
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan
a
2007
2008
2009
0
Depth (m)
200 400 600 800 1000
Apr 07
Jan 07
b
– 0.5
– 0.4
Oct 07
Jul 07 – 0.3
– 0.2
– 0.1
Jan 08 0
Apr 08 0.1
0.2
Jul 08 0.3
Oct 08 0.4
0.5
Mean error
0
Depth (m)
200 400 600 800 1000
Apr 07
Jan 07
c
0
0.2
Oct 07
Jul 07 0.4
0.6
0.8
Jan 08 1
Apr 08 1.2
1.4
Jul 08 1.6
Oct 08 1.8
2
RMS error
Fig. 22.4↜渀 a Mean (↜dotted line) and RMS (↜solid line) of the innovations for SSH. b Mean of the temperature innovations as a function of depth and time. c RMS of the temperature innovations as a function of depth and time for the FOAM reanalysis for the period January 2007–December 2008
622
M. Martin
An example of the use of Taylor diagrams for plotting innovation statistics is shown in Fig.€ 22.5 with results from a hindcast run of the ¼° resolution global FOAM system (Storkey et€al. 2010). This shows the statistics for a number of different regions for both SST and SSH. The SST statistics are only shown for a comparison with the AATSR data, although other satellite SST data were assimilated. The variability in both these variables is well-reproduced by the model in all regions, but the correlations and RMS differences are clearly regionally dependent with the Mediterranean region having the largest RMS errors and lowest correlation coefficient.
22.4.3 E valuation of Analyses and Forecasts Using Independent Data In operational assimilation systems, the aim is to provide the best possible estimate of the ocean state, and so all available data are assimilated. However, some data-sets are not available in real-time and so can be used in delayed mode to validate the results. An example of this is the RAPID array which measures sub-surface ocean properties in the North Atlantic in order to produce estimates of the Atlantic Meridional Overturning Circulation (AMOC). Qualitative inter-comparisons can be made between ocean model output of SSH and satellite ocean colour data (see for example Storkey et€al. 2010). These can help to show the performance of the systems in reproducing the position of mesoscale eddies and fronts, but it is difficult to produce robust quantitative statistics using this sort of technique. A method which is often used to validate ocean models in a hindcast setting is to withhold certain data from the data assimilation, and use this independent data for validating the results. This is a useful technique as it provides an independent check that the data assimilation system is working as expected. It is not possible to use this to assess the overall accuracy of the system as the unassimilated data would be assimilated in the operational system, but it can give a bound on the expected accuracy. An example of this technique is shown in Oke et€al. 2008 which shows results from the Bluelink Reanalysis system. Here some unassimilated Argo profiles are used to assess the RMSD in the assimilation run and the run without data assimilation. In all regions at almost all depths, the assimilation is improving the model’s representation of sub-surface temperature when compared to the non-assimilating model. Some data-sets provide information about variables which are not assimilated in most ocean forecasting systems at present. For instance, most of the current operational forecasting systems do not assimilate velocity data. Direct measurements of velocity are sparse, but there are some data in the tropical moorings and other time-series stations. There are also measurements of velocity from surface drifting
22â•… Ocean Forecasting Systems: Product Evaluation and Skill 0.1
1.5
0.2
0.3
0
0.7
0
1.4
1.0 0.8 20
1.
0.9
0.5
0.95
0.20
0.40
0.60
0.80
0.0 0.0
0.5 1.0 1.5 Dotted lines denote centred RMS (normalised wrt Obs) 0.1
0.2
0.3
0
1.6
Co
rre lati on 0.4 0.5 0.6 0.7
0
1.4
1.0 0.8 20
1.
0.9
0.5
0.95
+ Global North Atlantic Mediterranean Tropical Atlantic South Atlantic North Pacific Tropical Pacific South Pacific Indian Ocean Southern Ocean Arctic Stereo NATL12 IND12 MED12
0.20
0.60
0.0 0.0
0.80
0.99 1.00
Standard deviation (normalised wrt Obs)
+ Global North Atlantic Mediterranean Tropical Atlantic South Atlantic North Pacific Tropical Pacific South Pacific Indian Ocean Southern Ocean Arctic Stereo NATL12 IND12 MED12
0.99
1.5
b
rre lati on 0.4 0.5
0.40
a
Co
0.6
1.00
Standard deviation (normalised wrt Obs)
1.6
623
0.5 1.0 1.5 Dotted lines denote centred RMS (normalised wrt Obs)
Fig. 22.5↜渀 Taylor diagrams from a 2-year hindcast of the FOAM system for a SST comparison with AATSR data and b SSH comparison with a long-track altimeter data. The different colours and symbols represent the statistics for different geographical regions
624
M. Martin
buoys and these provide near-global coverage. These can be used as an independent check on the surface ocean currents, an important variable for a number of users. Surface drifters consist of a surface buoy which is attached to a subsurface drogue. This drogue is usually centred at 15€ m depth. The buoy measures temperature (and sometimes other ocean/atmosphere properties) and the position of the drifter is usually inferred from satellite transmission information. The SST data and position of the drifter are disseminated via the global telecommunications system (GTS). Three months of data from 1st January–31st March 2006 were quality controlled by checking the SST against climatology using a Bayesian technique, and by checking that the average daily velocity of the floats did not exceed 2€ m/s. The daily mean velocity values from drifter data were calculated by estimating the distance in the latitudinal and longitudinal directions between the first and last float positions during each day, and dividing the distance by the difference in their reporting time. The modelled velocity corresponding to the observed velocity was calculated by interpolating the model’s daily mean velocity fields to all of the observed drifter locations using a bilinear interpolation, and averaging the values for each day. There are a number of issues with estimated velocities from surface drifters, for example aliasing of inertial oscillations, inaccuracy of position data, unknown drogue depths, un-drogued data and different reporting frequencies. The technique described in the previous paragraph also introduces errors as the curvature in the path of the drifter is not taken into account. Other techniques for comparing the model and observed velocities exist. For example one could input the starting position for each drifter on a particular day, run the model forward to estimate its position at the end of the day, and compare that with the final observed position of the drifter. Statistics on these position errors could then be calculated and assessed. Various experiments were performed with the 1/9° resolution FOAM North Atlantic system (as it was in 2006, see Martin et€al. 2007 for details) in order to assess the impact of different aspects of the system on the surface currents. Figure€22.6 shows the Taylor diagrams for a sample of these experiments for the u and v components of the velocities in the North Atlantic. The first experiment (in light blue) was a re-run of the operational FOAM system which shows that the variability in the model was close to the observed variability but that the correlation was very low with a fairly high RMSD. When not assimilating altimeter data (dark blue), the model’s variability is much less, but the correlation coefficient is even worse. This implies that the altimeter assimilation is adding in variability to the model which is not naturally included in the model. One way of getting round this problem is to increase the viscosity in the model so that any spurious variability is damped. The results from a run of FOAM with an increased viscosity are shown in green. For comparison, the results from HYCOM and Mercator (as they were in 2006) are also shown in yellow and orange respectively. This shows improvements in the correlation and reduced RMSD compared to the other FOAM runs, giving similar results to HYCOM and Mercator.
22â•… Ocean Forecasting Systems: Product Evaluation and Skill 1.5
0.1
0.2
0.3
Co
rre
625
lat
ion
0.4
Standard deviation (normalised wrt Obs)
0.5 0.6 0.7 1.0
0.8
0.9 0.5
0.95
0.99
a
0.0 0.0 1.5
0.5 0.1
0.2
0.3
1.0 Co
rre
0.4
1.5
lat
ion
Standard deviation (normalised wrt Obs)
0.5 0.6 0.7 1.0 0.8
0.9 0.5
0.95
0.99
b
0.0 0.0
0.5
1.0
1.5
Fig. 22.6↜渀 Taylor diagrams for the a u and b v components of surface currents for various model runs during the period 1st January–31st March 2006 compared to velocity from surface drifters. Dark blue—FOAM with no altimeter assimilation; light blue—FOAM with altimeter assimilation; green— FOAM with altimeter assimilation and increased viscosity; yellow—HYCOM; orange—Mercator
626
M. Martin
22.4.4 Forecast Versus Analysis In order to assess the forecasts from ocean models, one can assume that the analysis produced by the data assimilation is providing a “best estimate”. The subsequent forecast can be compared against the analysis (at the correct time), and the differences between these fields can be used, over a large number of realisations, to assess the skill in the model forecast. Various statistics can be calculated based on these differences; the most commonly used are RMSD, mean and anomaly correlations, as described previously. It should be noted that these do not give the overall magnitude of the errors, as the analysis errors are not included, but they do provide information about the evolution of errors in time. The analysis errors should be computed separately (as described previously) and used in conjunction with these errors to provide information about the overall error in the forecasts. An example of the growth in the SSH forecast errors from the HYCOM/NCODA system (Metzger et€ al. 2009) is shown in Fig.€ 22.7 for various regions. Here, the median ACC and RMSD statistics are plotted as a function of forecast length out to 14 days. Globally, the model forecasts clearly have higher ACC and lower RMSD than the persistence forecasts throughout the 14-days. The picture is slightly different when looking at particular regions however. For instance, in the Kuroshio region, the forecast model is not providing much more skill than persistence due to the fact that the flow is dominated by mesoscale flow instabilities (rather than being dependent on the atmospheric forcing), although both forecast and persistence are more accurate than climatology throughout the period. In the Yellow Sea region where the ocean responds rapidly to the atmospheric forcing, a persistence forecast quickly becomes no better than climatology, whereas the forecast retains some skill out to at least 5 days. Another example of comparing forecasts with analyses is shown in Fig.€ 22.8 which shows August 2009 monthly average 5-day temperature forecast-analysis differences from the ¼° global FOAM system at 25 and 50€m depths. A number of features are apparent in these figures, but we focus here on the main broad-scale signal: at 25€m depth there is a clear negative bias in the northern mid-to-high latitudes, with a corresponding warm anomaly at 50€m depth. This dipole pattern indicates that heat is being mixed too vigorously in the model. This suggests that either the wind forcing is too strong, or the mixing scheme in the model is not representing the real-world mixing correctly. It is possible to independently validate the wind forcing, for example using scatterometer data. In this case it is thought that the main problem lies with the model’s mixing scheme, so the focus of model development here will be to improve this aspect of the model.
22.4.5 Case Studies for Particular Applications As described previously, ocean forecasting systems serve a large number of users. Among the most significant of these are the Navies, who are interested in a number
SSH anomaly correlation
22â•… Ocean Forecasting Systems: Product Evaluation and Skill .08˚ Global HYCOM (82.4) Median SSH Anomaly Correlation wldocn: number of forecasts 48 (01–Jun–2007 to 22–May–2008)
1
.08˚ Global HYCOM (82.4) SSH Median RMS error wldocn: number of forecasts 48 (01–Jun–2007 to 22–May–2008) 10
forecast
0.9
627
persistence
0.8
5
climatology forecast
0.7 0.6
SSH anomaly correlation
a
persistence 0
1
15
b
0
5
10
15
forecast length (in days) .08˚ Global HYCOM (82.4) SSH Median RMS error kurosh: number of forecasts 48 (01–Jun–2007 to 22–May–2008)
.08˚ Global HYCOM (82.4) Median SSH Anomaly Correlation kurosh: number of forecasts 48 (01–Jun–2007 to 22–May–2008) forecast
0.9
persistence
10
0.8
climatology
5
0.7
forecast persistence
c SSH anomaly correlation
10 forecast length (in days)
0.6 0
1
5
10 forecast length (in days)
15
.08˚ Global HYCOM (82.4) Median SSH Anomaly Correlation nwarab: number of forecasts 48 (01–Jun–2007 to 22–May–2008)
0
d
15
6
persistence
4
0.7
2
5
10
15
forecast length (in days)
1
10 5 forecast length (in days)
.08˚ Global HYCOM (82.4) SSH Median RMS error nwarab: number of forecasts 48 (01–Jun–2007 to 22–May–20 08)
8
0.8
e
0
forecast
0.9
0.6 0
SSH anomaly correlation
5
0
0
f
climatology forecast persistence 0
5
10
15
forecast length (in days)
.08˚ Global HYCOM (82.4) Median SSH Anomaly Correlation yelsea: number of forecasts 48 (01–Jun–2007 to 22–May–2008)
.08˚ Global HYCOM (82.4) SSH Median RMS error yelsea: number of forecasts 48 (01–Jun–2007 to 22–May–2008) 15
forecast
0.9
persistence
10
0.8
climatology
5
0.7
forecast persistence
0.6 0
g
5
10
forecast length (in days)
15
0
h
0
5
10
15
forecast length (in days)
Fig. 22.7↜渀 Median SSH anomaly correlation (↜left column) and median SSH RMSD (↜right column) against the verifying analysis as a function of forecast length for the global ocean (entire domain— top row), the Kuroshio (120–179°E, 21–55°N—second row), the northwest Arabian Sea (51–65°E, 15–26°N—third row) and the Yellow Sea (118–127°E, 30–42°N—bottom row). The red curves are HYCOM/NCODA forecasts, the cyan curves are for persistence of the nowcast and the black curves of RMSE are for the hindcast annual mean
of different outputs including information about sound speed in the ocean in order to model the acoustics (Metzger et€al. 2008, 2009). In order to produce accurate sound speed estimates, the temperature and salinity fields must be accurately determined, with the mixed-layer depth (MLD) and sonic-layer depth (SLD, Millero and Li 1994) of particular interest (amongst other parameters).
628
M. Martin
Fig. 22.8↜渀 Monthly average 5-day forecast temperature differences for FOAM for August 2009, compared with analyses at a 25€m and b 50€m depth
Metzger et€al. (2008, 2009) investigate the accuracy of the MLD and SLD forecasts in the HYCOM/NCODA system used by the US Navy. An example of this validation is shown in Fig.€22.9 which shows the mean and RMS errors in SLD as a function of forecast time for three regions. This shows that the model forecast and persistence are both producing more accurate estimates of SLD than is available from climatological estimates throughout the 14-day forecast. The skill of the model is generally similar to that of persistence, although this result is regionally dependent. The RMS errors generally show a large amount of variability which is most likely due to vertical interpolation errors, and could also be due to observation sampling issues.
22â•… Ocean Forecasting Systems: Product Evaluation and Skill 65 60 RMSE (m)
0 – 10 – 20
50
– 30
40
20
65
10
60
0 – 10 – 20
55 50
– 30
45
– 40
40 35
10 0
RMSE (m)
Mean Error (m)
55
45
RMSE (m)
Mean Error (m)
Mean Error (m)
10
– 10 – 20 – 30
629
2
4 6 8 10 12 Forecast length (days)
14
30
25 20
2
4 6 8 10 12 Forecast length (days)
14
Fig. 22.9↜渀 Error analysis of sonic layer depth (metre) as a function of forecast length based on 48 14-day forecasts by HYCOM/NCODA for regions MER4d (↜top), the western Pacific (↜middle) and the Arabian Sea (↜bottom). The left column shows mean error and the right column shows RMSD. The black curves are for HYCOM/NCODA forecasts, the blue curves are for persistence of the nowcast ocean state and the red curves are for the GDEM3 climatology. Note the y-axis differs between most plots
22.5â•…Summary and Conclusions An overview of methods which can be used to evaluate the accuracy and skill of ocean forecasts has been presented. Various statistical methods which can be used to perform evaluations have been defined, together with some useful diagrams for summarising related statistical information. A discussion on the importance of knowledge about the accuracy and quality of the observations used in the evaluations has also been given. Some examples of the application of the various statistical measures to GODAE ocean forecasting systems have been given. These were used to highlight the need to evaluate the ability of the model to reproduce the large-scale ocean circulation, the accuracy of the analyses, and the accuracy of the subsequent forecasts. The use
630
M. Martin
of independent data in assessing analyses and forecasts has also been presented, as has an example of validation directed at a particular user need. Various techniques which could be used to evaluate ocean forecasting systems have not been described in detail for different reasons. For example, it is possible to estimate a formal error estimate of the analysis using the Hessian of the cost function in variational data assimilation schemes. However, this is an expensive quantity to calculate and the output of the calculation is dependent on the input error covariance information which is usually not well known. For these reasons, it is not usually provided as an analysis error estimate. Similarly, for systems which run an ensemble of forecasts, the spread in the forecasts can be used to provide an estimate of the confidence which should be placed in the forecasts. The uncertainty in the initial conditions and the processes and parameterisations which are modelled can be sampled and the spread of the forecasts can then give statistical information on how much confidence should be placed in certain regions. However, the way in which the uncertainties in the system are sampled has a significant impact on the resulting forecast error estimates, and few operational ocean forecasting systems run an ensemble prediction system at present. Inter-comparison with other ocean forecasting systems can also provide useful information about the skill of a particular ocean forecasting system and insight into weaknesses that can easily be corrected. For more information on this subject, the reader is directed to the separate paper on inter-comparison methods. The evaluation of ocean forecast products is an important aspect of all the GODAE systems, and is continually being improved. It is hoped that common verification statistics will be produced routinely by all the systems over the coming years which will drive improvements to the systems themselves, and will also provide further insight into the most appropriate methods for their evaluation. Acknowledgements╇ The author would like to thank Joe Metzger, Nicolas Ferry and Peter Oke for their permission to reproduce results here. The author also gratefully acknowledges the FOAM team for input and useful discussions. The FOAM system was developed for the Royal Navy, and under the MERSEA and MyOcean Projects—partial support of the European Commission under Contracts SIP3-CT-2003-502885 and FP7-SPACE-2007-1 is gratefully acknowledged.
References Atger F (1999) The skill of ensemble prediction systems. Mon Weather Rev 127:1941–1953 Brier GW (1950) Verification of forecasts expressed in terms of probability. Mon Weather Rev 78:1–3 Cummings JA (2005) Operational multivariate ocean data assimilation. Q J R Meteorol Soc 131:3583–3604 Ferry N, Rémy E, Brasseur P, Maes C (2007) The Mercator global ocean operational analysis system: assessment and validation of an 11-year reanalysis. J Mar Syst 65:540–560 Ingleby NB, Huddleston MR (2007) Quality control of ocean temperature and salinity profiles— historical and real-time data. J Mar Syst 65:158–175
22â•… Ocean Forecasting Systems: Product Evaluation and Skill
631
Jolliff JK, Kindle JC, Shulman I, Penta B, Friedrichs MAM, Helber R, Arnone R (2009) Summary diagrams for coupled hydrodynamic-ecosystems model skill assessment. J Mar Syst 76:64–82 Locarnini RA, Mishonov AV, Antonov JI, Boyer TP, Garcia HE (2006) World Ocean Atlas 2005. In: Levitus S (ed) NOAA Atlas NESDIS 61. US Government Printing Office, Washington, p€182 Martin MJ, Hines A, Bell MJ (2007) Data assimilation in the FOAM operational short-range ocean forecasting system: a description of the scheme and its impact. Q J R Meteorol Soc 133:981– 995 Maximenko NA, Niiler PP (2005) Hybrid decade-mean global sea level with mesoscale resolution. In: Saxena N (ed) Recent advances in marine science and technology, 2004. PACON International, Honolulu, pp€55–59 Metzger EJ, Hurlburt HE, Wallcraft AJ, Shriver JF, Smedstad LF, Smedstad OM, Thoppil P, Franklin DS (2008) Validation test report for the Global Ocean Prediction System V3.0—1/12° HYCOM/NCODA: Phase I. Memorandum report No. NRL/MR/7320-08-9148, Naval Research Laboratory, Oceanography Division, Stennis Space Center, MS 39529-5004 Metzger EJ, Hurlburt HE, Wallcraft AJ, Shriver JF, Townsend TL, Smedstad OM, Thoppil P, Franklin DS (2009) Validation test report for the Global Ocean Forecast System V3.0—1/12° HYCOM/NCODA: Phase II. Memorandum report No. NRL/MR/7320-09-9236, Naval Research Laboratory, Oceanography Division, Stennis Space Center, MS 39529-5004 Millero FJ, Li X,(1994) Comments on “On equations for the speed of sound in seawater”. J Acoust Soc Am 95:2757–2759 Murphy AH (1988) Skill scores based on the mean square error and their relationships to the correlation coefficient. Mon Weather Rev 116:2417–2424 Murphy AH (1995) The coefficients of correlation and determination as measures of performance in forecast verification. Weather Forecast 10:681–688 Oke PR, Brassington GB, Griffin DA, Schiller A (2008) The Bluelink Ocean Data Assimilation System (BODAS). Ocean Model 21:46–70 Rio MH, Schaeffer P, Hernandez F, Lemoine JM (2005) The estimation of the ocean Mean Dynamic Topography through the combination of altimetric data, in-situ measurements and GRACE geoid: from global to regional studies. Proceedings of the GOCINA international workshop, Luxembourg Storkey D, Barciela RM, Blockley EW, Furner R, Guiavarc’h C, Hines A, Lea D, Martin MJ, Siddorn JR (2010) Forecasting the ocean state using NEMO: the new FOAM system. J Oper Oceanogr 3:3–15 Taylor KE (2001) Summarizing multiple aspects of model performance in a single diagram. J Geogr Res 106:7183–7192
Chapter 23
Performance of Ocean Forecasting Systems—Intercomparison Projects Fabrice Hernandez
Abstract╇ Ocean modelling, and more recently, ocean reanalysis or ocean forecasting system perform scientific assessment in order to evaluate errors and accuracy, but also to identify main drawbacks and possible improvements. Intercomparison has been a way to achieve such assessment among several numerical experiments. It is also a more robust approach for ocean state and forecast estimations. An historical overview of ocean model validation bringing to intercomparison activities is proposed here. Intercomparison projects performed over the last two decades by the oceanographic modelling community are presented and discussed here, in terms of objectives, and methodologies. Specific aspects for model, reanalysis, or ocean forecast experiment intercomparison are then detailed. Finally a particular focus is made on intercomparison studies performed in the framework of GODAE.
23.1â•…Introduction For the past 15 years, the development of Ocean Forecasting Systems (OFS) have been focusing in providing a continuous and routinely updated description of the ocean physical parameters for the past (hindcast1 and nowcast2 products), as well as in prediction mode (forecast products). Principal physical parameters of interest 1╇ Hindcast refers in the assimilation oceanographic community to ocean estimates obtained with an assimilated run where all observations are available, usually in delayed mode and numerical simulations performed over a past period. 2╇ Nowcast refers in the assimilation oceanographic community to ocean estimates obtained with an assimilated run in real time or near-real time where all possible observations are not yet available. This is the nominal “past estimates” that are provided by operational system over the previous days before the forecast.
F. Hernandez () Mercator Océan/IRD, Parc Technologique du Canal, 8–10 rue Hermes, 31520 Ramonville St. Agne, France e-mail:
[email protected] A. Schiller, G. B. Brassington (eds.), Operational Oceanography in the 21st Century, DOI 10.1007/978-94-007-0332-2_23, ©Â€Springer Science+Business Media B.V. 2011
633
634
F. Hernandez
being description of the water masses (temperature and salinity), of the currents (in the three dimensions), of sea level, of sea-state, of near-surface properties (like mixed layer depth, fronts) and of sea-ice. Heat and momentum exchanges with the atmosphere are also interesting meteorologists. More recently, by using coupled biogeochemical models, the ocean description is extended to ecosystem parameters from low to high trophic levels. Due to the sparseness of available ocean observations and due to errors attached to numerical models, the OFS development have tried to integrate observation description together with modelling approaches using assimilation methods. OFS are thus composed of numerical models of ocean dynamics, possibly coupled with seaice dynamics models and biogeochemical models, including forcing fields, together with ocean observations collecting systems, and assimilation procedures. The performance3 of such system depends on the robustness4, accuracy5 and reliability6 of these different components. This performance is thus appreciated from a user point of view by the accuracy and usefulness of ocean products delivered routinely by OFS (hindcasts, nowcasts, forecasts) for their respective applications. The OFS developed during the past years have first considered the ocean physical description. In many countries, local initiatives started to develop regional or coastal forecasting systems. In parallel, in the framework of GODAE (Global Ocean Data Assimilation Experiment, see https://www.godae.org/), some groups and countries worked to propose basin scale, or global description of the ocean dynamics. This second kind of forecasting systems is discussed here. More specifically, are discussed here the methodology proposed to evaluate the performance of eddy-permitting to eddy-resolving systems, where diurnal cycle and ocean high frequencies are not considered. Most of these systems rely on primitive equation ocean models, where tides dynamics are usually neglected (Dombrowsky et€ al. 2009). During the recent years, these systems benefitted from an ocean observability never reached before: the satellite altimetry together with ARGO, buoys and drifter programs strongly enhanced the mesoscale description since 2002 (Clark et€al. 2009). This observability promoted the development of state-of-the-art assimilation tools, and the implementation of mature multivariate methods (Cummings et€al. 2009). The GODAE system performance can be degraded by several causes, listed for their different components in Table€23.1. The four component listed here are different fields of ocean studies that have been usually studied separately. Thus, ocean modelling, as well as assimilation developments are usually associated with vali3╇ Performance has the same meaning than the title of this chapter, and is considered here in terms of usefulness and efficiency for users of ocean products provided by the OFS. In the framework of operational oceanography validation, a more specific definition is given later in this lecture notes. 4╇ Robustness (the quality of being able to withstand stresses, pressures, or changes in procedure or circumstance) is considered here in terms of OFS capacity to provide a consistent behavior and results under similar circumstances. 5╇ Accuracy is considered here as the degree of closeness of ocean estimates provided by the OFS to its actual true value. In the framework of operational oceanography validation, a more specific definition is given later in this lecture notes. 6╇ Reliability is considered here as the ability of the OFS to perform its required functions and provide ocean estimates under stated conditions while it is routinely operated.
23â•… Performance of Ocean Forecasting Systems—Intercomparison Projects
635
Table 23.1↜渀 List OFS components errors, that reduce the performance and increase ocean products errors Ocean model Numerical errors Physical parametrization and approximations (e.g., sub-grid parametrization) Explicitly not represented ocean processes (e.g., tides, diurnal cycle, surface gravity waves…) Errors on initial conditions External inputs Errors in forcing fields (atmospheric fluxes, river run-off errors) Bathymetry errors Climatology errors Boundary condition errors Observation Data accuracy level Data sparseness, aliasing effects Level of robustness of the multivariate estimation/correction Assimilation method Mismatch with the data representativeness Analysis shock Level of consistency for variational techniques (linear tangent model) in highly non-linear flows
dation studies to evaluate strength/drawbacks of new improvements. Most of the performance assessment methodologies applied in operational mode to OFS are derived from validation/evaluation techniques used separately by the research community on these components. For years, ocean modellers were solely evaluating their numerical results by (1) internal check, looking at consistency of ocean dynamics, or sensitivity studies to some parameters; and (2) external check, through the comparison of model results to reference studies or existing observations. Then intercomparison studies were scheduled, following example from the atmospheric modelling community.
23.2╅First Intercomparison Experiments The international Atmospheric Model Intercomparison Project (AMIP), in the framework of the World Climate Research Programme (WCRP) has provided a guidance for the oceanic modelling community. The aim of AMIP was to offer a comprehensive evaluation of the performance of atmospheric GCMs7 on climate and higher-frequency time-scales, and the documentation of their systematic errors. In a common modelling framework, that is, simulating the monthly variability of the atmospheric parameters for the 1990s decade, all climate modelling groups (more than 20 institutions around the world) provided their simulations in a standard way. Due to the participation of all groups in building up the assessment methodology, and the sustained reporting on evaluation of each experiment, AMIP has become the reference for atmospheric and climate performance assessment. An overview of AMIP is given in (Gates 1992). 7╇
GCM: Global Circulation Model.
636
F. Hernandez
Free coupled ocean/atmosphere numerical simulations are compared (usually monthly averaged parameters) against data (averaged in the same way), climatologies, or existing reference simulations. In particular ECMWF8, NCEP9 or COADS10 reanalysis, considered to be more realistic due to assimilation benefit. An ensemble approach was adopted. First each numerical simulation was individually evaluated (RMS11, correlation against the compared references). Then the ensemble mean, and its standard deviation were also evaluated. Ensemble approaches expect that individual simulations will present errors that are not correlated. In practice, this is not obviously true, if simulations are based on similar ocean models, similar forcings, etc…. However, by multiplying the number of simulations in the intercomparison, AMIP objective was clearly to get “not correlated” simulations. Such approach is presented in Fig.€ 23.1, taken from (Stammer et€ al. 2009)12 in the framework of the CLIVAR’s GSOP13. Models biased similarly bring to ensemble estimates also biased. With the era of model’s eddy-permitting capacity, different ocean modelling groups started to organize intercomparison experiment. The US-German Community Modelling Effort (CME), in support of the World Ocean Circulation Experiment (WOCE) started to infer model parametrization and sensitivity studies in modelling the North Atlantic basin (for a review, see Böning and Bryan 1996). The circulation and the eddy field were described in a limited way. Several causes were identified, among them boundary conditions, the representation of water exchanges and topographic controlled flows, overturning circulation and vertical mixing…. This experiment have been followed by the DYNAMO project, dedicated to offer intercomparison among three classes of ocean models of the North Atlantic Ocean in a similar numerical experiment framework (Meincke et€al. 2001). Forced identically, and configured over the same domain, a z-level, sigma-level and isopycnal vertical discretisation primitive equation models have been run in similar ways. The objective was to identify patterns of the North Atlantic Ocean circulation that were robust, and others that were sensitive to model parametrisation. Thus, one objective aimed to increase our knowledge of the Atlantic Ocean dynamics, the second was to improve ocean models, and share expertise among different modelling groups. Simulations were eddy-permitting (1/3° horizontal resolution). As far as possible, model parametrisations (i.e., lateral and vertical mixing, bottom friction, mixed layer turbulence, bathymetry, boundary conditions) were tuned to be similar, and initial conditions were provided by the Levitus climatology (for details, see Willebrand et€al. 2001). After a spin-up of 15 years, the last 5 years of the monthly-mean European Centre for Medium-Range Weather Forecasts. United States National Centers for Environmental Prediction. 10╇ Comprehensive Ocean-Atmosphere Data Set. 11╇ RMS: root mean square. 12╇ This OceanObs’09 community white paper is available at http://www.oceanobs09.net/cwp/index.php. 13╇ Global Synthesis and Observations Panel, (http://www.clivar.org/organization/gsop/synthesis/ synthesis.php). ╇ 8╇ ╇ 9╇
23â•… Performance of Ocean Forecasting Systems—Intercomparison Projects
637
Fig. 23.1↜渀 From Fig.€8 of (Stammer et€al. 2009): evaluating model quantity from multi-ensemble of results. The arrows illustrate the general expectance that assimilation of observations moves the results closer to the truth. The left panel show the ideal situation in which the ensemble spread and the distance to the ensemble mean provide useful measures while the right panel illustrates a biased case that is more realistic for the ensemble of present day synthesis
climatological forcing have been analysed following a protocol still considered nowadays for “consistency assessment” (explained later in this chapter): • Analysis of the meridional overturning circulation, that reflects the thermohaline circulation (mean annual values). Differences were analysed in term of deep flow and outflow/overflow representations, as well as diapycnal mixing effects. • Analysis of the overturning transport at 25°N. that also reflects the thermohaline circulation (mean annual values). Seasonal variation were also assessed in a specific study (Böning et€al. 2001). At this latitude, the western boundary current as well as the return circulations of the subtropical gyre are captured. Note that a particular effort has been put by the international community in order to have a sustained observation network of the flow across the Atlantic at that latitude. The RAPID array is a sustained program that provides data since 200414 (Cunningham et€al. 2007). The RAPID array uses standard observational techniques—moored instruments that measure conductivity, temperature and pressure, as well as bottom pressure recorders—to measure density and pressure gradients across the North Atlantic, from which one can readily calculate the basin overturning circulation and heat transport.
14╇
638
F. Hernandez
Meridional heat transport [PW]
1.5
1
0.5
0 20S
10S
EQ
10N
20N
30N
40N
50N
60N
70N
Fig. 23.2↜渀 Meridional heat transport in the North Atlantic Ocean from DYNAMO intercomparison (↜full line = LEVEL / dashed = ISOPYCNIC / dotted = SIGMA / dash-dotted = SIGMA-2). Values and errors bars given by Macdonald and Wunsch (1996). (Taken from Fig.€9 of Willebrand et€al. 2001)
• Analysis of the mean meridional heat transport, that reflects heat flux exchanges in a climatological aspect. Figure€23.2, taken from (Willebrand et€al. 2001) shows that models are underestimating the transport south of 20°N, compared to hydrographic data analysis, and that level and sigma models look less efficient in representing the transport in the subtropical gyre., due to their weaker Meridional Overturning Circulation (MOC) representation. • Analysis of mean surface circulation, associated with the mean geostrophic flow. Current at the surface and different depth are studied as well as vertically integrated transport. The Gulf Stream (transport across Florida Strait, separation at Cape Hatteras, the North West Corner flow), the North Atlantic Current, and the Azores Current representation are particularly discussed for the subtropical gyre (New et€al. 2001b). A dedicated study was performed for tropical currents in the western basin (South Atlantic Current, North Brazil Current, retroflection and North Atlantic Counter Current, eddies propagating into the Carribean current system) for the mean, and seasonal variations (Barnier et€al. 2001).
23â•… Performance of Ocean Forecasting Systems—Intercomparison Projects
639
• Analysis of the eddy field and its variability, associated with baroclinic and barotropic instabilities. The sea surface variability, as well as the eddy kinetic energy can be compared to satellite altimetry equivalent values (Stammer et€al. 2001). • Analysis of circulation at depth: pathways of the Mediterranean Waters, that impact the thermohaline circulation in the North Atlantic Ocean (New et€al. 2001a).
23.3â•…Evaluation and Intercomparison of Ocean Reanalysis With the availability of satellite altimetry in near real-time since the launch of ERS-1 (1991) and TOPEX/Poseidon (1992), assimilation techniques have been developed in order to provide more realistic descriptions of the ocean dynamics with ocean models. A first approach is to carry out reanalysis experiments, where models and assimilation are tuned to provide in the past the best description of the ocean circulation. Usually, the set of selected observations for assimilation is processed to remove possible biases and take into account differences in different types of observations. The set of forcing fields is also prepared in order to minimize errors, and long-term trend effects. Forcing estimates might merge observed and modelled parameters. During the experiment, successive intermediate runs might be performed in order to reduce errors identified in the meantime. And because state-of-the art ocean models are used, ocean reanalysis offer the most accurate description of the ocean for a set of “components” (i.e., choice of model and configuration, choice of observations and assimilation methodology). In fact, historically in the ocean community, the intercomparison of ocean reanalysis have been the first where objective was to be compared to the ocean truth. In the framework of GODAE and CLIVAR, the GSOP project aimed to intercompare different reanalysis computed over one to several decades (Fig.€23.3). One of the goal being to offer synthesis on ocean state estimation for climate research (Lee et€al. 2009a, b; Stammer et€al. 2009). The idea being that multi-model ensemble approaches can be useful to obtain better estimates of the ocean. In practice, the GSOP objectives are (1) to assess the consistency of the synthesis through intercomparison; (2) to evaluate the accuracy of the products, possibly by comparison to observations; (3) to estimate uncertainties; (4) to identify areas where improvements are needed; (5) to evaluate the lack of data that directly impacts the synthesis, and propose future observational requirements; (6) to work on new approaches, like coupled data assimilation. Another use of ocean reanalysis is to provide initial conditions for seasonal and climate forecasts. This is a much more “close-to-real-time-operation” application. The idea is to offer for present time, or for few weeks before, the best possible ocean description together with its error estimates, in order to start coupled ocean/atmosphere forecast for seasonal prediction (Balmaseda et€al. 2009).
640
F. Hernandez
z-Level Model
No Model
NCEP
ERA40
HOPE
Relax. Bias CORE corr. E-P.
Relax.
EN3
POP
OPA/NEMO
DePreSys
ECMWF
INGV
Mercator/ CERFACS
URDG
MOM
QSCAT GPCP
SODA
MIT
Relax. Relax.
GFDL
GODAS
K-7
GECCO
.25°x.25° 4D-Var
3D-Var/OI 1°x1° 2°x2°
DATA
Fig. 23.3↜渀 From (Stammer et€ al. 2009), Fig.€ 1, summarizing reanalyses taken into account by GSOP, sorted by forcing fields (↜green), type of ocean models (↜orange), assimilation methodology (↜pink), and résolutions (↜different blue)
For these two uses of reanalysis in ocean synthesis, errors listed in Table€23.1 are still relevant. One of the conclusions of the GSOP is that the full use of multiensemble assessment requires the detailed error information not only about data and models, but also about the estimated states. Figure€23.1 is illustrating that ocean estimates tend to cluster around methodologies and may not be independent from each other (see discussion in Stammer et€al. 2009). An important aspect of reanalysis accuracy, and the way intercomparison has to focus on, is their dependence on data to be assimilated in the past. Many ocean reanalysis are starting during the 1950s, when atmospheric reanalysis (NCEP and ECMWF ERA40) are available. Until 1978—first satellite with radiometer that provided Sea Surface Temperature (SST) with a global coverage—reanalysis can only rely on in-situ observations that are clearly under-sampling the ocean. As mentioned above, the ocean observability was strengthened with satellite altimetry in the 1990s. And since 2002, the ARGO array changed radically the ocean interior observability (e.g., Roemmich and Argo-Science-Team 2009). Note that atmospheric forcing accuracy has also been improved with satellite observations (radiometer for SST, heat content and exchanges in the atmosphere, and scatterometer for wind estimates). This lack of data in the past makes difficult any rigorous analysis of the ocean interannual and decadal variability.
23â•… Performance of Ocean Forecasting Systems—Intercomparison Projects
641
Another accuracy aspect is linked with multi-data assimilation approaches. Nowadays, most of the assimilation methods use multivariate scheme that corrects their background15 fields using information from temperature, salinity profiles, altimeter sea level measurements, SST from satellite and in-situ observations. Some also take into account satellite sea-ice data, satellite gradiometry, current measurements deduced from current meters or drifters etc…. In these assimilation schemes, every observation is impacting the model parameters. For instance, temperature observation should correct the salinity field, but also the sea-level, and reversely. Which means accuracy and intercomparison assessments have to considered carefully the relations between the corrected ocean parameters, and the observation errors in the framework of each forecasting system. Moreover, “representativeness” of data has to be taken into account in the assimilation scheme. For instance, coarse resolution models (e.g. 2° horizontal resolution), can clearly not reproduce ocean fronts and water mass distribution as observed by gliders on scale of few kilometres. The GSOP activity highlights most of these difficulties. A large number of studies have been performed using the reanalysis, among them, sea level variability, water mass pathways, variability of upper and mixed layer heat content, surface flux and run-off estimations, biogeochemistry, geodesy (see Lee et€al. 2009a for more details). Note that most of the topics are similar to those studied with free simulations (as mentioned above). In particular the MOC, corresponding to the regulation of the meridional heat transport that affects climate variability was subject to several analysis. Figure€23.4 provides a synthesis for the North Atlantic meridional heat transport. One can notice, compared to Fig.€23.2 that some reanalysis provide more accurate estimates compared to hydrography (Ganachaud and Wunsch 2000) in the subtropical gyre. It means that since the DYNAMO project, models, associated with data assimilation succeeded in improving the representation of the ocean general circulation. However, the spreading of the six estimates in Fig.€ 23.4 are larger than error bars from (Ganachaud and Wunsch 2000). Moreover, the four reanalysis based on ECCO are similarly below the reference, showing here correlated errors in ECCO systems that will strongly affect an ensemble mean. Figure€23.5 illustrates the difficulties in providing a robust evaluation of upper ocean heat content over more than 50 years. As mentioned earlier, the spreading before the 1970s seems associated with a lack of in-situ data. The ensemble standard deviation is reduced in the 1990s. However, since 2000, spreading appears again. This clearly raises the question of outliers with respect to the mean. Here, independent estimates should be used in order to evaluate reanalysis error levels. However, one can note a general tendency from all the time series: there is a clear warming of the upper ocean since the 1990s. The GSOP effort will continue in the future. Multi-model assessment and ensemble mean approach has been identified as the only way to provide reliable In the framework of assimilation the background is the state of the ocean model prior any correction by the assimilation method.
15╇
642
F. Hernandez
1.60
Heat transport (PW)
1.20
0.80
0.40
0.00 ECCO-SIO ECCO-JPL GFDL
–0.40 20°S
0°
20°N
40°N
ECCO-50y ECCO-GODAE INGV 60°N
80°N
Fig. 23.4↜渀 North Atlantic meridional heat transport, from Armin Koehl GSOP présentation at the CLIVAR/GODAE meeting on ocean synthesis evaluation, held at ECMWF, UK, in August 2006 (http://www.clivar.org/data/synthesis/intercomparison.php). (Point and error bars correspond to estimates from Ganachaud and Wunsch 2000)
ocean estimates. Which means that (a) intercomparison will still be used to evaluate discrepancies, and (b) that effort is needed to characterize uncertainties from each system. Data assimilation techniques should provide more robust control on analysis16 and innovations17. In parallel, the ocean model community is still working on improvements (see Griffies et€al. 2009 for a review). Moreover, work is still needed in order to reduce biases and make consistent historical dataset, but also clearly measure the impact of data type and availability on uncertainties (Heimbach et€al. 2009). The scientific assessment of these reanalysis will follow in a similar way. The main goal is still to characterise and understand the ocean medium and large scale patterns prior any further analysis. It means that the same ocean estimates analysed during CME or DYNAMO experiments will be evaluated first.
Here in the assimilation framework, the analysis is the production of an accurate image of the true state of the ocean at a given time, represented in a model as a collection of numbers. An analysis can be useful in itself as a comprehensive and self-consistent diagnostic of the ocean. It can also be used as input data to another operation, notably as the initial state for a numerical ocean forecast, or as a data retrieval to be used as a pseudo-observation. 17╇ The innovation is the discrepancies between observations and ocean model state, that is the vector of departures at the observation points. 16╇
23â•… Performance of Ocean Forecasting Systems—Intercomparison Projects
643
12m-rm seasonal anom: NATL Averaged temperature over the top 300m 0.6
0.4
ukdp ukoi cfcs2 cfas2 ecco50y
gfdl soda ecmfa ecmfc ukgs
ingv mrieccoSIO cfasa mct2
mct3 eccoJPLa eccoJPLc eccoMIT GMAO
0.2
0.0
–0.2
–0.4
–0.6
sdv ensm = 0.164 s/n ensm = 1.620 –0.8 1950
1960
1970
sdv all = 0.206 s/n all = 2.028
1980
spread = 0.101
1990
2000
Time
Fig. 23.5↜渀 Seasonal anomalies of integrated [0–300€m] température in the North Atlantic Ocean. Figure from Balmeseda and Weaver, GSOP présentation at the CLIVAR/GODAE meeting on ocean synthesis evaluation, held at ECMWF, UK, in August 2006 (http://www.clivar.org/data/ synthesis/intercomparison.php). Color code are indicated for each reanalysis. Gray shaded area correspond to ensemble mean standard déviation
23.4╅Intercomparison and Evaluation of Operational Ocean Forecasting System 23.4.1 D evelopment of Operational Ocean Forecasting System Evaluation The second use of data assimilation with ocean modelling has been dedicated to short terms ocean prediction. Operational oceanographic centres development is also related to the availability of satellite data. In the late 1990s, several groups had already proposed multivariate assimilation scheme enhancing ocean models capabilities, either based on quasi-geostrophic or primitive equation formulations (see Dombrowsky et€al. 2009 for a quick historical introduction). In the framework of GODAE, the main development of these groups focused on OFS providing daily estimates of hindcast, nowcast and short-term forecast18 of the ocean dynamics at 18╇
Short-term ocean prediction: between 5 days and 2 weeks.
644
F. Hernandez
mesoscale. That is, a description at length and time scales larger than 10€km and one day of the density field and water mass changes, of the currents (from surface Ekman currents to western boundary currents), and their respective transient effects in term of front, meanders, waves and eddy-like propagating features, from surface to depth. The objectives and potential applications of such OFS have been largely discussed with the terms of reference of GODAE (see Bell et€ al. 2009 for more details and references). However, one can mention ocean circulation description for synoptic to interannual studies, short term prediction for security (e.g. oil spill prediction, search and rescue activities), for water quality (by coupling with biogeochemical models, like algae bloom detection), for defence application (usually associated with acoustic modelling), or for fish stock assessment when coupled with efficient ecosystem and high trophic levels models. Evaluation methodology of OFS has first followed the path proposed by the modelling community, but had to take into account constraints that do not normally appear when performing model validation on academic project. First, by proceeding to the evaluation of the assimilation scheme, and its efficiency in providing accurate ocean analysis19. That is, more focused on accuracy than overall quality. In other words, where a certain level of quality is sought in pure modelling research (e.g., is there deep convection and Labrador Sea Water formed? a Gulf Stream overshoot? an acceptable meridional heat transport and Meridional Overturning Circulation?), assimilation experiments are tested on “realistic representation” where reference dataset are used to directly quantify error levels. A comprehensive error budget is also required for data assimilation results to be properly assessed. Assimilation schemes are more or less guided by background20 and observation errors, and the most sophisticated schemes provide robust analysis21 and forecast error estimates (Brasseur 2006; Cummings et€al. 2009). It is then required to verify the model error assumptions against dedicated error validation procedures. Second, by taking into account and measuring the impact of real-time constraints. That is, lack of data (observations that are not yet available during the assimilation time-window), and/or the low quality of these data, compared to reanalysis framework, where data are usually complete and fully controlled and corrected. Note also that in real time operations, forcing fields provided by weather forecast or atmospheric models might be less precise. And third, by focusing more specifically to the scientific assessment of forecast products, that is the evaluation of the performance and the predictability of the OFS. Performance is considered here not in its general definition, but more precisely associated with the benefit of using an ocean predicting model, together with an assimilation methodology that correct the ocean estimates produced by the OFS. Here the performance is a value of the usefulness of these different components for user’s interest and applications: to predict ocean current for the next week, why just not use a climatology? Why not applying a persistence approach, saying that See footnote 16. See footnote 15. 21╇ See footnote 16. 19╇ 20╇
23â•… Performance of Ocean Forecasting Systems—Intercomparison Projects
645
the ocean state next week, in a good approximation, is the same that the estimation computed today? In both cases, what is the added value of sophisticated tools like assimilation scheme and ocean models compared to climatology, to persistency approach? In practice, the idea is to evaluate forecast errors, compared to climatology or persistence errors, together with accuracy of the analysis (i.e., the efficiency of the assimilation scheme). Constraints appear also on technical/engineering aspects. Assessments have to be performed in real time, matching practical operational constraints, such as computer resource, storage capacity, and availability of reference values. Which means that the dataflow has to be monitored, since lack of input data for any technical reasons will directly impact the ocean estimates quality. Also, outputs from operational systems might be used for user-oriented applications (e.g., water quality, marine security or other societal use). Thus, the performance assessment methodology mentioned above has to rely on user requirements. Different applications might require different levels of accuracy. For instance, the accuracy of surface current forecasts dedicated to help search and rescue activities could not be matched by the operational systems, while the same ocean model may be used satisfactorily for a more general ocean study, or the climatology could be useful enough for some applications (e.g., tourist brochure). Thus, for all these reasons, using different model configurations and data assimilation methods, operational oceanography teams have tried to develop their tools for assessing the quality of outputs, in order to be able to provide “error bars” to users. Thanks to GODAE, these initiatives could be shared at the international level. An outlook of OFS validation is covered by the lecture of Martin (2011) during summer school. In this context, a common interest for intercomparison or collaboration on validation methods soon appeared among the different groups developing OFS. In the framework of the MERSEA Strand1 European Union (EU) project (2003– 2004), a first attempt was realized to intercompare eddy-permitting, basin scale ocean data assimilating systems. Hindcasts originating from the different systems were intercompared using climatology and historical high quality ocean datasets, like WOCE sections (Crosnier et€al. 2006). This validation methodology has been enhanced during the EU MERSEA Integrated Project (2004–2008, see http:// www.mersea.eu.org) on several aspects: (1) Perform routinely the validation, and thereby stimulate data processing and archiving centers to provide observations in real time; (2) Apply diagnostics that offer a robust scientific evaluation of each system, and select the most suitable diagnostics among those applied in research mode; (3) Evaluate both operational system performance and the products quality, taking into account user requirements (usually from short term to seasonal timescale applications); (4) Push for consistency of assessment among the different forecasting centres: applying similar diagnostics to the different systems, thus strengthening the overall assessment management activity through central team expertise; (5) Use this consistency to allow intercomparison of the operational systems, and thus design and implement a technical architecture that allows robust exchanges, interconnections, and interoperability between these systems. Which
646
F. Hernandez
is a milestone for implementing, in a consistent way, interoperable activities like ensemble forecasting. In the framework of GODAE, based on these advances for OFS scientific assessment, a special intercomparison exercise was decided, prepared, and carried out at the beginning of 2008 (Hernandez et€al. 2009). Some of the results are highlighted below.
23.4.2 Validation and Intercomparison Methodology The assessment methodology used ultimately for the GODAE intercomparison project is a direct heritage of the validation activity performed earlier in the framework of operational oceanography projects. It is based on two aspects (Crosnier and Le Provost 2007). First, «the philosophy»: a set of basic principles to assess the quality of OFS products/systems through a collaborative partnership: • Consistency: verifying that the system outputs are consistent with the current knowledge of the ocean circulation and climatologies. • Quality (or accuracy of the hindcast/nowcast): quantifying the differences between the system “best results” (analysis) and the sea truth, as estimated from observations, preferably using independent observations (not assimilated). • Performance (or accuracy of the forecast): quantifying the short term forecast capacity of each system, i.e. Answering the questions “does the forecasting system perform better than persistence and better than climatology?” • Benefit: end-user assessment of which quality level has to be reached before the products are useful for an application. Second, «the methodology»: a set of sharable tools for computing diagnostics, and a set of sharable standards to refer to, for assessing the products quality. Both tools and standards should be subject to upgrades and improvements in an operational framework. This methodology has been built using “metrics”: mathematical tools that compute scalar measures from systems outputs, compared to “references” (climatology, observations etc…). The metrics provide equivalent quantities extracted out of the different systems for the same geographic locations. Applied on different forecasting systems, they provide homogeneous and consistent sets of quantities that can be compared without depending to the specific configuration of each OFS (horizontal resolution, vertical discretization etc…). “Share-ability” is mandatory and allows each forecasting center to perform intercomparison and validation independently, using results from other centres. Metrics, are computed in a standardized way, the NetCDF file format using the COARDSCF convention is used, allowing time aggregation, easy and flexible manipulation, and self consistent meta-data representation. Distribution relies on internet communication protocols, basically through FTP. However, more user-friendly communication technologies based on OPeNDAP servers that can be visualized through a Live Access Server (LAS), through Dynamic Quick View portals or with similar
23â•… Performance of Ocean Forecasting Systems—Intercomparison Projects
647
Fig. 23.6↜渀 Summary of Class 2/3 metrics. All existing and available moorings, tide-gauges, XBT lines, WOCE/CLIVAR lines and others have been selected in order to define virtual sections and mooring points implemented in ocean models
clients that have now been widely adopted (Blower et€al. 2008). In practice, these technologies allow each forecasting centre to compute a considerable amount of diagnostics stored on the local servers of other centres. The total set of validation data do not need to be centralized requiring large storage capacities. Instead, for a given diagnostic, one can specifically gather the information spread across the different centres. Metrics are defined in four types, or “classes” (see Hernandez et€al. 2008 for more details): • Class 1 metrics, i.e. 3D standardized grids of temperature, salinity, currents, mixed layer depth, sea ice quantities and fluxes, can be directly compared to climatologies, but also at the surface to satellite observations (e.g., SLA, SST, or ice concentration). By using similar Class 1 grids, several OFS can intercompare their ocean estimates with a given reference dataset (example is provided in Fig.€23.8 in next section). • Class 2 metrics (virtual moorings and sections) are designed to match location of existing in-situ datasets as shown in Fig.€23.6. Then each time observations are provided (e.g., an XBT sections from a merchant ship), the Class 2 diagnostic can be performed routinely, and the model variable can be compared to “ground truth”. Figure€23.7 illustrates the use of Class 2 diagnostic for intercomparison between five systems in the Gulf of Cadiz. Compared to older WOCE hydrographic transects, it also allows a consistency assessment. Finally, it helps to assess improvements from two generation of Mercator systems. • Class 3 metrics concern derived quantities, like ocean transport, heat content, thermohaline circulation.
648
F. Hernandez
Fig. 23.7↜渀 Intercomparison of several ocean forecasting systems (↜Mercator1, TOPAZ, FOAM, HYCOM) during the European Project MERSEA Strand1 through a Class 2 salinity section averaged in September 2003 in the Gulf of Cadiz. Two WOCE lines are used as reference dataset. Further comparison was carried on when a new version of the Mercator system was developed (↜Mercator2)
• But to get closer to data, both for hindcasts and forecasts, Class 4 metrics were designed to build up a dataset of “model values equivalent to observations” for all OFS outputs: hindcast, nowcast and forecast. Thus, forecasting skill of OFS can be objectively evaluated. Class 4 diagnostics have been implemented in several centres for temperature, salinity (observations from Coriolis in-situ data centre), sea-ice concentration (maps from OSI-SAF22), sea level (satellite altimetry from AVISO23) and currents (from the Global Drifter Program). For all these diagnostics, a particular attention is paid to use independent observations, i.e., preferably not assimilated. Ideally, instead of satellite altimetry assimilated in most OFS, tide gauge data for sea level, or drifter or ADCP24 velocities for current. Table€23.2 summarizes the list of ocean/sea-ice parameters that can be evaluated with Class 4, and the corresponding data set. See Ocean & Sea Ice Satellite Application Facility at http://www.osi-saf.org/. See http://www.aviso.oceanobs.com/. 24╇ Acoustic Dopler Current Profiler. 22╇ 23╇
23â•… Performance of Ocean Forecasting Systems—Intercomparison Projects
649
Table 23.2↜渀 Ocean and sea-ice physical quantities, and corresponding available observations for validation in real time (RT) or delayed mode (DM) Data type Measurement In-situ temperature CTD (DM), XBT (RT), buoy (RT), mooring (RT/DM), TSG (DM), deep float (RT), glider (RT/DM) In-situ salinity CTD (DM), XCTD (DM), buoy (RT/DM), mooring (RT/ DM), TSG (DM), deep float (RT), glider (RT/DM) Sea surface temperature Satellite radiometer/radar (RT), TSG (DM), buoy (RT), mooring (RT/DM) Sea surface salinity TSG (DM), buoy (RT), mooring (RT/DM) [SMOS, Aquarius] (RT expected) Horizontal currents Drifters (RT), Current meter (DM), ADCP (DM) Satellite altimeter (RT), SAR (DM), High Frequency radar (DM), derived from SST (DM), derived from deep float displacement (DM) Sea level Tide gauges (RT), satellite altimeter (RT), GPS (to be tested) Ocean colour Satellite imagery (RT/DM) Sea Ice concentration, drift Satellite (RT) CTD conductivity temperature depth, XBT expendable bathythermograph, TSG thermosalinograph, XCTD expendable conductivity temperature depth
From Class 1, 2 and 3 metrics, the consistency and quality of each system could be deduced, or intercompared. For instance, daily section of operational run can be routinely compared to Class 2 historical section as illustrated in Fig.€23.7: in this case, the “general good looking” of the water masses distribution is verified against two historical WOCE lines: e.g., one expect that salinity signature of the Mediterranean waters appears at the proper depth. A system’s performance can be addressed using Class 4 metrics. The “benefit” could also be addressed using a set of Class 1, 2, 3 and 4 metrics. However, new “user-oriented” metrics might need to be defined to fully address this.
23.4.3 The GODAE Intercomparison Project Recently the GODAE Intercomparison Project has allowed to intercompare and perform accuracy and consistency assessment. The objectives of the project were to (a) demonstrate GODAE operational systems in operations; (b) share expertise and design validation tools and metrics endorsed by all GODAE operational centers; (c) evaluate the overall scientific quality of the different GODAE operational systems (results are summarized in Hernandez et€al. 2009). This project involved the majority of operational centres worldwide delivering daily ocean products, such as: BLUElink (Australia), HYCOM (USA), MOVE/ MRI.COM (Japan), Mercator (France), FOAM (United Kingdom), C-NOOFS (Canada), and TOPAZ (Norway) systems (Dombrowsky et€al. 2009; Hurlburt et€al. 2009). It provides a diversity of ocean models -4 types; global, or regional; based on
650
F. Hernandez
different vertical discretizations; eddy-permitting to eddy-resolving; coupled or not with sea-ice models; using different types of air-sea flux modelling. It provides also some diversity of assimilation techniques, using not the same kind of observations; proceeding to weekly or daily analysis or updates; based on sequential or variational approaches; based on single or ensemble analysis and predictions; applying or not “close to data” schemes like First Guess At Appropriate Time (FGAT) and Incremental Analysis Update (IAU) techniques (Bloom et€al. 1996). It was initially decided to analyse similarly the operational outputs of the different OFS involved. February, March and April 2008 was the selected period. In practice, all output could not be provided in real-time, and the scientific evaluation has been performed with some month delay. A series of observations and reference dataset have been used to assess the accuracy and consistency of the ocean products, using Class 1 and Class 2: • Weekly maps of Sea Surface Height (SSH) or Sea Level Anomalies (SLA) from AVISO satellite altimetry25. • Weekly maps of surface currents derived from satellite altimetry (Larnicol et€al. 2006). • The Levitus WOA 2005 climatology (Antonov et€al. 2006; Locarnini et€al. 2006). • The Mixed Layer Depth climatology (D’Ortenzio et€al. 2005; de Boyer Montégut et€al. 2004, 2007). • Daily sea-ice concentration from satellite from OSI-SAF26. • OSTIA GHRSST SST products (Donlon et€al. 2009). • In-situ temperature and salinity, provided by CORIOLIS. In practice, all groups could contribute to the intercomparison. Specific studies were carried out: in the north, the south and the tropical Atlantic, the western north Pacific, the tropical Pacific, and the Indonesian Seas. All group had access to all output and reference dataset. SST consistency and accuracy was verified against OSTIA maps. Water mass consistency was evaluated using the WOA 2005 climatology. Mixed Layer Depth consistency was verified with the climatology. Sea level and mean circulation were assessed against satellite altimetric maps. Mean and eddy kinetic energy were compared at the surface with SURCOUF maps (Larnicol et€al. 2006). Three months are rather short to infer the circulation patterns analysed in DYNAMO or reanalysis project. However, the consistency assessment allowed verifying if the “mean” circulation was like expected. For instance, in Fig.€23.8 the current analysis in the North Atlantic showed us the consistency of the subtropical and subpolar gyre circulation. One can note that the Azores Current appears for some systems, that the Gulf Stream extension does not spread similarly, or that the Labrador and the East-Greenland current are more or less intense. The use SURCOUF data allowed to status on the quality of the different outputs: eddy kinetic energy can be computed, and accuracy numbers given. However the high resolution systems 25╇ 26╇
See footnote 23. See footnote 22.
23â•… Performance of Ocean Forecasting Systems—Intercomparison Projects
651
Fig. 23.8↜渀 From (Hernandez et€al. 2008). Averaged eddy kinetic energy (m2/s2) from February to April 2008 for: TOPAZ (↜top-left), HYCOM (↜middle-left), C-NOOFS (↜middle-bottom-left), FOAM (↜top-right), Mercator High-res (↜middle-right), Mercator global (↜middel-bottom-right) and the observed SURCOUF product (↜bottom-left). Bottom-right: Time series of eddy kinetic energy box averaged in limited area around the Gulf Stream (80–60°W and 30–42°N)
652
F. Hernandez
seem to provide more energetic features than SURCOUF. Here we can suspect that SURCOUF currents are smoother than HYCOM. Which indicate than reference dataset have also to be taken with caution. Similar limitation appeared using OSTIA SST: OSTIA map can be dubious when satellite data are lacking. Thanks to OSTIA error estimates provided together with SST values, the intercomparison could focus on “valuable” areas. The full overview of this first intercomparison experiment is given in (Hernandez et€al. 2008, 2009). This first international intercomparison of OFS was limited to a short period, and a short set of ocean parameters. Impact of forcing field was not studied, neither the time-varying aspect of ocean features (eddy propagation, waves…) or sea-ice. The analysis was also limited to hindcast: forecast and performance metrics could only be assessed in a limited way. This initiative should carry on in the framework of the GODAE Ocean View project. More reference dataset will be made available in real-time soon, and the methodology, the metrics, are now adopted by most groups. This experiment has shown that intercomparison and evaluation of OFS could be performed in any part of the ocean. The three-months limited period could address the consistency, and accuracy of OFS for this season. The performance of the system, with regard of their particularities (resolution, model approximations, assimilation method…) started to be evidenced. Next step would be to carry on the intercomparison in term of multi-model ensemble assessment.
23.4.4 User Oriented Validation As mentioned earlier, most of the validation methodology proposed for ocean models and OFS is based on the “oceanographer point of view”. That is, evaluation of the large scale circulation, and smaller scale features in a general sense. Even if accuracy number and error bars can be produced by this approach, they might not fully satisfy some users. For instance, a merchant ship captain may not be satisfied with a daily averaged map of sea-ice concentration, instead, he might prefer a map of ice-edge and position of ice-extent, with the probability of ice-drift for the next day. Many examples could be mentioned, particularly concerning coupled physical/biogeochemical parameters that impact ecosystem behaviour, or coastal applications (e.g., De Mey et€al. 2009). Oil spill prediction has been one of the applications particularly studied. Major disaster pushed authorities to develop oil spill models. They were first driven by wind and waves effects. With the availability of ocean current forecast, new oil spill models have been developed. In the framework of MERSEA, simulated experiment at sea, together with oil spill modeling have been carried out. Intercomparison has been a key point: oil spill predictions were performed using different OFS current. It allows to check the robustness of the predictions, and ensemble forecast analysis was performed (see Hackett et€al. 2009 for a review). Similar studies were achieved for search and rescue drift-prediction models.
23â•… Performance of Ocean Forecasting Systems—Intercomparison Projects
653
23.5â•…Conclusion Due to easier exchange of data and numerical experiment results worldwide, the growing community of ocean modelling or OFS is more subject to mutual validation and collaborative work. Moreover, model and forecast evaluation benefit now from a shared methodology endorsed by the GODAE community. It is a first step toward the assessment approach implemented in the Numerical Weather Prediction community. More operational validation tools are planned to be implemented in the framework of the European MyOcean projects, and intercomparisons activities will be carried in GODAE Ocean View. Note also that validation needs the limited number of existing ocean data for accuracy assessment. Hence, more groups apply similar techniques, and tend to work together. Intercomparisons of ocean models and forecasts is thus coming a standard approach. However, there are still specific validation aspects with respect to academic studies, reanalysis evaluation, or ocean forecast performance assessment, corresponding to each specific framework. For instance, ocean forecasting systems have to deal more particularly with data availability and quality. The validation approach presented here, and proposed by the open ocean community, is slowly extended to the coastal and the biogeochemical modelling communities. Note that the ocean observing community, dealing with the need to infer the impact of future observation system is soliciting the OFS for impact studies where the validation methodology is used to assess the performance of the simulated networks.
References Antonov JI, Locarnini RA, Boyer TP, Mishonov AV, Garcia HE (2006) World ocean atlas 2005. In: Levitus S (ed) Salinity, vol€2. U.S. Government Printing Office, Washington, p€182 Balmaseda MA, Alves O, Arribas A, Awaji T, Behringer DW, Ferry N, Fujii Y, Lee T, Rienecker M, Rosati A, Stammer D (2009) Ocean initialization for seasonal forecasts. Oceanogr Mag 22:154–159 Barnier B, Reynaud T, Beckmann A, Böning CW, Molines J-M, Barnard S, Jia Y (2001) On the seasonal variability and eddies in the North Brazil current: insights from model intercomparison experiments. Progr Oceanogr 48:195–230 Bell MJ, Lefebvre M, Le Traon P-Y, Smith N, Wilmer-Becker K (2009) GODAE, the global ocean data experiment. Oceanogr Mag 22:14–21 Bloom SC, Takacs LL, da Silva AM, Ledvina D (1996) Data assimilation using incremental analysis updates. Mon Weather Rev 124:1256–1271 Blower JD, Blanc F, Cornillon P, Hankin SC, Loubrieu T (2008) Underpinning technologies for oceanography data sharing, visualization and analysis: review and future outlook. Final GODAE Symposium 2008: the revolution in global ocean forecasting GODAE: 10 years of achievement. Nice, France, GODAE, pp€301–310 Böning CW, Bryan FO (1996) Large-scale transport processes in high-resolution circulation models. In: Krauss W (ed) The warmwatersphere of the North Atlantic Ocean. Gebrüder Borntraeger, Berlin, pp€91–128
654
F. Hernandez
Böning CW, Dieterich C, Barnier B, Yanli J (2001) Seasonal cycle of meridional heat transport in the subtropical North Atlantic: a model intercomparison in relation to observations near 25°N. Progr Oceanogr 48:231–253 Brasseur P (2006) Ocean data assimilation using sequential methods based on Kalman filter. In: Chassignet EP, Verron J (eds) GODAE Summer school in ocean weather forecasting: an integrated view of oceanography. Springer, Dordrecht, pp€371–316 Clark C, Wilson S, Benveniste J, Bonekamp H, Drinkwater MR, Fellous J-L, Gohil BS, Lindstrom E, Mingsen L, Nakagawa K, Parisot F, Roemmich D, Johnson M, Meldrum D, Ball G, Merrifield M, McPhaden MJ, Freeland HJ, Goni GJ, Weller P, Send U, Hood M (2009) An overview of observing system relevant to GODAE. Oceanogr Mag 22:22–33 Crosnier L, Le Provost C (2007) Inter-comparing five forecast operational systems in the North Atlantic and Mediterranean basins: the MERSEA-strand1 methodology. J Mar Syst 65:354–375 Crosnier L, Le Provost C, MERSEA Strand1 team (2006) Internal metrics definition for operational forecast systems inter-comparison: examples in the North Atlantic and Mediterranean Sea. In: Chassignet EP, Verron J (eds) GODAE summer school in ocean weather forecasting: an integrated view of oceanography. Springer, Dordrecht, pp€455–465 Cunningham SA, Kanzow T, Rayner D, Baringer MO, Johns WE, Marotzke J, Longworth HR, Grant EM, Hirschi J, Beal LM, Meinen CS, Bryden HL (2007) Temporal variability of the Atlantic meridional overturning circulation at 26.5°N. Science 317:935–938 Cummings JA, Bertino L, Brasseur P, Fukumori I, Kamachi M, Martin MJ, Mogensen KS, Oke PR,Testut C-E, Verron J, Weaver A (2009) Description of assimilation methods used in GODAE systems. Oceanogr Mag 22:96–109 De Mey P, Craig P, Davidson F, Edwards CA, Ishikawa Y, Kindle JC, Proctor R, Thompson KR, Zhu J, GODAE Coastal and Shelf Seas Working Group (2009) Application in coastal modelling and forecasting. Oceanogr Mag 22:198–205 Dombrowsky E, Bertino L, Brassington GB, Chassignet EP, Davidson F, Hurlburt HE, Kamachi M, Lee T, Martin MJ, Mei S, Tonani M (2009) GODAE systems in operation. Oceanogr Mag 22:80–95 Donlon CJ, Casey KS, Robinson IS, Gentemann CL, Reynolds RW, Barton I, Arino O, Stark JD, Rayner NA, Le Borgne P, Poulter D, Vazquez-Cuervo J, Beggs H, Jones LD, Minnett P (2009) The GODAE high resolution sea surface temperature pilot project (GHRSST). Oceanogr Mag 22:34–45 Ganachaud A, Wunsch C (2000) Improved estimates of global ocean circulation, heat transport and mixing from hydrographic data. Nature 408:453–457 Gates WL (1992) AMIP: the atmospheric model intercomparison project. Bull Am Meteorol Soc 73:1962–1970 Griffies SM, Adcroft A, Banks H, Böning CW, Chassignet EP, Danabasoglu G, Danilov S, Deleersnijder E, Drange H, England M, Fox-Kemper B, Gerdes R, Gnanadesikan A, Greatbatch RJ, Hallberg RW, Hanert E, Harrison MJ, Legg SA, Little CM, Madec G, Marsland S, Nikurashin M, Pirani A, Simmons HL, Schröter J, Samuels BL, Treguier A-M, Toggweiler JR, Tsujino H, Vallis GK, and White L (2009) Problems and prospects in large-scale ocean circulation models. In: Fischer AS (ed) OceanOb’s 2009 Hackett B, Comerma E, Daniel P, Ichikawa H (2009) Marine oil pollution predication. Oceanogr Mag 22:168–175 Heimbach P, Forget G, Ponte RM, Wunsch C, Balmaseda MA, Awaji T, Baehr J, Behringer D, Carton JA, Ferry N, Fischer AS, Fukumori I, Giese BS, Haines K, Harrison E, Hernandez F, Kamachi M, Keppenne C, Köhl A, Lee T, Menemenlis D, Oke PR, Remy E, Rienecker M, Rosati A, Smith DE, Speer KG, Stammer D, Weaver A (2009) Observational requirements for global-scale ocean climate analysis: lessons from ocean state estimation. In: Fischer AS (ed) OceanOb’s 2009 Hernandez F, Bertino L, Brassington GB, Cummings JA, Crosnier L, Davidson F, Hacker P, Kamachi M, Lisæter KA, Mahdon R, Martin MJ, Ratsimandresy A (2008) Validation and intercomparison of analysis and forecast products. Final GODAE Symposium 2008: the revolu-
23â•… Performance of Ocean Forecasting Systems—Intercomparison Projects
655
tion in global ocean forecasting GODAE: 10 years of achievement. Nice, France, GODAE, pp€147–191 Hernandez F, Bertino L, Brassington GB, Chassignet EP, Cummings JA, Davidson F, Drévillon M, Garric G, Kamachi M, Lellouche J-M, Mahdon R, Martin MJ, Ratsimandresy A, Regnier C (2009) Validation and intercomparison studies within GODAE. Oceanogr Mag 22:128–143 Hurlburt HE, Brassington GB, Drillet Y, Kamachi M, Benkiran M, R. Bourdallé-Badie, Chassignet EP, Jacobs GA, Le Galloudec O, Lellouche J-M, Metzger EJ, Oke PR, Pugh TF, Schiller A, Smedstad OM, Tranchant B, Tsujino H, Usuii N, Wallcraft AJ (2009) High resolution global and basin-scale ocean analysis and forecasts. Oceanogr Mag 22:110–127 Larnicol G, Guinehut S, Rio M-H, Drévillon M, Faugère Y, Nicolas G (2006) The global observed ocean products of the french mercator project. International Symposium on Radar Altimetry: 15 years of altimetry, ESA/CNES Lee T, Awaji T, Balmaseda MA, Greiner E, Stammer D (2009a) Ocean state estimation for climate research. Oceanogr Mag 22:160–167 Lee T, Stammer D, Awaji T, Balmaseda MA, Behringer D, Carton JA, Ferry N, Fischer AS, Fukumori I, Giese BS, Haines K, Harrison E, Heimbach P, Kamachi M, Keppenne C, Köhl A, Masina S, Menemenlis D, Ponte RM, Remy E, Rienecker M, Rosati A, Schröter J, Smith DE, Weaver A, Wunsch C, Xue Y (2009b) Ocean state estimate from climate research. In: Fischer AS (ed) OceanOb’s 2009 Locarnini RA, Mishonov AV, Antonov JI, Boyer TP, Garcia HE (2006) World ocean atlas 2005, In: Levitus S (ed) Temperature, vol€1. U.S. Government Printing Office, Washington, p€182 Martin M (2011) Ocean Forecasting systems: product evaluation and skill. In: Schiller A, Brassington GB (eds) Operational oceanography in the 21st century. Springer, Dordrecht, pp 611–632 Meincke J, Le Provost C, Willebrand J (2001) DYNAMO. Progr Oceanogr 48:121–122 New AL, Barnard S, Herrmann P, Molines J-M (2001a) On the origin and pathway of the saline inflow to the Nordic Seas: insights from models. Progr Oceanogr 48:255–287 New AL, Jia Y, Coulibaly M, Dengg J (2001b) On the role of the Azores current in the ventilation of the North Atlantic Ocean. Progr Oceanogr 48:163–194 Roemmich D, Argo-Science-Team (2009) Argo: the challenge of continuing 10 years of progress. Oceanogr Mag 22:46–55 Stammer D, Böning CW, Dieterich C (2001) The role of variable wind forcing in generating eddy energy in the North Atlantic. Progr Oceanogr 48:289–311 Stammer D, Köhl A, Awaji T, Balmaseda MA, Behringer D, Carton JA, Ferry N, Fischer AS, Fukumori I, Giese BS, Haines K, Harrison E, Heimbach P, Kamachi M, Keppenne C, Lee T, Masina S, Menemenlis D, Ponte RM, Remy E, Rienecker M, Rosati A, Schröter J, Smith DE, Weaver A, Wunsch C, Xue Y (2009) Ocean information provided through ensemble ocean syntheses. In: Fischer AS (ed) Oceanob’s 2009 Willebrand J, Barnier B, Böning CW, Dieterich C, Killworth PD, Le Provost C, Jia Y, Molines J-M, New AL (2001) Circulation characteristics in three eddy-permitting models of the North Atlantic. Progr Oceanogr 48:123–161
Part VIII
Applications, Policies and Legal Frameworks
Chapter 24
Defence Applications of Operational Oceanography An Australian Perspective Robert Woodham
Abstract╇ Oceanographic conditions can affect naval operations in a variety of ways, and for this reason navies around the world have traditionally used oceanographic observations, and climatologies derived from them, for operational decision making. Rapid advances in global ocean observing systems since the 1990s, and more recently in operational ocean forecasting systems, offer substantial opportunities for improved decision making. The recent focus of many defence forces on information superiority has coincided with the availability of high resolution forecasts of oceanic physical properties. These oceanic data sets are being used to assess and forecast such properties as: sea surface height, temperature and salinity, for acoustic applications to undersea warfare; and oceanic currents and tidal streams, for Search and Rescue (SAR), mine warfare and amphibious applications. The Royal Australian Navy (RAN) is using ocean forecasts from the BLUElink global ocean modelling system, and a limited area ocean model, and is developing a very high resolution model for applications in the littoral zone, as well as integrating high resolution oceanographic data into sonar range prediction models. These military applications of operational oceanography are reviewed, and illustrated with examples from an Australian perspective.
24.1â•…Introduction It is perhaps self-evident that navies around the world are interested in ocean conditions. What may not be so obvious, however, is the variety of ways in which the ocean can affect naval operations. This chapter aims to describe oceanographic impacts on maritime operations, and how these can be assessed and forecast using operational oceanographic capabilities which have become available in recent years. Whilst its content is generally applicable to naval forces, it is presented from the viewpoint of the Royal Australian Navy (RAN), which has been closely involved R. Woodham () Directorate of Oceanography and Meteorology, Royal Australian Navy, Sydney, Australia e-mail:
[email protected] A. Schiller, G. B. Brassington (eds.), Operational Oceanography in the 21st Century, DOI 10.1007/978-94-007-0332-2_24, ©Â€Springer Science+Business Media B.V. 2011
659
660
R. Woodham
in the establishment of oceanographic observation and forecasting in Australia, as a partner in the ‘BLUElink’ project (Brassington et€ al. 2007). Jacobs et€ al. have recently published a more general overview of how operational oceanography is being used by navies throughout the world, which includes examples from the United States, the United Kingdom, France and Australia (Jacobs et€al. 2009). Harding and Rigney have previously published an overview of operational oceanography in the United States Navy (Harding and Rigney 2006). In general terms, military forces around the world have given increasing attention, in recent times, to the importance of basing decision making on the most comprehensive and up to date information available. This is partly due to the increasing capabilities offered by information and communications technologies (ICT), and partly due to a change in emphasis to manoeuvre warfare, rather than positional (or attritional) warfare. The focus on manoeuvre warfare has its roots in the latter part of the Cold War, when NATO realised that it must use force-multipliers if it was to overcome the numerical superiority of Soviet forces. These force multipliers included information superiority and the manoeuvrist approach. The related concept of ‘Network Centric Warfare’ (NCW), as distinct from a platform centric approach, envisages the rapid collection and dissemination of actionable information, using the latest technologies, to achieve information superiority throughout the battlespace. Environmental information, including Meteorological and Oceanographic (METOC) information, is regarded by modern navies as a vital component of information superiority and NCW, allowing naval forces to optimise their weapons, sensors and manoeuvre for the prevailing and forecast environmental conditions. For these reasons, the more technologically advanced world navies have been quick to take advantage of the recent rapid developments in operational oceanography, which have been described elsewhere in this volume. Improved oceanic observations, data management and forecast systems have all been applied to naval operations, in order to contribute to the goal of information superiority. This approach is particularly applicable in Australia, because oceanographic conditions in the region are so complex (see Fig.€ 24.1 for geographic locations referred to in this paragraph). The East Australian Current affects the Tasman Sea, spawning numerous warm- and cold-core eddies (Ridgway and Dunn 2003). The Leeuwin Current flows down the west coast and across the Great Australian Bight. The Pacific-Indonesian Throughflow affects the Timor and Arafura Seas and the Northwest Shelf. The Antarctic Circumpolar Current affects waters to the south of the Australian continent. Other oceanographic phenomena in the region include upwelling events (particularly along the Queensland coast, and the Bonney coast of South Australia), internal waves, solitons, extreme tidal ranges and abundant freshwater inflows, providing strong buoyancy forcing during the Northwest monsoon. Faced with the need to operate successfully in such complex waters, the RAN has been quick to appreciate the need to maintain a state-of-the-art oceanographic capability. It is working closely with partners, notably the Australian Bureau of Meteorology (BoM) and the Commonwealth Scientific and Industrial Research Organisation (CSIRO), to develop such a capability.
24â•… Defence Applications of Operational Oceanography
661
Fig. 24.1↜渀 Geographic locations in the Australian region referred to in the text
24.2â•…Impacts of the Ocean on Operations 24.2.1 Anti-Submarine Warfare (ASW) The need for specialist oceanographic expertise first came to be recognised by the RAN in the mid 1950s, when the Fairey Gannet Anti-Submarine Warfare (ASW) aircraft was first operated from the aircraft carrier HMAS MELBOURNE. Meteorological officers onboard MELBOURNE provided tactical oceanographic advice to the Gannet squadrons, using bathythermographic observations of the ocean as the basis for sonar performance predictions. This advice was used by the Gannet crews to determine the optimum deployment of buoys fitted with hydrophones (‘sonobuoys’), which they used in the acoustic detection and tracking of submarines.
662
R. Woodham
In making these acoustic assessments, the effects of the ocean’s thermohaline structure on the propagation of sound in water must be considered. The effects on sound speed of temperature, salinity and pressure are as follows: • Temperature—sound speed is higher in warmer water (4€ms−1 per 1°C) • Salinity—sound speed is higher in more saline water (1.4€ms−1 per 1 PSU) • Depth—sound speed is higher at greater pressure (1.7€ms−1 per 100€m) Acoustic propagation in water can be understood by imagining that sound propagates through a homogeneous medium in straight lines (the ‘raytrace’ approach) (Urick 1983). The refraction of sound rays is described by Snell’s Law, which states that, when a ray crosses a boundary between two media in which its speed of propagation ( v ) is different: sin θi v1 = , sin θr v2
where iâ•›, r are the angles of incidence and refraction. This means that sound in the sea is refracted towards areas of lower sound speed. The degree of refraction is also frequency dependent, being greater for higher frequencies. Snell’s Law can be applied qualitatively, to understand the acoustic properties of the water column, and hence determine optimum tactics, such as search or evasion plans. It can also be applied quantitatively, in sonar range prediction models, such as the RAN’s ‘Tactical Environmental Support System version 2’ (TESS 2). These models estimate detection ranges, based on ocean acoustics, the performance characteristics of sonar systems (such as operating frequencies, transmitted power, pulse length, processing losses and gains, etc), and a knowledge of target characteristics (such as target strength, depth, aspect, etc). Ray-tracing models are generally found to give good results at medium and high frequencies (above 1–2€ kHz). These frequencies are typically used by active sonars, which transmit a pulse of acoustic energy, and detect its echo (as distinct from passive sonars, which detect radiated noise from a target). Active sonars are fitted in ships and submarines, and can be deployed from aircraft as sonobuoys or, in the case of helicopters, on winches (‘dipping’ sonars). Consider a typical thermal profile of the ocean, such as the one taken from the central Tasman Sea shown in Fig.€24.2. This profile is from the ‘Ship Of Opportunity Programme’ (SOOP) dataset, and has been extracted from the Integrated Marine Observing System (IMOS) Ocean Portal. The top 20–30€m shows an isothermal profile in the mixed layer. Here, temperature and salinity are constant, but pressure will increase as depth increases, giving rise to a slight increase in sound speed. This will have the effect of refracting sound waves upwards towards the surface. If the sound frequency is sufficiently high, in comparison to the depth of the mixed layer, acoustic rays travelling through the water at small angles to the horizontal will be refracted upwards towards the surface, where they will be reflected. After reflection, the rays will again be refracted upwards towards the surface. This has the effect of trapping acoustic energy in the surface ‘duct’, which can give rise to low acoustic losses, and therefore long ranges. Because higher frequencies are refracted more, there is a ‘cut-off’ frequency, below which acoustic energy will not be trapped in the
24â•… Defence Applications of Operational Oceanography
663
Temperature Profile for the XBT (88605242) launched at lat/lon (-36.8165/156.9538) Line PX34 Sydney-Wellington on the: 27-Jan-2009 19:08:00 0 -100 -200
Depth in metres
-300 -400 -500 -600 -700 -800 -900 -1000 -1100 -3
0
5
10 15 20 Temperature in degrees Celsius
25
30
35
Fig. 24.2↜渀 Typical thermal profile of the ocean, taken from the central Tasman sea. (Data is from the Ship of Opportunity Programme (SOOP), and was obtained from the IMOS ocean portal)
duct. If the surface wind is light, surface losses due to scattering on reflection will be low, and very long ranges are possible. In order to take advantage of this ducting effect, hull-mounted active sonars in ASW frigates are generally designed to operate at frequencies which are high enough to be trapped by the surface duct, in order to maximise detection ranges against shallow submarines. Below the mixed layer, the water gets colder in the thermocline zone. Between the base of the mixed layer at around 30€ m, and the base of the thermocline at around 100€ m, the temperature has fallen by about 7°C (Fig.€ 24.2). This means that the sound speed will have increased by around 1.7€ms−1, due to the increasing pressure, but fallen by around 28€ms−1 due to the decreasing temperature. Overall there is a large decrease in sound speed, which means that acoustic energy will be refracted downwards. Between 100€m and 800€m, there is a drop in temperature of around 1°C per hundred metres (Fig.€24.2), which means that the sound speed will decrease by about 2.3€ms−1 per hundred metres. In the water column below the mixed layer, these various effects result in a downward-refracting profile, stronger in the main thermocline region, which means that acoustic energy will be refracted down towards the sea bed. If the sea bed is a good absorber of acoustic energy at the relevant frequency, acoustic propagation will be generally poor.
664
R. Woodham
Fig. 24.3↜渀 Sound speed profile through the Tasman sea at 155° East on 31 March 2008. (Data is from the Ocean Forecast Australia Model (OFAM). An anticyclonic eddy is evident at around 32°S)
Although the profile shown in Fig.€24.2 extends to around 850€m, the depth of water in this location is around 5,000€ m. Below 850€ m, there will come a point where the decrease in sound speed caused by the temperature lapse is negated by the increase in sound speed due to the pressure increase. In isothermal water, clearly the sound speed will increase with depth, and acoustic energy will start to be refracted up towards the surface. The consequences of an increasing sound speed at depth can be seen in Fig.€24.3, which shows a sound speed cross section through the Tasman Sea at longitude 155°E, from 30°S to 40°S. Temperature and salinity data has been obtained from the Ocean Forecast Australia Model (OFAM) (Brassington et€al. 2007) for 31 March 2008, and converted to sound speed using Mackenzie’s equation (Mackenzie 1981). There is a sound speed minimum at a depth of around 1,200€m, whilst below this depth, sound speed increases due to the small temperature lapse combined with the increasing pressure. At depth, the value of sound speed increases to be similar to that at the surface. The sound speed minimum at 1,200€m is associated with an acoustic channel, which is a very low-loss path. Above 1,200€m, sound tends to be refracted downwards, towards the channel axis. Below 1,200€m, sound tends to be refracted upwards, again towards the channel axis. A depth of 1,200€m would therefore be a good depth at which to position a hydrophone, in order to take advantage of this low-loss path in the acoustic detection of submarines.
24â•… Defence Applications of Operational Oceanography
665
Where the sound speed at depth exceeds the value at the surface, a ‘convergence zone’ may be experienced. This is a ring around the acoustic source, typically with a radius of around 25€miles, where sound is focussed by a caustic effect. This focussing of acoustic energy near the surface provides opportunities for greatly increased ranges, and it is even possible for multiple convergence zones to be present, giving even longer range detections. In addition to the vertical gradients of sound speed discussed so far, horizontal gradients of sound speed are caused by temperature and salinity gradients associated with fronts and eddies, and these can also have a large effect on the acoustic properties of the ocean. For example, an anticyclonic (warm core) eddy will have warmer water towards its centre, therefore there will be a lateral gradient of sound speed associated with the eddy. An anticyclonic eddy can be seen in Fig.€24.3, at around 32°S, with an associated sound speed maximum at a depth of around 200€m. If a surface ship outside the eddy is searching for a submarine in the centre of the eddy, using active sonar, the sound will be refracted away from the submarine, reducing the probability of detection. Similarly, if the ship and submarine are on the opposite sides of an oceanic front, detection ranges will be much reduced. Acoustic effects such as the ones described in this section have been well known, and have been the principal concern of naval oceanographers, for a long time. Recent advances in operational oceanography are starting to provide the highly detailed oceanic data required to enable acoustic assessments and forecasts to be made at greatly increased spatial and temporal resolutions, suitable for tactical applications. For example, a submarine wishing to evade acoustic detection can use such oceanographic data to identify a location in the thermocline beneath the mixed layer, where the water is not too deep and the bottom is a good absorber of low frequency noise. This will ensure that its radiated noise is directed down to the sea bed, where it is absorbed, hence minimising counter-detection ranges. ASW aircraft can use high-resolution oceanographic data to identify near-surface sound channels, deploying the hydrophones on their sonobuoys or dipping sonars in the channel in order to achieve the greatest possible detection ranges. Knowledge of the location of fronts and eddies enables ASW frigates to design the most effective search plans, armed with an accurate assessment of detection ranges. These are just a few examples of how the wealth of oceanographic data now available presents abundant opportunities for the ingenuity of naval oceanographers and tacticians to be stimulated. As well as temperature and salinity, ocean currents also have an effect on ASW, and should be considered by naval forces. Submarines can take advantage of currents to increase their speed over the ground, whilst keeping their engines at low power (and therefore operating quietly). In some cases, particularly in the Australian region, ocean currents can run at 3 or 4 knots (Roughan and Middleton 2002), so this effect can be significant. Ocean currents can also be taken into account in sonar range prediction systems, such as the RAN’s TESS 2, since they affect sound speed.
666
R. Woodham
24.2.2 Amphibious Warfare Amphibious operations can be very sensitive to weather and oceanographic conditions. The offloading of troops and equipment from specialist amphibious shipping to a beachhead involves transfers from ships to landing craft, and from landing craft to the beach itself. Most navies possess a range of relatively small watercraft for use during amphibious operations. Such activities are sensitive to sea state, swell and surf conditions, tidal streams, longshore currents and rips, which must all be assessed and forecast in order to ensure mission success. Many navies use sea, swell and surf models to predict oceanic conditions in the littoral environment, and hence assess their impact on amphibious operations. Figure€24.4 shows the output from an experimental implementation of the ‘Simulating Waves Nearshore’ (SWAN) wave model, and the US Navy’s ‘Surf’ model, which displays model output using a Geographic Information System (GIS). The model has been run over North Beach, Cronulla, which is on the east coast of New South Wales to the south of Sydney (Fig.€24.1). Figure€24.4 shows: significant wave height (grey contours) and direction (vectors); significant wave period (blue rasters); littoral currents (closely spaced arrows along the approach to the beach); wave trains (displayed in grey as representative wave crests); and breaker percentage (displayed as green for 15%). This information can be used in the planning phase of an amphibious assault, to compare the suitability of
Fig. 24.4↜渀 Sea and surf conditions forecast for North beach, Cronulla. See text (Sect.€24.2.2) for an explanation of the symbology
24â•… Defence Applications of Operational Oceanography
667
various beaches for the operation, or to predict conditions at the beach at the time of the assault. Depending on the nature of the assault, a suitable beach may be required to have negligible surf and manageable longshore currents, although a single line of low, spilling surf may be tolerated. The location of the beach centre and approach lanes can also be chosen, using model output of this type, to avoid rips. A knowledge of the location and strength of longshore currents in the boat lanes, at the time of the assault, will help the landing craft crews to make a successful approach and beaching. The RAN is developing a high resolution forecasting system, called the ‘Littoral Ocean Modelling System’ (LOMS), which will provide sea, swell and surf predictions at greater resolution and fidelity, and over larger domains, than the SWAN/ Surf implementation described above. It will provide a three dimensional characterisation of the wave conditions, at resolutions in the order of tens of metres.
24.2.3 Mine Warfare Mine Warfare operations include mine hunting (using specialist sonars and Remotely Operated Vehicles (ROVs)), mine sweeping, and mine clearance diving. These operations are generally conducted in littoral environments, which can be challenging due to the complexity of ocean conditions. Tidal streams are often strong, turbidity can affect visibility, and variations in the bottom type and thermohaline structure can make acoustic detection difficult. In order to achieve good detection and resolution of small objects, mine hunting sonars typically operate at relatively high frequencies (hundreds of kHz). This means that typical detection ranges are quite low, and so ocean models with horizontal resolutions in the order of tens of kilometres are unable to provide adequate resolution of the oceanic structure for these applications. The RAN uses a limited area oceanic model, called the ‘Relocatable Ocean Atmosphere Model’ (ROAM), which is described in Sect.€24.4.2 below, to generate forecasts at resolutions down to 1 or 2€km and the LOMS model will provide even higher resolution in the near future. A mine warfare variant of the TESS 2 sonar range prediction software, called TESS 2€M, provides acoustic assessments at the scales required by mine warfare applications. The main demands on naval oceanographers supporting mine warfare operations are often to assess and predict wind waves, swells and currents. Currents, both at the surface and at depth, depend on the tidal regime, wind driven flow and influence of the current structure in the adjacent deep ocean basin, all of which can be modelled by systems such as ROAM. ROVs and divers may be limited by the surface conditions and the strength of these currents. Forecasts are used to identify windows of opportunity, when wind waves, swells and current strengths are low enough that such activities will not be unduly hampered. Conditions of high turbidity can also hamper diving operations by reducing visibility. The thermohaline structure of the littoral water mass is of interest to mine hunting operations, since it affects the performance of high-frequency mine hunting sonars. It can have a substantial impact, particularly in the case of a salt wedge estuary or where there is strong tidal
668
R. Woodham
modulation. River outlets can affect the thermohaline structure on short timescales, for example when thunderstorm activity or heavy rain causes a sudden increase in outflow. Naval oceanographers must also be mindful of weather patterns, which can affect ocean conditions, and must predict events such as a sudden increase in sea state due to a frontal passage.
24.2.4 Submarine Operations Submarine operations require a knowledge of the locations of fronts and eddies (see Sect.€24.2.1 above), and the general thermohaline structure of the ocean, in order to identify the best tactics for detection, attacks and evasion. The strength and direction of ocean currents is also required, for the purposes of manoeuvre. A knowledge of conditions at the surface, such as wind waves and swells, enables the risk of counterdetection to be assessed, informing decisions on whether it is safe to raise a periscope or communications mast, or to recharge batteries by ‘snorting’. A knowledge of surface wind waves and precipitation is also required, in order to assess ambient noise from these sources, as this affects the performance of acoustic sensors.
24.2.5 Search and Rescue (SAR) Analyses and forecasts of ocean currents are invaluable in the assessment of drift during Search and Rescue (SAR) operations, by informing the design of effective search plans. Perhaps the most complex aspect of such calculations are associated with the drift of the object being searched for under the influence of the wind (or ‘leeway’) (Hackett et€al. 2006). Objects with different shapes, such as persons wearing lifejackets, survival rafts and lifeboats, experience different leeway effects. Even without the assistance of algorithms which account for leeway effects, a good approximation can often be obtained from ocean models which include currents, tidal streams and Ekman flow. In addition, knowledge of sea surface temperatures (SST) allows survival times to be estimated. Ocean modelling systems can even be used to investigate historical problems of this type, such as the search for the location of HMAS SYDNEY II, which was greatly assisted by oceanic drift calculations using BLUElink reanalysis data (Mearns 2009; Griffin 2009). The SYDNEY wreck site was located off Western Australia in April 2008, 66 years after the ship sank, with the tragic loss of her entire ship’s company.
24.2.6 Maritime Interdiction Operations The bulk of the chapter so far has concentrated on high-end warfighting applications of operational oceanography, such as prosecuting submarines, clearing mine-
24â•… Defence Applications of Operational Oceanography
669
fields and conducting amphibious assaults. Oceanographic products are also used, however, to provide routine support to lower tempo operations. Examples include maritime interdiction, patrol tasks and constabulary activities, which may be constrained by high sea states or heavy swells. A current example is the use of real-time satellite observations, and forecasts, of significant wave height to identify the risk of pirate attacks off the coast of Somalia. The correlation between pirate attacks and satellite observations of significant wave height has been established using historical data. A ‘stoplight’ diagram, based on these correlations and using forecast wave heights, is routinely provided to naval forces in the international Combined Task Force 150 (CTF 150), operating off the Somali coast. This force includes an Australian frigate. The stoplight product shows the risk of pirate attack in three categories: ‘probable’, ‘possible’ and ‘unlikely’ (Fig.€24.5).
24.3â•…Forecast Methods—Their Strengths and Weaknesses 24.3.1 Climatology Until the recent advent of operational oceanography, navies have had to rely on climatologies or point observations to make operational decisions (Jacobs et€al. 2009). Climatologies can be useful for planning purposes, but they are of limited use where oceanic variability is high. In the extreme case of a bimodal system, climatology shows the mean of the two modes, which may be a physical situation that never arises in reality (e.g. south or north of a front, inside or outside an eddy). Figure€24.6 illustrates the limitations of climatology, by showing the September monthly mean SST in the Tasman Sea as depicted by the World Ocean Atlas 2001, and the daily mean SST on 16 September 2009 from the BLUElink forecasting system. Conversely, where variability is low, or where it occurs on timescales longer than the averaging period (normally monthly), climatology can give a very good indication of expected conditions. Furthermore, the expected error of a forecast based on climatology is independent of the lead time of the forecast (see Martin (2010, Fig.€8b)). Forecasts based on deterministic models or persistence perform better, on average, than climatology in the early part of the forecast period. This is illustrated in Fig.€8b of Martin (2010), which shows the median RMS errors of global Sea Surface Height (SSH) forecasts based on climatology, persistence and a deterministic model. Climatology gives better guidance than other forecast methods, such as persistence or deterministic models, at long lead times (Murphy 1992), because forecasts based on persistence or unbiased deterministic models asymptote to twice the climatological variance at long time periods. This is because, once these forecasts are completely decorrelated from reality, they have errors resulting from having anomalies in the wrong places, as well as errors from not having anomalies in the right places (Kalnay 2003). For this reason, defence forces normally use climatological oceanographic data when conducting long-range planning.
Fig. 24.5↜渀 36€h Forecast of risk of pirate attacks off the Somali coast, based on significant wave height, valid 28 August 2009
670 R. Woodham
24â•… Defence Applications of Operational Oceanography
671
Fig. 24.6↜渀 Comparison of SST as depicted by climatology (↜upper panel) (World Ocean Atlas 2001, September) and an oceanic model (↜lower panel). (BLUElink Reanalysis, 16 September 2009)
24.3.2 Persistence Point observations, such as temperature profiles from eXpendable Bathy Thermograph (XBT) systems, have been used by navies for many decades to infer the acoustic properties of the water column. These observations are relatively simple to make, and do not require assistance from ashore. This approach amounts to a persistence forecast, that is, an assumption that the water properties will not change during the period for which the assessment is required. It also assumes that there is no spatial variation in temperature, so only range-independent sonar predictions can be made using this approach. Persistence forecasts can be expected to have lower errors than climatology at the start of the forecast period, but as the oceanic flow evolves from this initial state, the errors grow rapidly (Murphy 1992). Spatial
672
R. Woodham
variations, caused by the physical movement of the ship or aircraft making the observation, as well as temporal variations due to ocean dynamics, contribute to these errors. Nonetheless, a persistence forecast may be useful where uniform conditions can reasonably be expected, such as in high latitudes where mixed layers are very deep, or over continental shelves, provided the water is well mixed, and advection is minimal. Persistence is a more valid approach for short lead time forecasts, such as may be required for a Co-ordinated Anti-Submarine EXercise (CASEX) lasting two or three hours, than for longer lead times. Nevertheless, where spatial and temporal variability is great, such as in the waters around Australia, a persistence forecast may be misleading even on very short timescales. An ASW frigate which makes an XBT observation just inside or outside an eddy, for example, may soon experience very different acoustic conditions from those inferred from the XBT observation.
24.3.3 Deterministic Forecasts Deterministic forecasts of the ocean have only become available in relatively recent times (Bell et€al. 2000). Nevertheless, rapid progress has been made over the last decade, including the introduction of eddy-resolving models and the assimilation of new observational data. See Brassington (Brassington 2010) for a comprehensive review of progress in operational ocean forecasting. In Australia, the BLUElink ocean forecasting system commenced routine forecasting operations in August 2007 (Brassington et€al. 2007). Provided sufficient observational data is available, deterministic forecasts should have relatively small errors at the start of the forecast period. These errors will grow more slowly than persistence forecast errors, because the deterministic model is able to keep up with changes in the state of the ocean, by modelling its dynamic processes (see Fig€22.7b, Martin (2010)). Deterministic forecasts from systems such as BLUElink are highly detailed, providing variables such as temperature, salinity, currents and sea surface height at high spatial resolution for forecast periods of several days. They represent a huge advance on the persistence and climatological forecasts used by navies for many decades. In one sense, however, their strength is also their weakness, since it is difficult to transmit the high volumes of oceanic data now available from deterministic ocean forecasting systems from shore to ships and submarines, due to the bandwidth limitations of naval communications systems.
24.3.4 Ensemble Forecasts Ensemble forecasting is well established for Numerical Weather Prediction (NWP), but less so for oceanographic forecasting. Ensemble techniques are used to generate covariance matrices for oceanic data assimilation applications (Oke et€ al. 2005), and some ocean models have tangent linear and adjoint versions, which can be used
24â•… Defence Applications of Operational Oceanography
673
to generate ensembles of initial conditions. Ensemble forecasts of the ocean offer great benefits to military users, since they enable the expected accuracy of forecasts to be quantified at the start of the forecast period. Ensemble techniques can also be used to provide probabilistic forecasts, which assist military commanders by enabling them to understand operational risks. The computational demands of implementing an operational ensemble forecasting system can, however, be substantial.
24.4â•…Naval Applications of Deterministic Forecasts The ocean analysis and forecasting capability of the BLUElink system has been described by Brassington (Brassington 2010). This Section will describe how forecasts from the BLUElink system, including ROAM, are used by the RAN for operational decision making.
24.4.1 The BLUElink Global/Regional Model (OceanMAPS) The BLUElink Ocean Modelling, Analysis and Prediction System (OceanMAPS) is implemented at the Bureau of Meteorology (BoM) in Melbourne (Brassington 2010). It produces an analysis and 6-day forecast of ocean temperature, salinity, currents, sea surface height and mixed layer depth twice per week. Model output graphics are available from the BoM public website, and the model data itself is available to the RAN, and more generally for research purposes, from the BoM’s ‘Thematic Realtime Environmental Distributed Data Services’ (THREDDS) server, in Network Common Data Form (NetCDF) format. The OceanMAPS system is currently configured to give eddy-resolving resolution (10€km horizontally) over the Australian region (90°E–180°E and 16°N–75°S). Within this domain, OceanMAPS data is routinely used by the RAN to create oceanographic charts, which are available to naval personnel for a range of applications, including ASW, amphibious and mine warfare, passage planning and spatial awareness. An example of a ‘METOC Oceanographic Forecast Summary’ (MOFS) chart is shown in Fig.€24.7. From the MOFS chart shown in Fig.€24.7, it can be seen that the East Australian Exercise Areas (EAXA), shown as blue polygons, are dominated by a large anticyclonic feature at the southern extremity of the East Australian Current. There is a sharp temperature gradient associated with this feature, at around 35°S, where reduced sonar ranges may be expected. For an ASW exercise in this location, the ASW commander might decide to allocate search assets either side of the temperature gradient, in order to achieve an efficient search. The submarine commander may chose to remain in the core of the current associated with this temperature gradient, in order to evade detection. By moving to either side for brief periods, sonar performance can be improved so that the tactical picture can be compiled. The
674
R. Woodham
Fig. 24.7↜渀 ‘METOC Oceanographic Forecast Summary’ (MOFS) chart, showing SST and currents in the Tasman Sea for 17 May 2010. MOFS charts are routinely produced twice weekly, showing forecasts out to 6 days, and are used by a variety of naval personnel
24â•… Defence Applications of Operational Oceanography
675
currents can also be exploited by the submarine to increase speed over the ground. Air assets deploying lines of sonobuoys can apply knowledge of the current field to ensure that the buoy patterns are not sheared out of shape by the flow. In addition, specialist oceanographers (‘METOC’ officers) may be available to provide further insights into the acoustic properties of the area, using OceanMAPS data, and hence assist decision making.
24.4.2 Relocatable Ocean Atmosphere Model (ROAM) The Relocatable Ocean Atmosphere Model (ROAM) is used by the RAN to generate high resolution oceanic and atmospheric forecasts over limited domains of interest to the Australian Defence Force (ADF). ROAM is designed to be set up by non-expert users, with minimal input, anywhere in the Australian region (Herzfeld 2009), and is used routinely by RAN forecasters. The ROAM ocean model is initialised and forced by data from OceanMAPS, and is typically implemented at resolutions of 1–2€km. Figure€24.8 shows Sea Surface Temperature (SST) and currents calculated by ROAM for a domain in the vicinity of Hobart, Tasmania (see Fig.€24.1), which was used for the RAN mine warfare exercise ‘DUGONG’. Exercise ‘DUGONG’ involved the Mine Hunter Coastal (MHC) vessels HUON and DIAMANTINA, which provided mine sweeping and hunting capabilities, the auxiliary minesweeper BANDICOOT, clearance diving teams and US Navy salvage divers. It took place over two weeks in October 2009 in the Derwent River, and the approaches to Hobart. In this example, the current characteristics were of primary importance to the exercise, which involved an underwater survey of the historic wreck of MV Lake Illawarra in the Derwent River. The water temperature was also of interest to the diving teams, to ensure that they were suitably prepared for the prevailing conditions. ROAM was used to generate current forecasts at intervals down to one hour. Additionally, the ROAM atmospheric model provided high resolution forecasts of the wind strength and direction, also at one hour timesteps, which allowed changes in the sea state to be anticipated. These also proved to have a significant impact on the exercise. Note that Fig.€24.8 does not show the full resolution of the ROAM model, as it has been expanded to show conditions in Storm Bay. As well as providing oceanographic data for graphical products, the output from ocean forecasting systems can be used in sonar range prediction systems, in order to produce assessments and forecasts of acoustic conditions which take account of the spatial and temporal variability of the ocean environment. Figure€24.9 shows a series of sonar range predictions, which have been generated by the RAN’s Tactical Environmental Support System (TESS 2) using ROAM data at 1€ km resolution. The domain is in the vicinity of Jervis Bay, which is around 130€km south of Sydney (Fig.€24.1). It is an area where the RAN frequently conducts ASW and MW exercises. The sonar range predictions are displayed as ‘Probability of Detection’ plots (PODgrams), where a 90% or greater probability of detection is shown in red.
676
R. Woodham
Fig. 24.8↜渀 Sea Surface Temperature (SST) and current forecast produced by the ROAM system for the mine warfare Exercise DUGONG in October 2009
Figure€24.9 shows ROAM sea surface temperature and currents as the background. The three PODgrams are for an ASW frigate leaving Jervis Bay and tracking to the northeast, searching for a submarine at Periscope Depth (PD). Similar calculations may be run at any depth required by the user. The capabilities of the sonar used for the calculation are fictional. The PODgrams seem to make sense intuitively, since they show the greatest ranges inshore, where the water is shallow and with a relatively homogeneous thermohaline structure, and bottom losses are low from the sandy sea bed. Offshore, where the temperature gradient is greater, detection ranges are less. The scale of Fig.€24.9 can be gauged by considering that the current vectors are shown at the ROAM resolution of 1€km. The PODgrams have hollow centres because echoes cannot be received whilst the sonar is transmitting. This gives rise to
24â•… Defence Applications of Operational Oceanography
677
Fig. 24.9↜渀 Sonar performance predictions produced by ROAM and TESS 2 for 1000 UTC on 06 October 2009, in the vicinity of Jervis Bay, NSW. The background shows sea surface temperature (↜colour stretch and contours) and current vectors. The three ‘Probability of Detection’ plots (PODgrams) are over-plotted, with a probability of detection of 90% or more shown in red
a ‘dead zone’ of varying radius, depending on the duration of the transmitted pulse, and the speed of sound in water.
24.5â•…Summary Oceanographic data has been collected by the world’s navies for many years, and used to inform the planning and conduct of a range of naval operations. Perhaps the main preoccupation of the naval oceanographer is with the acoustic properties of the ocean, because acoustic detection is of great importance in Anti-Submarine Warfare (ASW) and Mine Warfare (MW). The effects of oceanic temperature and salinity, and the depth of water, on sound speed are well known. This means that oceanographic conditions can be used to infer acoustic properties, both qualita-
678
R. Woodham
tively by naval personnel, and quantitatively in sonar range prediction systems. The relatively recent advent of operational oceanography has made available a wealth of observational and forecast data at high spatial and temporal resolutions. Recent advances span ocean observation systems, data assimilation and deterministic forecasting models. These new datasets are being used by naval oceanographers to provide much improved characterisations of the physical structure of the ocean, in order to inform operational and tactical decision making. The time and space scales which are starting to be resolved allow oceanographic support to be provided in complex littoral environments, where there is a demand from amphibious and mine warfare operations. This new oceanographic capability is timely, given a broader trend towards information superiority in the more technologically advanced defence forces. This chapter has described oceanographic effects on ASW, Amphibious Warfare, Mine Warfare, submarine operations and lower tempo activities such as Search and Rescue (SAR) and maritime interdiction operations. The acoustic properties of the ocean have been outlined in some detail, using examples from the Tasman Sea. The strengths and weaknesses of various forecasting methods (climatology, persistence, deterministic and ensemble forecasts) have been described, from the perspective of naval forces. Finally, some examples have been given of the use of deterministic forecasts of the ocean, including ASW activities in the Tasman Sea, a Mine Warfare exercise in the approaches to Hobart, Tasmania, and the use of high resolution oceanographic data to generate range-dependent sonar predictions in the Jervis Bay exercise areas. The maturing international capability for operational oceanography presents a remarkable opportunity for the world’s navies and maritime forces, and this has been seized on by the RAN and other leading navies. As the resolvable time and space scales continue to reduce, and progress is made with the downscaling of global systems to coastal scales, the complexities of the littoral environment will continue to be unravelled. It is an exciting time to be a naval oceanographer. Acknowledgments╇ I would like to thank the International GODAE Summer School organizing committee, for their kind invitation to present a lecture on defence applications of operational oceanography at the International GODAE Summer School in Perth, Western Australia, during January 2010. This chapter is based on the lecture. Sincere thanks also to Lieutenant Commander Aaron Young, RAN, for his kind assistance with some of the figures. I am also very grateful to Stephen Ban and Lieutenant Commander Richard Bean, RAN, for helpful suggestions which have improved the text.
References Bell MJ, Forbes RM, Hines A (2000) Assessment of the FOAM global data assimilation system for real-time operational ocean forecasting. J Mar Syst 25:1–22 Brassington GB (2010) System design for operational ocean forecasting. In: Schiller A, Brassington GB (eds) Operational oceanography in the 21st century. Springer, Dordrecht
24â•… Defence Applications of Operational Oceanography
679
Brassington GB, Pugh T, Spillman C, Shulz E, Beggs H, Schiller A, Oke PR (2007) BLUElink— development of operational oceanography and servicing in Australia. J Res Pract Inf Technol 39(2):151–164 Griffin D (2009) Locating HMAS Sydney by back-tracking the drift of two life rafts. Bull Aust Meteorol Oceanogr Soc 22(5):138–140 Hackett B, Breivik O, Wettre C (2006) Forecasting the drift of objects and substances in the ocean. In: Chassignet EP, Verron J (eds) Ocean weather forecasting, 1st edn. Springer, Dordrecht Harding J, Rigney J (2006) Operational oceanography in the US Navy: a GODAE perspective. In: Chassignet EP, Verron J (eds) Ocean weather forecasting, 1st edn. Springer, Dordrecht Herzfeld M (2009) Improving stability of regional numerical ocean models. Ocean Dyn 59:21–46 Jacobs GA, Woodham RH, Jourdan D, Braithwaite J (2009) GODAE applications useful to navies throughout the world. Oceanography 22(3):182–189 Kalnay E (2003) Atmospheric modeling, data assimilation and predictability. Cambridge University Press, Cambridge Mackenzie KV (1981) Nine term equation for sound speed in the oceans. J Acoust Soc Am 70(3):807–812 Martin M (2010) Ocean forecasting systems—product evaluation and skill. In: Schiller A, Brassington GB (eds) Operational oceanography in the 21st century. Springer, Dordrecht Mearns DL (2009) The search for the Sydney. HarperCollins, Sydney Murphy AH (1992) Climatology, persistence and their linear combination as standards of reference in skill scores. Weather Forecast 7:692–698 Oke PR, Schiller A, Griffin DA, Brassington GB (2005) Ensemble data assimilation for an eddyresolving ocean model of the Australian region. Q J R Meterol Soc 131(613):3301–3311 Ridgway KR, Dunn JR (2003) Mesoscale structure of the mean East Australian current system and its relationship with topography. Prog Oceanogr 56:189–222 Roughan M, Middleton JH (2002) A comparison of observed upwelling mechanisms off the east coast of Australia. Cont Shelf Res 22(17):2551–2572 Urick RJ (1983) Principles of underwater sound. McGraw-Hill Book Company, USA
Chapter 25
Applications for Metocean Forecast Data— Maritime Transport, Safety and Pollution Brian King, Ben Brushett, Trevor Gilbert and Charles Lemckert
Abstract╇ This lecture outlines the recent advances in the incorporation of oceanic and atmospheric forecast datasets into specialized trajectory models. These models are used for maritime safety purposes and to aid in combating oil and chemical marine pollution events. In particular, the lecture examines in detail the system assembled by the authors for improving oil spill trajectory models (OSTM) and chemical spill trajectory models (CSTM) as part of the Australian Maritime Safety Authority’s (AMSA) role in Australia’s national plan to combat pollution of the sea by oil and other noxious and hazardous substances. The main topics of this lecture will include: • A summary of metocean forecast datasets currently being used operationally in the Australian region; • The incorporation of tidal current dynamics into ocean forecasting models; • Three case studies of utilising metocean forecast datasets in maritime trajectory models, a study of the Australian Maritime Safety Authority’s OSTM and CSTM systems (OILMAP, CHEMMAP and the Environmental Data Servers) being. − The Pacific Adventurer oil and chemical Spill, offshore Brisbane; − The Montara Well Head Platform Blowout, Timor Sea; − The towing of MSC Lugano off Esperence (WA)
25.1â•…Introduction The operational use of metocean (meteorological and oceanic) forecast datasets is necessary for the effective response to search and rescue (SAR) incidents, mitigation of pollutant spills at sea (such as oil or chemicals), and for the response to other maritime hazards (such as towing a stranded vessel to safety). To effectively model the likely drift pattern of a person lost at sea, the movement of a marine pollutant B. King () Asia-Pacific ASA, PO Box 1679, Surfers Paradise, QLD 4217, Australia e-mail:
[email protected] A. Schiller, G. B. Brassington (eds.), Operational Oceanography in the 21st Century, DOI 10.1007/978-94-007-0332-2_25, ©Â€Springer Science+Business Media B.V. 2011
681
B. King et al.
682
spill, or a stranded vessel’s movements, both wind and ocean current forecast datasets are required. Among the ocean current forecast models in use operationally in the Australian and greater Asia Pacific region are the US Navy Coastal Ocean Model (NCOM) and the Australian BLUElink model. Both of these models were developed for large to mesoscale ocean circulation, and as such neither model includes the effects of tidal currents. This lack of tidal current forcing limits the effectiveness of the models in shallow near coastal waters, where tidal currents are important and can be the dominant driving force in water circulation. Asia-Pacific ASA have developed an aggregation tool which is able to incorporate the effects of both coastal tidal currents and large scale oceanic currents, producing an effective current forecast dataset for both open ocean and coastal waters alike. There are several wind forecast models available operationally; the two used in this study were the US Global Forecast System (GFS) and the US Navy Operational Global Atmospheric Prediction System (NOGAPS). Asia-Pacific ASA has a dedicated environmental data server (EDS) called COASTMAP EDS. This server downloads, catalogues, stores and disseminates environmental and metocean forecast and hindcast datasets for use with ASA modelling software (such as SARMAP, OILMAP and CHEMMAP). Table€ 25.1 below outlines the specifics of each of the metocean forecast models operationally available for the Australian region on the EDS. The availability of several different forecast models provides an excellent opportunity to compare the various model outcomes of a particular drift scenario. If the outcomes are similar, then there is consensus between the datasets, and the modeller can be confident that the forecast is as accurate as possible. If there is a discrepancy between the forecasts, then there is no consensus, which suggests that the forecast may not be as reliable. In such a situation it is necessary for the modeller to further revise the input data based on field observations to ascertain which may be the most reliable forecast. Operational consensus forecasting has been used successfully in meteorology; however its application in oceanographic forecasting has been minimal thus far. This however is changing, and the adoption of consensus forecasting in the oceanographic community is increasing. Several case studies of the operational use of consensus forecasting are outlined in the following sections. The first
Table 25.1↜渀 Operational metocean forecast models Model Type Temporal Spatial Resolution (h) Resolution NCOM Currents 6 1/8° BLUElink Currents 24 1/10°â•›≤â•›2° GFS NOGAPS
Winds Winds
6 6
1/2° 1/2°
Spatial Extent Global Effectively (90°E–180°E, 75°S–16°N) Global Global
Update Frequency daily 2â•›×â•›weekly
Forecast Length (h) 72 144
4â•›×â•›daily 4â•›×â•›daily
180 144
25â•… Applications for Metocean Forecast Data
683
Fig. 25.1↜渀 Map showing the location of the incidents, Pacific Adventurer oil and chemical spills, Montara well head blowout, and MSC Lugano towing
relates to the Pacific Adventurer oil spill which occurred off Moreton Island, Queensland; the second was the Montara oil well blow out in the Timor Sea, and the final was the towing of the MSC Lugano off Esperance in Western Australia (see Fig.€25.1).
25.2â•…Review of Meteorological and Ocean Forecast Models 25.2.1 BLUElink Ocean Model The BLUElink project became operational in 2007 from the collaboration between the Australian Bureau of Meteorology (BoM), Royal Australian Navy (RAN) and the Commonwealth Scientific Industry Research Organisation (CSIRO 2010). Operationally, it is now under the management of the Australian Bureau of Meteorology. There are several components to the BLUElink system, including operational forecasts, reanalysis and data assimilation. The operational forecasts from BLUElink used in this study were derived from the Ocean Model Analysis and Prediction System (OMAPS-fc). This system uses the Ocean Forecasting Australia Model (OFAM) which is based on the Modular Ocean Model version 4 (MOM4) (Andreu-Burello et€al. 2010). The 3D model has a resolution of 1/10° (~10€km) in the Australian region (90°E–180°E, 75°S–16°N), with up to 2° resolution elsewhere around the globe, to reduce computational costs. There are 47 vertical layers, with the topmost 20 layers being 10€m thick. (Australian Bureau of Meteorology 2007) Data assimilation is controlled by the BLUElink Ocean Data Assimilation System (BODAS) which is an ensemble optimal interpolation (EnOI) scheme that assimilates Sea Surface Temperature (SST), Sea Surface Height (SSH) and temperature and salinity profiles. Atmospheric fluxes are currently provided by the BoM Global Atmospheric Prediction System (GASP) (Brassington et€al. 2009). The BLUElink system provides up to 144€hour forecasts of the sea surface current velocities, at 24€hour intervals.
684
B. King et al.
25.2.2 NCOM Ocean Model The Navy Coastal Ocean Model (NCOM) is a 3D global ocean current forecast model which was developed by the Naval Research Laboratory (NRL) and was transitioned to be run operationally by the Naval Oceanographic Office (NAVO). The forecast model is based on the Princeton Ocean Model (POM) and has global coverage with a horizontal resolution of 1/8°. Vertical resolution is controlled by an σ–z coordinate system with 19σ-coordinate layers in the upper 137€m (topmost surface layer thickness of 1€m) and 21 z-coordinate layers from 137€m to 5,500€m. Data assimilation is controlled by the Modular Ocean Data Assimilation System (MODAS) which assimilates temperature, salinity and SSH. Atmospheric forcing is provided by the Navy Operational Global Atmospheric Prediction System (NOGAPS) atmospheric fluxes (Barron et€al. 2007). NCOM provides a 72€hour forecast of the sea surface current velocities, at 6€hour intervals.
25.2.3 GFS Atmospheric Model The Global Forecasting System (GFS) is a global spectral numerical model operationally run by the US National Oceanic and Atmospheric Administration (NOAA). The T254 version (used in this study) provides global coverage with a horizontal resolution of 1/2° with 64 unequally spaced vertical layers. GFS model output consists of 10€m U and V wind velocities with a forecast length of up to 180 hours and a temporal resolution of 6€hours (Environmental Modelling Centre 2003).
25.2.4 NOGAPS Atmospheric Model The Navy Operational Global Atmospheric Prediction System (NOGAPS) is a spectral general circulation model (GCM) which has been under constant development at the NRL over the last 20 years. It is the principal source of atmospheric forcing for the US Navy ocean models (eg. NCOM) and short term numerical weather prediction (NWP). NOGAPS uses a one way coupling system to capture ocean–atmosphere interaction. NOGAPS has global coverage, with horizontal resolution ~1/2°. The forecast length of the NOGAPS product is 144 hours with temporal resolution of twelve hours (at 00 and 12 UTC) and updates at 06 and 18UTC to enable background forecasts, which are used in the analysis. Outputs from the model include momentum flux, both latent and sensible heat fluxes, precipitation, solar and long wave radiation and surface pressure, as well as 10 metre U and V wind velocities (Rosmond 1992; Rosmond et€al. 2002).
25â•… Applications for Metocean Forecast Data
685
25.3â•…Case Studies of the Operational Use of Meteorological and Ocean Forecast Datasets Three case studies involving the operational use of metocean datasets were investigated. Two were in response to pollutant spills, the first was the Montara well head blowout in the Timor Sea, and the second was the Pacific Adventurer oil and chemical spills off Moreton Island in Queensland whilst the third case study presented herein was the towing support of the disabled MSC Lugano off Esperance in Western Australia. The two oil spill studies demonstrate how consensus modelling has been used operationally, and show when consensus was reached, and when it was not.
25.3.1 Case Study 1—Pacific Adventurer In the early hours of the morning on the 11th of March 2009 the Pacific Adventurer encountered severe weather conditions (as a result of nearby Tropical Cyclone Hamish) whilst on route from Newcastle to Indonesia. As a result of the severe weather conditions, 31 shipping containers (containing a total of approximately 600 tonnes of ammonium nitrate) were lost overboard. Several of the containers ruptured the ship’s fuel tanks, which resulted in the loss of 270 tonnes of heavy fuel oil to the marine environment (Asia-Pacific ASA 2009). At the request of the Australian Maritime Safety Authority (AMSA), Asia-Pacific ASA provided modelling support to the response teams to determine the likely fates and possible shoreline strikes of the heavy fuel oil (HFO) and the dissolved concentrations of the ammonium nitrate in the water column.
25.3.1.1â•…Oil Spill Forecast Panels in Fig.€25.2 show the various model runs completed using OILMAP to determine the likely trajectory of the HFO. Environmental forecast data was sourced from the COASTMAP EDS. Specifically NCOM and BLUElink forecast ocean currents aggregated with tidal currents provided the current forcing, whilst GFS and NOGAPS wind forecast models provided wind forcing. To account for variability in the inputs (such as wind gusts) uncertainty particles are included in the model runs. These uncertainty particles are subjected to winds and water currents that have been varied by up to ±30% of their strength and ±30° in direction. The black dots represent the likely surface oil locations, the white dots represent the water surface swept by the oil, the light grey represents the uncertainty particles used by the model, and the red indicates the full extents of the shoreline oil stranding, as reported by Maritime Safety Queensland.
686
B. King et al.
Fig. 25.2↜渀 The four different model runs completed when forecasting the Pacific Adventurer spill. Top BLUElink plus Tides, Bottom NCOM plus Tides, Left GFS winds, Right NOGAPS winds
As shown, there is a general consensus between the model forecasts. All four model forecasts show that the shorelines on the northern end of Moreton Island and the beaches near Kawana will be impacted, with the possibility of shoreline impacts to the beaches both north and south of the Kawana Beach region. The best correlation between the model predicted shoreline impacts and observed shoreline impacts was attained by using NCOM predicted currents aggregated with tidal currents, and the GFS forecast winds (bottom left panel of Fig.€25.2). 25.3.1.2╅Chemical Spill Forecast The simulation of a mass release of the entire contents of all overboard containers was completed using the CHEMMAP software. This was indicative of a worst case scenario where all 31 of the lost containers would rupture expelling ammonium nitrate over a period of 4€hours after hitting the seabed. NCOM plus tides and GFS winds were used as the forcing data for the CHEMMAP model run. The CHEMMAP system predicted that a release of 600 tonnes of ammonium nitrate would quickly dissolve in the water column. The results are shown below in Fig.€25.3, which describes the re-projected location of the reported incident and the projected path of the simulated ammonium nitrate spill over 96€ hours. The key indicates the dissolved concentration of the chemical in the water column in milligrams per cubic meter, from the surface to depths divided into five layers. The concentrations of ammonium nitrate within the water column fell to 1€mg/L (1,000€mg/m3) within 4 days following the event. Due to the near seabed release, dissolved concentrations remain near the bottom well away from the surface where they might enter Moreton Bay.
25â•… Applications for Metocean Forecast Data
687
Fig. 25.3↜渀 Pacific Adventurer chemical spill showing concentration and location of dissolved ammonium nitrate 96€hours after release
25.3.2 Case Study 2—Montara Well Head Blowout During the morning of 21st of August 2009, well control at the Montara well head was lost. The Montara well head is located approximately 680€km west of Darwin off the Kimberly coast in Western Australia. An estimate of 400 barrels per day of crude oil was being discharged into the sea. The leak continued for 74 days discharging a total of 30,000 barrels until the well was successfully “killed” on the 3rd November 2009 (PTTEP Australasia). Asia-Pacific ASA provided modelling support throughout this incident. At the beginning there was no consensus between the forecast models, with a different direction of travel predicted for the NCOM plus tidal currents, the BLUElink plus tidal current forecast data, and the GSLA plus tidal current data. The GSLA currents are generated from mapping Gridded Sea Level Anomalies, which provide geostrophic flow estimates. This approach gives a good representation of the general circulation of the ocean, however as the produced current field uses measurements of sea level anomalies that can be up to several days old, it essentially produces a nowcast of the sea state, rather than a forecast. This can work
688
B. King et al.
well for large scale circulation which takes time to set up, and has time scales of the order of weeks to months; however GSLA currents are not able to reproduce meso to small scale circulation which have time scales of hours to days (CSIRO 2010). GSLA currents do however provide a good reference to validate forecast model (NCOM and BLUElink) performance at recreating the oceanic circulation. Two surface drifters were deployed to provide observed estimates of the currents. These revealed that the currents were tidally governed (as shown by the oscillations in the buoy trajectories). This indicates that for successful prediction of drift patterns of objects or oil in this region, the addition of the tidal component to the surface currents is vitally important. As the incident continued, the forecast datasets proved to better resolve the surface currents in the region when compared to several other drifter tracks, the location of predicted surface oil and observed surface oil, and when directly comparing the NCOM and BLUElink forecast current vectors with hindcast currents. Of the 13 weeks that the oil was tracked, approximately 10 weeks returned very good current forecast data. Each dataset (NCOM, BLUElink and GSLA) was tested against the over flight and satellite imagery to ensure the best forecasts were produced. Table€25.2 below shows the periods throughout the 92 days of the incident (from 21st August 2009 until 23rd November 2009) for which dataset was found to produce the most accurate forecast of oil movement. Forecast bulletins were produced routinely throughout the Montara event by APASA to outline the expected operational conditions, and likely whereabouts of oil. Refer to Appendix A for the reproduction of one of these forecast bulletins (for 29th October 2009).
25.3.3 Case Study 3—MSC Lugano Stranding The MSC Lugano is a 240€m container ship which was en route from Adelaide in South Australia to Fremantle in Western Australia. On the 31st of March 2008 it was disabled by an engine room fire and as a result, was in jeopardy of grounding off Esperance, Western Australia. Three tugs from nearby Esperance were called in to provide assistance, whilst another larger and better equipped tug was en route from Fremantle. The tugs took Table 25.2↜渀 Metocean forecast products used during the Montara well head blowout for oil spill forecast modelling Start End Days Wind Current 21/08/2009 30/10/2009 10 GFS GLSA+Tides 30/08/2009 27/10/2009 57 GFS BLUElink+Tides 27/10/2009 06/11/2009 10 GFS NCOM+Tides 06/11/2009 11/11/2009 5 GFS GSLA+Tides 11/11/2009 23/11/2009 12 GFS BLUElink+Tides
25â•… Applications for Metocean Forecast Data
689
Fig. 25.4↜渀 Snap shot of surface currents off Esperance Western Australia
the MSC Lugano in tow however they were not designed or equipped for deep ocean towing and ran into difficulty off Pt D’Entrecastreux whilst on a passage northward to Fremantle. The vessels were not making any headway due to very high surface current speeds and were at risk of losing the tow (Australian Transport Safety Bureau 2009). The Western Australian authorities advised the vessels to proceed further offshore into deeper water in an attempt to avoid the high current speeds and coastal hazards. However consensus ocean current forecast data (NCOM and BLUElink) indicated stronger currents offshore when compared to inshore. Upon further inspection of the forecast currents it was deemed that the tow remain closer to the shore in the more favorable current conditions. The tow was successfully completed on the 13th of April 2008. Figure€25.4 below shows a snap shot of the surface currents in the region at the time of the towing. Note the stronger southerly currents offshore of Cape Leeuwin, compared to the currents closer inshore to Cape Leeuwin.
25.4â•…Conclusions The growing view is that oceanographers should follow the best-practice methodology used by weather forecasters to take full advantage of the multiple wind and ocean forecasting datasets available. This is made particularly evident through the three case studies investigated above. Weather forecasters use all available datasets and assess each of them to develop a consensus of opinion from the various weather forecast models on what might occur. With multiple ocean forecasting datasets
690
B. King et al.
available now, the same approach can be applied, for example oil spill models rely on good forecasts of both currents and weather to accurately predict the oil’s future drift and potential impact zones. Both winds and currents are used as input data to ASA’s OILMAP and CHEMMAP spill models and have been able to successfully predict the movement of oil or chemicals over time if the forecast winds and currents have been accurate. The latest approach is to run the same spill scenario with different datasets. When consensus between forecast models is reached, the outcome gives a higher level of confidence in the spill predictions. If different forecast datasets result in disparate trajectories and outcomes, then there are multiple viable outcomes, and a low level of confidence in any one prediction. The spill forecasts can then be issued with a confidence indicator, based on the degree of consensus obtained from the multiple analyses performed. Field observations such as aircraft over flights, drifting buoys, or satellite-derived observations can all be used to help estimate errors in the forecast data. One such reason for not attaining consensus between forecast models is the location or positioning of mesoscale eddies. Mesoscale eddies have spatial extents in the order of tens of kilometers, where large scale eddies tend to have a spatial extent of greater than 100€km. As the two aforementioned global current forecast models (NCOM and BLUElink) have spatial resolutions of approximately 10€km they are essentially semi-mesoscale eddy resolving models. To adequately resolve mesoscale eddies, a resolution in the order of 5–6€km at a minimum is required. Problems arise with semi-mesoscale eddy resolving models when eddies are misplaced or even absent completely. Acknowledgements╇ This research was supported under the Australian Research Council’s Linkage Projects funding scheme LP0991159.
Appendix pill Forecast Bulletin for Montara Incident Issued Midday S 29-October-2009 for the Australian Maritime Safety Authority Over flight and satellite observations collected from the 24th–28th October 2009 have been used to update oil, oil patches and wax positions within the AMSA OILMAP Oil Spill Trajectory Model (OSTM). The recent satellite observations indicated that the slick was patches of oil/wax lying east and southeast of Montara extending to the south as patches (refer to Fig.€ 25.5). The winds have remained favourable over recent days which has seen the edge of the slick move parallel to the coast north-eastward rather than towards to coast. Using these observations, the latest wind and ocean forecast data has been incorporated to provide “search areas for oil and wax” for midday (Darwin Time) on the 30th and 31st of October 2009, as shown in Figs.€25.6 and 25.7 below. Please note that the brown dots in the figures
25â•… Applications for Metocean Forecast Data
691
Fig. 25.5↜渀 AQUA Satellite Observation at UTC0500 28th October 2009. The darker colour within the red circle is indicative of surface oil slick; the white colour within the yellow circle indicates cloud
Fig. 25.6↜渀 Forecast of surface oil (as represented by the orange spots) at 12€pm on the 30th October 2009. The surface currents are shown by the coloured arrows and the wind conditions are shown by the wind barbs
692
B. King et al.
Fig. 25.7↜渀 Forecast of surface oil (as represented by the orange spots) at 12€pm on the 31st October 2009. The surface currents are shown by the coloured arrows and the wind conditions are shown by the wind barbs
below indicate “search areas for oil and sheen”. The density of the brown dots in the figure below indicates the likelihood of finding oil or wax in the various locations around the Montara well site. Due to the containment and dispersant operations, far field predictions are typically for defining search areas for scattered weathered oil and wax patches which may no longer be visible on the water’s surface, hence this forecast is potentially a ‘worst-case’ depiction of the spill at this time. The wind conditions for Montara are expected to be north-westerly winds (4–12€ kts) for 30th October 2009, weakening from the north for 31st October, 2009. At the Montara well site, tidal oscillations are expected to be weak as we move through the neap tidal phase in the Timor Sea. The slick will generally drift southward over the forecast period. Fresh oil flows at Montara are predicted to be as follows: • 30th Oct 2009: Weak SSE flow at 9€am; Weak SSW flow at 3€pm (4–12 knot NW winds); • 31st Oct 2009: Weak SSW flow at 9€am; Weak NW flow at 3€pm (weak northerly winds); To the far north in deep waters (The Timor Trench), the Indonesian Thru Flow current continues to flow strongly WSW. This strong flow is now spinning anticlockwise current eddies along the northern shelf-break which are moving position, allowing deepwater flows to spill over the shelf and drive the slick around Montara generally southward over the forecast period.
25â•… Applications for Metocean Forecast Data
693
At Ashmore, Hibernia and Cartier Reefs, the forecast indicates that previously reported small patches of weathered wax will remain in the vicinity of Ashmore and Cartier Reefs. These patches were reported with dimensions of 50â•›×â•›50€m or less. For waters between the West Atlas rig and the Kimberly coastline, the forecast indicates that the oil patches should drift slowly southward. The southeastern most position of this part of the slick (last described as very scattered small patches of wax) will remain north of Holithuria Banks. These patches may no longer be visible on the water’s surface, and are not expected to reach any shorelines during the forecast period (APASA forecast bulletin 2009).
References Andreu-Burello I, Brassington G, Oke P, Beggs H (2010) Including a new data stream in the BLUElink Ocean Data Assimilation System. Aust Meteorol Oceanogr J 59:77–86 APASA forecast bulletin (2009, 29 October) Report provide at the request of the Australian Maritime Safety Authority duing the Montara Response Asia-Pacific ASA (2009) Independent assessment of the shoreline cleanup operations for the pacific adventurer oil spill. Asia-Pacific ASA report to Maritime Safety Queensland Australian Bureau of Meteorology (2007) BLUElink> Ocean model analysis and prediction system version 1.0 (OceanMAPSv1.0) technical specification. Canberra. Available from: http:// bom.gov.au/oceanography/forecasts/technical_specification.pdf. Accessed 5 March 2010 Australian Transport Safety Bureau (2009) Independent investigation into the engine room fire on board the Marshall Islands registered container ship MSC Lugano off Esperance Western Australia. Canberra. http://www.atsb.gov.au/media/51269/mo2008004.pdf. Accessed 19 March 2009 Barron CN, Birol Kara A, Rhodes RC, Rowley C, Smedstad LF (2007) Validation test report for the 1/8° global navy coastal ocean model nowcast/forecast system. Naval Research Laboratory, Stennis Space Centre Brassington GB, Pugh T, Oke PR, Freeman J, Andreau-Burrel I, Huang X, Warren G (2009) Operational ocean data assimilation for the BLUElink Ocean Forecasting System. Fifth WMO Symposium on the Assimilation of observations for meteorology, oceanography and hydrology, Melbourne, 5–9 Oct 2009 CSIRO (2010) Ocean surface currents and temperature news. http://www.cmar.csiro.au/remotesensing/oceancurrents/index.htm. Accessed 22 March 2010 Environmental Modelling Centre (2003) The GFS Atmospheric Model 28. http://www.emc.ncep. noaa.gov/gmb/moorthi/gam.html. Accessed 5 March 2010 PTTEP Australasia (2009) Frequently asked questions montara incident. West Perth. http://www. au.pttep.com/faq.asp#Q3. Accessed 2 Jan 2010 Rosmond TE (1992) The design and testing of the Navy Operational Global Atmospheric Prediction System. Weather Forecast 7:262–272 Rosmond TE, Tiexiera J, Peng M, Hogan T (2002) Navy operational global atmospheric prediction system (NOGAPS): forcing for ocean models. Oceanography 15(1):99–108
Chapter 26
Marine Energy: Resources, Technologies, Research and Policies John Huckerby
Abstract╇ Marine energy technologies have enjoyed a resurgence of development since the late 1990s and there are now widespread international activities to develop marine energy technologies and project deployments, principally in mid-latitude countries, where wave and tidal stream resources are more energetic. Substantial new deployments of tidal barrages, essentially comprising hydroelectric technologies driven by seawater, are under evaluation or construction in a number of countries. Technologies for extraction of heat energy from seawater by Ocean Thermal Energy Conversion (OTEC) and submarine geothermal energy are being developed more slowly, as are technologies for harnessing energy from salinity gradients and production of bio-fuels from marine biomass.
26.1â•…Introduction Marine energy technologies have enjoyed a resurgence of development since the late 1990s and there are now widespread international activities to develop marine energy technologies and project deployments, principally in mid-latitude countries, where wave and tidal stream resources are more energetic. Substantial new deployments of tidal barrages, essentially comprising hydroelectric technologies driven by seawater, are under evaluation or construction in a number of countries. Technologies for extraction of heat energy from seawater by Ocean Thermal Energy Conversion (OTEC) and submarine geothermal energy are being developed more slowly, as are technologies for harnessing energy from salinity gradients and production of bio-fuels from marine biomass. Early deployments have so far reported few environmental issues but such concerns must be comprehensively addressed. Extensive research and monitoring of deployments continues to ensure that impacts are avoided or minimized.
J. Huckerby () Power Projects Limited, Panama Street, PO Box 25456, Wellington 6146, New Zealand e-mail:
[email protected] A. Schiller, G. B. Brassington (eds.), Operational Oceanography in the 21st Century, DOI 10.1007/978-94-007-0332-2_26, ©Â€Springer Science+Business Media B.V. 2011
695
696
J. Huckerby
Governments are key investors in early-stage technology developments from research and development (R & D) and proof-of-concept developments to precommercial prototypes. They also set the regulatory framework, in which devices can be deployed and developed. This framework can contain a range of incentives designed to support maturing technologies through to commercial development. Marine energy developments are taking place in over 30 countries and the initial focus in NW European countries has broadened to become truly international. The most advanced developments in technologies, collaborative research and policies continues to occur in NW Europe. Nonetheless, at the time of writing, less than 300€MW of installed ocean energy generating capacity is presently operational (enough electricity to supply approximately 80,000 households annually). This paper begins by outlining the forms of ocean energy and the distribution of ocean energy resources. The development of ocean energy technologies is intimately linked to these resources and the operation of these technologies will have an impact on the surrounding environment. Space and resources of ocean energy will eventually need to be allocated in competition with other use of marine space and resources. To compete with existing energy generation technologies, ocean energy technologies—like other new generation technologies—will need a favourable political framework, through R & D grants, capital support, tariffs and allocation regimes, to promote and accelerate their contribution to global energy supply portfolios.
26.2â•…Forms of Ocean Energy For the purposes of this paper, ocean energy resources are defined as those energy resources, which use seawater as either the motive power or for its chemical or heat potential. There are at least six principal forms of ocean energy, which could be harnessed to produce electricity or other products. These forms are: 1. Wave Energy 2. Tidal Energy a. Tidal Rise and Fall b. Tidal Streams 3. Ocean Current Energy 4. Ocean Thermal Energy a. Ocean Thermal Energy Conversion (OTEC) b. Submarine Geothermal Energy 5. Salinity Gradient 6. Marine Biomass Some authors consider offshore wind energy as a form of ocean energy but it is derived from the movement of winds, rather than the kinetic movement of seawater. Offshore wind energy is thus not really a form of ocean energy and is not considered further here.
26â•… Marine Energy: Resources, Technologies, Research and Policies
697
The remaining seven forms of ocean energy have been under investigation for over 100 years and, with the exception of adaptation of hydro-electric dam technology for tidal barrages, are still relatively underdeveloped. Nonetheless, these disparate forms of energy are globally distributed and may offer significant opportunities to supplement or displace existing generation sources, particularly as costs for fossil fuel energy sources rise. The products of these forms of energy can be used for a number of different purposes: 1. Generation of electricity (AC and DC) 2. Production of pressurized and potable water 3. Production of heat 4. Production of hydrogen 5. Production of bio-fuels A number of ocean energy technologies are being developed to produce potable water, either directly or via generation of electricity to drive desalination plants (Jalihal and Kathiroli 2009). Ocean Thermal Energy Conversion (OTEC) is also being developed for use as seawater air conditioning, ‘district cooling’ and seawater enrichment of onshore mariculture operations (Nihous 2009).
26.3â•…Ocean Energy Resources Potential energy resources available from the ocean significantly exceed worldwide demand for energy but they are not presently accessible, either technically or as economically competitive alternatives to current low cost energy sources—coal, oil, gas and geothermal energy. Increasing the contribution of ocean energy to meet growing international energy demand will require substantial investment in research and development, demonstrations, deployments and diffusion of commercial technologies. Nonetheless, all forms of ocean energy are emissions-free (barring construction, deployment and decommissioning activities) and will become cheaper and more attractive alternatives to the presently predominant forms of fossil fuel as emissions trading, carbon taxes and the full cost of externalities are priced into the cost of fossil fuel sources. The limitations on the extent to which ocean energy will be developed are factors such as unit cost of generated electricity (compared with other renewable technologies), reliability and operations and maintenance costs.
26.3.1 Wave Energy Wave energy is present across the globe and can be harnessed as a combination of kinetic and potential energy of water particles. Waves are created by the action of winds passing over the surface of the ocean. Wave heights (and thus energy) are greatest in the sub-equatorial regions where the trade winds (such as the ‘Roaring Forties’) are strong and blow consistently in the same direction over long distances (Fig.€26.1).
Fig. 26.1↜渀 Global distribution of annual mean wave power. (Cornett 2008)
698 J. Huckerby
26â•… Marine Energy: Resources, Technologies, Research and Policies
699
In addition to the geographic variability indicated by Fig.€26.1, there are also seasonal and shorter-term variabilities in wave regimes, brought about by weather systems. In the higher latitude areas, background wave regimes may be sufficient to permit almost continuous generation. Extreme wave conditions, brought about by storms, may provide greater energy than average conditions but this energy may not be extractable if wave energy devices have to go into ‘survival’ mode. Nonetheless, waves are essentially integrated wind energy, which are thus more predictable than winds. Wave energy farms may produce more forecastable energy, thus enabling project developers to secure higher prices for their produced power. There are a number of computational wind wave models, which are used for wave forecasting (Greenslade and Tolman 2010).
26.3.2 Tidal Energy Tidal energy can be divided into two distinct forms: 1. Tidal rise and fall 2. Tidal streams or currents 26.3.2.1â•…Tidal Rise and Fall Tidal rise and fall energy is potential energy derived by height changes in sea level, caused by the gravitational attraction of the moon, the sun, and to a lesser extent other astronomical bodies, on oceanic water bodies. The effects of these tides are complex and most major oceans and seas have internal tidal systems, called amphidromic systems (Fig.€26.2). Each major ocean has its own internal circulation system, called ‘gyres’, which rotate anticlockwise in the northern hemisphere and clockwise in the southern hemisphere. There are a number of different tidal components, which operate at different phases (called Kelvin waves), the largest being the M2 component, due to the moon. Tidal range energy can be best harnessed nearshore, particularly in estuaries, where tidal rise and fall can be amplified as coastal waters shallow towards the coast. 26.3.2.2â•…Tidal Stream Energy The movement of ocean water volumes, caused by the changing tides, creates tidal stream energy. Kinetic energy can be harnessed usually near shore or particularly where there are constrictions, such as straits, islands and passes. Tidal stream energy results from local regular (diurnal at roughly 24€h periods and semi-diurnal at 12€h 25€min periods) flows caused by the tidal cycle. Spring tides occur when the gravitational attraction of the sun and moon act in the same
700
J. Huckerby
Fig. 26.2↜渀 Global distribution of tidal amplitude, i.e., M2 tidal constituent. (Ray 2007)
direction, whilst neap tides result when the attraction of the sun and moon operate in opposite directions. These tides are forecastable on an 18.5-year cycle. These tides cause kinetic movements, which can be accelerated near coasts, where there is constraining topography, such as straits between islands (Soerensen and Weinstein 2008).
26.3.3 Ocean Currents Open ocean current systems are driven by the latitudinal distribution of the winds and have a clockwise circulation in the northern hemisphere and a counter-clockwise circulation in the southern hemisphere. Such wind-driven currents operate at shallow depths (1,000€m) seawater to drive heat exchange processes (Nihous 2009). In most oceans or seas there is a marked drop in temperature between the surface and deeper water. This difference is most marked in tropical regions (between the Tropics of Capricorn and Cancer), where surface temperatures can be up to 20°C higher than underlying waters. Ocean thermal anomalies occur on a (Fig. 26.4) seasonal basis but there is increasing evidence of decadal anomalies in most oceans (Alves et€al. 2010). Whilst the seasonal variations are reasonably forecastable, longer-term variability is presently not fully understood, so forecasting is more uncertain. 26.3.4.2â•…Submarine Geothermal Energy In the 1980s ‘black smokers’ were discovered at mid-ocean ridges in the Atlantic Ocean. Black smokers are submarine geothermal systems, circulating seawater through hot, fractured volcanic rock, which is being formed and exposed at these mid-ocean ridges, where the earth’s ocean floors are expanding (Nihous 2010; also
Fig. 26.4↜渀 Worldwide average ocean temperature differences between 20 and 1,000 m water depths (colour palette is from 15º – mauve – to 25º C – red)
702 J. Huckerby
26â•… Marine Energy: Resources, Technologies, Research and Policies
703
Fig. 26.5↜渀 Volcanic activity at mid-ocean spreading ridges. (Gaba 2009)
see Alcocer and Hiriart 2008). The circulated seawater emerges from the volcanic rocks containing a range of minerals, which may contain gold, silver, copper, lead, zinc and other precious and rare metals. They also contain substantial quantities of heat, often reaching the seafloor at temperatures in excess of 350°C. It is this heat energy and the unusual mix of minerals that lead to the spectacular and unusual faunal assemblages, which are common at black smoker sites. Since the first discoveries, most mid-ocean ridges have been found to contain black smoker sites, some at remarkably shallow water depths. Some mid-ocean ridges come close to shore as, for example, does the Tonga-Kermadec Arc to the north coast of the North Island of New Zealand or the spreading ridge at the northern edge of the Gulf of California (Fig.€26.5). Like onshore geothermal energy, submarine geothermal energy should be forecastable and produce baseload electricity. Although individual submarine geothermal vents are ephemeral, regional production of geothermal fluids is usually predictable and forecastable.
26.3.5 Salinity Gradients Seawater is approximately 200 times more saline than fresh river water, derived from rain, snowmelt or groundwater and is delivered to the coast by major rivers. Global salinity differences arise from submarine and surface current movements (Fig.€26.6). The relatively high level of salinity in seawater thus establishes a pressure potential with sweet river water, which can be used to generate electricity or derive fresh (drinking) water from the seawater. This ‘osmotic’ pressure differential—equivalent to a hydraulic head over 120€ m—can be harnessed and used to drive a conventional Pelton Wheel turbine to generate electricity. It can also be used as a chemical potential to generate electricity directly.
704
J. Huckerby
Fig. 26.6↜渀 Average global sea surface salinity. (NASA 2009)
Whilst freshwater salt content is unlikely to be seasonably variable, the same is not true for ocean salinity. There is strong evidence of seasonal variability of salinity, which may also be related to decadal variations, such as the El Nino Southern Oscillation (Alves et€al. 2010). However, the impact of these seasonal and decadal fluctuations may be relatively small on potential power production. In any event the technologies are not sufficiently developed to confirm that salinity variation will lead to seasonal variations in energy production.
26.3.6 Marine Biomass The oceans are the largest source of biomass on earth. Man utilizes relatively little of this biomass, although overfishing of some species has rendered them endangered. Onshore biomass is principally used for the production of biofuels, although there are increasing concerns about the planting of biofuel crops to displace food crops.
26.4â•…Ocean Energy Technologies The range of ocean energy technologies is huge and varied for a number of reasons: 1. There are a number of different forms of ocean energy 2. There are many different ways to extract energy from seawater 3. Ocean energy technologies are at an early stage of development and a wide range of experimentation is continuing. 4. No dominant technologies have yet emerged
26â•… Marine Energy: Resources, Technologies, Research and Policies
705
Because there is a wide range of options for energy extraction and no dominant technologies, it is unlikely that ocean energy technologies will converge on a single device type, equivalent to the monopole tower, horizontal axis wind turbine generator with an upwind three-bladed rotor, which characterize the majority of wind turbines. Seawater is approximately 830 times denser than air at sea level. Consequently devices which seek to extract potential or kinetic energy from seawater movements are likely to be substantially smaller and more robust than wind turbines. The forces exerted by seawater are much greater than forces exerted by the wind. Presently the only nominally commercial ocean energy technology is the tidal barrage, which is effectively an existing technology—a hydroelectric plant—which utilizes the tidal range in river mouths, estuaries or embayments to generate a hydraulic head during either or both the ebb and flood tides, which can be used to generate electricity. All other ocean energy technologies have, at best, reached the pre-commercial demonstration phase and have yet to become commercial. However, considerable investment and research effort is being expended worldwide and new technologies are approaching commercial deployment, particularly wave and tidal stream technologies.
26.4.1 Classification of Ocean Energy Conversion Technologies There are a number of classification schemes for ocean energy conversion technologies. A primary classification can be made based upon the basic energy resource being harnessed: 1. Potential and kinetic energy in waves and currents 2. Chemical potential of seawater (salinity gradients) 3. Heat potential of seawater (ocean heat and geothermal heat) 4. Biological potential of seawater 26.4.1.1â•…Wave, Tidal and Ocean Current Technologies These technologies effectively utilize the potential energy (derived from wave height or tidal height differences) and kinetic energy (derived from water movement). Device technologies have four key features: 1. A stable platform or surface 2. A mobile working surface for the wave or current to work against 3. The mobile working surface must, at least partially, resist the wave or current action 4. The mobile working surface must be connected to some power take-off. Classification of wave energy devices can be made by consideration of the following characteristics: principle of operation, device location and mode of operation (Fig.€26.7; Falcão 2009). Names in bold refer to specific examples of devices in each class.
706
2VFLOODWLQJZDWHU FROXPQ ZLWKDLUWXUELQH
J. Huckerby
)L[HGVWUXFWXUH
,QEUHDNZDWHU6DNDWD0XWULNX )ORDWLQJ0LJKW\:KDOH2FHDQ(QHUJ\6SHUER\2FHDQOLQ[
)ORDWLQJ 2VFLOODWLQJERGLHV ZLWKK\GUDXOLFPRWRU K\GUDXOLFWXUELQHOLQHDU HOHFWULFDOJHQHUDWRU
(VVHQWLDOO\WUDQVODWLRQKHDYH $TXD%XR\ ,36%XR\)2:DYHERE3RZHU%XR\ (VVHQWLDOO\URWDWLRQ3HODPLV36)URJ6($5(9 (VVHQWLDOO\WUDQVODWLRQKHDYH $:6
6XEPHUJHG
2YHUWRSSLQJ ZLWKORZKHDG K\GUDXOLFWXUELQH
,VRODWHGB3LFR/,03(7
)L[HGVWUXFWXUH
5RWDWLRQERWWRPKLQJHG :DYH5ROOHU2\VWHU 6KRUHOLQHZLWKFRQFHQWUDWLRQ 7$3&+$1 ,QEUHDNZDWHUZLWKRXWFRQFHQWUDWLRQ 66*
)ORDWLQJVWUXFWXUHZLWKFRQFHQWUDWLRQ :DYH'UDJRQ
Fig. 26.7↜渀 Classification of wave energy devices. (Falcão 2009)
There are abundant publications with pictures of wave, tidal and other water current devices, almost all of which are conceptual and a few undergoing full-scale openocean deployments. Rather than duplicate these papers with pictures of devices, the reader is directed to the publications of the Executive Committee of the Ocean Energy Systems Implementing Agreement (OES-IA) and particularly the 2008 Annual Report (Brito-Melo and Bhuyan 2009). Further information on the range of wave, tidal and water current technologies can be found in the “Marine and Hydrokinetic Technology Database” of the United States Department of Energy (US DoE 2008). 26.4.1.2â•…Chemical Potential of Seawater Seawater has a higher salinity than all river water debouching into oceans. The opportunity to use this chemical potential to generate electricity was recognized in the nineteenth century but commercial technologies are still some way off. Nonetheless any major river entering the sea offers the potential for future deployment of salinity gradient technologies. There are two ways to extract energy from the salinity differences between river water and seawater: 1. Osmosis—the process is called Pressure Retarded Osmosis (PRO) 2. Reversed Electro-Dialysis (RED) PRO, sometimes called “osmotic power” exploits the chemical potential (i.e., salt concentration) between fresh water and seawater as pressure. Loeb developed
26â•… Marine Energy: Resources, Technologies, Research and Policies
707
Fig. 26.8↜渀 Operational principles of a PRO power plant. (Skråmestø and Skilhagen 2009)
the concept in the 1970s (Loeb and Norman 1975). Seawater and fresh water are brought together across semi-permeable membranes. The resultant pressure is in the range 24–26€bar, depending on the salt concentration in the seawater (Fig.€26.8). Filtering of both the seawater and fresh water are critical, as impurities easily reduce the efficiency of membranes. The world’s first pilot plant for PRO became operational at Tofte, Oslo Fjord, in SW Norway in October 2009. The plant, built and operated by Statkraft, combines river water and water from the fjord to produce up to 4€kW of electricity. Reverse electro-dialysis is a process, which utilizes chemical potential differences between two solutions, in this case seawater and fresh water brought into contact through an alternative series of anion and cation exchange membranes. The chemical potential generates a voltage over each membrane. This concept is being developed in a first prototype by Dutch researchers (Groeman and van den Ende 2007). 26.4.1.3â•…Heat Potential of Seawater The heat potential of seawater was recognized in the 1970s and is available in two forms:
708
J. Huckerby
1. Ocean Thermal Energy Conversion (OTEC) 2. Submarine Geothermal Energy OTEC technologies were first developed in the United States in the 1970s but languished after the oil price rises in the 1980s. OTEC takes deep ocean water, which tends to be at a steady temperature of c. 4°C and combines it—in a heat exchange process—with shallow surface water. The key component of the technology is the ‘cold water pipe’, usually a large-diameter (>1€m) plastic pipe, extending down for 1€km up which the deep cold ocean water is brought to the surface. Once at the surface, an open- or closed-cycle heat exchange process extracts heat energy, using a secondary fluid, such as ammonia (with a low boiling point) as the exchange fluid and converts it into mechanical energy (Fig.€26.9). Submarine geothermal energy could potentially be harnessed at those mid-ocean ridges, which are close to the surface and close to shore. Proposed technologies would be submarine heat exchange devices, which generate electricity on the seabed (Fig.€26.10; Hiriart 2008, personal communication). There are proposals to generate potentially drinking water on site to utilize its buoyancy relative to seawater to deliver the drinking water to a surface location.
26.4.1.4â•…Biological Production Various attempts have been made to develop technologies to harvest biomass from the sea for the production of biogas and biofuels (Brehany 1983). In the 1970s research in the United States focused on the harvesting of kelp but this languished in the 1980s as oil prices declined. More recently interest has shifted to the potential for open-ocean harvesting of marine algae for biofuels. The marine algae would essentially be ‘farmed’ by the chemical fertilization to enhance marine algae growth and concentration. At present there are no technologies capable of concentrating dispersed marine algae from their very low natural levels in the open ocean.
26.4.2 Predictability of Ocean Energy A key factor in the uptake of ocean energy will be predictability of produced energy (or water), as this will affect grid connections and the market price for electricity sold into local markets. Ocean currents, osmotic power, OTEC and submarine geothermal energy could potentially produce continuous, i.e., baseload, electricity, whilst tidal currents are forecastable for periods of days (with some modification due to weather). Even wave energy can be predicted 1–2 days in advance. All forms of ocean energy are less variable than wind energy.
26â•… Marine Energy: Resources, Technologies, Research and Policies
709
:$50 :$7(5 ,17$.( (;3$1',1*6(&21'$5< 9$325
&2/' :$7(5 287/(7
+($7(;&+$1*(56
785%,1(
(9$325$725
&21'(16(5
3803
6(&21'$514€knots) of considerable size (>80€m). Fast-rotating propellers, which are actively putting energy into the water, are problematic. Most ocean energy devices are at fixed locations (allowing for diurnal tidal movements), do not have fast-moving parts and are relatively small. Careful siting of ocean energy device farms outside known migration routes should minimize the potential for collision.
26.5.3.3â•…Noise and Vibration Noise generated by ocean energy devices is likely to be limited and potentially not much above ambient noise. Rotating turbines may cause low-frequency noise, particularly if blade tips reach speeds fast enough to cause cavitation i.e., creation
26â•… Marine Energy: Resources, Technologies, Research and Policies
713
and explosive collapse of air bubbles at blade tips. Vibration caused by rotating machinery is likely to be limited.
26.5.3.4â•…Electromagnetic Fields Several marine species, including sharks and rays, use weak magnetic and electrical fields for navigation and prey location. Electro-sensitive species may be attracted to or repelled by these fields. Devices, such as offshore seismic survey cables, which emanate electrical and magnetic fields, can be targeted by some species, in the belief that they are prey. Appropriate cable selection and shielding technologies can mitigate the effects of these fields.
26.5.3.5â•…Summary The maritime and offshore oil and gas industries have long experience with locating fixed or moving structures in the marine environment. Ocean energy converters are somewhat different in that they are essentially long-term static installations, some of which have passively moving rotors and blades. Their likely effects on benthic and pelagic species need substantial research and experience from deployment. Extensive monitoring by early deployment projects, e.g., the tidal current projects in the East River of New York (Verdant Power 2009) and Strangford Lough, Northern Ireland (MCT 2009) have so far shown that interactions and effects between devices and native marine fauna are limited and not threatening.
26.6â•…Space and Resource Allocation Marine energy generation is a new use for sea space. The gradual development of commercial technologies to utilize the various properties of seawater, described in previous sections, will create both a requirement for occupation of sea space and a valuation of this activity against existing and other new uses. Many countries have space and resource allocation regimes for other uses, such as oil and gas exploration. The legislative frameworks for each resource may be quite different, particularly where the occupation is nominally permanent and irreversible (e.g., creation of a tidal barrage). The legislative requirements for each resource, whether it be creation of shipping lanes, nomination of marine reserves, fishing quotas or award of oil and gas exploration permits, are quite different. Indeed they are usually customized to the particular resource and may be modified to meet with international best practice. Marine energy best practice is still under development and analogies to related industries, such as offshore oil and gas or shipping may not be directly applicable.
714
J. Huckerby
Some countries are therefore developing new legislative and regulatory practices to acknowledge the particular, if not unique, qualities of marine energy. The valuation of space for marine energy projects, as compared to other potential or actual uses of sea space, is in its infancy. So far competitive issues have not been too great but this probably reflects the relatively few deployments of marine energy converters that have taken place. Most projects have had the luxury of being granted space or access to marine energy resources on a ‘first come, first served’ basis. This will become less satisfactory as competition for space or particular resources becomes an issue. The United Kingdom’s Marine Estate has already held its first competitive bid round for marine energy permits in the Pentland Firth of NE Scotland (Crown Estate 2009). The United States has also put in place a permitting scheme administered by both the Minerals Management Service (MMS) and FERC, the Federal Energy Regulatory Commission (FERC 2009). Regulatory authorities have yet to establish a single valuation methodology for marine energy. Perhaps the most attractive approach will be a ‘best use solution’. This approach has been proposed for the management of onshore fresh water resources in New Zealand (NZBCSD 2008). The ‘best use solution’ uses a mixed statutory planning and market-based approach. The objective is to manage water (or potentially sea space and resources) in an integrated and sustainable way, taking into account all potential uses and users. Some countries have begun to use an integrated approach to allocation of offshore space and resources, called ‘Marine Spatial Planning’, which is defined as a process of allocating the spatial and temporal distribution of human activities in marine areas to achieve a range of environmental, economic and social objectives.
26.7â•…Political Framework for Ocean Energy Governments have a number of policy options for the promotion and acceleration and uptake of ocean energy. Although investment in ocean energy technology development is undertaken by a spectrum of organizations—from small-medium enterprises with a conceptual idea to major international utilities or energy companies with seed investments—it is government support that is driving the development of ocean energy. Recent major changes in international policy are favourable for ocean energy. The current proposals to replace the Kyoto Protocol with another binding treaty to include greater involvement of developing nations is driving national governments to consider renewable energy and energy efficiency initiatives. The development of emissions trading regimes, emissions reduction targets and potential implementation of carbon taxes are ‘levelling the playing field’ between conventional fossil fuel use and new renewables. These global initiatives are supportive but marine energy will only flourish in countries, where dedicated marine energy policy instruments are used to promote its uptake. Such dedicated instruments have been implemented in NW European countries to promote solar PV installations (in Germany and Spain) and wind energy.
26â•… Marine Energy: Resources, Technologies, Research and Policies
715
The same NW European countries (Scotland, United Kingdom, Ireland, Portugal and Spain) are leading the way with respect to marine energy. The key policy instruments that are influential in promoting marine energy are: 1. Legislated or aspirational targets for installed capacity of generation contribution from ocean energy 2. Government funding (from R & D to production incentives) 3. Infrastructure developments 4. Other incentives (Table€26.1)
26.7.1 Lifecycle Incentives for Ocean Energy No country offers all of the incentives outlined in Table€26.1, although Scotland and the United Kingdom come close (Scotland has its own set of incentives, separate from the rest of the UK). The governments in these countries have recognized the potential for marine energy as an energy supply option and potential export opportunity. Additionally they have understood the need to provide a development path for an industry, drawn from other disparate industries, which require an integrated set of policy incentives to promote involvement throughout the supply chain. The policies should also continue to provide incentives, as the industry matures. The introduction of production incentives in Scotland and the United Kingdom demonstrates in the increasing maturity of wave and tidal stream technologies being developed there. Similarly the international spread of marine energy testing centres and participation in standards development (see next section) indicate the development of an international industry.
26.7.2 International Initiatives for Ocean Energy At least 30 countries have active developments in marine energy, ranging from individual inventors pursuing their own concepts by prototype modeling to major government initiatives to develop multi-MW tidal barrages (e.g., in the United Kingdom and Russia). Whilst the NW European coastal countries have led developments since the 1970s, activities have spread around the world and some of the largest developments are planned in the N.W. Pacific, e.g., the 254€MW Sihwa tidal barrage, which will become operational in Korea in June 2010.
26.7.3 International Initiatives There are a number of regional and international initiatives for promotion and development of ocean energy.
716
J. Huckerby
Table 26.1↜渀 Government policy instruments for ocean energy
26.7.3.1â•…Ocean Energy Systems Implementing Agreement (OES-IA) The OES-IA is an inter-governmental initiative under the auspices of the International Energy Agency (IEA) in Paris. It presently has 16 member governments, who send representatives to the Executive Committee (ExCo). Australia joined the OES-IA in 2009 and Korea, South Africa and France are due to join early in 2010.
26â•… Marine Energy: Resources, Technologies, Research and Policies
717
The OES-IA ExCo meets twice a year to lead work programs, which will promote and accelerate the uptake of ocean energy. The Committee commissions Annexes, which are separate optional work programs, in which national governments can choose to participate, on specific issues. Presently, there are Annexes on: 1. Open-sea testing protocols 2. Grid connection of marine energy converters 3. Environmental impacts of marine energy converters The OES-IA publishes newsletters and annual reports as well as technical reports based the Annex work programs. These are all publicly available at http://www. iea-oceans.org.
26.7.3.2â•…IEC’s Technical Committee 114 The International Electrotechnical Commission, based in Geneva, has set standards for electrical, electronic and electromechanical equipment for over 100 years (http://www.iec.ch). In 2007 it decided to establish a Technical Committee (TC114) to establish standards for wave, tidal and other water current energy converters. TC114 currently comprises representatives for 16 country governments and is developing technical specifications, the precursors for standards, on the following subjects: 1. Marine energy terminology 2. Wave Device performance 3. Tidal stream device performance 4. Design criteria for marine energy converters 5. Wave and tidal energy resource characterization and assessment The first of these technical specifications is likely to be published in mid-late 2012.
26.7.4 European Initiatives 26.7.4.1â•…EquiMar and Predecessors The Equitable Testing and Evaluation of Marine Energy Extraction Devices in terms of Performance, Cost and Environmental Impact (EquMar) is a European Commission-funded consortium program with 22 partners, ranging from device developers to university researchers (http://www.equimar.org). The program is led by the University of Edinburgh. The purpose of EquiMar is to deliver a series of highlevel and detailed protocols for the equitable evaluation of marine energy converters. The project was commissioned in April 2008 and draft protocols are presently available. The project will run for three years and is on track to deliver final outputs by April 2011.
718
J. Huckerby
EquiMar follows early European Commission-funded research projects, like the Co-ordinated Action on Ocean Energy (CA-OE; http://www.ca-oe.net/home. htm) and WAVETRAIN, an initiative to train postgraduate students in ocean energy (http://www.wavetrain2.eu). 26.7.4.2â•…Waveplam The WAVe Energy PLAnning and Marketing project (Waveplam) is another European Commission-funded consortium program with eight partners developing tools, methods and standards to speed up the introduction of ocean energy into the renewable energy market (http://www.waveplam.eu). The project consortium includes European research organizers and device developers, who aim to address non-technological barriers to the establishment of ocean energy.
26.8â•…Trends and Growth in Ocean Energy The year 2008 was an important one for ocean energy. The world’s first ‘pre-commercial tidal demonstrator’, the Marine Current Turbines’ SeaGen tidal generator, began to feed electricity into the Northern Ireland Grid (Fig.€26.11a). Shortly afterwards, the world’s first wave farm array (of three Pelamis devices) became operational at Aguçadoura in northern Portugal (Fig.€26.11b).
Fig. 26.11↜渀 Recent marine energy deployments. a MCT’s SeaGen pre-commercial tidal demonstrator (Source: http://www.marineturbines.com/21/technology/), and b Pelamis Wavepower’s 3â•›×â•›750€ kW Pelamis array at Aguçadoura (Source: http://www.pelamiswave.com/content. php?id=149)
26â•… Marine Energy: Resources, Technologies, Research and Policies
719
There have been fewer deployments in 2009, perhaps the most notable being the deployment of the Aquamarine Oyster surge device at the European Marine Energy Centre in the Orkney Islands. A number of major energy companies (Total, Chevron) and utilites (RWE, Statkraft, Vattenfall and Fortum) have invested in ocean energy device or project developments and the venture capital community has remained involved. Statkraft, the Norwegian transmission system operator and generator, opened the world’s first prototype osmotic power plant. The US Department of Energy continued its investment in R & D projects. That funding covers a range of projects and, in 2009, includes funding for accelerated market developments. Some of the 2009 funding is dedicated to a rejuvenation of research into ocean thermal energy conversion. Other governments continue to support R & D projects and device developments, with a growing focus on providing energy for desalination or the direct production of drinking water from ocean energy. The Scottish Executive offered the first prize for ocean energy, called the Saltire Prize (GBP 10€million), to be awarded to the first commercially viable wave or tidal stream technology to generate more than 100€GWh of electricity over a continuous 2-year period. Undoubtedly, investments and developments in ocean energy have been affected by the world’s economic situation since the middle of 2007. As the world’s economies recover during 2010, activities deferred during 2008 should be resurgent. The growing numbers of device developments and international testing centres should lead to an acceleration and maturation of technology development to the first commercial devices. For the nascent technologies, such as OTEC and osmotic power, recent R & D and prototype investments should lead to more concrete developments in coming years. Lastly, a number of countries and organizations have proposed targets for installed generation capacity from ocean energy. Forecasts made in the early 2000s (e.g., Scottish Executive 2004) have proven too optimistic but ocean energy capacity is now growing. Presently, the total capacity—from all forms of ocean energy—is relatively small (c. 300€MW) with the largest contribution coming from the 240€MW La Rance Tidal Barrage in northern France. However, this total will almost double in 2011, when the 254€MW Sihwa barrage in Korea comes on stream.
References Alcocer SM, Hiriart G (2008) An applied research program on water desalination with renewable energies. Am J Environ Sci 4(3):190–197 Alves O, Hudson D, Balmaseda M, Shi L (2010) Seasonal and decadal prediction. In: Schiller A, Brassington GB (eds) Operational oceanography in the 21st century. Springer, Dordrecht Brehany JJ (1983) Economic and systems assessment of the concept of nearshore kelp farming for methane production. Parsons Co. and Gas Research Institute, Technical Report PB-82-222158 Brito-Melo A, Bhuyan G (eds) (2009) 2008 Annual Report of the International Energy Agency Implementing Agreement on Ocean Energy Systems (IEA-OES), February 2009 Charlier RH, Justus JR (1993) Ocean energies: environmental, economic and technological aspects of alternative power sources. Elsevier Oceanography Series. Elsevier, Amsterdam
720
J. Huckerby
Cornett AM (2008) A global wave energy resource assessment. Annual international offshore and polar engineering conference, Vancouver, BC, ISOPE-2008-579 Crown Estate (2009) Details of Pentland Firth BIDS Announced. Crown Estate press release, 8 June 2009. http://www.thecrownestate.co.uk/newscontent/92-pentland-firth-tidal-energyproject-3.htm. Accessed 16 Dec 2009 Dombrowsky E, Bertino L, Brassington GB, Chassignet EP, Davidson F, Hurlburt HE, Kamachi M, Lee T, Martin MJ, Mei S, Tonani M (2009) GODAE systems in operation. Oceanography 22(3):80–95 Falcão AF deO (2009) The development of wave energy utilization. In: Brito-Melo A, Bhuyan G (eds) 2008 Annual report of the Ocean Energy Systems implementing agreement. Lisbon, February 2009 FERC (2009) MMS/FERC guidance on regulation of hydrokinetic energy projects on the OCS. Federal Energy Regulatory Commission, 24 April 2009. http://www.ferc.gov/industries/hydropower/indus-act/hydrokinetics/pdf/mms080309.pdf. Accessed 16 Dec 2009 Gaba E (2009) World map in English showing the divergent plate boundaries (OSR—Oceanic Spreading Ridges) and recent sub aerial volcanoes. http://en.wikipedia.org/wiki/ File:Spreading_ridges_volcanoes_map-en.svg. Accessed 15 Dec 2009 Greenslade D, Tolman H (2010) Surface waves. In: Schiller A, Brassington GB (eds) Operational oceanography in the 21st century. Springer, Dordrecht Groeman F, van den Ende K (2007) Blue energy. Leonardo Energy. www.leonardo-energy.org Hiriart G (2008) Hydrothermal vents. PowerPoint presentation, November 2008 Jalihal P, Kathiroli S (2009) Utilization of ocean energy for producing drinking water. In: BritoMelo A, Bhuyan G (eds) 2008 Annual report of the Ocean Energy Systems implementing agreement. Lisbon, February 2009 Langhamer O (2007) Colonization of wave power device foundations by invertebrates. In: National Renewable Energy Laboratory and Natural Resources Canada (eds) IEA-OES workshop: potential environmental impacts of ocean energy devices: meeting summary report. Messina, Italy, 18 Oct 2007 Loeb S, Norman RS (1975) Osmotic power plants. Science 189:654–655 MCT (2009) www.marinecurrentturbines.com NASA (2009) Map of average global sea surface salinity. http://aquarius.nasa.gov/educationsalinity.html. Accessed 16 Dec 2009 Nihous G (2009) Ocean thermal energy conversion (OTEC) an derivative technologies: status of development and prospects. In: Brito-Melo A, Bhuyan G (eds) 2008 Annual report of the Ocean Energy Systems implementing agreement. Lisbon, February 2009 Nihous GC (2010) Mapping available Ocean Thermal Energy Conversion Resources around the main Hawaiian Island with state-of-the-art tools J Renew Sustain Energy 2, 043104 NOAA (2008) Map of major surface ocean currents. http://www.adp.noaa.gov/currents.jpg. Accessed 15 Dec 2009 NZBCSD (2008) A best use solution for New Zealand’s water problems. New Zealand Business Council for Sustainable Development, Auckland, August 2008 Ray R (2007) Scientific visualization studio, and television production NASA-TV/GSFC, NASAGSFC, NASA-JPL. http://en.wikipedia.org/wiki/Amphidromic_point. Accessed 25 Nov 2009 Scottish Executive (2004) Harnessing Scotland’s marine energy potential: Marine Energy Group (MEG) report 2004. Report by the Forum for Renewable Energy Development in Scotland Skråmestø OS and Skilhagen SE (2009) Status of technologies of harnessing salinity power and the current osmotic power activities. In: Brito-Melo A, Bhuyan G (eds) 2008 annual report of the Ocean Energy Systems implementing agreement. Lisbon, February 2009 Soerensen HC, Weinstein A (2008) Ocean energy: position paper for IPCC. Key note paper for the IPCC scoping conference on renewable energy. Lubeck, Germany. http://www.eu-oea.com/ euoea/files/ccLibraryFiles/Filename/000000000400/OceanEnergyIPCCfinal.pdf US DoE (2008) Marine and hydrokinetic technology database. http://www1.eere.energy.gov/ windandhydro/hydrokinetic/default.aspx. Accessed 15 Dec 2009 Verdant Power (2009) www.verdantpower.com
Chapter 27
Application of Ocean Observations & Analysis: The CETO Wave Energy Project Laurence D. Mann
Abstract╇ The latest full-scale version of the CETO wave energy converter (WEC) is described, along with its principle of operation, key features and site selection. At the time of writing, a full-scale prototype test site was under development at a coastal site approximately 37€km to the south west. Some pragmatic issues pertaining to the use of global wave model data and in-situ observations are discussed in the context of this commercial venture.
27.1â•…Hardware Overview The CETO wave energy converter is shown schematically in Fig.€27.1. Submerged buoys are connected to pumps that are tethered to the seabed in an array. As a wave disturbance passes overhead the buoys are heaved upwards and exert tension on the tethers forcing the pistons inside the pumps to move upwards-expelling fluid at high pressure. The high-pressure fluid, usually water, is piped into a manifold where it moves to shore. The pressurised water may be used to drive a turbine directly for electricity production or for production of desalinated water, or a combination of both. CETO thus distinguishes itself from other WEC’s in that the output of the offshore plant is not electricity but rather pressurised fluid. Energy conversion from hydraulic to electric takes place onshore with standard off-the-shelf plant- Pelton, or similar high-head turbines coupled to electric generators. CETO wave energy converters may be understood better when compared to a current snapshot of the competitors, as shown schematically in Fig.€ 27.2. Wave energy converters such as OPT’s Powerbuoy1 and Oceanlinx2 that are on the sur1╇ 2╇
OPT website: www.oceanpowertechnologies.com. OCEANLINX website: www.oceanlinx.com.
L. D. Mann () Carnegie Wave Energy Limited, Level 1, 16 Ord Street West Perth, WA, Australia 6005 Perth e-mail:
[email protected] A. Schiller, G. B. Brassington (eds.), Operational Oceanography in the 21st Century, DOI 10.1007/978-94-007-0332-2_27, ©Â€Springer Science+Business Media B.V. 2011
721
722
Fig. 27.1↜渀 CETO Schematic
Fig. 27.2↜渀 Schematic comparison of CETO with other WEC’s
L. D. Mann
27â•… Application of Ocean Observations & Analysis
723
face of the water are exposed to breaking waves during large seas, for example, in storms, whereas the CETO device and the AWS buoy3 are fully submerged during normal operation so that they are significantly less prone to damage than floating devices. Pelamis is a surface going device but is designed to tolerate very large seas4. Another metric for comparison of WEC’s is the simplicity of the power generation scheme. CETO, Oyster5 and the shore-based Limpet6 have the electricity generation equipment onshore rather than in the water and this leads to greater simplicity and should translate to better long-term reliability. Also, onshore electrical generation plant may be upgraded or modified without recourse to removal of the WEC’s from the water. Devices that are fully submerged in normal sea states, including CETO, are not able to access the full energy flux of the waves as can surface mounted devices such as OPT’s Powerbuoy. This penalty in energy capture efficiency may be offset by reduced operating costs in the form of maintenance, as submerged devices are subject to a lower rate of occurrence of breaking waves and repetitive stress loading compared to surface devices. CETO will typically be deployed up to several hundred metres from shore so the pipeline lengths are typically greater than those of the other hydraulic wave energy converter, the OYSTER, which operates in the breaking wave zone much closer to shore. This means that the CETO balance of plant design must pay close attention to the optimisation of the energy losses and minimise the cost of piping. Fortunately this is made easier by the selection of very high operating fluid pressures of around 7€ MPa which enables the hydraulic design of CETO to obtain acceptable piping losses in smaller diameter, and therefore cheaper, piping to shore.
27.2╅Installation Water Depth The CETO units are designed to operate in shallow waters of between 20 and 50€m depth. At these depths there is significant energy loss as deep-water swell waves propagate shoreward and lose energy due to friction with the seabed. Nevertheless there is still an appreciable wave resource available on the southern Australian coastline even after these losses are taken into consideration. The advantage of operating in relatively shallow waters is that sites are typically no more than two kilometres from shore and often significantly closer than that, helping to keep costs of piping fluid to shore within acceptable limits.
AWS website: www.awsocean.com. Pelamis website: www.pelamiswave.com. 5╇ OYSTER website: www.aquamarinepower.com. 6╇ LIMPET website: www.wavegen.co.uk. 3╇ 4╇
724
L. D. Mann
27.3â•…Prospective Site Identification The identification of prospective sites for the installation of wave energy convertors requires estimates of wave climate. In this context the wave climate constitutes ‘the resource’. Site selection involves not only the resource, of course, but also takes into account the overlays of onshore grid connectivity and competing ocean usage (for example state and federal marine parks). Based on these and other considerations, Carnegie Wave Energy Limited has selected, applied for, and obtained, wave exploration licenses in selected areas off the coast of Western Australia, South Australia, Victoria and Tasmania. In addition to this site-specific work, RPS METOCEAN was commissioned to produce broader estimates of the total deep water and shallow water wave resource for the southern Australian coastline. The estimated total deep-water wave resource for this coastline is 525€GW and the estimated shallow water resource is 171€GW. This shallow water estimate is still about three times greater than Australia’s national electricity consumption.
27.4â•…First Installation Site: Garden Island, WA The first full scale CETO devices will be deployed in an area near Garden Island off the coast of Western Australia as in Fig.€27.3. This is ultimately the planned site for a 5€MW wave farm comprising multiple CETO units that is expected to be one of the first (if not the first) commercial scale wave farm in the southern hemisphere. Grid connected power is expected to be available by 2012. The site is located in the Sepia Depression, an area of 20–25€m water depth between Five Fathom Bank and Garden Island. Garden Island houses HMAS Stirling, which is the Royal Australian Navy’s largest base. A memorandum of understanding has been signed with the Australian Department of Defence for collaboration, onshore space and potential power off take. The site is connectible into the Western Australian transmission grid (SWIS) via the facilities at Garden Island. The site was chosen because of its proximity to the port of Fremantle and marine support facilities in and around Cockburn Sound, as well as the ready access to the SWIS and the population centre of Perth. The wave resource isâ•›>â•›35€ MW/km on site with approximately 65% availability of 2€m waves and 90% availability of 1€m waves. The presence of Five Fathom Bank provides some sheltering from excessive swell waves while still maintaining a viable wave climate for a commercial wave farm. Significant wave height (Hs) inside the bank is limited to 4€ m, compared to 8€m outside, due to the attenuation effect of bank. The Sepia Depression location has a similar average wave climate to a fully exposed location outside of Five
27â•… Application of Ocean Observations & Analysis
725
Fig. 27.3↜渀 Garden Island site
Fathom Bank. Another advantage of the Sepia Depression is that it has a greater predictability of the sea state due to the sheltering effect and this translates to increased site accessibility. The first phase of full scale insitu evaluation will involve a single autonomous CETO unit. This installation will be used to gather performance and reliability data as well as validate storm survivability measures. A Waverider® buoy is anchored adjacent to the CETO mooring, and both the Waverider® data and the outputs from the WEC will be beamed back to shore via wireless link. Energy produced by the unit will be safely dispersed as heat into the surrounding seawater.
726
L. D. Mann
27.5â•…Specific Uses of Ocean Observations & Analysis The CETO Wave Energy Project has benefited from the vast store of oceanographic data, observations and analysis throughout the course of its development to date, and will continue to draw on the knowledge base as projects are developed worldwide. At this stage of project development, the utility of ocean observations and analysis has been restricted to site selection, validation and calibration studies. It is noted however that operational forecast products are expected to become increasingly relevant at the later stages of commercialisation. Good reviews relevant to the wider application of ocean observations and analysis to wave energy forecasting can be found in the papers of Moreira et€al. (2002); Bruck and Pontes (2006) and Tolman et€al. (2002) and Greenslade and Tolman (2010). Specifically to the CETO project, in 2003 Carnegie commissioned a survey of near-shore wave energy resource availability from a local oceanographic company WNI, with geographical scope restricted to the SW coastal areas of Western Australia from Geraldton around to Esperance. The survey was based on an in-house data set comprising parameterized data from an implementation of NOAA Wave Watch III (NWW3). The data covered the period January-1997 to August-2003, at 3 hourly intervals with spatial resolution of 1° by 1.25°. As a deep-water model, this data was considered relevant only at depths greater than approximately 50€m which in turn restricted attention to a subset of only 8 grid points within the geographic scope. Fortunately, these remaining 8 points corresponded or overlapped with areas along the WA coast that Carnegie had flagged as being of interest. However, this does highlight the limitations of using such a coarse grid to generate data. A finer grid of say 0.25° by 0.25° would have been desirable to account for some of the protection provided headlands, for example Cape Naturaliste. Again it was fortunate in this case that most of the swell waves that impact the south west coast of Western Australia are from the SW direction and not from the S or SSW so the shadowing effects of the land were not as severe as they might have been in other locations. More recently, Carnegie commissioned an independent report from ocean resource specialists RPS MetOcean, to provide an independent assessment of the near-shore wave energy resource availability at 17 potential development sites along the southern coastline of Australia. Wave data was sourced primarily from an implementation of NOAA Wave Watch III (NWW3) and compared to available measured data for seven sites across southern Australia for verification purposes and to examine localised effects on wave power and its availability. This study indicated that Australia has a potential near-shore wave energy resource of approximately 170,000€MW in water depths of 25€m (Fig.€27.4). This equates to approximately 4 times the total amount of installed power generation capacity nationally. The shallow water wave estimate represents the potentially available resource only and does not take into account the efficiency of extraction by a wave energy conversion device or accessibility of the wave resource. The issues inherent in the 2003 wave modelling exercise were recognised and addressed by this study. Specifically, points of interest were selected where, because of the bathymetry, the model was
27â•… Application of Ocean Observations & Analysis
727
Fig. 27.4↜渀 Wave resource estimates produced by Carnegie and RPS METOCEAN
able to deliver reliable deep-water results. Also points of interest were chosen only from gridded data points where there were no land shadowing effects. Thus, this later report represents a better estimate of the shallow water resource for Southern Australia. In summary, the result was an indicative ratio between the total available deep water and shallow water resource of the order of 3:1 (525:170€GW).
27.6â•…Limitations to the Use of Model Data Neither of the modelling exercises discussed above considered mesoscale effects such as sea breezes, and these are usually low pass filtered out of WWW3 raw data. As a result the modelled wave climate exclude higher frequency signals such that only swell states with wave periods typicallyâ•›≥â•›8€sec are represented. The experience of using model data provided interesting insights into the limitations of such data for determining suitable sites for wave energy converters. For example it was recognised that ‘mapping in’ of deepwater ocean data using computational algorithms taking into account bathymetry, shoaling effects and coastline morphology, would be useful for some locations but less effective for others. Importantly from a commercial perspective, is the consideration of diminishing returns with respect to the expense and quality of data. This trade-off between the increasing cost of modelling and the resultant highly processed information was
728
L. D. Mann
also influenced by the fact that a tri-axis wave accelerometer could be purchased for approximately the same cost as an extensive mesoscale wave data analysis. There the choice becomes one between having actual real data at site (but having to wait several months to collect a representative statistical wave data set) versus having highly interpreted and therefore slightly suspect data (but having it representing a longer epoch).
27.7â•…Pragmatic Approaches in a Commercial Context Carnegie in its development of CETO wave sites worldwide generally favours the approach of using coarse gridded WWW3 data as a guide to determining general feasibility, then making detailed selections based on deployed tri-axis wave measurement accelerometers, rather than ‘gridding in’ coarser wave data to finer scales. This approach works in practice because at scales finer than the typical WWW3 grid, the decision about the most suitable site for a wave energy converter is no longer solely dependent on wave resource alone; other considerations such as land and seabed access, and access to onshore grid connections come in to play. This pragmatic selection of sites, overriding purely wave resource factors, is evidenced in the selection of the Sepia Depression site discussed earlier. Here the site was selected partially for its sheltering ability but mostly for convenience and access. Another aspect of wave energy converter design that CETO, and indeed all devices, must address is: how adaptable the design is to accommodate the actual dynamic range of wave heights that are expected at locations where they are going to be deployed. The distribution of wave height along with the maximum wave height at a given site must be known in order to match the design to the site. In practice, the gathering of this detailed information could be an expensive and time consuming process if there is not already wave buoy data available at the exact location of deployment, or if surveys and analysis have not already been carried out. The verification of the operation of CETO at a technical level will involve the comparison of empirical measurement with the convolution output of the device power matrix and the wave matrix for the Sepia Depression site. This process allows the actual capacity factor to be compared with that predicted from the convolution of these two matrices, and is the key to commercial validation. It is important to note that the tools such as WWW3, while useful for site selection, are not in themselves sufficient to predict the energy output of CETO or of any other wave energy converter for that matter. This is because all wave energy converters around the world have not yet accumulated enough operation data to provide a simple predictor of integrated energy output based on historical or hindcast wave data statistics. This will emerge over years to come, but for now all wave energy converters will need to demonstrate ‘bankability’; that is, a sufficiently high average capacity factor for the particular wave farm to present a favourable return on investment.
27â•… Application of Ocean Observations & Analysis
729
Fig. 27.5↜渀 Results of wave resource analysis produced by Carnegie and commissioned from RPS METOCEAN for selected locations along the southern Australian coastline
Until sufficient operational data exists, most wave farms, including CETO, will be in this bankability demonstration mode. Furthermore, such demonstration in practice requires a wave buoy on site at the wave farm to correlate the input wave state with the output of the wave energy converter. Acknowledgments╇ Carnegie acknowledges the spport of the Western Australian government through their LEED funding program, which partially supports this work. RPS METOCEAN is acknowledged for providing the wave data and Mr. Tim Sawyer of Carnegie for the preparation and analysis of the data presented in Figs.€27.4 and 27.5.
References Bruck M, Pontes MT (2006) Wave energy resource assessment based on satellite data. Workshop on performance monitoring of ocean energy systems. http://pmoes.ineti.pt. Lisbon, Nov 2006 Greenslade D, Tolman H (2010) Surface waves. In: Schiller A, Brassington GB (eds) Operational oceanography in the 21st Century. Springer, New York Moreira NM, Oliveira Pires H, Pontes T, e Câmara C (2002) Verification of TOPEX-Poseidon wave data against buoys off the West Coast of Portugal. Proceeding. Conference on Offshore Mechanics and Arctic Engineering (OMAE02), paper 2002–28254. Oslo, Norway, 23–28 June 2002 Tolman HL, Balasubramaniyan B, Burroughs LD, Chalikov DV, Chao YY, Chen HS, Gerald VM (2002) Development and implementation of wind generated ocean surface wave models at NCEP. Weather Forecast 17:311–333