Remote Sensing of the Changing Oceans

Remote Sensing of the Changing Oceans DanLing (Lingzis) Tang Editor Remote Sensing of the Changing Oceans 123 Edi...

Author: DanLing Tang

80 downloads 1264 Views 19MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Remote Sensing of the Changing Oceans

DanLing (Lingzis) Tang Editor

Remote Sensing of the Changing Oceans

123

Editors DanLing (Lingzis) Tang Research Center for Remote Sensing and Marine Ecology/Environment (RSMEE) Key Laboratory of Tropical Marine Environmental Dynamics South China Sea Institute of Oceanology Chinese Academy of Sciences No.164 West Xingang Road 510301 Guangzhou People’s Republic of China [email protected] [email protected] Gad Levy NorthWest Research Associates Seattle Division 4118 148th Ave NE 98052 Redmond USA [email protected] Malcolm Heron Marine Geophysical Laboratory School of Computer Science, Mathematics James Cook University Townsville Australia [email protected]

James (Jim) Gower Institute of Ocean Sciences Fisheries and Oceans Canada Marine Environmental Quality Section West Saanich Road 9860 V8L 4B2 Sidney British Columbia Canada [email protected] Kristina B. Katsaros Division of Applied Marine Physics Rosenstiel School of Marine and Atmospheric Science Campus University of Miami 4600 Rickenbacker Causeway 33149 Miami USA [email protected] Ramesh Singh Schmid College of Science Chapman University One University Drive 92866 Orange USA [email protected]

ISBN 978-3-642-16540-5 e-ISBN 978-3-642-16541-2 DOI 10.1007/978-3-642-16541-2 Springer Heidelberg Dordrecht London New York © Springer-Verlag Berlin Heidelberg 2011 This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable to prosecution under the German Copyright Law. The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Cover design: deblik, Berlin Printed on acid-free paper Springer is part of Springer Science+Business Media (www.springer.com)

Ocean Satellite Remote in space, I keep my constant watch, Sensing your color, temperature, height and texture, Of all possible viewpoints, mine is the best. Now, The humans are changing the atmosphere, Changing the way you live and breathe. Oceans, caring for you is my only mission. My eyes are only on you. Guangzhou, China November, 2010

DanLing (Lingzis) Tang, James (Jim) Gower

v

This is Blank Page Integra

vi

Acknowledgements

The successful completion of this work “Remote Sensing of the Changing Oceans” is the result of the cooperation, confidence, and endurance of many people. This book is also an achievement in relation to the 9th Pan Ocean Remote Sensing Conference-PORSEC2008, which, through the help and support of many talented people, institutions and government departments, was successfully held in Guangzhou, China in December 2008. All the authors were participants of PORSEC2008. I thank the authors for their great contributions and their patience and effort to revise their chapters. I appreciate our editorial board members Drs. James (Jim) Gower, Gad Levy, Kristina B. Katsaros, Malcolm Lewis Heron, and Ramesh Singh, as well as all the other reviewers for their time, passion and ability to improve the book. I would also like to express my gratitude to Miss Paula Lei for her constant assistance during the entire process. Thanks to my team members. My heartfelt thanks also go to Dr. Johanna Schwarz for her coordination and patience, and to the Springer production team for taking care of the typesetting and layout of the book. We thank National Natural Science Foundation of China (40576053, 40811140533, 40976091, and 31061160190), Chinese Academy of Sciences (kzcx2-yw-226), and Guangzhou Association for Science and Technology, China, and Guangdong Natural Science Foundation (8351030101000002, 40976091, 2010B031900041), for their generous support for PORSEC2008 and for related research projects. Last but not least, I must thank Professor Sui GuangJun and Mr Sui Yi, for their understanding and support for my work.

May, 2010 Guangzhou, China

DanLing (Lingzis) Tang

vii


viii

Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . DanLing (Lingzis) Tang and Gad Levy Part I

1

Satellite Observation System and International Cooperation

2 Climate Data Issues from an Oceanographic Remote Sensing Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . Kristina B. Katsaros, Abderrahim Bentamy, Mark Bourassa, Naoto Ebuchi, James (Jim) Gower, W. Timothy Liu, and Stefano Vignudelli

7

3 Altimeter Observations of Sea Level and Currents off Atlantic Canada . . . . . . . . . . . . . . . . . . . . . . . . . . . . Guoqi Han

33

4 Eddy Statistics for the Black Sea by Visible and Infrared Remote Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Svetlana Karimova

61

5 Passive Ocean Remote Sensing by Near-Space Vehicle-borne GPS Receiver . . . . . . . . . . . . . . . . . . . . . . Wen-Qin Wang, Jingye Cai, and Qicong Peng

77

Part II

Global Changes

6 A Global Survey of Intense Surface Plankton Blooms and Floating Vegetation Using MERIS MCI . . . . . . . . . . . . . James (Jim) Gower and Stephanie King

99

7 Evaluating Sea Ice Deformation in the Beaufort Sea Using a Kinematic Crack Algorithm with RGPS Data . . . . . . . K. Peterson and D. Sulsky

123

8 Satellite Air – Sea Fluxes . . . . . . . . . . . . . . . . . . . . . . . . Abderrahim Bentamy, Kristina B. Katsaros, and Pierre Queffeulou

141

9 Remote Sensing of Oil Films in the Context of Global Changes . . Andrei Yu. Ivanov

169 ix

x

Contents

Part III Coastal Environment 10

Coastal Monitoring by Satellite-Based SAR . . . . . . . . . . . . . Antony K. Liu

195

11

Satellite Altimetry: Sailing Closer to the Coast . . . . . . . . . . . Stefano Vignudelli, Paolo Cipollini, Christine Gommenginger, Scott Gleason, Helen M. Snaith, Henrique Coelho, M. Joana Fernandes, Clara Lázaro, Alexandra L. Nunes, Jesus Gómez-Enri, Cristina Martin-Puig, Philip Woodworth, Salvatore Dinardo, and Jérôme Benveniste

217

12

Low Primary Productivity in the Chukchi Sea Controlled by Warm Pacific Water: A Data-Model Fusion Study . . . . . . . . Kohei Mizobata, Jia Wang, Haoguo Hu, and Daoru Wang

13

Medium Resolution Microwave, Thermal and Optical Satellite Sensors: Characterizing Coastal Environments Through the Observation of Dynamical Processes . . . . . . . . . . Domingo A. Gagliardini

239

251

Part IV Regional Observation 14

15

16

Satellite Observation on the Exceptional Intrusion of Cold Water and Its Impact on Coastal Fisheries Around Peng-Hu Islands, Taiwan Strait . . . . . . . . . . . . . . . . . . . . Ming-An Lee, Yi Chang, Kuo-Wei Lan, Jui-Wen Chan, and Wei-Juan Hsieh Comparison of the Satellite and Ship Estimates of Chlorophyll-a Concentration in the Sea of Japan . . . . . . . . . Elena A. Shtraikhert, Sergey P. Zakharkov, and Tatyana N. Gordeychuk Observed Interannual Variability of the Thermohaline Structure in the South Eastern Arabian Sea . . . . . . . . . . . . . Nisha Kurian, Joshua Costa, V. Suneel, V.V. Gopalakrishna, R.R. Rao, K. Girish, S. Amritash, M. Ravichandran, Lix John, and C. Ravichandran

Part V 17

18

281

293

305

Natural Hazards

Satellite Observations Defying the Long-Held Tsunami Genesis Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Y. Tony Song and Shin-Chan Han Tsunami Source Reconstruction by Topex/Poseidon Data . . . . . . Vladimir V. Ivanov

327 343

Contents

19

Scientific Research Based Optimisation and Geo-information Technologies for Integrating Environmental Planning in Disaster Management . . . . . . . . . . Hussain Aziz Saleh and Georges Allaert

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xi

359 391


xii

Contributors

Georges Allaert Institute for Sustainable Mobility, Ghent University, Vrijdagmarkt 10/301, 9000 Gent, Belgium, [email protected] S. Amritash National Institute of Oceanography, Regional Centre, Kochi, India, [email protected] Abderrahim Bentamy Institut Français pour la Recherche et l’Exploitation de la MER (IFREMER), Plouzané, France, [email protected] Jérôme Benveniste European Space Agency/ESRIN, Frascati, Italy, [email protected] Mark Bourassa COAPS and Florida State University, Tallahassee, FL, USA, [email protected] Jingye Cai School of Communication and Information Engineering, University of Electronic Science and Technology of China, Chengdu 610054, P. R. China, [email protected] Jui-Wen Chan Remote Sensing Laboratory, National Applied Research Laboratories, Taiwan Ocean Research Institute, Taipei, Taiwan, [email protected] Yi Chang Department of Environmental Biology and Fisheries Science, National Taiwan Ocean University, Pei-Ning Rd. Keelung 20224, Taiwan, [email protected] Paolo Cipollini Ocean Observing and Climate, National Oceanography Centre, Southampton, UK, [email protected] Henrique Coelho Hidromod, Lisbon, Portugal, [email protected] Joshua Costa National Institute of Oceanography, Dona Paula, Goa, India, [email protected] Salvatore Dinardo Serco/ESRIN, Frascati, Italy, [email protected] Naoto Ebuchi Institute of Low Temperature Science, Hokkaido University, Sapporo, Japan, [email protected] xiii

xiv

Contributors

M. Joana Fernandes Faculdade de Ciências, Universidade do Porto, Porto, Portugal, [email protected] Domingo A. Gagliardini IAFE, Casilla de Correo 67, Suc. 28 (C1428ZAA) Ciudad Autónoma de Buenos Aires, Argentina, [email protected] K. Girish National Institute of Oceanography, Regional Centre, Kochi, India, [email protected] Scott Gleason Ocean Observing and Climate, National Oceanography Centre, Southampton, UK, [email protected] Jesus Gómez-Enri Universidad de Cádiz, Cádiz, Spain, [email protected] Christine Gommenginger Ocean Observing and Climate, National Oceanography Centre, Southampton, UK, [email protected] V.V. Gopalakrishna National Institute of Oceanography, Dona Paula, Goa, India, [email protected] Tatyana N. Gordeychuk Pacific Oceanological Institute, Far Eastern Branch of the Russian Academy of Sciences, 43 Baltiyskay Street, Vladivostok 690041, Russia, [email protected] James (Jim) Gower Institute of Ocean Sciences, Fisheries and Oceans Canada, Sidney, BC, Canada, [email protected] Guoqi Han Fisheries and Oceans Canada, Northwest Atlantic Fisheries Centre, St. John’s, NL, Canada, [email protected] Shin-Chan Han Goddard Space Flight Center, National Aeronautics and Space Administration, [email protected] Wei-Juan Hsieh Remote Sensing Laboratory, Taiwan Ocean Research Institute, National Applied Research Laboratories, Taipei, Taiwan, [email protected] Haoguo Hu School of Natural Resources and Environment, Cooperative Institute for Limnology and Ecosystems Research, University of Michigan, Ann Arbor, MI, USA, [email protected] Andrei Yu. Ivanov P.P. Shirshov Institute of Oceanology, Russian Academy of Sciences, Nakhimovsky prospect, 36, Moscow, 117997, Russian Federation, [email protected] Vladimir V. Ivanov Institute of Marine Geology & Geophysics, Yuzhno-Sakhalinsk, Russia, [email protected] Lix John National Institute of Oceanography, Regional Centre, Kochi, India, [email protected]

Contributors

xv

Svetlana Karimova Space Research Institute of the Russian Academy of Sciences, 84/32 Profsoyuznaya St., Moscow, 117997, Russia, [email protected] Kristina B. Katsaros Rosenstiel School of Marine and Atmospheric Sciences, University of Miami, Miami, Florida and Northwest Research Associates, Bellevue, Washington, USA, [email protected] Stephanie King Institute of Ocean Sciences, Fisheries and Oceans Canada, Sidney, BC, Canada, [email protected] Nisha Kurian National Institute of Oceanography, Dona Paula, Goa, India, [email protected] Kuo-Wei Lan Department of Environmental Biology and Fisheries Science, National Taiwan Ocean University, Pei-Ning Rd., Keelung, 20224, Taiwan, [email protected] Clara Lázaro Faculdade de Ciências, Universidade do Porto, Porto, Portugal, [email protected] Ming-An Lee Department of Environmental Biology and Fisheries Science, National Taiwan Ocean University, Pei-Ning Rd., Keelung 20224, Taiwan; Remote Sensing Laboratory, National Applied Research Laboratories, Taiwan Ocean Research Institute, Taipei, Taiwan, [email protected] Gad Levy NorthWest Research Associates, Seattle Division, 4118 148th Ave NE, 98052 Redmond, USA, [email protected] Antony K. Liu National Taiwan Ocean University, Keelung, Taiwan; NASA Goddard Space Flight Center, Greenbelt, Maryland, USA, [email protected] W. Timothy Liu Jet Propulsion Laboratory, Pasadena, CA, USA, [email protected] Cristina Martin-Puig Starlab Barcelona S.L., Barcelona, Spain, [email protected] Kohei Mizobata Department of Ocean Sciences, Tokyo University of Marine Science and Technology, 4-5-7, Kounan, Minato-ku, 108-8477 Tokyo, Japan, [email protected] Alexandra L. Nunes Instituto Politécnico do Porto, Instituto Superior de Engenharia, Porto, Portugal, [email protected] Qicong Peng School of Communication and Information Engineering, University of Electronic Science and Technology of China, Chengdu 610054, P. R. China, [email protected] K. Peterson Sandia National Laboratories, Albuquerque, NM, USA, [email protected]

xvi

Contributors

Pierre Queffeulou Institut Français pour la Recherche et l’Exploitation de la MER (IFREMER), Plouzané, France, [email protected] R.R. Rao Naval Physical and Oceanographic Laboratory, Kochi, India, [email protected] C. Ravichandran National Institute of Oceanography, Regional Centre, Kochi, India, [email protected] M. Ravichandran Indian National centre for Ocean Information Services, Hyderabad, India, [email protected] Hussain Aziz Saleh Higher Commission for Scientific Research, P.O. Box 30151, Damascus, Syria; Institute for Sustainable Mobility, Ghent University, Gent, Belgium, [email protected]; [email protected]; [email protected] Elena A. Shtraikhert V.I.Il`ichev Pacific Oceanological Institute, Far Eastern Branch of the Russian Academy of Sciences, 43, Baltiyskaya Street, Vladivostok 690041, Russia, [email protected] Helen M. Snaith Ocean Observing and Climate, National Oceanography Centre, Southampton, UK, [email protected] D. Sulsky Department of Mathematics and Statistics, University of New Mexico, Albuquerque, NM, USA, [email protected] V. Suneel National Institute of Oceanography, Dona Paula, Goa, India, [email protected] Y. Tony Song Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109, USA, [email protected] DanLing (Lingzis) Tang Research Center for Remote Sensing and Marine Ecology/Environment (RSMEE), LED, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China, [email protected]; [email protected] Stefano Vignudelli Consiglio Nazionale delle Ricerche, Pisa, Italy, [email protected] Jia Wang NOAA Great Lakes Environmental Research Laboratory (GLERL), 4840 S. State Road, 48108 Ann Arbor, MI, USA, [email protected] Daoru Wang Hainan Marine Development and Design Institute, Hainan, China, [email protected] Wen-Qin Wang School of Communication and Information Engineering, University of Electronic Science and Technology of China, Chengdu, P. R. China, 610054; Key Laboratory of Ocean Circulation and Waves, Chinese Academy of Sciences, Qingdao, P. R. China, 266071, [email protected]

Contributors

xvii

Philip Woodworth Proudman Oceanographic Laboratory, Liverpool, UK, [email protected] Sergey P. Zakharkov Pacific Oceanological Institute, Far Eastern Branch of the Russian Academy of Sciences, 43 Baltiyskay Street, Vladivostok 690041, Russia, [email protected]


xviii

Reviewers

Bayler,

Eric

Berkelmans,

Ray

Brinkman,

Richard

Businger, Cherniawsky, Evans, Gower, Heron, Ianson, Katsaros, Kepert, Kwok, Levy, Liu, Lough,

Steven Josef Wayne James (Jim) Malcolm Lewis Debby Kristina Jeff Ron Gad Cho Teng Janice

Mests-Nunez, Meyers,

Alberto Gary

Rothlisberg,

Peter

Roughan, Singh, Skirving,

Moninya Ramesh William

Tang,

DanLing (Lingzis)

Tory, Trinanes,

Kevin Joaquin

Troitskaya,

Yuliya

National Oceanic and Atmospheric Administration (NOAA), USA Australian Institute of Marine Science (AIMS), Australia Australian Institute of Marine Science (AIMS), Australia University of Hawaii, USA Institute of Ocean Sciences, Canada NorthWest Research Associates (NWRA), USA Institute of Ocean Sciences, Canada James Cook University, Australia Institute of Ocean Sciences, Canada University of Miami, USA Bureau of Meteorology Research Centre, Australia Jet Propulsion Laboratory, NASA, USA NorthWest Research Associates (NWRA), USA Taiwan University Australian Institute of Marine Science (AIMS), Australia Texas A&M University-Corpus Christi, USA Commonwealth Scientific and Industrial Research Organization (CSIRO), Australia Commonwealth Scientific and Industrial Research Organization (CSIRO), Australia University of New South Wales (UNSW), Australia Chapman University, USA National Oceanic and Atmospheric Administration (NOAA), USA South China Sea Institute of Oceanology, Chinese Academy of Sciences, China Bureau of Meteorology Research Centre, Australia National Oceanic and Atmospheric Administration (NOAA), USA Institute of Applied Physics RAS, Russia

xix


xx

Acronyms

ACO ADCP ADEOS ADEOS-1/2 ALBICOCCA ALTICORE AMSRs AO APT ASCAT ASF AVHRR AVISO AZMP BoB BPSK BT CDOM CDRs CEOS CERSAT(IFREMER) CGDR CGDRs CIOM CMORPH CNES COASTALT CONAE CPC

Ant Colony Optimization Acoustic Doppler Current Profiler ADvanced Earth Observing Satellite ADvanced Earth Observing Satellites 1 and 2 ALtimeter-Based Investigations in COrsica, Capraia and Contiguous Areas value added satellite ALTImetry for COastal REgions Advanced Microwave Scanning Radiometers Announcement of Opportunity Automatic Picture Transmission Advanced SCATterometer Alaska Satellite Facility Advanced Very High Resolution Radiometer Archiving, Validation and Interpretation of Satellite Oceanographic data Atlantic Zone Monitoring Program Bay of Bengal Binary Phase Shift-Keyed Brightness Temperature Colored Dissolved Organic Material Climate Data Records Committee on Earth Observation Satellites Centre ERS d’Archivage et de Traitement (IFREMER) Coastal Geophysical Data Record Coastal Geophysical Data Records Coupled Ice-Ocean Model CPC MORPHing technique Centre National d’Etudes Spatiales (French Space Agency) ESA Development of COASTal ALTimetry COmisión Nacional de Actividades Espaciales (Argentine Agency for Space Activities) Climate Prediction Center xxi

xxii

CSA CTD CTPR CYR CZCS DEM DIRTH DLM DMU DORIS ECDIS ECMWF EICC EPA EP-TOMS ERS ERS-1/2 ESA ESDRs EUMETSAT EW FOCI GAC GAs GCOS GDR GDRs GEO GEOSS GHRSTT GIM GIMs GIS GLONASS GM GMES GMF GNSS GOCE GODAE GOSUD GPD GPM G-POD

Acronyms

Canadian Space Agency Conductivity-Temperature-Depth Clutter-to-Target Power Ratio Chang-Yuen Ridge Coastal Zone Color Scanner Digital Elevation Model Direction Interval Retrieval with Threshold Nudging Dynamically Linked Model De Montfort University Doppler Orbitography and Radiopositioning Integrated by Satellite Electronic Chart Display and Information System European Centre for Medium-Range Weather Forecasts East India Coastal Current Environmental Protection Agency Earth Probe – Total Ozone Mapping Spectrometer European Remote-sensing Satellite European Remote-sensing Satellites 1 and 2 European Space Agency Earth System Data Records EUropean Organization for the Exploitation of METeorological SATellites Early Warning Fisheries-Oceanography Coordinated Investigation Global Area Coverage Genetic Algorithms Global Climate Observing System Geophysical Data Record Geophysical Data Records Group on Earth Observations Global Earth Observation System of Systems Group for High Resolution SST Global Ionosphere Map GPS Ionosphere Maps Geographic Information System GLObal NAvigation Satellite System ERS-1 Geodetic Mission Global Monitoring for Environment and Security Global Model Function Global Navigation Satellite System Gravity field and steady-state Ocean Circulation Explorer Global Ocean Data Assimilation Experiment Global Ocean Surface Underway Data GNSS-derived Path Delay Global Precipitation Measuring mission GRID Processing On Demand

Acronyms

xxiii

GPS GRACE GRIP GSFC GSHHS

Global Positioning Systems Gravity Recovery And Climate Experiment Government Related Initiative Program NASA/Godard Space Flight Center Global Self-consistent, Hierarchical, High resolution Shoreline Database Gulf of St. Lawrence Harmful Algal Blooms High-Frequency Handicapped Person Transportation HaiYang (for Ocean in Chinese) satellite mission International Arctic Research Center Information Communication Technology Institut Francais de Recherche et de L’Exploitation de la Mer GFO Intermediate Geophysical Data Record IARC-JAXA Information System Initial Joint Polar System ENVISAT RA-2 Intermediate Marine Abridged Record Instituto Nacional de Pesquisas Espaciais Inertial Navigation Units International Ocean Colour Coordinating Group Intergovernmental Panel on Climate Change Institut pour la Recherche et le Développement International Satellite Cloud Climatology Project Integrated Side Lobe Ratio Japan Aerospace Exploration Agency Jet Propulsion Laboratory Low Earth Orbit Light Detecting And Ranging Maximum Chlorophyll Index Multi Channel Sea Surface Temperature Making Earth System data records for Use in Research Environments Middle Earth Orbit MEdium Resolution Imaging Spectrometer Meteorological Operational TOPEX-Poseidon Merged Geophysical Data Record Massive Influx Marginal Ice Zone Merged Local Area Coverage Mushroom-Like Currents Mixed Layer Depth Maximum Likelihood Estimator MODerate-resolution Imaging Spectroradiometer Multi-objective Optimisation Problems

GSL HABs HF HPT HY-2 IARC ICT IFREMER IGDR IJIS IJPS IMAR INPE INU IOCCG IPCC IRD ISCCP ISLR JAXA JPL LEO LIDAR MCI MCSST MEaSUREs MEO MERIS MetOp M-GDR MI MIZ MLAC MLCs MLD MLE MODIS MOPs

xxiv

MPL MSFC MWR NAEs NAO NASA NCEP NCAR NDBC NESZ netCDF NOAA NODC NPOESS NW NWP O&SI SAF OA OC4L ODAS ONI ONR OOPC OPR-2 OST OSTM OSTST OSVW PHI PIRATA PISTACH PMEL PO.DAAC POC PRN QC RA-2 RAIES RCS RGPS RMS RR RS

Acronyms

Main Processing Loop NASA Marshall Space Flight Center Microwave Radiometer Near-shore Anticyclonic Eddies North Atlantic Oscillation National Aeronautics and Space Administration National Center for Environmental Prediction National Center for Atmospheric Research National Data Buoy Center Noise Equivalent Sigma Zero network Common Data Form National Oceanic and Atmospheric Administration National Ocean Data Center National Polar- Orbiting Operational Environmental Satellite System North Western Numerical Weather Prediction Ocean & Sea Ice Satellite Application Facility Objective Analysis Ocean Color 4 version 4 Linear Ocean Data Acquisition System Oceanic Niño Index U.S. Office of Naval Research Ocean Observations Panel for Climate Ocean PRoduct level 2 Ocean Surface Topography Ocean Surface Topography Mission Ocean Surface Topography Science Team Ocean Surface Vector Wind Peng-Hu Islands Pilot Research Moored Array in the Tropical Atlantic Prototype Innovant de Système de Traitement pour les Applications Côtières et l’Hydrologie NOAA Pacific Marine Environmental Laboratory Physical Oceanography Distributed Active Archive Center Particulate Organic Carbon Pseudo-random Noise Quality Control Radar Altimeter 2nd generation on Envisat Envisat RA-2 Individual Echo and S-band data for ocean Radar Cross Section RADARSAT Geophysical Processor System Root-Mean-Square Reduced Resolution Remote Sensing

Acronyms

xxv

SA SAMOS

Simulated Annealing Shipboard Automated Meteorological and Oceanographic System Synthetic Aperture Radar Shelf-Basin Interactions South China Sea South Eastern Arabian Sea Sea-viewing Wide Field-of-view Sensor Sensor Geophysical Data Record Scanning Multichannel Microwave Radiometer Short Message Service Signal-to-Noise Ratio Shuttle Radar Topography Mission Sea Surface Height Special Sensor Microwave Imager Sea Surface Salinity Sea Surface Temperature Singular Value Decomposition Significant Wave Height Surface Water and Ocean Topography TOPEX/Poseidon Tropical Atmosphere Ocean Total Electron Content Thematic Mapper/ Enhanced Thematic Mapper Plus Top of the Atmosphere Tropical Ocean Global Atmosphere Tropical Rainfall Measuring Mission Tabu Search Taiwan Strait User-defined Coastal Geophysical Corrections Voltage Controlled Attenuator Vessel Monitoring System Vertical Polarization Warm Core Rings Normalized Water-Leaving Radiance Winter Monsoon Current WiNd Field World Ocean Circulation Experiment Wind Vector Cell Zenith Hydrostatic Delay Zenith Total Delay Zenith Wet Delay

SAR SBI SCS SEAS SeaWiFS SGDR SMMR SMS SNR SRTM SSH SSM/I SSS SST SVD SWH SWOT T/P TAO TEC TM/ETM+ TOA TOGA TRMM TS TS UCGC VCA VMS VV WCR WLR WMC WNF WOCE WVC ZHD ZTD ZWD

Chapter 1

Introduction DanLing (Lingzis) Tang and Gad Levy

Abstract The world’s oceans and their resources are vast, and they are changing globally. Rapid growth in population and development is leading to large demands on marine resources and fast changes in the oceanic environment. Satellite remote sensing can provide the opportunity to study these dynamic processes of oceans on a global scale. Keywords Ocean · Remote sensing · Change The world’s oceans and their resources are vast, and they are changing globally. Rapid growth in population and development is leading to large demands on marine resources and fast changes in the oceanic environment. The recognition that the ocean environment is changing and must be monitored, and that marine resources must be managed wisely, is gaining acceptance within the global scientific, political, and economic communities. Marine scientists, engineers, and resource managers need to have at their disposal extensive digital data sets for comprehensive scientific understanding and monitoring of the changes, as well as for planning operations and managing vital marine resources. These data are inherently spatial and lend themselves well to being exploited by satellite remote sensors. The study of the changing ocean encompasses several aspects, including biological, physical, and chemical properties, as well as the interaction of the ocean with land, the cryosphere, and the atmosphere. Satellite remote sensing has given the opportunity to study these dynamic processes of oceans on a global scale. This book features a selection of articles based on expanded presentations at the Ninth Biennial Pan Ocean Remote Sensing Conference (PORSEC2008) themed: “Oceanic Manifestation of Global Changes”, held in Guangzhou, China in December 2008. Primarily through volunteer efforts, with support from the D. Tang (B) Research Center for Remote Sensing and Marine Ecology/Environment (RSMEE), LED, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China e-mail: [email protected]; [email protected]

D. Tang (ed.), Remote Sensing of the Changing Oceans, C Springer-Verlag Berlin Heidelberg 2011 DOI 10.1007/978-3-642-16541-2_1,

1

2

D. Tang and G. Levy

host countries and national and international agencies that share its principles, the PORSEC Association has been holding biennial scientific meetings since 1992 in eight countries around the Pacific Rim and Indian Ocean. The goal of the meetings and the associated training courses is to further the understanding of the Earth’s ocean environmental processes by taking advantage of the unique perspective provided by satellite remote sensing technology. PORSEC strives to protect the ocean and atmosphere and promote sustainable use and development of oceanic and coastal resources. Pan Ocean Remote Sensing Conference Association (PORSEC) is an organization dedicated to helping developing nations stimulate their science programs using global remote sensing data. PORSEC has provided 20 years of effort with over 50 countries and agencies participating to advance the science capabilities in developing countries. A governing body of 60 global scientists provides the guiding effort and leadership to make this a successful movement. Originating in Japan in 1990, it is truly an international effort now. Because of the international nature of PORSEC and its unique mission to foster collaborations between scientists in developing and developed nations, its meetings provide fertile grounds for collaborations between international scientists. These collaborations result, in particular, in shared resources (computer codes, problem solving techniques), shared ideas, and shared data that eventually benefit research of the global community of scientists and institutions observing the Earth. As developing countries start making their own high-quality in situ measurements and launching their own satellite missions, this new wealth of data is critical in supporting regional and global research. For example, the Indian Oceansat II scatterometer will be useful in comparisons with the future US and Chinese scatterometers and European ASCAT, and, in tandem, those could provide synoptic coverage for global weather and climate models. These data are often made available to collaborators who have initiated joint research projects at PORSEC. PORSEC2008 brought together about 450 scientists from 36 countries, notably from the emerging satellite launching nations of China, India and Korea, providing the participants the opportunity to present their results and their Earth observing plans with remote sensing and to plan international collaborations. Most notably, PORSEC2008 had two special workshops. The Workshop on the Indian Ocean and South China Sea role in climate, the Pacific warm pool, monsoons, tropical cyclones and other hazards targeted initiation of cooperation programs based on the interactions/collaborations. The Workshop on data sampling, consistency, archiving, distribution, sharing, and international cooperation aimed to open discussion about what climate researchers need in terms of sampling of oceanographic and meteorological variables and consistency of data records over time. PORSEC is very committed to capacity building, education, and training of the next generation of remote sensing scientists. Extended review papers and in-depth articles deriving from results presented at PORSEC2008 have been compiled in this book. This book is a comprehensive account of the basic concepts, theories, methods and applications used in ocean satellite remote sensing. The book provides a synthesis of various new ideas and theories and discusses a series of key research

1

Introduction

3

topics in oceanic manifestation of global changes as viewed from space. A variety of research methods used in the analysis and modelling of global changes are introduced in detail along with numerous examples from around the world’s oceans. The authors review oceanic manifestation of global changes at different levels, including Global and Regional Observations, Natural Hazards, the Coastal Environment and related science issues, all from the unique perspective of Satellite Observation Systems. Thus, the book not only introduces the basics of oceanic manifestation of global changes, but also new developments in satellite remote sensing technology and international cooperation in this emerging field. This book consists of 19 chapters and is divided into five parts: (1) Satellite Observation System and International Cooperation; (2) Global Changes; (3) Coastal Environment; (4) Regional Observation; (5) Natural Hazards. Reflecting the-state-of-the-science, this book is suitable as a textbook for undergraduate and graduate students and as a useful reference book for researchers and practitioners in ocean science, environmental sciences, remote sensing sciences, marine ecology, and environmental management and planning.

Part I

Satellite Observation System and International Cooperation

Chapter 2

Climate Data Issues from an Oceanographic Remote Sensing Perspective Kristina B. Katsaros, Abderrahim Bentamy, Mark Bourassa, Naoto Ebuchi, James (Jim) Gower, W. Timothy Liu, and Stefano Vignudelli

Abstract In this chapter we review several climatologically important variables with a history of observation from spaceborne platforms. These include sea surface temperature and wind vectors, altimetric estimates of sea surface height, energy and water vapor fluxes at the sea surface, precipitation over the ocean, and ocean color. We then discuss possible improvements in sampling for climate and climate change definition. Issues of consistency of different data sources, archiving and distribution of these types of data are discussed. The practical prospect of immediate international coordination through the concept of virtual constellations is discussed and applauded. Keywords Oceanographic satellite sensors · Scatterometers · Altimeters · Microwave radiometers · Infrared and ocean color sensors · Winds · Sea surface temperature · Air-sea fluxes · International cooperation for climate quality data · Sampling · Consistency · Archiving and distribution

1 Introduction and Motivation This book chapter is a background paper for future action. It follows on a workshop held during the Pan Ocean Remote Sensing Conference 2008 in Guangzhou, China entitled: “International Coordination and Planning for Enhanced Climate Monitoring and Data Stewardship”. The interest in this subject by many remote sensing scientists has grown from observing how often we have missed optimizing the application of our satellite data sets for lack of communication and coordination. While intergovernmental coordination has been taken up by the

K.B. Katsaros (B) Rosenstiel School of Marine and Atmospheric Sciences, University of Miami, Miami, Florida and Northwest Research Associates, Bellevue, Washington, USA e-mail: [email protected]


7

8

K.B. Katsaros et al.

Committee on Earth Observation Satellites (CEOS), the Global Earth Observing System of Systems (GEOSS), Ocean Observations Panel for Climate (OOPC) and Global Climate Observing System (GCOS), the benefit and awareness still have to come down to the level of scientific investigations (e.g. See GEOSS, 2010; OOPC, 2010). In the 20th and 21st century we have created climate changes that may cause serious problems for humankind. We consider changes in the atmosphere, the ocean and the cryosphere intimately related, so they cannot be separated when discussing changes to the complex Earth system. That high energy use by a burgeoning population and increased affluence in many places is causing a warming climate system is now well proven according to the International Panel on Climate Change (IPCC) report 2007 (Solomon et al., 2007). This report states: Warming of the climate system is unequivocal. Most of the observed increase in global average temperatures since the mid-20th century is very likely due to the observed increase in anthropogenic greenhouse gas concentrations.

What are the manifestations of the climate and oceanographic changes in global and regional effects? This was the theme of the PORSEC 2008 conference. This article addresses the status of satellite observations on an international scale with emphasis on oceanographically important measurements. The subject is far too vast to review in this chapter, so it is our intention to focus on some mature and relevant climate/oceanographic variables observed from space and discuss some satellite missions and data management projects where cooperation may give major dividends in the near term. Satellite scientists have a grave responsibility to get the facts straight and be able to inform the leaders around the world about trends in climatic variables and uncertainties in our knowledge. We also need to understand where the changes may lead to serious hazards due to enhanced storm intensity and frequency, sea-level rise, modified monsoons, drought, flooding and more. Part of this responsibility is to work together to accelerate our accomplishments in finding the facts. This article will promote ideas for coordinated sampling of important data with satellite constellations, consistent and uniform calibrations and algorithms, long-term archiving of these data and efficient and convenient distribution for research and applications. We begin by reviewing the status of satellite remote sensing of certain oceanographic and climatically important variables in Sect. 2. Section 3 discusses some results obtained based on these observations aiming to demonstrate their significance and contribution to understanding the changing Earth system. Section 4 discusses four aspects of data collection, sampling, calibration, archiving and distribution. These must be well done for adequate collection and use of the valuable and expensive satellite data in aiding understanding of our global climate, the oceans and what the records tell us about changes. The fifth section is a summary and charge for action.

2

Climate Data Issues from an Oceanographic Remote Sensing Perspective

9

2 Parameters of Focus This section discusses the successes and desirable improvements for measurement of a select number of variables: surface wind speed and momentum flux over the ocean, sea surface temperature (SST), precipitation, air-sea energy fluxes including turbulent fluxes and radiative fluxes, sea surface height measurements and ocean color.

2.1 Winds Over the Oceans Instruments to infer the surface winds over the oceans from space have been in use for nearly 2 decades by the time this book is printed and were famously illustrated by several of the microwave instruments on the SEASAT satellite, launched in 1978 and operated for 3 months (e.g. Katsaros and Brown, 1991). This satellite carried the well-established visible and infrared sensors, as well as experimental instruments such as a microwave radiometer, a scatterometer, i.e. a radar system that measures the roughness of the sea from which wind vectors can be inferred. It also carried an altimeter, a range-measuring radar, which determines the height of the ocean relative to the satellite, from which currents and oceanic heat content can be derived. The shape of the signal also provides information on sea state. The Synthetic Aperture Radar, SAR, on SEASAT provided high-resolution views of the sea surface measuring waves, surface roughness (later converted to wind speed) and ice cover. It allowed detection of oceanic fronts, slicks and ships. These microwave instruments represented relatively new technologies and were mounted together on a polar orbiting satellite platform for the first time. All the microwave instruments depend on variations in the surface roughness in order to provide surface wind information. (Full discussion of these issues and others in this section can be obtained from the text by Robinson, 2004).The signals can be used to derive wind speed and in the case of the scatterometer also wind direction (e.g. Jones et al., 1982). The SEASAT Ku-band scatterometer data allowed scientists to learn how to interpret radar returns from 3 stick antennae, in terms of wind speed and direction basically by inverting an empirical model relating measured scatterometer backscatter coefficients and the surface wind speed and direction. The model calibration is mainly based on the use of in-situ and/or numerical model data collocated in space and time with scatterometer measurements (Bentamy et al., 1999; Stoffelen, 1999; Wentz et al., 2001). Since 1991 there have been several scatterometers in space, two from the European Space Agency, ESA, on the European Remote Sensing satellites 1 and 2, ERS 1/2, launched in 1991 and 1995, respectively. These scatterometers see well through clouds. The C-band scatterometers are relatively insensitive to the small ocean waves at low wind speeds, while the Ku-band instruments are more adversely impacted by clouds and precipitation, but respond better to the small waves or roughness elements at low wind speeds. (See the chapter by Bentamy et al. in this book). A Japanese – U.S. collaboration led to the launch

10


of Ku-band scatterometers, NSCAT (NASA scatterometer) on the Advanced Earth Observing Satellite ADEOS in 1996. Unfortunately, ADEOS had an early demise, which lead to the fast launch in 1999 of QuikSCAT, a satellite carrying the Seawinds scatterometer. This was a copy of the instrument designed for ADEOS II, which was launched in 2002. The Seawinds on QuikSCAT was very successful (e.g. Ebuchi et al., 2002; Katsaros et al., 2001). It collected data for a full 10 year period allowing many new analysis techniques and algorithms to be tested. For example, Long (2004) achieved higher spatial resolution by sophisticated processing of the original data. When ADEOS II was flying, there was a period of a tandem sampling by two identical scatterometers (Liu et al., 2008). The Seawinds scatterometer has a continuous 1,800 km wide swath, including the nadir region, which has proven very important. One such instrument views the global ocean once every 12 h at 50 degrees latitude, but less frequently in the tropics. The sampling with several scatterometers is further discussed in Sect. 4. Since October 2007, the new Advanced Scatterometer, ASCAT, onboard METOP-A, launched by EUMETSAT, is collecting data for an operational agency. (EUMETSAT is the European organization for meteorological satellites for weather, climate and environmental applications). It provides valuable surface wind information with high space and temporal resolutions over the global ocean using two C-band beams, each side of nadir. Figure 2.1 illustrates ASCAT wind retrievals and how it and QuikScat sampling enhance the coverage when analyzed together. Alternative sources of wind speed observations are microwave radiometers and Synthetic Aperture Radar (SAR). SAR cannot be used to obtain wind directions, except indirectly through patterns in the images, such as windstreaks; however, the consensus is that these directions are much less accurate and robust than scatterometer directions. The outstanding advantage is that SAR can retrieve very close to the coast, which is not the case with the current generation of scatterometers and radiometers. The additional great disadvantage is that access to SAR data is highly restrictive and costly. Microwave radiometers typically obtain only wind speed, although an attempt has been made to use multiple looks at the same footprint to derive direction also from the Stokes parameters of the emitted signal (using several look-angles and multiple polarization combinations). Such an instrument was launched as a pilot project (WindSat) in January 2003 (Smith et al., 2006). The hope was that polarimetric microwave radiometers could serve the surface wind sensing need at low cost over the ocean and replace the need for scatterometers, but consensus seems to be at this juncture that a passive instrument has too many limitations (Bourassa et al., 2010). The polarimetric radiometers suffer too much from raininduced interference, and are poor for the low to moderate wind speeds found over 50% of the Earth’s water surface. An active system works better through rain, and has better directional information which is very important for many applications, not least of which is storm analysis. The scatterometer is clearly the instrument of choice for oceanic surface wind and momentum information, except in the very near coastal regions where SAR would be useful if the data were made available. A review of the scatterometer calibrations and the impact of variations in retrievals upon air-sea flux estimates are provided in the chapter by Bentamy et al. in this book.

2


11

Fig. 2.1 ASCAT and QuikSCAT wind observations of hurricane Bill on August 19, 2009. The time separation between the two scatterometer measurements is about 2 h and 30 min

Winds are the most rapidly changing ocean surface variables. On average, QuikSCAT observed the ocean surface twice a day. However, locations near the poles are observed four to six times a day, and large areas around 20◦ are missed every fourth day. Several studies over the years have shown that diurnal variability of surface winds is very important for coupling the ocean’s mixed layer to the atmosphere. Ocean-Sat-2 carrying a scatterometer was launched on September 23, 2009 by India’s space agency and Hy-2 is an expected 2010 launch by China’s space agency. A constellation of inter-calibrated scatterometers is highly desirable to determine the diurnal cycle in surface wind. The sampling with several scatterometers is further discussed in Sect. 4.1.

2.2 Sea Surface Temperature This parameter often used singly to define changes in the climate has the longest history of measurements from space and has become so well established through the long-lived NOAA program of two satellites in morning and early afternoon polar orbits that it is almost taken for granted. Two satellite imagers with wide swaths (2,000 km) cover the whole Earth each day sensing in the atmospheric window regions of the infrared, 3.5–4.0 and 8–12 μm. In the latter window products are

12


nowadays limited to 10–11 μm to avoid the central ozone line inside the 8–12 μm “window”. Standard products have used empirical fits to surface observations by buoys or ships to correct for biases mostly caused by atmospheric variability not sensed by standard sensor suites. Microwave observations of SST through clouds were reported by Wentz et al. (2000) for the first time since SEASAT, but they were not available globally and routinely, since they were derived from the radiometer on the Tropical Rainfall Measuring Mission, TRMM. Since June 2002 microwave measurements of SST by the Advanced Microwave Scanning Radiometers, AMSRs, on NASA’s Earth Observation satellites in polar orbits have given valuable information on the SST below the heavily clouded regions of the Earth, which were eliminated in the purely IR- method (Reynolds et al., 2007). The increased sampling has allowed shorter times for averaging, now daily, and substantially improved the mean values. However, the statistics changed drastically, so that two products are produced now in order to avoid an unphysical jump in the data when the microwave SST’s became available. The article by Smith et al. (2008) reviews this new product of NOAA’s and complexities that have arisen due to two sensors that effectively measure different depths near the sea surface – of the order of – 0.2 mm for IR and 2 cm for microwaves. Zhang et al. (2009) review the total system for operational sea surface temperature production. The diurnal variation of sea surface temperature, especially the formation of a warm layer in low latitudes when the wind is weak, has lead to new concerns related to absolute “accuracy” and the need to define exactly what of many definitions of SST is desired for the climate record. The ubiquitous “cool” film is a thin layer, produced by an upward heat loss from the topmost layer at the interface (even when the net heat flux may result in heating of the ocean due to strong insolation). The shortwave radiation is absorbed over a depth of many 10 s of meters due to penetration by the sunlight and may be distributed over the whole thermocline due to turbulent mixing of the upper ocean in moderate to strong winds (e.g. Saunders, 1967; Katsaros, 1980; Schluessel et al., 1990; Gentemann et al., 2009). Even with strong turbulent mixing there remains a thin “cool film” of the order of a few millimeters near the interface. The infrared radiation measured by infrared remote sensing emanates from that micro-layer, while buoys, ships and the microwave sensors correspond to the temperature at a greater depth in the water, where there may remain a gradient of temperature, so that not even these sensors agree on a SST value. The concept of SST is therefore somewhat ill-defined. An international group was formed to consider these issues (Poulter et al., 2007); currently it is known as the Group for High Resolution SST, GHRSTT, (Donlon et al., 2009 – GHRSST is pronounced “GRIST” for convenience). It is an on-going project defined as follows: “GHRSST has four main tasks that are relevant to the development of the SST observing system: (1) Improved SST data assembly/delivery (2) Testing of SST data sources (3) Perform inter-comparison of SST products (4) Develop applications and data assimilation of SST to demonstrate the benefit of the improved observing system. GHRSST has successfully demonstrated that the requirements of the Global Ocean Data Assimilation Experiment, GODAE can be met and has been instrumental in defining the shape and form of the modern-era SST measurement system and user service over the last 10 years” – (Donlon et al., 2010).

2


13

A difficulty for climate studies that remains is that this long record, 1985–2009 cannot be fully consistent. Overlapping data from the different sources and methods of processing are therefore invaluable for inter-calibrations. For the future, agreements exist between The European Organization for the Exploitation of Meteorological Satellites, EUMETSAT, and NOAA that they share responsibility for operational polar orbiting satellites. The Initial Joint Polar System (IJPS) has the arrangement that EUMETSAT operates its polar-orbiting Meteorological Operational, METOP, satellite series in the morning orbit, and NOAA will continue to guarantee its satellites in the afternoon orbit. These agreements are likely to be extended for NOAA’s new series of satellites under the National Polar-Orbiting Operational Environmental Satellite System, NPOESS. The success of the NOAA satellite program for measuring SST (and visible data for cloud information) and the extensive use by fishermen and others for 3 decades is an encouraging example of exceptional success, and one may add, exemplary leadership in generously sharing these data by the USA. The widespread use of SST information has resulted in joint planning for the future. The IR-SST’s have been made available to anyone via direct read-out, the socalled APT (Automatic Picture Transmission) since the late 1970s, but these data were, of course, not qualified and fully calibrated. A so-called Pathfinder project has continued re-analysis of the high resolution NOAA data, adding refinements to algorithms and known corrections. The long-time global record dating from 1981 is available from http://www.nodc.noaa.gov/SatelliteData/pathfinder4km/. These new Version 5.0 data are being developed at University of Miami at the Rosenstiel School of Marine and Atmospheric Sciences and the National Ocean Data Center (NODC) and distributed in partnership with the NASA Physical Oceanography Distributed Active Archive Center (PO.DAAC). In this 4 km Pathfinder project, the entire time series has been reprocessed at the 4 km Global Area Coverage (GAC) level, the highest resolution possible globally up through about 2006 and interim data for 2007 through near present have also been generated. In addition to the IR sensors measuring in the window region of the spectrum on the NOAA polar orbiting satellites, such sensors were also mounted on the geostationary satellites since the early 1980s and could observe at 4 km resolution. They are mostly used for cloud/weather observations, but can provide SST as well, because frequent observations ameliorate the cloud interference, which is so serious for polar orbiting instruments. Multiple views of the same area as the clouds pass by may allow retrieving the SST between the clouds (e.g. Legeckis and Zhu, 1997). However, global sampling via geostationary satellites requires coordination between 6 such systems, and the footprints of these satellite observations do not reach to the poles.

2.3 Precipitation and Evaporation These two variables, precipitation and evaporation are possibly the most difficult of the variables to measure globally by any means – satellite measurements are

14


the only ones that could provide global coverage, but sampling of precipitation is a major problem due to its intermittent nature. Most records are based on indirect estimates using the cloud brightness or cloud top temperature as indicators of deep and raining systems (e.g. Adler et al., 2003). Direct measurements of precipitation over oceans are almost non-existent, and even attempts to put rain gauges on ships and buoys have had little success due to flow distortion by the ship’s structure. Buoys also have issues with contamination by sea spray, so today satellite estimates of precipitation are typically calibrated against coastal radar estimates of precipitation, which in turn are calibrated against ground based rain-gauge networks (Adler et al., 2003). In spite of these difficulties we do now have precipitation estimates based on inferences from the temperature and patterns of cold cloud tops and from microwave radiometer measurements of the cloud water particles and large particle scattering (These are airborne precipitation particles, which may or may not coincide with precipitation impacting on the sea surface.) A method for estimating the ground precipitation combining various methods called CMORPH (which stands for CPC Morphing technique, where CPC is Climate Prediction Center), has been developed for routine use (Joyce et al., 2004). It is flexible, because it can combine various measurements and their algorithms by “morphing” them together. A dramatic step forward in rain measurements from space was provided by the Tropical Rainfall Measuring Mission, TRMM (Simpson et al., 1988; Kummerow et al., 2000) which carried a narrow-swath rain radar and the TRMM Microwave Radiometer, TMI. A nice example of using the one spaceborne rain radar to date with algorithms calibrated by coastal radars is found in the climatological study of rainfall in tropical cyclones by Lonfat et al. (2004). This subject has come a long way thanks to TRMM, which points the way to better planned systems described in Sect. 4. Stephens and Kummerow (2007) discuss the improvements in estimates that are possible when clouds and precipitation are analyzed in concert. For the Earth’s general hydrologic balance and for changes in the ocean we need to know the net fresh water flux, evaporation minus precipitation, E-P. Reasonably good estimates of evaporation over the ocean using satellite data have been available since a method was proposed by Liu (1984). It uses the bulk air-sea flux models and wind speed, SST and estimates of atmospheric near surface humidity based on the column integrated water vapor content obtained from microwave radiometers. Over the past 25 years this method has been used, somewhat modified (e.g. Schultz et al., 1993, 1997) and incorporated with other air-sea energy flux estimates, as discussed in Sect. 2.4).

2.4 Air Sea Energy Fluxes The fluxes of energy across the air-sea interface, as turbulent fluxes of latent and sensible heat in the atmospheric boundary layer, and short and long wave radiative fluxes have sometimes been the focus of discussion about accuracy required of climate quality data. A value of 10 W m−2 on an annual time scale was established

2


15

as the needed accuracy in net air-sea transfer of energy by a committee in the early 1980s and has been the number suggested many times since; more recently in a chapter of the 2007 report of the Intergovernmental Panel on Climate Change, IPCC, (Bindoff et al., 2007). Finer accuracy is required in Polar Regions (Bourassa et al., 2010). These fluxes depend on many of the variables obtainable from space, so this accuracy requirement puts a heavy burden on the satellite observations over the global oceans. We are not today meeting this accuracy – or one could claim that we don’t even know if we could do it, since high quality net flux is not regularly derived. What we have is mostly one or two terms in the heat budget obtained by a certain group and/or technique over a limited period of time, or we have long-term records, whose consistency is not well proven and the accuracy is known to be less than this ideal. Another consideration is that the fluxes are derived quantities – not directly measured, so an error in their values is due to a compound effect of many variables and physical parameterizations (e.g. Fairall et al., 2010, OCEANOBS09 and references therein.) Nonetheless, many products exist today and are the consequence of diligent efforts over the past 3 decades or more. The earliest method to derive evaporation rate and latent heat flux from satellite data was the study by Liu (1984), who used the bulk aerodynamic formulas discussed by Fairall op cit. and data from the Scanning Multichannel Microwave Radiometer, SMMR (flown both on Seasat and on Nimbus) to estimate near surface atmospheric humidity. SST and surface wind speed are also required and available from satellites already in 1984 as described in Sects. 2.1 and 2.2. Further work to develop the satellite method for latent heat flux, emphasizing the concentration of humidity in the atmospheric boundary layer, were presented by (Schultz et al., 1993, 1997). The reader is referred to the work by the group at IFREMER for evaporative heat flux based mostly on satellite input data (Bentamy et al., 2003 and a chapter in this book). Alternatively, there are products (e.g. Yu et al., 2008), where numerical model, in situ observations, and satellite data are merged. Sensible heat flux has often been tied to the evaporative flux, because the near surface air-temperature, Ta , has remained a difficult variable to obtain from space and not obtained very satisfactorily from numerical models (Kubota and Shikauchi, 1995). New efforts to improve estimates of Ta and nearsurface humidity using multi-sensory microwave observations have been reported (Jackson et al., 2006, and work continues on improvements Gary Wick personal communication, 2009). Strong air- sea temperature contrasts over the warm ocean currents during cold air outbreaks remain difficult to assess, and may remain so for some time. Radiative transfer models have taken observations at the top of the atmosphere, TOA, and inferred the surface irradiances for shortwave irradiance at the air-sea interface (e.g. Gautier et al., 1980; Pinker and Lazlo, 1992). The International Satellite Cloud Climatology Project, ISCCP, is a basis for much of the estimation of surface solar irradiance (Rossow and Duenas, 2004). The variable atmospheric transmission due to aerosols, which are not adequately measured in the long-term record still imply large uncertainty (e.g. Slingo et al., 2006). For the downwelling longwave radiation, use of cloud information and the microwave data on liquid

16


water content shows promise (e.g. Schmetz, 1991; Brisson et al., 2001). Outgoing longwave radiation from the ocean simply requires the black body formula with a value for the surface emissivity of the full infrared spectrum, which however also requires information on the sea state. Community efforts to inter-compare the satellite products are underway within the SEAFLUX program (Curry et al., 2004).

2.5 Altimetry Satellite altimeters are designed to deliver relatively long revisit (10 days and more) and global views of sea surface height. The concept is well established since early missions GEOS-3 and SEASAT that date back to the 1970s. The launch of the TOPEX/Poseidon in 1992 provided the greatest impetus for satellite altimetry research in the 20th century. Its launch was followed by the Jason-1 (2001) and Jason-2 (2008). The European space Agency, ESA, satellites were launched in 1991, 1995, 2002 (ERS-1, ERS-2, ENVISAT) and the US Navy launched a series of Geosat satellites (1985 and 1998). Whilst some of these missions are still in active orbit and expected to continue operation for the foreseeable future, new missions are planned to be launched by space agencies over the next few years, e.g., CryoSat-2, AltiKa, Sentinel-3, HY-2, including new experimental concepts such as SWOT (Fu et al., 2010). Satellite altimetry is recognized as an essential component of the Global Earth Observation System of Systems. To date satellite altimetry has focused on the open ocean (Fu and Cazenave 2001), but recently the coastal ocean has emerged as an important domain for these data (see the chapter by Vignudelli et al. in this book). Only through a synergy from various altimeters can a thorough and complete characterization of mesoscale circulation be obtained (Cipollini et al., 2010). A valuable application has been the incorporation of altimetric data to estimate the depth of the warm water available as a heat source for hurricane intensification in the Caribbean (Goni et al., 2003; Goni et al., 2009) and western Pacific ocean (Lin et al., 2008). An uninterrupted flow of altimeter data has been accumulating, contributing to the ability to address scientific and societal challenges in the ocean. However, in some cases the various missions were designed without continuity in mind, with different observing strategies usually driven by their particular objectives of accuracy, spatial resolution and temporal revisit requirements. For example, the T/P and Jason series main objective was to generate the best estimates of sea level over time to serve climate monitoring, but these data can be also used to better understand the ocean circulation. As a result, data from the various satellite altimeters were processed independently. Nevertheless, the missions have common foundations concerning retrievals, orbits, geophysical corrections, data calibration and exploitation. Assimilation of altimeter data into operational oceanographic models and the increased precision (of the order of 1 cm in sea surface height), makes this technology ready for operational use. EUMETSAT and NOAA have already established cooperation agreements with regard to altimetry (see Sect. 4).

2


17

2.6 Ocean Color Ocean color optical sensors on satellites continue to provide significant data on surface chlorophyll concentrations. Value of the satellite data increases as users gain experience and a larger number of years are covered. Observations from space are quite different from measurements made from research ships, so the interpretation has taken some time to develop. A confounding problem is that chemical and biological elements both contribute to variations in color. Instruments with more spectral bands can help to address this issue in some cases. Also, newer instruments can detect spectral signatures which provide useful additional information, but are not yet widely applied. The ocean color science developed from the 4-band Coastal Zone Color Scanner, the CZCS, on the Nimbus 7 satellite, launched by NASA in 1978. This provided ocean-color bands at blue, blue-green, green and red wavelengths, sufficient for estimates of chlorophyll and water brightness using a simple atmospheric correction. After a data gap of 10 years, the short-lived Japanese OCTS and the US SeaWiFS were launched. SeaWiFS, which has only recently stopped providing data (August 2009) has 8 spectral bands, adding a UV band aimed at separating CDOM from chlorophyll, and near infrared bands to improve atmospheric correction. The simple and well-calibrated design of SeaWiFS has allowed it to collect more than 10 years of “Climate Quality” global ocean color data, “Climate Quality” referring to its ability to address the question, “How is global primary productivity changing under the long-term impact of increasing carbon dioxide in the atmosphere?” SeaWiFS was followed by NASA’s 36-band MODerate-resolution Imaging Spectroradiometer, MODIS, launched on both the Terra satellite (1999) and the Aqua satellite (2002). MODIS has 9 spectral bands giving water color information at 1 km spatial resolution, and also includes two bands with 250 m, and five bands with 500 m spatial resolution for “sharpening” images, three bands for atmospheric water vapor and 17 thermal bands for atmospheric and sea surface temperature. MODIS ocean color bands include two, at 665 and 673 nm, designed to measure chlorophyll fluorescence. MODIS also provides bands in the short-wave infrared which give improved data in silty water. The European Space Agency launched the 15-band MEdium Resolution Imaging Spectrometer, MERIS, in 2002. This has 11 water color bands at 300 m spatial resolution, but with much of the data available only at the reduced resolution of 1,200 m, computed by on-board averaging of 4 by 4 pixels. The additional two bands on MERIS, compared to MODIS, are at 620 nm and 709 nm. The 620 nm band is in a relatively wide gap in MODIS coverage. The 709 band provides an improved baseline for fluorescence measurements and also detects a peak in waterleaving radiance due to intense surface blooms and floating vegetation. The radiance of this peak can be computed as MCI, the Maximum Chlorophyll Index. There are numerous sensors for the ocean color, past, present and future as seen in Fig. 2.2, but additional consideration should be given to planning as discussed below. Satellite ocean color remote sensing today seems to aim mostly at a single product: surface chlorophyll concentration. Standard web pages, such as NASA’s “ocean

18


Fig. 2.2 Ocean color sensors from many agencies, table from IOCCG (2008)

color” site (http://oceancolor.gsfc.nasa.gov) guide users mostly to this, but also provide normalized water-leaving radiance, useful for detecting bright blooms. The new sensors have the ability to distinguish a wider range of targets, but have moved away from the stability and simplicity of SeaWiFS. Given the need for monitoring of long-term changes in surface chlorophyll and hence in primary productivity, there is a feeling that, expressed in verse: We’ve needed one more SeaWiFS for many, many years, Instead we bought two MODIS’s, a MERIS and a VIIRS.

VIIRS was designed as the replacement sensor for the more basic AVHRR imager, but because of program cut-backs it is now the effective replacement for SeaWiFS and MODIS. There have been many delays and compromises in its design. It lacks spectral bands provided by MODIS and MERIS and will not have the stability provided by SeaWiFS. This lack was the subject of a recent briefing note by Siegel, Yoder and McClain, “Thoughts about the Future of Satellite Ocean Color Observations” (October, 2008), which concluded “It appears likely that the ocean biology and biogeochemistry communities will face a multi-year gap in our climate data records.” In view of the importance of these data, international pressure is needed to address this shortcoming. For understanding the ocean carbon cycle, there is also a requirement for estimating concentration of Particulate Organic Carbon (POC) in the ocean. This represents the assemblage of living particles (bacteria, phyto- and zooplankton) and non-living material (detritus, fecal pellets, aggregates) that contribute to the biological pump (transfer of carbon from the upper layers to the deeper ocean by biological processes). POC sinks from surface waters to deeper layers, removing carbon from the

2


19

surface layer and supplying food to mesopelagic and benthic organisms. Despite its importance, POC concentrations and its variability over basin or global scales have been poorly assessed. (IOCCG, 2008). As well as the gaps in instrument capability, there are also gaps in algorithm development. Both MODIS and MERIS provide information on the chlorophyll fluorescence signal at 685 nm, but there are few regularly available products providing this to users. MERIS has a band at 709 nm which is useful for detection of intense blooms and floating vegetation (Gower et al., 2008), but again there is no standard product based on this. SeaWiFS, MODIS and MERIS all have bands near 410 nm designed to detect Colored Dissolved Organic Material (CDOM). Quantitative measurements are at present hampered by lack of a robust capability for aerosol characterization and hence estimation of aerosol radiance contribution to the measured signal at 410 nm. This is another potential capability of ocean color satellites that needs development. Separation of different species of phytoplankton has long been a goal of satellite remote sensing. One species group, coccolithophores, are easy to recognize by the bright, bluish-white color, caused by the many microscopic coccoliths shed during a bloom. Surface chlorophyll concentrations are often too low to be used for detection of these blooms. Other species are much harder to separate. There are special cases, for example where cyanobacteria bloom regularly in the Baltic Sea in July, and where Tricodesmium blooms regularly in the Red Sea and other tropical and sub-tropical waters. In other cases, successes are claimed in more conservative separation into “phytoplankton functional groups” (Nair et al., 2008) or into blooms of “large” and “small” cells (Sathyendranath et al., 2004).

3 The Value of Coordinated Data Sets The parameters discussed in Sect. 2 are part and parcel of our definition of climate. The classic definitions of land climates by Köppen and later Köppen-Geiger are based on surface air temperature and precipitation patterns and dates back more than 100 years. A modern review by Peel et al. (2007 and see references therein) discuss the problem of fitting all of the Earth’s land types into these categories. These two variables are very basic to oceanic climates as well. The variables we have included here are not exclusive, but include those oceanographic ones measured by satellites which are important and which already have a good record. Air temperature over the sea, is NOT among them, however, as noted above in discussion of air-sea fluxes. It is likely that we will have to rely on better atmospheric numerical model analyses and use of blending techniques to get global estimates of air temperature above the sea surface and proper atmospheric stability functions for evaluating the turbulent heat fluxes. No fully adequate estimation technique has been found that can be used for all latitudes and seasons. The White Paper by Donlon et al. (2010) and the GHRSST reports present numerous recommendations for international cooperation and further work for this variable. SST is the foundation of estimates of surface winds from scatterometers,

20


air-sea heat fluxes (latent and sensible heat fluxes) and outgoing longwave radiation. Climate studies depend heavily on time series of Sea Surface Temperature evaluations. Precipitation climatologies exist and have benefitted from TRMM radar data, which provides a great example for how a high quality measurement can validate a large body of supporting data (see, for example, the tropical cyclone rain estimates by Lonfat et al. (2004) mentioned in Sect. 2.3). Derivation of sea surface elevation and the energy content of the upper ocean has become quite mature science and the satellite altimeter sensors are at least semioperational since the beginning of the 21st century (Wilson et al., 2010). For a comprehensive view of the Earth, its oceans and climate we need at least all of the variables listed in Sect. 2 and chemical and biological measurements and many in-situ data as well. The approximately 3000 ARGO drifting buoys, which cycle in the top layers of the ocean, provide the supporting information to the satellite measurements for more fully describing the state of the upper ocean, and meeting the requirements of the Global Ocean Data Assimilation experiment, GODAE (Guinehut et al., 2009). Better evaluation of aerosols will be crucial for understanding any changes in the radiation balance at the sea surface. To provide these data for analysis we must coordinate the sampling and management from today onward and not allow missed opportunities to occur. For many measurements we have been in research and learning phase, but from 2010 onward we have enough experience to know what is needed and have the organizations to help coordinate the work internationally. Here we have not dealt with land, the upper atmosphere or the Polar Regions, which in themselves deserve the same attention to give us a comprehensive view of the climate on the Earth. Ocean color does not have a complete and consistent series of sensors in space as noted in Sect. 2.6, but it is emerging as an important measure of the status of the ocean’s health, It is not a measure of climate in the classical sense of Köppen definitions related to atmospheric weather patterns, but it has great importance for understanding the changes that are occurring in the oceans and especially in coastal regions due to the evolution of other variables.

4 The Issues of Concern for an adequate Climate Data Record in the future Our workshop focused on 4 main issues and they will now be discussed in turn:

4.1 Sampling Many of the instruments to obtain high resolution and high accuracy data, and all of the microwave instruments to date, have been mounted on polar orbiting satellites or satellites in other low Earth orbits, both because of power concerns at launch and for obtaining high resolution observations. (Visible and IR data can now be obtained

2


21

from geostationary satellites at 40,000 km at resolutions of the order of kilometers, but that was not so in the early days, and boosting the large antennas needed for microwave sensors to geostationary heights have not been attempted yet.) To sample the global ocean from geostationary heights require 6 satellites, each observing a 60 degree sector. We currently have this system in place, mostly motivated by weather observations, but they have contributed valuable data for climate research as well. A good example is the contributions to the International Satellite Cloud Climatology Project, ISCCP. Sampling of the global ocean by any microwave instrument more than once per day still requires multiple polar orbiting satellites for good coverage. Possible constellations and overlap between instruments abound and some activities are under way. A concept of virtual constellations has emerged through the Council of Earth Observation Satellites (See CEOS, 2008), which refers to post-facto arrangements rather than major planned and coordinated satellite launches, although international coordination is gaining ground. The aim is to ensure continuous time-series and when possible fill the gaps that risk occurring in these series due to political and economic pressures, if only one nation’s agency is solely responsible. The virtual constellation concept is ‘in support of the Group on Earth Observations (GEO, http://earthobservations.org), objectives and as a component of the Global Earth Observation System of Systems (GEOSS). A Constellations is a coordinated set of space and/or ground segment capabilities from different partners that focuses on observing a particular parameter or set of parameters of the Earth system. The CEOS Constellation for Ocean Surface Topography (OST) goal is to implement a sustained systematic capability to observe the surface topography of global oceans from the basin scale to the mesoscale (=100 km). The surface topography from satellite altimeters and the upper-ocean density field from Argo profiling floats (which currently number around 3000) are oceanic analogues to the surface pressure from barometers and the density field from atmospheric profilers. Observations of these two fundamental state variables are necessary for understanding the dynamics of the oceans, assessing their role in climate and developing an operational forecast capability. (Wilson et al., 2010)

A multi-mission, accurate and consistent altimetry data base would give a more complete picture of the ocean surface than would be possible with a single satellite. It also would help to meet the needs of the Intergovernmental Panel on Climate Change to rely on high quality altimeter climate data records. This calls for re-visiting and rigorously reprocessing of the original records, including application of homogeneous algorithms, inter-calibration (during tandem and overlapped periods) with the adoption of internationally accepted standards. The series of satellite altimeters have been operated in de facto constellation, although in the absence of agreements between the various space agencies. However, the issue might become even more challenging in future since more missions are being planned (Fig. 2.3), with broader sensor variety, different data rates, increasing complexity, etc. The use of several satellite altimeters coordinated in a “virtual constellation” would optimize resources in operation and exploitation and would address possible emerging data gaps (Wilson et al., 2010).

22


08

09

10

11

12

13

HY-2A China

14

15

16

17

18

19

20

21

22

HY-2B, -2C, -2D

Saral/AltiKa India/France

ENVISAT ESA CRYOSAT-2 ESA

Sentinel-3A, 3B Europe

Sentinel-3C, -3D

GFO-2 (US Navy) SWOT/WaTER-HM USA/Europe Jason evolution demonstrator

Jason-1 Fr./USA

Jason-3 Europe/USA Jason-CS

Jason-2 Europe/USA In orbit

Approved

Planned/Pending approval

Needed

Fig. 2.3 Ocean Surface Topography Missions constituting a virtual constellation

The CEOS Constellation for Ocean Surface Vector Wind (OSVW) will collect observations of vector winds over the global ice-free oceans from multiple satellites and distribute them within sufficiently short time interval to make them useful for forecasting, but also provide the appropriate products for retrospective analysis and research. With India’s and China’s scatterometers in addition to METOP and Seawinds or a follow on, we could have 4 scatterometers in space simultaneously. Figure 2.4 by Liu et al. (2008) provides a graphic illustration of the optimal spatial coverage and mean re-visit times for various combinations of swaths from these 4 scatterometers, if orbits are ideally timed. The three constellations formally agreed upon by CEOS are the topographic constellation quoted above, and one for precipitation and ocean color, respectively. The proposed precipitation constellation has grown out-of the usefulness of the TRMM radar to tie other precipitation estimates together, working mainly with a calibration function for climate issues. As TRMM aged, a follow-on mission was proposed, the Global Precipitation Measuring mission (GPM); the Japanese Aerospace and Exploration Agency, JAXA and NASA playing important roles with others included. It will have a central major satellite carrying a rain-radar, and would be accompanied by numerous, small satellites providing frequent coverage of the globe with wide-swath scanning microwave radiometers (Fig. 2.5 is an illustration of the GPM concept compared to the TRMM constellation.) A US National Research Council Report describes the GPM mission with focus on NOAA’s role (Committee on the Future of Rainfall Measuring Missions, 2007).

2


23

Fig. 2.4 Top panel: Calculated spatial coverage as a fraction of the Earth’s surface by combinations of current and future scatterometers; Bottom panel: Revisit times for a position on the Earth as a function of how many scatterometers are contributing data

24


Fig. 2.5 Artist’s rendition of the constellation visualized for the Global Precipitation Measuring mission with the very successful TRMM mission, which has already proven the concept

The Virtual Constellation for ocean color is being implemented (see Fig. 2.2). Its objective is to provide calibrated radiances at key wavelength bands. Crosscalibration is an important aspect of a constellation and is currently being used by several ocean color instruments. One can hope that the valuable record from SeaWifs will encourage many nations to contribute an instrument. A possible scenario can be found in the latest report from the International Ocean Colour Coordinating Group, IOCCG, (IOCCG, 2008, and www.ioccg.org) Similar proposals for virtual constellations exist for atmospheric composition and land surface imaging, but are not yet implemented. The scatterometer constellation also requires further coordination. The emerging space programs in India and China have many of the established important climatic parameters in their portfolios, so it is fervently hoped that they will soon be front and center in these virtual constellations with their data. Innovation in sampling techniques of individual instruments and new choice of frequencies or combinations of more than one microwave frequency are likely to develop from these interactions and the all important calibration of any sensor benefits tremendously by access to data from other instruments in the constellation, team meetings and discussions. The USA /NASA soon learned how valuable input from many users could be for improving data and products, so the concept of instrument teams was developed more than 3 decades ago. Currently many other space agencies, notably ESA and JAXA, follow this method of enlarging the user groups.

4.2 Consistency We include in the concept of consistency both accuracy, which refers to the absolute value of a variable and the concept of understandable and systematic errors. The

2


25

uncertainty due to random errors in an individual measurement is not very problematic for climate records, since the averaging done to obtain climatic long-term time-series will tend to eliminate them in the final product. Systematic errors leading to biases are more serious, because they can imply a climatic trend, which may instead be due to varying transmission properties of the atmosphere or degradation of the satellite sensor. A typical example is the biases in sea surface temperature that were discovered for the tropical Atlantic ocean. They were found to be caused by the aerosol clouds emanating from the Saharan desert in seasonal dust storms. Similar effects occur over the Indian Ocean and were well identified after the eruption of the Pinatubo volcano. When the physics is understood and we have available supportive data about the radiative effects of the aerosol in the atmosphere, we can correct for these effects. The difficulty is that the bias errors are often not fully understood from the beginning and lead to unfounded speculation about regional climate changes. Due to the intense concern about climate change and the important role of SST in identifying a warming climate signal, much has been learned about the atmospheric aerosol and its effects on satellite data. In addition to effects on the interpretation of infrared signals for SST, the radiative transfer models that derive sea surface short wave irradiance from measurements of solar irradiance above the atmosphere, so called Top of the Atmosphere, TOA, values, must also include the aerosol effect (e.g. King et al., 1999). Only approximate methods have been used to date due to limited input data on the aerosol. Older data sets will have only approximate aerosol corrections, since direct measurements of the aerosol’s presence were not obtained with older instruments of limited spectral resolution. Early satellite instruments were not always calibrated for absolute values. For example visible radiation was post-calibrated using the reflection of the Earth’s surface in the White Sands desert of New Mexico (e.g. Catherine Gautier, personal communication, 1980s). Such development often required field programs to determine the exact reflectivity of the sands in several spectral bands. Many other satellite instruments have been calibrated post-launch with in situ measurements. The corner reflectors used to test satellite radars is another example. The jungles of the Amazon in Brazil have been used as a black body substitute for calibrations and consistency checks on microwave radiometers in space. Buoy and ship data provide the standard inter-comparison data sets for many of the satellite sensors. Two programs have been instituted to ensure research quality ship observations: the Global Ocean Surface Underway Data (GOSUD) and the Shipboard Automated Meteorological and Oceanographic System (SAMOS). The SAMOS initiative is working to improve access to calibrated, quality-controlled, surface marine meteorological data collected by automated instrumentation on research vessels (primarily) and select merchant ships, while GOSUD focuses on the collection, quality evaluation, and distribution of near surface ocean parameters (salinity and sea temperature) from vessels. The importance of ongoing quality control was illustrated by the discovery of a bias in the ERS-2 scatterometer surface wind vectors compared to the winds from ERS-1and subsequent scatterometers. This was found and corrected by comparison with collocated buoy winds.

26


A most valuable approach for obtaining a consistent climate data record is to plan for overlap between a new satellite instrument in a series and the previous one by arranging orbits such that this can happen in the beginning of a satellite mission. Inter-comparisons of this sort are invaluable. The continuous calibration of all the members of the constellation in the GPM rain-radar program, as illustrated in Fig. 4.3 is a case in point. For altimeters the practice has been to have crossing orbits for some time, where the cross-over points clearly and rapidly show any differences. Coordination of these efforts has been led by the Group on Earth Observations. The intergovernmental Group on Earth Observations (GEO) is a voluntary partnership of governments and international organizations, providing a framework within which to develop new projects and coordinate Earth observation strategies and investments. As of June 2009, GEO’s Members include 79 Governments and the European Commission. In addition, 56 intergovernmental, international, and regional organizations with a mandate in Earth observation or related issues have been recognized as Participating Organizations. GEO Members and Participating Organizations are working towards the realization of a coordinated, comprehensive, sustained Earth observation system of systems called the Global Earth Observation System of Systems (GEOSS). The aim is to enable societal benefits of Earth observations, including advances in scientific understanding in the nine Societal Benefit Areas (Disasters, Health, Energy, Climate, Water, Weather, Ecosystems, Agriculture, and Biodiversity).

4.3 Archiving and Distribution These two topics, archiving and distribution, are closely connected. They depend on institutions that are well established and have solid financial support. Today, much of the archiving is distributed with links between operators and certain groups working on establishing good meta-data records. These give important information about the data so that researchers can make informed decisions about whether and how the data are relevant to their work. This subject is developing as more long-term records become established. We did not as a group expect to propose any changes, but we note here some sites that can be helpful. Several of them have so-called Help-Desks that allow the person seeking information on accessing the center’s data to have direct contact with helpful and knowledgeable persons. This is a crucial service. Currently, there are many efforts to work out differences between sensors in consecutively launched space-borne instruments. This has lead to formation of groups to ensure that the resulting Earth System Data Records (ESDRs) and especially Climate Data Records (CDRs), are consistent in terms of calibration of the instruments and data products. The latter aspect involves comparison of methods and algorithms for evaluation of the raw data. A new NASA sponsored program: Making Earth System data records for Use in Research Environments (MEaSUREs), has several active projects. The program focuses on finding consensus for algorithms, best practices and evaluation of errors and limitations of the data (Martha Maiden and other sponsors of a session at the 2009 Annual Meeting of the American Geophysical Union, personal communication, 2009). This effort ought to include all national agencies investing in spaceborne geophysical measurements.

2


27

The Committee on Earth Observation Science (CEOS) consists of members of most (or all) space agencies, and the forum has had successes such as the Virtual Constellation concept discussed in Sect. 4.1). However, many important variables have not received this attention with full international cooperation yet. A large obstacle is the considerable costs and man-power required for the re-processing to correct a time-series when improvements have been identified and agreed upon. The data records maintained at Institut Francais de Recherche et de L’Exploitation de la Mer, IFREMER, based on the ERS1/2 since the beginning in 1991 have been reprocessed several times as algorithms were improved and bias errors were discovered. NASA has an on-going program of re-analysis, which is illustrated best by the Pathfinder program. The US has the National Climate Data Center, which maintains and distributes all manner of climate-related data. Emphasis on a web-based meta-data system, from which the data sources can be found, even though they may be distributed in different centers, is a good and practical idea. Furthermore, the U.S. maintains several Distributed Active Archiving Center’s, DAACs, of which the one for Physical Oceanography, the PO.DAAC, is of most interest here: http://podaac.jpl.nasa.gov. A continuous and well supported system of centers for climate data around the world is to be encouraged. We must, however, find ways to support these efforts, without allowing them to entrain all resources from the climate research enterprise, especially as the data sets proliferate and algorithms may multiply. Judicious coordination that does not stifle innovation and valuable discovery of problems with the data and yet does not neglect good maintenance and stewardship of already collected data, is the goal. The maintenance issue may seem mundane, but is actually both a complex and demanding undertaking. An obvious requirement is clear and effective communication between the data/computer specialists and the scientists and other users. The optimal functioning of the system is in everyone’s interest, but it is wise to bear in mind the different mind-sets of the many communities that need to work together for the goal of a complete, easy to use and understandable climate data record, which is within reach of anyone who needs to or wishes to know. A good English word for this goal is “transparency”, like a beautiful window into the facts.

5 Possible Future Developments The changing climate of the Earth and the serious consequences for mankind and other living creatures on the planet makes it important that scientists arrive at a reasonable way to work together to constantly improve the climate data records. To face the challenges of climate change, decisions must be based on reliable Earth observation data. On one side there is the need of reprocessing existing archives and on the other to ensure continuity of crucial data sets. Through the Global Monitoring for Environment and Security (GMES) initiative (major details at www.gmes.info) Europe is developing its own system to monitor the state

28


and evolution of Earth’s atmosphere, land, sea and ice. The GMES infrastructure builds on a dedicated constellation of satellites, called the Sentinels, expected to be operational during the 2013–2023 time frame. Satellite remote sensing is a natural focus for international efforts, since every satellite launch requires major national organizations behind them, so that the infrastructure for cooperation and collaboration already exists. Many new satellite missions are already international cooperative ventures with at least two countries involved; for instance the soon-to-be launched Aquarius with Argentina and the US collaborating. Above we mentioned the EUMETSAT and NOAA collaboration for operational polar orbiting satellites, and the geostationary placement of the 6 satellites also has required coordinated planning for many years. The nations that have more recently become mature satellite data collectors, e.g. India and China now have a great opportunity to contribute in significant ways to the improvement in sampling, which is so crucial for climate records. We would like to charge all like-minded readers to join us in fostering this spirit of a truly global community of Earth scientists, who work for a comprehensive network of climate observations that give credence to statements made to governments regarding the most important issue of our time – the changing climate regimes on this planet. Acknowledgements We appreciate the contribution of: Figure 2.2 by Dr. Stanley Wilson. We are grateful for support by the U.S. National Science Foundation, NSF, by NASA and ESA for the PORSEC 2008 meeting and by the Chinese hosts, especially The South China Sea Institute of Oceanology, Chinese Academy of Sciences and the chair of the Local Organizing committee, Dr. Danling (Lingzis) Tang, who made arrangements for the Workshop on Climate Data possible on the last day of the conference in Guangzhou, China, December 6, 2008. K. Katsaros gratefully acknowledges support from an NSF research grant to University of Miami, Rosenstiel School, ATM 0631685.

References Adler RF, Huffman GJ, Chang A, Ferraro R, Xie P, Janoviak 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 Hydromet 4:1147–1167. Bentamy A, Queffeulou P, Quilfen Y, Katsaros KB (1999) Ocean surface wind fields estimated from satellite active and passive microwave instruments. IEEE Trans Geosci Remote Sens 37(5):2469–2486. Bentamy A, Katsaros KB, Mestas-Nuñez AM, Drennan WM, Forde EB, Roquet H (2003) Satellite estimates of wind speed and latent heat flux over the global oceans. J Clim 16(4):636–656. Bindoff NL, Willebrand J, Artale V, Cazenave A, Gregory J, Gulev S, Hanawa K, Le Quere C, Levitus S, Nojiri Y, Shum CK, Talley LD, Unnikrishnan A (2007) In: Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M, Miller HL (eds) Observations: oceanic climate change and sea level, in climate change 2007. The physical science basis, contributions of working group i to the fourth assessment report of the intergovernmental panel on climate change, Cambridge University Press, Cambridge, UK, New York. Bourassa MA, Bonekamp H, Chang P, Chelton D, Courtney J, Edson R, Figa J, He Y, Hersbach H, Hilburn K, Lee T, Liu WT, Long D, Kelly K, Knabb R, Lindstorm E, Perrie W, Portabella M,

2


29

Powell M, Rodriguez E, Smith D, Stoffelen A, Swail V, Wentz F (2010) Remotely sensed winds and wind stresses for marine forecasting and ocean modeling. In: Hall J, Harrison DE, Stammer D (eds) Proceedings of the “OceanObs’09: sustained ocean observations and information for society” conference, vol. 2. Venice, Italy, 21–25 September 2009, ESA Publication WPP-306. doi:10.5270/OceanObs09.cwp.08 Brisson A, Le Borgne P, Marsouin A (2001) Downward long-wave irradiance product manual. Version 1.1. Ocean and Sea Ice SAF, November 2001. CEOS (2008) The earth observation handbook: climate change special edition 2008. http://www.eohandbook.com/. Accessed. Cipollini P, Benveniste J, Bouffard J, Emery W, Gommenginger C, Griffin D, Høyer J, Madsen K, Mercier F, Miller L, Pascual A, Ravichandran M, Shillington F, Snaith H, Strub T, Vandemark D, Vignudelli S, Wilkin J, Woodworth P, Zavala-Garay J (2010) The role of altimetry in Coastal Observing Systems. 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.16 Committee on the Future of Rainfall Measuring Missions/Board on Atmospheric Sciences and Climate, Division of Earth and Life Studies (2007) NOAA’s role in space-based global precipitation estimation and application, report. National Academies Press, Washington, DC, pp 131. ISBN 13: 978-0-309-10298-8 and 10: 0-309-10298-7. Curry JA, Bentamy A, Bourassa MA, Bourras D, Brunke M, Castro S, Chou S, Clayson CA, Emery WJ, Eymard L, Fairall CW, Kubota M, Lin B, Perrie W, Reeder RR, Renfrew IA, Rossow WB, Schulz J, Smith SR, Webster PJ, Wick GA, Zeng X (2004) SEAFLUX. Bull Am Met Soc 85:409–424. Donlon CJ, Robinson I, Casey KS, Vazquez-Cuervo F, Armstrong F, Arino O, Gentemann C, May D, LeBorgne P, Piollé J, Barton I, Beggs H, Poulter DJS, Merchant CJ, Bingham A, Heinz S, Harris A, Wick G, Emery B, Minnett P, Evans R, Llewellyn-Jones D, Mutlow C, Reynolds R, Kawamura H, Rayner N (2009) The global ocean data assimilation experiment (GODAE) high resolution sea surface temperature pilot project (GHRSST-PP), 2007. Bull Am Meteorol Soc 88(8):1197–1213. doi:10.1175/BAMS-88-8-1197 Donlon C, Casey K, Gentemann C, LeBorgne P, Robinson I, Reynold R, Merchant C, LlewellynJones D, Minnett P, Piolle J, Cornillon P, Rayner N, Brandon T, Vazquez J, Armstron E, Beggs H, Barton I, Wick G, Castro S, Hoeyer J, May D, Arino O, Poulter D, Evans R, Mutlow C, Bingham A, Harris A (2010) Successes and challenges for the modern sea surface temperature observing system. 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.24 Ebuchi N, Graber HC, Caruso MJ (2002) Evaluation of wind vectors observed by QuikSCAT/SeaWinds using ocean buoy data. J Atmos Oceanic Tech 19(12):2049–2062. Fairall C, Barnier B, Berry D, Bourassa M, Bradley E, Clayson C, de Leeuw G, Drennan W, Gille S, Gulev S, Kent E, McGillis W, Quartly G, Ryabinin V, Smith S, Weller R, Yelland M, Zhang H (2010) Observations to quantify air-sea fluxes and their role in climate variability and predictability. 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.27 Fu LL, Cazenave A (2001) Satellite altimetry and earth sciences: a handbook of techniques and applications, Academic, San Diego, CA. Fu L, Alsdorf D, Rodriguez E, Morrow R, Mognard N, Lambin J, Vaze P, Lafon T (2010) The surface water and ocean topography (SWOT) Mission. 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.33 Gautier C, Diak G, Masse S (1980) A simple physical model to estimate incident solar radiation at the surface from GOES satellite data. J Appl Met 19:1005–1012.

30


Gentemann CL, Minnett PJ, Sienkiewicz J, DeMaria M, Cummings Y, Doyle JD, Gramer L, Barron CN, Casey KS, Donlon CJ (2009) MISST: Multi-sensor improved sea surface temperature project. Oceanography 22(2):76–87. Global Earth Observing System of Systems (GEOSS) (2010) http://www.earthobservations.org/ geoss.shtml. Accessed. Goni G, Black P, Trinanes J (2003) Using satellite altimetry to identify regions of hurricane intensification, AVISO Newsletter, May, 2003. Goni G, DeMaria M, Knaff J, Sapson C, Ginnis I, Bringas F, Mavume A, Lauer C, Lin I-I, Ali MM, Sandery P, Ramos-Buargue S, Kang K, Mehra A, Chassignet E Halliwell G (2009) Applications of satellite-derived ocean measurments to tropical cyclone intensity forecasting. Oceanography, The Quarterly Journal of the Oceanography Society 22(3):8. Guinehut S, Coatanoan C, Dhomps A-L, Le Traon P-Y Larnicol G (2009) On the use of satellite altimeter data in Argo quality control. J Atmos Oceanic Technol 26:395–402. Gower JFR, King SA, Goncalves P (2008) Global monitoring of plankton blooms using MERIS MCI. Int J Remote Sens 29:6209–6216. IOCCG (2008) Why ocean colour? The societal benefits of ocean-colour technology. In: Platt T, Hoepffner N, Stuart V, Brown C (eds) Report number 7, Dartmouth, Canada, p 144. Jackson DL, Wick GA, Bates JJ (2006) Near-surface retrieval of air temperature and specific humidity using multisensory microwave satellite observations. J Geophys Res 111:D10306. doi:10129/2005JD006431. Jones WL, Schroeder LC, Boggs DH, Bracalente EM, Browb RA, Dome GJ, Pierson WJ, Wentz FJ (1982) The geophysical evaluation of remotely sensed wind vectors over the ocean. J Geophys Res 87:3297–3317. Joyce RJ, Janowiak JE, Arkin PA, Xie P (2004) CMORPH: A method that produces global precipitation estimates from passive microwave and infrared data at high spatial and temporal resolution. J Hydromet 5:487–503. Katsaros KB (1980) The aqueous thermal boundary layer. Bound Layer Met 18:107–127. Katsaros KB, Brown RA (1991) Legacy of the Seasat mission for studies of the atmosphere and air-sea-ice interactions. Bull Am Met Soc 72(7):967–981. Katsaros KB, Forde EB, Chang P, Liu WT (2001) QuikSCAT facilitates early detection of tropical depressions in 1999 hurricane season. Geophys Res Lett 28:1043–1046. King MD et al. (1999) Remote sensing of tropospheric aerosol from space: past, present and future. Bull Am Met Soc 80(11):2229–2259. doi:10.1175/1520-0477(1999)0802.0CO;2. Kubota M, Shikauchi A (1995) Air temperature at ocean surface derived from surface-level humidity. J Oceanogr 51:619–634. Kummerow C, and Coauthors (2000) The status of the tropical rainfall measuring mission (TRMM) after two years in orbit. J Appl Met 39:1965–1982. Legeckis R, Zhu T (1997) Sea surface temperatures from the GOES-8 geostationary satellite. Bull Am Met Soc 78(9):1971–1983. Lin I-I, Wu CC, Pun I-E (2008) Upper ocean thermal structure and the western North Pacific category 5 typhoons Part I. Ocean features and category 5 typhoons’ intensification. Mon Wea Rev 136:3288–3306. Liu WT, Tang W, Xie X, Navalgund RR, Xu K (2008) Power density of ocean surface wind from international scatterometer tandem missions. Int J Remote Sens 29:6109–6116. Liu WT (1984) Estimation of latent heat flux with SEASAT-SMMR, a case study. in North Atlantic. In: Gautier C, Fieux M (eds) Large-scale oceanographic experiments and satellites. Reidel Press, Dordrecht, pp 205–221. Lonfat M, Marks FD, Chen SS (2004) Precipitation distribution in tropical cyclones using the tropical rainfall measuring mission (TRMM) microwave imager: a global perspective. Mon Wea Rev 132:1645–1660. Long DG (2004) Reconstruction of high resolution ocean wind vectors from low resolution scatterometer measurements. Proceedings of the SPIE vol. 5562 Image Reconstruction from Incomplete Data III, Bellingham, WA, SPIE, 196–207.

2


31

Nair A, Sathyendranath S, Platt T, Morales J, Stuart V, Forget M-H, Devred E, Bouman H (2008) Remote sensing of phytoplankton functional types. Remote Sens Env 112:3366–3375. National Research Council (NRC) (2008) Ensuring the climate record from the NPOESS and GOES-R Spacecraft: elements of a strategy to recover measurement capabilities lost in program restructuring; committee on a strategy to mitigate the impact of sensor Descopes and Demanifests on the NPOESS and GOES-R Spacecraft, National Research Council. ISBN13: 978-0-309-12184-2. 190pp. http://books.nap.edu/catalog.php?record_id=12254#orgs. Accessed. Ocean Observation Panel for Climate (OOPC) (2010) http://oc-goes-opc.org/state_of_the_ocean/. Accessed. Peel MC, Finlayson BL, McMahon TA (2007) Updated world map of the Köppen- Geiger climate classification. Hydrol Earth Syst Sci 11:1633–1644. Pinker RT, Lazlo I (1992) Modeling surface solar irradiance for satellite applications on global scale. J App Met 31:194–211. Poulter C, Merchant J, Bingham A, Heinz S, Harris A, Wick G, Emery B, Minnett P, Evans R, Llewellyn-Jones D, Mutlow C, Reynolds R, Kawamur H, Rayner N (2007) The global ocean data assimilation experiment (GODAE) high resolution sea surface temperature pilot project (GHRSST-PP). Bull Am Met Soc 88(8):1197–1213. doi:10.1175/BAMS-88-8-119. Reynolds, RW, Smith TM, Liu C, Chelton DB, Casey KS, Schlax MG (2007) Daily highresolution-blended analyses for sea surface temperature. J Clim 20:5473–5496. Robinson IS (2004) Measuring the oceans from space: theoretical principles and methods of satellite oceanography. Springer, Berlin, Heidelberg, New York, p 646. Rossow WB, Duenas EN (2004) The international satellite cloud climatology project, ISCCP, Web site. Bull Am Met Soc 85(2):167–172. Sathyendranath S, Watts L, Devred E, Platt T, Caverhill C, Maass H (2004) Discrimination of diatoms from other phytoplankton using ocean color data. Mar Ecol Prog Ser 272:59–68. Saunders PM (1967) The temperature of the ocean-air interface. J Atmos Sci 24:269–273. Schluessel P, Emery WJ, Grassl H, Mammen T (1990) On the bulk-skin temperature difference and its impact on satellite remote sensing of sea surface temperature. J Geophys Res 95: 13341–13356. Schmetz J (1991) Retrieval of surface radiation fluxes from satellite data. Dyn Atmos Oceans 16:61–72. Schultz J, Schlüssel P, Graßl H (1993) Water vapor in the atmospheric boundary layer over oceans from SSM/I measurements. Int J Remote Sens 14:2773–2789. Schultz J, Meywerk J, Ewald S, Schlüssel P (1997) Evaluation of satellite-derived latent heat fluxes. J Clim 10:2782–2795. Simpson J, Adler RF, North GR (1988) Proposed tropical rainfall measuring mission (TRMM) satellite. Bull Am Met Soc 69:278–295. Slingo A, Ackerman TP, Allan RP, Kassianov EI, McFarlane SA, Robinson GI, Barnard JC, Miller MA, Harries JE, Russell JE, Dewitte S (2002) Observations of the impact of a major Saharan dust storm on the atmospheric radiation balance. Geophys Res Lett 33:L24817. doi:10.10292006GL027896. Smith CK, Bettenhausen M, Gaiser PW (2006) A statistical approach to Windsat ocean surface wind vector retrieval. IEEE Geosci Remote Sens Lett 3:164–168. Smith TM, Reyonolds RW, Peterson TC, Lawrimore J (2008) Improvements to NOAA’s historical merged land-ocean surface temperature analysis (1880–2006). J Clim 21:2283–2296. Solomon S et al. (eds) (2007) Climate change 2007, the physical science basis. Contribution of working group 1 to the fourth assessment report of the intergovernmental panel on climate change. Cambridge University Press, Cambridge. Stephens GL, Kummerow CD (2007) The remote sensing of clouds and precipitation from space: a review. J Atmos Sci 64:3742–3765. Stoffelen A (1999) A simple method for calibration of a scatterometer over the ocean. J Atmos Ocean Technol 16:275–282.

32


Yu L, Jin X, Weller RA (2008) Multidecadal global flux Data Sets from the Objectively Analyzed Air-Sea Fluxes (OAFlux) Project: Latent and sensible heat fluxes, ocean evaporation, and related surface meteorological variables, OAFLUX project technical report (OA-2008-01), Woods Hole Oceanographic Institution, January 2008. Wentz FJ (2001) Advanced algorithms for QuikScat and SeaWinds/AMSR. IGARSS 2001, Sydney, NSW, Australia. Wentz FJ, Gentemann C, Smith D, Chelton D (2000) Satellite measurements of sea surface temperature through clouds. Science 288:847–850. Wilson S, Parisot F, Escudier P, Fellous J, Benveniste J, Bonekamp H, Drinkwater M, Fu L, Jacobs G, Lin M, Lindstrom E, Miller L, Sharma R, Thouvenot E (2010) The 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, Italy, 21–25 September 2009, ESA Publication WPP-306. doi:10.5270/OceanObs09cwp.92 Zhang HM, Reynolds RW, Lumpkin R, Molinari R, Arzayus K, Johnson M, Smith TM (2009) An integrated global observing system for sea surface temperature using satellites and in situ data: research to operations. Bull Am Met Soc 90:31–38.

Chapter 3

Altimeter Observations of Sea Level and Currents off Atlantic Canada Guoqi Han

Abstract The advent of high-accuracy (a few centimeters) satellite altimetry has provided an unprecedented opportunity for monitoring and investigating sea level and circulation variability off Atlantic Canada. A number of exploratory studies have combined satellite altimetry with hydrographic measurements, other remotesensing data, and ocean model results to understand sea level and circulation variability at various temporal and spatial scales, e.g. seasonal and interannual variability of the Labrador Current, interannual variability of coastal, shelf, and slope sea level, and the Gulf Stream warm core rings. The Labrador Current as part of the coastal currents of the North Atlantic subpolar gyre dominates the circulation along the Atlantic Canadian shelves. The studies of the Labrador Current provide good examples of coastal applications of satellite altimetry, a challenging yet highpriority area facing the international science community. This chapter provides a detailed review on advances in applications of satellite altimetry to the coastal and shelf circulation off Atlantic Canada within the last decade and a brief discussion on challenges and prospects of coastal altimetry. Keywords Satellite altimetry · Sea level · Currents · Volume transport · Eddies · Seasonal and interannual variability · Labrador current · Gulf stream · Fish recruitment · Dispersion · Coastal altimetry · Labrador sea · Gulf of St. Lawrence · Scotian shelf · Newfoundland shelf · Scotian slope

1 Introduction Dominant flow features off Atlantic Canada are the equatorward Labrador Current along the shelf-edge and upper continental slope and to a much lesser strength along the Newfoundland and Labrador coast (Loder et al., 1998). The Labrador Current

G. Han (B) Fisheries and Oceans Canada, Northwest Atlantic Fisheries Centre, St. John’s, NL, Canada e-mail: [email protected]


33

34

G. Han

is part of the coastal current associated with the North Atlantic subpolar gyre. It carries colder and fresher water of Arctic origin and has a strong direct influence on the shelf circulation. The Gulf Stream and the North Atlantic Current interact with the shelf-edge Labrador Current off Newfoundland Slope. The Labrador Current Extension interacts with the Gulf Stream off Nova Scotia. Significant interannual variations in physical environments have been observed off Atlantic Canada, e.g., interannual variability of the Labrador Current transport (Han and Tang, 2001) and of the coastal and shelf sea level off Atlantic Canada (Han, 2002). These variations are attributable to strong influences of large-scale ocean circulation (the subpolar and subtropical boundary currents), changes in surface winds and heat fluxes, and ice formation/melting. Besides, meanders and frontal eddies are pinched from the Gulf Stream/North Atlantic Current, generating significant circulation variability in both time and space and resulting in intensive shelf/slope/deep-ocean interactions and exchanges (Han et al., 2008). All these variations on the interannual and decadal scales are presumably directly, indirectly or partially influenced by the dominant mode of atmospheric variability in the North Atlantic, the North Atlantic Oscillation (NAO), which is defined as the sea level pressure difference between the Azores High and the Icelandic Low (Hurrell, 1995). Historically, knowledge of the physical oceanography off Atlantic Canada has come largely from: tide-gauge data at a few permanent coastal stations; hydrographic data, mostly from fisheries surveys; short-term current measurements mainly by the petroleum industry (e.g. Gregory and Bussard, 1996); and moored measurements from a few dedicated field programs. The Atlantic Zone Monitoring Program (AZMP) launched by Fisheries and Oceans Canada in late 1990s (Therriault et al., 1998) has regularly collected physical, chemical and biological oceanographic data at fixed stations and transects on the Atlantic Canadian coastal seas. Satellite observations are vital in complementing the in situ monitoring program. Ocean models have integrated observational data and provided an improved spatial description of seasonal-mean circulation (e.g. Tang et al., 1996; Han et al., 1997; Han et al., 2008). Hindcast, nowcast, and forecast ocean models have been and are being developed with an emphasis on physical state variables (sea level, currents, temperature and salinity). The advent of satellite altimetry opened a new era for the monitoring and study of sea level and circulation off Atlantic Canada, in spite of various challenges associated with the coastal application of satellite altimetry. Han et al. (1993) used Geosat altimetry data to study surface currents over the Scotian Shelf, providing one of the first-ever examples of applying satellite altimetry to shelf circulation. Since then, a series of applications of satellite altimetry have been carried out over the Atlantic Canadian coastal and shelf seas (e.g. Han et al., 2002; Häkkinen and Cavalieri, 2005). These studies, which often combined satellite altimetry data with in situ observations and ocean model results, significantly advanced our knowledge of sea level and circulation variability off Atlantic Canada. There are other relevant altimetry-based studies for the Northwest Atlantic (e.g. Häkkinen and Rhines, 2004; Brandt et al., 2004; Volkov, 2005), however focusing on deep-ocean currents. The measurement principles and oceanographic applications of satellite altimetry

3

Altimeter Observations of Sea Level and Currents off Atlantic Canada

35

are explained in detail by Fu and Cazenave (2001) and articles highlighting its coastal and shelf applications are also available (Han, 1995; Han, 2006a). Recently, there have been increasing interests in exploring utility of coastal altimetry, with three dedicated international workshops on coastal altimetry by the European Space Agency. This chapter reviews the methods and results from the altimetry-related studies that are clearly focused on the coastal, shelf and slope processes off Atlantic Canada. These studies were carried out by scientists in Fisheries and Oceans Canada in the past decade. We hope that our experiences can benefit the international effort in coastal altimetry. Next we briefly introduce the principle of altimeter measurements and outline the methods used for deriving geostrophic surface currents from altimetry and for calculating volume transport from altimetry and hydrography. A satellite altimeter is a nadir-looking active microwave sensor (Robinson, 2004). Its downward transmitted signal pulse reflects from the sea surface back to an altimeter antenna. The round-trip time and the propagation speed of the electromagnetic waves are used to compute the range between the antenna and the ocean surface, subject to various atmospheric corrections including ionospheric delay and wet and dry tropospheric delays and oceanographic effects such as the sea state bias, inverse barometric response, the elastic ocean tides, solid Earth tides and pole tides. From the altimetermeasured range, the instantaneous sea surface relative to a reference surface, such as an ellipsoid, can be determined if a satellite orbit relative to the reference surface is known. The sea surface topography relative to the marine geoid due to ocean dynamic circulation including the temporal averages can also be obtained. Although the marine geoid is not well determined at dominant spatial scales of coastal and shelf processes, repeated observations can provide a measurement of the temporal variability of the sea surface height since the geoid can be treated as time-invariant for oceanographic applications (Han, 2006a). The computation of the sea surface current, depth-averaged current and volume transport is based on the time independent momentum equation with the nonlinear and horizontal friction terms neglected. The equation for velocity v (positive northward), perpendicular to an east-west transect, is given by − fv = −

1 ∂τ 1 ∂p + ρ0 ∂x ρ0 ∂z

p = g ρ0 ζ + g

0

ρdz

(1)

(2)

z

where x is the horizontal coordinate along the transect positive eastward, f is the Coriolis parameter, p is the pressure, z is the vertical coordinate positive upward with z = 0 at the mean sea level, g is the gravity acceleration, ρ is the density of water, ρ 0 is the reference density, τ is the x-component of the shear stress, and ζ is the altimetric sea surface height referenced to an ocean geoid, corrected for various atmospheric and oceanic effects.

36

G. Han

The surface geostrophic current is determined solely by sea surface slope: v(x) =

g ∂ζ f ∂x

(3)

The geostrophic current at any depth z is given by g g ∂ζ − f ∂x ρo f

v(x, z) =

z

0

∂ρ dz ∂x

(4)

Integrating Eq. (1) over the depth and neglecting the bottom boundary layer, we obtain the depth-averaged velocity V: 1 g ∂ζ + V(x) = f ∂x f

0 −H

b=

1 ∂b dz + ∂x Hf

0 −H

z

∂b 1 dz − τ0 ρ ∂x 0 fH

(5)

g [ρ(x, z) − ρ(z)] ρ0

(6)

where H is the local water depth, b is the buoyancy parameter, τ 0 is the alongtransect component of surface wind stress, and ρ (z) is a reference density obtained by averaging ρ across the transect at a given depth. The cumulative volume transport T from x1 to x2 is given by T=

g f

x2

x1

∂ζ 1 Hdx + ∂x f

x2

H x1

0

−H

∂b 1 dzdx + ∂x f

x2

x1

0

−H

z

∂b 1 dzdx − ρ0 f ∂x

x2

x1

τ0 dx (7)

Satellite altimetry provides an estimate of ζ and thus the surface geostrophic current. Determination of the geostrophic current at depth or of the volume transport requires further information that can be provided by hydrographic data and/or models.

2 The Labrador Current Transport in the Western Labrador Sea 2.1 Seasonal Variability Earlier efforts to simulate the circulation in the Labrador Sea, based on relatively simple dynamics and coarse model resolutions, have resulted in useful information on the transport and sea level variability due to wind forcing. Thompson et al. (1986), using the topographic Sverdrup relationship, obtained a seasonal range of 7 Sv off Labrador, with the transport being mainly along the lower continental slope. Greatbatch and Goulding’s (1989) wind-driven barotropic model of the North Atlantic revealed an increase of the Labrador Sea transport by 5 Sv from July to January. Greatbatch et al. (1990) and Han et al. (2008) studied the influence of the local and remote forcing on the seasonal sea level and current variation

3


37

on the Newfoundland and Labrador Shelf, showing the importance of the North Atlantic wind forcing. From hydrographic measurements Lazier and Wright (1993) estimated an annual range of 4 Sv associated with the Labrador Current over the shelf edge and upper continental slope, with a peak in fall. Han and Tang (1999) introduced the use of altimeter data to the study of the Labrador Current. They investigated seasonal variability of the Labrador Current by combining 3.5 years of TOPEX/Poseidon (T/P) altimeter measurements with concurrent wind data and climatological density data. A linearized momentum equation (see Eq. (1)) was used to compute vertically averaged velocities and depthintegrated volume transports normal to selected sections (Fig. 3.1) across the western Labrador Sea. The sea surface was used as the level of known motion derived geostrophically from the altimetric data. Along-track T/P data from December 1992 to April 1996 were averaged for four seasons: winter (January, February, March), spring (April, May, June), summer (July, August, September), and fall (October, November, December). The wind stresses based on the NCEP/NCAR (National Center for Environmental Prediction / National Center for Atmospheric Research) reanalysis project (Kalnay et al., 1996) were averaged over the same period to

Greenland

3000m

200m 60

r

do

nt re ur

C

bra

Latitude

or

ad

br

La

NA

La

56

or ad r b La a Se

HA

4000m

52

NF Newfoundland

48

−65

−60

−55

1000m −50

−45

Longitude Fig. 3.1 Map showing the study area of Han and Tang (1999). The dotted lines denote the T/P ground tracks. Three transects (thick lines) across the Labrador Sea were selected for analysis: the Hamilton (HA), Nain (NA), and northern Newfoundland (NF) Sections. The location of the Labrador Current (thick arrows) is also depicted. From Han and Tang (1999)

38

G. Han

generate seasonal-mean values and interpolated onto the selected sections. The density was interpolated from Tang and Wang’s (1996) seasonal-mean climatology of temperature and salinity of 1/6o by 1/6o resolution. Error analyses were carried out to estimate the uncertainty in altimetric measurements and geophysical corrections, and in density data. The Labrador Current transport anomalies over the shelf break and upper continental slope have a seasonal range of 5 Sv at the Hamilton Section (Figs. 3.1 and 3.2a). Note that a negative anomaly indicates a stronger southward Labrador Current. Hence, the Labrador Current is stronger in winter and fall and weaker in spring. Over the lower continental slope, the transport has a smaller seasonal range of 3 Sv (Fig. 3.2b). The total transport from the 300-m to 2,500-m isobath has a seasonal range of 6.5 Sv at the Hamilton Section (Fig. 3.2c). The calculations for the Nain and northern Newfoundland Sections (see Fig. 3.1) show that the total transport evolves in a similar way, largest in winter and fall, and smallest in spring (Han and Tang, 1999). However, there is a significant reduction in the seasonal range from 17 Sv at the Nain Section to 6.5 Sv at the Hamilton Section (300-m to 2,500-m isobath). The seasonal range at the northern Newfoundland Section is 5 Sv. Since the transport variability may not be limited to the 2,500-m isobath, the integration of transport was extended seaward to the deepest ocean.

Fig. 3.2 Volume transport anomalies over (a) the upper slope, (b) the lower slope, and (c) the upper and lower slope for the Hamilton Section. The total transport (thick lines) is the sum of the barotropic (crosses), baroclinic (open circles) and wind-driven (dashed lines) components. The standard errors associated with the altimetric sea surface height anomaly (vertical bars) and with the density (vertical lines in the upper right corner) are also shown. From Han and Tang (1999)

3


39

It was found that the seasonal transport range at the Nain and northern Newfoundland Sections changed little, but the transport at the Hamilton Section increased from 6.5 Sv to 10 Sv. The decrease of the seasonal transport range from the Nain to Newfoundland Sections may be related to the circulation pattern in the southwestern Labrador Sea (Lazier, 1994; Tang et al., 1996).

2.2 Interannual Variation of Volume Transport Whereas Han and Tang (1999) studied seasonal variability in the Labrador Current using climatological hydrographic and altimetric data, Han and Tang (2001) investigated its interannual variability using data from a WOCE (World Ocean Circulation Experiment) hydrographic section along with T/P altimeter data under the geostrophic approximation. Two ascending altimeter tracks (Fig. 3.3) were

Fig. 3.3 Map showing the study area of Han and Tang (2001): HB and HBN are two T/P ascending ground tracks that straddled the AR7W section in the western Labrador Sea. The five open circles on HBN indicate the outer limits of integration in the calculation of the volume transport, from the deepest sea westward at an interval of 25 km. The corresponding outer limits for HB (for clarity only the easternmost and westernmost locations are shown) are also shown. From Han and Tang (2001)

40

G. Han

Fig. 3.4 Temporal variation of the Labrador Current transport anomalies (west of the open circles in Fig. 3.3). The total transport anomaly (thick lines) is the sum of the barotropic (circles) and baroclinic (squares) components. The standard errors associated with the barotropic and baroclinic components are shown as vertical lines. Adapted from Han and Tang (2001)

selected that approximately overlap the WOCE hydrographic section, which was sampled in late spring/early summer each year from 1993 to 1999. While the instantaneous CTD sections were used, the altimetric data were averaged for April–September. The resulting cumulative southward transports for the time series from 1993 to 1999 are presented in Fig. 3.4. Here a positive anomaly corresponds to a larger than average southward transport. The interannual range of the total transport is about 6 Sv, which is comparable to the seasonal range of 10 Sv (Sect. 2.1). The years 1993, 1994, 1995, and 1999 have larger than average transports; while the years 1996 and 1998 have smaller than average transports. The barotropic and baroclinic transports are nearly out of phase. As a result, the total transport anomaly has interannual variations smaller than either of the two components. The total transport in the Labrador Sea circulation is positively correlated with the winter NAO index (Han and Tang, 2001). Häkkinen and Rhines (2004) showed similar results using altimetry and current meter data.

3 Sea Level in the Gulf of St. Lawrence Seasonal and interannual variability of sea level in the Gulf of St. Lawrence (GSL) was examined by Han (2004a), using T/P sea surface height anomalies from 1992 to 1999. A local tidal model (Lambert et al., 1991) on a 0.25-degree grid was used for the tidal correction. The altimetric sea surface height anomalies were

3


41

Table 3.1 Annual harmonics from in situ tide-gauge measurements and T/P data. The phase indicates the year day (month) when the annual sea level is highest. From Han (2004a) Tide Gauge

Sept-iles Renard River Charlottetown Lower Escuminac

T/P

Amplitude (cm)

Phase

Amplitude (cm)

Phase

4 3.5 4.3 3.5

253 (Sep) 257 (Sep) 317 (Nov) 268 (Sep)

3.8 3.6 5 3.7

248 (Sep) 255 (Sep) 309 (Nov) 287 (Oct)

analyzed using a modified response analysis method (e.g. Cartwright and Ray, 1990; Han et al., 2002), in order to separate an annual cycle from variations at alias frequencies of oceanic tides. The annual cycle has average amplitude of 5 cm and peaks in September– November (Han, 2004a). The T/P-derived annual cycle agrees approximately with tide-gauge measurements (Table 3.1), with the root-mean-square (RMS) amplitude and phase differences of 0.4 cm and 11 d, respectively. The agreement points to the ability of the T/P altimetry’s observing seasonal cycle in the GSL. The seasonal-mean sea level anomalies, after the annual cycle and tidal residuals (including the semi-annual cycle) were removed from altimetric data, were calculated and smoothed with a temporal five-point moving filter for three regions: the northeastern Gulf, the Laurentian Channel (depicted as thick segments in Fig. 3.5), and the southern Gulf (Han, 2004a). The altimetric sea level in the GSL had a range of 5–10 cm and was generally higher during the period of 1996–1997. However, there were notable regional differences in terms of magnitude and timing. Tide-gauge data (adjusted for the inverse barometric effect) averaged at the three stations in the southern gulf for the same period exhibited similar interannual variations, but with an additional rising trend (Han, 2004a). The T/P altimeter measures a geocentric height, while the tide gauge measures the sea level relative to the local seabed. The removal of the effect of the post-glacial rebound from the tide-gauge data (Fig. 3.6, thick solid line) leads to better agreement with the altimetric observations (averaged on Track 033 and 126 in the southern gulf) both in magnitude and in phase. The rising trend from the tide gauges is estimated to be 0.8 cm/year, much larger than 0.2–0.4 cm/year from Tushingham and Peltier’s (1991) model and from longer tide-gauge data (e.g. Douglas, 1991). A number of factors of both oceanographic and atmospheric origins may contribute to the GSL sea level variability on the interannual scale (Han, 2004a). The GSL circulation is not only a response to the freshwater run-off from the St. Lawrence River and to regional atmospheric forcing, but also an integrated part of the northwestern North Atlantic coastal currents: Labrador Shelf water through the Strait of Belle Isle (Petrie et al., 1988) and Labrador Slope water through the Laurentian Channel. Instead of the fresh water runoff and regional surface forcing, the variation of the Labrador Current transport and the fluctuation of the Gulf Stream position may be the primary cause for the interannual sea level variability in the GSL (Han, 2004a).

42

G. Han

Fig. 3.5 Map showing the study area of Han (2004a). Open circles are the locations of coastal tide-gauge stations. The thin lines stand for the T/P ground tracks. The 100-m, 200-m, and 500-m isobaths are also shown. AC: Anticosti Channel; AI: Anticosti Island; CT: Charlottetown; EC: Esquiman Channel; JCS: Jacques Cartier Strait; LE: Lower Escuminac; MB: Miramichi Bay; MI: Magdalen Islands; MS: Magdalen Shallows; P.E.I.: Prince Edward Island; RR: Renard River; SBI: Strait of Belle Isle; SI: Sept-isle. From Han (2004a)

Fig. 3.6 Seasonal-mean sea-level anomalies from coastal tide-gauge data (dashed thick line: before de-trending; solid thick line: after de-trending) and altimetry (thin line) averaged in the southern GSL. From Han (2004a)

3


43

4 Labrador Current Over the Newfoundland Slope 4.1 Surface Circulation Variability Petrie and Anderson (1983) estimated mean flows and transports, and their fluctuations, on the Newfoundland Shelf and Slope from various data sources. There have been a number of modeling studies focused on the shelf scale features, e.g. Greenberg and Petrie’s (1988) barotropic model for the mean circulation on the Grand Bank and Han et al.’s (2008) baroclinic circulation model for the monthly-mean circulation. Han (2006b) used T/P sea-surface height data for the period from mid-1992 to early 2002 to investigate circulation variability across the Newfoundland Slope south and east of Newfoundland. An along-track low-pass Butterworth filter with a width of 46 km was applied to the sea surface height data. From the T/P SSH anomalies, geostrophic surface current anomalies perpendicular to the track (positive westward) were derived. The estimated geostrophic current anomalies have an RMS error of ~4 cm/s. At the crossover locations of the descending and ascending tracks the total (both directional components) current anomalies were calculated. Total RMS current variability (Fig. 3.7) increases offshore (Han, 2006b). Typical values are 15–30 cm/s over the southwestern (SW) and northeastern (NE) Slopes and 30–50 cm/s over the southeastern (SE) Slope. Differences between the upper (200–1,000m) and lower (1,000–3,000m) slope are not significant for the NE Slope. On average, there appears to be no systematic difference in ratios between the descending and ascending tracks. The total current variability over the Newfoundland Slope may be estimated by a factor of 1.5 from the cross-track current variability of either the descending or ascending tracks. There are significant variations at various time scales (notably seasonal and interannual changes) in the current anomalies (Fig. 3.8) at crossovers I (water depth of about 200 m) and D (water depth of about 1,000 m) (See Fig. 3.7 for locations). The altimetric current anomalies have substantial along-isobath and cross-isobath components, though the model mean currents are essentially along the isobath (Han, 2006b). The along-isobath current at I is directed equatorward year round, while that at D reverses from time to time. The equatorward current can be up to 70 cm/s at D and 30 cm/s at I.

4.2 Interannual Variability Significant interannual variation of the along-isobath currents is evident in Fig. 3.8. Further, Han (2006b) calculated monthly-mean unit-depth volume transport to nominally represent the strength of the Labrador Current. After annual and semiannual cycles were removed, the monthly means were averaged to obtain seasonal means by season and year, which were then smoothed by five-point moving filter.

44

G. Han

Fig. 3.7 Variability of the altimetric current anomalies at crossovers from 1992 to 2002 plotted as twice the RMS values. The 200-m, 1,000-m, 2,000-m, 3,000-m and 4,000-m isobaths are also shown. From Han (2006b)

There are substantial interannual changes besides the seasonal cycle in the nearsurface Labrador Current over the Newfoundland Slope. The Labrador Current was intensified on the NE and SE Newfoundland Slope in 1996/1997 and on the SW Newfoundland Slope in 1997, consistent with an enhancement of the baroclinic Labrador Current off central Labrador in 1995/1996 (Han and Tang, 2001) and a Labrador Current pulse traveling through the Scotian Slope in 1997/1998 (Han, 2002). Han and Tang (2001) also found that the baroclinic Labrador Current to be negatively correlated with the NAO.

3


45

Fig. 3.8 Time series of along-isobath (Vt, thin line) and cross-isobath (Vn, thick line) currents at crossovers I and D on the northeast and southeast upper slopes of the Grand Bank, respectively. See Fig. 3.7 for exact locations. The T/P curves are shifted to match model means (dashed lines). From Han (2006b)

Circulation model results (Han et al., 2008) showed the bifurcation of the mean Labrador Current north of Flemish Pass (J in Fig. 3.7) with one branch eastward north of the Flemish Cap (O in Fig. 3.7) and the other southward through Flemish Pass. The T/P results indicate the eastward transport north of the Flemish Cap is positively correlated with the westward flow south of the Cap on the interannual scale, both representing the same anticyclonic partial eddy (Han, 2006b). They were weaker in the mid-1990s and stronger in early and late 1990s, nearly out of phase with the shelf-edge Labrador Current, which is consistent with Pickart et al.’s (1999) finding on the interannual variability of the slope water system.

46

G. Han

4.3 Interannual Variability and White Hake Recruitment White hake (Urophycis tenuis) is a temperate demersal fish species. On the Grand Bank of Newfoundland, it is distributed mainly to the southwest where water temperatures are warmest. Survey data from the 1970s to 2000s indicate that mature adult females aggregate on the southwestern Grand Bank slope in spring, firstyear juveniles settle on the southern, shallow Grand Bank in autumn, and an extremely strong recruitment of the 1999 year class. Han and Kulka (2009) studied the dispersion of egg, larvae and young juveniles in general and explained the strong recruitment of the 1999 year class in particular using dispersal models based on reconstructed circulation fields from an ocean model and satellite altimeter data. For an average year, climatological monthly-mean circulation fields from a finiteelement model (Han et al., 2008) were used for the dispersal modelling (Han and Kulka, 2009). Altimetry-derived current anomalies in conjunction with the finiteelement model solution were used to reconstruct monthly-mean circulation fields for 1999. The monthly geostrophic current anomalies for April–September of 1999 were derived from weekly satellite altimetric sea level anomalies. The satellite data provide surface current anomalies only. To estimate subsurface current anomalies, two scenarios were considered: (1) the current anomalies are constant in the vertical; and (2) they are linear in the vertical, with the bottom current equal to zero. For each scenario the vertical profiles of the current anomalies were added onto the long-term model mean current to obtain the total currents for April–September of 1999. The dispersion simulations treated eggs, larvae and juveniles as particles that follow the water movement under monthly-mean circulation and M2 tidal currents (Han and Kulka, 2009). Survey data at various life stages of white hake were used to determine regions for release of eggs and for settling of juveniles. The major release region was the southwestern slope region from the 100-m to 500-m isobath and from 56◦ W to 50◦ W. The nursery ground was defined as the area from the 100-m isobath to 45◦ N and from 52◦ W to 50◦ W (Fig. 3.9). The release date was April 1. The tracking time was 180 d, ending on September 29. All three releases at the 50-m depth show a rapid increase of particles in the nursery area nearly immediately (Fig. 3.10a) (Han and Kulka, 2009). The result under the depth-invariant current anomaly scenario has high fluctuations and a large number of particles inside the nursery area during the first 3 months, whereas the result under the depth-linear assumption shows an initial increase followed by a relatively steady decline. The weakened Labrador Current and the increased on-bank flow in 1999 (not shown) are major factors responsible for the increased nursery settlement. Clearly, the availability for settlement in 1999 is much higher than that for the monthly-mean case (i.e., a normal year). For the 1-m release, the contrast between 1999 and a normal year is even more striking (Fig. 3.10b). Note that the first 1–2 months after release are more important when the swimming capability of larvae and young juveniles is low relative to ocean currents.

3


47

Fig. 3.9 Horizontal distribution of particles at the start of tracking on the southwest slope (shaded). The nursery area on the south Grand Bank spans from the 100-m isobath to 45◦ N and from 52◦ W to 50◦ W (polygon). The 100-m, 200-m, 1,000-m, and 3,000-m isobaths are depicted. Adapted from Han and Kulka (2009)

5 Scotian Shelf and Slope 5.1 Scotian Slope Current and Eddy Variability Geostrophic current anomalies derived from T/P altimetry indicate anti-cyclonic eddies over the Scotian Slope from time to time (Han, 2004b). By combining the T/P current and frontal analysis data, 24 snapshots of warm core rings (WCR) over the Scotian Slope were identified for 1999, each of which had an average radius of greater than 75 km with one of the seven tracks closely passing by its centre. The relative vorticity of these WCRs varies from 0.81 l/s × 10−5 to 2.58 × 10−5 l/s with a mean value of 1.47 l/s × 10−5 l/s, consistent with Joyce’s (1984) estimate. The RMS current variability ranges from 34 cm/s to 80 cm/s with an average of 59 cm/s. There is little correlation between the radius and the relative vorticity or between the radius and the RMS current variability. Altimetric results were compared with in situ measurements (Han, 2004b). ADCP (Acoustic Doppler Current Profiler) and CTD (Conductivity-TemperatureDepth) data were collected on a section across a Gulf Stream/WCR front on 65.5◦ W

48

G. Han

Fig. 3.10 The temporal evolution of the percentage of particles within the nursery area in 1999 and a normal year (under the monthly-mean circulation) for (a) the 50-m release and (b) the 1m release. The particles were released on April 1 and tracked for 6 months. Juveniles become demersal usually after September. Adapted from Han and Kulka (2009)

off southwest Nova Scotia (Smith et al., 1999). The ADCP section covered only half of the northern side of the WCR (Fig. 3.11). The best T/P observations of the ring are from Track 071 on September 30, 1999, approximately passing along its major axis. The maximum cross-track speed is about 1.5 m/s, located approximately 60 km to the northeast of the rotational center. The maximum ADCP speed normal to Track 071, located at 65.50◦ W and 40.92◦ N, is about 1.8 m/s, in fair agreement with the altimetric maximum. Increased discrepancies towards the northern edge of the ring are attributed to the location mismatch and apparent elliptisity of the ring. The above comparison is valid despite of the 2-day time difference since the centre of the WCR moved very slowly.

5.2 Insights from T/P-Jason1 Tandem Mission Han (2004c) used T/P-Jason1 tandem data to study the shelf-edge current and kinematics of the Gulf Stream WCRs. From January 2002 to July 2002, the T/P and

3


49

GS Ring : 27−Sep−1999 42

2

0 05

01

40 cm/s 41.5

41

Latitude°N

40.5

40

39.5

39

15

−67

−66

20

−65 Longitude°W

10

38 −68

07

1

Altimetry ADCP

9

38.5

−64

25

−63

−62

30

Fig. 3.11 Comparison of the altimetric cross-track current anomalies (thin arrows) on Track 071 on September 30, 1999 and ADCP data (thick arrows) on September 27–28, 1999. Also depicted are positions of the Gulf Stream rings (solid lines), the shelf/slope front (thick dashed lines), the Gulf Stream northern boundary (dash-dotted lines) and satellite-measured sea surface temperature distribution (gray image) on September 27, 1999. From Han (2004b)

Jason1 missions had the same ground tracks with the latter leading slightly in time, which provides a good opportunity of evaluating the consistency of the T/P and Jason1 data. After a linear interpolation of the T/P data to Jason1 time, RMS sea level variability (Fig. 3.12a and b) was calculated for T/P and Jason1. The sea level variability increases offshore and westward, with RMS magnitude of 5–10 cm over the continental slope and of 50 cm in the deep ocean. The T/P and Jason1 data are highly correlated (Fig. 3.12d). The RMS difference is much smaller than either the T/P or Jason1 variability over most areas (Fig. 3.12c), with an average of 3–4 cm and a limited spatial variation. By assuming the noise in the T/P and Jason1 time series to be uncorrelated and to be of equal variance, the noise was estimated as half the variance of the difference time series, i.e., about 2–3 cm. The signal-to-noise

50

G. Han

Fig. 3.12 (a) Twice the T/P RMS sea level variability; (b) Same as (a) but for Jason1; (c) Twice the RMS difference between T/P and Jason1; (d) T/P and Jason1 correlation coefficients (only those significantly different from zero at the 95% confidence level are shown); (e) Twice the square root of T/P signal-to-noise variance ratio for T/P; and (f) Same as (e) but for Jason1. The 200-m, 1,000-m, 2,000-m, 3,000-m, and 4,000-m isobaths are also shown. From Han (2004c)

3


51

variance ratio (Fig. 3.12e and f) varies from ~100 (very high) in the deep ocean (due to elevated sea level variability towards the Gulf Stream northern boundary) to ~5 (moderate) over some areas of the upper continental slope (Fig. 3.12a and b). A similar analysis carried out for the geostrophic surface current anomalies indicates that the variability increases offshore in both T/P and Jason1 data from 10–20 cm/s to 50–70 cm/s, with the RMS difference smaller and relatively uniform (Han, 2004c). Overall the T/P and Jason1 correlation for currents is lower compared to that for sea level, but remains statistically significant for the majority of locations. The signal-to-noise variance ratio increases from 2 to 5 over the shelf and slope to 20 in the deep ocean. The offshore increase is primarily due to the increased current variability associated with the proximity to the Gulf Stream and with more frequent WCR occurrence. Gulf Stream WCRs were observed during the tandem period (Han, 2004c). T/P and Jason1 currents show consistent anti-cyclonic eddies, approximately located where WCRs were observed in the frontal analysis data. The RMS magnitude of WCR’s rotational currents and the vorticity of its inner core agree well between T/P and Jason1 as well. The RMS current magnitude changes from 39 cm/s to 75 cm/s with an average of 55 cm/s, close to Han’s (2004b) estimate of 60 cm/s. The relative vorticity varies from 0.7 × 10−5 l/s to 1.8 × 10−5 l/s with a mean value of 1.43 l/s × 10−5 l/s, consistent with Joyce’s (1984) and Han’s (2004b) estimates.

5.3 Interannual Variability of Slope Water Circulation By combining T/P geostrophic surface current anomalies with a climatological mean circulation field of a finite element model, Han (2007) reconstructed nominal absolute currents over the Scotian Slope for 1992–2002. The winter-mean surface currents clearly show that both the shelf-edge current and the slope current were strongest in 1998, with a maximum speed of 30 cm/s. The current in 1996 was notably different, with weaker and broader southwestward flows from the 200-m to 4,000-m isobath. The interannual current variability was attributed to the interplay between the Labrador Current strength and the position of the Gulf Stream (Han, 2007).

5.4 Interannual Sea Level Variability off Nova Scotia: Shelf vs. Slope Han (2007) also compared the interannual sea level variations between the Scotian Shelf and Slope. For each cross-slope ground track (see Fig. 3.12) the T/P sea level anomalies at each location were first averaged seasonally and then spatially for the four slope segments based on bathymetry: (i) 200–1,000 m, (ii) 1,000–3,000 m, (iii) 3,000–4,000 m and (iv) 4,000–4,500 m. The seasonal, spatially averaged anomalies were further smoothed using a temporal five-point moving filter.

52

G. Han

Over the shelf edge and upper continental slope between the 200-m and 1,000-m isobaths, the sea level showed a rising trend overall (Han, 2007). Over the lower continental slope and rise there was significant interannual sea level variation, with an increasing magnitude towards deeper waters. Sea level variations on all four tracks showed a significant fall from 1994 to 1996, with a rapid rebound after 1997. Han (2002) reported that the sea level change over the Scotian Slope offshore of the 1,000-m isobath was almost out of phase with that over the Scotian Shelf. The time-latitude plot of sea level for the central Scotian Shelf track 088 (See Fig. 3.12 for location) shows that the sea level structure across the shelf or the slope was quite coherent, but the difference between the shelf and slope was notable in certain periods (Fig. 3.13). In particular, there were large positive onshore slope anomalies across the shelf edge and the upper continental slope in 1997 and early 1998, which indicated intensified shelf-edge currents during this period since the mean shelf-edge flow is directed southwestward (Han et al., 1997). Han (2007) showed that the interannual sea level variability over the Scotian Slope is in part related to the Gulf Stream north-south position (Han, 2007). The Gulf Stream northern boundary and the shelf-slope front were in their most northern

Fig. 3.13 Seasonal-mean T/P sea level anomaly (in cm) distribution in the time-latitude domain for track 088, including both the Scotian Shelf and Slope. The seasonal-mean anomalies are smoothed with a temporal five-point moving filter. The dashed line indicates the location of the shelf break. See Fig. 3.12 for the track location. From Han (2002)

3


53

position in the early 1990s until the end of 1995 (Fig. 3.14a), when the winter NAO index (Osborn, 2004) was higher (Fig. 3.14b) (Han, 2007). An abrupt southward movement of the Gulf Stream occurred from 1995 to 1996, when the NAO weakened dramatically. A typical mean steric height profile across the Scotian Slope is illustrated in Fig. 3.15 based on Tang and Wang’s (1996) temperature and salinity climatology. The sea level increases toward the coast and the deep ocean with the lowest value near the 3,000-m isobath. From the north-south shift of the Gulf Stream and shelf/slope front (and by neglecting any local dynamical adjustment and interaction with the Labrador Current) the location of the lowest sea level would be more onshore in the early 1990s and more offshore in 1996 (Fig. 3.15). As a result, the sea level over the lower continental slope (the shelf) in 1996 would be lower (higher) than that in the early 1990s. Altimetric sea level variability over the shelf was similar to that in tide-gauge data (adjusted for the inverse barometric effect) at Halifax (Fig. 3.16a) (Han, 2002). The tide-gauge data were further de-trended to remove the post-glacial rebound

Fig. 3.14 (a) Annual anomalies of the Gulf Stream northern boundary (square) and the shelf/slope front (circle) positions averaged for 55–65◦ W, (b) the winter (December–March) NAO index, calculated as the normalized winter sea level pressure difference between the Azores High and the Icelandic Low. From Han (2007)

54

G. Han

Fig. 3.15 Sea level profiles along Track 088 across the Scotian Shelf and Slope. The thick solid line is the steric height based on Tang and Wang (1996) annual-mean density climatology. The dash-dotted and dashed lines represent the northern and southern Gulf Stream position scenarios, respectively. The triangles are the spatial averages north and south of the lowest point for the former scenario and the crosses are those for the latter. The grey line is the bottom topography. See Fig. 3.12 for the track location. From Han (2007)

effect. The de-trended tide-gauge data are in approximate agreement with the spatially averaged altimetric observations both in magnitude and in phase (Fig. 3.16b). The correlation coefficient is 0.8 and the RMS difference 1.4 cm. The circulation variability over the Scotian Shelf/Slope is in part a response to the fluctuations of the Labrador Current and the Gulf of St. Lawrence outflow and to shelf-ocean exchange associated with the fluctuation of the Gulf Stream (Han,

3


55

Fig. 3.16 (a) Seasonal-mean sea level anomalies derived from tide-gauge data at Halifax (63.59◦ W and 44.66◦ N) and St. John’s (52.71◦ W and 47.56◦ N); (b) De-trended tide-gauge data at Halifax (solid line) and averaged T/P sea level anomalies nearby Halifax (dashed line). H: highest events. L: lowest events. From Han (2002)

2002). The contributions of the local wind stresses and air pressure to the interannual sea level variability were small (Thompson, 1986) and local atmospheric heat fluxes could not account for the interannual temperature variability over the Scotian Shelf (Umoh and Thompson, 1994). The interannual variability of the St. Lawrence River runoff cannot explain the sea level change in the T/P data and tide-gauge data; Instead, it was conjectured that the interannual sea level variability along the Nova Scotia coast was related to fluctuations of the baroclinic Labrador Current and the Gulf Stream position (Han, 2002). In 1993–1994 when the NAO was strongest, the Gulf Stream was in its most northern position off Nova Scotia (Fig. 3.14a), with the sea level over the Scotian Shelf being the lowest (Figs. 3.13 and 3.16). Afterwards, the Gulf Stream retreated southward, and the sea level over the shelf started to rise. The baroclinic Labrador Current in 1995–1996 was strongest off central Labrador (Fig. 3.4). This strongest current event, called as the cold Labrador Current pulse, did not arrive in the Newfoundland offshore until early 1997, as suggested in St. John’s tide gauge data (a sea level fall in early 1997, Fig. 3.16a). The sea level fall at Halifax had a 9-month lag relative to that at St. John’s. The further penetration onto the Scotian Shelf was complicated by the presence of the deeper Laurentian Channel and direct interactions with offshore warm slope water. Altimetric currents and hydrographic data (not shown) indicated the arrival of the cold Labrador Current pulse off the eastern Scotian Shelf in late 1997. The timing of the cold water arrival corresponds well with that of the altimetric and tide-gauge sea level fall. After the

56

G. Han

passage of the cold water pulse, the coastal sea level remained low (Figs. 3.13 and 3.16) because the Gulf Stream repositioned towards the north with the NAO being intensified (Figs. 3.14 and 3.15; Han, 2007). The magnitude of the interannual variability over the Scotian Shelf in the 1990s was comparable to the range of the annual cycle (Han et al., 2002). While the former seemed to be related to the fluctuations of the Labrador Current strength and the Gulf Stream position, the latter was mainly associated with the thermal expansion and the St. Lawrence River runoff fluctuations.

6 Discussion Atlantic Canadian coastal seas are one of the regions where satellite altimetry has helped significantly advance knowledge of sea level, currents and eddy variability. In particular, the altimetry provides a unique tool for monitoring the seasonal and interannual variability of the Labrador Current volume transport, for studying the characteristics of the Gulf Stream WCRs and their impacts on slope water circulation over the Scotian Slope, and for understanding seasonal and interannual coastal sea level variability and underlying mechanisms. Nevertheless, as for other coastal seas, significant challenges and great potentials remain for exploring full capacity of satellite altimetry off Atlantic Canada. In recent years, coastal altimetry has become one of the major directions of satellite altimetry. The COASTALT project funded by the European Space Agency aims to define and develop a prototype software processor for Envisat and to test it in a few European areas. The PISTACH project, supported by the Centre National d’Etudes Spatiales (CNES) of France, provided reprocessed global Jason2 altimeter data in the coastal zone, to generate high-resolution (~350 m, 20 Hz) along-track measurements (improved continuity from the open ocean up to the coastline). There are other coastal altimetry projects off the United States and Canada supported by the National Aeronautics and Space Administration. Appropriate re-tracking methods must be implemented and evaluated to ensure accuracy. Various environmental corrections such as for water vapor effects remain problematic. There are needs for accurate coastal tide models and for them to be seamlessly integrated with the global deep ocean tide model for de-tiding. Highly variable meteorological and oceanographic conditions on relatively wide Atlantic Canadian shelves make coastal altimetry there even more challenging. For climatic studies on the decadal scale, a longer data record is required. The recent launch of Jason2 will help build longer time series, and more follow-up missions are required. For the shelf circulation, the spatial resolution in the past and present missions is not sufficient. To overcome the spatial resolution issue, wideswath ocean altimetry (Fu, 2003) is to be aboard on the proposed SWOT (Surface Water and Ocean Topography) mission. For the Atlantic Canadian shelves, the presence of ice in winter and spring causes a large portion of missing data. Highresolution circulation models with advanced data assimilation scheme may alleviate these problems.

3


57

Finally, a long-standing challenge is the difficulty to obtain accurate absolute ocean currents from satellite altimetry. To overcome the difficulty it is required to have an accurate marine geoid in addition to accurate altimetric sea surface height. At the present we are constrained to use satellite altimetry mainly for temporal current variability only. This is because the marine geoid (which is time-invariant from the oceanographic perspective) is not accurate enough at spatial scales of O (100 km). The Gravity field and steady-state Ocean Circulation Explorer Mission (GOCE) will provide a marine geoid at 1–2 cm accuracy on the 100 km scale (Drinkwater et al., 2007), which translates to an impressive accuracy of 1–2 cm/s for the altimetry-derived absolute mean current at the mid latitude. With this, we will be in a much better position to monitor and examine many important currents using altimetry, such as the Labrador Current and the Labrador Sea gyre. Assimilation of such altimetric sea surface heights relative to the geoid will then directly improve both the mean circulation and temporal variability in our basin-scale and regionalscale models. In addition, the use of regional ocean circulation model (Foreman et al., 2008; Han, 2009) and regional gravity data may help improve mean sea surface topography and geoid at scale of O (10 km) in the coastal and shelf seas. Acknowledgements Kyoko Ohashi, Jim Helbig and Fraser Davidson provided comments on an early version of this review paper. Helpful suggestions were received from Kristina Katsaros. This work is partially supported by the Canadian Space Agency Government Related Initiative Program and the Canadian Program for Energy, Research and Development.

References Brandt P, Schott FA, Funk A, Martins CS (2004) Seasonal to interannual variability of the eddy field in the Labrador Sea from satellite altimetry. J Geophys Res 109:C02028. doi:10.1029/ 2002JC001551. Cartwright DE, Ray RD (1990) Oceanic Tides from Geosat Altimetry. J Geophys Res 95: 3069–3090. Douglas BC (1991) Global sea level rise. J Geophys Res 96: 6981–6992. Drinkwater MR, Haagmans R, Muzi D, Popescu A, Floberghagen R, Kern M, Fehringer M (2007) The GOCE gravity mission, ESA’s first core earth explorer. In: Proceedings of 3rd international GOCE user workshop, Frascati, Italy, 6–8 November, 2006, ESA SP-627, ISBN 92-9092-9383, pp 1–8. Fu LL (ed) (2003) Wide swath altimetric measurement of Ocean surface topography, wide swath altimeter working group. JPL publication 03-2002, JPL, California. Fu LL, Cazenave A (ed) (2001) Satellite altimetry and earth sciences, a handbook of technique and applications, Academic, San Diego, CA, p 463. Foreman MGG, Crawford WR, Cherniawsky JY, Galbraith J (2008) Dynamic ocean topography for the northeast Pacific and its continental margins. Geophys Res Lett 35:L22606. doi:10.1029/2008GL035152. Greatbatch RJ, deYoung B, Goulding A, Craig J (1990) On the influence of local and North Atlantic wind forcing on the seasonal variation of sea level on the Newfoundland and Labrador Shelf. J Geophys Res 95:5279–5289. Greatbatch RJ, Goulding A (1989) Seasonal variations in a linear barotropic model of the North Atlantic driven by the Hellerman and Rosenstein wind stress field. J Phys Oceangr 19:572–595. Greenberg DA, Petrie BD (1988) The mean barotropic circulation on the Newfoundland shelf and slope. J Geophys Res 93:15541–15550.

58

G. Han

Gregory DN, Bussard C (1996) Current statistics for the Scotian shelf and slope. Can Data Rep Hydrogr Ocean Sci 144:iv + 167 pp. Häkkinen S, Cavalieri DJ (2005) Sea ice drift and its relationship to altimetry-derived ocean currents in the Labrador Sea. Geophys Res Lett 32:L11609. doi:10.1029/2005GL022682. Häkkinen S, Rhines PB (2004) Decline of subpolar North Atlantic circulation during the 1990s. Science 304:555–559. Han G (1995) Coastal tides and shelf circulation by altimeter. In: Ikeda M, Dobson FW (eds) Oceanographic application of remote sensing, CRC Press, Boca Raton, FL, pp 45–55. Han G (2002) Interannual sea level variability in the Scotia-Maine region in the 1990s. Can J Remote Sens 28(4):581–587. Han G (2004a) Sea level and surface current variability in the Gulf of St. Lawrence from satellite altimetry. Int J Remote Sens 25:5069–5088, 2004. Han G (2004b) Scotian Slope circulation and eddy variability from TOPEX/Poseidon and frontal analysis data. J Geophys Res 109:C03028. doi:10.1029/2003JC002046. Han G (2004c) TOPEX/Poseidon–Jason comparison and combination off Nova Scotia. Mar Geodes 27:577–595. Han G (2005) Wind-driven barotropic circulation off Newfoundland and Labrador. Cont Shelf Res 25:2084–2106. Han, G. (2006a) Altimeter surveys, coastal tides and shelf circulation. In: Schwartz ML (ed) Encyclopedia of coastal science. Springer, Dordrecht, pp 27–28. Han G (2006b) Low-frequency variability of sea level and currents off Newfoundland. Adv Space Res 38:2141–2161. Han G (2007) Satellite observations of seasonal and interannual changes of sea level and currents over the Scotian Slope. J Phys Oceanogr 37:1051–1065. Han G (2009) Mean dynamic topography and surface circulation in the Western Labrador Sea and Newfoundland Offshore: Satellite observations and Ocean modelling. Int J Remote Sens (in press). Han G, Hannah CG, Smith PC, Loder JW (1997) Seasonal variation of the three-dimensional circulation over the Scotian Shelf. J Geophys Res 102:1011–1025. Han G, Ikeda M, Smith PC (1993) Annual variation of sea surface slopes over the Scotian Shelf and Grand Banks from Geosat altimetry. Atmos-Ocean 31:591–615. Han G, Kulka D (2009) Dispersion of eggs, larvae and pelagic juveniles of white hake (Urophycis tenuis) in relation to ocean currents of the Grand Bank: a modelling approach. J Northwest Atl Fish Sci 41:183–196. Han G, Lu Z, Wang Z, Helbig J, Chen N, deYoung B (2008) Seasonal variability of the Labrador Current and shelf circulation off Newfoundland. J Geophys Res 113:C10013. doi:10.1029/2007JC004376. Han G, Tang CL (1999) Velocity and transport variability of the Labrador Current determined from altimetric, hydrographic, and wind data. J Geophys Res 104(C8):18047–18057. Han G, Tang CL (2001) Interannual variations of volume transport in the western Labrador Sea based on TOPEX/Poseidon and WOCE data. J Phys Oceanogr 31:199–211. Han G, Tang CL, Smith PC (2002) Annual variations of sea surface elevations and currents over the Scotian Shelf and Slope. J Phys Oceanogr 32:1794–1810. Hurrell JW (1995) Decadal trends in the North Atlantic Oscillation: regional temperatures and precipitation. Science 269:676–679. Joyce TM (1984) Velocity and hydrographic structure of a Gulf Stream Warm-core ring. J Phys Oceanogr 14:936–947. Kalnay E, Kanamitsu M, Kistler R, Collins W, Deaven D, Gandin L, Iredell M, Saha S, White G, Woollen J, Zhu Y, Chelliah M, Ebisuzaki W, Higgins W, Janowiak J, Mo KC, Ropelewski C, Wang J, Leetma A, Reynolds R, Jenne R, Joseph D (1996) The NCEP/NCAR 40-year reanalysis project. Bull Amer Met Soc 77(3):437–471. Lambert A, Billyard AP, Pagiatakis SD (1991) Numerical representation of ocean tides in Canadian waters and its use in the calculation of gravity tides. In: Jentzsch G (ed) Proceedings of the

3


59

workshop on high precision tidal data processing, Bulletin International des Marees Terrestres. 110:8017. Lazier JRN (1994) Observations in the northwest corner of the North Atlantic Current. J Phys Oceangr 24:1449–1463. Lazier JRN, Wright DG (1993) Annual velocity variations in the Labrador Current. J Phys Oceangr 23:659–678. Loder JW, Petrie BD, Gawarkiewicz G (1998) The coastal ocean off northeastern North America: a large-scale view. In: Brink KH, Robinson AR (eds) The global coastal Ocean: regional studies and synthesis. The Sea, vol. 11, chap. 5. John Wiley & Sons, Inc., New York, pp 105–133. Osborn TJ (2004) Simulating the winter North Atlantic Oscillation: the roles of internal variability and greenhouse gas forcing. Clim Dyn 22:605–623. Petrie B, Anderson C (1983) Circulation on the Newfoundland continental shelf. Atmos-Ocean 21:207–226. Petrie BD, Toulany B, Garrett C (1988) The transport of water, heat and salt through the strait of Belle Isle. Atmos-Ocean 26:234–251. Pickart RS, McKee TK, Torres DJ, Harrington SA (1999) Mean structure and interannual variability of the slopewater system south of Newfoundland. J Phys Oceanogr 29:2541–2558. Robinson IS (2004) Measuring the oceans from space, Springer, Berlin, New York, pp 669. Smith PC, Boyce R, Petrie L (1999) Report on C.S.S. Parizeau Cruise 99-028, Bedford institute of oceanography, Dartmouth, NS, Canada. Tang CL, Gui Q, Peterson IK (1996) Modeling the mean circulation of the Labrador Sea and the adjacent shelves. J Phys Oceangr 26:1989–2010. Tang CL, Wang CK (1996) A gridded data set of temperature and salinity for the northwest Atlantic Ocean. Can Data Rep Hydrog Ocean Sci 148:iv + 45 pp. Therriault J-C, et al. (1998) Proposal for a northwest Atlantic zonal monitoring program. Can Tech Rep Hydrogr Ocean Sci. 194:57 pp. Thompson KR (1986) North Atlantic sea-level and circulation. Geophys J R Astr Soc 87:15–32. Thompson KR, Lazier, JRN, Taylor B (1986) Wind-forced changes in Labrador Current transport. J Geophys Res 91:14261–14268. Tushingham AM, Peltier WR (1991) Ice-3G: a new global model of ate pleistocene deglaciation based on geophysical predictions of post-glacial relative sea level change. J Geophys Res 96:4497–4523. Umoh JU, Thompson KR (1994) Surface heat flux, horizontal advection, and the seasonal evolution of water temperature on the Scotian Shelf. J Geophys Res 99:20403–20416. Volkov DL (2005) Interannual variability of the altimetry-derived eddy field and surface circulation in the extratropical North Atlantic Ocean in 1993–2001. J Phys Oceangr 35:405–426.

Chapter 4

Eddy Statistics for the Black Sea by Visible and Infrared Remote Sensing Svetlana Karimova

Abstract Examination of vortical circulation features in the Black Sea surface waters is presented based on satellite optical and infrared images. The sequence of images used comprises the period from September 2004 to December 2008. The images were obtained by the Advanced Very High Resolution Radiometer (AVHRR) and Moderate Resolution Imaging Spectrometer (MODIS) sensors. Our analysis allowed us to gain new insights into Black Sea mesoscale circulation features. In particular, information on the spatial distributions of structures such as the Rim Current meanders, mushroom-like currents, near-shore anticyclonic eddies, and chains of shear eddies was obtained. As a result, a better understanding of Black Sea dynamics has been achieved. Keywords The Black Sea · Seawater circulation · Mesoscale eddies · AVHRR · MODIS · Sea surface temperature · Chlorophyll a concentration · Water-leaving radiance

1 Introduction The Black Sea ranks among the most interesting water bodies of the global oceans. It has an extremely dynamical, mesoscale-dominated circulation and a highly eutrophic ecosystem. Another essential reason for studying the basin is its poor ecological conditions that are the result of its limited water exchange with the adjacent basins, weak vertical mixing, and a significant contamination from river discharges, waste from city and tourist resorts, oil and other discharges from shipping and oil terminals. Because most of the contamination comes from the shore and nearshore regions of the sea, the processes of horizontal mixing and cross-shelf water exchange are of great importance. S. Karimova (B) Space Research Institute of the Russian Academy of Sciences, 84/32 Profsoyuznaya St., Moscow, 117997, Russia e-mail: [email protected]


61

62

S. Karimova

The commonly assumed scheme of the Black Sea general circulation based on decades of hydrographic surveys includes (i) a basin scale boundary current cyclonically flowing along the continental slope (the Rim Current), (ii) the Bosporus, Sakarya, Sinop, Kizilirmak, Batumi, Sukhumi, Caucasus, Kerch, Crimea, Sevastopol, Danube, Constantsa, and Kaliakra anticyclonic eddies in the coastal zone of the sea (Oguz et al., 2005), (iii) near-shore anticyclonic eddies (NAEs) between the Rim Current and the shore (Oguz et al., 1993). Traditionally, this scheme also includes a few semi-permanent cyclonic gyres in the deep part of the sea, but there are some disagreements on the number (2, 3 or more) of these sub-gyres. Moreover, those gyres were not confirmed during the drifter experiment carried out in 1999–2003 (Enriquez et al., 2005; Poulain et al., 2005). Mesoscale circulation in the Black Sea is represented by meanders, anticyclonic and cyclonic vortices, pinched off eddies, vortex dipoles, filaments and jets. Numerous observations demonstrate that the mesoscale variability is very important in the transport of scalars especially in the coastal-deep basin water exchange across the Rim Current. Satellite imagery has made possible the detection of the various nonstationary mesoscale dynamical features that contribute to the exchange. Nearshore anticyclonic eddies are a characteristic circulation feature in this sea. Remote sensing contributed significantly to the studies of this form of mesoscale variability. Seasonal variability was clarified, and the role of eddies in the distribution of chlorophyll a concentration was described. The study of small-scale circulation features in the near-coastal zone of the Black Sea was made possible recently with synthetic aperture radar (SAR) data (Lavrova and Bocharova, 2006; Lavrova et al., 2008). As one can see the satellite observations provide a good resource for increasing our present level of knowledge on the mesoscale circulation in the Black Sea. However the satellite datasets were used in some rather limited way, e.g., for complementing some hydrographic observations (Zatsepin et al., 2003) or for analyzing short-term data sequences (Sur and Ilyin, 1997; Afanasyev et al., 2002). The present study attempts to characterize the Black Sea basin and mesoscale circulation from the Advanced Very High Resolution Radiometer (AVHRR) and Moderate Resolution Imaging Spectrometer (MODIS) images. As a result, a better understanding of the Black Sea dynamics has been achieved and a renewed comprehensive scheme of the general mesoscale circulation has been drawn up. The remainder of this paper is structured as follows. After a short description of the dataset and methodology used we discuss the different types of vortical structures detected in the satellite-derived images (the Rim Current, its meanders and quasi-permanent eddies; NAEs; mushroom-like currents; eddy chains). Finally, conclusions are presented in Sect. 7.

2 Data and Methods The work described is based mainly on the processing and analysis of satellite data (IR and sea color) with a spatial resolution of the order of 1 km and a temporal resolution of a few hours (up to six passes per day). The dataset includes the

4

Eddy Statistics for the Black Sea by Visible and Infrared Remote Sensing

63

following satellite images entirely covering the Black Sea: • AVHRR NOAA Sea Surface Temperature (SST) images obtained since September 2004–December 2008; total number of images is about 3,000; • AVHRR MetOp-2 SST images obtained during January–December 2008; total number is about 100 images; • MODIS Aqua SST, normalized water-leaving radiance (WLR) at 551 nm, and chlorophyll a concentration images obtained from April 2006 to December 2007; total number is about 250 images. The data are provided by the Remote Sensing Department of the Marine Hydrophysical Institute (Sevastopol, Ukraine) (http://dvs.net.ua/). Tracking ocean currents from space is possible in two different ways. The first – direct method – can be applied if there are some tracers in the water such as suspended organic or inorganic matter, ice floes, etc. Unlike infrared and visible data, SAR-data can reflect seawater circulation through surfactant – natural or anthropogenic – slicks. Because of their high surface resolution these data can represent small-scale eddies with diameter of a few kilometers or fragments of the mesoscale eddies. Another way – indirect – consists in the estimation of the current velocity using satellite altimeter data. This method is beyond the scope of this paper (see e.g., Korotaev et al., 2003). The most adequate products to track mesoscale circulation features are infrared thermal images derived from the AVHRR sensor, which is onboard NOAA polar orbiting satellites. Spatial and spectral resolutions of this sensor provide SST fields that allow monitoring vortical structures as small as 20 km in diameter. Other product used is WLR at 551 nm derived from the MODIS sensor onboard the Aqua satellite. As WLR at 551 nm is affected by dissolved and particulate matter present within the water column, the contrasts of WLR fields are subjected to large spatial and seasonal variations. As a result the maximum value of WLR is usually observed in June when coccolithophore bloom takes place. In Fig. 4.1 MODIS WLR at 551 nm and chlorophyll a concentration charts obtained during coccolithophore bloom (Fig. 4.1a and b) and in normal conditions (Fig. 4.1c and d). On June 20, 2006, during the maximum extent of coccolithophore bloom, the prevailing magnitude of WLR was within 1.5–2.5 mW·cm−2 ·μm−1 ·sr−1 (Fig. 4.1a). The typical for the Black Sea basin WLR magnitude can be accessed from Fig. 4.1c presenting the distribution of WLR on July 20, 2006, when the bloom was over. As we can see from Fig. 4.1c, maximum of WLR was hardly exceeding 1.2 mW·cm−2 ·μm−1 ·sr−1 . The third type of products used is chlorophyll a concentration. In general, these images are applicable for the purposes of circulation study only in the near-coastal zone where the pigment concentration is especially high. Two examples are given in Fig. 4.1 providing the possibility to compare the way in which circulation features were manifested in simultaneously obtained WLR and chlorophyll a concentration fields. Nevertheless, sometimes chlorophyll a concentration fields can provide very

64

S. Karimova

Fig. 4.1 Manifestation of the surface circulation features in MODIS WLR and chlorophyll a concentration images during a coccolithophore bloom (a and b respectively) and in the normal conditions (c and d). Two upper images were obtained on June 20, 2006 at 11:10 GMT, the lower ones – on July 20, 2006 at 11:25 GMT. Letters mark the most prominent vortical structures: A – an anticyclonic eddy, B – the Sevastopol quasi-permanent eddy, C – a mushroom-like current, D – the Batumi quasi-permanent eddy, E – the Sinop anticyclonic eddy

important information on the NAEs that cannot be retrieved by any other means (see some details and examples in Sect. 4). The main obstacle associated with using optical imagery is the limited availability of clear sky scenes which does not allow the continuous monitoring of circulation features. In Fig. 4.1 and in all other MODIS images presented hereafter, areas covered by clouds are designated by white color used also as a land mask. In AVHRR SST images cloud cover is manifested in a different way: dense cover is white while in the edges it can be represented by the lowest temperature colors; we use purple and dark blue colors. Sometimes it is really difficult to define whether purple color means water temperature or just the edge of the clouds and this can be confusing while trying to retrieve precise SST data from the images. However, water circulation patterns are visible only in the areas unhindered by clouds, so the purple colors should not be confusing this aspect of the analysis. Extraction of the vortical structures discussed below was performed manually. In thermal images the position of maximum temperature gradient with spiral or circular shape was used as an eddy border; in ocean color images such a border was coincident with flow, which had orbital velocity maximum characterized also by radiance maximum. Automated or semi-automated methods are hardly possible to use, because at times eddy identification is complicated even for a trained person familiar with the Black Sea hydrodynamics background.

4


65

3 Observations of the Rim Current and its Meanders. Quasi-Permanent Eddies The most striking point about the Rim Current is that this basin-scale circulation feature confirmed by numerous hydrological observations is rarely evident in satellite images. It has been shown (see e.g., Gill, 1982) that the eddy kinetic energy would be much larger than that of the original gyre. So, this could be one of the reasons why eddies generated by the Rim Current are much easier to observe than the Rim Current itself. Another reason concerns the spatial scale of the Black Sea basin. Due to comparatively small basin extent in latitude there are relatively homogeneous waters, so the Rim Current waters show very weak contrast and can hardly be detected (unlike e.g., the Gulf Stream). Nevertheless, sometimes in the cold season there is enough thermal contrast between the Rim Current and adjacent waters, so satellite data can provide some opportunities to track the Rim Current especially in the western part of the sea (e.g., see Fig. 4.2a). Due to the mentioned reasons, well-expressed manifestations of the Rim Current meanders in satellite images are rare. The region with the most frequent formation of an anticyclonic meander lie to the south-west of the Sevastopol zone of high hydrodynamic instability. One such meander is presented in Fig. 4.2b, marked with the letter C. In the generalized scheme (Fig. 4.4d) this eddy is mentioned as Western eddy (meander). Both model data and observations manifest Black Sea circulation seasonal cycles (Korotaev et al., 2003; Poulain et al., 2005). Strong winds during the winter season cause an increase in the basin-scale Ekman transport. As a result, the circulation of the Rim Current is more clearly defined and intense during winter than in summer.

Fig. 4.2 Examples of the surface water circulation patterns typical for cold (a) and warm (b) seasons. (a) image captured by AVHRR NOAA-18 on February 25, 2008 at 11:08 GMT; (b) image captured by AVHRR NOAA-18 on September 7, 2006 at 23:33 GMT. Letters mark some mesoscale structures filling the zone attributed to the Rim Current: A – a mushroom-like structure, B – the Rim Current stream, C – the Rim Current meander, D and E – anticyclonic eddies, F – the Batumi quasi-permanent eddy

66

S. Karimova

The eddy activity is more pronounced at different scales in the summer season (Sur and Ilyin, 1997; Zatsepin et al., 2003; Shcherbak et al., 2008). Satellite observations confirm this idea. The typical pattern of the Black Sea surface water circulation in cold season is well represented by this AVHRR NOAA-18 image acquired on February 25, 2008 at 11:08 GMT (Fig. 4.2a). In the image, the Rim Current is manifested as the warmest waters and shown by bright green color. One can see that it goes just along the coastline and very close to it. As has been mentioned above in Sect. 2, white, purple and dark blue fragments in the eastern part of the basin mean cloud cover. In summer, closely packed vortical structures fill this region. Some of them are associated with the meandering Rim Current, while others are produced by atmospheric forcing and buoyancy fluxes (Sur and Ilyin, 1997). Some examples of such mesoscale features and the Rim Current trajectories are presented in Fig. 4.2b. This SST image was obtained by NOAA-18 on September 7, 2006 at 23:33 GMT. In the western part of the image one can observe a large mushroom-like structure (A), adjacent to the Rim Current stream (B) that forms an anticyclonic meander (C) at the Bulgarian coast. In the eastern part there are three anticyclonic eddies of different sizes: the smaller one at the Crimean coast (D), the bigger to the south from the Kerch Peninsula (E), and the Batumi eddy (F) in the south-easternmost part of the basin. Another example of the warm season typical circulation is given in Fig. 4.1a and b. The most prominent circulation features are: an anticyclonic eddy in the vicinity of the Bosporus Strait (A), the Sevastopol quasi-permanent eddy (B), a mushroomlike structure (C, arrows show the centers of the antycyclonic and cyclonic parts), the Batumi quasi-permanent eddy (D), and the Sinop anticyclonic eddy (E). Some interesting results were obtained on so called Black Sea quasi-permanent anticyclonic eddies that originated from the combined effect of hydrodynamic instability of the Rim Current and basin configuration. The Sevastopol, Batumi and Caucasus eddies are traditionally regarded as the greatest and most permanent of the quasi-permanent eddies. Satellite observations have shown that the Sevastopol eddy is not just one distinct anticyclonic vortex as it has traditionally been thought to be (Sur and Ilyin, 1997; Afanasyev et al., 2002; Korotaev et al., 2003; Zatsepin et al., 2003; Oguz et al., 2005; Poulain et al., 2005; Stanev, 2005). Though sometimes this structure looks like a single eddy with 100–120 km diameter, more often it represents a whole system of closely packed eddies, mushroom currents and eddy dipoles. So, in Fig. 4.3 some examples of the Sevastopol eddy modifications are presented: (a) a “classical” onecore eddy – A; (b) two anticyclonic eddies designated by B and C; (c) an eddy (D) and a mushroom-like structure (E); (d) a chain of three anticyclonic eddies (F) and an associated mushroom-like structure (G); (e) two (three?) associated mushroomlike structures (H); (f) an anticyclonic eddy (I) with a number of associated smaller cyclonic ones (J). The Batumi quasi-permanent eddy lifetime is much greater than that of the Sevastopol eddy. Unlike the Sevastopol quasi-permanent eddy, the Batumi one more

4


67

Fig. 4.3 Different vortical structures generated at the place of the Sevastopol quasi-permanent eddy detected in AVHRR images obtained (a) May 29, 2008 at 18:46 GMT; (b) April 17, 2005 at 04:42 GMT; (c) June 2, 2006 at 03:48 GMT; (d) November 4, 2004 at 15:46 GMT; (e) June 3, 2008 at 03:01 GMT. A – a single anticyclonic eddy, B and C – a chain of two anticyclonic eddies, D – an anticyclonic eddy with an attached mushroom-like current (E), F – a chain of three anticyclonic eddies, in which one of the eddies is a part of a mushroom-like current (G), H – three closely packed mushroom-like structures, I – an anticyclonic eddy with a series of attached cyclonic eddies (I)

often has the form of a well-shaped anticyclonic structure (see Figs. 4.1a, D and 4.2b, F). Nevertheless, on the periphery of the Batumi eddy there are also multiple attached eddies and mushroom-like currents. In Fig. 4.4b and partly in Fig. 4.4c one can see how the vortical structures practically visualize the Batumi eddy boundary. Some authors, on the basis of the numerical modeling and satellite altimeter observations, traditionally include in the number of the quasi-permanent anticyclonic eddies some other coastal eddies (e.g., Caucasus, Crimea, Bosphorus, Sakarya, Sinop, Kizilirmak, Kaliakra, Danube and Constantsa) (Korotaev et al., 2003; Oguz et al., 2005; Stanev, 2005). According to the observations presented some of them can hardly be referred to as quasi-permanent ones. The anticyclonic structures observed in the Caucasus near-coastal zone differ greatly in size and location and generate irregularly. So, it is hardly possible to regard any of these eddies as quasi-permanent.

68

S. Karimova

Fig. 4.4 Schemes of different vortical structures detected in AVHRR and MODIS images obtained from September 2004 to December 2008: (a) near-shore anticyclonic eddies; (b) mushroom-like currents; (c) chains of shear eddies. In these images the darker the color the more frequently vortical structures were observed. (d) – generalized scheme of quasi-permanent and most frequent non-stationary vortical structures: A – the Anatolian anticyclonic eddies, B – the Batumi quasipermanent eddy, C – the Caucasus quasi-permanent eddy, CA – the regions of cyclonic vorticity along the Anatolian coast, K – the quasi-permanent shear eddy chain to the south from the Kerch Peninsula, N – near-shore anticyclonic eddies, S – the Sevastopol zone of high hydrodynamic instability, W – the Western quasi-permanent eddy (meander)

The Crimea quasi-permanent eddy is considered to be generated along the southern coast of the Crimean Peninsula (see e.g., Korotaev et al., 2003). Satellite observations revealed that anticyclonic eddies really generated in this region – both as elliptical NAEs and as spiral-like eddies e.g., that marked with D in Fig. 4.2b (see schemes presented in Fig. 4.4). However, most frequently this zone was occupied by the anticyclonic shear eddy chains due to baroclinic instability (Fig. 4.4c). For more details on eddy chains please refer to Sect. 6. As for the eddies located along the southern coast (Bosporus, Sakarya, Sinop, Kizilirmak), it is also not quite true to recognize these four eddies as quasipermanent because usually the Black Sea southern coast has a great number of the well-developed anticyclonic eddies. More detailed information on the eddy chains detected along the Anatolian coast is provided in Sect. 6. The Danube, Constantsa, and Kaliakra anticyclonic eddies were not detected in the observations presented in this chapter.

4


69

4 Near-Shore Anticyclonic Eddies Near-shore anticyclonic eddies (NAEs) form within a zone of coastal anticyclonic current vorticity (anticyclonic convergency zone) between the coast and the midstream of the Rim Current. They stretch along the coast due to lateral friction, so their most prominent feature is an elongated shape. Because of NAEs formation, the zone of coastal anticyclonic current vorticity has a bimodal current regime (i.e. back-and-forth motions along the shore when passing the NAE) (Titov, 2002). NAEs growth gives rise to large meanders that could either detach and propagate in the open sea or stagnate for some time in coastal areas. Their separation from the coast and transformation into open sea eddies could provide horizontal mixing of the upper layer waters and result in deflection of the Rim Current offshore, formation of large meanders of the current around the eddies, and its branching when rounding such features (Zatsepin et al., 2003). It is traditionally considered that NAEs are more pronounced and stable at the Caucasian and Anatolian coasts but they can form along the entire Black Sea coast, however, most frequently along the Caucasian one (Ovchinnikov et al., 1986). The present study shows that NAEs arise only in the regions where (i) the width of the shelf is minimal and (ii) the Rim Current goes closely to the shoreline, i.e. along Caucasian and Bulgarian coasts. Anticyclonic eddies quasi-permanently generated along the Anatolian coast are of special origin, so they are examined in Sect. 6. In Fig. 4.4a a generalized scheme of NAEs detected within the framework of the present study is given. In this scheme all the NAEs observed are marked with a grey ellipse and after that the ellipses are overlapped. As a result areas with frequent NAE formation look in the scheme as black patches. As we can conclude from the scheme, most frequently NAEs generate along the Caucasian coast and southeastern coast of the Crimean Peninsula though sometimes they can arise along the Bulgarian coast. Morphometric parameters of the NAEs detected are as follows: The longer axis of NAEs varies approximately between 30 and 150 km with an average of about 60 km; for the smaller axis, the Figures are 20, 75 and 50 km, respectively. Another discovered characteristic feature of NAEs is that along with a single NAE or chained medium size NAEs, we can often observe them combining in pairs as shown at Fig. 4.5. Letters A and B mark the position of the pair-forming NAEs centers. Usually such pairs include NAEs of about 90–120 km in length.

5 Mushroom-Like Currents Vortex pairs (dipoles) or mushroom-like currents (MLCs) are regularly observed on thermal and colour imagery of the coastal ocean. Many dipole observations have been reported by Fedorov and Ginzburg (1986). It was determined that the formation of a dipole at the sea surface is due to a uniform reaction of the upper layer

70

S. Karimova

Fig. 4.5 Some examples of double NAEs detected in satellite images obtained (a) by AVHRR NOAA-18 on April 16, 2006 at 10:44 GMT; (b) by AVHRR NOAA-15 on April 10, 2008 at 14:11 GMT; (c) by MODIS Aqua (normalized water-leaving radiance, 550 nm) on November 24, 2006 at 10:40 GMT; (d) by AVHRR NOAA-15 on August 4, 2006 at 15:11 GMT. Letters A and B mark the center of the pair-forming eddies

to any spatially localized forcing from the atmosphere (wind-forcing) or due to the oceanic dynamics itself (e.g., instability of currents, pulsing exchange through straits). Dipoles represent the important effective mechanism of horizontal mixing and transport of heat and mass. The horizontal dimensions of MLCs detected differ between 50 km and 250 km with a typical size of about 90 km. In most cases, the width of the “mushroom cap” exceeds the length of the “mushroom stem” by about 20–50%. Coherent dipole structures are often observed along unstable boundary currents, such as the Black Sea Rim Current system, typically excited by density or wind impulses. The present examination of MLCs in the Black Sea basin showed the peculiar spatial distribution of these vortex structures. In Fig. 4.4b a scheme of the observed MLCs location is given. An MLC was approximated by two grey ellipses representing cyclonic and anticyclonic parts of MLC and a grey triangular marking the MLC’ jet position. When different MLCs overlap the colour in the scheme becomes darker. As we can see from the scheme (Fig. 4.4b), most frequently MLCs can be detected in the Rim Current zone along the western coast of the sea with the maximal density in the region to the south-west of the Crimean Peninsula. Such pattern

4


71

of MLCs spatial distribution confirms the fact that the region of the Sevastopol quasi-permanent eddy is a zone of high hydrodynamic instability as has been shown before. Another region of frequent MLCs formation is located along the Caucasian coast. Such a peculiar spatial distribution suggests two different factors of the Black Sea MLC generation, namely: (i) the instability of the current that apparently dominates along the western seacoast and (ii) wind forcing that is great along the Caucasian coast especially affected by north-westerly winds, “bora” events.

6 Eddy Chains In this study we define eddy chains in a wide sense; number of eddies in the chain can be as small as only two. Analysis of satellite images made it possible to subdivide eddy chains into different groups: (i) chains of shear eddies, (ii) eddy chain along the Anatolian coast, and (iii) small-scale eddy chains. The generalized scheme of the first group eddy chains is given in Fig. 4.4c. In this scheme all eddies in the detected chains were represented as circular grey-colour figures. As one can see from the scheme, practically all the chains are generated in the zone of the Rim Current. Chains of shear eddies are considered as horizontal versions of the KelvinHelmholtz instability. These eddies have marked spiral forms and the centres of eddies are located along a straight line. Eddies forming a chain can be both cyclonic and anticyclonic. The chains of shear eddies can originate from different causes: when the currents separate from the shelf or shore, in the cases of a sharp change in the shoreline configuration, at fronts, in areas of local current shear, etc. The region where chains of shear eddies were frequently detected is one to the south of the Kerch Strait. In this region ten eddy chains were detected in thermal images. The chains are situated above the continental slope between 44◦ N and 45◦ N and stretched from east to west. All eddies in these chains are anticyclonic; number of eddies in the chains varied from 2 to 4; diameter of eddies was within 30–100 km. The chains were generated in different seasons but most of them in April and May. The second most frequent region of eddy chain formation is the near-coastal zone along the Caucasian coast. Within a stripe of near-coastal waters of approximately 100 km in width, nine chains of shear eddies were detected. All of them are cyclonic with an eddy diameter between 30 km and 90 km; number of eddies in the chains was 2 or 3. All the chains in this region were observed in the cold season (from September to March). A characteristic feature of this type of eddy chains is their large total length compared to the diameter of eddies forming the chain. Figure 4.6a shows a typical chain of cyclonic shear eddies. This AVHRR NOAA-12 image was obtained on March 13, 2005 at 13:58 GMT. The chain is made up by 3 eddies with diameters not exceeding 60 km while the chain stretches out about 400 km.

72

S. Karimova

Fig. 4.6 Chains of eddies different in origin: (a) chain of cyclonic shear eddies (A) detected in AVHRR NOAA-12 image obtained on March 13, 2005 at 13:58 GMT, (b) anticyclonic eddies of the Anatolian coast (B) and the adjacent cyclonic eddies (C). The AVHRR NOAA-18 image was obtained on December 4, 2007 at 10:26 GMT

Eddies of the Anatolian coast. The schemes shown in Fig. 4.4c and d reveal that all the southern near-coastal zone of the Black Sea is a region of quasi-permanent generation of eddies. The most numerous group among these eddies is made up of anticyclonic eddies with a diameter of 60–100 km. The generalized scheme of their location is presented in Fig. 4.4d where they are marked with the letter A. These eddies have spiral shape and form very closely to the coast; frequently well-formed spiral eddies are observed in the numerous semicircular bays along the eastern part of the coast as it is shown in Fig. 4.6b, letter B; arrows depict the eddy centres. This satellite image was obtained by AVHRR NOAA-18 on December 4, 2007 at 10:26 GMT. As anticyclonic eddies of the Anatolian coast originate under the current baroclinic instability and effects of the complicated shoreline configuration, they represent a special transitional type between NAEs and chains of shear eddies. The sequence of AVHRR images obtained in February 2008 allowed us to follow the evolution of an anticyclonic eddy in the vicinity of the Bosporus Strait. Originally there was some cold water intrusion about 60 km length distinctively marked in the image obtained 20 February 20, 2008 at 08:11 GMT (Fig. 4.7a). During the following 24 h the linear intrusion was curling clock-wise until an enclosed ring was formed (Fig. 4.7b and f). During that period, the speed of the orbital motion was strikingly high and ranged between 1.2 km and 4.0 km per hour. At the same time the speed of movement downstream in the Rim Current was about 1 km per hour. On the third day of the observations it became clear that another anticyclonic eddy of the same size had formed downstream in the Rim Current and the deformation of the first eddy began (Fig. 4.7i). The last image obtained before clouds prevented further observations shows that the first eddy was almost destroyed and the second one continued its motion along the coast (Fig. 4.7p). Another characteristic feature of this region is a cyclonic eddy chain generation that sometimes can be observed to the north from the row of the anticyclonic

4


73

Fig. 4.7 Evolution of an anticyclonic eddy retraced by the sequence of AVHRR images obtained from February 20 to February 25, 2008

eddies. Chains of this type are schematically shown in Fig. 4.4c; in Fig. 4.4d they are marked with “CA”. In Fig. 4.6b such a chain of two cyclonic eddies marked with the letter C is shown. Despite all the differences in their shape and location these anticyclonic and cyclonic eddies are caused by the same mechanism – baroclinic instability. This mechanism of cyclonic vorticity transfer into the interior of the basin was successfully simulated in the laboratory experiments and described by Blokhina and Afanasyev (2003).

7 Conclusions In this paper, we show the potential of MODIS and AVHRR imagery for studying circulation features with spatial scale exceeding 20 km. Our analysis allowed us to gain new results on the spatial distribution of the Black Sea mesoscale vortical structures. The most important of them are summarized below. It was shown that the region to the south-west of the Crimean Peninsula is a quasi-permanent zone of high current instability that results in the formation of the

74

S. Karimova

Sevastopol anticyclonic eddy with numerous attached eddies along with mushroomlike currents, eddy dipoles, etc. The only other eddy that can be regarded as quasipermanent is the Batumi one. The region where NAEs are generated most frequently is along the Caucasian coast. Another possible place of NAE formation is the Bulgarian near-coastal waters (Fig. 4.4a). Near-coastal zone along the Anatolian coast gives rise to the quasi-permanent chains of anticyclonic spiral eddies that should be differentiated both from NAEs and from the chains of shear eddies. Mushroom-like currents and eddy dipoles are most frequently observed in the Rim Current zone to the north of the western and eastern near-coastal waters of the Black Sea (Fig. 4.4b). The region to the south of the Kerch Strait above the continental slope is a zone of quasi-permanent formation of chains of cyclonic shear eddies. Other areas with frequent chains of cyclonic shear eddies arising are near-coastal waters along the Caucasian and Anatolian coasts (Fig. 4.4c). As a result a generalized scheme of quasi-permanent and most frequent nonstationary vortical structures is presented (Fig. 4.4d). In this scheme, there are three well-known quasi-permanent structures: the Batumi (marked with the letter B), Sevastopol (S) and Caucasus (C) eddies and one which has not yet been well studied – Western meander (W). Due to the multiple possible modifications of the Sevastopol eddy it was symbolized by two anticyclonic eddies. The Caucasus and Western eddies were drawn with a dashed line, because the former’s generation was not regular and the latter was a seasonal structure (usually it formed in autumn). Also there are provided the symbols for NAEs (N), Anatolian anticyclonic eddies (A) as well as the Kerch and Anatolian eddy chains (K and CA respectively). The Rim Current can be regarded as flowing around all the anticyclonic structures designated in the scheme from the inner side of the basin. Acknowledgements This work was partly supported by the RFBR grants #08-05-00831, # 1005-00428 and the Russian Federation President grant MK-927.2009.2. Satellite images were processed and provided by S. Stanichny, D. Soloviev, E. Kalinin, the Marine Hydrophysical Institute, Sevastopol, Ukraine.

References Afanasyev Y, Kostianoy A, Zatsepin A, Poulain P-M (2002) Analysis of velocity field in the eastern Black Sea from satellite data during the Black Sea ’99 experiment. J Geophys Res 107(C8):3098. doi:10.1029/2000JC000578. Blokhina M Afanasyev Y (2003) Baroclinic instability and transient features of mesoscale surface circulation in the Black Sea: laboratory experiment. J Geophys Res 108(C10):3322. doi:10.1029/3003JC001979. Enriquez C, Shapiro G, Souza A, Zatsepin A (2005) Hydrodynamic modelling of mesoscale eddies in the Black Sea. Ocean Dyn 55:476–489. Fedorov K, Ginzburg A (1986) “Mushroom-like” currents (vortex dipoles) in the ocean and in a laboratory tank. Ann Geophys 4:507–516. Gill A (1982) Atmosphere-ocean dynamics. Academic, Orlando, FL, 662 pp. Korotaev G, Oguz T, Nikiforov A, Koblinsky C (2003) Seasonal, interannual and mesoscale variability of the Black Sea upper layer circulation derived from altimeter data. J Geophys Res 108(C4):3122. doi:10.1029/2002JC001508.

4


75

Lavrova O, Bocharova T (2006) Satellite SAR observations of atmospheric and oceanic vortex structures in the Black Sea coastal zone. Adv Space Res 38(10):2162–2168. Lavrova O, Mityagina M, Bocharova T, Gade M (2008) Multisensor observation of eddies and mesoscale features in coastal zones. In: Barale V, Gade M (eds) Remote sensing of the European Seas. Springer, Heidelberg, pp 463–474. Oguz T, Latun V, Latif M, Vladimirov V, Sur H, Markov A, Ozsoy E, Kotovshchikov V, Eremeev V, Unluata U (1993) Circulation in the surface and intermediate layers of the Black Sea. Deep Sea Res 40:1597–1612. Oguz T, Tugrul S, Kideys AE, Ediger V, Kubilay N (2005) Physical and biogeochemical characteristics of the Black Sea. In: Robinson AR, Brink KH (eds) The Sea, vol 14B, Chapter 33. Harvard University Press, Cambridge, MA, pp 1331–1369. Ovchinnikov I, Titov V, Krivosheya V (1986) New data on temporal variability of currents from long-term measurements at a stabilized buoy on the Black Sea shelf. Dokl Akad Nauk SSSR 5:286. Poulain P-M, Barbanti R, Motyzhev S, Zatsepin A (2005) Statistical description of the Black Sea near-surface circulation using drifters in 1999–2003. Deep-sea research, part I: oceanographic research paper, vol 52, no 12. December 2005, pp 2250–2274. Shcherbak S, Lavrova O, Mityagina M, Bocharova T, Krovotyntsev V, Ostrovskii A (2008) Multisensor satellite monitoring of seawater state and oil pollution in the northeastern coastal zone of the Black. Int J Remote Sens 29(21):6331–6345. doi:10.1080/01431160802175470. Stanev E (2005) Understanding Black Sea dynamics: Overview of recent modeling. Oceanography 18(2):56–75. Sur H, Ilyin Y (1997) Evolution of satellite derived mesoscale thermal patterns in the Black Sea. Prog Oceanogr 39:109–151. Titov V (2002) Morphometric parameters and hydrophysical characteristics of near-shore anticyclonic eddies in the Black Sea, Meteorol Gidrol 4:67–73. Zatsepin A, Ginzburg A, Kostianoy A, Kremenetskiy V, Krivosheya V, Stanichny S, Poulain P-M (2003) Observations of Black Sea mesoscale eddies and associated horizontal mixing. J Geophys Res 108(C8):3246. doi:10.1029/2002JC001390.

Chapter 5

Passive Ocean Remote Sensing by Near-Space Vehicle-borne GPS Receiver Wen-Qin Wang, Jingye Cai, and Qicong Peng

Abstract To avoid possible ocean disasters, e.g., the massive 2004 Indian tsunami, effective prediction techniques are required. Inspired by the advantages of nearspace vehicles, this chapter investigated the system concept of passive ocean remote sensing by near-space vehicle-borne GPS receivers. This system involves placing passive receivers inside a near-space vehicle, which operates in conjunction with GPS illuminators. Rather than emitting signals, it relies on opportunistic illuminators. This is particularly attractive, because it is desirable for such a sensor to provide persistent monitoring but without impacting existing electronic systems. Note that near-space is defined as the space region between 20 km and 100 km, which can offer many new capabilities that are not accessible for satellites and airplanes. The system models, signal processing algorithms, and motion compensation are described, along with technical challenges and possible solutions. Keywords Ocean remote sensing · Near-space vehicle · Passive remote sensing · Environment monitoring · GPS radar

1 Introduction In recent years, the frequency of ocean disasters has shown rapid increase (Kouchi and Yamazaki, 2007; Shinde and Gahir, 2008). Examples of this trend are related to floods, tsunamis and hurricanes (Tralli et al., 2005). The tsunami that killed thousands of people in the coastal area of India, Indonesia, Thailand and Srilanka has brought the awareness that we cannot any more take ocean disasters as something unavoidable. Methods and strategies along with effective techniques have to be developed to predict ocean disasters, without impacting normal human activities. For these reasons, ocean disaster monitoring has received much recognition (Oo et al., 2005; Meyer et al., 2006; Bovolo and Bruzzone, 2007). W.-Q. Wang (B) School of Communication and Information Engineering, University of Electronic Science and Technology of China, Chengdu, P. R. China, 610054; Key Laboratory of Ocean Circulation and Waves, Chinese Academy of Sciences, Qingdao, P. R. China, 266071 e-mail: [email protected] D. Tang (ed.), Remote Sensing of the Changing Oceans, C Springer-Verlag Berlin Heidelberg 2011 DOI 10.1007/978-3-642-16541-2_5,

77

78

W.-Q. Wang et al.

When electromagnetic signals scatters off ocean surface, the scattering process changes the characteristics of the propagation signals in a way that is dependent on the reflecting surface. These changes contain information on the sea surface waves and indirectly on the near-surface meteorological conditions. This was quantitatively demonstrated over half a century ago (Cox and Munk, 1954). This fact has encouraged the scientific community in the last years to consider the passive bistatic retrieval of data, which can potentially provide more information compared to monostatic approaches. An alternative way is using global positioning systems (GPS) to sense sea roughness (Garrison et al., 1998). The possibility of using GPS signals reflected off ocean surface and received by air- or spaceborne sensors for different measurements has been investigated by several authors (Zavorotny and Voronovich, 2000; Heise et al., 2008). These preliminary results open a door to scatterometric applications for wind direction retrieval, as well as other applications (Garrison et al., 1998, 2002; Gleason et al., 2005). However, the persistent coverage that we are desperately wanted for ocean remote sensing is still not available from spaceborne and airborne receivers. It is impossible for our limited non-geosynchronous earth orbit (GEO) satellites to provide a constant, staring presence on a timescale of days, weeks, or months over a selected target or an area of interest (Meyer et al., 2006; Suchandt et al., 2006). Most low earth orbit (LEO) satellites have a specific target in view for less than ten minutes at a time and a low revisiting frequency. Using multiple satellites for persistent coverage is prohibitive expensive. In contrast, conventional airplanes cannot fly too high. As a consequence, physical limitations due to orbital mechanics and fuel consumption prevent long-term persistence for both spaceborne and airborne receivers. Fortunately, these requirements can be satisfied with near-space vehicleborne receivers (Wang et al., 2007). Near-space defined as the space region between 20 km and 100 km is too high up for conventional airplanes, and too low for LEO satellites (Allen, 2006). However, the vehicles operating in near-space can offer many capabilities that are critical to emerging ocean remote sensing applications, but not accessible for satellites and airplanes. Literature search in near-space reveals that most published work concentrated on designing near-space vehicles, including balloons and maneuvering vehicles (Marcel and Baker, 2007). Little work on the use of near-space vehicle-borne sensors for communication and navigation applications has been reported (Guan et al., 2007). Even less work has been placed on near-space vehicle-borne remote sensing applications (Wang, 2007a; Galletti et al., 2007). This chapter investigated the system concept of passive ocean remote sensing by near-space vehicle-borne GPS receivers. This system involves placing passive receivers inside a near-space vehicle, which operates in conjunction with GPS illuminators. Rather than emitting signals, this passive remote sensing system relies in opportunistic illuminator. This is particularly attractive, because it is desirable for such a sensor to provide persistent monitoring but without impacting normal human activities, health and commerce (Bindlish et al., 2009; Shokr et al., 2008; Boukabara et al., 2007). The remaining sections are organized as follows. In Sect. 2, the motivations of this chapter are formed. The system concept and signal model of near-space

5

Passive Ocean Remote Sensing by Near-Space Vehicle-borne GPS Receiver

79

vehicle-borne GPS passive remote sensing are formulated in Sect. 3. Taking tsunami detection as an example, its application in ocean remote sensing is described in Sect. 4. Next, the power budget analysis is investigated in Sect. 5, followed by technical challenges and possible solutions in Sect. 6. Finally, Sect. 7 concludes the whole chapter, along with a short discussion of some future work.

2 Background Existing ocean remote sensors are scatterometers, altimeters, synthetic aperture radars, and radiometers. (1) Scatterometer (Nie and Long, 2008): A scatterometer is an active radar measuring the normalized backscatter cross section of the ocean surface. An example of a scatterometer is the SeaWinds instrument on the QuikSCAT satellite. However, it may produce very inaccurate results under rain condition due to strong attenuation, volume scattering, and perturbation of sea surface. (2) Altimeter (Kurtz et al., 2008): An altimeter is an active radar used to determine the altitude of the ocean surface. The pulse delay and echo amplitude from an altimeter can be used as a measure of the ocean surface measurement. An example altimeter for ocean topographic measurements is the TOPEX. However, altimeter using single frequency cannot remove the effects of ionospheric delay on the radar pulse. (3) Synthetic aperture radar (SAR): SAR uses antennas placed on moving platforms, and then mathematically combines the separate signals transmitted when the antenna moves. An example SAR for ocean remote sensing is the Seasat. However, short-wavelengths are not well imaged by conventional SAR systems. (4) Radiometer (Shi et al., 2006): A passive microwave radiometer is an instrument measuring radiance, or the radiation emitted from the ocean surface. From the emissivity of the ocean surface, the salinity can be estimated. However, passive radiometer cannot separate out the effects of sea surface roughness from conductivity. It appears that near-space vehicle-borne GPS passive remote sensing may provide a promising solution to ocean remote sensing due to the following reasons (see Fig. 5.1) (Tomme, 2005; Wang, 2008):

2.1 Persistent Regional Coverage Near-space is above the troposphere, where most weather occurs, there are no clouds, thunderstorms, or precipitation in near-space. As shown in Fig. 5.2, the near-space vehicles considered in this chapter include free-floaters and maneuvering vehicles. The free-floaters can stay at a specific site almost indefinitely and

80

W.-Q. Wang et al.

Fig. 5.1 Near-space definition and its advantages while compared to space including GEO, middle Earth orbit (MEO) and LEO, and airspace

NASA’s helios

USA army’s high altitude airship

(a)

(b)

Conceptual version of Navy’s HAARR

Air-ballasted maneuvering vehicle

(c)

(d)

Fig. 5.2 Several typical near-space vehicles

provide a stationary observation platform. The maneuvering vehicles can fly to and station-keep over a specified position. Moreover, these vehicles can use conventional propellers and unconventional buoyancy-modification schemes to propel themselves, enabling them to overcome possible winds in the near-space.

5


81

2.2 Low Cost When cost is the concern, near-space vehicles have no peers. Their inherent simplicity, recoverability, relative lack of requirement for complex infrastructure, and lack of space-hardening requirements all contribute to this strong advantage. If the payloads they carry has malfunction, they can be brought back and repaired. Additionally, operating in near-space eliminates the expense involved with infrastructure construction.

3 System Concept and Signal Model GPS broadcasts a civilian-use carrier signal at 1.57542 GHz, referred to as “L1”. Additionally, a second carrier signal for military use is also broadcasted at 1.2276 GHz referred to as “L2”, but civilian reception of this signal requires complicated signal processing algorithms. These signals are constantly scattered off earth surface, and are known to contain valuable and varied information on the targets or environment. Note that, although the passive receiver placed on nearspace vehicle may be stationary, an aperture synthesis can still be achieved with the motion of the GPS satellite only (He et al., 2005). This configuration offers two significant advantages, one is regional persistence, and the other is bistatic observation. Bistatic observation can provide several specific advantages, like the exploitation of additional bistatic information (Kuang and Jin, 2007) and improved detection capability of slowly moving targets (Li et al., 2007). Objects detection in heterogeneous environments can take advantage of reduced retro-reflector effects (Fernandez et al., 2006). The segmentation and classification of ocean surface and volume scatters are alleviated by comparing the spatial statistics of the monostatic and bistatic scattering coefficients. Bistatic observation may also increase the radar cross section (RCS) of manmade object and/or the sensitivity to specific scattering centres of objects composites. Furthermore, bistatic observation in a forward scattering geometry also has a potential for systematic vegetation monitoring. As shown in Fig. 5.3, the passive receiver consists of two channels. One is fixed to collect the direct-path signal, which can be used as the reference signal for matched filtering. This signal is sampled in a delayed window that can be predicted using the near-space vehicle-borne receiver’s position information. The other channel is used for remote sensing. The PRN (pseudo-random noise) code modulation of the GPS carrier and the relative velocity between the GPS satellite and near-space vehicleborne receiver allow the receiver to extract the reflected signal components. Note that another way using two or more near-space receivers. Each receiver performs its own matched filtering with its received signal. The total results can then be combined to provide a consistent image or detection. Additionally, a configuration using a single-channel receiver is also feasible. In this case, the received signals contain the energy from both the direct-path channel and the scattered channel. Once they are separated, successful matched filtering can then be obtained.

82

W.-Q. Wang et al.

Fig. 5.3 System configuration with two-channel receiver direct-path

GPS transmitters

near-space receiver

Rt Rr

θ

H o

As an example, we consider the L1 signal (Grewal et al., 2001) sL1 (t) =

2PI d (t) c (t) cos (2π fL1 t + θL1 ) +

2PQ d (t) p (t) sin (2π fL1 t + θL1 ) (1)

where PI and PQ are the respective carrier power for the in-phase and quadraturephase components, d (t) is the 50-bps data modulation, c (t) and p (t) are the respective C/A pseudorandom code (short code) and P pseudorandom code (long code) waveforms, fL1 is the L1 carrier frequency in Hz per second, and θL1 is a common phase shift in radians. The quadrature carrier power PQ is approximately 3dB less than PI . Then, the received GPS scattered signal is xr (t) = αr sL1 (t − τr )

(2)

where α r and τ r are the attenuation of the scattered signal and the corresponding delays in time, respectively. Using the reference function h (t) = s∗L1 (−t)

(3)

where ∗ denotes a complex conjugate operator, from matched filtering we have sout (t) = xr (t) ⊗ h (t) = αr sL1 (t − τr ) ⊗ s∗L1 (−t)

(4)

where ⊗ denotes a convolution operator. Note that the reference function in Eq. (3) can be either extracted from the direct-path channel or extracted from the scattered channel. As both C/A and P codes have low cross-correlation, from Eq. (4) we have sout (t) = αr sC/A (t − τr ) ⊗ s∗C/A (−t) + αr sP (t − τr ) ⊗ s∗P (−t) where sC/A (t) and sP (t) with variable t are defined, respectively, by

(5)

5


sC/A (t) = sP (t) =

83

2PI d (t) c (t) cos (2π fL1 t + θL1 )

(6a)

2PQ d (t) p (t) sin (2π fL1 t + θL1 )

(6b)

The subsequent receiver processing involves performing the cross correlation between the scattered electric field obtained from the antenna and delayed for τ , with a locally generated replica of the GPS signal for a specific pseudorandom noise code and compensation the carrier frequency fL1 Ti Y (t0 , τ , fL1 ) =

u t0

2π jf + τ · [PRN t0 + t e L1

+ t

t0 +t

]dt

(7)

0

where PRN (t) is the binary phase shift-keyed (BPSK) modulation of the PRN code, Ti is the integration time, and u (t) is simply PRN (t) at a time offset caused by propagation from the satellite to the receiver, modulated onto a carrier of exp (−2π jfdo t) with a Doppler-shifted frequency fdo . For the GPS signal scattered from ocean surface, the scattered field contains contributions from all points on the ocean surface with different delays and Doppler frequencies. In this case, the voltage signal at the receiver can be represented by (Zavorotny and Voronovich, 2000) Y(t0 , τ , fL1 ) = e 0 −2π j(f0 −fL1 ) Ti D(r) τ (r, t0 ) S f (r, t0 ) g ((r, t0 + τ )) d2 r t

(8)

where D (r) is the footprint function of the receiving antenna in terms of complex amplitude, r is projection of the position vector from the scatterer on the surface to the receiver onto the mean sea level, which is assumed to be locally flat. 0 is a constant phase offset that is not be relevant, and exp −2π j (f0 − fc ) t represents the frequency offset between f0 . is the autocorrelation of PRN code

(τ ) ≈

| 1 − |τ τc , |τ | ≤ τc 0, |τ | > τc

(9)

S (f ) =

sin (π fTi ) −π ·j·f ·Ti e π fTi

(10)

and S is expressed as

0 is the position vector from the transmitter to the scatterer on the If defining R is the position vector from the scatterer on the surface to the receiver, surface and R τ , f , and g (r, t) can then be represented, respectively, by (You et al., 2004) τ (r) = τ −

R0 (r) + R (r) c

(11a)

84

W.-Q. Wang et al.

f (r, t) = fd (r, t) + f0 − fc

(11b)

q2 exp jK (R0 + R) g (r, t) = − qz 4π j R0 R

(11c)

where fd is the Doppler shift for a point on the surface at r relative to the reference carrier frequency fL1 , g is the polarization- and phase-delay-sensitive function, is the polarization-sensitive reflection coefficient, and q = (q⊥ , qz ) is the scattering vector in the bisector direction, c is the speed of light.

1000

800

800

600

600

Fig. 5.4 Comparative C/A correlation results: left is the auto-correlation and right is the cross-correlation

Rik

Rkk

1000

400

400

200

200

0

0 0

500 Lag

1000

0

500 Lag

180 160

range resolution ρr [m]

140 120 100 80 60 40 20 0

0

20

40

60 80 100 120 bistatic angle γ [degree]

Fig. 5.5 Measure resolution versus its bistatic angle

140

160

180

1000

5


85

From Eq. (5) we can see that, the output results are the auto-correlations of C/A and P codes. To obtain good matched filtering results, the used waveforms (here are the C/A and P codes) should have high auto-correlation and low cross-correlation. As both C/A and P codes have high auto-correlation, as shown in Fig. 5.4, successful matched filtering can be obtained from Eq. (4). Correspondingly, the measure resolution is determined by ρr =

1 c 2Br cos (γ /2)

(12)

where Br is the GPS signal bandwidth, and γ is the bistatic angle. As an example, Fig. 5.5 shows the potential range resolution.

4 Example Application: Tsunamis Detection The interaction of electromagnetic field with rough surfaces and their capability to scatter the signals has been widely studied for different applications. It is possible to split the scattering process into two contributions: one is the specular term, and the other is the diffuse term. The former is characterized by its high directivity, while the second contribution spreads the signal into a wide range of low power scattered directions. The diffuse component of the scattering is caused by the roughness of the surface. The interaction of an electromagnetic field with ocean surface and the capability of the latter to scatter that field has been mainly studied in a monostatic geometry. However, bistatic ocean scattering has not been fully investigated due to its complexity (Kuang and Jin, 2007). Possible approaches to deal with this problem are Kirchhoff Model (Fung et al., 2001), KM Geometric Optics Limit (Ulaby et al., 1982) and Small Perturbation Method (Sobieski et al., 1991). An example application is tsunami detection. Tsunamis can be originated by earthquakes, submarine landslides, volcanic eruptions, meteorite impacts or by a combination of these factors (Ross, 2009). Tsunami waves have sufficient energy to across entire oceans, which travel outward rapidly with small detectable height in the deep ocean. But, due to a tsunami wave’s shoaling near shore, a tsunami that is visibly unnoticeable in the deep ocean will become more detectable if closer to shore (Myers et al., 2008). It is reported that the Sumatra tsunami had a recorded wave height of 60–80 cm in the deep ocean, but had a maximum wave height of 15 m near Banda Aceh (Borrero, 2005). As tsunami is a surface gravity wave with a wavelength much larger than the ocean depth, the wave develops in three stages (Myers et al., 2008): one is the formation due to the initial cause and propagation near the source, the second is the free propagation of the wave in the open ocean at a large depths, and the third is the propagation of the wave in the region of continental shelf and shallow coastal water. So, the wave behavior in each specific region should be understood in order to properly detect, predict, and model the tsunami wave’s activity.

86

W.-Q. Wang et al.

A rigorous treatment of tsunami detection requires the full Navier-Stikes equation, together with the general equation of continuity and the associated boundary conditions. Nevertheless, many essential characteristics can be derived from a simplified model, which is obtained by assuming incompressible, inviscid flow, ignoring Coriolis and centrifugal effects, setting aside consideration of acoustic waves in the fluid, and side-stepping the possibility of significant ambient currents. Note that the wave amplitude will be small except at the immediate coastal zone (Anderson, 2008). The potential flow φ can be determined by 2 φ+ ∇xy

∂2 φ=0 ∂z2

(13)

where (x, y, z) is the coordinate. Once the kinematic boundary conditions at the free surface and on the bottom, and the dynamic boundary condition at the free surface are imposed, Eq. (13) can be solved successfully. Hence, tsunami can be detected from electromagnetic investigation. It is has been proved that, there is a link between the tsunami wave amplitude and the microwave radar cross section (RCS) (Kouchi and Yamazaki, 2007), and significant variations (a few dB) of the RCS synchronous with the sea level can be found both at C and Ku band in the geophysical data record of the altimetry satellite Jason-1. From the microwave active remote sensing view point, the oceanographic detectable features of a tsunami wave are summarized in (Galletti, 2007): (1) tsunami wave height, (2) tsunami orbital velocity, (3) tsunami-induced RCS modulation. The first two parameters belong to an ocean wave, and the third is a geophysical feature that arises from complex hydrodynamic processes. The behavior of tsunami-induced RCS modulations depends upon bathymetry, meteorological factors and sea-state. It has been proved the capability of microwave radar systems to detect internal waves, which are generally triggered by tides and it is worth noting that they can be found on radar images for about any condition of wind speed and depths of water (Le Caillec, 2006; Rodenas and Garello, 1998; Hogan et al., 1996). Along with tides, tsunamis are shallow water waves and have the potential to trigger internal waves. As such, tsunami-induced RCS modulations can be used to predict future tsunami. This involves not only detection but also an estimate of the tsunami magnitude. Thus a microwave sensor has the possibility to detect tsunami by comparing the pre- and post- quake patterns, as shown in Fig. 5.6.

Fig. 5.6 Tsunami-induced radar cross section modulations synchronous with the tsunami wave amplitude

5


87

5 Power Budget Analysis Using GPS satellite as an illuminator presents a problem of signal detectability because the received GPS signal is very weak. As the final remote sensing performance is directly related to the signal-to-noise ratio (SAR), power budget analysis is necessary. For radar image formation, the total data samples are processed coherently to produce a single image resolution cell. The thermal noise samples can be taken as independent from sample to sample within each pulse. After coherent range and azimuth compression, the image SNR is given by (He et al., 2005) SNRimage = 0 ·

Ar σ0 1 Rr · · ·η 2 4π Rr KT0 Fn vs · ρa

(14)

where 0 = 3 × 10−14 Wt/m2 , σ 0 is the RCS parameter, Rr is the distance from the receiver to a given target, K is the Boltzmann constant, T0 is the system noise temperature, Fn is the noise figure, Ar is the effective area of the receiver antenna, and η is the loss factor. If synthetic aperture is applied, the potential azimuth resolution is ρa =

λRt vs Ts

(15)

where Ts is the integration time, we then have SNRimage =

0 Ar σ0 Ts η 4π R2r Fn KT0

(16)

As an example, assuming a typical system with the following parameters: σ0 = 20m2 , Ts = 1000s, T0 = 300, η = 0.5, and Fn = 2dB, then the calculated SNR is illustrated in Fig. 5.7. Here the SNR is favorable owing to an essentially long integration time. Note that this SNR can be further improved by using non-coherent integration of signals from more than one receiver channel. Another quantity directly related to radar image performance is the noise equivalent sigma zero (NESZ), which is the mean RCS necessary to produce an SNRimage of unity. The NESZ can be interpreted as the smallest target cross section which is detectable by the SAR system against thermal noise. Setting SNRimage = 1, Eq. (16) gives NESZ =

4π R2r Fn KT0 0 Ar Ts η

(17)

Assuming again the same parameters as above, the calculated NESZ is illustrated in Fig. 5.8. This result shows that a comparable RCS requirement to current radar systems is possible.

88

W.-Q. Wang et al.

Fig. 5.7 SNR analysis

Fig. 5.8 NESZ analysis

However, the final SNR performance varies with the change of incidence angle. As the GPS transmitted signal is circular polarization, if considering the reflected signal as two linear polarization signals H (horizontal polarization) and V (vertical polarization), we have H =

sin (βin ) − sin (βin ) +

(εo − j60λεr ) − cos2 (βin ) (εo − j60λεr ) − cos2 (βin )

(18)

5


H =

(εo − j60λεr ) sin (βin ) − (εo − j60λεr ) sin (βin ) +

(εo − j60λεr ) − cos2 (βin ) (εo − j60λεr ) − cos2 (βin )

89

(19)

where βin is the incidence angle, ε0 is the dielectric constant, and εr is the electrical conductivity coefficient. In an alike manner, if considering the reflected signal as two circular polarization signals co (coherent polarization) and cr (cross polarization), we have co =

H + V 2

(20)

cr =

H − V 2

(21)

As an example, suppose εo = 25 and εr = 4.5, the simulated reflection coefficients of these two cases are given in Figs. 5.9 and 5.10, respectively. To obtain good SNR performance, an incidence angle of 20◦ ∼ 30◦ is preferred in actual systems. Another consideration is clutter, which can be assumed to enter the system via the antenna sidelobes only. The transmitted GPS signal is a spread spectrum signal with a chip rate of 1.023 M Hz, the clutter power at the receiver antenna can be represented by 2 ρr λ Gsl σc · log 1 + · Pc = 0 · 2 Rr 4π

(22)

Normalized amplitude of reflection coefficient

1 0.9 0.8 0.7 0.6 0.5 ΓH

0.4

ΓV

0.3 0.2 0.1 0

0

10

20

30

40

50

60

70

80

Incidence angle [°]

Fig. 5.9 Normalized amplitude of reflection at the case of linear polarization

W.-Q. Wang et al.

Normalized amplitude of reflection coefficent

90 1 0.9 0.8 0.7 0.6 0.5

Γco

0.4

Γcross

0.3 0.2 0.1 0

0

10

20

30 40 50 60 Incidence angle [°]

70

80

Fig. 5.10 Normalized amplitude of reflection at the case of circular polarization

where σ c and Gsl are the RCS of the clutter per unit area and the sidelobe gain of the antenna, respectively. From radar equation we get the clutter-to-target power ratio (CTPR) as CTPR =

Pc σc Gsl ρr = 2π · · · R2r · log 1 + Pr σ0 Gr Rr

Fig. 5.11 Calculated clutter-to-target power ratio

(23)

5


91

To estimate the clutter power, we suppose that Gsl = −10dB, σc = −20dB, and the other parameters are same as above. Figure 5.11 gives the clutter-to-target power ratio for different values of Rr and Ar (Ar = λ2 Gr /4π ). The results show that the clutter contains almost as much power as the target returns. This situation can be improved by reducing the magnitude of the antenna sidelobe gain. Moreover, the CTPR can be further improved owing to subsequent range compression and azimuth compression.

6 Challenges and Possible Solutions The passive remote sensing sensor discussed in this paper is a bistatic configuration, hence it is subject to the problems and special requirements that are either not encountered or encountered in less serious form for current monostatic SAR systems. The biggest challenge lies in the synchronization between the near-space vehicle-borne receiver and the GPS satellites: phase synchronization, the receiver and the satellites must be coherent over extremely long periods of time; spatial synchronization, the receiving antenna and the transmitting antennas must simultaneously illuminate the same spot on the ground. Consequently, there is no cancellation of low-frequency phase noise as in a monostatic radar, where the same oscillator is used for modulation and demodulation. We can express the synchronization errors as T φe =

2π (ft − fr )dt

(24)

0

where ft and fr denote the GPS transmit carrier frequency and receiver’s demodulation frequency, respectively, T is the integrated time and should be greater than one aperture time. A typical requirement for the maximum tolerable ISLR (integrated sidelobe ratio) is −20 dB (Wang, 2006). Unfortunately, the phase synchronization errors are usually random and too complex to apply autofocus algorithms. Bistatic radar synchronization processing has received much recognition in recent years (Younis et al., 2006; Wang, 2007b; Krieger and Moreira, 2006). Oscillator phase noise may not only defocus the radar image, but also introduce significant positioning errors along the scene extension, so some synchronization technique or compensation algorithms must be applied. One possible solution is the direct-path signal based synchronization technique described in (Wang et al., 2008). The direct-path signal of transmitter is received with one appropriative antenna and divided into two channels, one is used to synchronize the sampling clock, and the other is down-converted and used to compensate the phase synchronization errors. Finally, the residual synchronization errors can then be compensated with auto-focusing algorithms. Another problem is motion compensation. For successful near-space vehicleborne passive remote sensing, strict relative position or altitude is required.

92

W.-Q. Wang et al.

GNSS transmitters near-space receiver

Rt

VCA

Tra

nsp

ond

er

DDS

Rr o

Fig. 5.12 Transponder based motion compensation

However, as a matter of fact, problems arise due to the presence of atmospheric turbulence, which introduce aircraft trajectory deviations from the nominal position, as well as altitude (roll, pitch, and yaw angles). For current radar systems, the motion compensation is usually achieved with GPS and INU (Inertial Navigation Units). However, for near-space vehicle-borne passive receiver the motion measurement facilities may be not reachable, the conventional motion sensors based motion compensation techniques may be not applicable any longer, so some new efficient motion compensation algorithms must be developed. To reach this aim, we can use the transponder proposed by (Weiβ, 2002; Weiβ and Berens, 2004), as shown in Fig. 5.12, to extract the possible motion errors. This transponder consists of a low-noise amplifier followed by a bandpass filter. A voltage controlled attenuator (VCA) is used to modulate the radar signal in a manner that the retransmitted signal will show two additional Doppler frequencies. Thereafter the signal will be amplified to an appropriate level and retransmitted towards the near-space receiver. This transponder can be seen as an amplitude modulator, that is sc (t) = α + β cos (2π fm t + ϕm ) · s0 (t)

(25)

where fm is the transponder modulation frequency, ϕ m the starting phase and s0 (t) the GPS transmitted signal. The corresponding near-space sensor received signal can be represented by sr (t) = ss (t) + α + β cos (2π fm t + ϕm ) · sm (t)

(26)

where ss (t) and sm (t) denote the un-modulated part and the echo of the transponder without the amplitude modulation, respectively. After applying an Fourier transform to Eq. (26), we then have

5


93

Sr (f ) = Ss (f ) + αSm (f ) β β + e−jϕm Sm (f + fm ) + ejϕm Sm (f − fm ) 2 2

(27)

We can notice that, the retransmitted signal will show the original GPS signal and two additional Doppler frequencies, one positive and one negative shifted, allowing to extract the motion errors without clutter interferences.

7 Conclusion As near-space vehicles can provide a remote sensing more responsively and more persistently than satellites and airplanes, this chapter investigated the passive ocean remote sensing by near-space vehicle-borne receiver for persistently regional remote sensing applications. This involves placing a passive radar receiver placed inside a near-space vehicle to operate in conjunction with opportunistic illuminators. In this chapter, GPS satellites is used as the opportunistic illuminators. The system configuration, potential applications, technical challenges and possible solutions are investigated. There was a doubt that the ocean-scattered GPS signals would be too weak to be detected from near-space vehicle-borne receivers. However, this is not the case and ocean-scattered GPS signals can be detected on a regular basis. The number of available GPS surface reflections and the low cost of this technology provide a potential to perform very dense sampling at high temporal resolution across small features of the ocean surface. The described passive remote sensing sensor can provide significant applications ranging from regional climate weather models to marine ocean safety, such as wind speed estimation, wave height measurement, and ocean mean square height estimation. We do not advocate eliminating satellites or airplanes; however, in some circumstances near-space vehicles are indeed the best choice for providing some specific remote sensing applications such as persistently regional ocean remote sensing applications. Acknowledgements This work was supported in part by the Specialized Fund for the Doctoral Program of Higher Education for New Teachers under contract number 200806141101, the Open Fund of the Key Laboratory of Ocean Circulation and Waves, Chinese Academy of Sciences under Contract number KLOCAW0809, and the Open Fund of the Institute of Plateau Meteorology, China Meteorological Administration under contract number LPM2008015.

References Allen EH (2006) The case for near-space, Aerospace America, 22:31–34. Anderson S (2008) Prospects for tsunami detection and characterization with HF skywave radar. Proceedings of international radar conference, IEEE, Adelaide, SA, pp 639–645. Bindlish R, Crow WT, Jackson TJ (2009) Role of passive microwave remote sensing in improving flood forecasts. IEEE Geosci Remote Sens Lett 6:112–116.

94

W.-Q. Wang et al.

Borrero JC (2005) Field data and satellite imagery of tsunami effects in Banda Aceh. Science 308:1596. Boukabara SA, Weng F, Liu Q (2007) Passive microwave remote sensing of extreme weather events using NOAA-18 AMSUA and MHS. IEEE Trans Geosci Remote Sens 45:2228–2246. Bovolo F, Bruzzone L (2007) A split-based approach to unsupervised change detection in large-size multitemporal images: application to tsunami-damage assessment. IEEE Trans Geosci Remote Sens 45:1658–1679. Cox C, Munk W (1954) Measurements of the roughness of the sea surface from photographs of the suns glitter. J Optical Soc Am 44:838–850. Fernandez PD, Cantalloube H, Vaizan B, Krieger G, Horn R, Wendler M, Giroux V (2006) ONERA-DLR SAR bistatic campaign: planning, data acquisition, and first analysis of bistatic scattering behavior of natural and urban targets. IEE Proc Radar Sonar Navig 153:214–223. Fung AK, Zuffada C, Hsieh CY (2001) Incoherent bistatic scattering from the sea surface at L-band. IEEE Trans Geosci Remote Sens 39:1006–1012. Galletti M, Krieger G, Thomas B, Marquart M, Johannes SS (2007) Concept design of a near-space radar for tsunami detection. Proceedings of IEEE geoscience and remote sensing symposium, IEEE, Barcelona, pp 34–37. Garrison JL, Katzberg SJ, Hill MI (1998) Effect of sea roughness on bistatically scattered range coded signals from the global positioning system. Geophys Res Lett 25, 2257–2260. Garrison JL, Komjathy A, Zavorotny V, Katzberg S (2002) Wind speed measurements using forward scattered GPS signals. IEEE Trans Geosci Remote Sens 40:50–65. Gleason S, Hodgart S, Sun Y, Gommenginger C, Mackin S, Adjrad M, Unwin M (2005) Detection and processing of bistatically reflected GPS signals from low earth orbit for the purpose of ocean remote sensing. IEEE Trans Geosci Remote Sens 43:1229–1241. Grewal MS, Weill LR, Andrews AP (2001) Global positioning systems: inertial navigation and integration, John Wiley & Sons, New York, ISBN: 0-471-20071-9. Guan MX, Guo G, Li L (2007) A novel access protocol for communication system in near space. Proceedings of wireless communication network mobile computation conference, IEEE, Shanghai, China, September 2007, pp 1849–1852. He X, Cherniakov M, Zeng T (2005) Signal detectability in SS-BSAR with GNSS non-cooperative transmitter. IEE Proc Radar Sonar Navig 152:124–132. Heise S, Wickert J, Beyerle G, Schmidt T, Smit H, Cammas JP, Rothacher M (2008) Comparison of water vapor and temperature results from GPS radio occultation aboard CHAMP with MOZAIC aircraft measurements. IEEE Trans Geosci Remote Sens 46:3406–3411. Hogan GG, Chapman RD, Watson G, Thompson DR (1996) Observations of ship-generated internal waves in SAR images from Loch Linnhe, Scotland, and comparison with theory and in situ internal wave measurements. IEEE Trans Geosci Remote Sens 34:532–542. Kouchi K, Yamazaki F (2007) Characteristics of tsunami-affected areas in moderate-resolution satellite images. IIEEE Trans Geosci Remote Sens 45:1650–1657. Krieger G, Moreira A (2006) Spaceborne bi- and multistatic SAR: potential and challenges. IEE Proc Radar Sonar Navig 153:184–198. Kuang K, Jin YQ (2007) Bistatic scattering from a three-dimensional object over a randomly rough surface using the FDTD algorithm. IEEE Trans Antennas Propag 55:2302–2312. Kurtz NT, Markus T, Cavalieri DJ, Krabill W, Sonntag JG, Miller J (2008) Comparison of ICESat data with airborne Laser altimeter measurements over Arctic sea ice. IEEE Trans Geosci Remote Sens 46:1913–1924. Le Caillec JM (2006) Study of the SAR signature of internal waves by nonlinear parametric autoregressive models. IEEE Trans Geosci Remote Sens 44:148–158. Li G, Xu J, Peng YN, Xia XG (2007) Bistatic linear antenna array SAR for moving target detection, location, and imaging with two passive airborne radars. IEEE Trans Geosci Remote Sens 45:554–565. Marcel MJ, Baker J (2007) Interdisciplinary design of a near-space vehicle. Proceedings of southeast conference, IEEE, Richmond, VA, March 2007, pp 421–426.

5


95

Meyer F, Hinz S, Laika A, Weihing D, Bamler R (2006) Performance analysis of the TerraSAR-X traffic monitoring concept. ISPRS J Photogr Remote Sens 61:225–242. Myers RG, Draim CJE, Cefola PJ, Raizer VY (2008) A new tsunami detection concept using space-based microwave radiometery. Proceedings of IEEE geoscience and remote sensing symposium, IEEE, Boston, MA, pp IV.958–IV.961. Nie C, and Long DG (2008) A C-band scatterometer simultaneous wind/rain retrieval method. IEEE Trans Geosci Remote Sens 46:3618–3631. Oo KS, Mehdiyev M, Samarakoon L (2005) Potential of satellite data in assessing coastal damage caused by South-Asia tsunami in December 2005 – a field survey report. Asia J Geoinformatics 5:16–37. Rodenas JA, Garello R (1998) Internal wave detection and location in SAR images using wavelet transform. IEEE Trans Geosci Remote Sens 36:1494–1507. Ross RE (2009) Tsunami alert system starts up in Indonesia. IEEE Spectr 46:18–19. Shi JC, Liang LM, Zhang LX, Chen KS, Wigneron JP, Chanzy A, Jackson TJ (2006) Physically based estimation of bare-surface soil moisture with the passive radiometers. IEEE Trans Geosci Remote Sens 44:3145–3153. Shinde YS, Gahir HK (2008) Dynamic pressure sensing study using photonic crystal fiber: application to tsunami sensing. IEEE Photon Tech Lett 20:279–281. Shokr M, Lambe A, Agnew T (2008) A new algorithm (ECICE) to estimate ice concentration from remote sensing observations: an application to 85-GHz passive microwave data. IEEE Trans Geosci Remote Sens 46:4104–4121. Sobieski P, Guissard A, Baufays C (1991) Synergic inversion technique for active and passive microwave remote sensing of the ocean. IEEE Trans Geosci Remote Sens 29:391–405. Suchandt S, Eineder M, Breit H, Runge H (2006) Analysis of ground moving objects using SRTM/X-SAR data. ISPRS J Photogr Remote Sens 61:209–224. Tomme EB (2005) Balloons in today’s military: an introduction to near-space concept. Air Space J 19:39–50. Tralli DM, Blom RG, Zlotnicki V, Donnellan A, Evans DL (2005) Satellite remote sensing of earthquake, volcano, flood, landslide and coastal inundation hazards. ISPRS J Photogr Remote Sens 59:185–198. Ulaby FT, Moore RK, and Fung AK (1982) Microwave remote sensing: active and passive. Artech House, Inc., ISBN 0-89006-191-2. Wang WQ (2006) Analytical modeling and simulation of phase noise in bistatic synthetic aperture radar systems. Fluct Noise Lett 6:L297–303. Wang WQ, Cai JY, Peng QC (2007) Near-space SAR: a revolutionary microwave remote sensing mission. Proceedings of Asia-Pacific synthetic aperture radar conference, IEEE, Huangshan, China, October 2007, pp 127–131. Wang WQ (2007a) Application of near-space passive radar for homeland security, Sens Imag An Int J 8:39–52. Wang WQ (2007b) Approach of adaptive synchronization for bistatic SAR real-time imaging. IEEE Trans Geosci Remote Sens 45:2695–2700. Wang WQ (2008) Near-space passive remote sensing for homeland security: potential and challenge. Proceedings of XXI international society for photogrammetry and remote sensing symposium, ISPRS, Beijing, China, July 2008, pp 1021–1027. Wang WQ, Ding CB, Liang XD (2008) Time and phase synchronization via direct-path signal for bistatic synthetic aperture radar systems. IET Radar Sonar Navig 2:1–11. Weiβ M (2002) A new transponder technique for calibrating wideband imaging radars. Proceedings of Europe synthetic aperture radar symposium, EUSAR, Cologne, Germany, pp 493–495. Weiβ M, Berens P (2004) Motion compensation of wideband synthetic aperture radar with a new transponder technique. Proceedings of IEEE geoscience and remote sensing symposium. IEEE, Anchorage, AK, USA, pp 3649–3651.

96

W.-Q. Wang et al.

You HZ, Garrison JL, Heckler G, Zavorotny VU (2004) Stochastic voltage model and experimental measurement of ocean-scattered GPS signal statistics. IEEE Trans Geosci Remote Sens 42:2160–2169. Younis M, Metzig R, Krieger G (2006) Performance prediction of a phase synchronization link for bistatic SAR. IEEE Geosci Remote Sens Lett 3:429–433. Zavorotny VU, Voronovich A (2000) Scattering of GPS signals from the ocean with wind remote sensing application. IEEE Trans Geosci Remote Sens 38:951–964.

Part II

Global Changes

Chapter 6

A Global Survey of Intense Surface Plankton Blooms and Floating Vegetation Using MERIS MCI James (Jim) Gower and Stephanie King

Abstract The MERIS imager on the European Envisat satellite has spectral bands which give a new capability for detection of blooms and aquatic vegetation. We use MERIS data to compute MCI (Maximum Chlorophyll Index), which measures the radiance peak at 709 nm in water-leaving radiance, indicating the presence of a high surface concentration of chlorophyll a against a scattering background. The index is high in “red tide” conditions (intense, visible, surface, plankton blooms) and also when aquatic vegetation is present, leading to a “red edge” step increase in radiance. A bloom search based on MCI has resulted in detection of a variety of events in marine waters and lakes round the world, as well as detection of extensive areas of pelagic vegetation (Sargassum spp.) in the ocean, previously unreported in the scientific literature. Global MCI composite images are produced daily from all MERIS (daylight) passes of Reduced Resolution (RR) data, starting soon after MERIS launch, in June 2002. This paper describes the composites and gives examples of plankton bloom events that they have detected. Keywords Satellite images · MERIS · Plankton blooms · Sargassum · Trichodesmium

1 Introduction The MEdium Resolution Imaging Spectrometer (MERIS) was launched on the European Envisat satellite in June 2002 (ESA, 2008; Rast et al., 1999). We use MERIS to compute the MCI (Maximum Chlorophyll Index). This shows the amplitude of a peak near 709 nm in the radiance spectrum of light reflected from the earth’s surface, which several authors have associated with high levels of chlorophyll a in ocean, coastal and lake water targets, such as plankton blooms and floating or benthic plants (Gitelson et al, 1992; Yacobi et al., 1995; Gower et al. 1999, 2005, J. Gower (B) Institute of Ocean Sciences, Fisheries and Oceans Canada, Sidney, BC, Canada e-mail: [email protected]


99

100

J. Gower and S. King

2008b). The index is unique to MERIS among satellites for monitoring large ocean and coastal areas, in that a band near 709 nm is not present on either MODIS or SeaWiFS. We have demonstrated use of the MCI to detect plankton blooms (Gower et al., 2005), floating Sargassum (Gower et al., 2006) and “superblooms” of Antarctic diatoms associated with platelet ice (Gower and King, 2007). In the future, we expect the global composites to lead to more frequent detection of a wider variety of phenomena. The MCI (Gower et al., 2005) is computed as radiance at 709 nm above a linear baseline defined by radiances at 681 nm and 753 nm. In our global search, this is calculated only for pixels for which radiance at 865 nm is less than 15 mW m–2 sr–1 nm–1 to eliminate land pixels and areas of strong sun glint, haze or cloud: If L865 < 15 mW m−2 · sr−1 · nm−1 , then MCI = L709 − L681 − 0.389(L753 − L681 )

(1)

where L865 represents Level 1 radiances (as measured at the satellite) at a wavelength of 865 nm, and similarly for other wavelengths, and the factor 0.389 represents the wavelength ratio (709–681)/(753–681). The MCI therefore indicates an excess radiance at 709 nm above this baseline, which can indicate a water-leaving radiance spectrum showing a peak at 709 nm. Models indicate that this type of spectrum is characteristic of intense surface plankton blooms in which high concentrations of phytoplankton are distributed in near-surface waters. In this case, the absorption by chlorophyll reduces radiance at wavelengths shorter than 700 nm, while absorption by water reduces radiance at wavelengths longer than 720 nm, leading to a radiance peak at the wavelength of minimum absorption, near 709 nm. Models also show that vegetation under a shallow layer of water, including coral reefs, can give rise to a spectral peak near 709 nm, also giving a positive MCI signal. MCI can also be high due to presence of a “red-edge” spectrum, in which water-leaving radiance shows a step increase near 700 nm with increasing wavelength. This type of spectrum is characteristic of land vegetation, for which the red-edge (maximum rate of increase of observed radiance with wavelength) usually occurs at a wavelength of about 720 nm. Over water with MERIS, we observe an apparent step increase between 681 nm and 709 nm suggesting a shorter wavelength position. The MCI as defined in Eq. (1), will give a high value in this case as well. We interpret red-edge type spectra as showing presence of buoyant slicks of either phytoplankton or macroalgae, eg Sargassum. MCI computed using Eq. (1) will tend to show negative features where the radiance at 681 nm is raised by chlorophyll a fluorescence. A “broad” MCI can be computed using radiance at 665 nm instead of 681 nm, which avoids this problem (see for example, Fig. 6.8 below), but in the intense blooms in most of these examples the fluorescence signal is relatively small. MCI values are computed from level 1 spectral radiance data before atmospheric correction, since the events we are studying typically give radiances too high to be handled by atmospheric-correction algorithms. In the few cases where we have been able to compute MCI from atmospherically-corrected (Level 2) data, reflectance

6

A Global Survey of Intense Surface Plankton Blooms and Floating Vegetation

101

values at each wavelength in Eq. (1) were used instead of radiances. The resulting images are nearly identical to the Level 1 results, with values diverging from a mean linear transformation between reflectance and radiance by only a few percent. In most cases, however, the Level 2 atmospheric correction fails over these events. The present prototype global composite includes a small bias correction as a function of local sun elevation, not shown in Eq. (1), which compensates an observed annual variation in the average MCI signal. In presenting spectra of the blooms, we compare top-of-atmosphere spectra of the high-MCI event itself, and of a nearby reference area of clear water. We plot the difference spectrum on the same Figure, using an expanded scale to show the radiance changes due to the blooms. In Figures that follow, spectra showing topof-atmosphere radiance for an area giving high MCI signal are plotted in red, and spectra for nearby clear water are plotted in green. The difference spectra are plotted in blue using an expanded scale (values on right axis). Although the MCI will not detect all of the harmful algal blooms (HABs) that are causing increasing damage to aquaculture, tourism and coastal water quality, it is clear that it has a role in monitoring some of the surface, high-chlorophyll events, which are widely publicized as “red tides.” It also has a role in mapping floating vegetation and blooms in ice. The global composites extend the search for these events to the limits imposed by MERIS data collection, sun-glint and global cloud cover.

2 Meris Global MCI Composite Data The global MCI composites are produced using GRID computing capabilities at ESRIN, according to specifications based on detection of bloom and vegetation events in a number of locations round the world, including the coasts of Canada, the USA, India, Chile and Antarctica (Gower et al., 2008a). In Fig. 6.1, the daily composite shows the 14 descending (daylight) passes which make up the MERIS coverage pattern for a single day. The compositing procedure preserves the maximum MCI value at each location, but only high-latitude areas have more than a single measurement in one day. Data gaps between passes are shown as white. For this day, there is a missing pass on the right side of the composite, and a shorter gap nearer the centre of the Figure. Land, cloud and areas affected by sun-glint in the data are masked to black. At this season, there is adequate sunlight to collect data north of latitude 64 N, where adjacent swaths overlap, while extreme southern coverage is cut-off at a minimum sun elevation. Sun-glint obscures the east side of the swaths between latitudes −10 and +50. MCI values are coded from dark blue (low), through light blue and yellow to red and then white (high). The lower panel of Fig. 6.1 shows the result of combining daily images to give the composite for the month of July 2006. Note that most of the image is featureless, as expected since the MCI is detecting “extreme” events. In both panels, a bloom event can be seen in the southern Baltic. Composites have a nominal 5 km

102


Fig. 6.1 Example of MERIS global composite MCI images (top) for July 7, 2006, and (bottom) for the month of July 2006. In both cases, the composite preserves the maximum MCI value for each pixel. A major bloom event in the southern Baltic (Fig. 6.2) can be seen at the upper centre of both images

resolution. Each composite pixel shows the maximum MCI value computed for any RR MERIS (1.2 km) pixel assigned to that composite pixel. The reduced spatial resolution compared to the 1.2 km of MERIS RR may result in smaller bloom events being spatially distorted, but the use of the maximum values in the composite preserves the record of their occurrence. The monthly composites of MCI signal at 5 km spatial resolution can also be analyzed by computing the frequency distribution (histogram) of MCI values in predefined areas. Pixels that include coastlines and other fixed areas where MCI is observed to be high, such as coral reefs, are masked in all months. We name the sum of the number of MCI values above threshold, multiplied by the amount by which MCI exceeds its background value (in mW m–2 nm–1 sr–1 ) as “MERIS count.” That is: MERIS count =

m=∞

(m − b)n(m)

(2)

m=b+t

where n(m) is the number of pixels in the area having an MCI value of m, b is the background value of MCI corresponding to open water and t is the threshold value. The MERIS count can then be used to assess annual and interannual variations.

6


103

3 Detection of Plankton Blooms After inspecting nearly 7 years of MERIS global MCI data, we have found several areas in which blooms are regularly made visible by MCI (Table 6.1). In Table 6.1, the satellite provides data on bloom intensity, area and season. Identifications are where species have been confirmed at the times of our observations. Questioned identifications are where species have been observed in past years. Due to the regularity of many of these events, we expect that the species or species mix, to be roughly the same from year to year. In other cases events have not been previously reported and we have no information on species We hope to eventually fill in the “Species” column in the table with help from local observers. In addition to these areas, MCI has detected signals from Antarctic superblooms, Sargassum, lakes and cosmic rays, which we discuss separately in the next 4 sections. Figure 6.2 shows the bloom in the Baltic Sea near peak intensity in July 2005. In this case the bloom is easily visible in both the true colour (RGB) and MCI images. MCI gives a cleaner signal in the sense that it is less confused by cloud, haze and sun-glint, so that bloom intensity can be computed over a given area using Eq. (2). Spectra (bottom) show how MCI detects the bloom, either at the highest concentrations where the red-edge signature (A) suggests vegetation (phytoplankton in a surface slick) in air, or at low concentrations where the 709 peak (C) suggests phytoplankton dispersed in water. The spectrum B shows an intermediate form. Figure 6.3 shows the time series of bloom intensity for all months from June 2002 to April 2009, computed using Eqs. (1) and (2). Bloom intensity peaked strongly in July of 2002, 2003, 2005, 2006 and 2008, with the highest intensity in 2005 (Fig. 6.2). The peak in 2007 is much smaller, but still in July. Only in 2004 did a small peak occur in August. At these high latitudes, satellite coverage is increased Table 6.1 Areas in which blooms are regularly detected by MERIS MCI. The medium/high intensity classification is somewhat subjective. Large area blooms tend to cover areas more than 1000 km across, medium 200–1000, and small less than 200 km. Seasonal patterns are well-defined by month Area

Intensity

Area

Species

Peak month

Baltic Sea Strait of Georgia Arabian Sea West Coast of India Maldives Persian Gulf Red Sea North-West Australia Gulf of Carpentaria Great Barrier Reef Polynesia Yellow Sea Mozambique Channel Cape Town, Coast, North

High High High Medium Medium High High High High High Medium High High High

Medium Small Large Large Large Medium Medium Medium Medium Medium Large Small Small Small

Cyanobacteria Various

July June–September January/February April/May April/May April/May June/July November–February September October/November November–February August/September January February–April

Trichodesmium? Trichodesmium? Trichodesmium? Trichodesmium?

Trichodesmium

Sargassum?

104


Fig. 6.2 (Top) MERIS RR data for July 10 2005 showing a strong and extensive cyanobacteria bloom in the southern Baltic Sea surrounding the island of Gotland. Both the true colour (RGB) and MCI images show the bloom, but MCI is not confused by cloud or haze and emphasizes areas where the bloom is close to the surface. Spectra (bottom) are shown for the highest concentrations of the bloom giving a red-edge signature (A) decreasing through (B) to low concentrations (C). The 709 nm peak appears in (B) and (C)

to the point where some coverage from MERIS is available on most days (Fig. 6.4). Data similar to Fig. 6.3 could be computed with time steps down to a few days, to detect smaller changes in bloom timing from year to year. Figure 6.4 shows the Baltic Sea on a series of days in June and July 2006. Fig. 6.3 shows that in 2006, blooms were relatively less intense, and Fig. 6.4 shows them covering a smaller area than Fig. 6.2, mostly to the south of the island of Gotland. The bright bloom (high MCI) event starts on about June 30 and continues intermittently to about July 17. The Baltic blooms are confirmed to be intense surface blooms of cyanobacteria (E. Graneli, Kalmar University, pers. comm.) as observed in previous years. Sudden disappearances of the bloom after July 8 and

6


105

500000 450000

Relative count

400000 350000 300000 250000 200000 150000 100000 50000 0 2002

2003 2004

2005

2006

2007

2008

2009

2010

Start of Year

Fig. 6.3 Time series of MERIS counts for the area 55.9–59.7 N, 17.1–21.2E, including the area shown in Fig. 6.2. MCI data show that cyanobacteria blooms peak in July in most years in the southern Baltic

Fig. 6.4 Segments of the daily global composites from 29 June (060629, top left) to July 18, 2006 (060718, bottom right) covering the southern Baltic. Global composites are produced in a “lat/long” projection, causing longitudes to be stretched by about a factor 2 near 60 N, the approximate latitude of Fig. 6.4

106


13 may correspond to surface winds mixing buoyant cyanobacteria cells down into the water column. These blooms are at present being tracked using AVHRR satellite image data, which provide a single spectral band (570–670 nm) for mapping the blooms, with a second band (670–900 nm) that can be used to separate atmospheric effects assuming water-leaving radiances are low. However, in many cases, this condition is not met. The additional spectral bands of MERIS can provide significant improvements in feature discrimination. Figure 6.5 shows a major bright bloom event with a strong MCI signature in water inside the Great Barrier Reef in north-east Australia. The “red edge” type spectrum shows that the bloom consists of a buoyant slick of floating vegetation or phytoplankton in air, rather than below the water surface. The spectrum of the coral reef, by contrast, shows a small peak at 709 nm due to vegetation under water. The time series of MCI signal for this area (a strip of water 125 km wide and 500 km long

Fig. 6.5 (Top) MERIS RR data for October 5, 2008 showing a bloom extending along the shore, out of Repulse Bay, NE Australia in MCI (right). The true colour (RGB) image (left) shows patches of reef, sun glint offshore and silty water in Broad Sound, but no visible sign of the bloom. Spectra (bottom) are of the coral of the Great Barrier Reef (green arrow) and the bloom (red arrow)

6


107

parallel to the coast) is shown in Fig. 6.6. The October 2008 event is the strongest measured by MERIS so far. We can show images similar to Figs. 6.2 and 6.5, and time series plots similar to Figs. 6.3 and 6.6 for all of the areas listed in Table 6.1. We also see less regularlyoccurring high-MCI blooms. For example, Fig. 6.7 shows a bloom indicated by an extended bright patch of high MCI water off the coast of China immediately seaward of the high-sediment plume of the Chiang-Jiang river on September 12, 2008. This is the highest MCI signal detected by MERIS in this area in the nearly seven years of observation so far. Spectra show a significant increase in back-scattered radiance from 500 nm to 800 nm. The strong MCI signal is apparently due to absorption at 665 nm and 681 nm, presumably by chlorophyll a, which also absorbs strongly at wavelengths shorter than 520 nm. We interpret the spectrum as indicating high chlorophyll concentrations in surface water with significant sediment load. The most recent monthly global composite (April 2009) showed increasing Sargassum in the Gulf of Mexico, and blooms in other expected areas such as the Red Sea, Arabian Sea and Gulf of Carpentaria (Table 6.1). One area in which we had not seen such an extensive bloom before was off the coast of Namibia at about 20S latitude. Search of the data showed that the peak of the bloom was observed on April 4, 2009. Its extent was much reduced by the seventh, though it was visible again on the 13th and 14th. Another clear image on the 26th showed no high MCI signals. Figure 6.8 shows four products derived from MERIS: RGB, FLH, MCI, and Broad MCI, with two spectra for areas indicated by arrows. The RGB product is a true-colour representation of the area, providing the most intuitive imagery and showing clouds, land, and in the present case, a bright bloom as indicated by the top

40000 35000

Relative count

30000 25000 20000 15000 10000 5000 0 2002

2003

2004

2005 2006 2007 Start of Year

2008

2009

2010

Fig. 6.6 A time series of the relative bloom intensity for part of the area in Fig. 6.5, showing the high value reached in October 2008, with smaller maxima in November of 2003 and 2007, and October 2006

108


Fig. 6.7 MERIS RR image of a high-MCI bloom near the edge of the Chiang Jiang River plume on September 12, 2008. The area of high MCI signal is just outside the edge of the sediment plume shown in the true colour (RGB) image. Spectra below show that the MCI signal is due to absorption by chlorophyll a

spectrum and arrow at about 17.5S. This bloom also shows significant FLH signal, suggesting that it is not due to Coccolithophores. The other spectra have significant peaks at 709 nm, indicating very high concentration of chlorophyll a in a large area of surface bloom. The “broad MCI” image is derived from a relation similar to Eq. (1), but using the MERIS band at 665 nm instead of 681 nm to avoid the effect of chlorophyll fluorescence.

4 Antarctic Superblooms Antarctic scientists have reported “superblooms” in which very high concentrations of algae (chlorophyll concentrations up to 200 mg · m–3 ) occur in or near land-fast ice, close to the southern, minimum limit of ice extent (Smetacek et al., 1992).

6


109

Fig. 6.8 Intense surface bloom off the coast of Namibia at about 20S latitude on April 4, 2009, showing spectra (left) of a bright bloom (top) and high MCI blooms (bottom). The bright bloom (top arrow) is visible in the true colour image and also in the fluorescence image (FLH). The high MCI blooms are centred in an area of high FLH. The “Broad MCI” image avoids negative signals (black) due to the effect of FLH on MCI

Blooms are most often of centric or pennate diatoms, and are associated with ice platelets in near-surface water. The diatoms appear to have bloomed in a water layer 1–2 m below floating sea ice, attached to the platelets. Blooms are observed when an icebreaker penetrates the ice, or in the case of satellites, when surface currents advect the bloom into open water. Observers on ice-breakers commonly note the brown colour of the undersides of overturned floes as well as the brown water in leads, and splashed onto surface snow. Figure 6.9 shows the monthly MCI composite computed for Antarctica for February 2007. This is a re-projection of data similar to the lower panel of Fig. 6.1. An area in the south-west Weddell Sea shows high MCI signal. This is where a superbloom event was reported in February/March 1968, and near where events were reported in other coastal areas of the Weddell Sea in October/November 1986 and Febrary/March 1983 (Smetacek et al., 1992). The spectrum in Fig. 6.10 shows that the high MCI signal is due to a strong radiance peak at 709 nm with an associated dip at 665 nm and 681 nm, similar to the spectra in Fig. 6.7. This dip can be interpreted as a reduction in the radiance back-scattered by ice, due to absorption by chlorophyll a near 670 nm. The time series of total MCI signal summed over all longitudes south of 60S latitude (Fig. 6.10) shows an apparent increase in this type of bloom in Antarctic waters between 2003 and 2009.

110


Fig. 6.9 An area of high MCI index (upper left), indicating a high levels of chlorophyll a among ice in the Weddell Sea. The image is a monthly composite of daily global MCI composites for February 2007. The inset shows spectra in the high-MCI area (red) compared to a nearby openwater area (green). The difference (blue) is plotted using the right-hand axis. Other areas off Wilkes Land (bottom right) and along the ice edge off Queen Maud Land (top) also show high values of the index

5 Floating Sargassum Benthic species of Sargassum are common in many areas of the world. Freelyfloating (pelagic) species, Sargassum natans and fluitans, are common in the Gulf of Mexico and the tropical north Atlantic, but they have only recently been detected in satellite images (Gower et al., 2006). The major reason for its non-detection in the past has been the lack of a combination of sensor bands that provides a definitive signal in the presence of cloud, haze and sun glint. This is now rectified by the MCI of MERIS. Since the first detection using satellite imagery of Sargassum in the north-western Gulf of Mexico in May and June of 2005, we have collected several images of dense aggregations in the Gulf Stream extension area of the western Atlantic east of Cape Hatteras, in October and November of 2006 and 2007, such as Fig. 6.11.

6


111

350

MERIS count (thousands)

300 250 200 150 100 50 0 2002

2003

2004

2005

2006

2007

2008

2009

2010

Start of Year

Fig. 6.10 The time series of total bloom intensity south of 60S summed over all longitudes. Data show an apparent increase in total occurrence of this type of bloom in March of the years 2003–2009. MCI also indicates a smaller (spring) peak in October/November of each year

The difference spectrum (blue) in the inset shows the “red edge” characteristic of land vegetation, with a shift in wavelength of the effective “red edge” position due to water absorption. The spectrum in Fig. 6.11 results in an MCI value of about 2.0 mW/(m2 · nm · sr). The peak in the visible spectrum (400–700 nm) at 620 nm is consistent with the brown colour of Sargassum. Interpretation as Sargassum is also based on the shape of the patches (especially the lines extending to over 100 km in length), continuity of patterns over a several-month period, and lack of indication in the satellite data of any high background chlorophyll concentration (for example by FLH, as in Fig. 6.8), which would indicate a dense concentration of phytoplankton. This Sargassum can be tracked using the global MERIS MCI data set provided by ESA’s G-POD system. Results show an annual cycle of Sargassum distribution in the Gulf of Mexico and North Atlantic, with considerable interannual variability. The satellite data do not show any sign of similar populations of long-lived pelagic Sargassum in other oceans of the world, though we note a possible population in Mozambique Channel (Table 6.1). The monthly composites of MCI signal at 5 km spatial resolution for the Gulf of Mexico and north-west Atlantic were analyzed by computing the frequency distribution (histogram) of MCI values in each one-degree square, and assuming that all MCI values exceeding the mean ocean background value in that one-degree square by a threshold amount, indicate presence of Sargassum. The threshold value, t in Eq. (2), was taken as 0.4 mW m–2 nm–1 sr–1 . The resulting count was then taken as proportional to the total amount of Sargassum in each one-degree square.

112


Fig. 6.11 Floating Sargassum in the Gulf Stream near 63 W, 37 N imaged as MCI on October 22, 2007 by ESA’s ocean sensor, MERIS at its full spatial resolution of 300 m. Colours indicate MCI value as shown by the colour bar. The Sargassum is collected into patches and long lines by surface convergence and shear. The difference spectrum in the inset shows the “red edge” characteristic of land vegetation, with the apparent red edge position shifted to a shorter wavelength

Figure 6.12 shows the spatial distributions of MERIS counts of Sargassum for the months June to September (left to right), the season when Sargassum is observed to leave the Gulf and move into the Atlantic by way of the Loop Current and Gulf Stream. High concentrations of Sargassum are indicated in the northern Gulf of Mexico for most years in July, and Sargassum then appears in a broad area of the Atlantic to the east of Cape Hatteras (35–40 N, 45–75 W) in July and August, as best shown in 2005–2008. In 2005, the large amount of Sargassum in the Gulf in June to August has vanished in September. Figure 6.13 shows plots of estimated total amounts of Sargassum in the Gulf of Mexico (15–30 N, 80–100 W) and the western North Atlantic (22–40 N, 20–80 W), for each month from June 2002 to December 2008. MERIS counts are scaled to millions of tons (wet weight) by comparison with ship observations (Parr, 1939, Butler et al., 1983). Highest amounts are in May, June, July in the Gulf of Mexico, with values increasing through 2003 and 2004 to a maximum in 2005. Amounts in the Atlantic increase each year after July and drop back to low values in the spring of the following year, with the exception of 2008. Figure 6.14 summarises the seasonal movement of Sargassum in this area as shown by our observations. Sargassum grows in the northwest Gulf of Mexico between March and June each year and appears in the Atlantic, north and east of Cape Hatteras, starting in about July. In all years except 2008, we observe low total Sargassum amounts in the Atlantic before this annual injection from the Gulf of Mexico. Observations for 2003–2007 suggest that most Sargassum has a life-time of one year or less, with the major “nursery area” being in the northwest Gulf of

6


113

Fig. 6.12 MERIS counts of Sargassum for the years 2002–2008 (top to bottom) for months June to September (left to right) for the area 10–50 N latitude, 18–98 W longitude. Pixels measure one degree in latitude and longitude. The background dark blue colour corresponds to no detections. Increasing amounts are indicated by the colour sequence shown in Fig. 6.11. One-degree squares where MCI gives strong signals from coral reefs and other benthic vegetation are masked to black. Land at 0.25 degree spatial resolution is shown in grey

Mexico. If Sargassum were longer-lived, we would expect to observe more evidence of Sargassum in the Atlantic northeast of the Bahamas, in February to May in these years. Only in 2008 do observations show significant concentrations in this area, with circulation back into the Gulf Stream. The 2008 circulation pattern would be consistent with the traditional picture of the Sargasso Sea as being the main repository of Sargassum biomass. Satellite images clearly provide greatly improved data coverage compared to ship surveys of Sargassum, but with limitations due to spatial resolution, cloud cover and sun glint. We note that the satellite may miss significant quantities of Sargassum if it is too evenly distributed or mixed beneath the surface by wind. In the future, satellite observations can continue to provide a lengthening time series of data of the type we present here, and can play an important role in selecting the sampling pattern for any future ship survey.

114


Total Sargassum (million tons)

7 6

Gulf of Mexico 5

Atlantic

4 3 2 1 0 2002

2003

2004

2005 2006 Start of year

2007

2008

2009

Fig. 6.13 Plots of total amounts of Sargassum for the Gulf of Mexico (15–30 N, 80–100 W) and the western Atlantic (22–40 N, 40–80 W), for the period June 2002 to December 2008. Values are MERIS counts converted to tons by comparison with ship observations. In the Gulf of Mexico large amounts were detected in May, June, and July of 2003, 2004 and 2005, with a maximum Sargassum year in 2005. In the Atlantic, amounts increase after July and usually drop back to low values by March, but higher amounts were observed throughout 2008. Very little Sargassum was detected in 2002

6 Taihu Lake, China Eutrophic lakes can have more extreme optical properties than ocean and coastal waters. Blooms showing a high MCI signature are common in many lakes. We choose Taihu Lake in China as an example of the contribution MERIS can make to studying bloom events in such bright and highly variable water. Taihu Lake is the third largest freshwater lake in China. It covers an area of 2,420 km2 , 68 km north to south, 56 km west to east between Jiangsu and Zhejiang Provinces, about 100 km west of Shanghai. The Lake forms the heart of an important network for agricultural and drinking water supply, draining into the Yangtze River (Chang Jiang). The lake’s environmental problems include accelerated eutrophication caused by nitrogen and phosphorus enrichment, leading to frequent blooms and oxygen depletion. These have affected drinking water supplies to nearby cities of Wuxi and Suzhou. Figure 6.15 shows a MERIS FR image of the lake on April 25, 2008. The RGB image (top left) shows most of the lake to be bright with suspended matter. This could be inorganic sediment or a bright bloom of a species not indicated by MCI. The MCI image (top right) shows a strong bloom along the southwest shore (white and red), in water that is darker than most of the lake. This bloom gives spectra of two types, those that show the red edge near 700 nm (A, lower left) indicating a surface slick, and those that show the isolated peak at 709 nm (B, lower right) indicating a bloom dispersed in surface water.

6


115

Fig. 6.14 Simplified outline diagram showing the average seasonal extent of Sargassum, based on MERIS count distributions by month. In each year, Sargassum is first observed in the Gulf of Mexico in about March, by July it is still present in the Gulf, but has also appeared off Cape Hatteras. It then moves east and then south west in the Atlantic, as shown. In the period 2003–2007, MERIS sees relatively little Sargassum in the Atlantic between March and June except in 2008 (Fig. 6.13)

In our initial attempt to study the long-term changes in Taihu Lake, we used the MCI computed with Eq. (1) as part of our global data set. However, the water in this lake is often so bright that the near-infrared threshold radiance is exceeded. In addition, for Taihu Lake, aerosol loading is often high, also increasing these radiances. G-POD was therefore specially tasked to produce monthly composites of the small area, about 100 km square, centred on Taihu Lake, at the higher spatial resolution of 1,200 m, corresponding to the “reduced resolution” (RR) of MERIS and without applying the threshold at 865 nm. The resulting monthly time series of MCI images is shown in Fig. 6.16. Statistics of bloom occurrence can be deduced from the MCI data once a threshold signal has been selected as indicating a bloom, corresponding to the value “b” in Eq. (2). For Taihu Lake this value was set to the apparently constant) winter value and counts were computed with t = 0. Figure 6.17 shows the average MCI count over the whole area of the Lake and nearby open water in Fig. 6.17. MERIS shows that this average count has increased over the period June 2002 to March 2009. The values need to be compared with in-situ observations of blooms and nutrient inputs to the Lake in order to understand the observed changes. Figure 6.18 shows the improved coverage given by the modified algorithm, which removes one cause of an apparent seasonal cycle. The true annual cycle is still

116


Fig. 6.15 MERIS FR images and spectra of blooms in Taihu Lake on April 25, 2008. The RGB image (top left) shows most of the lake to be bright with suspended sediment. The MCI image shows a strong bloom along the southwest shore (white and red), giving spectra that near 709 nm show both the red edge (A, lower left) indicating a surface slick, and the isolated peak (B, lower right) indicating a bloom dispersed in surface water

strong, as shown in Fig. 6.19, peaking in both May and September. MERIS detects very little bloom activity in December to March, with a minimum in January and February. Activity rises to a maximum in May, with a secondary peak in late summer and fall (August, September, October). Blooms in Taihu Lake have also been monitored by MODIS and other satellites using the NDVI computed from bands at 645 nm and 855 nm (Ma et al., 2008, Xu et al., 2008). Compared to MODIS, MERIS has the advantage of the additional band at 709 nm, better defining the red edge spectral shape and determining the MCI (Maximum Chlorophyll Index). On the other hand, MODIS has the advantage of a wider swath width, leading to about twice the spatial coverage of MERIS and the possibility of imaging the Lake on more days. MODIS also has the advantage of providing data from two satellites on each day with swaths offset in time and space.

6


117

Fig. 6.16 Monthly time series of MCI signal in Taihu Lake from June 2002 to March 2009. Data are derived from global composite images at 1,200 m resolution formed from all available MERIS data

7 6

Average MCI

5 4 3 2 1 0 2002

2003

2004

2005 2006 2007 Start of year

2008

2009

2010

Fig. 6.17 Average MCI signal of the lake showing increased bloom activity since 2002. Black squares indicate the February to November average for each year. This has increased by about a factor 2.8 over the time of observation (average of years 2006–2008 compared to 2002–2004)

The improved spectral information of MERIS compared to MODIS, gives quantitative information for many bloom events in Taihu Lake. The time series computed from these values for 2002–2008 shows a significant increase in bloom intensity over this period, suggesting a serious and continuing deterioration of water quality in the lake. The time series from MERIS is so far relatively short, but will continue.

118


Fraction imaged

1 0.8 0.6 0.4 0.2 0 2002

2003

2004

2005 2006 2007 Start of year

2008

2009

2010

Fig. 6.18 MERIS MCI coverage by month, comparing the standard global method (Eq. (1), black line) with the modified method for the high-signal Taihu Lake (red line). With the standard method, coverage was lower in spring and summer, apparently due to more haze, cloud and brighter water 3.5 3

Average MCI

2.5 2 1.5 1 0.5 0 1

2

3

4

5

6 7 8 Month of year

9

10

11

12

Fig. 6.19 Average annual cycle of bloom frequency in Taihu Lake, based on observations of MCI from June 2002 to October 2008. The thin black line shows the annual cycle of signals measured by the standard global algorithm (Eq. (1)) compared to the algorithm with threshold removed and higher spatial resolution (red line)

7 False Alarms for Bloom Detection As noted above, high MCI signals in the Gulf of Mexico and north-west Atlantic are often due to Sargassum. This can be considered a “false alarm” for phytoplankton bloom detection, though it can be argued that both targets contain high concentrations of chlorophyll-a, which give rise to “red-edge” spectra in both cases. Many

6


119

blooms also give the “709 peak” type of spectrum, which we have not observed from Sargassum. Targets in Mozambique Channel, listed in Table 6.1, may be partly or entirely Sargassum. We also observe strong signals from coral reefs and other areas of marine and aquatic vegetation. Again, these indicate chlorophyll a, but not plankton blooms A more definitely “false” signal can be caused by cosmic ray hits on elements of the CCD detectors of MERIS. Since the detectors cover all wavelengths and all look directions of MERIS, the hits can cause a false high radiance at any apparent wavelength or direction. Since cosmic rays are due to single particles, the signals will affect single CCD elements and should tend to show as bright single pixels on images formed from individual bands. They are especially noticeable in products such as MCI or FLH which emphasize small relative differences. In MCI and FLH, they can also show as dark single pixels if they increase the apparent radiance of a baseline. Figure 6.20 shows the global spatial distribution of bright single pixel MCI events computed for the month of July 2006. These are located by examining 3 by 3 pixel areas of the global daily composites (Fig. 6.1 upper panel) and counting areas where the eight edge pixels all have valid MCI (that is, excluding “no-data”, cloud or land), where the standard deviation of the eight edge pixels is below a threshold S, and where the centre MCI value exceeds the average of the eight edge pixels by more than a threshold V. This will select for pixels that are bright relative to uniform neighbouring pixels. For Fig. 6.20, S = 0.05 mW/(m2 · sr · nm) and V = 0.3 mW/(m2 · sr · nm). Colours from dark blue to green, yellow, red and white show increasing numbers of events in a given one-degree area. Because of the geometrical projection used for the composites, this algorithm will detect fewer events at higher latitudes, cutting off at about 60 degrees north and south latitude.

Fig. 6.20 The global spatial distribution of single pixel events in MCI detected by MERIS, showing the higher concentrations in the South Atlantic Anomaly for the month of July 2006

120


The coloured area in Fig. 6.20 corresponds to a large number of single-pixel events over the South Atlantic off Brazil at the position of the South Atlantic Anomaly. This is observed in all months. The anomaly causes high cosmic ray intensity at the altitude of Envisat. In other parts of the world in Fig. 6.20, most events are due to vegetation on isolated coral and other reefs. MCI values are computed only over cloud-free water, so that persistent cloud (for example, off Peru) blocks detection of events.

8 Conclusions The ability of the MCI index derived from MERIS data to detect and map the variety of events shown here, indicates the importance of the spectral band at 709 nm. Although the MCI will not detect all of the harmful algal blooms (HABs) that are causing increasing damage to aquaculture, tourism and coastal water quality, it clearly has a role in detecting and monitoring some surface, high-chlorophyll events, occurring under clear-sky conditions. It also has a role in mapping blooms in ice and turbid water and in mapping floating vegetation. The present nearly seven years of data shows blooms over an increasing range of areas and gives improving statistics on seasonal and interannual variability. The data record is not yet long enough to show climate-related and other longer-term trends. Future observations based on MCI depend on maintaining the capability provided by MERIS. The present US sensors SeaWiFS and MODIS and the planned future sensor VIIRS lack the band at 709 nm. The major limitation of MERIS, especially when compared to MODIS, is in its relatively narrow swath width, which at 1,150 km is about half that of MODIS. This is not wide enough to allow daily coverage of all areas. Also, there are at present two MODIS instruments in orbit, on the Terra and Aqua satellites, again increasing coverage. Ideally, future versions of the MERIS sensor would image over an increased swath width, and would also have two sensors in orbit at all times. Acknowledgements This work was supported by Fisheries and Oceans Canada and by the Canadian Space Agency (CSA) under the GRIP (Government Related Initiative Program). Satellite imagery was provided by ESA under the Announcement of Opportunity (AO) program for scientific research applications of MERIS imagery. Processing of the global composite data is by ESA’s G-POD (GRID Processing On Demand) system. We are grateful to students Lindsay Orr, Sara Statham and Sara Fissel from the University of Victoria (BC, Canada) Co-operative Education program for the processing of some images and spectra.

References Butler JN, Morris BF, Cadwallader J, Stoner AW (1983) Studies of Sargassum and of the Sargassum community. Bermuda biological station, special publication no. 22, pp 1–85. European Space Agency (ESA) (2008) http://envisat.esa.int/instruments/meris/. Accessed Jan 2009. Gitelson A (1992) The peak near 700 nm on radiance spectra of algae and water: relationships of its magnitude and position with chlorophyll concentration. Int J Remote Sens 13:3367–3373.

6


121

Gower JFR, Doerffer R, Borstad GA (1999) Interpretation of the 685 nm peak in water-leaving radiance spectra in terms of fluorescence, absorption and scattering, and its observation by MERIS. Int J Remote Sens 9:1771–1786. Gower JFR, King S, Borstad GA, Brown L (2005) Detection of intense plankton blooms using the 709 nm band of the MERIS imaging spectrometer. Int J Remote Sens 26:2005–2012. Gower JFR, Hu C, Borstad GA, King S (2006) Ocean color satellites show extensive lines of floating Sargassum in the Gulf of Mexico. IEEE Trans Geosci Remote Sens 44:3619–3625. Gower JFR, King S (2007) An Antarctic ice-related “superbloom” observed with the MERIS satellite imager. Geophys Res Lett 34. doi:10.1029/2007GL029638. Gower JFR, King SA, Goncalves P (2008a) Global monitoring of plankton blooms using MERIS MCI. Int J Remote Sens 29:6209–6216. Gower JFR, King SA, Borstad GA, Brown L (2008b) The importance of a band at 709 nm for interpreting water-leaving spectral radiance. Can J Remote Sens 34:287–295. Ma R, Kong F, Duan H, Zhang S, Kong W, Hao J (2008) Spatio-temporal distribution of cyanobacteria blooms based on satellite imageries in Lake Taihu, China. J Lake Sci 20:687–694. Parr AE (1939) Quantitative observations on the pelagic Sargassum vegetation of the western north Atlantic. Bulletin of the Bingham Oceanographic Collection, Peabody Museum of Natural History, Yale University, no 6, 94 pp. Rast M, Bezy JL, Bruzzi S (1999) The ESA medium resolution imaging spectrometer MERIS a review of the instrument and its mission. Int J Remote Sens 20:1681–1702. Smetacek V, Scharek R, Gordon LI, Eicken H, Fahrbach E, Rohardt G, Moore S (1992) Early Spring phytoplankton blooms in ice platelet layers of the southern Weddell Sea, Antarctica. Deep-Sea Res I 39:153–168. Xu, J, Zhang B, Li F, Song K, Wang Z (2008) Detecting modes of cyanobacteria bloom using MODIS data in Lake Taihu. J Lake Sci 20:191–195. Yacobi YZ, Gitelson A, Mayo M (1995) Remote sensing of chlorophyll in Lake Kinneret using high spectra resolution radiometer and Landsat TM: Spectral features of reflectance and algorithm development. J Plankton Res 17:2155–2173.

Chapter 7

Evaluating Sea Ice Deformation in the Beaufort Sea Using a Kinematic Crack Algorithm with RGPS Data K. Peterson1 and D. Sulsky

Abstract Sea ice in the Arctic plays an important role in the Earth’s climate and has been an early indicator of global warming. Remote sensing data, particularly synthetic aperture radar (SAR) imagery of Arctic regions, can help us understand the complex processes controlling sea ice dynamics and thermodynamics. The RADARSAT Geophysical Processor System (RGPS) was developed by the Polar Remote Sensing Group at the Jet Propulsion Laboratory (JPL) to extract sea ice motion data from SAR imagery. A set of points initially forming a regular grid is tracked providing displacements at each point over time. If the set of points in the original configuration is viewed as the vertices of square cells then the motion of the points determines the deformation of the cells. With this interpretation, grid quantities such as divergence, shear, and vorticity can be approximated using the nodal displacements. In this work an alternative approach for evaluating the deformation of sea ice from these data is proposed that uses the deformation gradient and a kinematic crack algorithm. With this approach, the deformation in a cell is assumed to be due solely to a displacement discontinuity caused by a crack. Then a minimization approach is used to find the crack dimension and orientation for each cell that best fits the nodal displacements. The kinematic crack algorithm is illustrated using RGPS data for a period of 16 days in February and March of 2004 for a region in the Beaufort Sea lying between Banks Island in the East and Point Barrow in the West. Keywords Sea ice · RADARSAT Geophysical Processor System · Cracks · Beaufort Sea · Kinematics · Finite strain · Strong discontinuities · Deformation gradient

K. Peterson (B) Sandia National Laboratories, Albuquerque, NM, USA e-mail: [email protected] 1 Sandia is a multiprogram laboratory operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Company, for the U. S. Department of Energy’s National Nuclear Security Administration under Contract DE-AC04-94AL85000.


123

124

K. Peterson and D. Sulsky

1 Introduction The effects of global warming are being seen dramatically in the Arctic where the loss of sea ice is both an indicator of climate change and, through feedback mechanisms, a driver of climate change. The extent and thickness distribution of the sea ice play important roles in the energy balance of the Arctic where sea ice mediates the heat transfer between the ocean and atmosphere. Sea ice acts as an insulator and as it freezes in the winter, it would grow to an equilibrium thickness of about 2–3 m if it were stationary. However, the ice is not stationary. Its motion is driven by ocean currents and atmospheric winds. In order to move, the ice cover continuously breaks up and refreezes. The leads, or cracks in the ice, can be a kilometer wide and up to hundreds of kilometers long. New ice is formed primarily in leads where open water, exposed to the cold atmosphere, freezes quickly and ejects brine in the process. Thus, leads additionally dominate the brine flux into the ocean mixed layer. As leads close, ice piles up into pressure ridges, or is forced down into keels, creating thicker ice and increasing drag. In addition, the greater albedo of the ice compared to the water results in more solar radiation and a cooler atmosphere. Because of this interplay, climate models must account for coupled atmosphere, ice and ocean dynamics, and ice models must include the effects of leads and ridging, as well as the thermodynamic processes of melting and freezing. Remote sensing, in particular spaceborne, synthetic aperture radar (SAR) imagery from the RADARSAT satellite, has played an important role in understanding the complex way sea ice responds to dynamic mechanical and thermal forcing. RADARSAT’s wide-swath mode has provided basin-scale, high-resolution (~100 m) SAR maps of the western Arctic Ocean. The SAR data are acquired and assembled into images at the Alaska Satellite Facility (ASF) in Fairbanks, Alaska. The imagery is then processed in the RADARSAT Geophysical Processor System (RGPS) developed at the Jet Propulsion Laboratory (Kwok and Cunningham, 2000; Kwok et al., 1990). The sampling period is typically about 3 days; however, there are periods and areas that are more finely sampled. Taking advantage of this detailed temporal and spatial information, RGPS uses kinematic analyses to examine ice motion and deformation (Kwok, 2006, Kwok and Cunningham, 2002; Kwok et al., 1995). These analyses are able to resolve long linear features in the pack ice that evolve with time and can generally be associated with leads. The activity, persistence, orientation, and the length scale of these discontinuities are remarkable and enlightening. It is possible to observe the temporal development of basin-wide deformation in the form of ice opening, closing and shear, and to estimate ice production and thickness (Kwok, 2006, Kwok and Cunningham, 2002). The large-scale expressions of the deformation of the ice cover resulting from localized and smallscale deformations are also important for use in validating the results of sea ice models (Kwok, 2006; Kwok et al., 2008; Lindsay et al., 2003). As more satellite data become available it is important to consider what the appropriate metrics are for analyzing the data and comparing with numerical simulations. The RGPS produces measures such as divergence, vorticity and shear, which

7

Sea Ice Deformation Using a Kinematic Crack Algorithm

125

are calculated under the assumption of continuous displacement fields. If the majority of the deformation in the ice is focused in leads, which are cracks where the displacement is discontinuous, these measures may not be appropriate. Moreover, these quantities are rates (units of 1/day). So, for example, the divergence field shows regions that are currently opening or closing, but does not show regions that are open but not currently deforming. Additional methods for evaluating remote sensing data for the Arctic could provide a more complete picture of the evolving ice cover. In this article we extend the RGPS kinematic analysis by considering the deformation gradient and its determinant. This analysis is motivated by standard kinematic descriptions of large deformations of solids. The determinant of the deformation gradient measures area changes in two dimensions and identifies regions that have diverged or converged. Given the importance of leads in ice deformation, it is also of interest to re-examine the deformation with an eye toward identifying discontinuous motion. The main contribution of this article is a new method for decomposing RGPS products into fields that depict the amount of lead opening and gives their spatial orientation and distribution. The next section describes the current RGPS products relevant to the analysis in this article. In Sect. 3 the deformation gradient is defined and its use in examining persistence of leads is illustrated for a region of the Beaufort Sea. Section 4 discusses the kinematics of a discontinuous displacement field and Sect. 5 presents an algorithm to determine the crack that best fits the observed ice deformation. This kinematic crack algorithm is illustrated for the same region of the Beaufort Sea. Concluding remarks are given in Sect. 6.

2 RADARSAT Geophysical Processor System Data The fundamental quantity produced by RGPS is trajectories of sea-ice points. At an initial time, a set of points forming a square grid is located in a SAR image. In images resulting from subsequent satellite passes, the original points are found again using area-based and feature-based tracking. The time separation between repeat observations is variable and is based on available coverage. The time interval between successive images is called the timestep. This procedure provides the trajectory of each point as these points move with the ice cover. Since the same set of points is tracked over an ice season, RGPS provides a densely sampled Lagrangian picture of the motion, similar to what would be obtained by a large array of drifting buoys. See Fig. 7.1 for an example. Secondary procedures in RGPS derive estimates of ice deformation. If the set of points in the original configuration is viewed as vertices of cells then the motion of the points determines the deformation of the cells. With this interpretation, grid quantities can be computed that help provide a picture of the rate and type of ice deformation. Measures such as the divergence, shear, and vorticity can be approximated using the cell vertex velocity. Since these cells are Lagrangian, it is known (for example, from the finite element literature) that remeshing is required for accurate results when cell deformations are large. In RGPS, nodes are added locally

126


(a)

(b)

Fig. 7.1 Satellite view of small region with grid at (a) initial and (b) final times

to avoid problems due to mesh distortion. For example, if the relative change in length of one of the line segments making up the cell boundary exceeds a threshold (eg. 1.1) then a point is added along the edge (Kwok et al. 1995). If (u, v) are components of the velocity of a point (x, y) then the divergence, shear, and vorticity are defined as ∂u ∂v + divergence = ∂x ∂y 1/2 ∂u ∂v 2 ∂u ∂v 2 shear = + (1) − + ∂x ∂y ∂y ∂x vorticity =

∂v ∂u − . ∂x ∂y

If (x, y) represents the current position of a point tracked by the RGPS and (¯u, v¯ ) represents the displacement of the point over a timestep, t, then the new position of the point is (x + u¯ , y + v¯ ). If the velocity of the point is assumed constant over the timestep, the relationship between velocity and displacement is (u, v) = (¯u, v¯ )/t. Discrete approximations of the divergence, shear and vorticity for an RGPS cell can be constructed from the velocity using the following formulae for the partial derivatives

∂u ∂x ∂u ∂y ∂v ∂x

h = h = h =

Nvert 1 1 (ui+1 + ui )(yi+1 − yi ) a 2

1 a 1 a

i=1 N vert

i=1 N vert

i=1

1 (ui+1 + ui )(xi+1 − xi ) 2 1 (vi+1 + vi )(yi+1 − yi ) 2

(2)

7


∂v ∂y

h =

127

Nvert 1 1 (vi+1 + vi )(xi+1 − xi ). a 2 i=1

In these formulae, the brackets and the superscript h denote discrete approximations of the average value of the derivative over the cell, where a is the cell area, Nvert is the number of cell vertices making up the cell, ui and vi are the components of the velocity at vertex i, xi and yi are the components of the position vector for vertex i, and the vertex index is cyclical (i.e. 1 + Nvert = 1) (Kwok and Cunningham, 2002). The cell area, a, can be approximated as Nvert 1 a= (xi yi+1 − xi+1 yi ). 2

(3)

i=1

The RGPS deformation product consists of cell area changes and spatial derivatives computed using the ice velocity at the cell vertices. Associated with each cell is a unique identifier, the time of its creation, the map coordinates of the cell center, the displacement of the cell center between this and the last observation, the timestep, and the approximate derivatives as defined above. An illustration of RGPS data is shown in Fig. 7.1, where satellite views of a 50 km by 50 km region of Arctic ice have a 5 km × 5 km RGPS grid superimposed. The time span between the first and second observation is 18.5 h and the satellite images were recorded in mid May 2002. This region undergoes a fairly simple deformation over this time, which can be seen in the grid displacements where a large shear band is visible. Using the discrete definitions of divergence, shear, and vorticity shown above, plots of these fields for this small region are shown in Fig. 7.2. Notice the primary band of deformation visible in all three plots. For larger regions the deformation is typically much more complex and therefore not so easy to interpret. To illustrate analysis of a larger region over several timesteps, the RGPS ice deformation product over 16 days from February 23 to March 10, 2004 is used for a region in the Beaufort Sea, which lies between Banks Island in the East and Point Barrow in the West. During this time, daily satellite coverage is obtained. Plots of divergence for day 54 (23 February), day 62 (3 March), and day 69 (10 March) of 2004, are shown in Fig. 7.3 where the orange region is land, the North coast

(a)

(b)

Fig. 7.2 Small ice region (a) divergence, (b) shear, and (c) vorticity

(c)

128

K. Peterson and D. Sulsky DivergenceDay62

0.1 800

0.08 0.06

600

0.06

600

0

200

400

0.02 0

200

−0.02 0

0 −200

−0.1

0.02 0

200

−0.02 0

−0.04 −0.06

−0.06 −200

−0.08 −400 −2400 −2200 −2000 −1800 −1600 −1400

0.04 400

−0.04

−0.06 −200

0.06

−0.02

−0.04

0.1 0.08

600

0.04 Y (km)

Y (km)

0.02

800

0.08

0.04 400

DivergenceDay69

0.1

Y (km)

DivergenceDay54

800

−0.08

−0.08 −400 −2400 −2200 −2000 −1800 −1600 −1400

−0.1

−400 −2400 −2200 −2000 −1800 −1600 −1400

X (km)

X (km)

X (km)

(a)

(b)

(c)

−0.1

Fig. 7.3 Beaufort Sea region divergence (a) day 54, (b) day 62, (c) day 69 of 2004. The orange region is the North coast of Alaska between Banks Island in the East and Point Barrow in the West. West is at the top of the figures

of Alaska. In the plots a yellow color indicates little or no deformation in a cell. Red indicates positive divergence or areas of opening and blue indicates negative divergence or areas of closing. These plots depict area changes over the timestep. Similar plots of the shear and vorticity for the same region are shown in Figs. 7.4 and 7.5. By definition, shear is always positive. Red indicates regions of large shear. Vorticity measures the local rotation. Positive (counterclockwise) rotation is shown in red and negative (clockwise) rotation is indicated by blue. The scale for each of the quantities is 1/day. Note that linear features in the data are apparent when plotting these quantities, indicating localized deformation in leads. Note also that the divergence, shear, and vorticity are rates and therefore the plots only display regions that are deforming over the timestep of 1 day. Regions with areas of open water that were active previously, but are quiescent over this time will not be visible on the plots. There is little deformation apparent in Figs. 7.3c, 7.4c and 7.5c, but leads may still be present. In the following sections, we consider the deformation gradient which can provide information about persisting leads. ShearDay54

ShearDay62

800

0.08 0.06

0.06

600

0

Y (km)

Y (km)

400

0.02

200

0.02 0

200

0

−0.04

0

−0.04

−200

X (km)

(a)

0.02 0

200

−0.02 0

−0.04 −0.06

−200 −0.08

−0.08

−0.08 −0.1

0.04 400

−0.06

−0.06

−400 −2400 −2200 −2000 −1800 −1600 −1400

0.06

600

−0.02

−0.02

−200

0.08

0.04

0.04 400

0.1 800

0.08

Y (km)

600

ShearDay69

0.1

0.1 800

−400 −2400 −2200 −2000 −1800 −1600 −1400 X (km)

(b)

−0.1

−400 −2400 −2200 −2000 −1800 −1600 −1400

−0.1

X (km)

(c)

Fig. 7.4 Beaufort Sea region shear (a) day 54, (b) day 62, (c) day 69 of 2004. The orange region is the north coast of Alaska between Banks Island in the East and Point Barrow in the West. West is at the top of the figures

Sea Ice Deformation Using a Kinematic Crack Algorithm VorticityDay54

VorticityDay62 0.1

800

0.06

0.06

600

400 Y (km)

Y (km)

0

0

−0.04

0 200

0

−0.04

0.02 0

200 −0.02 0

−0.04

−0.06 −200

−0.06 −200

−0.08 −0.1

0.04 400

−0.02

−0.06

−400 −2400 −2200 −2000 −1800 −1600 −1400

0.06

600

0.02

−0.02

−200

0.08

0.04

0.02

200

0.1 800

0.08

0.04 400

VorticityDay69 0.1

800

0.08

600

129

Y (km)

7

−0.08 −400 −2400 −2200 −2000 −1800 −1600 −1400

−0.1

−0.08 −400 −2400 −2200 −2000 −1800 −1600 −1400

X (km)

X (km)

(a)

(b)

−0.1

X (km)

(c)

Fig. 7.5 Beaufort Sea region vorticity (a) day 54, (b) day 62, (c) day 69 of 2004. The orange region is the north coast of Alaska between Banks Island in the East and Point Barrow in the West. West is at the top of the figures

3 Continuum Kinematics Measures of local deformation play a prominent role in the analysis of solids and their constitutive behavior. Analysis of solids typically involves a Lagrangian description of the motion. Specifically, for the two-dimensional ice cover, assume the ice occupies a region 0 = (0) ⊂ R2 initially and (t) ⊂ R2 for t > 0. In a continuum, material points in the original configuration are labeled with coordinates X. At time t > 0 the current position of the point that started at X is x = ϕ(X, t) with ϕ(X, 0) = X. The function ϕ is called the deformation mapping of the material. Let X be a point and X + d X be a neighboring material point, with x + dx being the corresponding spatial point. Then for small dX, we have xi + dxi = ϕi (X + dX, t) ∼ ϕi (X, t) +

∂ϕi ∂ϕi (X, t)dXJ = xi + (X, t)dXJ ∂XJ ∂XJ

(4)

which leads to the differential relation dxi =

∂ϕi (X, t)dXJ . ∂XJ

(5)

Here the indices i and J represent vector components and run from 1 to 2. Thus, the deformation mapping of an infinitesimal material vector dX at X is completely determined by the deformation gradient Fi,J =

∂ϕi (X, t). ∂XJ

(6)

The deformation gradient is a key quantity in continuum mechanics and is the derivative of the current position with respect to the original position. It can also be written in direct notation as F(X, t) = ∂ϕ(X, t)/∂X = GradX ϕ.

(7)

130


The deformation gradient is a linear transformation with positive determinant and has a 2 × 2 matrix representation for the two-dimensional ice cover. The Jacobian of the transformation from original to current coordinates is J(X, t) = det(F(X, t)) >0. The Jacobian transforms area elements in the original configuration, dA, to area elements in the current configuration, da = J dA. The time derivative of the Jacobian is related to the divergence of the velocity field, ∂J/∂t = J∇ · v. A discrete, deformation gradient for each RGPS cell can be estimated using information from the RGPS deformation product. Let X represent the initial position of a cell center. If the state of the cell at the time it is first recorded is used as an undeformed reference configuration, then the deformation gradient is the identity tensor, I, at that time. At a later time, tn , the cell center is located at xn = ϕ(X, tn ). At the next observation time, tn+1 , the cell center is located at xn+1 = ϕ(X, tn+1 ). The deformation gradient at time tn+1 can be updated from the deformation gradient at the previous observation at time tn using the chain rule Fn+1 =

∂ϕ(X, tn+1 ) ∂xn+1 ∂xn ∂xn+1 ∂ϕ(X, tn ) ∂xn+1 n F · = = = ∂X ∂xn ∂X ∂xn ∂X ∂xn

(8)

Now if we write xn+1 = xn + u(xn , tn ) where u(xn , tn ) is displacement between xn+1 and xn , then Fn+1 = f n+1 Fn

(9)

where f n+1 = I +

∂u(xn , tn ) . ∂xn

(10)

Cell approximation of the displacement derivatives in the above formula for the tensor f n+1 can be found by multiplying the velocity derivatives in the RGPS deformation product by the timestep, t. Thus, we can easily update the approximate deformation gradient for each cell using Eqs. (9) and (10). Determinant FDay54

Determinant FDay62

800

800

800 1.2

1.2

1.2

600

600

600 1.1

1.1

1.1 400

200

0.9

0

−200

−400 −2400 −2200 −2000 −1800 −1600 −1400 X (km)

(a)

1

200

0.9

0

0.8

−200

0.7

−400 −2400 −2200 −2000 −1800 −1600 −1400

Y (km)

400 1

Y (km)

400 Y (km)

Determinant FDay69

1

200

0.9

0

0.8

0.8

−200

0.7

−400 −2400 −2200 −2000 −1800 −1600 −1400

X (km)

(b)

Fig. 7.6 Beaufort Sea region det F (a) day 54, (b) day 62 and (c) day 69

X (km)

(c)

0.7

7


131

Figure 7.6 shows the det F calculated over the 16 day period of observation from February 23 to March 10, 2004, and displayed for days 54, 62 and 69 in 2004. The plot in Fig. 7.6a gives similar information to Fig. 7.3a since the rate of opening is equivalent to the accumulated opening after 1 day. Fig. 7.6b and c show the areas where opening and closing persist and can be contrasted with the rate of opening shown in Fig. 7.3b and c. In this manner we see that the information given by det F gives additional insight into the ice deformation.

4 Discontinuous Kinematics Since leads are formed by cracks in the ice, the displacement associated with a lead is discontinuous. The article by Coon et al. (2006) introduced the idea of separating the displacement into the sum of a continuous and discontinuous part. An analysis of RGPS deformations was performed assuming that the predominant term was the jump discontinuity in displacement. This procedure fit a jump in displacement, or a crack, in each RGPS cell that could best account for the observed deformation. However, the ideas in (Coon et al., 2006) are based on theories that are appropriate for small deformations of the ice. Since sea ice can undergo large deformations, it is of interest to extend the analysis to that regime. Large deformation theories in solid mechanics are based on the deformation gradient. Part of our motivation to analyze ice deformation in terms of predicting where leads are located in cells comes from a new elastic-decohesive constitutive model for sea ice (Schreyer et al., 2006). In this constitutive model, the ice is an elastic solid for small strains, but the ice fails and a displacement discontinuity occurs when the traction on a surface reaches a critical value. The fracturing of ice directly models the formation of leads. This constitutive model can be formulated in the framework of strong discontinuities developed by Simo et al. (1993) and applied to a finite strain regime by Armero and Garikipati (1996) and Garikipati (1996). Since the constitutive model predicts leads, it is useful to process observations of ice to extract lead formation in order to validate the model. As in the last section, consider a local region ⊂ R2 with coordinates labeled X, which is the reference configuration of the ice. Let ϕ : 0 ×[0, T] → R2 be the map that defines the deformation. Now consider a local region where a discontinuity is formed along a surface with unit normal N as shown in Fig. 7.7. Assume that the − discontinuity splits this local region into two parts, + and . The deformation ¯ and a jump discontinuity, ϕ and mapping can be divided into a smooth part, ϕ, can be written as ¯ ϕ(X, t) = ϕ(X, t) + ϕ(X, t) H (X) where the Heaviside function on is defined as

1 for X ∈ + H (X) = · 0 for X ∈ −

(11)

(12)

132


Fig. 7.7 Ice region with discontinuity surface

The deformation gradient corresponding to ϕ(X, t) is then calculated as F = GradX ϕ = GradX ϕ¯ + GradX ϕ H + ϕ ⊗ N δ

(13)

where δ is the delta function along and the tensor product (⊗) is defined so that (a ⊗ b = ai bj ) for any vectors a and b. Now define the regular part of F as F = GradX ϕ¯ + GradX ϕ H

(14)

F = F + ϕ ⊗ N δ .

(15)

and then

Alternatively, a multiplicative decomposition of the deformation gradient can be defined as F = F(I + F

−1

( ϕ ⊗ N )δ )

(16)

In the following section, this form will be used in the kinematic crack algorithm.

5 Finite Strain Crack Kinematics Using the formalism of strong discontinuities, assume that the region is associated with an RGPS grid cell undergoing deformation, where displacements at the vertices of the cell are known. To find the best fit discontinuity for the displacement field, the deformation gradient on each cell must be approximated from the RGPS data and then fit to a deformation gradient of the form shown in Eq. (16). First, however, the singularity in the deformation gradient must be regularized over the cell. To do this consider an integral over a grid cell and use the definition of the δ-function to obtain

7


F

−1

ϕ ⊗ N δ d =

F

133

−1

ϕ ⊗ N dS.

(17)

This relationship implies a length scale defined as

d

L=

(18)

dS

When calculating the jump in displacement using finite strain crack kinematics, this length scale is approximated by the cell spacing and the regularized deformation gradient over the cell is 1 −1 F =F I+ F ϕ ⊗ N . L h

(19)

Define a scaled displacement jump as J = F−1 [ϕ]/L and the deformation gradient simplifies to Fh = F I + J ⊗ N .

(20)

For the kinematic crack calculations the change in area of the cell is assumed to be entirely due to a displacement discontinuity. This implies that the regular portion of the deformation gradient is an area preserving transformation and, therefore, can only be a rotation. In the first simple case, assume the amount of rotation is zero and therefore F is equal to the identity, I. The deformation gradient over a cell will then be of the form (I + J ⊗ N). For each cell also assume that the discontinuity passes through the center and that J is constant over a linear . With these assumptions, Fh − I is expected to be a rank one matrix formed from the two vectors J and N. Starting with the discrete cell value, Fh , calculated from the RGPS data, obtain the singular value decomposition (SVD) of Fh − I defined as Fh − I = UVT

(21)

where U and V are unitary matrices and = diag(σ1 ,σ2 ) where σ1 ,σ2 are the singular values. For a rank one matrix, σ2 is identically zero, and therefore, the error associated with how far (Fh − I) is from rank one is just the magnitude of σ2 . Writing the components of the assumed form of Fh − I using the components of J = (J1 , J2 ) and N = (N1 , N2 ) gives Fh − I =

h h F11 − 1 F12 h h F21 F22 − 1

=

J1 N1 J1 N2 J2 N1 J2 N2

.

(22)

134


The SVD in component form is F −I= h

U12 U22

U11 U21

σ1 0 0 σ2

V11 V21

V12 V22

.

(23)

The closest rank one matrix to Fh − I is the matrix produced from the SVD with σ2 = 0 (Golub and Van Loan, 1996). In terms of the components of U and V this matrix becomes U11 σ1 V11 U11 σ1 V21 . (24) U21 σ1 V11 U21 σ1 V21 Assuming that N is normalized, the components of J and N can be equated to the SVD components in the following way J1 = σ1 U11 J2 = σ1 U21 N1 = V11 N2 = V21

(25)

Then the spatial jump is calculated as ϕ = LFJ = LJ

(26)

Plots of the cracks predicted for the small Arctic region discussed above are shown in Fig. 7.8. Each cell has a crack, which is assumed to go through the center of the cell and extend the length of the cell. The dimensions and orientation of the quadrilaterals representing the cracks in the figure are obtained from [ϕ] and N. The cracks can both open and shear depending on the normal and tangential components

1.5

x 105

1.5

1.4

1.4

1.3

1.3

1.2

1.2

1.1

1.1

1

1

0.9 −5.6

−5.5

−5.4

−5.3

−5.2

(a)

−5.1

−5

−4.9 x 105

x 105

0.9 −5.6

−5.5

−5.4

−5.3

(b)

−5.2

−5.1

−5

−4.9 x 105

Fig. 7.8 Calculated Discontinuities for Small Region with F = I (a) Cutoff = 0, (b) Cutoff = 0.8 km

7


135

of the jump. The normal component is in the direction of N and the tangential component is orthogonal to N. A red color in the plots indicates an opening mode where [ϕ] · N > 0 and a blue color indicates a closing mode where [ϕ] · N < 0. Note that this calculation assumes that there is a crack in every cell as shown in Fig. 7.8a. However, a cutoff for cracks with less than a minimum opening magnitude can be chosen to limit the display to substantial cracks as shown in Fig. 7.8b where the cutoff has been set to be 0.8 km. The cutoff is used under the assumption that cells predicted to have small cracks probably to do not crack in reality. (In a dynamic, mechanical analysis, as opposed to this kinematic analysis, cells undergoing a small deformation would not crack, but would deform elastically, or possibly plastically.) In the next more difficult case, F is assumed to be a general rotation, R, which must be determined in addition to J and N. In order to determine J and N for this case, first calculate the singular value decomposition of Fh as Fh = R(I + J ⊗ N) = U1 1 VT1 ·

(27)

1 = UT1 R(I + J ⊗ N)V1 = UT1 RV1 + UT1 RJ ⊗ NV1 .

(28)

Solving for 1 we get

Since U1 and V1 are unitary matrices, R = UT1 RV1 is a rotation. Additionally, define new vectors J = UT1 RJ and N = NV1 . Then 1 can be written as 1 = R + J ⊗ N.

(29)

In this form it is apparent that 1 − R is a rank one matrix. Therefore, the next step in the algorithm is to find an intermediate rotation matrix, R, such that 1 − R is as close to rank one as possible. The matrix R can be written as R=

cos θ sin θ − sin θ cos θ

(30)

for an unknown θ. Then 1 − R can be written as sin θ σ1 − cos θ 1 − R = − sin θ σ2 − cos θ

(31)

where σ1 and σ2 are the components of 1 . For 1 − R to be rank one, the second singular value of the SVD for the matrix must be zero. R can be found by minimizing the second singular value of 1 − R, which is equivalent to minimizing the smallest eigenvalue of (1 − R)T (1 − R). This matrix can be written as (1 − R)T (1 − R) = and has eigenvalues (λ1,2 ),

σ12 − 2σ1 cos θ + 1 sin θ (σ2 − σ1 ) sin θ (σ2 − σ1 ) σ22 − 2σ2 cos θ + 1

(32)

136


2λ = σ22 − 2(σ2 + σ1 ) cos θ + σ12 + 2± 1,2 2 σ22 − 2(σ2 + σ1 ) cos θ + σ12 + 2 + 4 sin2 θ (σ1 − σ2 )2 . − σ12 − 2σ1 cos θ + 1 σ22 − 2σ2 cos θ + 1

(33)

The value of θ producing the minimum eigenvalue is calculated numerically using a Matlab minimization subroutine fminbnd. Given this minimizing θ, R is calculated from Eq. (30). Next, the SVD of 1 − R is calculated as T

1 − R = UV .

(34)

As in the previous case with no rotation, the closest rank one matrix to 1 − R is the matrix produced from the above SVD with the second singular value equal to zero. Assuming this matrix is equal to

U 11 σ 1 V 11 U 11 σ 1 V 21 U 21 σ 1 V 11 U 21 σ 1 V 21

,

(35)

the components of the intermediate vectors J and N are J 1 = σ 1 U 11 J 2 = σ 1 U 21 . N 1 = V 11 N 2 = V 21

(36)

R, J, and N can then be calculated from the intermediate values as R = U1 RVT1 J = R T U1 J . N = V1 N

(37)

Finally, to plot the jump in spatial coordinates, calculate [ϕ] as [ϕ] = LFJ = LRJ.

(38)

Plots of the cracks predicted for the small Arctic region using the algorithm allowing a general rotation for F are shown in Fig. 7.9. As before, cracks that are colored red indicate an opening mode and cracks that are colored blue indicate a closing mode. The first figure plots cracks in all cells and the second uses a cutoff of 0.8 km as in the previous case. Note that the majority of cracks along the large deformation band are similar using the two techniques. This method can also be applied to the Beaufort RGPS data. However, in this case there is not a single deformation over some time interval, but a series of data for multiple days between 23 February and 10 March. The best fit discontinuities have been calculated using the total deformation gradient for each day for the Beaufort region and are shown using a cutoff of 0.8 km for day 54 in Fig. 7.10a and for subsequent

7


1.5

x 105

1.5

1.4

1.4

1.3

1.3

1.2

1.2

1.1

1.1

1

1

0.9 −5.6

−5.5

−5.4

−5.3

−5.2

−5.1

−5

(a)

x 105

0.9 −5.6

−4.9

137

−5.5

−5.4

−5.3

−5.2

−5.1

−5

(b)

x 105

−4.9 x 105

Fig. 7.9 Calculated discontinuities for small region with no restrictions on F (a) cutoff = 0, (b) cutoff = 0.8 km CracksDay62

CracksDay69

800

800

600

600

600

400

400

400

200

Y (km)

800

Y (km)

Y (km)

CracksDay54

200

200

0

0

0

−200

−200

−200

−400 −2400

−2200

−2000

−1800

−1600

−1400

−400 −2400

−2200

−2000

−1800

−1600

−1400

−400 −2400

−2200

−2000

−1800

X (km)

X (km)

X (km)

(a)

(b)

(c)

−1600

−1400

Fig. 7.10 Beaufort Sea region predicted leads (a) day 54, (b) day 62, and (c) day 69

days 62 and 69 in Fig. 7.10b and c. As before cracks that are colored red indicate opening and cracks that are colored blue indicate a closing mode. Additionally, the dimensions of the quadrilaterals representing the cell cracks in the figures are oriented along the direction of the crack and have a width determined by the jump in displacement, [ϕ]. Coherent regions of cracks with similar directions and opening magnitudes can be seen in Fig. 7.10, particularly along the coast. This may be indicative of the existence of larger continuous cracks in the region and is analogous to the coherent crack orientations for the smaller region experiencing the simple shearing motion displayed in Fig. 7.9. Note that the regions of activity are similar to those in the plots of the deformation gradient where a measure of accumulated opening and closing is represented. However, the crack representation provides, in addition to the accumulated opening and closing, a more detailed portrait of the orientation of the cracks and the components of the cumulative opening and closing along the crack direction.

138


6 Conclusions Spaceborne remote sensing and the products provided by RGPS have given us remarkable new insight into the dynamics of sea ice motion and deformation. Among other purposes, these data and insights are invaluable in creating and validating new models. Conversely, new ice models can motivate alternative methods for extracting information from remote sensing data. Consideration of solid models appropriate for representing large mechanical deformation of sea ice, led us to extend the RGPS deformation product to include calculation of the deformation gradient. The determinant of the deformation gradient measures accumulated area changes and shows persistent converged or diverged regions. Motivated by a new constitutive model for sea ice that treats ice as an elastic solid which can fail and form leads, this article also describes a new method to extract information from RGPS data about the likely size and orientation of leads. The calculation of the crack size and orientation is done through a straightforward optimization procedure assuming that all the deformation in a cell results from a displacement discontinuity. Results from this calculation could be compared with sea ice model results using new metrics described in Levy et al. (2008). The algorithm presented in this article for associating a crack through the center of a cell to account for the observed deformation of that cell, gives a unique prediction of the size and orientation of the crack. However, if one examines a cell with four vertices, for example, and just considers the displacement of the four points over a timestep, there are multiple ways to account for that motion using jumps in displacement with different combinations of opening and shear centered at different locations in the cell. Thus, more work needs to be done before we conclude that the predictions of this algorithm are accurate. Ultimately, we may find that a cell-by-cell analysis is inadequate and a more global analysis is necessary to faithfully predict lead extent and orientation. The assumption that the crack passes through the center of the cell could be relaxed and then continuity of cracks between adjacent cells could be enforced instead. This might give a better approximation of the orientation of large cracks that pass through several cells. Acknowledgments Thanks to Ron Kwok of JPL for providing the RGPS data used in this analysis and for many enlightening discussions about sea ice. Thanks also to Giang Nguyen for assistance in analyzing some of the data. The support of the National Science Foundation under grant ARC0621173 is also gratefully acknowledged.

References Armero F, Garikipati K (1996) An analysis of strong discontinuities in multiplicative finite strain plasticity and their relation with the numerical simulation of strain localization in solids. Int J Solids Struct 33:2863–2885. Coon M, Kwok R, Levy G, Pruis M, Schreyer H, Sulsky D (2006) Arctic ice dynamics experiment AIDJEX assumptions revisited and found inadequate. J Geophys Res 112:C11S90. doi:10.1029/2005JC003393.

7


139

Garikipati K (1996) On strong discontinuities in inelastic solids and their numerical simulation. Doctoral dissertation, Stanford University. Golub GH, Van Loan CF (1996) Matrix Computations. The Johns Hopkins University Press, London. Kwok R (2006) Contrasts in sea ice deformation and production in the Arctic seasonal and perennial ice zones. J Geophys Res 111. doi:10.1029/2005JC003246. Kwok R, Cunningham GF (2000) RADARSAT geophysical processor system data user’s handbook. JPL report, JPL D-19149. Kwok R, Cunningham GF (2002) Seasonal ice area and volume production in the Arctic Ocean: November 1996 through April 1997. J Geophys Res 107(C10). doi:10.1029/2000JC000469. Kwok R, Curlander JC, McConnell R, Pang SS (1990) An ice-motion tracking system at the Alaska SAR facility. IEEE J Ocean Eng 15:44–54. Kwok R, Hunke EC, Maslowski W, Menemenlis D, Zhang J (2008) Variability of sea ice simulations assessed with RGPS kinematics. J Geophys Res 113. doi:10.1029/2008JC004783. Kwok R, Rothrock DA, Stern HL, Cunningham GF (1995) Determination of ice age using Lagrangian observations of ice motion. IEEE Trans Geosci Remote Sens 33(2):392–400. Levy G, Coon M, Nguyen G, Sulsky D (2008) Metrics for evaluating linear features. Geophys Res Lett 35:L21705. doi:10.1029/2008GL035086. Lindsay RW, Zhang J, Rothrock DA (2003) Sea ice deformation rates from satellite measurements and in a model. Atmos-Ocean 41(1):35–47. Schreyer H, Monday L, Sulsky D, Coon M, Kwok R (2006) Elastic- decohesive constitutive model for sea ice. J Geophys Res 111:C11S26. doi:10.1029/2005JC003334. Simo JC, Oliver J, Armero F (1993) An analysis of strong discontinuities induced by strainsoftening in rate-independent inelastic solids. Comput Mech 12:277–296.

Chapter 8

Satellite Air – Sea Fluxes Abderrahim Bentamy, Kristina B. Katsaros, and Pierre Queffeulou

Abstract This chapter addresses the estimation of global surface winds, surface wind stress, latent heat flux, and sensible heat flux over the oceans with high spatial and temporal resolution using satellite radar and radiometer measurements. An overview of the physics of remotely sensed data, of methods and algorithms used to retrieve surface fluxes is provided. The retrievals are used to estimate regular in space and time surface parameters, requested for oceanic forcing function, over global ocean. The characteristics of the former are investigated at global and regional scales. Keywords Remotely sensed data · Turbulent fluxes · Air sea interaction · Calibration and validation · Scatterometers · Altimeters · Radiometers · Surface wind vectors · Sea state

1 Introduction The large exchanges of energy between ocean and atmosphere through air-sea fluxes at the interface, the absorption of radiation from the sun in the upper ocean, and the redistribution of heat by the ocean circulation at all time and space scales, characterize the main role of the ocean in climate variability. Surface fluxes of momentum, heat, and water vapor provide some of the dominant processes that are involved. For several reasons and especially at large scales, the measurement of the relevant oceanic surface properties is quite difficult. It is common to use parameterization methods to estimate surface fluxes based on the knowledge of some basic variables such as surface wind, sea surface temperature, air temperature, and surface and air humidity. The latter can be estimated from buoy, ship, and satellite data. We usually rely on the bulk aerodynamic formulae that parameterize the fluxes in terms of the A. Bentamy (B) Institut Français pour la Recherche et l’Exploitation de la MER (IFREMER), Plouzané, France e-mail: [email protected]


141

142

A. Bentamy et al.

observed mean quantities. The surface fluxes derived from satellite observations are expressed as follows: τ = (τx , τy ) = ρ CD U(u, v)

(1)

Qlatent = −lρ CE U(qa − qs )

(2)

Qsens = −ρCp CT U(Ta − Ts )

(3)

where τ is the vector wind stress with zonal τx and meridional τy wind stress components; Qlatent and Qsens are the latent and sensible heat fluxes; U is the magnitude of the surface wind vector (wind speed) at 10 m height under neutral stratification which has zonal u and meridional v vector wind components; qa and qs are the air and surface (or saturation) specific humidity; Ta is the dry bulb temperature; Ts is the sea surface temperature; l is the coefficient for latent heat of evaporation considered as constant 2.5 Jkg–1 ×106 Jkg–1 ; ρ is the air density at observation level, calculated from mean surface temperature and sea-level pressure using the ideal gas equation with a correction for the virtual temperature to compensate for the behavior of moist air; CP the specific heat at constant pressure is approximated to be constant 1.0 Jkg–1 K–1 ×103 Jkg–1 K–1 . CD and CE and CT are the bulk drag coefficient, the transfer coefficient for water vapor, and the transfer coefficient for sensible heat, respectively. Since our estimates of the surface fluxes are based on the bulk approach, their quality would be related to the accuracy of surface wind, air and sea temperature, and of air and near surface humidity. This paper describes the methods and the algorithms used to retrieve these parameters from radar and radiometer measurements onboard polar orbiting satellites.

2 Remotely Sensed Data 2.1 Scatterometer 2.1.1 General Topics Since 1991 five scatterometers have been launched onboard polar-orbiting satellites: European Remote Sensing Satellites 1 and 2 (ERS-1/2), Advanced Earth Observing Satellites 1 and 2 (ADEOS-1/2), QuikScat and METOP. The scatterometer is an active radar sending microwave pulses to the ocean surface and measuring the power backscattered from surface roughness. The backscatter is mainly related to the small centimeter waves on the surface. Indeed, it was established that the ocean surface ripples are in equilibrium with local wind stress. Jones et al. (1978) showed, based on measurements from aircraft experiments that for incidence angle greater than 20◦ , the backscatter coefficient increases with respect to wind speed. They also demonstrated the anisotropic characteristics of the scattering. It was established that the backscatter coefficient σ 0 is not only a function of wind speed, but also of

8

Satellite Air – Sea Fluxes

143

wind direction relative to the radar azimuth. The scatterometer is the unique radar providing wind speed as well as wind direction over the global ocean. The study of the relationship between σ 0 measurements and the surface wind vector is still ongoing: indeed, many current works aim to establish a physical backscatter model. However, the theory relating wind speed to wave generation and equilibrium spectrums is not well developed. Therefore, only empirical models are currently determined and used to establish a relation between the backscatter coefficient and wind speed and direction for some specific incidence angles, radar azimuth, and polarization. The European Space Agency launched two scatterometers using identical instruments onboard ERS-1 (August 1991) and ERS-2 (April 1995). Both are composed of three antennas (fore-, mid-, and aft-beam) operating at C-band (5.33 GHz) with only vertical polarization (VV). ERS scatterometers scan a 500 km swath on one side of the satellite, and measure at three azimuth angles: 45◦ , 90◦ , and 115◦ . The incidence angle varies from 17◦ to 46◦ for the mid beam and from 25◦ to 57◦ for fore- and aft-beams (Fig. 8.1). The scatterometer swath is divided into cells of 50 km × 50 km separated by 25 km distance. Hereafter, the scatterometer cell over the ocean is referred to as a wind vector cell (WVC). Over each WVC, a backscatter coefficient might be provided by each antenna. They are used to calculate speed and direction through inverse and direct models. Two kinds of ERS scatterometer winds are available. Near real time data processed by ESA, and off line processed, archived, and distributed by the Centre ERS d’Archivage et de Traitement (CERSAT/IFREMER). The latter are called WNF (WiNd Field). The calibration and validation of the algorithms were performed with dedicated buoy data during the RENE91 experiment, with the National Oceanic and Atmospheric Administration (NOAA) National Data Buoy Center (NDBC) buoys and the Tropical Ocean Global Atmosphere (TOGA) Tropical Atmosphere Ocean (TAO) buoys. The accuracy of the wind speed and direction derived from the IFREMER algorithm is about 1 m/s and 14◦ . The validation of the off-line wind products indicated that, at low wind speeds, data are less accurate in wind speed and direction determination (Graber et al., 1996). In August 1996, the National Aeronautics and Space Administration (NASA) launched the scatterometer called NSCAT on board Japanese satellite ADEOS-1

METOP-A

Fig. 8.1 Satellites carrying scatterometers launched since 1991

144

A. Bentamy et al.

(or Midori). It is in a circular orbit with a period of 101 min at an inclination of 98.59◦ and at a nominal height of 796 km with a 41-day repeat cycle. NSCAT had two 600 km wide swaths located on each side of the satellite track and separated by 300 km (Fig. 8.1). It operated at 14 GHz (Ku band). Its fore-beam and aft-beam antennas pointed at 45◦ and 135◦ to each side of the satellite track, respectively. The mid-beam pointed at 65◦ and 115◦ depending on the NSCAT swath. The fore and aft-beams provide σ 0 measurements with vertical polarization and incidence angle varying between 19◦ and 63◦ . The mid-beam provided two σ 0 measurements corresponding to vertical and horizontal polarizations with an incidence angle varying between 16◦ and 52◦ . The spatial resolution of the instrument on the earth’s surface was about 25 km. Following theADEOS-1 breakdown, NASA launched the SeaWinds scatterometer onboard the QuikSCAT satellite on 19th July 1999. This satellite operated for 10 full years. QuikScat/SeaWinds had a rotating antenna with two differently polarized emitters: the H-pol with incidence angle of 46.25◦ and V-pol with incidence angle of 54◦ (Fig. 8.1). The inner beam had a swath of about 1400 km, while the outer beam swath was 1800 km width. The spatial resolution of SeaWinds (oval footprint) was 25 km × 35 km. The latter were binned over the scatterometer swath into WVC of 25 km × 25 km. There are 76 WVC across the satellite swath, and each contains the center of 10–25 measured σ 0 . The remotely sensed wind vectors are estimated from the scatterometer σ 0 over each WVC using the empirical model QSCAT-1 relating the measured backscatter coefficients to surface winds. The standard SeaWinds wind retrievals are referenced as L2B products. They have been calculated using the standard scatterometer method based on the Maximum Likelihood Estimator (MLE) (JPL, 2001). The scatterometer retrieval algorithm estimates several wind solutions for each wind cell. In general, there are four solutions. The ambiguity removal method is then used to select the most probable wind solution. The latter are used in this study. To improve the wind direction, especially in the middle of a swath, where the azimuth diversity is quite poor, an algorithm called Direction Interval Retrieval with Threshold Nudging (DIRTH) is used too. SeaWinds is a Ku band radar. Therefore, rain has a substantial influence on its measurements. Previous studies (Sobieski et al., 1999) showed that the rain impact may attenuate the scatterometer signal, resulting in wind speed underestimation, or change the surface shape due to raindrop impact and splatter, leading to an overestimation of the retrieved winds. The SeaWinds wind products involve several rain flags determined from the scatterometer observations and from the collocated radiometer rain rate onboard other satellites. An identical SeaWinds scatterometer was launched by NASA onboard the second Japanese satellite, ADEOS-2, in December 2002. It operated until June 2003. The QuikScat/SeaWinds surface wind estimations will be indicated by QuikScat hereafter. The latest remotely sensed surface wind-measuring instrument is the Advanced SCATterometer (ASCAT). It was launched aboard the European Meteorological Satellite Organization (EUMESAT), MetOp-A on October 19, 2006. Scientific and technical documentation related to ASCAT physical measurements as well as to

8


145

ASCAT derived products may be found at the EUMETSAT web site http://www. eumetsat.int/Home/Main/Publications/Technical_and_Scientific_Documentation/ Technical_Notes/ and under EUMETSAT Ocean & Sea Ice Satellite Application (O&SI SAF) web site (http://www.osi-saf.org/). MetOp is in a circular orbit (near synchronous orbit) for a period of about 101 min, at an inclination of 98.59◦ and at a nominal height of 800 km with a 29-day repeat cycle. ASCAT has two swaths 550 km wide, located on each side of the satellite track, separated by 700 km. It operates at 5.3 GHz (C band). Its fore-beam and aft-beam antennas point at 45◦ and 135◦ on each side of the satellite track, respectively. The mid-beam antennas point at 90◦ . The ASCAT beams measure normalized radar cross sections with vertical polarization, σ0 , which are a dimensionless property of the surface, describing the ratio of the effective echoing area per unit area illuminated. The fore and aft-beams provide backscatter coefficient measurements at incidence angle varying between 34◦ and 64◦ . The mid-beams provide σ 0 measurements at incidence angle varying between 25◦ and 53◦ . Two Backscatter coefficient spatial resolutions are available over global ocean: 25 km and 12.5 km.

2.1.2 Scatterometer Wind Retrievals Retrieving wind velocity from sea state is a not trivial inverse problem. Indeed, results obtained via boundary-layer theories give relations to link a given wind vector over the sea surface to momentum exchange between air and sea. This momentum is then related to sea roughness properties (wave height, slope, etc.). Nevertheless, the inverse problem (from a value of sea roughness to an associated wind vector) is not yet fully based on theory. Another attempt deals with global ocean wind sea retrieval. Indeed, the actual knowledge of the atmospheric boundary layer is more concerned with a sea at an equilibrium state than for specific regions (closed seas with limited fetch). The model function which relates a wind vector to a sea state has then to be a Global Model Function (GMF). The general GMF form used for scatterometers is based on a truncation of the Fourier expansion of σ 0 over the azimuth angle range: σ 0 (U, ϕ, θ , P) = A0 (U, θ , P)+A1(U, θ , P)×cos (ϕ)+A2(U, θ , P)× cos (2ϕ)

(4)

Where ϕ is the difference between the wind direction and measurement azimuth, U the wind speed, θ the incidence angle, and P the polarization. The GMF and the inverse algorithm are supposed to be valid for the global oceans. Therefore some local events (in space and time) that might modify the ripple wave spectrum and then degrade the scatterometer retrieved wind vectors are not explicitly taken into account. Examples of such effects include: the interaction of short waves with longer ocean surface waves, the damping of waves trough natural or artificial surface slicks, the impact of the atmospheric boundary layer stability on the generation of ripple waves, and other local sea state characteristics. The impact of such perturbations might be detected through quality control procedures.

146

A. Bentamy et al.

Fig. 8.2 Behaviour of the backscatter coefficient (σ 0 ) measured by the ERS scatterometer as a function of relative wind direction for three wind speeds (columns) and three incidence angles ranges (rows). The solid line indicates σ 0 estimated from GMF Eq. (1) while dots indicate the measured σ 0

The determination of GMF model coefficients A0, A1, and A2 is performed using a control optimal method minimizing the difference between measured and simulated (from GMF) backscatter coefficients. The latter are estimated from collocated buoy and/or numerical weather prediction (NWP) wind speed and direction. Figure 8.2 shows an example of behavior of measured (dots) and simulated (line) σ 0 as a function of wind direction for three wind speed and incidence angle ranges. The maximum σ 0 values are reached for relative wind direction of 0◦ (upwind) and 180◦ (downwind). The minimums are located at 90◦ and 270◦ (crosswind). The determination of surface wind speed and direction from the knowledge of measured backscatter coefficients over a given WVC, requires some assumptions. First we assume that measured σ 0 are expressed as σ 0 = σP0 + ε

(5)

where σP0 represents « truth » for the backscatter coefficient and ε is the error related to instrument and physics of the measurement, surface conditions, and to the calibration and validation procedures. ε is assumed Gaussian with zero mean and variance δ ε .

8


147

It is also assumed that σP0 is related to GMF through: 0 σP0 = σmod + εmod

(6)

σ 0 mod is backscatter coefficient value estimated from (4), and εmod is the model error assumed Gaussian with δεmod variance. For a given wind speed and direction over WVC, the difference between measured and simulated backscatter coefficients is calculated: 0 Δ = σ 0 − σmod

(7)

Assuming that instrumental and model errors are independent, Δ is gaussian with zero mean and variance δΔ = δε + δε mod Therefore the probability density function of Δ constrained by σ 0 becomes: 1 Δ2 P(Δ/σ ) = P(Δ/{U, ϕ}) = √ exp − 2δΔ 2π δΔ 0

(8)

Let us consider N to be the number of σ 0’s over WVC (3 in the case of ERS), and that the corresponding Δ ’s are independent. The conditional probability is then provided by: P (Δ1 . . . ΔN /{U, ϕ}) =

N i=1

N

Δ2 1 i √ exp − 2δΔi 2π δ Δi i=1

(9)

The maximum likelihood estimator (MLE) criterion implies that the solution {U, ϕ} is the local minimum of P. In general, over each WVC the wind speed and direction solutions are determined as a maximum of the following function: J(U, ϕ) =

N

(σ 0 − σ 0 mod (U, ϕ))2 i

i=1

i

δ Δi

+ ln (δΔi )

(10)

J is related to P through a logarithmic transform. The algorithm proposes up four solutions, called ambiguities. The most probable vector is indicated as the selected wind vector for the specific WVC. This selection is mainly based on the MLE and quality control (QC) use (See for instance Quilfen, 1995; Stoffelen et al., 1997; Freilich and Dunbar, 1999; Thiria et al., 1993). Examples of selected wind speed and direction derived from QuikScat measurements are shown in Fig. 8.3. Each panel presents wind speed and direction estimated over WVC of an available QuikScat swath crossing the Mediterranean Sea. 2.1.3 Scatterometer Wind Accuracy The accuracies of scatterometer retrieval wind speed and direction are commonly determined through comparisons with buoy wind measurements. Four buoy

148

A. Bentamy et al.

Fig. 8.3 Example of retrieved wind speed (in colour) and direction (arrow) estimated over QuikScat swath from 12th to 15th November 2001. The approximated swath date is shown in the top left area of each panel (Year, Month, Day, Hour)

networks are used to estimate the quality of the retrieved scatterometer wind vectors: the National Data Buoy Center (NDBC) buoys-off the U.S. Atlantic, Pacific and Gulf coasts maintained by the National Oceanic and Atmospheric Administration (NOAA); the Tropical Atmosphere Ocean (TAO) buoys located in tropical Pacific Ocean and maintained by the NOAA Pacific Marine Environmental Laboratory (PMEL); the European buoys-off European coasts called ODAS and maintained by U.K. Met office and Meteo-France; and the Pilot Research Moored Array in the Tropical Atlantic (PIRATA) moored in the Tropical Atlantic ocean and maintained by the Institut pour la Recherche et le Développement (IRD), the Instituto Nacional de Pesquisas Espaciais (INPE), and PMEL. Collocated Buoy and Scatterometer Data For each buoy network and scatterometer (ERS-1/2, NSCAT, and QuikScat), the spatial collocation between anemometer and remotely sensed data is achieved by selecting satellite WVCs which fall within a 2◦ ×2◦ square centered around the buoy location (Longitude and latitude). Temporal collocation is performed by

8


149

choosing the buoy observation closest to the time of satellite overpass. In general the buoy observations are hourly reported. Hourly PIRATA data are calculated from 10-minute observations. Only available and validated (based on quality control procedures) buoy and scatterometer data are used in the comparisons. Furthermore, for accuracy purposes, only buoys located in deep water and far enough from coast are considered because no shallow water effects are taken into account. The calculation of buoy wind speed at 10 m height in neutral conditions is performed using boundary layer model (Liu et al., 1979). For the four networks, only hourly buoy wind speed and direction estimates are used in the scatterometer/buoy wind comparisons. Statistical Parameters Comparison procedures are based on the following statistical parameters: X = E(X) σX = E(X − E(X))2 γX =

(11) (12)

E(X − E(X))3 (E(X − E(X))2 )3/2

(13)

E(X − E(X))4 (E(X − E(X))2 )2

(14)

KX =

X stands for wind speed (or wind component) variable. E stands for the first conventional moment. The surface wind variable is often considered as stochastic. Therefore, it may be described using linear moments in addition to using conventional moments; the advantage of using linear moments is their small sensitivity to erroneous measurements and/or estimates that yields outliers in data (Hosking, 1990). The nth linear moment is defined as 1 λn =

x(F)Pn−1 (F) dF

(15)

0

F is the probability function of X, x(F) is the inverse function of F, called quantile function, P∗n are orthogonal polynomial functions related to the Legendre polynomials through P∗n = Pn (2s − 1), s∈ [0,1]. Finally, the comparisons are also characterized by the linear regression coefficients; Let X and Y be two wind variables (model and satellite), the linear dependence between them is described as: Y˜ = aX + b a=

Syx and b = E(Y) − aE(X) Sxx

(16)

150

A. Bentamy et al.

Syx = E(Y − E(Y))E(X − E(Y)) and Sxx = E(X − E(X))2 As the error characteristics of the model and the scatterometer are not known, it is more efficient to estimate the symmetrical regression coefficients: Y˜ s = as X + bs as =

Syy Sxx

(17)

the correlation coefficient is defined as Syx . ρ=√ SxxSyy

(18)

For wind direction, the parameters mean difference (9), standard deviation of the difference (10), and vector correlation (11) are used. They take into account the circular behaviour of such variables. −1 < sin(Db − Ds) > D = tan (19) < cos(Db − Ds) > Db and Ds are the collocated model and satellite wind directions, respectively. σD = sin−1 (ε) (1 + 0.1547 ε 3 ) From Yamartino (1984) ε = 1 − (( sin(δD))2 + (cos(δD))2 ) Where sinδD =< sin(Db − Ds) > and cos(δD) =< cos(Db − Ds)> ρ 2 = Tr (Σ11 )−1 Σ12 (Σ22 )−1 Σ21 ij

(20)

(21)

is the cross-covariance matrix of the wind vector and Tr states for matrix track.

Results Figures 8.4 and 8.5 show scatterometer wind speed and direction accuracy results, respectively. They show scatter plots of the comparisons of ERS-1, ERS-2, and QuikScat scatterometer wind speeds and directions with 10-m neutral winds derived from buoys moored in Atlantic (including NDBC and ODAS), : Pacific (NDBC), and in Tropical (including TAO and PIRATA) zones. The remotely sensed and buoy winds compare well. In general, correlation coefficients exceed 0.8 and rms differences are lower than 2 m/s for wind speed and 25◦ for wind direction. The main discrepancies are found for low wind speed conditions. Excluding buoy winds less than 5 m/s, the rms values drop to 1.2 m/s for wind speed and 18◦ for wind direction. However, the mean differences indicate a slight underestimation of ERS, and an overestimation of QuikScat wind speeds with respect to buoy measurements. Indeed, the bias values are about 0.4 m/s, 0.7 m/s, and –0.4 m/s for ERS-1, ERS-2,

8


151

Fig. 8.4 Comparison of the wind speeds (left panel) and directions (right panel) observed by ERS1 (top), ERS-2 (middle), and QuikScat (bottom) scatterometers with 10-m buoy winds moored in the Atlantic Ocean (first column), the Pacific Ocean (second column), and in the Tropical oceans (third column)

Fig. 8.5 As Fig. 8.4 for wind direction comparisons

152

A. Bentamy et al.

and QuikScat, respectively. Using a large database involving a collocated buoy and satellite data set, empirical models are under development to reduce the remotely sensed wind biases.

2.2 Special Sensor Microwave Imager (SSM/I) Since 1990 the SSM/I radiometers onboard the DMSP F10, F11, F13, F14, and F15 satellites provide measurements of the surface brightness temperatures at frequencies of 19.35, 22.235, 37, and 85 GHz (hereafter referred to as 19 GHz, 22 GHz, 37 GHz, and 85 GHz), respectively. Horizontal and vertical polarization measurements are taken at 19 GHz, 37 GHz, and 85 GHz. Only vertical polarization is available from 22 GHz. Due to the choice of the channels operating at frequencies outside strong absorption lines [for water vapor] (50–70 GHz), the radiation observed by the antennae is a mixture of radiation emitted by clouds, water vapor in the air and the sea surface, as well as radiation emitted by the atmosphere and reflected at the sea surface. For estimation of the 10-m wind speed from SSM/I brightness temperatures, we used an algorithm published by Bentamy et al. (1999). This algorithm is a slightly modified version of that published by Goodberlet et al. (1989) that includes a water vapor content correction. The SSM/I wind speeds are calculated over swaths of 1394-km width, with a spatial resolution of 25 km×25 km. Previous studies investigated the accuracies of the retrieved SSM/I winds through a comparison with wind speed and direction measured by moored buoys in several oceanic regions (Bentamy et al., 2002). The retrieved wind speed was calculated from brightness temperature measurements provided by NASA Marshall Space Flight Center (MSFC). The standard error values of SSM/I wind speeds with respect to the buoy winds are less than 2 m/s. The bias values do not exceed 0.2 m/s. The SSM/I measurements are also commonly used to estimate rain rate, latent and sensible heat fluxes. Several methods for estimating such parameters have been discussed in the literature (see for instance Liu, 1986; Miller and Katsaros, 1992; Schulz et al., 1993; Schlüssel et al., 1995; Bentamy et al., 2003). The calculation of latent and sensible heat fluxes from satellite measurements is mainly based on the use of bulk formulae (Eqs. (2) and (3)). It requires the knowledge of surface wind speed, the specific air and surface humidity, and the sea surface and air temperatures. 2.2.1 Specific Air Humidity Several authors have investigated the estimation of specific air humidity (qa ) from microwave radiometer measurements. Liu (1986), used 17 years of soundings from ship and ocean-island stations to show that qa (not necessarily at a 10-m height) is well correlated with the integrated water vapor content, W, which can be derived from SSM/I brightness temperatures. This method provides accurate values of global monthly-averaged qa but exhibits a systematic bias grater than 2 g kg–2 in the Tropics, as well as in the mid and high latitudes. To reduce this bias, Miller and

8


153

Katsaros (1992) derived regressions of the air-sea humidity difference as a function of W. Their model improves the estimation of instantaneous values but it is limited to the northwest Atlantic. Schulz et al. (1993) provided a model to estimate the SSM/I precipitable water of the lowest 500-m layer of the planetary boundary layer (bottom-layer-integrated water vapor WB instead of W). The calibration of the SSM/I WB is based on 542 globally distributed soundings derived from meteorological field experiments. In addition, they derived a linear relationship between WB and qa . Ataktürk and Katsaros (1998) applied the Schulz et al. (1993) model to individual estimations and found that it overestimated qa values in the subtropics. Schlüssel et al. (1995), using a larger dataset of soundings, determined a new version of the Schulz model. In this model, qa is derived directly from SSM/I brightness temperature measurements. Several of the inverse models relating the specific humidity of air and SSM/I brightness temperature measurements were investigated through comparison with observations of qa from ships. The model described by Schulz et al. (1993, 1997) provides better agreement with in situ qa estimates than previous models. However, comparisons performed by Bentamy et al. (2003) showed seasonal and regional biases between ship and satellite qa calculated using the Schulz model. In the North Atlantic, this bias was about –0.22 gkg–1 during the summer season, while in the winter and spring seasons it was about 0.7–0.8 gkg–1 . Comparisons between ship and ODAS buoy qa estimates did not show such biases. Therefore, to minimize these biases between satellite and in situ air specific humidity, a sample of 1,000 pairs of collocated SSM/I brightness temperatures and ship data was used to estimate new values for the coefficients in the Schulz model. The collocation is performed over the global oceans, using all available and validated satellite (F10, F11, F13, and F14) and ship data during the period October 1996-September 1997. The collocated ship qa data are divided into bins of 0.5 gkg–1 . From each qa class, 20 of the collocated ship/satellite data were randomly selected. The qa model coefficients were determined by minimizing the squared differences between observed qa (from ship) and estimated qa (from satellite). The new model and its coefficients are provided by the following equation: qa = a0 + a1 T19 V + a2 T19H + a3 T22 V + a4 T37 V

(22)

where a0 = −55.9227, a1 = 0.4035, a2 = −0.2944, a3 = 0.3511, and a4 = −0.2395. The remaining collocated ship/satellite data are used to compare in situ and remotely sensed qa estimates. As expected, the comparisons of the statistical parameters are improved using the new qa model. On average, the bias is reduced by 15% and is no longer statistically significant. The rms difference between satellite and ship qa estimates is now 1.40 instead of 1.70 gkg–1 . Over the North Atlantic Ocean (80% of ship data are located in this region), the maximum values of the difference bias between satellite and ship qa is about 0.25 g kg–1 and is found during the summer season, where qa values are high.

154

A. Bentamy et al.

2.2.2 Latent Heat Flux The remotely sensed surface wind speed and specific air humidity, described above, are used to estimate latent heat flux. The calculation of Qlatent is performed hourly, with a spatial resolution of 1◦ in latitude and 1◦ in longitude. This resolution is consistent with that of the Reynolds daily gridded maps used for SST retrieval. Prior to calculating Qlatent , all available data (winds, SSTs, and brightness temperatures), sampled within a 1◦ ×1◦ grid point of a satellite swath during a given hour, are averaged, and the two first statistical moments are computed. Over each grid point located within each SSM/I swath, the available U10 , Ts , T19 V , T19H , T22 V , and T37 V are used to estimate the instantaneous latent heat flux values through Eq. (2). In cases when the SSM/I wind speeds are not valid, scatterometer winds calculated over the same grid point and within a 3-h window are temporally interpolated to the time of the SSM/I observations. On average, the percentage of individual latent heat fluxes estimated with scatterometer wind speeds is about 15% for NSCAT and 9% for ERS-2. This number increases in tropical areas (10◦ S–10◦ N) to 19% for NSCAT and to 12% for ERS-2. Examples of latent heat flux estimations over satellite swaths are shown in Fig. 8.6. Several assumptions have been made for the calculations described above. The SST at a grid point is assumed constant over a day. The surface pressure P0 is assumed to be at a constant value of 1013.25 hPa. Air temperature at 10 m, T10 , is taken to be Ts −1.25 K. The impact of these assumptions on bulk latent heat flux estimation has been investigated with buoy measurements, which provide surface pressure, air temperature, and sea surface temperatures. The possible error (uncertainty) due to these assumptions is generally less than 2.5%.

Day 1

Day 2

Day 3

Fig. 8.6 Three days of latent heat flux (left column), specific surface and air humidity difference (middle), and surface wind speed (right column) estimated from satellite measurements

8


155

2.3 Altimeter Satellite altimeters routinely provide along-track measurements of surface wind speed (no direction) and significant wave height (SWH). Five altimeters which have various instrumental configurations are considered in this study: ERS; Topex/Poseidon; Jason; GFO; and Envisat. The use of remotely sensed wind and SWH in the future should potentially lead to more refined wind stress field analysis at global and regional scales. 2.3.1 Altimeter Swh Validation Although altimeter SWH is calibrated and validated during dedicated commissioning phase operations, after-launch long-term monitoring of the quality of the estimated geophysical parameters is needed (Queffeulou, 2003). Biases and trends are commonly observed on altimeter SWH measurements. For instance, biases of about 50 cm between TOPEX and ERS-1 and −20 cm between TOPEX and GEOSAT Follow-On (GFO) have been observed. A trend example is the TOPEX side-A SWH trend of about 40 cm between 1996 and beginning of 1999, which has been attributed to drift in the electronics. Biases are also observed on the two recent altimeters on board Jason and ENVISAT (Queffeulou, 2004). To correct for biases and trends, methods have been developed using buoy and cross altimeter data comparisons. The buoy data from the US NDBC, the Canadian MEDS, and the European networks were used in these comparisons. Details are given in (Queffeulou, 2003, 2004). Table 8.1, from (Queffeulou, 2004), gives proposed corrections to be applied to the altimeter SWH data. These corrections were established for the following altimeter data: ERS-2 Ocean PRoduct level 2 (OPR-2), TOPEX-Poseidon Merged Geophysical Data Record (M-GDR), GFO Intermediate Geophysical Data Record (IGDR), Jason Geophysical Data Record (GDR) and ENVISAT RA-2 Intermediate Marine Abridged Record (IMAR). Correcting the data greatly reduces the differences between the various satellite data sets. There are still some differences between SWH, at global scale, but these Table 8.1 Summary of the proposed linear corrections to altimeter SWH measurements (SWH_cor = a ∗ SWH + b). n = number of comparison data points Satellite

Reference

n

ERS-2 TOPEX-A1 TOPEX-B Poseidon GFO Jason ENVISAT

Buoys Buoys Buoys Buoys TOPEX GFO GFO

12070 2562 7826 752 15974 6332 1428

1 TOPEX

A

b

1.0642 1.0539 1.0237 0.9914 1.0625 1.0587 1.0526

0.0006 −0.0766 −0.0476 −0.0103 0.0754 −0.0571 −0.1991

side-A has to be further corrected as a function of cycle number, for cycle 98 to 235: swhcor = swh + poly3(98) − poly3(cycle) with poly3(x) = ai ×xi and a0 = 0.0864; a1 = −6.0426×10−4 ; a2 = −7.7894×10−6 ; a3 = 6.9624×10−8

156

A. Bentamy et al.

are reduced to about 10 cm, and might be attributed to the variability resulting from the different geographical samples of the various altimeters Note that SWH altimeter validations and corrections are regularly updated. Recent results can be found in Queffeulou and Croizé-Fillon (2010). 2.3.2 Validation Over the Western Mediterranean Sea The validations given in Sect. 2.3.1 were performed for the global ocean, using available buoy measurements. It could be reasonably suggested that regional validations are needed in order to take into account particular characteristics such as short fetch area, high wind variability and swell predominance. A study (Queffeulou et al., 2004) illustrates the particular SWH variability over the Western Mediterranean Sea. The TOPEX SWH measurements were compared to the data from four buoys. One of the buoys is in the Atlantic Ocean, west of Brittany (“Brittany”, 47.5◦ N 8.5◦ W); the three other buoys are located in the Western Mediterranean Sea: in the Gulf of Lion (“Lion”, 42.1◦ N 4.7◦ E), between the Italian coast and Corsica (“Corsica”, 43.4◦ N 7.8◦ E), and south of the Balearic Islands (“Mahon”, 39.72◦ N 4.44◦ E), respectively. The TOPEX Brittany SWH comparison shows general good agreement and low scatter. Data off the Gulf of Lion are also in good agreement, though the number of data points is only 22, over a SWH range limited to 4 m. The Corsica results show an underestimate of TOPEX SWH values above 1.5 m, and larger altimeter variability than in previous cases. The interpretation is not obvious: in this area the variability of wind speed and direction is high, and the buoy is located close to the coast, leading to unusual short fetch conditions, which could affect the accuracy of the altimeter algorithm. There are also only three comparison data above 2 m SWH, and the accuracy of the buoy measurement could also be involved. Analysis of the Mahon comparisons showed that the wave direction relative to the islands has to be taken into account for altimeter validation. For some directions of the wave field, the altimeter location can be modified by the presence of the island (sheltering, refraction) while at the buoy location, the wave field is less affected by the island. The particular examples discussed above illustrate the necessity for a careful analysis of the data over such closed seas and short fetch conditions. 2.3.3 Altimeter Wind Speed Validation For ERS, TOPEX and GFO, buoy wind speed comparisons were performed (Queffeulou, 2003) and linear corrections were proposed. Jason and ENVISAT RA-2 wind speed were validated using collocated data with buoy and GEOSAT FO. Jason wind speed is underestimated by about 1 m/s relative to buoy data, and by 1.2 m/s, relative to GFO. ENVISAT wind speed is also underestimated relative to both buoy and GFO measurements by about 0.7 m/s and 0.8 m/s, respectively. The relation between Jason and GFO wind speed is non-linear. The wind speed algorithms used are different: GFO uses the classical modified Chelton and Wentz

8


157

algorithm based on σ 0 wind speed dependence, while the Jason algorithm was developed from TOPEX and QuikScat data using both SWH and σ 0 as input. This last algorithm have been tuned to Jason data (Zieger et al., 2009). As for SWH, altimeter wind speed validations and data corrections are regularly updated (Queffeulou and Croize-Fillon, 2010).

3 Ocean Forcing Function 3.1 Remotely Sensed Flux Analysis Oceanographers are particularly interested in turbulent fluxes available at regular space and time intervals (i.e. gridded fields). The objective analysis of satellite wind and latent heat flux observations is based on the kriging method described by Bentamy et al. (1996). The method is applied to surface winds and latent heat flux fields separately. The aim is to calculate global daily, weekly, and monthly averaged flux parameters on a spatial grid of 0.5◦ ×0.5◦ or 1◦ ×1◦ (latitude×longitude) resolution. The interpolation scheme uses a spatial and temporal structure function describing the behavior of the variables. The algorithm provided by Bentamy et al. (2002) is used to calculate gridded wind fields. The structure function for latent heat flux is determined using spatial and temporal correlation scales calculated from satellite observations that are about 1510 km and 65 h, respectively. These parameters are then used to evaluate the weights of the satellite observations required to estimate the weekly value, depending on their spatial and temporal position relative to the grid point under analysis. As can be expected, the number of these observations is a function of latitude. On average, more than 336 observations are used at a grid point. The lowest numbers are found in the western part of the tropical Pacific Ocean (about 120 observations). The numbers of day and night observations are about the same. Figure 8.7 shows examples of weekly latent heat flux and wind speed fields over the tropical Atlantic Ocean. During the period 4–24 November 1996, the trade winds in both the North and South Atlantic reach mean weekly-averaged speeds of 8–10 m/s with the associated latent heat fluxes at about 200 W/m2 . Consistently higher in the northeasterly trade wind region than in the southeast trades, all three weeks illustrate the coherence between the wind and latent heat flux patterns.

3.2 Accuracy of Surface Wind Analysis The investigation of the accuracy of gridded surface parameters estimated from satellite data is only illustrated with the accuracy results related to surface wind fields. As for satellite observations, the accuracy of the gridded satellite flux analysis is determined through comparisons with buoy and numerical atmospheric estimates. For instance the comparisons between buoy and scatterometer averaged winds use

158

A. Bentamy et al.

Fig. 8.7 Sequence of three weeks in November 1996 showing pairs of weekly averaged maps of surface wind speed and latent heat flux. The week of November 11–17 (upper right) shows the effects of a high wind event blowing from the Atlantic Ocean towards the Gulf of Mexico with intensely enhanced latent heat flux (from Katsaros et al., 2003)

the following standard statistical data analysis: The wind speed, zonal wind component, and meridional wind component are assumed to be random variables which could be characterized by their moments. For this purpose, the two conventional moments of each variable are estimated. Moreover, some statistical parameters are calculated to assess the comparisons between satellite gridded wind fields and buoy averaged winds.

8


159

3.2.1 Global Comparisons Tables 8.2–8.4 provide summary statistics of wind speed comparisons. The wind speed correlation coefficients are significantly high and range from 0.85 to 0.89. The rms values of the buoy-satellite differences do not exceed 1.16 m/s over the NDBC and TAO networks, but are higher for ODAS comparisons: 1.48 m/s for NSCAT, and 1.66 m/s for ERS-2. This is mainly due to a smaller number of comparison data points and to high wind variability in the ODAS area. Furthermore, the statistics Table 8.2 Comparison of averaged weekly wind speed and direction estimated from NDBC buoy measurements and from ERS-1, ERS-2 and NSCAT scatterometer observations. Bias, root mean square (Rms), correlation coefficient (ρ).. and the standard deviation characterizing the difference between buoy and scatterometer averaged wind speeds and directions are provided Wind speed (m/s)

Wind direction

Data SET

BuoyWind speed range (m/s)

Length

Bias (m/s)

Rms (m/s)

ρ

Bias (deg)

Std (deg)

NDBC/ERS-1

0−24 0−5 5−10 >10 0−24 0−5 5−10 >10 0−24 0−5 5−10 >10

3281 320 2603 358 1921 142 1581 198 522 28 444 50

0.02 −0.14 0.05 −0.0 0.35 0.06 0.37 0.40 −0.37 −0.54 −0.37 −0.32

1.16 1.03 1.16 1.31 1.15 0.82 1.16 1.26 1.02 0.94 1.01 1.15

0.86 0.74 0.83 0.76 0.86 0.75 0.83 0.77 0.88 0.76 0.85 0.79

3 5 3 3 6 0 6 6 8 3 8 7

35 47 34 30 33 47 33 25 25 29 26 15

NDBC/ERS-2

NDBC/NSCAT

Table 8.3 Comparison of averaged weekly wind speed and direction estimated from TAO buoy measurements and from ERS-1, ERS-2 and NSCAT scatterometer observations Wind speed (m/s)

Wind direction

Data SET


Length

Bias (m/s)

Rms (m/s)

ρ

Bias (deg)

Std (deg)

TAO/ERS-1

0−24 0−5 5−10 >10 0−24 0−5 5−10 >10 0−24 0−5 5−10 >10

10047 3262 6693 92 6737 1925 4736 76 1780 515 1246 19

0.29 −0.14 0.47 0.24 0.56 0.06 0.75 0.76 −0.26 −0.70 −0.08 0.03

0.89 0.85 0.91 0.92 1.03 0.84 1.10 1.14 0.92 1.18 0.79 0.82

0.86 0.76 0.84 0.70 0.86 0.75 0.85 0.78 0.92 0.74 0.83 0.78

3 1 5 8 3 4 5 7 5 2 7 10

31 51 17 9 27 45 16 10 20 33 11 5

TAO/ERS-2

TAO/NSCAT

160

A. Bentamy et al.

Table 8.4 Comparison of averaged weekly wind speed and direction estimated from ODAS buoy measurements and from ERS-2 and NSCAT scatterometer observations Wind speed (m/s)

Wind direction

Data SET


Length

Bias (m/s)

Rms (m/s)

ρ

Bias (deg)

Std (deg)

ODAS/ERS-2

0−24 0−5 5−10 >10 0−24 0−5 5−10 >10

222 10 155 57 194 6 118 70

−0.73 −1.26 −0.61 −0.83 −0.65 −1.29 −0.62 −0.57

1.69 2.01 1.68 1.50 1.52 2.07 1.44 1.47

0.84 0.72 0.80 0.80 0.89 0.72 0.81 0.86

1 31 3 4 2 14 1 9

38 75 39 22 30 76 30 22

ODAS/NSCAT

calculated by several meteorological centers (ECMWF, CMM, UKMet) indicate that ODAS buoy wind speeds tend to be underestimated according to meteorological wind analysis (see ftp://ftp.shom.fr/meteo/qc-stats, site maintained by P. Blouch). The statistical parameters are also calculated in bins of 5 m/s of the buoy wind speed. Their values show small dependence on the NDBC and TAO wind speed. The bias is slightly positive for ERS and negative for NSCAT in all the wind speed ranges. The analysis carried out on collocated data, shows that the slopes calculated over each buoy network and against buoy wind estimates, are similar regardless of which of the three scatterometer wind products is used for comparison. For NDBC (Table 8.2), buoys and scatterometers correlate closely, as expressed by slopes (b and bs) of about 1 and intercepts of about zero. For TAO in the tropical Pacific Ocean, slopes are about 0.90, suggesting an overestimation of low wind speed and an underestimation of high wind speed by scatterometer wind fields compared to TAO winds. In the North Atlantic area, the slopes of the scatter plots are close to 1, whereas the intercepts are about 0.5, indicating that the scatterometer wind fields are consistently high compared to ODAS weekly averaged wind speeds. The calculation of statistical parameters for the ODAS buoy wind speed ranges shows that their values are made variable by the outlying points at low and high wind speeds. No systematic wind direction bias is found, and the overall bias and standard deviation in terms of the mean angular difference are less than 8◦ and 38◦ , respectively. These results are consistent with the calibration/validation of scatterometers against buoys (Graber et al., 1996; Caruso et al., 1999). For instance, in the Pacific tropical area, where the wind direction is quite steady, the standard deviation of wind direction calculated for buoy wind speed higher than 5 m/s does not exceed 17◦ . 3.2.2 Time Series The agreement between averaged wind fields from scatterometers and buoys can be studied using time series. Figure 8.8 shows examples of weekly averaged time series of wind speed at three buoy locations in the NDBC and TAO arrays, respectively.

8


161

Fig. 8.8 Time series of buoy (green), ERS-1 (red), ERS-2 (blue), NSCAT (cyan) wind speed at three buoy locations

162

A. Bentamy et al.

They indicate that the matchups are strongly correlated, and their geographical features compare well. The lowest correlation values (less than 0.91) are found in the TAO array. At the 95◦ W–2◦ N TAO (Fig. 8.8c), the difference is consistent and the bias is about 1 m/s. This may be related to the south equatorial current effect on scatterometer backscatter coefficient measurements (Quilfen et al., 2001). Indeed, the buoy samples the absolute wind, whereas the scatterometer samples the relative wind. The highest discrepancy between TAO and scatterometer winds (bias greater than 1.5 m/s) occurred between May and December 1998. During this period, several scatterometer retrieval winds are not valid (especially during May and June 1998), and the TAO buoy moored at this location reported high variable winds of about 7 m/s, exceeding climatology by 1 m/s. The standard deviation of weekly averaged buoy wind speed varies between 0.9 m/s and 1.9 m/s (72% of standard deviation values are great than 1.2 m/s). Furthermore, the analysis of oceanic current measured at 110◦ W, 2◦ N indicate that its magnitude is about 50 cm/s from May through December 1998, while for 1992 until 1997 and during the same months, the current magnitude is on average 30 cm/s. The comparisons between NDBC and scatterometer averaged wind speed time series do not exhibit any systematic bias (an example is shown in Fig. 8.8a). At some locations a seasonal variation in the differences is found. The bias tends to be positive in winter and negative in summer. This may be related to the dependence of wind speed residuals on buoy wind speed ranges illustrated by the results of Table 8.2. For ODAS (not shown), scatterometer averaged wind speeds are consistently higher than buoy estimates. However, the bias tends to be large between October and December 1996, when the correlation coefficient is about 0.69. The latter is lower than for the whole period. by a factor of 22%.Some discrepancies between buoys and scatterometers are related to the sampling errors of scatterometer wind fields. For instance, between July and August 1996, the ERS-2 error exceeds 2 m/s due to the relatively small number of scatterometer observations available to estimate the gridded fields. Finally, the dependence of the residuals on the buoy latitude is investigated. More than 80% of the latitudinal differences are less than 0.5 m/s. Between 8◦ S and 2◦ N latitudes (TAO array), the bias (buoy minus scatterometer) is positive and continuous with increasing latitude. This dependency is consistent with results shown above and might be due to current and sea state. From 5◦ N to 45◦ N, a slightly decreasing bias is exhibited. At high latitudes, where the wind is highly variable, scatterometer weekly wind speeds tend to be overestimated against buoy estimates. This is mainly related to the methods used to average wind data from scatterometers and buoys, and to the sampling scheme. The analysis of the rms behavior with latitude confirms the latter results. Indeed, most of the values of the rms difference between buoys and scatterometers are below 1.2 m/s, except at latitudes above 45◦ N. To examine the agreement between weekly averaged scatterometer and buoy winds as a function of buoy latitude, the correlation coefficient for each latitude is calculated. The main results of these statistical parameters are higher than 0.8 for all latitudes and the differences between them are not significant at the 95% confidence level.

8


163

3.2.3 Scatterometer/Ecmwf Averaged Wind Comparisons In this section, the new mean weekly and monthly scatterometer wind fields are compared to the ECMWF operational surface wind analyses. Like several National Weather Prediction (NWP) systems, ECMWF is a very complex analysis system which is continually being improved. It assimilates measurements from a variety of sources: satellites, buoys, and ships. It is important to notice that ECMWF products are not used as a “ground truth” for surface winds. However, they represent the main known wind features at various scales and in all oceanic basins. Their use allows the investigation of scatterometer wind field patterns over a given ocean basin and/or time period. Furthermore, as the scatterometer data are uniformly processed, they can be used to evaluate the impact of the numerous changes that have occurred in the ECMWF forecast-analysis system. The mean weekly and monthly ECMWF wind speed, zonal component and meridional component are computed from the 6-hourly global analysis datasets on 1◦ .125×1◦ .125 grid. The scatterometer sea ice mask is used to avoid ice. The comparisons are performed over the global ocean for all December and June months of the ERS-1, ERS-2 and NSCAT periods. Only ECMWF wind speeds above 3 m/s and estimated over oceanic regions are used. Statistics of the comparisons are summarized in Table 8.5. The correlations for wind speed, zonal wind component, meridional wind component, and wind direction are high and exceed 0.89. For wind direction, the biases are small, while the rms values are 28◦ for ERS-1, 26◦ for ERS-2, and 17◦ for NSCAT. Even if the wind speed biases are rather low, ERS-1 and NSCAT are biased high by about 0.50 m/s compared to ECMWF, while the corresponding rms values are 1.40 m/s for ERS-1 and 1.03 m/s for NSCAT. The number of high wind condition events derived from ERS-1 and NSCAT is high with respect to ECMWF. More than 6.5% of ERS-1 and NSCAT wind speed estimates exceed 15 m/s and this percentage drops to 4.5% for ECMWF. Comparisons between ECMWF and ERS-2 provide the lowest bias and rms values of 0.04 m/s and 0.96 m/s, respectively. Most of significant discrepancies between ECMWF and scatterometers are located at high latitudes in both hemispheres poleward of 60◦ . However, in some cases of low correlations are found in middle latitudes. For instance, the correlation coefficient, calculated in the South Atlantic region between 35◦ S and 45◦ S for the period December 7–13, 1992 is 0.42. For this week and

Table 8.5 Comparison of averaged weekly wind speed and direction estimated from ECMWF wind analysis and from ERS-1, ERS-2 and NSCAT scatterometer observations Wind speed (m/s)

Wind direction

Data SET

Bias (m/s)

Rms (m/s)

ρ

Bias (deg)

Std (deg)

ECMWF/ERS-1 ECMWF/ERS-2 ECMWF/NSCAT

−0.39 0.04 −0.57

1.42 0.96 1.03

0.89 0.94 0.92

1 0 5

28 26 17

164

A. Bentamy et al.

region, the kriging error measuring the quality of weekly averaged winds does not exceed 1 m/s. The annual mean profiles, estimated as longititudinal averages of the scatterometer and ECMWF winds in 1◦ latitude bands, indicate that scatterometer and ECMWF wind features compare well. The highest wind values are found in the 50◦ S–60◦ S and 50◦ N–60◦ N bands. Lowest winds occur within equatorial regions. For instance, at 53◦ S, scatterometer and ECMWF provide wind speed averages of about 9.5 m/s, while at 0◦ the annual mean wind speed is about 5 m/s. The highest differences exceeding 0.5 m/s are found in the 55◦ N–65◦ N belt. However, such a calculation indicates that scatterometer wind speeds are greater than ECMWF estimates almost everywhere. Figure 8.9 displays examples of latitudinal weekly scatterometer and ECMWF wind speed comparisons. The time series are calculated from 1◦ ×1◦ gridded fields integrated over three 20◦ latitude bands over the Atlantic Ocean. They show that the correlation is high and roughly constant over the whole period. Scatterometer and ECMWF winds exhibit similar wind features. In particular, the examples do not show any disturbing oscillations in scatterometer winds. Furthermore, such calculations confirm that the ERS-1 scatterometer records higher winds than ECMWF. The maximum differences between ERS-1 and ECMWF winds occurred between December 9, 1991 and February 24, 1992, corresponding to many missing data in scatterometer observations due to the ERS-1 scatterometer calibration/validation process. However, the calculation of the relative differences ((Wecmwf – Wscat ) / (Wecmwf +Wscat )/2) indicate that on average their values in equatorial regions decrease from 12% to 2% between March 1992 and September 1994, while in high latitudes these values are nearly steady and are about 5%. For ERS-2, the differences between ECMWF and scatterometer winds are the lowest. ERS2 scatterometer measurements have been assimilated within the ECMWF analysis scheme since April 1996. Except in the Atlantic sector of the Southern Ocean, average weekly winds estimated from NSCAT observations are higher than ECMWF wind estimates. The variability of the difference between ECMWF and scatterometer weekly wind fields is investigated in terms of rms differences (Figures not shown). Excluding periods when there are many missing scatterometer data, the average rms difference in wind speed is less than 1.5 m/s in the middle and tropical latitudes. In high latitudes and due to high wind variability, the rms difference values are high and about 2 m/s. Similar geographical features are found in terms of wind components. As expected, the rms difference between ECMWF and ERS2 is 0.5 m/s lower than the rms difference between ECMWF and ERS-1. The analysis of the rms difference patterns according to time indicates that there is a decreasing trend mainly related to ECMWF model changes (ECMWF, 1993). Furthermore, the rms features are highly correlated to seasonal wind variability. For instance, in high northern latitudes the rms differences are lower between April and September with a mean value of about 0.8 m/s for wind speed. The behavior of the rms differences between ECMWF and NSCAT weekly wind speed and components is found to be quite comparable to that estimated from ECMWF and ERS-2 differences.

8


165

Fig. 8.9 Time series of averaged ECMWF (green), ERS-1 (red), ERS-2 (blue), NSCAT (cyan) wind speed estimated over three oceanic areas

166

A. Bentamy et al.

4 Summary A brief review of the methods for extracting surface winds from scatterometers, altimeters, and radiometers has been given. The allowance for wind conditions, sea state, and atmospheric effects has been discussed and empirical corrections have been outlined. The surface wind retrievals are used to enhance the determination of turbulent flux components such as wind stress and latent heat flux. In this study, only scatterometer and SSM/I winds in combination with the specific air humidity retrieved from the radiometer brightness temperatures are used for estimating surface fluxes. They allow the determination of accurate weekly and monthly turbulent flux field over global ocean. In future studies, retrievals from altimeters and from ASCAT scatterometer will also be considered to improve the spatial and temporal resolutions as well the quality of the forcing function components. Weekly and monthly flux data including, wind speed, zonal and meridional components, wind stress and the associated components, latent and sensible heat fluxes, are freely available at the following addresses: • ERS-1/2 L2b winds: http://cersat.ifremer.fr/fr/data/discovery/by_parameter/ocean_wind/ers_wnf • ERS-1/2 L4b wind products: http://cersat.ifremer.fr/fr/data/discovery/by_parameter/ocean_wind • NSCAT and QuikSCAT L2b wind products http://podaac.jpl.nasa.gov/DATA_PRODUCT/OVW • NSCAT L4b wind products: http://cersat.ifremer.fr/fr/data/discovery/by_parameter/ocean_wind/mwf_nscat • QuikSCAT L4b wind products: http://cersat.ifremer.fr/fr/data/discovery/by_parameter/ocean_wind/mwf_ quikscat • Satellite turbulent fluxes: ftp://ftp.ifremer.fr/ifremer/cersat/products/gridded/flux-merged/flux/data/

References Ataktürk SS, Katsaros KB (1998) Estimates of surface humidity and wind speed obtained from satellite data in the stratocumulus regime in the Azores region. In: Brown RA (ed) Remote sensing of the pacific ocean by satellites. Southwood Press, Marrickville NSW, Australia, 16–22. Bentamy A, Quilfen Y, Gohin F, Grima N, Lenaour M, Servain J (1996) Determination and validation of average field from ERS-1 scatterometer measurements. Global Atmos Ocean Sys 4:1–29. Bentamy A, Queffeulou P, Quilfen Y, Katsaros K (1999) Ocean surface wind fields estimated from satellite active and passive microwave instruments, IEEE Trans Geos Remote Sens 37(5): 2469–2486. Bentamy A, Quilfen Y, Flament P (2002) Scatteromter wind fields: a new release over the decade 1991–2001. CJRS 28(3):424–430.

8


167

Caruso M, Dickinson S, Kelly K, Spillane M, Mangum L, McPhaden M, Stratton L (1999) Evaluation of NSCAT scatterometer winds using Equatorial Pacific buoy observations. Technical Report, Applied Physics Laboratory, University of Washington, Seattle, WA, USA, 60pp. ECMWF (1993) http://www.ecmwf.int/products/data/operational_system/evolution/evolution_ 1993.html Freilich MH, Dunbar S (1999) The accuracy of the NSCAT 1 vector winds: comparisons with National Data Buoy Center buoys. J Geophys Res 104:11231–11246. Goodberlet MA, Swift CT, Wilkerson JC (1989) Remote sensing of ocean surface winds with the Special Sensor Microwave/Imager. J Geophys Res 94:14547–14555. Graber HC, Ebutchi N, Vakkayil R (1996) Evaluation of ERS-1 scatterometer winds with wind and wave ocean buoy observations. Tech. Report, RSMAS 96-003, Division of Applied Marine Physics, RSMAS, University of Miami, FL, USA, 58 pp. Hosking JRM (1990) L-moments: analysis and estimation od distributions using linear combinations of order statistics. J R Statist Soc. Ser B 52:105–124. Jones WL, Wentz FJ, Schroeder L (1978) Algorithm for inferring wind stress from Seasat-A. J Spacecr Rockets 15:368–374. JPL (2001) QuikScat science data product users manual (version 2.0). Jet Propulsion Laboratory Publication, Pasadena, CA, 84pp (Available online at http://podaac.jpl.nasa.gov/quikscat) Katsaros KB, Mestas-Nuñez AM, Bentamy A, Forde EB (2003) Wind bursts and enhanced evaporation in th tropical and subtropical Atlantic Ocean. In: Goni G, Malanotte-Rizzoli P (eds) Interhemispheric water exchange in the atlantic ocean. Elsevier Oceanographic Series, 463–474. Liu WT, Katsaros K, Businger JA (1979) Bulk parametrization of air-sea exchanges of heat and water vapor including the molecular constraints at the interface. J Atmos Sci 36:1722–1735. Liu WT (1986) Statistical relation between monthly precipitable water and surface-level humidity over global oceans. Mon Wea Rev 114:1591–1602. Miller DK, Katsaros K (1992) Satellite derived surface latent heat fluxes in rapidly intensifying mariner cyclone. Mon Wea Rev 120:1093–1107. Queffeulou P (2003) Cross-validation of ENVISAT RA-2 significant wave height, sigma0, and wind speed. IFREMER Final report, May. Queffeulou P, Bentamy A, Guyader J (2004) Satellite wave height validation over the Mediterranean Sea, Proceedings of the ENVISAT & ERS symposium, Salzburg, Austria, 6–10 September 2004. Queffeulou P (2004) Long-term validation of wave height measurements from altimeters. Mar Geodes 27:495–510. Queffeulou P, Croizé-Fillon D (2010) Global altimeter SWH data set, version 7, May, ftp://ftp.ifremer.fr/ifremer/cersat/products/swath/altimeters/waves/ Quilfen Y (1995) ERS-1 off-line wind scatterometer products. IFREMER Tech Rep 75 pp. Quilfen Y, Chapron B, Vandemark D (2001) The ERS Scatterometer Wind Measurement Accuracy: Evidence of Seasonal and Regional Biases. J Atmos Ocean Technol 18(10):1684–1697. Schlüssel P, Schanz L, English G (1995). Retrieval of latent heat flux and long wave irradiance at the sea surface from SSM/I and AVHRR measurements. Adv Space Res 16:107–115. Schulz J, Schlüssel P, Grassl H (1993) Water vapor in the atmospheric boundary layer over oceans from SSM/I measurements. Int J Remote Sens 14:2773–2789. Schulz J, Meywerk J, Ewald S, Schlüssel P (1997) Evaluation of satellite-derived latent heat fluxes. J Clim 10:2782–2795. Sobieski P, Craeye C, Bliven L (1999) Scatterometric signatures of multivariate drop impacts on fresh and salt water surfaces. Int J Remote Sens 20(11):2149–2166. July 1999. Stoffelen A, Anderson D (1997) Scatterometer data interpretation: Estimation and validation of the transfer function CMOD4. J Geophys Res 102:5767–5780.

168

A. Bentamy et al.

Thiria S, Mejia C, Badran F, Crepon M (1993) A neural network approach for modeling nonlinear transfer functions: Application for wind retrieval from spaceborne scatterometer data. J Geophys Res 98:22827–22841. Yamartino RJ (1984) A comparison of several “single-pass” estimators of the standard deviation of wind direction. J Clim Appl Met 23:1362–1366. Zieger S, Vinoth J, Young IR (2009) Joint calibration of multiplatform altimeter measurements of wind speed and wave height over the past 20 years. J Atmos Ocean Technol 26(12).

Chapter 9

Remote Sensing of Oil Films in the Context of Global Changes Andrei Yu. Ivanov

Abstract Since the middle of the 20th century, during the global industrialization, crude oil has become a major source of energy, and the problem of oil spills in the sea has become a public concern. Spilled oil interacts with marine environments, causing damage to marine ecosystems and influencing the ocean-atmosphere interaction. The impact of accidental spills is catastrophic for coastal zones, and has far-reaching consequences, not realized or anticipated previously. Sea surface oil films play a significant role in important climate processes such as the exchange of momentum, heat and gas between the ocean and the atmosphere. However, nowadays there is no estimating method for determining how fully and how frequently the ocean is covered by oil films, even when slicks are imaged from space. Remote sensing is believed to offer an effective mean for accounting of the global impact of oil films, which could lead to an improvement of representing air-sea interactions in climate models. Among remote sensing techniques, spaceborne Synthetic Aperture Radar (SAR) is the primary remote sensor used for oil spill surveillance. Oil films floating on the sea surface are detectable by SARs, because they damp the short surface waves that are responsible for the radar backscattering. The European and Asian seas have been chosen as test basins to work out an oil spill mapping technology using SAR images, and these studies demonstrated the high potential of SAR-equipped satellites. An approach for oil spill mapping has been also developed. It is based on studies conducted within the analysis framework of a geographic information system (GIS). GIS-made oil spill maps can be a valuable source of information about oil spill distribution, statistics and sources. They allow scientists to identify the most polluted waters – an important step in development of monitoring scenario. Finally, a basis for the determination of volumes and extent of floating oil based on satellite remote sensing is provided. This paper reviews and discusses this problem in the context of global changes.

A.Y. Ivanov (B) P.P. Shirshov Institute of Oceanology, Russian Academy of Sciences, Nakhimovsky prospect, 36, Moscow, 117997, Russian Federation e-mail: [email protected] D. Tang (ed.), Remote Sensing of the Changing Oceans, C Springer-Verlag Berlin Heidelberg 2011 DOI 10.1007/978-3-642-16541-2_9,

169

170

A.Y. Ivanov

Keywords Climate changes · Air-sea interface · Oil pollution · Oil spills · Biogenic slicks · Remote sensing · SAR images · GIS approach · Oil spill distribution maps

1 Introduction In the last several decades, significant advances have been made in studying air-sea fluxes over the ocean (Toba, 2003; Garbe et al., 2007). The study of the boundary between the atmosphere and the ocean is very important. This air-sea interface, supporting marine life, is one of the most physically, chemically and biologically active interfaces, and plays a major role in the exchange processes of gases, materials, and energy (Kraus, 1972). Many scientists consider it a “mirror” of climate changes. A considerable amount of new research over the last decade has led to wider understanding of the vital importance of the interface between the ocean and the atmosphere, and how it may interact with the processes involved in global changes (Liss and Duce, 1997; Toba, 2003; Garbe et al., 2007). However, predictions of future climate changes based on existing knowledge vary greatly, and detailed forecasts are still subject to debate (Toba, 2003). One key uncertainty is caused by the lack of accurate knowledge of transport processes in the air-water interface, especially in the presence of oil films, both natural and man-made, which pose the main resistance to transfer processes between the ocean and the atmosphere. Modeling and predictions of global climate can only be improved by gaining a more complete understanding of the transporting mechanisms across the air-sea boundary. However, these mechanisms and the interactions involved are very complex and still not fully understood. Oil in the sea, or “oil spills”, are the releases of liquid petroleum hydrocarbons into the marine environment due to human activity, and are a form of oil pollution. In the ocean, oil is frequently released into coastal waters. The oil and petroleum products consist of different mixtures of liquid materials, including crude oil and/or refined petroleum products. Oil is also released into the marine environment from natural geological sources, known as oil seeps (or seepage) on the seafloor. In the sea, crude oil and most oil products are subjected to weathering processes (Fig. 9.1), which include spreading, drifting, dispersion, evaporation of lighter fractions, chemical transformation, emulsification, photo- and bio-oxidation, etc. Because oil floats on the sea surface, it affects the marine life and limits the lifecycle of plankton organisms. This affects the food chains in the marine ecosystems and eventually the fauna population (Oil in the sea, 2003). Oil films covering the ocean surface from time to time can now be detected and resolved at sufficient temporal and spatial resolution, because they smooth the sea surface roughness (Fig. 9.2). Satellite-derived oil spill products have the appropriate resolution and can be used instead of conjecture and hypothesized “products” that often do not appropriately resolve oil films variability. In the last several decades, synthetic aperture radar (SAR) images were extensively used to obtain statistical and quantitative information about oil pollution

9

Remote Sensing of Oil Films in the Context of Global Changes

171

Advection by sea current

Fig. 9.1 Oil in the sea: main oil transportation and weathering processes. © SINTEF

Smooth Surface: very low wind (< 2.5 m/s)

Slightly Rough Surface: Frequent wind conditions

Very Rough Surface: strong wind (> 13 m/s)

incident radiation diffuse scattering

reflected radiation

Oil spill smoothing water surface

Oil spill cannot be detected

Good condition to detect oil spills

Fig. 9.2 Sea surface conditions and detectability of oil spills Source: © ESA/Eduspace

Oil spill mixed in the water

172

A.Y. Ivanov

(Pavlakis et al., 1996; Gade and Alpers, 1999; Lu et al., 1999; Pavlakis et al., 2001; Ivanov and Zatyagalova, 2008; Shi et al., 2008). In response to the challenges by oil spill in the sea in the 20–21st centuries, a number of projects for utilization of SAR data of different swaths and resolution for oil spill mapping were started since, 1991 (Pavlakis et al., 1996). Many cases of oil spill pollution have been documented and a number of regional monitoring campaigns have been carried out (see, e.g., Bern et al., 1992; Pavlakis et al., 1996; Espedal et al., 1998; Gade and Alpers, 1999; Lu et al., 1999; Pavlakis et al., 2001; Ivanov and Zatyagalova, 2008). Many years of practical research have shown that the SAR is one of the best satellite sensors for oil spill monitoring. Consequently, wide-swath SARs aboard the SAR-equipped satellites have become important tools for control and monitoring of marine environments due to their all-weather and day-and-night capability. They provide high resolution, wide swath and relatively short revisit time coverage, and are the most suitable tools not only for regular monitoring, but also for mapping of oil spills in the sea. Evidently, oil pollution of the sea is major environmental problem. The first approach to the oil spill mapping problem was developed by Pavlakis et al. (1996), who showed usefulness of a large set of SAR images acquired over the Mediterranean Sea. Gade and Alpers (1999) analyzed more than 400 ERS-2 SAR images covering the North Sea, Baltic Sea and the Gulf of Lions. They concluded that these waters were polluted mainly by shipping. Lu et al. (1999), for the first time compiled a statistical spatial distribution map of oil pollution for the Southeast Asian waters after analyzing 2,500 middle resolution ERS-1/ERS-2 SAR images; they also concluded that the waters were most polluted along main shipping lanes. Finally, it became clear that SAR images are very useful not only for locating the areas where tankers are washed and ships are discharged engine room waters, but for collecting information about spatial distribution of oil films on meso-, and even global scale.

2 Marine Oil Films and Their Role in Ocean-Atmosphere Interaction Natural and man-made oil slicks affect the physical properties of the air-sea interface, and, in turn, have impact on a number of important processes determining the ocean-atmosphere interaction. Among those are: energy transfer from wind to waves, retardation of evaporation and convective exchange, sea surface temperature (SST) variability, gas exchange, formation of skin-layer and foam on the sea surface and others (Kraus, 1972; Monin and Krasitskii, 1985). Existing knowledge about the effects of oil films on the marine environment, ocean-atmosphere interaction and global warming, in the current circumstances of global changes, is insufficient (Monin and Krasitskii, 1985). These effects are studied only partially and locally. Their global impact and their role in global changes are not well understood. Their regional and global estimates are very approximate, because the real extent and

9


173

distribution of oil slicks as well as oil spills in the world’s ocean is mostly unknown. These processes have to be re-studied and re-estimated at regional and global scales. The sea-water surface is always contaminated by oily organic materials of different sorts (Monin and Krasitskii, 1985). Extremely thin layers of natural organic films have thickness comparable to the effective diameter of a single molecule and are formed by surface-active materials, such as some fatty acids and alcohols (Levich, 1962). These materials are made up of molecules with both hydrophilic and hydrophobic parts, and they will spread on the water surface and tend to form a layer one molecule in thickness. These layers continuously change the physical-chemical properties of the air-sea interface (Levich, 1962). Such a monomolecular layer of some oily substances can prevent the escape of the water molecules into the atmosphere vapor. For example, for a small tank, depending on weather conditions, this can reduce the evaporation rate by more than 50% (Kraus, 1972). But for large water regions this effect is limited by the difficulty of maintaining of unbroken oil layers on large sea surfaces. At the sea surface, in particular at moderate wind conditions, organic films, although never absent from large areas, are always patchy or broken up by currents and surface waves (Monin and Krasitskii, 1985). Their effect on evaporation is also poorly understood (Monin and Krasitskii, 1985). Moreover they may have some effect on SST fluctuations and hence on convection in the low atmosphere and mixing in the upper ocean layer (Monin and Krasitskii, 1985). They also can influence the generation of waves on the sea surface (Alpers and Hühnerfuss, 1989). Organic and oil films change the local gas exchange rates through the water interface (Monin and Krasitskii, 1985). “Oil” is a general term used to denote petroleum products which mainly consist of hydrocarbons. “Oil spills” is a term that often refers to marine oil spills, when oil is released into the open ocean or coastal waters, due to natural or human activity (ITOPF, 2009). Oil spills consist of liquid petroleum hydrocarbons, and are a form of oil pollution. Ship accidents and operational discharges, river run-off, natural seeps, offshore production, and atmospheric fallout are considerably large sources of oil pollution (Oil in the sea, 2003). It is well-known that oil floating on the sea surface originates from different sources. Clemente-Colón and Yan (2000) divided these oil substances into two major categories: natural and man-made (Fig. 9.3). Each category then can be further subdivided into those of biogenic origin and those of mineral origin. Oil films and slicks on the sea surface have been extensively studied theoretically and experimentally starting from the beginning of 60 s (see, e.g., Levich, 1962; Monin and Krasitskii, 1985; Scott, 1986; Alpers and Hühnerfuss, 1989; Ermakov et al., 1992; Gade et al., 1998; Espedal et al., 1998).

Man-made oil slicks are typically caused by accidental spills (collisions and groundings) or by illegal dumping of oil and petroleum products. Most man-made oil pollution comes from different human activities (Fig. 9.4–9.7), but public attention is focused mostly on oil tankers (ITOPF, 2009). Fortunately, there were only 21 major tanker crashes since 1967 with oil spills of over 30,000 tons (ITOPF, 2009). Such spills usually take months or even years to clean up. There is a decreasing number of large spills, and detailed analysis is available for each of them (ITOPF, 2009). From the ITOPF’s statistics it is evident that most spills result from routine

174

A.Y. Ivanov

Oil in the sea

Natural

Biogenic

Man-made

Mineral

Surfactants

Biogenic

Seepage oils

Mineral

Fish and vegetable oily products

Crude oil and oil products

Fig. 9.3 Simple classification of oil products forming oil slicks on the sea surface

Fig. 9.4 The COSMO-SkyMed SAR image showing oil spill after shipwreck in the Yellow Sea (September 2008). © ASI/Italian MoD

operations such as discharging, loading and bunkering that normally occur at shipping lanes, in roads, ports, or at oil terminals. The majority of these operational spills are small (700 tons) has decreased during the last 38 years (ITOPF, 2009). The mineral oil spills may be of a variety of materials, including crude oil, refined petroleum products (light products: petrol, kerosene, diesel fuel, local fuel blends;

9


175

Sea of Azov

3

3

Ukraine 4 Kerch

7

9

4

4 Kerch Bay

7 2 Chushka Split

5

2 1

Taman Bay Isl. Tuzla

6

Tuzla Dam 1

x Russia

Black Sea

8

Fig. 9.5 Oil pollution in the Kerch Strait after the Volgoneft-139 tanker shipwreck (about 1,300 tonnes of fuel oil released) as seen on the TerraSAR-X image (16/11/2007, res. 3 m, pol. VV): h – location of the bow of the tanker; 1 – oil trail from the bow penetrating into the Kerch Strait; 2 – oil accumulated in the west part of Taman Bay further propagating into the strait; 3 – oil patches in the Sea of Azov on the exit from the Kerch Strait; 4 – oil pollution along the Ukrainian coast of the strait; 5 – oil pollution accumulated in Kerch Bay; 6 – weathered fuel oil possibly processed by sorbents; 7 – oil patches at the west shoreline of Chushka Split; 8 – ship-made oil spill; 9 – fairway (figure from Ivanov, 2010). © InfoTerra

176

A.Y. Ivanov

Fig. 9.6 Chronic oil pollution at the oil production site “Oil Stones”(Azerbaijan) in the Caspian Sea as imaged by Radarsat-2 on 11 and 12/07/2009. © MDA/CSA

heavy products: intermediate (IFO) & heavy (HFO) fuel oils) or by-products, ships’ bunkers and lubricates, oily refuse or oil mixed waste (CEDRE, 2004). Their appearance, physical characteristics (density, viscosity, pour point) and behavior (e.g., spreading, evaporation, dispersion and emulsification) depend on their composition. Shipping is the main source of oil spills in the open sea (Gade and Alpers,

9


177

Fig. 9.7 Ship-made oil spill in the Japan Sea on the ERS-1 SAR image. © ESA

1999; Lu et al., 1999; Pavlakis et al., 2001; Ivanov and Zatyagalova, 2008), but much man-made oil pollution also comes from land-based activity (Alpers and Espedal, 2004). Oil pollution from routine ship operations includes ballast water, tank washing residues and other oil mixtures from the engine room and bilge waters. Such pollution is also known as slops. Sludge includes engine room waste and foul bilge water from all types of ships. Most ships discharge their oily effluents en route, leaving linear spills (Pavlakis et al., 2001; Alpers and Espedal, 2004). Estimates of total volumes of oils released every year in the marine environment vary widely, but are considered to have decreased from 6.0 million in the early 1970s to 1.3 million tons at present (ITOPF, 2009). However, these estimates are still very uncertain and the real situation is poorly understood, particularly with regard to oil discharges from land-based sources, which is a large and still unknown part of oil pollution (Oil in the sea, 2003). The spatial distribution of these oil slicks marks the international and domestic shipping lanes and oil producing regions (Lu et al., 1999; Alpers and Gade, 1999; Ivanov et al., 2004; Shi et al., 2008). For this reason, the main problem for ocean remote sensing is to first detect and identify these slicks, and then to map their spatial distribution. Man-made biogenic oil slicks are produced by the discharge of organic matter resulting from human activities (Fig. 9.8), such as fish and food processing onshore and offshore (Espedal, 1998; CEDRE, 2004; ITOPF, 2009). Fish oils together with fish waste are released directly from fishing vessels or fish processing plants onshore. Vegetable oils are oil extractions from plants and fruits, such as sunflower, soybean, rapeseed, olive, castor, corn, palm nut, etc. According ITOPF (2009),

178

A.Y. Ivanov

Fig. 9.8 Spill from a palm oil production plant in Colombia. © Elastec/ITOPF

Fig. 9.9 Seepage slicks in the SW Caspian Sea as imaged by Envisat on 9/08/2003. © ESA

9


179

the fate, behavior and environmental impact of vegetable oil spills on the marine environment are not as widely appreciated as the impact of mineral oils. Natural mineral oil slicks are the result of bottom seepage or oil (hydrocarbon) seeps, which are those places on the seafloor, where liquid or gaseous hydrocarbons come to the sea surface through fractures and holes in the bottom. Oil seeps are quite common and are found in many places of the world’s ocean (e.g., in the Gulf of Mexico, Santa Barbara Channel, Caspian Sea; (Oil in the sea, 2003)). Oil, upon reaching the surface after being emitted from the bottom seep, has a tendency to spread out into thin layers, which cause oil slicks on the sea surface (Figs. 9.9 and 9.10) that can easily be detected by remote sensing techniques (MacDonald et al., 1993). It may seem unusually high, but almost half of the oil detected in the ocean is emitted from natural sources. Kvenvolden and Cooper (2003) report that natural seepage introduces into the marine environment from 0.2 to 2.0 million tons of crude oil per year. This amounts to about 47% of all crude oil currently entering

Fig. 9.10 Seepage slicks on Lake Maracaibo (Venezuela) on the Terra/ASTER image Source: http://earthobservatory.nasa.gov/IOTD/view.php?id=3533

180

A.Y. Ivanov

the marine environment, while mankind is responsible for the rest. Tracking and locating these slicks are helpful for two reasons. On the one hand, the seeps can indicate local hydrocarbon deposits, and indeed, a number of oil fields were discovered from observations of oil seeps (Oil in the sea, 2003). On the other hand, on a historic timeline, they may somehow influence the direction and magnitude of climate changes. Natural biogenic films, also known as surfactants due to their surface-active characteristics, are bio-products of natural life in the ocean. Natural biogenic substances are normally released into the marine environment and produced by plankton organisms and fishes during their life-cycle. They consist of surface-active, mainly organic compounds (hydrolyzed amino acids, proteins, fatty acids and alcohols, lipids, etc.) that are naturally secreted by marine organisms (Monin and Krasitskii, 1985; Alpers and Espedal, 2004). They (also referred as natural slicks) form an extremely thin layer on the sea surface, the so-called “oil microlayer”, which can be seen, when the wind is low (2–4 m/s), as patches of mirror-flat water (Monin and Krasitskii, 1985; Alpers and Espedal, 2004). In general, the biogenic films are only

Fig. 9.11 Biogenic slicks in the East China Sea on the Envisat SAR image of 24/08/2005. © ESA

9


181

about one molecular layer thick (~10−9 m) (Levich, 1962), and only few liters of surface-active material can cover an area of 1 km2 (Alpers and Espedal, 2004). The probability of encountering natural biogenic films is strongly enhanced, especially at times when the biological productivity is high, i.e., during and after phytoplankton blooms. In some cases the part of the sea covered by natural films can be very large, as it often the case in the inland and marginal seas (Fig. 9.11). When their concentration is very low, surfactants may also be present without directly showing up as slicks (da Silva et al., 2000). Of course, all these oily substances form layers of different thickness on the sea surface, very efficiently damp the short surface (capillary-gravity) waves and for this reason can be easily detected by remote sensing techniques, primarily by imaging radars.

3 Remote Sensing of Oil Films Fingas and Brown (1997) reviewed the airborne and spaceborne sensors and evaluated them in terms of their usefulness in responding to oil spills. The remote sensing techniques of marine oil spills include applications of optical (cameras and lasers), infrared (IR) and microwave (radiometers), and radar (real aperture radars – SLAR and synthetic aperture radars – SAR) sensors on board aircrafts or satellites. While airborne sensors are limited by small area of survey and high cost, satellite sensors have other kind of limitations. Satellite systems used visible bands are susceptible to false detections of oil due to sun altitude, clouds, shallow bottom topography and floating seaweeds. Moreover, traditional satellite optical sensors (e.g., SeaWiFS, MODIS, AVHRR) have not been of much use for oil spill detection due to spatial resolution of about 1 km (Fingas and Brown, 1997). Middle-resolution optical sensors (e.g., ETM/Landsat, HRV/SPOT, LISS/IRS etc.) do not provide daily observations, and the data are of limited spatial coverage and expensive (Fingas and Brown, 1997). Nevertheless, under cloudless conditions optical images demonstrate a good capability, as on Terra/ASTER image with seepage slicks on Lake Maracaibo (Venezuela) in Fig. 9.10. Other remote sensors, such as IR (or IR/UV) systems, microwave radiometers and fluorometric lasers have similar inherent weaknesses (lack of sufficient spatial resolution and operational applicability) and can be used on aircraft (Fingas and Brown, 1997). Therefore, there are situations when these techniques are unable to detect and mapping oil spills. It is in these cases that radar remote sensing is required. On SAR images, the dark signatures can be used to detect oil films on the sea surface and to distinguish them from different non-film phenomena. However, SAR data are not available on daily basis, are limited by wind speed (Bern et al., 1992; Ivanov et al., 1998; Alpers and Espedal, 2004), and have other disadvantages (such as their being costly and prone to many interferences; (Fingas and Brown, 1997)). SAR systems rely on the detection of variations of the sea surface roughness. The possibility of detecting an oil spill in a SAR image depends on the fact that oil films

182

A.Y. Ivanov

decrease the backscattering property of the sea surface, damping the short gravitycapillary waves that results in dark signatures, which contrast with the grey tone of the surrounding, normally backscattering sea surface (Alpers and Hühnerfuss, 1989; Gade et al., 1998). SARs used for monitoring of oil pollution usually operate at incidence angles from 20◦ to 60◦ , i.e., at those angles where the radar backscattering can be described by Bragg’s scattering theory (Valenzuela, 1978). At wind speeds 14 m/s), it causes the oil spill to disappear from the sea surface due to mixing down into the water column. Nevertheless, definite detection is difficult because SAR doesn’t measure any film characteristics and detection is possible in a narrow range of wind speeds (Bern et al., 1992). As previously mentioned, there are three main categories of slicks recognized on SAR images: oil pollution (the thickest slicks), seepage slicks (i.e. created by oil coming from seepages), and natural biogenic films (the thinnest slicks). The signatures of the second and third types can be mixed with those of oil spills produced by vessels, rigs or pipelines (Fig. 9.4–9.7). Discrimination between them and oil spills requires analysis of slick appearance (shape, size, area, dB-contrast, edge type, texture), environmental conditions (wind, currents, precipitation), contextual information about slick position relative to surrounding objects (ships, ship lanes, rigs, platforms, seeps) and other concomitant oceanic and atmospheric phenomena (internal waves, upwelling, grease ice, algae blooms etc.) (Espedal, 1998; Alpers and Espedal, 2004). Discrimination between them is a main and important task of ocean remote sensing (Brekke and Solberg, 2005), especially in the context of global changes. To sum it up, remote sensing of oil spills with a SAR and their delectability is, to some degree, a combination of: (1) environmental parameters (wind speed and sea state), (2) SAR parameters (frequency, incident angles, polarization, resolution), (3) oil film characteristics (oil type and film thickness), and (4) slick characteristics extracted from a SAR image (contrast, shape and dimension). Evidently, SAR systems play a major role in remote sensing of oil spills. For detection, a SAR has to have radar frequency, incident angles and polarization, which cover the frameworks of Bragg scattering mechanism. Moreover, resolution and swath width of SAR images have to be sufficient for routine oil spill monitoring. Wind speed (sea state) considered as critical limitations of SAR seems to be the same for all remote sensing systems. Wind speed >2.5 m/s is sufficient to produce background sea surface roughness allowing an oil spill became detectable. Wind speed >12 m/s, called an upper wind speed limit, prevents the oil spill detection.

9


183

Wind strongly affects the shape and appearance of the oil patch (Espedal and Wahl, 1999; CEDRE, 2004). Oil film characteristics also play important role in spill detection. The thinnest biogenic slicks are visible on the sea surface up to 4–5 m/s, but aren’t at wind speed >6 m/s (will tend to mix down into the water column and become undetectable by SAR). Seepage slicks are still detectable at winds of 6–7 m/s. The same is for the oil spills of middle thickness, but heavy oil products and crude oil (the thickest films) can be detectable on the sea surface up to 12 m/s, and even 14 m/s (Ivanov et al., 1998; Alpers and Espedal, 2004). Thus, wind speed can sometimes be a natural filter for oil films. When oil is in the sea, mineral oil films are subject to the actions of the wind, currents, and physical-chemical transformations, called weathering processes (Fig. 9.1). After some time, oil slicks become dispersed and undetectable by remote sensing techniques. This time depends on the type of the oil, its volume and film thickness, and on the wind and wave conditions (Alpers and Espedal, 2004). It usually varies from a half of day for seepage slicks and ship discharges, to several weeks or even months for crude oil. Slick characteristics visible on a SAR image such as contrast, shape and dimension are very important in remote sensing of oil spills. Apparently, the contrast or oil slick – surrounding ocean backscatter ratio is a complex result of oil film parameters, and wind speed & sea state. Slick shape and size are results of action of currents and winds and their history (Espedal and Wahl, 1999). Thin oil films break up into filaments or windrows due to circulation of currents and wind (Alpers and Espedal, 2004; CEDRE, 2004), whereas crude oil and heavy fuel oils are viscous and usually remain in relatively compact patches. Moreover, all oil patches are subjected to spreading, transportation, mixing with water and weathering processes (Fig. 9.1); they transform and decay by different processes at the air-sea interface. The appearance of seepage slicks and remnant oil pollution sometimes is similar to those of natural biogenic slicks. And sometimes it is not easy to discriminate oil spills from seepage slicks only using backscatter contrast, shape and size (Fig. 9.12). As oil slicks become smaller, thinner and less distinct, classification is less precise, and it is not always possible to understand a definitive origin. The thinnest biogenic slicks can act as ideal tracers. In convergent zones of currents and eddies they render visible the current patterns. Seepage slicks and oil spills persist over wider wind conditions and for longer time periods than natural biogenic slicks. The thinner oil films the quicker a slick disappears from the sea surface when wind increases. However, in many cases, the discrimination of mineral oil patches from natural surface films is possible by their shape and size (Alpers and Espedal, 2004). Not only oil films on the sea surface can create dark features in the SAR images of the sea surface. There are a variety of hydrodynamic, aerodynamic and biological phenomena in the upper ocean and the low atmosphere, which damp short surface waves with the strength of the oil and create, the so-called “look-alikes” (Espedal, 1998). The term “look-alikes” unites all slicks that look like oil slicks, which are produced by phenomena in the ocean and the atmosphere, and detected

184

A.Y. Ivanov

Fig. 9.12 Typical oil pollution from different sources in the Yellow Sea on the ERS-2 SAR images of 13/04/2004. © ESA

9


185

by SAR as dark patches. (Espedal, 1998). Typical phenomena producing look-alikes are: currents, upwelling, ship-made turbulence, internal waves, SST and wind stress variations, precipitation, atmospheric gravity waves, grease ice, algae blooms, floating seaweeds, sperm and eggs of marine animals, and shallow bottom topography (CEDRE, 2004; CEDRE, 2007). Under moderate wind conditions surface manifestations produced by these phenomena can be imaged as dark signatures with similar shape and size, and even contrast. Their manifestations on SAR images may cause false detections (Espedal, 1998; Alpers and Espedal, 2004). Further information and discussions on the typical appearance and detectably of oil spills on SAR images, possibilities to identify man-made oil spills on a complex image background as well as overview on these topics a reader can find, e.g., in Espedal (1998), Alpers and Espedal (2004), CEDRE (2007). In addition, remote sensing can be a powerful tool to control applicability of international conventions on marine oil pollution. According to the MARPOL Convention’s protocols, seagoing ships may discharge oily mixtures into the ocean at a rate of 16 l per km and at a distance longer than 80 km from a shore (MARPOL, 2009). But evidence shows that it is common practice to exceed this limit. The majority of oil pollution incidents takes place just beyond the territorial waters, and therefore beyond the attention of pollution authorities and environmental agencies (Pavlakis et al., 2001). To sum it up, it should be noted that despite different interferences, ambiguities and look-alikes environmental scientists and oceanographers faced with, a problem of remote sensing of oil films seems to be practically solved. Operational monitoring systems for oil pollution using satellite images are recognized as national/regional priorities. Many national space agencies and pollution authorities demonstrate SAR’s potential and use of SAR images to optimize oil pollution control, monitoring and surveillance. The process of oil spill identification consists of several and now well defined steps (Brekke and Solberg, 2005). Image analysts use visual analysis, semi-automatic and automatic methods to detect and identify oil spills and their sources. They provide subsequent users with an accurate and objective first guess at the presence of oil on SAR image/sea surface. By wide use of contextual information about surrounding objects in the sea, a discrimination procedure can be significantly facilitated. Simultaneously wind field and ship positions can be extracted from the same SAR image. Outputs, which are wind, ship and oil spill’ distribution maps, are converted in to shape files and than are sent to different catalogues, archives and end users. All these data are available for analysis from the beginning of the 1990s.

4 Mapping of Oil Films in the Context of Climate Changes A considerable amount of new research over the past 10 years has led to an understanding of vital importance of the interface between the ocean and the atmosphere, and how it may interact with the processes involved in global changes. Recent

186

A.Y. Ivanov

research has shown that oil spills, both natural (biology- and geology-made) and man-made, have a far greater impact on a changing climate than previously thought (Kraus, 1972; Monin and Krasitskii, 1985; Liss and Duce, 1997; Toba, 2003). It has been shown that major oil spills may have impact on the marine environment, even at test sites hundreds of kilometers from the accidents. For example, some findings show that oil spills in Europe, such as the Prestige tanker disaster off the coast of Spain in 2002, have consequences on marine environments far from the scene of the initial pollution (Garcia Perez, 2003). Evidently, prior to the investigation of the climate changes, the detail distribution of oil films over wide areas in the ocean and physical-chemical properties of the air-sea interface under their influence have to be meticulously studied (Monin and Krasitskii, 1985). In this connection, only long-term mapping of oil films enables scientists to see a correlation between total volume and extent of oil films in the ocean and climate changes. Oil spill researchers also showed that oil spill distribution and extent may have a significant effect on the properties of the ocean-atmosphere interaction (Monin and Krasitskii, 1985). The consistent high changes of a number of phenomena in the climatic system for the past 50 years may be due to human activity-induced oil pollution of the world ocean. If this is the case, it would mean that marine environment is vulnerable to human activities on two counts: directly through oil pollution from illegal discharges and accidents, and indirectly through global climate changes. As discussed above, biogenic slicks and seepage slicks are natural in the ocean and are permanent factors in the ocean-atmosphere system. Man-made oil pollution comes from a variety of sources, such as routine ship operations and accidents. The first two factors are taken into consideration in global climatic or ocean-atmosphere scales. The last is probably not. This is because the real coverage and spatial extent of oil spills are still not clearly understood. In other words, the distribution of natural oil slicks as well as oil spills is a crucial environmental issue. To understand that, one needs to collect as much data covering the most of seas of the world ocean as possible for at least the last 30 years. This can be realized only though remote sensing and particularly through a GIS, which allows scientists to collect SAR images, manage detected oil spills and link them with appropriate sources (Ivanov and Zatyagalova, 2008). Evidently, remote sensing is a powerful tool to control better marine oil pollution. To minimize the risks of false identification/classification of dark features on SAR images a number of special algorithms have been developed (see, e.g., reviews by Brekke and Solberg, 2005 and Topouzelis, 2008). These algorithms are based on various features and parameters available from a SAR image and taken into account environmental parameters and contextual information, such as wind, traveling ships, offshore oil and gas objects. This methodology usually uses objectoriented approach and image segmentation techniques (Brekke and Solberg, 2005; Topouzelis, 2008). Another approach developed worldwide and summarized by Ivanov and Zatyagalova (2008) (see also literature cited herein), called GIS-approach, uses a wide set of detailed geographic & oceanographic (coastline, bottom topography,

9


187

wind, currents, etc.), geologic-geophysical (oil and gas formations, oil fields, seepage, etc.) and industrial (onshore and off shore oil & gas infrastructure, shipping infrastructure, etc.) information about a marine basin. All this information can be arranged in a GIS, allowing manipulation of all kind of spatial information including remote sensing images. For example, images of oil spills collected over a long period of time will show in GIS risk zones or hotspots, where chronic oil pollution occurs and accumulates. Moreover, GIS-approach allows scientists to calculate spatial distribution (extent) and total area of detected oil films that, under certain assumptions about thickness of films, can give information about total volumes of spilled oil. Further monthly/annually trends in oil pollution can be analyzed. Finally, knowledge obtained in long-term monitoring based on GIS-approach can be used in climatic models enabling us to see an impact of man-made pollution on trends in climate changes. GIS-approach is also considered to be a tool for improving oil spill identification and classification in SAR imagery. First of all, a number of limitations of SAR for oil spill detection are being recognized. Second, because operational detection service based on SAR images still depends on an operator’s experience, such analysis is considered to help in identification of oil sources. It is proposed, therefore, that GISapproach and the use of geographic, remote sensing, contextual and other ancillary data and information can make an important aid to interpret correctly the slicks signatures, providing a framework for analysis (Ivanov and Zatyagalova, 2008). A GIS database for the marine basins can be compiled of data of several sources. Vector shoreline and bathymetric data can be obtained from the national geophysical data centers, by digitizing detailed nautical maps, or extracting from global/regional topographic models. Both oil and gas production infrastructure offshore and industrial/municipal infrastructure onshore can be compiled from data of regional/local archives or corporative databases. Main shipping lanes can be also obtained from nautical maps. Fishing/aquaculture areas, coastal and marine restricted, protected and vulnerable zones are responsibility of corresponding authorities. All datasets and geographic information have to be compiled together and placed into a generally used geographic projection using GIS-software. Thus, GIS, in providing an efficient storage, retrieval, analysis and visualization of geographic, oceanographic, environmental and industrial information, is considered to support oil spill detection and identification. Fig. 9.13 displays a window of the GIS constructed for oil spill applications in the Yellow Sea and East China Sea. Such GIS was created using different sources, such as maps, charts, satellite images, raster and vector digital data. On Fig. 9.13, only layers related to oil production and transportation, i.e., shipping lanes and oil production platforms are displayed. One can load a set of SAR images into a GIS and display information, which is necessary for the support of identification of oil spill sources, and decision about extent of oil pollution. In order to validate a methodological approach, a number of projects have been carried out. In these projects, the mapping of oil slicks/spills was undertaken in the Sea of Okhotsk/Japan Sea, Caspian Sea, Black Sea, Yellow and East China Sea, and in the Gulf of Thailand (Ivanov and Zatyagalova, 2008; Shi et al., 2008).

188

A.Y. Ivanov

Fig. 9.13 An example of GIS-made oil spill distribution map for the Yellow Sea and East China Sea (main regional/international ship lanes are also shown in red) (figure from Shi et al., 2008)

5 Conclusion Natural and man-made oil slicks affect the physical properties of the air-sea interface, and, in turn, impact on the processes, which determine the ocean-atmosphere interaction. Among them are: energy transfer from wind to waves, evaporation and convection, SST variability, gas exchange, skin-layer and foam on the sea surface. Existing knowledge about the effects of oil films on ocean-atmosphere interactions and global warming, in the current situation of global changes, is insufficient. Their regional and global impacts are estimated very approximate, because the real distribution of oil slicks as well as of oil spills in the world ocean is indeed unknown.

9


189

Significant advances have been made in the last decade both in estimating air-sea interaction over the ocean and in developing methods for detection and identifying of oil spills. Oil films covering the ocean surface can now be resolved with at sufficient temporal and spatial resolution by using remote sensing techniques. It is expected, in the near future, to use satellite-derived oil spill products that have the appropriate resolution instead assimilation products that often do not appropriately resolve oil films variability. Now remote sensing, being a critical element of any ocean monitoring system, plays an increasingly important role in oil slick monitoring. In its present state, optical satellite imagery does not offer much potential for oil spill remote sensing. The imaging radars have become an important tool in the monitoring of marine oil spills due to their high resolution, all-weather and all-day capability. The potential of spaceborne SARs for oil spill detection has been demonstrated many times in the coastal zones and in the open ocean via observations of oil films floating on the sea surface. In general, this monitoring based on SAR images reveals the dramatic scales of oil pollution in the world ocean. It was also shown that SAR allows detection, localization and furnishing with information on the oil spills. Modern multi-swath, multi-resolution and multi-polarization SAR-equipped satellites, such as Radarsat, TerraSAR-X, COSMO-SkyMed and future Sentinel, will play a significant role in providing not only real data and information on accidents and shipwrecks (Ivanov, 2010), but on oil film extent. Although conventional oceanographic technologies can provide more of the required data, the remote sensing can provide wider spatial and temporal coverage, new information products and effective solutions for monitoring of natural slicks and man-made oil spills. Through multi-temporal imaging, remote sensing techniques can provide information on slick distribution on a global and regional scale. It seems in the context of global changes only remote sensing can answer a number of questions faced by applying the approaches and methods developed recently. A GIS-approach for the mapping of oil slicks and spills in the sea is becoming an important tool, whereas a set of oil spill distribution maps, when combined and generalized, can be some kind of a response to the challenges of the 21st century.

References Alpers W, Espedal H (2004) Oils and surfactants. In: Synthetic aperture radar marine user’s manual, chap 11. NOAA/NESDIS, U.S. Department of Commerce, Washington, DC, pp 263–276. Alpers W, Hühnerfuss H (1989) The damping of ocean waves by surface films: a new look at an old problem. J Geophys Res 94(C5):6251–6265. Bern T-I, Wahl T, Andersson T, Olsen R (1992) Oil spill detection using satellite based SAR: experience from a field experiment. Proceeding of the 1st ERS-1 symposium, Cannes, France, 4–6 November 1992, no 2, pp 829–834. Brekke C, Solberg AHS (2005) Oil spill detection by satellite remote sensing. Remote Sens Environ 95:1–13.

190

A.Y. Ivanov

CEDRE (2004) Operational guide: aerial observation of oil pollution at sea. Centre de Documentacion de Recherche et d’Expérimentetion sur les Pollutions Acidentelles des Eaux http://www.cedre.fr/en/publication/aeri/aeri.php. Accessed 15 March 2010. CEDRE (2007) Use and analyse of satellite SAR images for oil spills detection. Centre de Documentacion de Recherche et d’Expérimentetion sur les Pollutions Acidentelles des Eaux http://www.cedre.fr/en/publication/workshop/guide-sat.pdf. Accessed 15 March 2010. Clemente-Colón P, Yan X-H (2000) Low-backscatter ocean features in synthetic aperture radar imagery. Johns Hopkins Apl Technical Digest 21(1):116–121. da Silva JC, Ermakov SA, Robinson IS (2000) Role of surface films in ERS SAR signatures of internal waves on the shelf. 3. Mode transitions. J Geophys Res 105:24089–24104. Ermakov SA, Salashin SG, Panchenko AR (1992) Film slicks on the sea surface and some mechanisms of their formation. Dyn Atmos Oceans 16:279–304. Espedal HA (1998) Oil spills and its look-alikes in ERS SAR imagery. Earth Observ Remote Sens, Russ Acad Sci 5:94–102. Espedal HA, Johannessen OM, Johannessen JA, Dano E, Lyzenga DR, Knulst JC (1998) COASTWATCH’95: ERS 1/2 SAR detection of natural film on the ocean surface. J Geophys Res 103(C11):24969–24982. Espedal HA, Wahl T (1999) Satellite SAR oil spill detection using wind history information. Int J Remote Sens 20(1):49–65. Fingas MF, Brown CE (1997) Review of oil spill remote sensing. Spill Sci Tech Bull 4(4):199–208. Gade M, Alpers W (1999) Using ERS-2 SAR images for routine observation of marine pollution in European coastal waters. In: Science of the total environment 237/238. Elsevier Science B.V. London, pp 441–448. Gade M, Alpers W, Hühnerfuss H, Wismann VR, Lange PA (1998) On the reduction of the radar backscatter by oceanic surface films: scatterometer measurements and their theoretical interpretation. Remote Sens Environ 66(1):52–70. Garbe CS, Handler RA, Jähne B (eds) (2007) Transport at the air-sea interface: measurements, models and parameterizations. Springer, New York. Garcia Perez JD (2003) Early socio-political and environmental consequences of the Prestige oil spill in Galicia. Disasters 27(3):207–223. ITOPF (2009) http://www.itopf.com. Accessed 8 December 2009. Ivanov AY (2010) The oil spill from a shipwreck in Kerch Strait: Radar monitoring and numerical modeling. Int J Remote Sens 31(17–18):4853–4868. Ivanov AY, Fang M, He M-X, Ermoshkin IS (2004) An experience of using Radarsat, ERS-1/2 and Envisat SAR images for oil spill mapping in the waters of the Caspian Sea, Yellow Sea and East China Sea. Proceedings of the Envisat & ERS symposium, 6–10 September 2004, Salzburg, Austria (ESA SP-572, April 2005). Ivanov AY, Litovchenko KT, Ermakov SA (1998) Oil spill detection in the sea using Almaz-1 SAR. J Adv Mar Sci Tech Soci 4(2):281–288. Ivanov AY, Zatyagalova VV (2008) A GIS approach to mapping of oil spills in a marine environment. Int J Remote Sens 29(21):6297–6313. Kraus EB (1972) Atmosphere-ocean interaction. Oxford University Press, London. Kvenvolden KA, Cooper CK (2003) Natural seepage of crude oil into the marine environment. Geo-Mar Lett 23:140–146. Levich VG (1962) Physicochemical Hydrodynamics. Prentice-Hall, Englewood Cliffs. Liss PS, Duce RA (eds) (1997) The sea surface and global change. Cambridge University Press. Litovchenko K, Ivanov A (2006) Oil spills on the Almaz-1 and ERS-1 SAR images: Results from the DOSE-91 experiment. In: Gade M, Hühnerfuss H, Korenowski G (eds) Surface Slicks and remote sensing of air-sea interaction, Springer, Berlin, Heidelberg, pp 299–313. Lu J, Lim H, Liew SC, Bao M, Kwoh LK (1999) Oil pollution statistics in Southeast Asian waters compiled from ERS synthetic aperture radar imagery. Earth Observ Quart 61:13–17. MacDonald IR, Guinasso NL Jr, Ackleson SG, Amos JF, Duckworth R, Sassen R, Brooks JM (1993) Natural oil slicks in the Gulf of Mexico visible from space. J Geophys Res 98(C9):16351–16364.

9


191

MARPOL (2009) http://www.imo.org/conventions/contents.asp?doc_id=678&topic_id=258. Accessed 19 December 2009. Monin AS, Krasitskii VP (1985) Phenomena on the ocean surface. Leningrad: Gidrometeoizdat (in Russian). Oil in the sea (2003) III: Inputs, fates, and effects. National Academy Press, Washington, DC. Pavlakis P, Sieber AJ, Alexandry S (1996) Monitoring oil spill pollution in the Mediterranean with ERS SAR. Earth Observ Quart 52:13–16. Pavlakis P, Tarchi D, Sieber AJ, Ferraro G, Vincent G (2001) On the monitoring of illicit vessel discharges: a reconnaissance study in the Mediterranean Sea. Report EUR 19906 EN, European Commission/JRC. Scott JC (1986) Surface films in oceanography. ONRL Workshop report c-11-86. Shi L, Ivanov AY, He M-X, Zhao C (2008) Oil spill mapping in the western part of the East China Sea using synthetic aperture radar imagery. Int J Remote Sens 29(21):6315–6329. Toba Y (ed) (2003) Ocean-atmosphere interactions. Kluwer Academic Publishers, Tokyo. Topouzelis KN (2008) Oil spill detection by SAR images: Dark formation detection, feature extraction and classification algorithms. Sensors 8:6642–6659. Valenzuela GR (1978) Theories for the interaction of electromagnetic and oceanic waves – A review. Bound Layer Meterol 13:61–85. Wahl T, Skøelv Å, Andersen JHS (1994) Practical use of ERS-1 SAR images in pollution monitoring. Proceedings of the IGARSS’94, 4:1954–1956.

Part III

Coastal Environment

Chapter 10

Coastal Monitoring by Satellite-Based SAR Antony K. Liu

Abstract The ability of Synthetic Aperture Radar (SAR) for monitoring surface signatures of swells, wind fronts, bottom features, oil slicks, and eddies has been amply demonstrated. The combined use of IR (AVHRR), ocean color (SeaWiFS, MODIS), and SAR (ERS-1/2, ENVISAT, RADARSAT) data can provide frequent high resolution coverage of the coastal area for the evolution study of oceanic processes. This chapter presents examples of these applications. During the Bering Sea Fisheries-Oceanography Coordinated Investigation (FOCI) field test, simultaneous satellite SAR data and in-situ measurements from moorings and ships were collected. Ships and their wakes are commonly observable in high-resolution satellite SAR imagery. This can be useful in national defense intelligence, shipping traffic, and fishing enforcement, especially when combined with the Vessel Monitoring System (VMS). SAR images can track the movement of the ice edge and floes in the marginal ice zone (MIZ). Results are relevant to climate change and fishery management. Sea-ice motion derived by satellite data can be used to interpret the Arctic ice retreat in the summer of 2007. Underwater bathymetry has been mapped by satellite remote sensing (SAR, SPOT, and LANDSAT) in the Spratly Islands of South China Sea (SCS) for ship navigation. As presented by these case studies, the use of SAR-derived observations can supply valuable information for the protection of the environment. Keywords Synthetic Aperture Radar · Fisheries-oceanography · Vessel monitoring system · Sea-ice motion · Pollution and hazard protection

A.K. Liu (B) National Taiwan Ocean University, Keelung, Taiwan; NASA Goddard Space Flight Center, Greenbelt, Maryland, USA e-mail: [email protected]


195

196

A.K. Liu

1 Introduction Remote sensing with repeated coverage is the most efficient method to monitor and study marine productivity and pollution. The mapping of mesoscale ocean features in the coastal zone is a major potential application for satellite SAR data, especially for the ScanSAR on RADARSAT with 500 km swath (Beal and Pichel, 2000). The use of SAR-derived observations to track eddies, surface temperature-related features, and river and estuarine plumes can aid in the management of fisheries and environment monitoring (Liu et al., 1994a). Especially in the Alaska coast area, uniform cold sea surface temperatures and cloud cover preclude AVHRR measurements of surface temperature features, and obscure ocean color observations. SAR is a side-looking imaging radar usually operating on either an aircraft or a spacecraft. The ability of a SAR to provide valuable information on the type, condition, and motion of the sea-ice and surface signatures of swells, wind fronts, and eddies has been amply demonstrated (Fu and Holt, 1982; Liu and Wu, 2001). With all-weather, day/night imaging capability, SAR penetrates clouds, smoke, haze, and darkness to acquire high quality images of the Earth’s surface. This makes SAR the frequent sensor of choice for cloudy coastal regions (Wu et al., 2000; Liu et al., 2003). Space agencies from U.S., Canada, and Europe use SAR imagery on an operational basis for sea ice monitoring, and for the detection of icebergs, ships, bathymetry, and oil spills (Liu et al., 2007). Satellite remote sensing with repeated coverage is the most efficient method to monitor and study coastal environment, marine productivity, underwater bathymetry, and pollution. The mapping of mesoscale ocean features in the ocean is one of major potential applications for satellite data. An eddy is a current that usually moves in a circular path; it may develop where currents encounter obstacles or where one flow passes another. For fisheries application in the Alaska area, eddies frequently occur over the sea valley west of Kodiak Island, as revealed by satellite imagery, buoy trajectories, and moorings. FOCI studies indicate a tendency for the coincidence of larval patches and mesoscale eddies and that for early larvae, presence within an eddy improves survival. High abundances of walleye pollock larvae often reside in eddies (Schumacher and Kendall, 1995). Ocean features such as eddies, fronts, and ice edges can result in changes in water temperature, turbulence, or transport and may be the primary determinant of recruitment to fisheries. The survival of larvae is enhanced if they remain on the continental shelf and ultimately recruit to nearshore nursery areas. Features such as fronts and eddies can act to retain larval patches within the shelf zone. For coastal monitoring application, mesoscale eddies are key features associated with the ice margin and are usually attached to the ice edge. Satellite observations of mesoscale features can play a crucial role in ocean-ice interaction study. Using SAR imagery, the spatial variability of the ice cover and current field can be observed (Liu et al., 1997b; Yu et al., 2006). Also, the wavelet transform-based ice tracking method has been developed and used to SSM/I, NSCAT, QuikSCAT, and AMSR-E data to obtain daily sea-ice drift information for both the Arctic and Antarctic (Liu and

10

Coastal Monitoring by Satellite-Based SAR

197

Cavalieri, 1998; Liu et al., 1999). Based on these satellite-derived maps, Arctic seaice advance/retreat can be observed and even predicted, especially in the summer. Ships and their wakes are commonly observable in high-resolution SAR imagery from satellite (Liu et al., 1996). The use of satellite images to monitor the commercial and fishing activities has been developed for VMS. Furthermore, satellite remote sensing can be a very useful tool for environment monitoring, especially for red tides, pollution control, ship navigation, and hazard protection as demonstrated by Liu et al. (2007) in the Spratly Islands of SCS. A series of selected SAR applications for coastal monitoring demonstrated as the case studies will be presented in this Chapter. In Sect. 2, the satellite remote sensing application on fisheries is first briefed with FOCI project as an example based on the monitoring of eddy/front in the coast of Alaska, Then in Sect. 3, ocean-ice interaction processes in the MIZ by waves and mesoscale features, such as upwelling and eddies, have been studied using ERS-1 SAR imagery. Also, the sea-ice motion derived by satellite microwave data is used to interpret the summer 2007 Arctic ice retreat in Sect. 4. VMS is briefly reported in Sect. 5 for the detection of ships and of ship wakes by means of remote sensing in the areas of shipping traffic, and fishing enforcement. Section 6 describes the bathymetry mapping by using satellite remote sensing, especially in the remote area such as Spratly Islands of SCS. As demonstrated in Sect. 7, Satellite remote sensing can also be a very useful tool for pollution and hazard protection, such as oil spills. Finally, coastal monitoring by using multiple sensors has been discussion for international collaboration to share the satellite data.

2 Fisheries-Oceanography Coordinated Investigation (FOCI) FOCI program was established by the National Oceanic and Atmospheric Administration (NOAA) in 1984 to examine the physical and biological factors that affect the walleye pollock fishery in Alaska. Walleye pollock is one of the world’s largest single-species fisheries with catches in Alaska annually exceeding 1 million metric tons. Understanding the dynamics of fish populations requires cooperative research among scientists from many different disciplines. In particular, studies in fisheries oceanography focus on the relationships between variations in fish populations and the marine environment. A major goal of FOCI is to understand natural changes in the abundance of walleye pollock and to provide this information to fishery managers. FOCI scientists realize this goal by integrating field, laboratory, and modeling studies to determine how biological and physical environmental factors influence walleye pollock in Alaska. The survival of larvae is enhanced if they remain on the continental shelf and ultimately recruit to near-shore nursery areas. Features such as fronts and eddies can act to retain larval patches within the shelf zone. So, the location of eddy formation coincides with the spawning region. Formation of three to four eddies each month during spring assures that some eggs hatch in an eddy. As a result of lack of dispersion, high abundance of larvae often occur in eddies. Some eddies tend to

198

A.K. Liu

remain nearly stationary for weeks and help retain larvae on the shelf. The integration of wind mixing, stratification within an eddy, and larval behavior is important to the subsequent survival of walleye pollock larvae. To monitor the frequency and location of eddies, scientists use moorings, AVHRR images, and high resolution SAR (Schumacher et al., 1991). The density of larvae contained in the eddy was estimated to be one order of magnitude greater than in surrounding waters. For example, Fig. 10.1 show (a) AVHRR image of Shelikof Strait obtained on May 1, 1992 from NOAA-11, (b) Sea surface temperature (◦ C) and (c) geopotential topography (dyn-m), both derived from a 38-station (crosses) CTD survey that took place

Fig. 10.1 (a) AVHRR image of Shelikof Strait obtained on May 1, 1992 from NOAA-11. (b) Sea surface temperature (◦ C) and (c) geopotential topography (dyn-m), both derived from a 38-station (crosses) CTD survey that took place on April 27–29, 1988. (d) ERS-1 SAR image of current features obtained on May 1, 1992. The island on the right-hand side of panel is the Kodiak Island

10


199

Fig. 10.2 ERS-1 SAR image of lower Shelikof Strait, acquired on October 23, 1991 showing a spiral eddy. The island on the right-hand edge of image near the eddy is Kodiak Island

on April 27–29, 1988. Figure 10.1d shows an ERS-1 SAR image of current features obtained on May 1, 1992. The island on the right-hand side of panel is the Kodiak Island. Another ERS-1 SAR image of lower Shelikof Strait, acquired on October 23, 1991 shows a spiral eddy in Fig. 10.2. In this image, an eddy with a diameter of approximately 20 km is visible due to low wind conditions. The island on the righthand edge of image near the eddy is the Kodiak Island. The eddy is characterized by spiraling curvilinear lines which are most likely associated with current shears, surface films, and to a lesser extent temperature contrasts. Field observations have revealed a connection between eddies and larvae. In Fig. 10.3, contours of larval abundance lie in close proximity to ARGO buoy trajectories in red and pink lines in the Bering Sea. The larval count of over 10,000 in the eddy area is ten times more than that in the surrounding area outside of eddy. Physical data show minimal exchange of water between eddies and adjacent waters, permitting estimates of mortality that reflect only predation and/or starvation. Mortality rates are low in eddies compared with mortality rates in other areas and in model simulations. Figure 10.4 shows the acoustic backscatter diagram of an

200

A.K. Liu

Fig. 10.3 Contours of larval abundance (rough counts in white numbers) and buoy trajectories (in red and pink lines) in the Bering Sea showing high concentrations of larvae and eddies are often observed together. The yellow dots are sampling stations

eddy in the Gulf of Alaska from EK500 echo sounder. Notice that the eddy boundary and ocean bottom (about 200 m depth) are both clearly identified with strong backscattering. Within the eddy backscatter is low, apart from some accumulative larger scatterers near the bottom. This is quite different from the uniform pattern of small scatterers outside the eddy. The use of SAR-derived observations to track eddies and fronts can supply valuable information and can aid in the management of the fishing industry.

3 Marginal Ice Zone (MIZ) The areas of the polar region ice cover which lie close to an open ocean boundary is generally called the marginal ice zone (MIZ). In this region the continuous ice cover characteristic of the central basin is broken up into floes by the flexural stress of waves and swell penetrating into the ice from the ocean. Ocean-ice interaction processes in the MIZ by waves and mesoscale features, such as upwelling and eddies, have been studied using ERS-1 SAR imagery and numerical models

10


201

Fig. 10.4 Acoustic backscatter diagram of the edge of an eddy (sloping dark line) in the Gulf of Alaska from EK500 echo sounder. The thick heavy curve near the bottom of the diagram is the ocean bottom in 200 m water depth. At depths from about 60 to 100 m, backscatter in the eddy (left) is lower than outside it (right)

(Liu et al., 1994b). Satellite observations of mesoscale features can play a crucial role in ocean-ice interaction study. Using SAR imagery, the spatial variability of the ice cover and current field can be observed (Liu et al., 1997b; Yu et al., 2006). Mesoscale eddies are key features associated with the ice margin and are usually attached to the ice edge. The surface effect models associated with grease ice, wavecurrent interaction, and atmospheric instability due to upwelling have been used to interpret the mesoscale features of eddies and fronts from SAR images. Typical scales of these eddies were 20–40 km. Rotation was mainly cyclonic with a maximum speed of up to 40 cm/s. Eddies play important roles in the distribution of heat, mass, and momentum fluxes in polar regions and in the control of the ice edge and its locations (Yu et al., 2009). They may directly affect the loss of ice by moving ice into warmer water which enhances ice melt and thereby the lateral mixing of heat. The edge of the sea-ice has been found to be highly productive for spring bloom and fishery feeding. In the Bering Sea, fish abundance is highly correlated with yearly ice extent because for their survival many species of fish prefer the cold pools left behind after ice retreat. SAR images are very useful for tracking the movement

202

A.K. Liu

Fig. 10.5 RADARSAT ScanSAR image collected over the Bering Sea on February 29, 2000 showing sea ice (grey, top), with open water in the bottom half of the image. The front visible in the open water is due to wind, but may be influenced by atmospheric stability due to colder surface water near the ice edge

of the ice edge. For demonstration, Fig. 10.5 shows a RADARSAT ScanSAR image near the ice edge in the Bering Sea collected on February 29, 2000 (centered at 59.6◦ N and 177.3◦ W). The sea-ice pack with ice bands extending from the ice edge can be clearly seen as the bright area because sea-ice surface is rougher than ocean surface (dark area). In the same image, a front is also visible and may be associated with the ice edge to the north. The cold water near the ice edge stabilizes ocean surface and appears as darker area compared with the other side of the front where it shows up as brighter area due to higher wind and higher sea states. In the Chukchi Sea, a sequence of seven SAR images from September 27 and October 3–18, 1991, with a 3-day interval have been studied for ice-edge advance/retreat. Each SAR image is 100 × 100 km in size and the pixel sizes are 100 m. The ice edge in each SAR image is delineated using two-dimensional Mexican-hat wavelet transform as an edge detector with a scale of 1.6 km, and the maximum intensity gradient lines above a threshold (70%) were extracted from the image (Liu et al., 1994b). The locations of the ice edge as determined from the SAR images are summarized in Fig. 10.6. The SAR coverage is in the rectangular box, and the locations of the four current meter moorings are indicated by solid circles for

10


203

Fig. 10.6 Summary of ice edge locations measured in the SAR coverage area by using a sequence of 7 ERS-1 SAR images of MIZ in the Chukchi Sea with 3-day intervals. Day numbers in 1991 are indicated. 270 is September 27. 291 is October 18. The locations of the four current meter moorings are indicated by solid circles

comparison. Figure 10.6 demonstrates that the ice edge motion is highly dynamic, with hundreds of kilometers of advance/retreat in 3 days. When the ice edges are relatively stationary, the formation of a mesoscale eddy near the meandering ice edge was clearly evident in the original imagery (not shown).

4 Summer Retreat of Arctic Sea-Ice The minimum of arctic sea-ice extent in the summer of 2007 was unprecedented in the historical record. Based on the model simulations to investigate the causes of this ice extent minimum, Lindsay et al. (2008) find that even though the 2007 ice extent was strongly anomalous, the loss in total ice mass was not. This unprecedented retreat of first-year ice during summer 2007 was enhanced by strong poleward drift over the western Arctic (Ogi et al., 2008). The anomalous sea-ice retreat in the summer of 2007 occurred mainly on the Pacific side of the Arctic basin. The motion of sea-ice is mainly wind-driven. The anti-cyclonic Beaufort Gyre re-circulates ice in the western Arctic, allowing it to become older and hence thicker, while the Transpolar Drift Stream advects ice across the basin and out through Fram Strait.

204

A.K. Liu

Fig. 10.7 Monthly Arctic sea-ice motion map derived from AMSR-E data in a grid of 100 km × 100 km for May (a) 2005, and (b) 2007

10


205

A large area of anomalous advective ice mass loss in the Pacific sector extends towards the North Pole. This strong transpolar drift is also observed by the drift of buoys and in the AMSR-E passive microwave measurements (Kwok, 2008). Wavelet transforms are analogous to Fourier transform but are localized both in frequency and time. A two-dimensional wavelet transform is a highly efficient band-pass data filter, which can be used to separate various scales of processes (Liu et al., 1997a). For effective identification and tracking of the common features in a pair of chosen images, a two-dimensional Mexican-hat wavelet transform is applied to the images with several spatial scales corresponding to extracted features and filtering out noise in the data (Wu and Liu, 2003). Filtered images have been examined for matching features by using templates. Matched templates are then readily converted to motion vectors and block-averaged onto a chosen grid (Liu et al., 2000b). The wavelet transform-based ice tracking method has been developed and used for SSM/I, NSCAT, QuikSCAT, and AMSR-E data to obtain daily sea-ice drift information for both the Arctic and Antarctic. The overall comparison of satellitederived ice motion with Arctic buoy data shows good agreement (Zhao and Liu, 2002, 2007). Figure 10.7 shows the monthly sea-ice motion map in the Arctic for May (a) 2005, and (b) 2007 derived from AMSR-E data in a grid of 100 km × 100 km. The blue arrows indicate ice velocities derived from satellite data. In a typical year

Fig. 10.8 A typical VMS/GIS chart/map for monitoring commercial and fishing vessels near Keelung Harbor in the north of Taiwan

206

A.K. Liu

(e.g., 2005), the anti-cyclonic Beaufort Gyre and Transpolar Drift Stream in the western Arctic are generally weak, so the ice extent and cover persist. However, in the year 2007 as shown in Fig. 10.8, the anti-cyclonic Beaufort Gyre and Transpolar Drift Stream in the western Arctic are relatively strong in April/May. This strong motion may setup the pre-condition to push the ice extent for an unprecedented retreat during the summer 2007. Such sea-ice information in the spring may be used for the prediction and early warning for the ice condition in the following summer, especially in the Arctic coastal areas.

5 Vessel Monitoring System (VMS) Conventional VMS consist of a beacon (transponder) located onboard a vessel capable of automatically reporting the vessel’s position through a satellite communications link to a shore-based terminal. Several U.S. and foreign fisheries require fishing vessels to carry VMS units as a condition of licensing and operating within the fishery. As an example, Fig. 10.8 shows a typical VMS/GIS chart/map for monitoring/tracking the commercial and fishing vessels near Keelung Harbor in the north of Taiwan. The European Union (EU) has been at the forefront of the move to use satellite to monitor fishing activities. In this connection, the basic function of VMS is to provide reports of the location of a vessel at regular intervals. Electronic devices, or “blue boxes”, are installed on board vessels. These devices automatically send data to a satellite system which transmits them to a ground station which, in turn, sends them to the fisheries monitoring center. Several researchers have studied integrated systems involving space-based SAR augmented with other space-based sensors such as AVHRR and optical sensors. The need for improved surveillance and control of commercial fishing on a global scale is urgent. Ships and their wakes are commonly observable in the high-resolution SAR imagery from satellite (Liu et al., 1996). In general, ship is a very effective corner reflector, so ship can be easily observed as a bright spot in the SAR image. Detection of ships and of ship wakes by means of remote sensing can be useful in the areas of shipping traffic, and fishing enforcement (Montgomery, 2000). Complementing these ship and wake detection technologies is the fact that there are instances where significant surface slicks are evident in the SAR image. Figure 10.9 shows an ERS-2 SAR image (50 × 50 km) in the SCS with two ships (bright dots), their ship wakes, and oil spill probably dumped by another ship at the earlier time. Ships can be easily detected by thresholds in the SAR imagery as bright dots in the SAR image. The detection algorithms of fishing or commercial vessels with SAR have been well studied (Pichel et al., 2004). The blended SAR and VMS position information will quickly show vessels that are not reporting by VMS. With such information, patrol craft can be vectored to the suspect vessel for identification and for determining its legal status. Occasionally, the ship in the SAR image remains almost invisible, and only trailing dark turbulent wakes are seen (Liu et al., 1996). Currently, there are two SAR sensors on different satellites, ERS-2 and ENVISAT, having acquisition time

10


207

Fig. 10.9 ERS-2 SAR image of 50 km × 50 km in the SCS with two ships (bright dots), their ship wakes, and oil spill probably dumped by another ship at the earlier time

Fig. 10.10 ENVISAT and ERS-2 SAR subscenes (28 km × 28 km) collected on April 27, 2005 north of Philippines in the Luzon Strait. The faint ship and its wake in green box near the eddy can be tracked easily

offset around 28 min with almost the exactly same path. That is, ERS-2 is following ENVISAT with a 28-min delay, which is a good time-scale for ship and ocean mesoscale feature tracking. Figure 10.10 shows ENVISAT and ERS-2 28 km × 28 km SAR subscenes obtained on April 27, 2005 north of Philippines in the Luzon

208

A.K. Liu

Strait. The images cover the area from longitude of 20.61◦ to 20.86◦ , and latitude of 122.09◦ to 122.34◦ . The mystery ship and its wake in the boxes near the eddy can be tracked easily in these figures. Then, the ship speed is estimated from the distance between ship locations in each SAR image and SAR acquisition time interval (28 min) to be 5.94 m/s. Very low backscattering of the ship configuration may have hidden the faint ship from view, or the wake could have been formed, instead, by an underwater vehicle. The eddy is characterized by surrounding white streaks and is probably generated by current interaction with near-by islands. Ocean surface drift near such an eddy has been derived by wavelet tracking using multiple SAR sensors (Liu and Hsu, 2009).

6 Bathymetry Mapping for Navigation The typical ways for bathymetry charting are ship multi-beam survey, airborne light detecting and ranging (LIDAR), and satellite mapping. Multi-beam surveys are slow and expensive. LIDAR is 20–100 times faster than ship multi-beam and even more cost effective. However, LIDAR only works in clear and shallow-water depth, so a vital role for ship multi-beam still exists. Satellite techniques do not meet the high standard of multi-beam survey and LIDAR, but do have clear advantages in cost and speed. They are more practical for some remote areas such as the Spratly Islands. Satellite bathymetry techniques can be used as a preview of a large area that will later be surveyed by multi-beam or LIDAR, and would indicate where the ship or LIDAR need to concentrate. Satellite mapping approach using SAR senses the interaction of a current with bottom bathymetry. So SAR does not depend on water clarity, but it does require tidal current exceeding about 0.5 m/s. It also needs ground truth calibration to invert SAR measurements into accurate water depth (Liu et al., 2007). C-MAP is a global electronic chart specifically designed to meet the needs of merchant marine vessels operating worldwide. Electronic Chart Display and Information System (ECDIS) consists of a database of electronic charts, together with the hardware and software needed to display simultaneously the charts and the ship’s own position (obtained from a GPS or another positioning sensor), and to perform navigational tasks such as route planning, route monitoring, measurement of distances on the chart, etc. Virtually all areas of interest to merchant marine navigation, fishing and pleasure boating are currently covered by the database, which includes at present data from more than 7,000 nautical charts. The database is however, being continuously increased both in terms of area coverage and chart detail. Spratly (Nansha) Islands in the South China Sea comprise 104 islands, reefs, cays, and banks. The area containing the islands stretches 810 km from north to south and 900 km from east to west. Taiping Island (Itu Aba) is the biggest island in the Spratlys, one of the northern Spratly Islands, and since 1955 claimed and held by Taiwan. Figure 10.11a shows a C-Map at a scale of 1:50,000 covering the area of coordinate: Lat: 10.30◦ N to 10.40◦ N; Lon: 114.34◦ E to 114.45◦ E. Both Taiping Island, Centre Cay, and surrounding bathymetry are clearly charted.

10


209

a

b

Fig. 10.11 (a) C-Map at a scale of 1:50,000, and (b) overlay with the ERS-2 SAR image (11 km × 11 km) collected on December 21, 2005, over Taiping Island in the South China Sea with coordinates from Lat: 10.30◦ N to 10.40◦ N; Lon: 114.34◦ E to 114.45◦ E. Notice that the island location has a shift of 500 m to the south as compared with the C-Map

210

A.K. Liu

Figure 10.11b is the overlay of C-Map on the SAR image (11 × 11 km) (Hsu et al., 2008). Notice that the location of the island in SAR image is shifted about 500 m to the south from the location in C-Map. The accuracy of this mapping technique is limited only by the persistence of the features and by the spatial resolution and navigational accuracy of satellite data. This error is consistent with the GPS survey result compared with the commercially used navigation charts (TWD67, WGS84). Besides Taiping Island, more islands and reefs have also been studied, and this kind of mismatch is typical between satellite images and navigational charts (Liu et al., 2007). The pixel size is 12.5 m and the resolution of SAR is 25 m. The surface signature of island size and reef area could be affected by the tide period, wind condition, and wave/current interaction around the bathymetry features. Taiping Island and the Centre Cay with coral reef around the island can be clearly identified at the upper part of image. Ocean waves refracted by the island and reef can also be observed easily. In a closer look, a boat (white dot) with oil slick (dark patches) are seen near the Taiping Island. With repeated coverage, spaceborne high-resolution SAR instruments provide the most efficient means to monitor and study the changes in important elements of the marine environment (Liu et al., 2006). The accuracy of marine charts is critical to the ship navigation, especially in a little used area that has never been systematically charted. Recent surveys using advanced technology, such as GPS navigation and multi-beam acoustic swath mapping systems, can definite improve the bathymetry map for all oceans. But in SCS, because of pirates and political issues the satellite imaging with high-resolution sensors could be a very powerful tool for mapping and monitoring navigation dangerous zones, especially in the denied or remote areas. Continuous monitoring of the change of bathymetry is critical to ship navigation, especially in the areas where water is shallow and ship traffic is heavy. A detailed study on bathymetry and sea bottom material based on SAR, SPOT images and ship survey in the Spratly Islands has been reported by Liu et al. (2007).

7 Pollution and Hazard Protection In a recent oil spill in the South China Sea, a tanker sank in rough seas on August 11, 2006 off the coast of Guimaras Island, about 312 miles southeast of Manila as shown in Fig. 10.12a. About 528,000 gallons of industrial fuel was leaking from the accident. A central Philippine island province declared a “state of calamity” following what authorities called the country’s worst spill. Faced with a potential “environmental catastrophe,” the Philippine coast guard called for a national mobilization of resources to mitigate the impact of the large amount of leaking fuel. After the accident, the oil spill had extended and covered a very large area. Lots of coastal areas were affected by the spill from about 17 to as much as 50 nm northeast of spill. However, six months after the oil spill in February 2007, the coastal areas are almost recovered with cleaning efforts (Liu et al., 2007).

10


211

a

b

Fig. 10.12 (a) Location of Guimara oil spill and polluted area; (b) ENVISAT ASAR image collected on August 24, 2006 over the Guimaras Island, Philippines. The location where the ship sank and the extent of the spill can be clearly identified to the south-southeast of the Guimaras Island as indicated by arrows

SAR images of the polluted area would be helpful to map out the extent of spill. For demonstration, Fig. 10.12b shows an ENVISAT ASAR image collected on August 24, 2006. Guimaras Island is the small island between two big islands, Panay and Negros on the right-hand side of image. The location where the ship

212

A.K. Liu

sank and the extent of the spill can be clearly identified to the south-southeast of the Guimaras Island as indicated by arrow. It has been observed that the sunk tanker was still leaking oil after a week. As shown in this SAR image, the oil spills are still visible after almost 2 weeks from the accident. Therefore, satellite-based SAR can be a very useful tool for environment monitoring, especially for pollution control and hazard protection. As an example to track/monitor, oil spills near Point Barrow, Alaska in November 1997 has been captured by RADARSAT ScanSAR (Liu et al., 2000a). The oil spill may have originated from a vessel that appears in the image on November 2, and the oil slick has been delineated by wavelet transform as an edge detector in Fig. 10.5a. The shape of oil slick is well-defined, apparently as the result of low wind and calm

a

b

c

d

Fig. 10.13 RADARSAT ScanSAR subscenes collected over Point Barrow, Alaska on November (a) 2, (b) 3, (c) 9, 1997, and (d) a map of Point Barrow and its vicinity to summarize the approximated locations of oil slicks and their tracks as they drifted further offshore due to the coastal wind

10


213

sea. On November 3, the wind had obviously become stronger with wind streaks as shown in Fig. 10.13b, and the oil slicks had broken into a series of elongated patches. Six days later on November 9, the oil slicks had drifted further offshore toward northwest and become two elongated patches as shown in Fig. 10.13c. Figure 10.13d shows a map of Point Barrow and its vicinity to summarize the approximated locations of oil slicks and their tracks as they drifted further offshore due to the coastal wind. The wind direction is the same direction of wind streaks as observed in the SAR images. An automated ocean feature detection, extraction and classification scheme for SAR imagery has been developed by Wu and Liu (2003) based on wavelet analysis. Only linear ocean and ice features in SAR imagery was attempted and demonstrated for the purpose of automated screening and extraction. The algorithms for the automatic detection of oil spills in SAR images have been developed by Solberg et al. (1999). The developed framework consists of first detecting dark spots in the image, then computing a set of lookalike features for classification based on statistical modeling. In addition, knowledge about wind field, sea states, and slick surroundings should be taken into account. Further development and improvement by incorporating and utilizing prior knowledge about the behavior of dark features may further increase the accuracy for operational use.

8 Discussion SAR has the unique capability of operating in day or night and under all weather conditions. With repeated coverage, spaceborne SAR instruments provide the most efficient means to monitor and study the changes in important elements of the marine environment. As demonstrated in this study, the use of SAR-derived observations to track eddies, fronts, ice edges and oil slicks can supply valuable information and can aid in the management of the fishing industry and the protection of environment. In overcast coastal areas in the tropical and subtropical regions, the uniformly warm sea surface temperature and persistent cloud cover preclude optical and infrared measurement of surface temperature features, and obscure ocean color observations. The mapping of ocean features by SAR in these challenging coastal regions is, therefore, a potentially major application for satellite-based SAR, particularly for the wider-swath ScanSAR mode. Therefore, satellite data provide unique information for studying the health of the Earth system, as well as critical data for natural hazards and resource assessments, especially for coastal monitoring. However, the temporal coverage (repeated cycles) of SAR images are usually low. Thus sequential satellite images from a single polar-orbiting SAR sensor may not be used to track ocean surface features with short coherent time periods. Currently, there are two SAR sensors on different satellites, ERS-2 and ENVISAT, having almost the exactly same path and an acquisition time difference about 28 min, that is a time period within the coherent time periods of ocean mesoscale feature. The study by Liu et al. (2006) and (Liu and Hsu, 2009) has demonstrated that a pair of ERS-2 and ENVISAT SAR images collected approximately 28 min

214

A.K. Liu

apart over same locations can be jointly used in deriving ocean surface drift by the wavelet-based tracking method. Also, this multi-sensor approach is expected to be useful in other applications on feature tracking for the combination of various satellite sensors with similar resolution. The prospect of satellite data collection extending well into the next century gives impetus to current research in satellite applications in ocean science and opens the doors to change detection studies on decadal time scales, especially using SAR (Liu et al., 2008). The next step is to move into the operational use of satellite-based SAR data in near-real-time to complement ground measurements. The challenge is to increase cooperation in the scheduling, processing, dissemination, and pricing of SAR data from all SAR satellites between international space agencies. Such cooperation might permit near-real-time high-resolution coastal SAR measurements of sufficient temporal and spatial coverage to impact ocean monitoring for selected heavily populated coastal regions. SAR imagery is a useful tool to locate oceanic features over extensive areas in coastal oceans and therefore to aid in the management of environment and resources. It is necessary to bear in mind that each satellite image is a snapshot and can be complemented with buoy and ship measurements. Ultimately, these data sets should be integrated by numerical models. Such validated and calibrated models will prove extremely useful in understanding a wide variety of ocean coastal processes. Acknowledgments The author would like to thank Prof. Ming-Kuang Hsu and Dr. Yunhe Zhao for providing assistance to this study, and Prof. Ming-An Lee of the National Taiwan Ocean University (NTOU) for providing Fig. 10.8. This work was supported by the U.S. Office of Naval Research (ONR) and National Aeronautics and Space Administration (NASA). The author is a Principal Investigator on ESA and CSA projects; all ERS-2 SAR and ENVISAT ASAR data are copyrighted by ESA, and RADARSAT data is copyrighted by CSA. The author is now a Visiting Professor of NTOU, and the support from Taiwan’s National Science Council (NSC) is also highly appreciated.

References Beal RC, Pichel WG (eds) (2000) Coastal and marine applications of wide swath SAR. Johns Hopkins APL Tech Digest 21:176 pp Fu L, Holt B (1982) Seasat views oceans and sea ice with synthetic aperture radar. JPL Publication 81–120, Pasadena, CA, 200 pp Hsu M-K, Liu AK, Zhao Y, Hotta K (2008) Satellite remote sensing of Spratly Islands using SAR. Int J Remote Sens 29:6427–6436 Kwok R (2008) Summer sea ice motion from the 18 GHz channel of AMSR-E and the exchange of sea ice between the Pacific and Atlantic sectors. Geophys Res Lett 35:L03504. doi:10.1029/2007GL032692 Lindsay R, Zhang J, Schweiger A, Steele M, Stern H (2008) Arctic sea ice retreat in 2007 follows thinning trend. J Clim. doi:10.1175/2008JCLI2521.1 Liu AK, Cavalieri DJ (1998) Sea-ice drift from wavelet analysis of DMSP SSM/I data. Int J Remote Sens 19:1415–1423 Liu AK, Ho C-R, Liu C-T (2008) Satellite remote sensing of South China sea. Tingmao, Taipei, 312 pp Liu AK, Hotta K, Hsu M-K (2007) Satellite remote sensing of Spratly Islands. Tingmao, Taipei, 97 pp

10


215

Liu AK, Hsu M-K (2009) Deriving ocean surface drift using multiple SAR sensors. Remote Sens 1:266–277. doi:10.3390/rs1030266 Liu AK, Martin S, Kwok R (1997b) Tracking of ice edge and ice floes by wavelet analysis of SAR images. J Atmos Oceanic Tech 14:1187–1198 Liu AK, Peng CY, Chang SY-S (1997a) Wavelet analysis of satellite images for coastal watch. IEEE J Oceanic Eng 22(1):9–17 Liu AK, Peng CY, Chang Y-S (1996) Mystery ship detected in SAR image. EOS Trans AGU 77(3):17–18 Liu AK, Peng CY, Schumacher JD (1994a) Wave-current interaction study in the Gulf of Alaska for detection of eddies by SAR. J Geophys Res 99:10075–10085 Liu AK, Peng CY, Weingartner TJ (1994b) Ocean-ice interaction in the marginal ice zone using SAR. J Geophys Res 99:22391–22400 Liu AK, Wu SY (2001) Satellite remote sensing: SAR. In: Steele JH, Thorpe SA, Turekian KK (eds) Encyclopedia of ocean sciences. Academic, London, pp 2563–2573 Liu AK, Wu SY, Tseng WY, Pichel WG (2000a) Wavelet analysis of SAR images for coastal monitoring. Can J Remote Sens 26:494–499 Liu AK, Wu SY, Zhao Y (2003) Wavelet analysis of satellite images in ocean applications. In: Chen CH (ed) Frontiers of remote sensing information processing, Chap 7. World scientific, Singapore, pp 141–162 Liu AK, Zhao Y, Esaias WE, Campbell JW, Moore T (2000b) Ocean surface layer drift revealed by satellite data. EOS Trans AGU 83:61–64 Liu AK, Zhao Y, Hsu M-K (2006) Ocean surface drift revealed by tracking SAR images. EOS Trans AGU 87:233, 239 Liu AK, Zhao Y, Wu SY (1999) Arctic sea ice drift from wavelet analysis of NSCAT and SSM/I data. J Geophys Res 104:11529–11538 Montgomery DR (2000) International fisheries enforcement management using wide swath SAR. Johns Hopkins APL Tech Digest 21:141–147 Ogi M, Rigor IG, McPhee MG, Wallace JM (2008) Summer retreat of Arctic sea ice: Role of summer winds. Geophys Res Lett 35: L24701. doi:10.1029/2008GL035672 Pichel WG, Clemente-Colon P, Wackerman CC, Friedman KS (2004) Ship and wake detection. In: Jackson CR, Apel JR (eds) Synthetic aperture radar marine user’s manual, NOAA, Washington, DC, Chap 12, pp 277–303 Schumacher JD, Barber WE, Holt B, Liu AK (1991) Satellite observations of mesoscale features in Lower Cook Inlet and Shelikof Strait, Gulf of Alaska. NOAA Technical Report ERL 445 PMEL 40, 18 pp Schumacher JD, Kendall AW Jr (1995) An example of fisheries oceanography: walleye pollock in Alaska waters. Rev Geophys 33:1153–1163 Solberg AHS, Storvik G, Solberg R, Volden E (1999) Automatic detection of oil spills in ERS SAR images. IEEE Trans Geosci Remote Sens 37(4):1916–1924 Wu SY, Liu AK (2003) Toward an automated ocean feature detection, extraction and classification algorithm for SAR imagery. Int J Remote Sens 24:935–951 Wu SY, Liu AK, Leonard GH, Pichel WG (2000) Ocean feature monitoring with wide swath SAR. Johns Hopkins APL Tech Digest 21:122–129 Yu J, Liu AK, Zhao Y (2006) Sea ice motion and deformation in the marginal ice zone through SAR. In: Ip W-P (ed) Advanced in geosciences, vol 5. World Scientific, Singapore, pp 41–47 Yu J, Yang Y, Liu AK, Zhao Y (2009) Dynamics of wave and ice interaction in the marginal ice zone of the Bering Sea. Int J Remote Sens 30:3603–3611 Zhao Y, Liu AK (2002) Validation of sea ice motion from QuikSCAT with those from SSM/I and buoy. IEEE Trans Geosci Remote Sens 40:1241–1246 Zhao Y, Liu AK (2007) Interaction of arctic sea-ice drift and atmospheric surface pressure. J Oceanogr 63:505–515

Chapter 11

Satellite Altimetry: Sailing Closer to the Coast Stefano Vignudelli, Paolo Cipollini, Christine Gommenginger, Scott Gleason, Helen M. Snaith, Henrique Coelho, M. Joana Fernandes, Clara Lázaro, Alexandra L. Nunes, Jesus Gómez-Enri, Cristina Martin-Puig, Philip Woodworth, Salvatore Dinardo, and Jérôme Benveniste

Abstract In this chapter we review the history of coastal altimetry. We illustrate the challenges associated with data processing, improvement and exploitation, including: (1) what altimeter data are available today and what are the issues in coastal zones; (2) what efforts are underway to fill the gaps in coastal altimetry and what still needs to be done; (3) how coastal altimetry can be used in support of coastal oceanography. After nearly two decades of data collection near coasts, the planned reprocessing of the multi-mission global record now appears to be necessary for full exploitation of satellite altimetry for coastal oceanography. We will focus on the European research efforts, in particular the main outcomes of the COASTALT project, by showcasing improved corrections (with special emphasis on the wet tropospheric effect), waveform analysis and novel retracking techniques, as well as the structure of the new processor for Envisat RA-2 coastal records. This is of interest to a broad range of data integrators who will be able to use the improved altimeter data in their operational products or services. Keywords ALBICOCCA · ALTICORE · Altimeter data corrections · Climate change · Coastal currents · Coastal zone altimetry · COASTALT · Delay-Doppler · Digital elevation model · DORIS · Envisat · ERS-1 · ERS-2 · Geoid · Global navigation satellite system · GNSS-derived Path Delay · GNSS · Jason-1 · Jason-2 · Mediterranean Sea · Microwave radiometer (MWR) · Microwave radiometry · Ocean currents · PISTACH · Retracking · River and lake level · SAR mode · Satellite radar altimetry · Sea surface topography · Sea level · Sea level anomaly · Sea surface height · Tide gauge · Tides · TOPEX/Poseidon · Validation · Water vapour · Wave height · Waveforms · Wet tropospheric correction · Wind speed

S. Vignudelli (B) Consiglio Nazionale delle Ricerche, Pisa, Italy e-mail: [email protected]


217

218

S. Vignudelli et al.

1 Introduction The coastal zone is the unusual part of the Earth where land, sea and air come together. Demands for information on the status of coastal waters are rising in response to pressure from population growth and climate change. However, the situation is more complex than for open oceans in terms of monitoring requirements. Only through the combination of modelling tools and multiple data sets created from space, air, land and ocean-based Earth Observation systems, can the gaps, in space and time, be filled and a thorough and complete characterization of the coastal environmental changes obtained. Right now, there are several environmental satellites providing one of the most extensive and most continuous archives ever available, and interoperability standards are removing the barriers to information flow that have traditionally separated disciplines and domains. Most of the current satellites orbiting in space and monitoring the marine environment owe their existence to open ocean needs (e.g., altimetry, scatterometry, etc.). Their use in the coastal zone often requires the development of new retrieval algorithms, improved corrections, reprocessing, etc. before taking full benefit from the data.

2 Satellite Altimetry in the Open Ocean Radar altimetry is a tool primarily designed for the measurement of ocean surface topography globally from space and for repeatedly monitoring its change (sea level rise/fall). Beginning with the launch of the ERS-1 and TOPEX/Poseidon (T/P) satellites in 1991 and 1992 and continuing with the ERS-2, Geosat Follow-On, Envisat and Jason-1/2 missions, satellite altimetry has proved successful as a global tool for monitoring sea level (see also the chapter by Katzaros et al. in this book) as well as the extent of ice caps. Whilst several of these missions are still operating and are expected to continue in operation in the future, new missions are planned for launch by space agencies over the next few years (CryoSat-2, AltiKa, Sentinel3, HY-2). Satellite altimetry has had exceptional success over the open ocean, the domain for which it was originally designed. In this endeavour, the unique combination of day/night and all weather operation and global coverage means that satellite altimetry is a key component of the Global Earth Observation System of Systems (GEOSS).

3 Satellite Altimetry in Coastal Zones The coastal zone is one of the new frontiers for satellite altimetry. It is primarily in this zone that the effects of rising sea levels, storm surges and changing coastal ocean dynamics are impacting human activities and coastal ecosystems. Any observing system that can deliver useful information in this crucial region must be exploited, even more so if it allows reprocessing of long time series of archived

11

Satellite Altimetry: Sailing Closer to the Coast

219

data and helps to establish long-term trends and climatology. Such is the case for altimetry, a success story over the open ocean (Fu and Cazenave, 2001), but still profoundly under-exploited in coastal areas, where the processing strategy used in the open ocean has not been of much success in retrieving accurate information. Processing and data correction issues have so far resulted in systematic flagging and rejection of 17 years of data. This limitation is due to the complicated conditions encountered in coastal zones, including the proximity to land and the influence of the sea bed on coastal dynamics, making it difficult to directly extract useable information from the altimeter waveforms. The processing of satellite data is generally challenging in the immediate vicinity of coastal areas; in the case of satellite altimetry, data retrieval must address several problems: (1) retracking (important within 10 km of the coast), (2) more accurate wet troposphere path delay correction, (3) better modelling of tidal and atmospheric effects. The advantage of current radar altimetry for coastal studies is that it can fill gaps in the vast areas around tide gauges, which are running continuously, but in only a limited number of places. Future missions will be designed with higher resolution capabilities than their predecessors, benefiting from advances in technology (e.g. delay-Doppler and Interferometry) in concert with the multiple orbital configurations that are emerging and, possibly, miniaturised instruments on constellations of satellites.

4 History of Coastal Altimetry The first attempt to retrieve near-shore altimetric data by customised processing was that of Manzella et al. (1997), who recomputed the wet tropospheric correction for ERS-1 altimeter data over the Corsica Channel by recalibrating the model correction with the closest available radiometric estimate; they also recovered measurements that had been flagged as bad due to the sigma-0 value being high (in turn due to a smooth sea surface). Crout (1998) reviewed the potential of T/P over the coastal ocean, using both 1 Hz and 10 Hz data but achieved no improvement in correction and noticed that, over a flat coastal topography, a useful signal is retrievable closer to shore than in the case of rough terrain. An extensive study of coastal altimetry, taking into account all issues in reprocessing, was carried out by Anzenhofer et al. (1999). They described the generation of coastal altimeter data and were the first to analyse in detail various retracking algorithms and their implementation. They then presented some examples based on ERS waveform data, and concluded with some recommendations for better (local) tidal modelling, careful screening of data, improvement of the wet tropospheric correction and retracking. The idea of customised tidal modelling was followed by Vignudelli et al. (2000) in another study over the Corsica Channel, this time using 1 Hz T/P data in combination with current meter and tide gauge data. Results were encouraging, showing that with simple improvements in processing, the signal recovered at seasonal time scales was in good agreement with in situ measurements and permitted useful oceanographic conclusions. This paper prompted the joint

220


French-Italian ALBICOCCA (ALtimeter-Based Investigations in COrsica, Capraia and Contiguous Areas) initiative for coastal altimetry; one of its outcomes was the generation of a coastal product in the Northwest Mediterranean (Vignudelli et al., 2006) by adopting ad hoc data filtering and screening techniques (Roblou and Lyard, 2004) in combination with state of the art tidal and atmospheric modelling (Lyard and Roblou, 2003). This allows better characterisation of the variability at seasonal scales, with the findings outlined in Vignudelli et al. (2005). The development work has been evolving within the ALTICORE (value added satellite ALTImetry for COastal REgions) initiative (Lebedev et al., 2007). The overall aim of ALTICORE was to build up capacity for provision of altimeter-based information in support of coastal ocean studies in European Seas (Mediterranean, Black, Caspian, White and Barents). Several other studies have dealt with the limitations of, and possible improvements to, coastal altimetry in recent years, including Brooks et al. (1998), Deng et al. (2002a), Dong et al. (2002), Fernandes et al. (2003) and Liebsch et al. (2002). The topic of coastal altimetry was also well represented by many contributions at the 2006 Venice symposium on “15 years of Progress in Radar Altimetry” (see the papers by Morrow et al., Freeman and Berry, Birol et al., Bouffard et al., Cipollini et al., Madsen and Hoyer, Han and Li, Mathers and Fernandes, Bonnefond et al., etc. in the symposium proceedings (Benveniste and Ménard, 2006)) and was the subject of a dedicated International Workshop in Beijing in July 2006. A very detailed assessment of the state of the art for post-processing coastal altimetry (i.e. altimetry that does not rely on retracking, but only on the improvement of the Geophysical Data Records (GDRs) and/or on using higher rate GDR data) is given in J. Bouffard’s PhD thesis (Bouffard, 2007). Comparisons with a regional circulation model (Bouffard et al., 2008a) highlight improvements with respect to the standard altimeter product. Moreover, the use of multi-mission data, illustrated over the Corsican Channel area in Bouffard et al. (2007a, 2007b, 2008b), allows some degree of monitoring of signals at a sub-seasonal scale. From several of the studies mentioned above, a reasonable consensus emerges that further improvements of near-shore altimetric records will have to rely not only on post-processing of higher rate data, but more importantly, on preprocessing, i.e. retracking of the waveforms or use of retracked higher rate data; one example of this for ERS-2 data has recently been proposed by Deng and Featherstone (2006). Another example, by Yi et al. (2006), successfully uses Ku-band T/P data, with various retrackers, to measure water level changes over the Louisiana wetlands. Quesney et al. (2007) are proposing new waveform models based on neural networks for the retracking of coastal waveforms. Before applying the retracking procedure, some filtering techniques can also be applied to reduce the noise in the raw waveform. In Ollivier et al. (2005), a method of noise reduction (based on a Singular Value Decomposition technique) has been applied to improve the results of a mean-square fitting of radar altimeter echoes. The oceanic parameters extracted by this method show a smaller standard deviation, which enables physical analyses of the sea surface with sampling on the satellite track of 350 m instead of the current 7 km (1 Hz); this improvement is crucial for application to the coastal zone. A set

11


221

of 11 different retracking algorithms, covering a comprehensive range of waveform shapes, corresponding to diverse contributions by ocean/land/ice to echoes and accounting for different states of the water surface, has been developed in recent years at De Montfort University (DMU). The technique employs a rule-based expert system, which identifies each echo shape and assigns it to one of the eleven different retrackers. This has been applied to the ERS-1 Geodetic Mission (GM) to monitor global river and lake height (Berry, 2002; Berry et al., 2005; Freeman and Berry, 2006), and then extended to encompass Envisat data (Benveniste and Berry, 2004; Garlick et al., 2005). These results demonstrate that altimetry can be used operationally to measure and monitor global land surface hydrology (river and lakes). Retracking has also been used for geodetic purposes, i.e. for gravity field retrieval in coastal areas. The coastal complications in deriving the marine gravity field from satellite altimetry were first examined by Andersen and Knudsen (2000); retracking was used over the Taiwan straits by Deng et al. (2002b), and recently by Andersen et al. (2005) using ERS-1 GM mission data. Finally, we ought to mention the RAIES Project (Gommenginger et al., 2005), an ESA-funded study for the scientific exploitation of Envisat RA-2 Individual Echo and S-band data for ocean, coastal, land and ice applications; one of the RAIES outcomes has been an updated retracker for Envisat RA-2 data (Gómez-Enri et al., 2006, 2007). Recent advances in improving corrections combined with developments in data processing make it possible to increase the quality and quantity of data from coastal zones (Cipollini et al., 2008). By improving the processing chain a larger number of altimeter ground points otherwise flagged as unusable can be successfully retrieved even from areas very close to the coast. As an example, some T/P tracks were selected along the coasts of the North West Mediterranean region. When compared to the standard AVISO products (Fig. 11.1), the improved processing provides data nearer to coasts and even fills gaps that are present in the open sea. A great impetus has been given to the field by the recent launch of two major projects devoted to the development of coastal altimetry products for specific missions: PISTACH, funded by the French Space Agency (CNES) concerning coastal altimetry processing for Jason-1 and Jason-2 (Mercier, 2008); and COASTALT (www.coastalt.eu), funded by the European Space Agency (ESA) to design and implement a prototype coastal altimetry processor for Envisat. In parallel, NASA is recognising the importance of the topic and is sustaining coastal altimetry research through specific R&D projects in response to the last OSTST call in 2008. It is now clear that, by overcoming the technological problems and extending the capabilities of current and future altimeters to coastal zones, altimeter derived measurements of sea level, wind speed and significant wave height can play an important role in coastal ocean observing systems (Cipollini et al., 2010). The new “coastal altimetry” community, which is inherently interdisciplinary, has already held three well-attended international workshops in Silver Spring (Smith et al., 2008), Pisa (Benveniste and Vignudelli, 2009) and Frascati (Vignudelli and Benveniste, 2010). The community is also contributing to a topical book with review chapters on specific topics related to coastal altimetry, e.g. retracking (Gommenginger et al., 2011), wet troposphere (Obligis et al., 2011), tides (Ray et al., 2011), atmospheric effects (Woodworth et al., 2011) and

222


Fig. 11.1 Comparison between the number of valid TOPEX/Poseidon ground points obtained along the coasts of NW Mediterranean when using standard AVISO processing (left panel) and improved ALBICOCCA/ALTICORE processing (right panel)

technological prospects (Raney and Phalippou, 2011) as well as chapters showing case studies of applications in regional seas (e.g. Mediterranean, Black, Caspian, White and Barents Sea) and around the coasts of the US, China and Australia.

5 The COASTALT Initiative COASTALT is essentially a scientific study leading to experimental Envisat RA-2 products over a few pilot coastal areas surrounding Europe. However the techniques developed under the COASTALT framework are a first step towards the complete reprocessing of Envisat coastal altimetry products globally. It also prepares the way for the exploitation of data from future altimetry missions such as CryoSat-2 and Sentinel-3. These missions will have inherently improved coastal zone capabilities by virtue of the adoption of a delay-Doppler instrument, also known as a SAR altimeter (Raney, 1998). For the pilot coastal areas, COASTALT is producing a coastal data set with an along-track resolution of around 350 m over a period from 2002 until now.

5.1 User Requirements for Coastal Altimetry The initial phase of the project consisted of an evaluation of the requirements for new coastal products by means of a survey within the community of potential

11


223

users. This provided valuable recommendations on: (1) Desired quantities – there is widespread interest not only in sea surface height (SSH), but also in significant wave height (SWH) and to a lesser extent in wind speed; (2) Domain and posting rate – users asked for along-track data as close as possible to the coast, with the maximum posting rate compatible with an acceptable signal-to-noise ratio, which may depend on the application. This, in practice, implies that data should be processed at the maximum posting rate available for standard raw altimetric waveforms (18 Hz for Envisat RA-2), leaving to the user to decide the level of averaging most suited to their particular application; (3) Level of detail – a significant number of users asked for the fields to be provided with individual corrections (HF dynamics for example) to facilitate their use in synergy with 2D and 3D models, plus quality flags, error budget and some clear documentation on the characteristics and limitations of products; (4) Data formats – NetCDF being the preferred format. Despite COASTALT being an experimental (i.e. not operational) project, the above recommendations have helped to design a prototype product already in line with the needs of users, facilitating its future extension and further development.

5.2 Assessment of Geophysical Corrections The objective of expanding the use of altimetry in coastal areas is only possible if the magnitudes and scales of the various correction terms are understood. Consequently, it is clear that one must always judge how effective the corrections terms may be in determining the physical signals of interest in coastal areas, and if it is possible to tolerate these uncertainties in an analysis. Nevertheless, while some coastal studies are already possible, our objective is to expand the range of possible applications of altimeter data. That can only be done by reducing the uncertainties of the various correction terms. In the current state of the art, the dry tropospheric correction is less critical than other corrections, however this does not imply that the dry tropospheric correction is a closed issue. The problem of computing this correction becomes the problem of securing the best possible measurements of sea level pressure. With the current widespread use of the ECMWF (European Centre for Medium-Range Weather Forecasts) reanalysis fields at 1/4◦ and 6 h resolution, the accuracy of this correction is of the order of 1 cm, and remains of the same order even in coastal areas, as the spatial scales of variability of air pressure are barely affected by the ocean/land transition (contrary to what happens to water vapour in the wet tropospheric correction). Improvements are still possible, and research is welcome in future projects, for instance, to investigate shear at the land/sea interface, the Gibbs effect in the models, effects of night/day mass transport and effects of higher temporal resolution in the models. The ionospheric delay correction tries to account for the fact that the speed of a radio pulse transmitted from an altimeter is less than the velocity of light by an amount proportional to the number of free electrons in the ionosphere (Total Electron Content, TEC) and is inversely related to the square of the radio pulse frequency. The magnitude of the correction varies from day to night, from summer

224


to winter, and as a function of the solar cycle. The TEC can be estimated from comparing measurements at two frequencies. Such measurements include: (1) Dual frequency range measurements by the altimeter itself (e.g. the dual frequency instruments on board T/P, Jason-1 and -2, and Envisat); (2) Dual frequency DORIS measurements as provided for T/P, Jason-1 and -2 and Envisat. Early experience in the T/P Science Working Team demonstrated that corrections (1) and (2) appeared to be consistent at the 1 cm level in the open ocean, with an absolute accuracy for both also of the order of 1 cm, e.g., Zlotnicki (1994). If this level of accuracy were to apply also in coastal areas, then either technique would certainly be adequate for COASTALT, given the magnitude of the uncertainty of other correction terms in coastal areas. The near-global coverage of DORIS beacons means that estimates of ionospheric correction based on (2) would be available virtually worldwide, including in most coastal areas. However, more recent experience has demonstrated that (1) and (2) are less consistent during periods of high solar activity (several cm difference, R. Scharroo, private communication, 2008). The ionospheric correction can also be estimated from GPS Ionosphere Maps (GIMs), such as the ionosphere correction based on GIMs provided on Envisat GDRs. The main contribution to the sea surface height error budget in coastal zones remains that from the path delay introduced by tropospheric water vapour (wet tropospheric correction). However, in all those applications needing the removal of tides and high-frequency (HF) oceanic and atmospheric signals (due to air pressure and wind effects), large errors in tidal models and in the models used to obtain high-frequency and inverse barometer corrections remain a problem. Sea surface elevations in shallow seas and coastal areas can depart considerably from those predicted by the tide and by the “inverse barometer” model of sea level response to air pressure changes. These departures are due to the role of winds, with wind set-up proportional to wind stress divided by water depth, and to dynamical adjustments to air pressure (e.g. Ponte et al., 1991). Surges can be metres in magnitude. For the removal of tidal and atmospheric effects, the issue is the accuracy of the global models used for the corrections, as these models have limited resolution for coastal applications and are prone to large errors in coastal areas. However, there are examples of successful coastal applications linked to more accurate tidal analyses. For example, Cherniawsky et al. (2001) showed that simple along-track analyses of tides in T/P data provide much better description of coastal currents, waves and eddies (e.g., Foreman et al., 1998; Crawford et al., 2000, 2002; Cherniawsky et al., 2004), when compared to using global tidal models. Furthermore, along-track or altimeter-to-altimeter crossover analyses provide more accurate estimates of tidal constituents for use in regional and coastal data-assimilating tidal models (Foreman et al., 2000, 2006; Sutherland et al., 2005). Indeed, for wet tropospheric correction, the very first issue that one has to face is the complete absence of microwave radiometer-derived correction in a strip of a few tens of kilometres along the coast. The radiometer-derived correction is collected at a 1 Hz rate and, over the open ocean, its expected error is about 1 cm (to be consistent with the description used for the other corrections) (Obligis et al., 2005). Obviously we will use this information up to where it is available (i.e. where

11


225

land starts impinging on the radiometer footprint), at some tens of km’s from the coast. But landwards of this point, we need to either extrapolate the correction or use a model. There are basically three options: (1) Estimate the correction based on an atmospheric model (such as ECMWF), adjusting the correction values so that there is continuity with the closest valid radiometer-derived correction over open ocean (this is the so-called Dynamically Linked Model approach used by Fernandes et al., 2003; Mercier, 2004); (2) Model and hence remove the influence of land for specific coastal areas in the radiometer readings, for example using the methods described by Desportes et al. (2007); (3) Use a correction based on the combination of Global Navigation Satellite System (GNSS) derived path delays with valid microwave radiometer (MWR) measurements and ECMWF model-based wet delays (Fernandes et al., 2010). Brown (2010) is also developing a new approach for the retrieval of wet tropospheric delay on coastal areas, which is being successfully tested at NASA/JPL on Jason-2/OSTM Advanced Microwave Radiometer. The Dynamically Linked Model (DLM) is the most obvious approach and is actually implemented in the COASTALT product. Fundamentally, it relies on corrections interpolated from a large-scale atmospheric reanalysis model, and somehow linked to the last few values of the available radiometer-derived correction before land starts entering the radiometer footprint. The model of choice is without doubt the ECMWF model, which is readily available in the GDRs, and was also chosen as reference by PISTACH for the tropospheric correction. There are two strategies depending on the configuration of the track segment over which the radiometer wet tropospheric correction is missing (or flagged as bad), which we refer to as the wet tropospheric gap. The algorithms implementing these two strategies are described below. Figure 11.2 shows the two typical configurations: (1) island type or ‘doubleended’ track segment, with valid radiometer measurements on each side of the track segment. In this case, the model field is adjusted to the radiometer field, at the beginning and end of the land contaminated segment, by using a linear adjustment (using time as interpolation coordinate); (2) Continental coastline type or ‘single-ended’ track segment, where valid radiometer measurements exist only on one side of the track segment. In this case, the model field is adjusted to the radiometer field on the side where valid radiometer data exist, using a bias correction. Figure 11.3 shows an example of the application of the DLM approach along Envisat ground tracks around European coasts. The COASTALT project, however, also investigated a different approach, the GPD (GNSS-derived Path Delay) approach. The methodology, described in detail by Fernandes et al. (2010), combines ZWD (zenith wet delay or wet tropospheric correction) estimates derived from GNSS measurements, with valid measurements from the microwave radiometer and data from a numerical weather model (such as ECMWF). The methodology is based on a linear space-time objective analysis (OA) technique whereby, at each altimeter ground-track point with invalid MWR measurement, a ZWD value is estimated by merging nearby (in space and time) valid MWR data with ECMWF-derived and GNSS-derived independent estimates, while taking into account the accuracy of each data set. A quantification of the error associated with each estimate is provided simultaneously by the methodology.

226


Fig. 11.2 Typical configurations of wet tropospheric gaps near islands (left panel) and continental coastline (right panel)

It is worth pointing out that the atmospheric quantity derived from the GNSS measurements is the zenith total delay (ZTD) caused by the atmosphere. To get the ZWD (wet tropospheric correction), the ZTD has to be corrected for the zenith hydrostatic delay (ZHD or dry tropospheric correction), which can be computed with an accuracy of a few millimetres, either from in situ reliable pressure data or from global grids of a model such as ECMWF. To be used for coastal altimetry, the ZWD values must be further reduced from the GNSS station elevation to sea level elevation. The methodology has been applied in the west Iberian and Mediterranean regions for Envisat data. An example of the results obtained at a location near the Portuguese South-Western Coast (see Fig. 11.4 for location) is reported in Fig. 11.5. Wet delay estimates remain nearly constant in the light blue shaded areas (wet delay values shown on the y-axis on the left side). The OA technique thus outputs wet delay estimates (black dots) at all along-track positions, including those within the grey or blue shaded areas, providing that these data points have an altimeter land-ocean flag equal to 0. Results are extremely promising and it is recommended that this correction be further explored and possibly included in future versions of the COASTALT product. However, the GPD estimates are highly dependent on the spatial and temporal distribution of the three independent data sets. For a global implementation of the method, a densification of the coastal GNSS stations would be required (with a station every 100 km, preferably equipped with meteorological sensors).

11


227

Fig. 11.3 Example of the application of the DLM approach along Envisat ground tracks around European coasts. Blue indicates corrected points, and red indicates uncorrected points

Fig. 11.4 Case study of application of GNSS-derived path delay method to obtain the wet tropospheric correction near the Portuguese South-Western Coast. On the left panel, blue and red dots represent valid and invalid Envisat Microwave Radiometer wet delay values and the cyan circle indicates the LAGO GNSS station. On the right panel, the blue dots represent model observations and the black triangle shows the location where the zenith wet delay (ZWD) is estimated

228


Fig. 11.5 Comparison of wet tropospheric corrections (y-axis left) along the Envisat track selected in Fig. 11.4 computed using ECMWF model (blue dots), Microwave Radiometer (red dots) and GNSS-derived Path Delay (black dots). Grey boxes indicate radiometer data flagged as “land” and blue boxes indicate radiometer data flagged as “bad” according to standard data quality flags (y-axis, right)

5.3 RA-2 Envisat Waveform Analysis and Retracking A comprehensive and systematic analysis of the return waveforms in the coastal zone was made as part of the COASTALT project. The main objective was a local analysis of Envisat RA-2 waveforms as the satellite approaches the coast and islands (Gómez-Enri et al., 2009, 2010). The joint use of Digital Elevation Model (DEM), coastline masks and Landsat images helped to assess waveforms at land-to-ocean and ocean-to-land transitions. We used a 3-arc second DEM product (90 m horizontal resolution) from the US Geological Survey based on SRTM (Shuttle Radar Topographic Mission) data. Note that it is possible to get higher resolution DEM for some areas, with horizontal accuracy around 10 m. It is a matter of debate what level of accuracy is really necessary for coastal altimetry. We present here some examples of return waveform analysis from the North Western (NW) Mediterranean Sea (Fig. 11.6). In this analysis, we show the inconsistencies between the water depth field and the land/ocean flag available in the GDR product at 1 Hz. Thus, negative depths (denoting ocean) were found in positions, which the land/ocean flag declared as ‘land’. Figure 11.6 shows two examples of this in the NW Mediterranean. The upper-left figure is a segment of the descending orbit 00022, crossing the coast of Tuscany near Cecina (Italy) and then Elba Island. The 1 Hz measurement analysed is declared as ‘land’ in the GDR, but a visual inspection of the shape of the

11


229

Fig. 11.6 Two examples of bad categorization of 1 Hz measurements. Upper left figure: Track segment of descending orbit 00022 over Italy showing the position of the 1 Hz measurement. Upper right figure: Shape of twenty 18 Hz waveforms corresponding to the first 1 Hz measurement over ocean. Lower-left: Track segment of an ascending orbit over France. Lower right: Shape of twenty 18 Hz waveforms corresponding to the last 1 Hz measurement over ocean

18 Hz Ku-band waveforms (20 subplots in upper-right figure) shows a number of ocean-like waveforms. The same can be observed in the ascending orbit over Port-Saint-Louis-du-Rhône (France) (lower-left figure) with the 1 Hz measurement analysed being considered as ‘land’. Again, the shape of the twenty 18 Hz Ku-band waveforms (lower right figure) shows that many exhibit ocean-like features, but with clear signs of land contamination. Three physically-based waveform retrackers are implemented and run in parallel in the COASTALT processor. The first one is the well-known conventional Brown ocean waveform retracker (Brown, 1977) which works well for altimeter waveforms

230


over the open ocean and (typically) up to ~10 km from the coast. The second one is the specular Beta-parameter retracker with an exponential trailing edge (Deng and Featherstone, 2006). This functional form is well suited to fit waveforms with a rapidly decaying trailing edge. The algorithm fits a 5- or 9-parameter function to pick up specular returns from one or two scattering surfaces. Finally, the third retracker is an experimental mixed Brown-Specular retracker, which aims to address the retracking of coastal waveforms which exhibit a specular peak embedded within a Brown-type ocean waveform. Such highly variable waveform shapes were observed frequently during the analysis of waveforms in the coastal zone. As an example, Fig. 11.7 focuses on a selected Envisat descending track (pass 0160) crossing Great Britain and featuring several ocean/land and land/ocean transitions. Figure 11.8 shows a 2D representation of the altimeter waveforms along that segment at Ku-band. We note the rapid transition from stable Brown-like ocean waveforms to a region of highly variable waveforms over land and a slower return to Brown-like echoes. We observe the complete loss of a leading edge around latitude 52.6◦ N in all cycles for both Ku- and S-band (Fig. 11.8, upper panel). This corresponds to a section of the track where the altimeter travels parallel to the coast. The corresponding waveforms fitted using the COASTALT Brown retracker are presented in Fig. 11.8 (lower panel) and show how the Brown model fits the waveforms well over the ocean, but not close to ocean/land and land/ocean transitions.

Fig. 11.7 Example showing Envisat descending track #0080 (pass 0160) crossing Great Britain. Waveform data shown in Fig. 11.8 are extracted along the red segment

11


231

Fig. 11.8 Measured waveforms (Ku-band) along the red segment in Fig. 11.7 (upper panel) and the corresponding fitted waveforms using the COASTALT Brown retracker (lower panel). X-axis is along track, y-axis is gate number and colour scale represents waveform amplitude

5.4 The COASTALT Processor The core of COASTALT is the design and coding of the prototype software processor, i.e. the software code that performs the retracking with the three models described previously, and generates the improved coastal altimetry products (which we refer to as the Coastal Geophysical Data Records, or CGDRs). This code also generates some improved corrections. The processor consists of a suite of programs, modules and functions written in Fortran and C. An important aspect of the

232


COASTALT processor, which differentiates it from other altimeter processing tools, is its flexibility: users can incorporate their local corrections and test new retrackers. This approach takes into account the needs for a global product (with some corrections computed globally) and the need of users for customisable regional coastal products (given that the best corrections are often only available locally). In more detail, the COASTALT processor consists of two functional units, which are both run as stand-alone applications: (1) A baseline module where the processing options are controlled by the user at run-time through an editable configuration file. The baseline processor components, interfaces and data flow are shown in Fig. 11.9. The Main Processing Loop (MPL) controls all other system blocks. The information and flags in the configuration file, read in immediately after the start of execution by the Init module, determine which blocks are called. The MPL reads through the entire SGDR data file, processing the individual entries if they belong to the coastal region mask. This mask is customisable, or can be switched off altogether, by the user. Then, for each waveform in both Ku- and S-band, the MPL calls three different waveform retrackers (one based on the Brown waveform model, one based on a specular waveform shape, and a mixed Brown + specular retracker). Subsequently – and for each retracker output – the processor computes the new 18 Hz corrections (e.g. ionosphere) and the wet tropospheric correction with the DLM algorithm. Finally, the processor outputs the retracked parameters from all three retrackers and all corrections to the CGDR NetCDF file. (2) An optional User-defined Coastal Geophysical Corrections (UCGC) module. The UCGC module is provided as a stand-alone program for users interested in including their own user-defined geophysical corrections in the COASTALT

Fig. 11.9 A snapshot showing the logical framework of the COASTALT Baseline Processor

11


233

product. Examples of such user-defined corrections include: tidal corrections from regional models, GPS-derived ionospheric corrections, land emissivitybased wet tropospheric corrections. The COASTALT processor reads data from the ESA Envisat RA-2/MWR SGDR products and reprocesses the waveforms using three retrackers, to generate new high-resolution (18 Hz) fields. Figure 11.9 shows the software logical model of the baseline processor, which illustrates the data flow for the main fields extracted from the SGDR file and their use in the processing. The processor also generates new geophysical corrections from these new retracked data, as well as higher rate geophysical corrections by interpolation, as are necessary to correct the higher rate range data. The output product contains all the relevant original and new fields, in a single file per pass, in a self-describing NetCDF format. The baseline COASTALT product includes fields that can be computed in any region, based on data from the altimeter itself or from instruments mounted on the same platform, and from global models. This baseline product does not include fields that require region-specific local information, such as a local tidal model or in situ observations. However, such additional region-specific fields may be added to the product by the user, using the standalone UCGC product enhancer module. The output product has been designed to provide access to the original and the retracked values of range, significant wave height and backscatter, together with the geophysical corrections that rely on them (such as the ionospheric correction). The output product contains the original data from the SGDR, to enable users to readily compare the SGDR and COASTALT values. One enhancement of the source data is the provision of all original 1 Hz geophysical corrections at the higher (18 Hz) data rate, by interpolation of the original 1 Hz values. A coastal mask identifies those data segments in the SGDR file, which require coastal processing. The COASTALT mask is global and time independent. It consists of a 0.1 by 0.1 deg latitude and longitude bitmap based on GSHHS (Global Self-consistent, Hierarchical, High resolution Shoreline Database; see http://www.soest.hawaii.edu/wessel/gshhs). Its spatial coverage is chosen to be deliberately conservative to ensure that all data in the coastal zone are included at all times (i.e. accounting also for temporal changes in coastline position and water depth due to e.g. tides). The coastal mask identifies coastal data segments in the SGDR file, which are subsequently processed, retracked and exported to the COASTALT output product. Only data within the coastal mask are included in the COASTALT output product. However, several coastal data segments can be present in the output for any given SGDR file.

6 Applications of Coastal Altimetry The objective of expanding the use of altimetry to coastal areas has to be considered in the context of the applications in which the altimeter data are to be employed, and of the magnitudes and spatial scales of the physical processes involved, together

234


with the magnitudes and scales of the various correction terms. Some coastal work is already possible with existing corrections. For example, the study of slope currents requires a precision of the order of 1 cm over 10 km for features spanning 100 km. At first sight, this is impossible to address given that some correction terms have uncertainties much larger in magnitude e.g. surge corrections with uncertainties of typically 5–10 cm. However, the correction in this case will vary over a much larger spatial scale than the signal of interest, and so studies of slope currents are already possible. The objective is to expand the range of possible applications of altimeter data (Cipollini et al., 2010). That can only be done by reducing the uncertainties of the various correction terms. When altimetry is used to study shelf sea dynamics, there is no clear distinction between tide/surge/inverse barometer corrections, and the clearest interpretation is possible only for sea level (uncorrected except for path-delay corrections and sea state bias) or for sub-surface pressure (sea level corrected with a pure inverse barometer correction). It is clear that the form of processing depends on the particular application and indeed, even altimeter data without any marine correction may still be useful. The COASTALT and PISTACH initiatives for the establishment of a multi-mission global coastal altimetry record are still at a relatively early stage of development. Once the existing 17-year archive of altimetry data is reprocessed, the most immediate application of coastal altimetry will be to look at the coastal sea level. This has two interrelated aspects: (1) long-term sea level change due to climate change and (2) tides. Thanks to reliable estimates of sea level and of its gradient over the continental shelf, an immediate application will be the construction of a global atlas of the statistics of sea level and surface current variability over the continental shelves of the world. But the most ambitious application of the surface dynamic topography from coastal altimetry is to estimate and forecast the three-dimensional ocean state through data assimilation. At the most general level, coastal altimetry in combination with modelling tools and other data sources is essential for applications such as monitoring surges and coastal set-up, measuring long term coastal sea level variations and providing current observations for erosion and sediment transport studies, ship routing and coastal defence design and operation.

7 Concluding Remarks A decade of research in coastal altimetry has provided some insights into the possible use of existing altimeter data sets in the oceanic coastal environment. Traditionally, altimeter data close to the coast were discarded because the processing was difficult and some corrections were inaccurate. Given the proximity to land, satellite-retrieved results may also be contaminated (within 10 km from land). Recently, however, the development of new techniques for reprocessing these data has begun, including the application of local, more accurate, models for their correction. Improvements in the retrieval of information from the shape of radar altimeter ocean averaged waveforms (retracking) have been identified. The COASTALT initiative is contributing to the generation of consistent, high-quality coastal altimeter

11


235

data from the Envisat mission through a series of actions including retracking and new operational corrections. The new coastal altimetry products will provide crucial input for coastal observing systems: the experience gained in optimising existing coastal altimetric data will guide the design of future instruments. Altimetry is a legitimate component of operational coastal observing and modelling systems and can play a significant role in these systems now that the obstacles in retrieving coastal SSH, wave and wind data are being overcome. But continued support is needed for coastal altimetry, including continued efforts to promote the exploitation of existing data sets (a real asset) and new initiatives to take advantage of forthcoming missions. This way, the multi-decadal record of coastal altimetry will not only be sustained, but also be improved in both quality and extent, and the transition to operational systems will be complete. Acknowledgments This chapter has been done in the framework of the COASTALT Project (N. 20698/07/I-LG) funded by the European Space Agency.

References Andersen OB, Dreyer S, Knudsen P, Berry PAM, Mathers EL, Trimmer R, Kenyon S (2005) Deriving 2HZ ERS-1 geodetic mission altimetry for gravity and marine geoid purposes. ESA SP-572, pp 1147–1151 Andersen OB, Knudsen P (2000) The role of satellite altimetry in gravity field modelling in coastal areas. Phys Chem Earth A Solid Earth Geod, 25(1):17–24 Anzenhofer M, Shum CK, Rentsh M (1999) Costal altimetry and applications, Tech. Rep. n. 464, Geodetic Science and Surveying, The Ohio State University, Columbus, OH Benveniste J, Ménard Y (2006) Proceedings of the “15 Years of Progress in Radar Altimetry” Symposium, Venice, Italy, 13–18 March 2006, ESA Special Publication SP-614 Benveniste J, Berry PAM (2004) Monitoring river and lake levels from space. ESA Bull 117: 36–42 Benveniste J, Vignudelli S (2009) Challenges in coastal satellite radar altimetry. Eos Trans AGU 90(26). doi:10.1029/2009EO260007 Berry PAM (2002) Global river and lake height monitoring using satellite altimeter data. Int J Hydropower Dams 9(6):52–54 Berry PAM, Garlick JD, Freeman JA, Mathers EL (2005) Global inland water monitoring from multi-mission altimetry. Geophys Res Lett 32(16):1–4 Bouffard J (2007) Amélioration de l’altimétrie côtière appliquée à l’étude de la circulation dans la partie nord du basin occidental Méditerranéen. Phd Thesis, LEGOS, Toulouse, France, 223 pp Bouffard J, Hermann M, Vignudelli S, Marsaleix P, Birol F, Lyard F, Ménard Y, Cipollini P (2007b) Improved satellite altimetric data dedicated to coastal areas: applications over the Northwestern Mediterranean. European Geophysical Union (EGU) General Assembly, Vienna, Austria, vol IX, pp 10004, 15–20 April Bouffard J, Vignudelli S, Cipollini P, Lyard F, Gasparini GP, Birol F (2007a) A view from multimission satellite atimetry over the coastal ocean: application to the Ligurain Sea and Corsica Channel, OSTST meeting, Hobart, Australia Bouffard J, Vignudelli S, Cipollini P, Menard Y (2008b) Exploiting the potential of an improved multi-mission altimetric dataset over the coastal ocean. Geophys. Res. Lett. doi:10.1029/2008GL033488 Bouffard J, Vignudelli S, Hermann M, Lyard F, Marsaleix P, Ménard Y, Cipollini P (2008a) Comparison of ocean dynamics with a regional circulation model and improved altimetry in the northwestern Mediterranean. Terr Atmos Oceanic J. doi:10.3319/TAO.2008.19.1-2.117(SA)

236


Brooks RL, Lockwood DW, Lee JE, Handcock DW, Hayne GS (1998) Land effects on TOPEX radar altimeter measurements in Pacific Rim coastal zones. In: Brown RA (ed) Remote sensing of the Pacific by satellites. Earth, Ocean. & Space Pty., Glebe, Australia, pp 175–198 Brown GS (1977) The average impulse response of a rough surface and its applications. IEEE J Oceanic Eng 2:67–74 Brown S (2010) A novel near-land radiometer wet path delay retrieval algorithm: application to the Jason-2/OSTM advanced microwave radiometer. IEEE Trans Geosci Remote Sens 48(4). doi:10.1109/TGRS.2009.2037220 Cherniawsky JY, Foreman MGG, Crawford WR (2001) Ocean Tides from TOPEX/ POSEIDON sea level data. J Atmos Oceanic Tech 18(4):649–664 Cherniawsky JY, Foreman MGG, Crawford WR, Beckley BD (2004) Altimeter observations of sea level variability off the West Coast of North America. Int J Remote Sensing 25:1303–1306 Cipollini P, Vignudelli S, Lyard F, Roblou L (2008) 15 years of altimetry at various scales over the Mediterranean. In: Barale V, Gade M (ed) Remote sensing of the European seas. Springer, Heidelberg, Germany, pp 295–306 Cipollini P, Benveniste J, Bouffard J, Emery W, Gommenginger C, Griffin D, Høyer J, Madsen K, Mercier F, Miller L, Pascual A, Ravichandran M, Shillington F, Snaith H, Strub T, Vandemark D, Vignudelli S, Wilkin J, Woodworth P, Zavala-Garay J (2010) The role of altimetry in coastal observing systems, OceanObs09 community white paper. 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.16 Crawford WR, Cherniawsky JY, Foreman MGG (2000) Multi-year meanders and eddies in Alaskan Stream as observed by TOPEX/Poseidon altimeter. Geophys Res Lett 27(7): 1025–1028 Crawford WR, Cherniawsky JY, Foreman MGG, Gower JFR (2002) Formation of the Haida-1998 Oceanic Eddy off the West Coast of Canada. J Geophys Res 107(C3). doi:10.1029/2001JC000876 Crout RL (1998) Coastal currents from satellite altimetry. Sea Technol 8:33–37 Deng XL, Featherstone WE (2006) A coastal retracking system for satellite radar altimeter waveforms: application to ERS-2 around Australia. J Geophys Res – Oceans 111: C06012. doi:10.1029/2005JC003039 Deng XL, Featherstone WE, Hwang C, Berry PAM (2002a) Estimation of contamination of ERS2 and POSEIDON satellite radar altimetry close to the coasts of Australia. Mar Geod 25(4): 249–271 Deng XL, Featherstone WE, Hwang C, Shum CK (2002b) Improved coastal marine gravity anomalies in the Taiwan Strait from altimeter waveform re-tracking. In: Proceedings of the International Workshop on Satellite Altimetry for Geodesy, Geophysics and Oceanography, Wuhan, 8–13 Sept 2002 Desportes C, Obligis E, Eymard L (2007) On the wet tropospheric correction for altimetry in coastal regions. IEEE Trans Geosci Remote Sens 45(7):2139–2149 Dong X, Woodworth PL, Moore P, Bingley R (2002) Absolute calibration of the TOPEX/POSEIDON altimeter using UK tide gauges, GPS, and precise local geoid-differences. Mar Geod 25:189–204 Fernandes MJ, Bastos L, Antunes M (2003) Coastal satellite altimetry – methods for data recovery and validation, 3rd meeting of the international gravity and geoid commission “Gravity and Geoid 2002”, Thessaloniki, August 26–30 Fernandes MJ, Lázaro C, Nunes AL, Pires N, Bastos L, Mendes VB (2010) GNSS-derived Path Delay: an approach to compute the wet tropospheric correction for coastal altimetry. Geosci Remote Sens Lett 7(3). doi:10.1109/LGRS.2010.2042425 Foreman MGG, Crawford WR, Cherniawsky JY, Gower JFR, Cuypers L, Ballantyne VA (1998) Tidal correction of TOPEX/POSEIDON altimetry for seasonal sea surface elevation and current determination off the Pacific Coast of Canada, J. Geophys. Res 103(C12): 27979–27998

11


237

Foreman MGG, Crawford WR, Cherniawsky JY, Henry RF, Tarbotton M (2000) A highresolution assimilating tidal model for the Northeast Pacific Ocean. J Geophys Res 105: 28629–28652 Foreman MGG, Cummins PF, Cherniawsky JY, Stabeno P (2006) Tidal energy in the Bering Sea. J Mar Res 64:797–818 Freeman J, Berry PAM (2006) A new approach to retracking ocean and coastal zone multi-mission altimetry. In Benveniste J, Ménard Y (eds) Proceedings of the “15 Years of Progress in Radar Altimetry” Symposium, Venice, Italy, 13–18 March 2006, ESA Special Publication SP-614 Fu LL, Cazenave A (2001) Satellite altimetry and earth sciences: a handbook of techniques and applications. Academic, San Diego, CA Garlick JD, Berry PAM, Mathers EL, Benveniste J (2005) The Envisat/ERS river and lake retracking system. ESA SP-572, pp 2117–2122 Gómez-Enri J, Gommenginger C, Villares P, Challenor P, Srokosz M, Benveniste J, Drinkwater M (2006) National Oceanography Centre, Southampton retracking scheme. In Benveniste J, Ménard Y (eds) Proceedings of the “15 Years of Progress in Radar Altimetry” Symposium, Venice, Italy, 13–18 March 2006, ESA Special Publication SP-614 Gómez-Enri J, Vignudelli S, Quartly G, Gommenginger C, Benveniste J (2009) Bringing satellite radar altimetry closer to shore. In: Newsroom, Society of Photo-Optical Instrumentation Engineers (SPIE), 1–3 September. doi:10.1117/2.1200908.1797 Gómez-Enri J, Vignudelli S, Quartly GD, Gommenginger CP, Cipollini P, Challenor PG, Benveniste J (2010) Modeling Envisat RA-2 waveforms in the coastal zone: case-study of calm water contamination. Geosci Remote Sens Lett 7(3). doi:10.1109/LGRS.2009.2039193 Gommenginger C, Challenor P, Quartly G, Gómez-Enri J, Srokosz M, Caltabiano A, Berry P, Mathers L, Bennett J, Cotton D, Carter D, Leduc I, Rogers C, Benveniste J (2005) RAIES: Envisat RA-2 individual echoes and S-band data for new scientific applications for ocean, coastal, land and ice remote sensing. ESA SP-572, pp 233–242 Gommenginger C, Thibaut P, Fenoglio-Marc L, Challenor P, Deng X, Gómez-Enri J, Quartly G, Gao Y (2011) Re-tracking altimeter waveforms near the coasts: a review of retracking methods and some applications to coastal waveforms. In: Vignudelli S, Kostianoy A, Cipollini P, Benveniste J (eds) Coastal Altimetry. Springer, Heidelberg 560 pp. doi:10.1007/978-3-64212796-0_4 Lebedev S, Sirota A, Medvedev D, Khlebnikova S, Vignudelli S, Snaith HM, Cipollini P, Venuti F, Lyard F, Cretaux JF, Birol F, Bouffard J, Roblou L, Kostianoy A, Ginzburg A, Sheremet N, Kuzmina E, Mamedov R, Ismatova K, Alyev A, Mustafayev B (2007) Exploiting satellite altimetry in coastal ocean through the ALTICORE project. Russ J Earth Sci 10: ES1002. doi:10.2205/2007ES000262 Liebsch G, Novotny K, Dietrich R, Shum CK (2002) Comparison of Multimission Altimetric Sea-Surface Heights with Tide Gauge Observations in the Southern Baltic Sea. Mar Geod 25:213–234 Lyard F and Roblou L (2003) Composite sea level prediction in the Mediterranean sea – comparisons with observation. Mercator Quaterly newsletter, 10:1–11 Manzella GMR, Borzelli GL, Cipollini P, Guymer TH, Snaith HM, Vignudelli S (1997) Potential use of satellite data to infer the circulation dynamics in a marginal area of the Mediterranean Sea. ESA SP-414 Proceedings of 3rd ERS symposium – space at the service of our environment, Florence, Italy, 17–21 March, vol 3, European Space Agency, pp 1461–1466 Mercier F (2004) Amélioration de la correction de troposphère humide en zone côtière. Rapport Gocina, CLS-DOS-NT-04-086 Mercier F (2008) Improved Jason-2 altimetry products for coastal zones and continental waters (PISTACH project), OSTST Meeting, Nice, France. http://www.aviso.oceanobs.com/fileadmin/ documents/OSTST/2008/Mercier_PISTACH.pdf Obligis E, Desportes C, Eymard L, Fernandes J, Lázaro C, Nunes A (2011) Tropospheric corrections for coastal altimetry. In: Vignudelli S, Kostianoy A, Cipollini P, Benveniste J (eds) Coastal altimetry. Springer, Heidelberg, 560 pp. doi:10.1007/978-3-642-12796-0_6

238


Obligis E, Eymard L, Tran N, Labroue S, Femenias P (2005) First three years of the microwave radiometer aboard ENVISAT: in-flight calibration, processing and validation of the geophysical products. J Atmos Ocean Technol 23(6):802–814 Ollivier A, Le Bihan N, Lacoume JL, Zanife OZ (2005) Improving speckle filtering with SVD to extract ocean parameters from altimeters radar echoes. In: Proceedings of Fourth International Conference on Physics in Signal and Image Processing (PSIP), 31 Jan – 2 Feb 2005, Toulouse, France, pp 157–162 Ponte RM, Salstein DA, Rosen RD (1991) Sea level response to pressure forcing in a barotropic numerical model. J Phys Oceanogr 21:1043–1057 Quesney A, Jeansou E, Lambin J, Picot N (2007) A new altimeter waveform retracking algorithm based on neural networks. OSTST meeting, Hobart, Australia Raney RK, Phalippou L (2011) The future of Coastal Altimetry. In: Vignudelli S, Kostianoy A, Cipollini P, Benveniste J (eds) Coastal Altimetry. Springer, Heidelberg, 560 pp. doi:10.1007/ 978-3-642-12796-0_20 Raney RK (1998) The delay/Doppler radar altimeter. IEEE Trans Geosci Remote Sens 36: 1578–1588 Ray RD, Egbert GD, Erofeeva SY (2011) Tide predictions in shelf and coastal waters: status and prospects. In: Vignudelli S, Kostianoy A, Cipollini P, Benveniste J (eds) Coastal Altimetry. Springer, Heidelberg, 560 pp. doi:10.1007/978-3-642-12796-0_8 Roblou L, Lyard F (2004) Retraitement des données altimétriques satellitaires pour des applications côtières en Mer Méditerranée, Tech. Rep. POC-TR-09 – 04., Pole d’Océanogr. Côtière, Toulouse, France, 15 pp Smith WHF, Strub T, Miller L (2008) First coastal altimetry workshop. Eos Trans AGU 89(40):380 Sutherland G, Garrett C, Foreman M (2005) Tidal resonance in Juan de Fuca Strait and the Strait of Georgia. J Phys Oceanogr 35(7):1279–1286 Vignudelli S, Benveniste J (2010) Coastal altimetry progresses toward applications. Eos Trans AGU 91(5). doi:10.1029/2010EO050006 Vignudelli S, Cipollini P, Astraldi M, Gasparini G, Manzella GMR (2000) Integrated use of altimeter and in situ data for understanding the water exchanges between the Tyrrhenian and Ligurian seas. J Geophys Res 105(8):19649–19663 Vignudelli S, Cipollini P, Roblou L, Lyard F, Gasparini GP, Manzella GMR, Astraldi M (2005) Improved satellite altimetry in coastal systems: case study of the Corsica Channel (Mediterranean Sea). Geophys Res Lett 32: L07608. doi:10.1029/2005GL022602 Vignudelli S, Snaith HM, Lyard F, Cipollini P, Birol F, Bouffard J, Roblou L (2006) Satellite radar altimetry from open ocean to coasts: challenges and perspectives. 5th Asia-Pacific remote sensing symposium, United States society of photo-optical instrumentation engineers (SPIE), Panaji, Goa, India, 13–17 November 2006, vol 6406, 64060L 1–12. doi:10.1117/12.694024 Woodworth PL, Horsburgh K (2011) Surge models as providers of improved “inverse barometer corrections” for coastal areas. In: Vignudelli S, Kostianoy A, Cipollini P, Benveniste J (eds) Coastal Altimetry. Springer, Heidelberg, 560 pp. doi:10.1007/978-3-642-12796-0_7 Yi Y, Lee H, Ibaraki M, Shum CK (2006) Louisiana wetland monitoring using TOPEX/ POSEIDON altimetry. Eos Trans. AGU 87(52), Fall Meet. Suppl., Abstract H23A-1464 Zlotnicki V (1994) Correlated environmental corrections in Topex/Poseidon, with a note on ionospheric accuracy. J Geophys Res 99:24907–24914

Chapter 12

Low Primary Productivity in the Chukchi Sea Controlled by Warm Pacific Water: A Data-Model Fusion Study Kohei Mizobata, Jia Wang, Haoguo Hu, and Daoru Wang

Abstract The Sea-viewing Wide Field-of-view Sensor (SeaWiFS) has identified a broad low chlorophyll-a (chl-a) area in the Chukchi Sea since 2002. High sea surface temperature from 2002 (more than 5◦ C), which resulted in a long duration of open water, was also detected by satellite. An intensified ocean color front at the southwest Chukchi Sea near the Siberian Coast indicates nutrient depletion in the Alaska Coastal Current and its branches. A low chl-a area started to emerge in the Hope Valley in June, and then expanded to the Herald Shoal and Hanna Shoal during July and August. The evolution pattern of low chl-a area is consistent with the variability of the pathway of the Pacific water simulated by a Coupled Ice-Ocean Model (CIOM). These results suggest that the summer phytoplankton bloom from 2002 to 2005 was suppressed by the dominance of warm nutrient-poor water from the Pacific, and by the deepening of the surface mixed layer by strong wind stress. During the summer of 2004, a phytoplankton bloom was detected at the ice edge when the sea surface wind field was relatively calm. Our results imply that the iceedge bloom was induced due to weak wind speeds, which produce shallower upper mixed layer, favoring the ice-edge bloom. Keywords Chlorophyll · Chukchi Sea · Pacific water · SeaWiFS · Coupled ice-ocean model · Primary productivity

1 Introduction The Chukchi Sea is a seasonal sea ice zone connecting the Bering Sea to the Arctic Ocean (Fig. 12.1). The warm Pacific water through the Bering Strait melts the sea ice during spring and summer, and high chlorophyll biomass has been observed at J. Wang (B) NOAA Great Lakes Environmental Research Laboratory (GLERL), 4840 S. State Road, 48108 Ann Arbor, MI, USA e-mail: [email protected] D. Tang (ed.), Remote Sensing of the Changing Oceans, C Springer-Verlag Berlin Heidelberg 2011 DOI 10.1007/978-3-642-16541-2_12,

239

240

K. Mizobata et al.

Fig. 12.1 Chukchi Sea bathymetry map. Only bathymetry inside the study area is shown (in meters)

the ice edge (Wang et al., 2005a; Hill and Glenn, 2005). According to Springer and McRoy (1993), the annual primary production is up to 125–800 gC m−2 in the south Chukchi Sea, and the phytoplankton biomass is tightly connected to the benthic biomass, which supports large populations of benthic-feeding marine mammals and birds (e.g., Grebmeier and Dunton, 2000; Dunton et al., 2005). Thus phytoplankton is an important factor for the Chukchi Sea marine ecosystem. In the Chukchi Sea, both the ice-edge bloom and open-water bloom contribute to primary production and carbon flux (Coyle and Cooney, 1988). The Chlorophyll-a (chl-a) value as an index of phytoplankton biomass fluctuates due to various physical forcings. The relationship among chl-a, sea ice, sea surface temperature, and wind in the Chukchi/Beaufort Sea was examined by Wang et al. (2005a). They found a relatively high correlation between chl-a and ice in May and attribute it to an ice-edge bloom. However the mechanism controlling the chl-a value and distribution is still poorly known due to a lack of in-situ data. The Chukchi Sea is strongly affected by the inflow of water from the Pacific through the Bering Strait (e.g., Codispoti et al., 1991; Woodgate et al., 2005). Northward transport of the Pacific water influences the ocean circulation and transports nutrient-rich water (Walsh et al., 1989; Weingartner et al., 1998). This northward advection is strongest in the summer (Rudels, 2001). During the sea ice melting season, a phytoplankton bloom usually occurs as light intensity increases. In the summer, a low chl-a area is common in the Chukchi Sea, while a relatively high chl-a area is sustained in the coastal region. Since 2002, the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) ocean color sensor, which has provided data since

12

Low Primary Productivity in the Chukchi Sea Controlled by Warm Pacific Water

241

September 1997, has captured a wide low chl-a area in the Chukchi Sea. An increase in heat flux and volume transport through the Bering Strait has also been reported since 2002 (Woodgate et al., 2006). The years 2002 and 2005 had record low summer ice extent in the Pacific Arctic (Wang et al., 2009a), which was due to the positive Dipole Anomaly (DA) in the Arctic. The wind anomaly associated with the positive DA is northward winds, which had the following important effects: (1) advects warm, humid air from the south to the Chukchi Sea, directly causing more melting, (2) drives sea ice away from the Pacific Arctic (Beaufort-Chukchi and the Eastern Siberian-Laptev seas) to the Atlantic Arctic (Wang et al., 2009a), (3) promotes the an increase in Bering Strait volume and heat transport (Mizobata et al. 2010), and (4) increases the ice/ocean albedo feedback, leading to accelerating ice melting (Wang et al., 2005c). The phytoplankton community is highly sensitive to changes in physical properties, especially in the continental shelf, such as in the Chukchi Sea. Therefore, phytoplankton dynamics is probably influenced by Pacific water during the ice melting season. In this study, we explore how Pacific water affects summer chl-a distribution and its magnitude in the Chukchi Sea (66–75◦ N, 178–155◦ W) between 2002 and 2005, using remotely sensed imagery and a Coupled Ice-Ocean Model (CIOM, Wang et al., 2002, 2005b, 2009b).

2 Data and Methods 2.1 Satellite Remote Sensing Data In this study, we employed three kinds of satellite data from 1998 to 2005 derived from the Sea-viewing Wide Field-of-view Sensor (SeaWiFS), the Special Sensor Microwave Imager (SSM-I), and the Advanced Very High Resolution Radiometer (AVHRR). We used the SeaWiFS daily dataset provided by the NASA/Godard Space Flight Center (GSFC) OceanColor Web (Feldman and McClain, 2005, http:// oceancolor.gsfc.nasa.gov/) to construct monthly 9 km × 9 km Chl-a images. The Ocean Color 4 version 4 Linear (OC4L) algorithm, which was proposed by Wang and Cota (2003), was utilized to estimate chl-a values. Using a SeaWiFS OC4L chl-a map, we calculated the area of low chl-a (less than 0.5 mg m−3 ). To estimate the sea ice area (km2 ), we utilized the SSM-I sea ice concentration maps estimated by the NASA team algorithm. The dataset was acquired at the National Snow and Ice Data Center website (http://nsidc.org/data/nsidc-0002.html). To avoid algorithm errors due to a melt pond or weather effects, only sea ice concentration more than 30% was used. After remapping the sea ice concentration map to fit the SeaWiFS 9 km resolution map, ice distribution was superimposed on the SeaWiFS OC4L chl-a map. We also used the AVHRR Pathfinder sea surface temperature (SST) data version 5.0 to estimate the average SST in the Chukchi Sea. The AVHRR SST dataset

242

K. Mizobata et al.

was provided by the NASA/Jet Propulsion Laboratory/Physical Oceanography Distributed Active Archive Center. Using SST images, the interannual variability of averaged SST in the Chukchi Sea was calculated to compare with the low chl-a area and the sea ice area.

2.2 The NCEP/NCAR Reanalysis Dataset We used the NCEP/NCAR Reanalysis dataset to describe the sea surface wind field over the Chukchi Sea during summers from 2002 to 2005 and to use in the coupled ice-ocean circulation model. Both monthly atmospheric climatology and daily atmospheric forcing (Sea surface wind, Sea level pressure, Longwave/Shortwave

Fig. 12.2 Sea surface chlorophyll-a pattern and sea ice area (more than 30% sea ice concentration) during summer (from June to August) from 2002 to 2005. Chl-a and sea ice concentration were derived from the Ocean Color 4 Linear (Wang and Cota, 2003). Sea ice concentration more than 30% is superimposed. Red arrow indicates sea surface wind vector derived from the NCEP/NCAR monthly dataset

12


243

radiation, Humidity, U-/V-wind at 10 m, Air temperature) were derived from the Climate Diagnosis Center website (http://www.cdc.noaa.gov/cdc/reanalysis/ reanalysis.shtml). The NCEP/NCAR monthly sea surface wind field was superimposed on the SeaWiFS OC4L chl-a map (Fig. 12.2).

2.3 The Description of the Coupled Ice-Ocean Model (CIOM) To simulate the ice-ocean circulation, we applied the 3-D Coupled Ice-Ocean Model to the Chukchi Sea (CIOM, Wang et al., 2002, 2005b, 2009b; Hu and Wang, 2010). The sea ice components of the CIOM are a thermodynamic model based on multiple categories of ice thickness distribution (Thorndike et al., 1975; Hibler, 1980) and a dynamic model based on a viscous-plastic sea ice rheology (Hibler, 1979). In this study, ten ice categories (0, 0.2, 0.5, 1, 1.5, 2, 3, 4, 5 and 6 m) were used. The Princeton Ocean Model (Blumberg and Mellor, 1987) was used as the ocean component of the CIOM. The model was spun up with temperature and salinity (PHC) of Steele et al. (2001) for the first four years under monthly atmospheric climatology. After spin-up integration, we ran the CIOM under daily atmospheric forcing data from 2002.

3 Chl-a Distribution, Sea Surface Wind, SST, and Sea Ice Cover Between 2002 and 2005 Figure 12.2 illustrates chl-a distribution, sea ice area, and sea surface wind pattern in the Chukchi Sea during late spring to summer (June, July, and August) from 2002 to 2005. The white area, gray area, and red arrows represent lack of SeaWiFS data, the sea ice area more than 30%, and the sea surface wind, respectively. In spite of missing chl-a data due to a gap of observing time and of spatial resolution between the SeaWiFS and SSM-I, horizontal chl-a distribution was clearly and mostly captured. The SeaWiFS revealed a wide low chl-a area and its interannual variability in the Chukchi Sea from 2002 to 2005 (Fig. 12.2b, c, d, e, i, j, k, and l). In 2002, a low chl-a area started to appear in the Hope Valley in June (Fig. 12.2a) and expanded to the Siberian Coast and northern Chukchi Sea in July and August (Fig. 12.2b and c). In June 2003, a low chl-a area already existed in the south Chukchi Sea near the Bering Strait, in Hope Valley, and off Cape Lisburne (Fig. 12.2d). There was an extensive low chl-a area in July (Fig. 12.2e), similar to that in 2002, and remained in parts of the northern Chukchi Sea (Herald Valley and west Hanna Shoal) through August (Fig. 12.2f). In June 2004, a low chl-a area showed up in the southern Hope Valley (Fig. 12.2g), similar to June 2002. In July, there was a low chl-a area in the Hope Valley and south of Hanna Shoal (Fig. 12.2h), and was widely distributed over the Hope Valley, Herald Shoal and Hanna Shoal through August (Fig. 12.2i). In June 2005, a low chl-a area emerged in the southern Hope Valley and the area between Herald Shoal and Hanna Shoal (Fig. 12.2j). One month later

244

K. Mizobata et al.

in July, a low chl-a area was seen in the Chukchi Sea, excluding the south Siberian Coast (Fig. 12.2k). In August 2005, a low chl-a area still remained in the northern Chukchi Sea (Fig. 12.2l). Regardless of the interannual variability of the low chl-a area, there is a pattern to its occurrence. The development of the low chl-a area begins in the Hope Valley or the southern Hope Valley, expands to the Herald Shoal and the Hanna Shoal and finally covers the Chukchi Shelf. Additionally, the low chl-a at the south Chukchi Sea near the Bering Strait changed to a relatively high chl-a area earlier than in the north Chukchi Sea, indicating chl-rich water was advected northwardly. In the deep basin area, the low chl-a area also occurred in the open water area, which seems to be caused by sea surface winds in July and August. The SeaWiFS chl-a imagery also exhibited the location of a chl-rich area in the Chukchi Sea. High chl-a is seen at the southeastern Siberian Coast and along the Alaskan Coast. Along the Alaskan Coast, a high chl-a area more than 1 mg m–3 is always detected, but its width is variable. A chl-a area along the Alaskan Coast usually occurs from the coastline to the 40 m isobath. In some cases, this band became fairly narrow (Fig. 12.2b, c, e, and k). High chl-a water along the Alaskan Coast is always detected in the Barrow Canyon during July and August. In August, chlrich water occasionally appeared in the Hope Valley, Hanna Shoal, or shelf break area where sea ice existed one month before (Fig. 12.2f, i, and l). At the Siberian Coast near the Bering Strait, chl-rich water was always maintained, but its magnitude was variable. A most prominent phytoplankton bloom of more than 10 mg m–3 was seen in June 2003 (Fig. 12.2d). In July, chl-a values usually decrease, but chl-a distribution sometimes shows a regional bloom in August. Thus, integrated chl-a in this area, which is potentially important for benthic biomass, highly depends on a regional bloom. A high chl-a area was also detected along the ice edge in July 2004, when gentle sea surface winds blew over the open water area. The ice-edge bloom is hard to detect during summer using satellite datasets with a different resolution. This ice-edge bloom was distributed between 40 m and 50 m water depths. In 2003 and 2005, there are also relatively high chl-a areas in parts of the ice-edge areas (Figs. 12.2e and k). Figure 12.3a shows interannual variability of the low chl-a area (≤0.5 mg m−3 ; gray bar) and averaged SST (less than 1,000 m water depth; black line) in the Chukchi Sea between 1998 and 2005. The averaged SST in the Chukchi Sea shelf of less than 1,000 m water depth was estimated. A low chl-a area has increased between June and September since 2002. Woodgate et al. (2006) has reported the increase in heat flux, volume transport, and fresh water flux through the Bering Strait detected by the mooring observations. Although SST data includes values between Wrangel Island and the Siberian coast, where the cold Siberian Coastal Current flows (Weingartner et al., 1999), high averaged SST (about 5–6◦ C) in the Chukchi Sea was detected since 2002. In 1998 and 1999, high SST was also similar to 2002 or 2003, while low SST (less than 4◦ C) was found in 2000 and 2001. The SST trend is not consistent with the variability of the low chl-a area, but high SST during summer 2002 coincides with results of Woodgate et al. (2006). The SST is highly variable, affected by local wind and heat exchange between the ocean and atmosphere. Thus we calculated the sea ice area and the period of the open water in the Chukchi

12


245

Fig. 12.3 Interannual variability of (a) low chl-a area (less than 0.5 mg m−3 ) and averaged sea surface temperature (excluding the area around the Wrangel Island) in the Chukchi Sea less than 1,000 m water depths, and (b) sea ice area in the entire study area

Sea (Fig. 12.3b), which is directly influenced by the Pacific water and local wind anomaly (Wang et al., 2009a). The sea ice area trend is well synchronized with that of the SST. The period of open water, however, became longer since 2002. Those high SST periods in 2002, with a relatively long duration of open water and less ice cover, indicate the increase in heat flux through the Bering Strait (Woodgate et al., 2006) and rapid sea ice retreat in the Chukchi Sea. These phenomena should be related to the large-scale DA-derived southerly wind anomaly (Wang et al., 2009a).

4 Possible Mechanisms for CHL-A Variability The locations of low chl-a areas in the Chukchi Sea are roughly consistent with the pathway of the ACC and its branches, which has been described by ship surveys and modeling studies (Weingartner et al., 2005; Winsor and Chapman, 2004). Also the low chl-a area and water temperature? have increased since 2002. Those results suggest a link between the phytoplankton dynamics and the warm Pacific water. Then we applied the 3-D CIOM to the Chukchi Sea, to explain the advection of the Pacific water. We successfully simulated both ice and ocean circulation in the Chukchi Sea in 2002 using the CIOM with NCEP/NCAR daily forcing. The simulated ocean circulation pattern is consistent with the previous hydrographic survey (Weingartner et al., 2005) and modeling study (Winsor and Chapman, 2004). The simulated sea ice area is also similar to that derived from the SSM-I sea ice concentration maps (not shown), except for sea ice distribution over the Hanna Shoal and Siberian coastal area near the Wrangel Island. A seasonal cycle of sea ice was well reproduced by the CIOM (Fig. 12.4).

246

K. Mizobata et al.

Fig. 12.4 CIOM-simulated seasonal cycle of ice cover in 2002 (red), in comparison with SSM/I measurements (blue) in 2002. The gray lines are the daily sea ice areas each year (9 years, i.e., nine curves) from January 1997 to December 2005 with the combined SMSP and SSM/I measurements and the black line is the climatological daily sea ice area averaged for the same 9-year period

Figure 12.5 shows sea ice cover (gray, sea ice concentration more than 30%), ice velocity (red arrow), SST and water velocity at 10 m water depth (black arrow) on June 20, July 20, and August 28. In June, the warm Pacific water flows along the Alaska Coast, Hope Valley, and the area between Herald Shoal and Hanna Shoal, where a low chl-a area and open water start to emerge (Fig. 12.5a). In July and August, warm Pacific water spreads to cover the Chukchi Sea shelf (Fig. 12.5b and c), which is consistent with a wide low chl-a area in 2002 (Fig. 12.2b and c). Recent hydrographic surveys conducted by the 2002 Western Arctic Shelf-Basin Interactions (SBI) Processes Cruises (http://sbi.utk.edu/) revealed low nitrate and ammonium concentrations on the eastern side of the Bering Strait, where the ACC flows during summer (Codispoti et al., 2005), while a high chl-a area, which is always maintained at the southeast Siberian Coast (Wang et al., 2005a), is due to a nutrient-rich environment supported by the Anadyr Current (Springer and McRoy, 1993). Therefore a strong ocean color front between the western side and eastern side of the south Chukchi Sea near the Bering Strait indicates nutrient-poor water in the ACC. In the Alaskan Coastal area in July of 2003 and 2005, the phytoplankton

12


247

a)

b)

c)

Fig. 12.5 The pathway of the Pacific Summer Water and sea ice cover (sea ice concentration more than 30%) on (a) June 20, (b) July 20, and (c) August 28 in 2002 simulated by the IARC Coupled Ice-Ocean Model (CIOM) using the NCEP/NCAR daily atmospheric forcing. Red and black arrows show ice velocity and oceanic water velocity at 10m water depth, respectively

248

K. Mizobata et al.

bloom was probably retrained by low-nutrient Pacific water. Additionally, extensive open water and an increase in heat/fresh water flux from 2002 will easily promote the occupation of low-nutrient Pacific water and induce a stable surface stratified layer over the continental shelf in the Chukchi Sea because of buoyancy. Thus, our results indicate that the Pacific water will determine an emerging pattern of a low chlorophyll area like “oceanic desert” during summer, whereas chl-rich water is advected by the ACC in late summer in the Bering Strait (Fig. 12.2c, f, and h). On the other hand, a high chl-a area at the ice edge in July 2004, which looks like the ice-edge bloom, cannot be explained by the warm low-nutrient Pacific water. Figure 12.5b and c indicate that ice-originated water tends to remain at the ice edge and in the eastern Herald Shoal because of the ocean circulation pattern, which is slightly modified by the Pacific water. The low SST area on August 28, 2002 simulated by the CIOM (Fig. 12.5c) is similar to the distribution pattern of ice-edge bloom at the Hanna Shoal in 2002 in spite of different atmospheric forcings. We can assume that weak wind, warm surface water from the Pacific, and freshened sea water (from ice melting) probably induced a salinity stratification and high nutrient situation resulting in a phytoplankton bloom in July 2004 (Niebauer et al., 1990; Hill et al., 2005). A smaller ice-edge bloom and relatively strong winds in July 2002, 2003, and 2005 suggest that strong winds destruct surface stratified water and deepen the low-nutrient mixed layer; but this idea needs further validation using measurements and model simulations. At the eastern side of the Northwind Ridge in the basin, there was also low chl-a area in August 2004 and 2005 (Fig. 12.2i and l). These low chl-a areas are due to the deep polar mixed layer, rather than the Pacific water. Because the polar mixed layer is characterized as having quite low nitrate and anmonium concentrations (Codispoti et al., 2005; Hill et al., 2005).

5 Concluding Remarks This study investigates the influence of the warm Pacific water not only on the sea ice area (e.g., Shimada et al., 2006), but also on phytoplankton dynamics in the Chukchi Sea using satellite remote sensing and the CIOM. In the surface layer, a low chl-a “oceanic desert” in the Chukchi Sea became larger since 2002 due to the flow of warm Pacific water through the Bering Strait resulting in the low-nutrient warm stratified surface layer, which restrains a nutrient supply from the subsurface or bottom layer. The comparison of the July chl-a magnitude at the ice edge and the wind field from 2002 to 2005 imply that the ice-edge bloom is promoted under a calm wind field. A relatively strong wind field similar to that in July 2002, 2003, and 2005 probably deepens the nutrient-poor layer, resulting in less opportunity for phytoplankton to utilize light and nutrient. Therefore, the amount of carbon flux from the surface layer decreased since 2002. Also the period of open water and the phytoplankton bloom occurred earlier than those before 2001, which may affect the survival rate and strategy of marine species in the Chukchi Sea. But the impact of the warm Pacific water on the total carbon flux in whole water column is ambiguous due to the lack of an in-situ dataset. At the Barrow canyon, Pickart et al. (2005)

12


249

observed a high fluorescence at 20–30 m water depth in summer. Intensified chl-a distribution is often found at the subsurface layer when water column is stratified. The information of nutrient distribution, amount of decomposition/grazing of ice algae, and relevant physical parameters is needed, which is beyond the scope of this study. To further investigate the changes of chl-a distribution and lower trophic level ecosystems in response to changes in ice-ocean circulation and climate, we are applying an ice-ocean-ecosystem model to the Chukchi Sea in the near future, along with on-going in situ hydrographic surveys. Additionally, the SeaWiFS OC4 linear algorithm, which was proposed by Wang et al. (2005a), made it possible to carry out this work. In the Chukchi Sea, the newlyimproved regional ocean color algorithm for the current main ocean color sensor Moderate Resolution Imaging Spectroradiometer (MODIS) is urgently needed. Acknowledgements A part of this study is supported by the Japan Aerospace Exploration Agency (JAXA) through the program of Arctic Research projects using the IARC (International Arctic Research Center)-JAXA Information System (IJIS). J.W. and K.M. also appreciate support from the RUSALCA Modeling Project of the NOAA Office of Arctic Research. The manuscript contents are solely the opinions of the authors and do not constitute a statement of policy, decision, or position on behalf of NOAA or the U. S. Government. This is GLERL contribution 1552.

References Blumberg AF, Mellor GL (1987) A description of 3-D coastal ocean circulation model. In: Heaps NS (ed) Coastal and estuarine sciences 4: 3-D coastal ocean models. American Geophysical Union, Washington, DC, pp 1–16 Codispoti LA, Flagg C, Kelly V, Swift JH (2005) Hydrographic conditions during the 2002 SBI process experiments. Deep Sea Res II 52:3199–3226 Codispoti LA, Friederich GE, Sakamoto CM, Gordon LI (1991) Nutrient cycling and primary production in the marine systems of the Arctic and Antarctic. J Mar Sys 2:359–384 Coyle K, Cooney RT (1988) Estimating carbon flux to pelagic grazers in the ice-edge zone of the eastern Bering Sea. Mar Biol 100:41–49 Dunton KH, Goodall JL, Schonberg SV, Grebmeier JM, Maidment DR (2005) Multi-decadal synthesis of benthic-pelagic coupling in the western arctic: role of cross-shelf advective processes. Deep Sea Res II 52:3462–3477 Feldman GC, McClain CR (2005) Ocean color web. In: Kuring N, Bailey SW (eds) SeaWiFS reprocessing, NASA Goddard Space Flight Center. October. http://oceancolor.gsfc.nasa.gov/ Grebmeier JM, Dunton KH (2000) Benthic processes in the northern Bering/Chukchi seas: status and global change. Impacts of changes in sea ice and other environmental parameters in the Arctic. Report of the Marine Mammal Commission Workshop, 15–17 February, Girdwood, Alaska. Marine Mammal Commission, Bethesda, MD, pp 61–71 Hibler WD III (1979) A dynamic thermodynamic seaice model. J Phys Oceanorgr 9:817–846 Hibler WD III (1980), Modeling a variable thickness sea ice cover. Mon Wea Rev 108: 1943–1973 Hill V, Glenn C (2005) Spatial patterns of primary production on the shelf, slope and basin of the Western Arctic in 2002. Deep Sea Res II 52:3344–3354 Hill V, Glenn C, Stockwell D (2005) Spring and summer phytoplankton communities in the Chukchi and Eastern Beaufort Seas. Deep Sea Res II 52:3369–3385 Hu H, Wang J (2010) Modeling effects of tidal and wave mixing on circulation and thermohaline structures in the Bering Sea: Process studies, J Geophys Res 115:C01006, doi:10.1029/2008JC005175

250

K. Mizobata et al.

Mizobata K, Shimada K, Saitoh S, Wang J (2010) Estimation of heat flux through the eastern Bering Strait. J Oceanogr 66(3):405–424. doi:10.1007/s10872-010-0035-7 Niebauer HJ, Alexander V, Henrichs S (1990) Physical and biological oceanographic interaction in the spring bloom at the Bering Sea marginal ice edge zone. J Geophys Res 95:22229–22241 Pickart RS, Weingartner TJ, Pratt LJ, Zimmermance S, Torresa DJ (2005) Flow of wintertransformed Pacific water into the Western Arctic. Deep-Sea Res II, 52:3175–3198 Pa S, Ha X, Dong Z, Wang J, Yu F, (2002) Hydrographic features and variability of the Oceanic front near Prydy Bay in the Antartic Ocean (Submitted to J Phys Oceanogr) Rudels B (2001) Arctic Basin circulation. In: Encyclopedia of ocean sciences. Academic, San Diego, CA, pp 177–187 Shimada K, Kamoshida T, Itoh M, Nishino S, Carmack E, MacLaughlin F, Zimmermann S, Proshutinsky A (2006) Pacific Ocean inflow: influence on catastrophic reduction of sea ice cover in the Arctic Ocean. Geophys Res Lett 33: L08605. doi:10.1029/2005GL025624 Springer AM, McRoy CP (1993) The paradox of pelagic food webs in the northern Bering Sea. III. Patterns of primary production. Cont Shelf Res 13:575–579 Steele M, Morley R, Ermold W (2001) PHC: a global ocean hydrography with a high-quality Arctic Ocean. J Clim 14:2079–2087 Thorndike AS, Rothrock DA, Maykut GA, Colony R (1975) The thickness distribution of sea ice. J Geophys Res 80(C5):4501–4513 Walsh JJ et al. (1989) Carbon and nitrogen cycling within the Bering/Chukchi Seas: source regions for organic matter effecting AOU demands of the Arctic Ocean. Prog Oceanogr 22:259–277. doi:10.1016/0079-661(89)90006-2 Wang J, Cota GF (2003), Remote-sensing reflectance in the Beaufort and Chukchi seas: observations and models, Appl Opt 42:2754–2765 Wang J, Cota GF, Comiso JC (2005a) Phytoplankton in the Beaufort and Chukchi Seas: distribution, dynamics, and environmental forcing. Deep Sea Res II 52:3355–3368 Wang J, Ikeda I, Zhang S, Gerdes R (2005c) Linking the northern hemisphere sea ice reduction trend and the quasi-decadal Arctic Sea Ice Oscillation. Climate Dyn 24:115–130. doi:10.1007/s00382-004-0454-5 Wang J, Liu Q, Jin M (2002) A user’s guide for a coupled ice-ocean model (CIOM) in the PanArctic and North Atlantic Oceans. In: International Arctic research center-frontier research system for global change, Tech. Rep. 02-01. International Arctic Research Center, Fairbanks, AK, pp 1–65 Wang J, Liu Q, Jin M, Ikeda M, Saucier FJ (2005b) A coupled ice-ocean model in the pan-Arctic and the northern North Atlantic Ocean: simulation of seasonal cycles. J Oceanogr 61:213–233 Wang J, Hu H, Mizobata K, Saitoh S (2009b) Seasonal variations of sea ice and ocean circulation in the Bering Sea: a model-data fusion study. J Geophys Res 114: C02011. doi:10.1029/2008JC004727 Wang J, Zhang J, Watanabe E, Mizobata K, Ikeda M, Walsh JE, Bai X, Wu B (2009a) Is the Dipole Anomaly a major driver to record lows in the Arctic sea ice extent? Geophys Res Lett 36:L05706. doi:10.1029/2008GL036706 Weingartner TJ, Aagaard K, Woodgate R, Danielson S, Sasaki Y, Cavalieri D (2005) Circulation on the north central Chukchi Sea shelf. Deep Sea Res II 52:3150–3174 Weingartner TJ, Cavalieri DJ, Aagaard K, Sasaki Y (1998) Circulation, dense water formation, and outflow on the northeast Chukchi shelf. J Geophys Res 103:7647–7661 Weingartner TJ, Danielson S, Sasaki Y, Pavlov V, Kulakov M (1999) The Siberian Coastal Current: a wind- and buoyancy-forced Arctic coastal current. J Geophys Res 104:29697–29713 Winsor P, Chapman DC (2004) Pathways of Pacific water across the Chukchi Sea: a numerical model study. J Geophys Res 109: C03002. doi:10.1029/2003JC001962 Woodgate RA, Aagaard K, Weingartner T (2005) A year in the physical oceanography of the Chukchi Sea: moored measurements from autumn 1990–1991. Deep Sea Res II 52:3116–3149 Woodgate RA, Aagaard K, Weingartner TJ (2006) Interannual changes in the Bering Strait fluxes of Volume, Heat and Freshwater between 1991 and 2004. Geophys Res Lett 33:L15609

Chapter 13

Medium Resolution Microwave, Thermal and Optical Satellite Sensors: Characterizing Coastal Environments Through the Observation of Dynamical Processes Domingo A. Gagliardini

Abstract Synthetic Aperture Radar (SAR) satellite sensors can provide relevant information about a variety of features related to dynamical processes. Due to the high resolution of available SAR sensors, it is possible to detect details of circulation and small-scale processes which are not observable by other satellite sensors frequently used for ocean research, including the Advanced Very High Resolution Radiometer (AVHRR) and the Sea-viewing Wide Field-of-view Sensor (SeaWiFS). Additionally LANDSAT-Thematic Mapper (TM)/ Enhanced Thematic Mapper Plus (ETM+) thermal and optical channels can be used to observe sea surface temperatures and surface layer ocean color (upwelled radiance), as well as sun glint patterns of surface roughness (reflected radiance) at a high spatial resolution, comparable to that of SAR. A large amount of LANDSAT TM/ETM+ and ERS-SAR data were processed to observe in detail some dynamical phenomena in the coastal environment of the Southwestern Atlantic Ocean, including fronts, internal waves, eddies, upwelling and bathymetric signatures. The purpose of this chapter is to present and discuss examples of these phenomena observed in three dissimilar sectors located along a broad range of latitudes (from 33◦ S to 54◦ S). Results indicate that, independently of the study area, medium resolution optical, thermal, and microwave sensors can provide relevant information about the properties of coastal environments by the observation of ocean dynamic processes. In addition, this study demonstrates that each sensor can capture different aspects or properties of the same process. Collectively, all this information provides a better understanding of the characteristics of coastal waters, highlighting their influence on marine biodiversity. Keywords Medium resolution sensors · Coastal environments · Dynamical processes D.A. Gagliardini (B) IAFE, Casilla de Correo 67, Suc. 28 (C1428ZAA) Ciudad Autónoma de Buenos Aires, Argentina e-mail: [email protected]


251

252

D.A. Gagliardini

1 Introduction Coastal environments of the Southwestern Atlantic Ocean from 33◦ S to 55◦ S are highly productive, allowing the existence of important seabird and marine mammal breeding assemblages, fish and crustacean spawning and nursery areas, and extensive beds of macroalgae and mollusks. The main oceanographical forcings in the area are: the outflow of the La Plata River, the Magellan Strait and some smaller effluents and nearshore, shelf and offshore currents, the latter including the Brazil and Malvinas (Falkland) currents (Fig. 13.1). These different water masses interact with each other and with the sea floor, originating a series of processes that strongly influence the dynamical and biological characteristics of these environments. In particular, numerous fronts of different origins are produced, giving rise to highly productive areas of great biological importance. Studies using satellite data have greatly enhanced the knowledge of these coastal environments in recent years, including various aspects associated with physical and biological oceanography (e.g., Armstrong et al., 2004; Acha et al., 2004; Saraceno et al., 2004, 2005; Garcia et al., 2005; Romero et al., 2006; Rivas, 2006; Rivas et al., 2006; Garcia and Garcia 2008; Piola and Romero, 2004; Piola et al., 2008; Dogliotti et al., 2009). Most of these studies have been conducted with information provided by the AVHRR, the SeaWiFS and the Moderate Resolution Imaging Spectroradiometer (MODIS) sensors. However, the low resolution of these sensors limits their ability to observe dynamic phenomena close to the coastline, or

Fig. 13.1 Study area: (a) Southwestern Atlantic Ocean (adapted from Piola and Rivas, 1997); (b) Sector 1 (S1); (c) north of Sector 2 (S2); (d) south of Sector 2 (S2), (e) Sector 3 (S3)

13

Medium Resolution Microwave, Thermal and Optical Satellite Sensors

253

particular details of interesting processes taking place in the open sea. Additional information, obtained with better resolution, is needed to overcome this limitation. This is particularly meaningful in the context of current interests and concerns regarding climate change and ecosystem based-management. The Argentine Agency for Space Activities (CONAE) inaugurated its own receiving station in 1997, and since then has provided free-of-charge images for research and teaching purposes. This has increased substantially the availability of satellite information for the region, facilitating oceanic research. However, most of the studies were conducted with low resolution data; only a few took advantage of LANDSAT- TM/ETM+ and ERS- SAR images (e.g., Gagliardini and ClementeColón, 2004a, 2004b; Gagliardini, et al., 2004a, 2004b, 2005; R. Amoroso and D. A. Gagliardini, preprint, 2009). These papers, however, addressed the capability of both satellite systems in oceanographic research, rather than in the identification of specific environmental characteristics. On the contrary, this contribution emphasizes the actual capabilities of those systems, currently considered of medium resolution (http://wgiss.ceos.org/lsip/overview.shtml), to study such characteristics. The results reported here are based on the analysis of a large number of LANDSAT-TM/ETM+ and ERS-SAR images covering three dissimilar sectors of the Southwestern Atlantic Ocean, over a broad latitudinal range (from 33◦ S to 55◦ S, Fig. 13.1a). Processing of these data allowed the detailed observation of a variety of features related to different dynamical phenomena, and illustrates the valuable information that can be obtained from sun glint patterns. This procedure made it possible to identify currents, fronts, internal waves, eddies, natural biogenic surfactants and bathymetric signatures in the visible, infrared and microwave bands. The purpose of this contribution is to present and discuss examples of the phenomena that characterize each of the sectors selected for detailed scrutiny, and to highlight differences among them. Additionally, it is shown that each sensor, alone or in combination with others, can strongly contribute to a better knowledge of coastal water dynamics that influence ecological processes. The results reported here are based on the analysis of a large number of LANDSAT-TM/ETM+ and ERS-SAR, images covering three dissimilar sectors of the Southwestern Atlantic Ocean, over a broad latitudinal range (from 33◦ S to 55◦ S, Fig. 13.1a). A list of these images is presented in Table 13.1 (see appendix) where satellite, sensor, path, row and date corresponding to each image are indicated. Processing of these data allowed the detailed observation of a variety of features related to different dynamical phenomena, and illustrates the valuable information that can be obtained from sun glint patterns. This procedure made it possible to identify currents, fronts, internal waves, eddies, natural biogenic surfactants and bathymetric signatures in the visible, infrared and microwave bands. Some results obtained over Sector 2 (Fig. 13.1) are also compared with SeaWiFS products. The purpose of this contribution is to present and discuss examples of the phenomena that characterize each of the sectors selected for detailed scrutiny, and to highlight differences among them. Additionally, it is shown that each sensor, alone or in combination with others, can strongly contribute to a better knowledge of coastal water dynamics that influence ecological processes.

Satellite/sensor

LANDSAT5-TM LANDSAT5-TM LANDSAT7-ETM LANDSAT7-ETM ERS2-SAR + LANDSAT5-TM LANDSAT5-TM LANDSAT5-TM LANDSAT5-TM LANDSAT5-TM LANDSAT5-TM ERS2-SAR + LANDSAT5-TM LANDSAT5-TM LANDSAT7-ETM LANDSAT7-ETM LANDSAT7-ETM LANDSAT7-ETM LANDSAT5-TM LANDSAT5-TM LANDSAT5-TM ERS2-SAR + ERS2-SAR + ERS2-SAR + ERS2-SAR + ERS2-SAR + ERS2-SAR −

Figure

2b 2c 3b 3c 3d 4b 4c 4d 4e 4f 4g 4h 4i 4j 5b 5c 5d 5e 5f 5g 6b 6c 6d 6e 6f 6g 6h

224-84/85 221/86/87/88 226-89 226-89 20605-4455 221-86/87/88 221-86/87 221-87 221-88 221-84 221-84 23854-4311 221-84 221-84 227-89 226-89-90 227-89 227-93 227-91 226-89 221-83 24713-4329 19474-4455 25901-4347 19474-4455 24212-4347 22187-6399

Path/row 30/11/1998 06/09/1998 28/01/2007 28/01/2001 30/03/1999 06/09/1998 06/09/1998 06/09/1998 06/09/1998 12/11/1999 12/11/1999 12/11/1999 28/11/1999 28/11/1999 21/12/2001 12/12/1999 21/12/2001 07/02/2002 25/01/2000 25/12/1999 28/11/1999 11/01/2000 10/01/1999 03/04/2000 10/01/1999 07/12/1999 19/07/1999

Date 10b 10c 11b 11c 11d 11e 11f 11g 11h 11i 11j 11k 11l 11m 11n 11o 11p 12b 12c 12d 12e 12f 12g 12h 12i 12j 12k

Figure LANDSAT5-TM LANDSAT5-TM LANDSAT7-ETM LANDSAT7-ETM LANDSAT5-TM LANDSAT7-ETM LANDSAT7-ETM LANDSAT7-ETM LANDSAT7-ETM LANDSAT7-ETM LANDSAT7-ETM ERS2-SAR − ERS2-SAR − ORBVIEW2-SEAWIFS ORBVIEW2-SEAWIFS ORBVIEW2-SEAWIFS ORBVIEW2-SEAWIFS LANDSAT5-TM ERS-SAR + LANDSAT5-TM LANDSAT5-TM ERS-SAR + LANDSAT5-TM LANDSAT5-TM ERS-SAR + LANDSAT7-ETM LANDSAT5-TM

Satellite/sensor 227-90 227-90 227-89 227-89 227-89 227-89 227-89 227-89 227-89 227-89 227-89 27269-6327/3645 28271-6345/6327 S2003121151853 S2002206151104 S2002150155022 S2004252155502 226-97 21335-4689 227-97 227-97 25615-4689 227-97 227-97 12818-4689 227-97 227-97

Path/row

14/01/2002 14/01/2002 15/08/1998 11/09/1999 31/10/1997 25/01/2000 27/09/1999 30/05/2002 16/10/2003 19/09/2002 02/06/2003 08/07/2000 16/09/2000 01/05/2003 25/07/2002 30/05/2002 08/09/2004 30/03/1997 20/05/1999 15/10/1997 15/10/1997 14/03/2000 15/09/1997 22/04/1997 02/10/1997 02/10/2001 15/09/1986

Date

Table 13.1 satellite, sensor, path, row and date of the images corresponding to each figure. Types of orbit for ERS satellite are indicated with + for descending and – for ascending mode. The time of passage over the studies sector in both mode are approximately between 13:53–14:00 and 03:30–03:37 UT respectively. LANDSAT always pass in descending mode between 13:16 and 13:23 UT

254 D.A. Gagliardini

Satellite/sensor

LANDSAT7-ETM ERS2-SAR − LANDSAT5-TM ERS2-SAR + ERS2-SAR + ERS2-SAR + ERS2-SAR + ERS2-SAR + ERS2-SAR +

Figure

7b 7c 8b 8c 8d 8e 9b 9c 9d

227-89 26267-6327 221-84 24670-4364 13104-4293 25901-4365 26631-4311/4329 25901-4347/4365 28134-4311

Path/row 21/12/2001 29/04/2000 24/12/1997 08/01/2000 22/10/1997 03/04/2000 24/05/2000 03/04/2000 06/09/2000

Date 13b 13c 13d 13e 14b 14c 14d 15b 15c

Figure

Table 13.1 (continued)

LANDSAT7-ETM ERS-SAR + LANDSAT7-ETM ERS2-SAR − ERS2-SAR + ERS2-SAR + ERS2-SAR + ERS2-SAR + ERS2-SAR +

Satellite/sensor

227-89 27891-4437 227-89 14973-6345/6327 331130-4563 331130-4563 27891-4563 21335-4689 21335-4689

Path/row

17/01/2000 20/08/2000 17/01/2000 02/03/1998 21/08/2001 21/08/2001 20/08/2000 20/05/1999 20/05/1999

Date

13 Medium Resolution Microwave, Thermal and Optical Satellite Sensors 255

256

D.A. Gagliardini

2 Study Area Marine coastal environments extend from the coastline to the outer edge of the continental shelf (Johannessen, 2000). The region considered in this study, ranging from the boundary between Uruguay and Brazil (33◦ 45 S) to the Atlantic coast of Tierra del Fuego (55◦ S) (Fig. 13.1a), spreads along approximately 200 km of Uruguayan and 4,700 km of Argentinean coastlines. Given the large extension of the region of interest, the resolution and path width of the sensors used, three sectors were selected to illustrate and analyze characteristics of dynamical processes (Fig. 13.1a–e).

2.1 Sector 1 Sector 1 (S1, Fig. 13.1a and b) ranges latitudinally from 33◦ 45 S to 40◦ S and longitudinally from 58◦ W to nearly 50◦ W. The La Plata River, receiving freshwater from the second largest South American basin, is 320 km long; its width increases from 38 km in the upper region to 230 km at the mouth. A turbidity front (e.g., Fig. 13.2b) is created at the inner border of its external zone by the flocculation of suspended matter at the tip of the salt wedge, and by the resuspension of sediments due to tidal currents friction at the bottom (Framiñan and Brown, 1996). The discharge of freshwater into the Southwestern Atlantic is approximately 23,000 m3 s−1 , inducing a large scale buoyant plume that during winters extends about 1,000 km beyond the estuary, spreading along the coasts of Argentina, Uruguay and Brazil (Piola and Romero, 2004). The dynamics of the oceanic zone of this area are strongly

Fig. 13.2 Turbidity fronts in S1: (a) Map of S1 indicating the observed areas; (b) La Plata River Turbidity Front (upwelling image); (c) Brazil-Malvinas Confluence Turbidity Front (upwelling image). Arrows indicate fronts

13


257

influenced by the encounter of the Brazil Current, carrying warm, saline, and oligotrophic waters, with the Malvinas (Falkland) Current, characterized by low salinity, cold, and nutrient-rich Sub-Antarctic waters. This encounter (Fig. 13.2c) is referred to in the literature as the Subtropical Convergence (Deacon, 1937) or the BrazilMalvinas Confluence (Olson et al., 1988). Multiple meanders, eddies, and filaments (Legeckis and Gordon, 1982; Gordon, 1989) result in a large mixing zone, with thermal fronts that exceed 8◦ C over a distance of a few kilometers. This confluence and its zone of influence is a highly dynamical region that constitutes an important biogeographical boundary between organisms of subtropical and Sub-Antartic origin (Bogazzi et al., 2005). This sector overlaps the Common Argentina/Uruguay Fishing Zone (Fig. 13.1b), of particular economical relevance due to the presence of commercially valuable fish and high diversity of invertebrates, as well as breeding colonies of marine mammals.

2.2 Sector 2 Sector 2 (S2, Fig. 13.1a, c, and d), located off the coast of northern Patagonia, ranges approximately from 40◦ 30 S to 47◦ 30 S. It encompasses a wide variety of environments, including gulfs, bays, islands, islets, low banks, mud flats, cliffs, sandy beaches and rocky shores. The shoreline of this area is characterized mainly by the presence of San Matias (18,000 km2 ), San Jose (814 km2 ), Nuevo (2,200 km2 ) and San Jorge Gulfs (39,340 km2 ). The Valdes Peninsula, connected to the mainland trough the Ameghino Isthmus, is also a conspicuous feature (Fig. 13.1c) of great significance for the conservation of marine wildlife. It consists of an extensive (3,600 km2 ) plateau of arid land with some salt lakes that extends 100 km into the Atlantic Ocean. The main oceanographic characteristics are a general northeastward circulation and a large amplification of oceanic semidiurnal tides towards the coast, with high energy dissipation. This dissipation, caused by the turbulent mixing generated by the interaction of tidal currents over the coastal shallow topography, forces larger thermal fronts at the external zone of San Matias Gulf, northeast of Valdes Peninsula, and at both ends of the San Jorge Gulf (Bogazzi et al., 2005). There are also some minor fronts originated by the interaction of tidal currents with the internal water of San Jose Gulf and with elevated or depressed sea bed irregularities such as islands, headlands, canyons or broken coastlines (Gagliardini et al., 2004a). All these fronts give rise to highly productive coastal waters (Carreto et al., 2007), important for regional and local economies and for the well-being of the marine life. Many species of seabirds, marine mammals, migratory birds, fishes, mollusks, and crustaceans breed and forage along this coastal region. One of the peculiarities of San Matias Gulf is the presence of extensive fields of sand waves at its entrance (Pierce et al., 1969; Gagliardini et al., 2005). Off its north coast, where tidal currents may reach 1 ms−1 , there are sand waves with an average height of 4 m (some may reach up to 7 m) and between 80 m and 240 m length (Achilli and Aliotta, 1992). In the southern region of the Gulf, currents can reach

258

D.A. Gagliardini

almost 2 ms−1 and sand waves are much higher, usually referred to as “giant waves”. Some of them can reach 17 m height and a length of 680 m; average height is approximately 10 m.

2.3 Sector 3 Sector 3 (S3, Fig. 13.1a and e), much smaller than S1 and S2, covers just the San Sebastian Bay, a shallow and protected coastal environment located on the Atlantic coast of Tierra del Fuego, between 53◦ S and 53◦ 20 S. A wide valley formed by glacier activity during the Pleistocene was reshaped by the sea into the present semicircular configuration of the bay, 55 km long and 40 km wide. At the north the bay is closed by a 17 km long gravel arrow spit named El Paramo (Fig. 13.1e), formed by wave-induced southward transport of sand and gravel. The spit shelters intertidal mudflats (Fig. 13.1e) up to 10 km wide, cut by meandering tidal channels that reach a depth of over 3 m and can be up to 50 m in width (Isla et al., 1991). The unprotected southern coast is steeper and narrower, with coarser sediment and boulders (López Gappa and Sueiro, 2006). Winds from the west, with velocities that exceed 65 km h–1 during more than 200 days per year, together with longshore currents flowing south, are the main dynamic drivers. Tidal range varies from 3.2 m to 10.5 m forcing a clockwise gyre towards the north (Vilas et al., 1986–1987). From an ecological point of view, San Sebastian Bay is a prime habitat for many species of birds and a place for non-reproductive congregations of migratory birds, particularly charismatic species that visit the area during summer for feeding (Yorio, 1998). The lack of a harbor and the muddy nature of the coast impose constrains on research and fishing activities. The benthic invertebrate biota is dominated by mussels and barnacles, and pinnipeds and cetaceans have been observed inside the bay. San Sebastian bay lies in the most important hydrocarbon extraction zone of Tierra del Fuego, surrounded by hundreds of oil wells (López Gappa and Sueiro, 2006). Oil is loaded to tankers from a buoy in the southern part of the bay, often under rough sea conditions. An oil spill is always a potential risk that could have long-lasting ecological effects (Malagnino et al., 1994). Therefore, understanding the dynamic of the coastal zone is important for planning and developing contingency plans.

3 Data Sources Missions ERS1/2-SAR and LANDSAT-TM/ETM+ are presented in this section. Also, the most relevant characteristics of the SAR, optical and thermal data used in this chapter are discussed.

13


259

3.1 SAR Data ERS1-SAR was launched in a Sun-synchronous near-polar orbit in 1991 with a mean altitude of 785 km, while ERS2-SAR was placed in a similar orbit in 1995. Both SAR sensors were C-band radiometers (5.3 GHz) that illuminate the ground with a 100 km wide swath and a spatial resolution of approximately 25 m. These right-looking radars collect data in ascending and descending orbits and have a repeat cycle period of 35 days. Far exceeding its expected lifetime, the ERS-1 mission was ended on March 10, 2000 because of a failure in the on-board altitude control system. ERS-2 is still operational after 14 years in orbit, with all its instruments still functional and providing good data. SAR is a side-looking imaging radar that emits a series of microwave pulses towards the earth, in a direction perpendicular to its flight path. Images are builtup by computing the intensity and time-delay of the backscattered signals. These depend primarily on the roughness and dielectric properties of the surface under observation, and on its distance from the radar. In the case of the ocean, the surface roughness is imposed by wind-generated capillary waves and small surface gravity waves (in the order of the radar wavelength, i.e., centimeters) and the returned signal is due to the resonance between both type of waves in agreement with Bragg’s scattering theory (Valenzuela, 1978). Thus, SAR or microwave images, represent, as a first approximation, the sea surface roughness conditions given by the spatial distribution of these capillary and small ocean waves known as Bragg waves. Different oceanographic phenomena interacting with these waves generate a particular distribution of roughness or signature by which processes, such as currents, fronts, internal waves, eddies, natural biogenic surfactants and bathymetric signatures can be identified and studied. Some examples are shown in Figs. 13.3, 13.4, 13.5, and 13.6. Wind speeds of approximately 3–12 ms−1 provide the roughness conditions required for the observation of most ocean features in the microwaves region. At much lower wind speeds, wave effects decrease, the sea surface is smoother

Fig. 13.3 Turbidity fronts north of S2: (a) Map of S2 indicating the observed areas; (b) Chubut River plume front (upwelling image); (c) and (d) Negro River plume front (upwelling and microwaves images respectively). Arrows indicate fronts

260

D.A. Gagliardini

Fig. 13.4 Thermal front in S1. (a) Map of S1 indicating the observed areas; (b) Brazil-Malvinas Confluence Front (thermal image); (c), (d), and (e) zooms of the three areas indicated in 4b (SST images); (f) SST image of a thermal front; (g) sun glint image of the area covered by 4f; Arrows indicate fronts (h) SST image of a thermal front; (i) sun glint image of the area covered by 4f; (j) microwaves image of the area covered by 4h

Fig. 13.5 Thermal fronts in S2. (a) map showing the observed areas; (b)–(d) San Matias, Valdes and San Jose thermal fronts (SST images); (e) and (f) South and north of San Jorge Gulf thermal fronts (SST images); (g) Negro River thermal front (SST image). Arrows indicate fronts

13


261

Fig. 13.6 Natural biogenic surfactants in S1. (a) map of S1 indicating the observed areas; (b)–(h) surfactants (b) is a sun glint image and (c)–(h) are microwaves images)

reflecting energy away from the sensor, and the non returned signal generates dark areas in the microwave imagery. At higher wind speeds the affected area is characterized by a significant background cluttering that reduces the contrast of ocean signatures in the image. The distinct signatures caused by the modulation of capillary and short surface gravity waves due to ocean circulation are most evident under the wind speed range indicated earlier.

3.2 Optical Data The optical data used in this work were provided by the LANDSAT-TM and LANDSAT-ETM+ missions. The first was launched in a Sun-synchronous nearpolar orbit in 1984, with a mean altitude of 750 km, while the ETM+ was placed in a similar orbit in 1999 with an 8-day shift. The spectral characteristics of this sensor are three channels in the visible bands (0.45–0.52 μm, 0.52–0.60 μm, 0.63–0.69 μm), three in the near and middle infrared, (0.76–0.90 μm, 1.55–1.75 μm, 2.08–2.35 μm), and one in the far or thermal infrared (10.40–12.50 μm). All of them have a swath width of approximately 185 km and a 30 m spatial resolution, except the thermal channel which has a resolution of 120 m in TM and 60 m in ETM+. The latter also has an enhanced thermal band and a

262

D.A. Gagliardini

panchromatic channel (0.52–0.90 μm) with 15 m resolution. Unfortunately TM is approximately 25 years old and its sensors have been deteriorating with the consequent reduction in accuracy and reliability. In 2003, a problem arose in the Scan Line Corrector of the ETM+ sensor and as a consequence only the central part of the images (approximately 22 km wide) continued to have a quality comparable to that of earlier imagery. When the solar radiation reaches the sea surface it is separated into two components. A portion is reflected into the atmosphere and is referred to as “sun glint,” while the remainder is refracted as it penetrates the water. This last component interacts with water molecules and suspended matter; part is absorbed and part backscattered. The latter, known as upwelling radiance, emerges from the surface and is detected simultaneously with the specularly reflected component. Therefore, an oceanic optical image represents the spatial pattern of the detected electromagnetic energy made up by both, the upwelling and the sun glint radiances. As shown below, there are cases in which one or the other prevails; the respective images will be indicated as upwelling or sun glint images. The intensity of the detected upwelling radiance depends on the scattering and absorption processes of sun light in the ocean water and provides information on water constituents in the upper ocean layer. On the other side, the intensity of the reflected fraction is a function of the view angle of the sensor (geometry of observation), the sun azimuth and elevation angles (geometry of illumination), and the roughness of the water surface generated by wind (Hennings et al., 1994). Sun glint is very sensitive to the three indicated parameters. Sun glint reaches its maximum value when the sensor observes the side of the image closer to the sun and its minimum in the opposite side. When wind increases from 3 ms−1 to 12 ms−1 the roughness of the sea surface increases, and this component also increases sharply. Finally, when solar elevation and azimuth solar angles are important, sun glint dominates over water-leaving radiance. In the area under study this requirements are only satisfied around the summer solstice, when the sun elevation and azimuth angle for LANDSAT passages are respectively within the ranges 49–58◦ and 54–69◦ . When sun glint is not present, only the upwelling radiance is measured, therefore the upwelling image represents the distribution of suspended matter in the upper ocean layer. In the opposite case, when sun glint prevails, the sun glint image represents the distribution of sea surface roughness, as in the case of microwaves. Thus the retrieval of information about the in-water constituents is severely compromised, often impossible. Indeed, some authors refer to sun glint only as a factor contaminating radiation, and try to estimate and eliminate it to recover the information of interest. This is very common in ocean color studies and sea surface temperature (SST) determination with low resolution sensors (1.1 km at the nadir), such as SeaWiFS and MODIS (e.g., Wang and Bailey, 2001; Kumar, 2004; Ottaviani et al., 2008). This action may not be convenient in the case of TM/ETM+ data because, as it will be shown later, the resolution of this sensor (30 m) permits, similarly to SAR, the observation of a large variety of meso-scale ocean processes through sun glint detection.

13


263

3.3 Thermal Data Thermal data used here in this work were those obtained by the channel 6 of TM and ETM+. Atmospheric gases, mostly water vapor, absorb part of the radiation emitted by the sea surface leading to an underestimation of the surface temperature. An accurate correction for atmospheric absorption is difficult in the case of LANDSAT images because they have only one infrared thermal channel, and given the lack of simultaneously collected field data are lacking. Therefore, the Brightness Temperature (BT) was estimated from channel 6 of the TM/ETM+ sensor using the Planck equation to convert measured energy to temperature, assuming a coefficient of water emissivity of 1, and that atmospheric effects are homogeneous over the study area. The corresponding images are indicated as thermal images. Even though the SST calculated in this case is underestimated, its spatial distribution and differences between locations are correct.

4 Observations of Medium-Small Scale Hydrodynamic Processes Identified with Medium Resolution Sensors This section addresses the contribution of microwaves, upwelling, sun glint and thermal images to the identification of ocean features generated by medium-small scale hydrodynamic processes identified in the three selected sectors over the Argentinean coast.

4.1 Fronts Fronts in the ocean can be described as sharp boundaries generated by the encounter of water masses with different properties. They are zones of transition characterized by strong gradients in water properties, generally temperature, density or both. Fronts are usually vertically inclined interfaces between two water masses, where nutrient rich waters are moved up, enriching the photic zone and thus enhancing primary production. If the front is sufficiently long-lived, populations of herbivorous zooplankton will increase, promoting secondary production (Acha et al., 2004). High food availability at frontal zones attracts nektonic organisms (fish, squid, etc.), transferring the energy to higher trophic levels, such as large fish, seals, whales and dolphins (Acha et al., 2004). Benthic invertebrates also take advantage of primary production and detritus generation in the photic zone (Largier, 1993; Mann and Lazier, 1996). In the costal zone, strong convergence velocities frequently accumulate floating matter along the convergence line. Flotsam includes detritus such as dust, foam and timber (Bowman, 1978). Fronts can be classified based on the process that originates them. According to Acha et al. (2004) in the Argentinean coastal zone there are fronts associated with

264

D.A. Gagliardini

a variety of phenomena: currents convergence, tides, the shelf-break, upwelling, plumes, and geomorphic features such as headlands and islands. Most of these types are found in S1 and S2, being the reason of their biological productivity and high biodiversity. Another way to classify fronts is taking into account the properties of the interacting water masses, e.g., turbidity, temperature and salinity. Currently only turbidity and thermal fronts can be detected by medium resolution optical and thermal sensors, due to sharp changes in the upwelled and emitted radiances. The La Plata River turbidity front is the main turbidity front in the whole study area. It is forced by the interaction between very turbid fresh waters and ocean water. The location of its north extreme oscillates along the Uruguayan coast, while the southern portion takes different shapes over Sanborombon Bay. The complete front can be observed in Fig. 13.2b, captured with two TM frames. A less intense turbidity front is shown in Fig. 13.2c, where the interaction of part of the confluence BrazilMalvinas can be observed. Turbidity plume fronts are shown in Fig. 13.3b and c, where the turbid freshwaters from Negro and Chubut Rivers encounter ocean water. This type of front can also be identified in SAR images due the difference in Bragg waves originated by dissimilar water density on each side of the front (Fig. 13.3d). Different thermal fronts observed in Sector 1 are illustrated in Fig. 13.4. Cold water is represented in light tones and warm water in dark tones. Figure 13.4b show part of the convergence front generated by the Brazil-Malvinas confluence. In this case the difference in SST between the cold and warm water masses can reach values of up to 8◦ C within a distance of a few kilometers, Fig. 13.4c–e zoom into the areas indicated in Fig. 13.4b. Highly dynamic and complex interactions can be observed in detail, with water being interchanged through an intricate array of intrusions, eddies, rings, and filaments. The resolution of this sensor allows the clear observation of cold water eddies of less than 20 km in diameter (Fig. 13.4e). Different thermal fronts observed in Sector 1 are illustrated in Fig. 13.4. Cold water is represented in light tones and warm water in dark tones. Figure 13.4b show part of the convergence front generated by the Brazil-Malvinas confluence. In this case the difference in SST between the cold and warm water masses can reach values of up to 8◦ C within a distance of a few kilometers, Fig. 13.4c–e zoom into the areas indicated in Fig. 13.4b. Highly dynamic and complex interactions can be observed in detail, with water being interchanged through an intricate array of intrusions, eddies, rings, and filaments. The resolution of this sensor allows the clear observation of cold water eddies of less than 20 km in diameter (Fig. 13.4e). Thermal fronts are not only observable using thermal channels but also with optical ones. One way is detecting the presence of a white line formed by the accumulation of foam or other buoyant materials along the boundary between the two water masses with different temperatures (Fig. 13.4i). For example, Fig. 13.4f and g show a case where this difference is of 1◦ C. Another way is by the detection of sun glint, the same way as with microwave backscattering. A shear of velocity currents occurs at the frontal zone, increasing surface roughness and altering Bragg waves. This fringe of enhanced roughness produces also an increase in sun glint or microwave backscattering. As a consequence, linear features can be

13


265

observed in both types of images at the same place where the thermal front appears in the thermal image (Gagliardini and Clemente-Colón, 2004a, b). This can be observed in Fig. 13.4h–j, where subsets of nearly coincidental LANDSAT-TM and ERS-SAR images (taken 25 min apart from each other) are shown. Bright lines appear in Fig. 13.4i and j, matching the spatial location of thermal fronts observed with thermal images (Fig. 13.4h). Both cases illustrate how thermal images can contribute to the interpretation of sun glint and microwave images. In the case of SAR, the incidence angle of microwave pulses is kept constant (23◦ at the scene centre) and the sea surface is always observed with the same illumination geometry. The situation is very different for TM because the position of the sun, and therefore its illumination geometry, changes along the day and the year and, as a consequence, the prevalence of the upwelling or sun glint radiances changes over time. In particular, over the study region sun glint prevails during summer. This can be observed comparing Figs. 13.2c and 13.4i, both images obtained over the same area but on different dates. The first was obtained on September 6, 1998, at 12:55 UTC with solar elevation and azimuth angles of approximately 34◦ and 47◦ , and the second on November 12, 1999, at the same time but with solar elevation and azimuth angles of approximately 54◦ and 68◦ . The first one, an upwelling image, shows typical upper layer structures of ocean color, providing information about both the distribution of suspended material and upper circulation patterns. The second shows clearly a sun glint image of a thermal front (strikingly similar to the SAR image, Fig. 13.4j), captured at the same place 25 min later (Gagliardini and Clemente-Colón, 2004a). Figure 13.5b–d show, respectively the San Matias, Valdes and San Jose thermal fronts. The last two are forced by turbulent mixing of tidal currents over a shallow coastal topography. Fronts start developing in the early spring, as the seasonal thermocline develops, and persist through the autumn, as stratification vanishes (Bogazzi et al., 2005). They reach their maximum intensity in summer, when the steepness of the gradient is maximal. The maximum difference in SST between the two interacting water masses, reached in January-February, is in the order of 2.5◦ C for the San Matias and Valdes fronts, and 1.5◦ C for the San Jose front. The San Jorge, Chubut and Rio Negro coastal thermal fronts, which are also clearly observable in the October image, are shown in Fig. 13.5e–g. They are forced by a combination of circulation, coastal topography and strong tidal currents. The difference in SST between the cold and warm water masses reaches a maximum of approximately 1.5◦ C during summer, in both cases. Figure 13.5f shows the intense turbulence associated with the upwelling of cold water, much more turbid than the water in its surroundings.

4.2 Natural Biogenic Surfactants Plankton and fish release natural biogenic surfactants into the ocean. Turbulence, such as that created by eddies and upwelling, leads to convective motions that can bring this organic material to the surface, where it may remain as a microlayer or

266

D.A. Gagliardini

natural surface film (Vesecky and Stewart 1982). As the concentration of surfactant molecules on the surface film increases, it generates a surface tension that inhibits the development of Bragg waves (Johannessen et al., 1994; Clemente-Colón and Yan, 1999, 2000). In consequence, their presence will be observed in microwave or sun glint images as dark filaments, or as dark patches with filamentary borders caused by wind effects. According to Clemente-Colón and Yan (2000), surfactants at the surface become undetectable under high speed wind conditions (over 6 ms−1 ) because they tend to mix down into the water column, and when wind speed is below 3 ms−1 because water surface does not cause backscattering of incident radiation. Natural biogenic surfactants are common in Sector 1 and can be clearly observed in Fig. 13.6 where a variety of examples on this sector can be observed. This can be associated with the high productivity of the region. In Fig. 13.6b–g surfactants are observed as black filaments that cover the whole area describing the water surface movements while in Fig. 13.6 h they are represented by a black patch. Surfactants in Fig. 13.6 g and h are surrounded by a white windy area. In Sector 2 (Fig. 13.7a), on the contrary, natural biogenic surfactants are not as frequent as in Sector 1, having been observed only two times. Figure 13.7b and c show the presence of both events, in San Matias and San Jose Gulfs, where their occurrence matches relatively productive areas.

Fig. 13.7 Natural biogenic surfactants north of S2: (a) map of north S2 indicating the observed areas: (b)–(c) Natural biogenic surfactants in San Matias and San Jose Gulfs (microwaves images)

4.3 Internal Waves Internal waves consist of solitary wave trains generated by the interaction between tidal currents and abrupt topographical features, propagating near the surface in stratified zones (Johannessen et al., 1994). The hydrodynamic interaction of such waves with the roughness of the surface gives rise to a sequence of alternating fringes of convergence and divergence, which are alternating strips of rough and smooth sea surface (Alpers, 1985). Their signature in microwave and sun glint images is an alternation of light and dark fringes. When organic substances that

13


267

Fig. 13.8 Internal waves in S1: (a) map of S1 indicating the observed areas: (b)–(e) internal waves ((b) is a sun glint image, (c)–(e) are microwaves images)

strongly reduce the amplitude of the Bragg waves are trapped in the convergence zone, their signature consists of dark stripes over a uniformly bright background. Internal waves are very common in Sector 1 and begin to be observed in spring when the thermocline begins to develop, some examples are presented in Fig. 13.8. It can be observed that they appear in small packets of waves with their constant phase lines (troughs and crests) somewhat curved radiating from a nearby source. Secondary effects associated with internal waves include the transport of water along with suspended material such as sediments, nutrients and larvae, as well as contaminants. Such is the case of internal tidal bores that play an important role in the transport of larvae and other organisms to the near shore, influencing the development of benthic communities (Leichter et al., 1996). Moreover, high frequency internal waves have been shown to influence the spatial distribution of plankton, effectively controlling nutrient dispersal in coastal regions (Leichter et al., 1996). Also, they are correlated with the transport of pelagic larvae of benthic invertebrates and fish (Franks, 1997).

4.4 Eddies, Vortex Dipoles and Jets Eddies are circular currents running reversely to the main current. They can induce upward displacement and inject nutrients into the euphotic zone, which results in the accumulation of phytoplankton biomass in the overlying water (e.g., Pingree et al., 1979; McGillicuddy et al., 1999; Crawford et al, 2007; Mahadevan et al., 2008). They are usually manifested in SAR images as a result of the interactions between eddy currents and Bragg waves, outlining their spiral shape. Figure 13.9b–d represent spiral eddies formed in Sector 1, inside and outside La Plata River. They can also be indirectly revealed by the presence of a natural film of sediments trapped within spiraling lines associated with the orbital motion as it is shown in Fig. 13.9d. Eddies can also be observed in optical and thermal images as a signature of the presence of turbid or cold water in the spiral circulation of clear or hot water

268

D.A. Gagliardini

Fig. 13.9 Eddies in S1: (a) map of S1 indicating the observed areas: (b)–(d) eddies (microwaves images)

Fig. 13.10 Eddy in north S2: (a) map of north S2 indicating the observed area: (b)–(c) eddy in Nuevo Gulf (upwelling and thermal images)

respectively, or vice-versa. An eddy with a diameter of approximately 45 km can be observed in Sector 2, in Nuevo Gulf through optical and thermal TM channels (Fig. 13.10b and c). Eddies of approximately 4 km of diameter or less can be clearly observed with optical channels on the west side of San Jose Gulf. They are revealed by turbulent fluxes associated with the intrusion of tidal currents, and their interaction with the sea floor and the south coast of the gulf. Besides, smaller eddies are clearly seen outside the gulf in Fig. 13.11b–h, including, self-propagating vortex dipoles and tidal jets. Some of these patterns are also observed in thermal images (Fig. 13.11i and j) and microwave images (Fig. 13.11k and l). The formation of these structures outside the San Jose Gulf and inside the San Matias Gulf, which is deeper than 100 m, is a consequence of the tidal flow passing through the narrow mouth of the gulf. Under these conditions the vorticity produced can lead to the generation of jets and two large counter-rotating vortices that form a coherent dipole structure. This structure can self-propagate away from the channel transporting suspended material over long distances. Inside the gulf the generation of these structures is related to the interaction between tidal currents and San Jose Gulf mouth and topography during tidal cycles (R. Amoroso and D. A. Gagliardini, preprint, 2009).

13


269

Fig. 13.11 Eddies, vortex dipoles, tidal jets inside and outside San Jose Gulf: (a) map of north S2 indicating the observed areas; (b)–(d) small eddies observed in the west side and outside the gulf; (e) jets; (f)–(h) jets, self-propagating dipoles and eddies (upwelling images); (i)–(j) self-propagating dipoles and eddies (SST images); (k)–(l) jets and eddies (microwaves images); (m)–(p) similar structures as those shown in panel b-l but with an optical low resolution sensor (SeaWiFS)

Similar scenarios to those shown in Fig. 13.11b–l are also observable with an optical sensor of lower resolution such as SeaWiFS (Fig. 13.11m–p). However, the signature of the structures is diffuse and difficult to interpret in low resolution images alone, although not when they are analyzed together with images obtained with medium resolution sensors. This approach can be very useful for completing temporal data series when data is scarce, or to assist in the interpretation of some structures not clearly observed with low resolution sensors.

270

D.A. Gagliardini

Fig. 13.12 Eddies, vortex dipoles, and tidal jets at San Sebastian Bay: (a) map of S3; (b)–(c) eddies formed in the north of the bay (upwelling and microwave images) obtained during flood tide; (d)–(e) upwelling images showing meandering channels and their associated jets, generated during ebb tide; (f) microwave image of the scenario seen in 12e; (g)–(k) images showing the generation of a vortex dipole, (g)–(h) and (j)–(k) upwelling images and (i) microwave image

Figure 13.12b and c are upwelling and microwave images respectively showing the semicircular configuration of San Sebastian bay in S3, separated from the adjacent shelf by a long gravel arrow spit (El Paramo). Both images were obtained during flood tide. The mud flat, covered during the flood, is clearly observable in these images with different tonalities, revealing differences in humidity and grain size. Figure 13.12b clearly shows the trajectory of incoming tidal currents described by a dark fringe that generates an eddy in the north area of the bay, also seen in Fig. 13.12c through the dark line generated by biogenic surfactants or an oil dump.

13


271

Figure 13.12d shows the meandering channels (Isla et al., 1991) and their associated jets, generated during ebb tide. It can be seen that these jets point towards the north in the south sector, indicating the northward direction of alongshore currents. This orientation is lost towards the end of the bay, where the outflow becomes perpendicular to the mud flat, following the direction of the currents parallel to the spit. Figure 13.12e is a zoom into Fig. 13.12d indicated by a rectangle showing the different shapes of the jets, with their sinuosity influenced by the depth and width of the channels. Figure 13.12f is a microwave image of the scenario seen in Fig. 13.12e, showing clearly the shape of these channels. Figure 13.12g, i–k (upwelling images) show a pattern similar to that described earlier for San José Gulf during the generation of a vortex dipole, in this case produced by the interaction of the ebb current with the El Paramo spit.

4.5 Bathymetric Signatures Bathymetric signatures are imaged by backscattered microwaves or sun glint when the interaction of ocean currents with bottom topography produces variations in surface current velocities, therefore modulating the Bragg waves. When a tidal current approaches a linear sand bank or sand wave, an increase in speed occurs over the ascending side flowing over the top of the irregularity. This process generates divergence and a decrease in the amplitude of the Bragg waves, whereas a decrease in speed in the descending side is produced and the inverse effect is generated. This implies first a reduction and then an increase in the backscattered intensity. In microwave or sun glint images, this results in narrow bright and dark linear features over a uniform background, as in the case of internal waves. (Gagliardini et al., 2005) An inverted sequence of the bright and dark patterns is observed when the direction of tidal currents is reversed. The opposite effect will be observed for depression features. In both cases this results looks like a 3d image of the bottom irregularities. Bathymetric features observed at the entrance of San Matias Gulf are shown in Fig. 13.13. Figure 13.13b (sun glint image) and c (microwaves image) reveal small sand dunes located north of the entrance, at 20–40 m depth. Notice that Fig. 13.13b shows more features than Fig. 13.13c. The reason is that optical images do not suffer from the speckle noise characteristic of coherent SAR observations, allowing a better observation of weaker features. Under this condition, the optical sensor can help expanding the results obtained with SAR sensors (Gagliardini et al., 2004b). Sun glint and microwave images are shown in Fig. 13.13d and e respectively, where “giant” sand dunes located at 40–60 m depth can be visualized. It can be clearly seen that both images show the same features. This is due to the fact that currents are more intense and the dunes are higher and larger. The difference between the bright-dark sequences is due to the fact that one image was registered during ebb tide (Fig. 13.13d) and the other during flood tide (Fig. 13.13c). At the right side of these images an elevation of 18 m is observed as a dark fringe during ebb tide and as a bright fringe during flood tide.

272

D.A. Gagliardini

Fig. 13.13 Bathymetric features at the San Matias Gulf entrance: (a) map of north S2 showing the areas covered in figures (b)–(e): (b)–(c) sun glint and microwave images of small sand dunes located north of the gulf entrance; (d)–(e) sun glint and microwave images of giant sand dunes and an elevation located east of the gulf entrance (double lined black arrows indicate the tidal current directions; doted black arrow indicates the same point in both images, white arrow indicates the elevation located at east)

Fig. 13.14 Bathymetric features at the San Jorge Gulf entrance: (a) map of south S2 with the areas covered in figures b, c and d; (b) microwaves image of the elevation of 40 m above the sea floor indicated with arrows; (c)–(d) microwaves images of ripples (both arrows indicating the same features) localized at the edges of the elevation on different dates ((c), zoom of the area indicated in Fig. 13.14b)

Figure 13.14 shows bathymetric features at the entrance of San Jorge Gulf. An elevation of 40 m above the sea floor, with the top at a depth of 6 m can be observed in Fig. 13.14b. Figure 13.14c and d show ripples located along this elevation. Figure 13.15 shows bathymetric features identified at the entrance of San Sebastian Bay. In Fig. 13.15b the most remarkable structure is the fringe of a white

13


273

Fig. 13.15 Bathymetric features at San Sebastian Bay: (a) map of S3; (b) microwave image of the bay during flood; (c) zoom of the area squared in b

signal (observable on the right side) caused by an intense microwave backscattering generated by strong sea surface roughness. This is the result of the flow of tidal currents over remains of moraines with its highest point located at a depth of four meters. The shape of these topographical features strongly influences the inflow and outflow of water during ebb and flood tides. Figure 13.15c is a zoom of a sector indicated on Fig. 13.15d, where the ripples located over the moraines remain, and the bright-dark sequence due to the flood tide can be seen in detail. Some ocean waves represented by lines perpendicular to the bathymetric structure can be seen in the north, indicating the inflow of ocean water.

5 Discussion and Conclusions This contribution shows how multi-sensor medium resolution data in the microwave, optical and thermal ranges can contribute to reveal and monitor dynamical patterns associated with different coastal environments. Among those are thermal fronts of intermediate and large size, illustrated by the San Matias and San Jorge Gulfs and the Brazil-Malvinas convergence, and plume fronts such as the La Plata, Negro and Chubut Rivers. Although some of these frontal systems had been previously observed with low resolution satellites, they had never been scrutinized before with the level of detail shown here. On the other hand, the resolution of these systems allowed the identification, for the first time, of features associated with various relatively small-scale processes, such as jets, eddies and upwelling. Besides, detection of sun glint and microwave backscatter made it possible to observe other fronts, upwelling, internal waves, eddies and seabed irregularities with similar resolution. Therefore, the two medium-resolution satellite systems considered here constitute powerful tools for characterizing coastal environments, providing significant information about a large variety of ocean dynamic processes. The use of these data, however, requires considering some constrains inherent to the necessity of daylight and cloud-free conditions for optical and thermal sensors, as well as specific wind

274

D.A. Gagliardini

conditions for microwave sensors. This limitation implies that it is not always possible to obtain information on the right place and right moment. This situation has been overcome in this study because of the large number of images available in the CONAE’s archives, started in 1997. This circumstance makes it possible to conduct long-term studies (e.g. annual variations in SST), and also within very short but recurrent periods (e.g. tidal cycles). Currently, in addition to the ERS-SAR and LANDSAT-TM/ETM+, other satellite systems can provide information similar to that used in this chapter, expanding the contributions of medium resolution sensors in this type of work and increasing the possibility of getting more data in a short period of time. Besides the LANDSAT-TM/ETM+, there are seven other optical medium resolution systems, which together form the CEOS (Committee on Earth Observation Satellites) Land Surface Imaging Portal Constellation (http://wgiss.ceos.org/lsip/overview.shtml). These missions, operated by nine space agencies, work in visible, near-infrared, shortwave infrared and thermal infrared wavelengths with spatial resolution between 10 m and 100 m. In addition to ERS2-SAR, there are microwave satellite systems launched some years ago such us Envisat-ASAR and RADASAT. A large variety of satellite systems have also been recently launched, among them, the Japanese ALOS, the German TerraSAR-X, the Chinese Huangjing-1C SAR, and the ItalianArgentine Earth Observation Disaster Management System (SIASGE), a satellite constellation which will be integrated by four SAR, two optical Italian satellites and two SAR Argentinean satellites. Also, the integration of medium and low resolution images has to be considered as a realistic possibility in order to obtain data at a higher temporal frequency. Low resolution sensors, such AVHRR and Sea-WiFS, can obtain more than one image per day, covering an area much larger than that observed by the medium resolution sensors but with a coarse resolution of 1.1 km. Hence, small features are vaguely seen, and additional information is needed for their interpretation. They could become meaningful, however, if the same feature is also detected with higher resolution sensors with similar characteristics, like TM /ETM+ and ERS-SAR, which can assist the analysis of low resolution data and the interpretation of diffuse patterns. Once this is achieved, tracking the evolution of phenomena under study will be also possible with low resolution data. Besides, as the latter cover a much larger area, it is possible to identify the relationship of the phenomenon of interest with the surroundings, and even with large-scale processes. Therefore, the combined analysis of both types of data enhances the information content of each category of data. Finally, it should be stressed that the results presented here demonstrate that medium-resolution optical, thermal and microwave sensors can identify and provide significant information about a large variety of little or completely unknown small-scale dynamical processes. These sensors, with their ability to detect sediment distribution, surface temperature and surface roughness, offer different types of information complementing, enlarging and reinforcing each other’s usefulness, independently of the characteristics of the study area. The same happens when this type of data are used together with low resolution data. This is of particular

13


275

importance in highly productive waters, where the information provided by these sensors allows researchers to relate oceanographic processes with regional ecology. Acknowledgments I wish to thank my colleagues MSc Haydée Karszenbaum, Dr. Lobo Orensanz and Lic. Ricardo Amoroso for the critical reading of the drafts of this chapter. Their detailed comments and suggestions have contributed greatly to the final version. I am grateful for the assistance of many colleagues but particularly my research assistants Paula Giudici and Lic. Nora Glembocki. Also, I acknowledge PhD DanLing (Lingzis) Tang and PhD Paula for their confidence and patience in waiting for the final manuscript. I would like to express my gratitude to the National Commission for Space Activities and the European Space Agency for providing the images used in this work, also the National Council on Scientific and Technical Research and the National Agency for Scientific and Technological Promotion for the financial support.

References Acha EM, Mianzan HW, Guerrero RM, Favero M, Bava J (2004) Marine Fronts at the continental shelves of austral South American physical and ecological processes. J Mari Sci 44:83–105 Achilli SM, Aliotta S (1992) Características morfológicas de ondas de arena de la plataforma continental al sur de la provincia de Buenos Aires, Argentina. Proceedings of the 3ras Jornadas Geológicas Bonaerenses, La Plata, Buenos Aires, Argentina, pp. 207–212 Alpers W (1985) Theory of radar imaging of internal waves. Nature 314:245–247 Armstrong RA, Gilbes F, Guerrero R, Lasta C, Benavidez H, Mianzan H (2004) Validation of SeaWiFS-derived chlorophyll for the Rio de la Plata Estuary and adjacent waters. Int J Remote Sens 25(7–8):1501–1505 Bogazzi E, Baldoni A, Rivas A, Martos P, Reta R, Orensanz JM, Lasta M, Dell P’Arciprete, Werner F (2005) Spatial correspondence between areas of concentration of Patagonian scallop (Zygochlamys patagonica) and frontal systems in the southwestern Atlantic. Fish Oceanogr 14(5):359–376 Bowman MJ (1978) Introduction and historical perspective. In: Bowman MJ, Esaias WE (eds) Oceanics fronts and coastal processes. Springer, New York, pp 2–5 Carreto JI, Carignan MO, Montoya NG, Cucchi Colleoni AD (2007) Ecología del fitoplancton en los sistemas frontales del Mar Argentino. Publicaciones Especiales INIDEP No 1393, El Mar Argentino y sus recursos, Mar del Plata, Argentina 5, pp 11–31 Clemente-Colón P, Yan XH (1999) Observations of east coast upwelling conditions in synthetic aperture radar imagery. IEEE Trans Geosci Remote Sens 37(5):2239–2248 Clemente-Colón P, Yan XH (2000) Low-Backscatter ocean features in synthetic aperture radar imagery. Johns Hopkins APL Tech Dig 21:116–121 Crawford WR, Brickley PJ, Thomas AC (2007) Mesoscale eddies dominate surface phytoplankton in northern Gulf of Alaska. Prog Oceanogr 75(2):287–303 Dogliotti AI, Scholss IR, Almandoz GO, Gagliargini DA (2009) Evaluation of SeaWiFS and MODIS chlorophyll-a products in the Argentinean Patagonian Continental Shelf (38o S–55o S). Int J Remote Sens 30:251–273 Deacon GER (1937) The hydrology of the southern ocean. Discov Rep 15:1–124 Framiñan MB, Brown OB (1996) Study of the La Plata Turbidity Front, Part 1: spatial and temporal distribution. Cont Shelf Res 16(10):1259–1282 Franks PJS (1997) Spatial patterns in dense algal blooms. Limnol Oceanogr 42(5 part 2): 1297–1305 Gagliardini DA, Aliotta S, Dogliotti AI, Clemente-Colón P (2005) Identification of bed forms through ERS-SAR images in San Matias Gulf, Argentina. J Coast Res 21(1):193–201 Gagliardini DA, Amoroso RO, Dell’Arciprete P, Yorio P, Orensanz JM (2004a) Detection of smallscale coastal oceanographic processes through LANDSAT-TM/ETM+ images: implications for

276

D.A. Gagliardini

the study of biological processes along the patagonian coasts of Argentina. Int J Biodivers Oceanol Conserv 68(2):194–200 Gagliardini DA, Clemente-Colón P (2004a) A comparative assessment on the use of SAR and high resolution optical images in ocean dynamics studies. Int J Remote Sens 25(7–8):1271–1275 Gagliardini DA, Clemente-Colón P (2004b) Ocean feature detection using microwave backscatter and sun glint observations. Int J Biodivers Oceanol Conserv 68(2):180–185 Gagliardini DA, Dogliotti AI, Karszenbaum H, Grings F (2004b) Comparison of bathymetric features detected by ERS2-SAR and LANDSAT-TM data over San Matias Gulf, Argentina. Int J Biodivers Oceanol Conserv 68(2):201–208 Garcia CAE, Garcia VMT (2008) Variability of chlorophyll-a from ocean color images in the La Plata continental shelf region. Cont Shelf Res. doi:10.1016/j.csr.2007.08.010 Garcia CAE, Garcia VMT, McClain CR (2005) Evaluation of SeaWiFS chlorophyll algorithms in the Southwestern Atlantic and Southern Oceans. Remote Sens Environ 95:125–137 Gordon A (1989) Brazil-Malvinas Confluence-1984. Deep-Sea Res 36:573–585 Hennings I, Matthews J, Metzner M (1994) Sun glitter radiance and radar cross-section modulations of the sea bed. J Geophys Res 99:16303–16326 Isla FI, Vilas FE, Bujalesky GG, Ferrero M, Gonzalez Bonorino G, Arche Miralles A (1991) Gravel drift and wind effects on the macrotidal San Sebastian Bay, Tierra del Fuego, Argentina. Mar Geol 97:211–224 Johannessen JA (2000) Coastal observing systems: the role of Synthetic Aperture Radar. Johns Hopkins APL Tech Dig 21:41–48 Johannessen JA, Digranes G, Espedal H, Johannessen OM, Samuel P, Vachon P (1994) SAR Ocean features catalogue. European Space Agency Special Pub. SP-1174 Kumar A (2004) Observations of intense sun glint in the daytime mid-infrared MODIS channelsfeasibility for sea surface temperature retrievals. Gayana 68(2):344–347 Largier JL (1993) Estuarine fronts: how important are they? Estuaries 16(1):1–11 Legeckis RV, Gordon AL (1982) Satellite observations of the Brazil and Falkland currents-1974 to 1976 and 1978. Deep-Sea Res 29:375–401 Leichter JJ, Wing SR, Miller SL, Denny MW (1996) Pulsed delivery of subthermocline water to Conch Reef (Florida Keys) by internal tidal bores. Limnol Oceanogr 41(7):1490–1501 López Gappa J, Sueiro MC (2006) The subtidal macrobenthic assemblages of Bahía San Sebastián (Tierra del Fuego, Argentina). Polar Biol. doi:10.1007/S00300-006-0225-3 Mahadevan A, Thomas LN, Tandon A (2008) Comment on “Eddy/Wind Interactions Stimulate Extraordinary Mid-Ocean Plankton Blooms”. Science 320(5875):448. doi:10.1126/science. 1152111 Malagnino E, Strelin JA, Gagliardini DA, Jacobo-Berlles J (1994) Analysis of geomorphic features of San Sebastian Bay, Tierra del Fuego and determination of the vulnerability to oil spill, using ERS-1 SAR data. Int Soc Photogramm Remote Sens Comm VII 7a:228–234 Mann KH, Lazier JRN (2006) Dynamic of marine ecosystems, biological-physical interactions in the ocean. Blackwell, Cambridge, MA McGillicuddy DJ Jr, Johnson RJ, Siegel DA, Michaels AF, Bates N, Knapp AH (1999) Mesoscale variability of ocean biogeochemistry in the Sargasso Sea. J Geophys Res 104:13381–13394 Olson D, Podestá GP, Evans RH, Brown O (1988) Temporal variation in the separation of Brazil and Malvinas Currents. Deep-Sea Res 35(12):1971–1990 Ottaviani M, Spurr RK, Stamnes L, Wei L, Wenying S, Wiscombe W (2008) Improving the description of sunglint for accurate prediction of remotely sensed radiances. J Quant Spectrosc Radiat Transf 109(14):2364–2375 Pierce JW, Sieguel FR, Urien CM (1969) Topografía submarina del Golfo San Matías. Proceedings of the VI Jornadas Geológicas Argentinas, Buenos Aires, Argentina, Actas III, pp 127–140 Piola AR, Rivas AL (1997) Corrientes en la Plataforma Continental, Publicaciones Especiales INIDEP No 1393 El Mar Argentino y sus recursos, Mar del Plata, Argentina, vol 1, pp 119–132 Piola AR, Romero SI (2004) Analysis of space-time variability of the Plata River plume. Gayana 68:482–486

13


277

Piola AR, Romero SI, Zajaczkowski U (2008) Space-time variability of the Plata plume inferred from ocean color. Cont Shelf Res 28:1556–1567 Pingree RD, Holligan PM, Mardell GT (1979) Phytoplankton growth and cyclonic eddies. Nature 278:245–247. doi:10.1038/278245a0 Rivas AL (2006) Quantitative estimation of the influence of surface thermal fronts over chlorophyll concentration at the Patagonian shelf. J Mar Syst 63:183–190 Rivas AL, Dogliotti AI, Gagliardini DA (2006) Satellite-measured surface chlorophyll variability in the Patagonian Shelf. Cont Shelf Res 26:703–720 Romero SI, Piola AR, Charo M, Garcia CAE (2006) Seasonal to interannual variability along the Patagonia Shelf and shelf-break based on satellite ocean color data. J Geophys Res. doi:10.1029/2005JC003244 Saraceno M, Provost C, Piola AR (2005) On the relationship of satellite retrieved surface temperature fronts and chlorophyll-a in the Western South Atlantic. J Geophys Res 110:C11016. doi:10.1029/2004JC002736 Saraceno M, Provost C, Piola AR, Bava J, Gagliardini DA (2004) Brazil Malvinas frontal system as seen from 9 years of advanced very high resolution radiometer data. J Geophys Res. doi:10.1029/2003JC002127 Valenzuela GR (1978) Theories for the interaction of electromagnetic and ocean waves-a review. Bound Layer Meteorol 13:61–85 Vesecky J, Stewart RH (1982) The observation of ocean phenomena using imagery from the SEASAT synthetic aperture radar: an assessment. J Geophys Res 87:3397–3430 Vilas F, Arche A, González Bonorino G, Isla F, Ferrero H (1986–1987) Sedimentación intermareal en la bahía de San Sebastián, Tierra de Fuego, Argentina. Acta Geol Hisp 22:253–260 Wang M, Bailey SW (2001) Correction of sun glint contamination on the SeaWiFS, Ocean and Atmosphere Products. Appl Opt 40(27):4790–4798 Yorio P (1998) Región 5. Zona costera patagónica. Los humedales de la Argentina, clasificación, situación actual, conservación y legislación, Wetland Internacional, 46:137–167

Part IV

Regional Observation

Chapter 14

Satellite Observation on the Exceptional Intrusion of Cold Water and Its Impact on Coastal Fisheries Around Peng-Hu Islands, Taiwan Strait Ming-An Lee, Yi Chang, Kuo-Wei Lan, Jui-Wen Chan, and Wei-Juan Hsieh

Abstract We used satellite-derived sea surface temperature (SST) data of the winters of 1996–2008 to examine the exceptional intrusion of China Coastal Current into the Taiwan Strait (TS). The long term observation reveals an exceptional cold water intrusion into the southern TS happened in February 2008. The warm Kuroshio Branch Current, which dominates the water around Chang-Yuen Ridge year round, was restricted to the southern Strait. Comparing the SST and wind speed during El Niño/La Niña events, we found that SST was warmer in the El Niño winters (1998, 2003, 2007) than in the La Niña winters (1996, 2000, 2008), and wind speed was more intensive in the La Niña winters than in the El Niño winters. It is suggested that in the winter of 2008, the strong and continuous northeasterly wind caused by La Niña event probably drove the cold Mainland China Costal Current more southward to penetrate into the southern TS north of the Chang-Yuen Ridge and a portion of this current intruded eastward south of the Peng-Hu Islands. The low SST event also significantly damaged marine life and cage aquaculture, causing the death of more than 73 m of resident and coral reef fishes; and at the same time brought increased abundance of migratory species. Keywords Sea surface temperature · China coastal current · El Niño · La Niña · Kuroshio current

1 Introduction Sea surface temperature (SST) in the Taiwan Strait (TS) is irregularly affected by extreme monsoons occurring during El Niño/La Niña events. This phenomenon is characterized by increased water temperature during El Niño, and decreased temperature during La Niña (Kuo and Ho, 2004; Chang et al., 2009). SST in the M.-A. Lee (B) Department of Environmental Biology and Fisheries Science, National Taiwan Ocean University, Pei-Ning Rd., Keelung 20224, Taiwan; Remote Sensing Laboratory, National Applied Research Laboratories, Taiwan Ocean Research Institute, Taipei, Taiwan e-mail: [email protected] D. Tang (ed.), Remote Sensing of the Changing Oceans, C Springer-Verlag Berlin Heidelberg 2011 DOI 10.1007/978-3-642-16541-2_14,

281

282

M.-A. Lee et al.

Fig. 14.1 (a–d) Extraordinary death of fishes due to the unusual intrusion of cold water into the waters around Peng-Hu Islands in February of 2008; (e) The EPA personnel removed dead fishes by truck. Pictures are reproduced from Hsieh et al. (2008) and the 2008 annual report of Fishery Agency, Taiwan (Anonymous, 2008)

southern TS is usually higher than 20◦ C in winter, but dropped to 13◦ C between late January and middle February in 2008. This extreme thermal anomaly has been proposed to have a great impact on marine ecology and fisheries. In January and February of 2008, a large number of resident fish was found dead on the beaches around Peng-Hu Islands (PHI), southern TS (Fig. 14.1), including more than 183 species in 58 families and a large amount of high-priced species, such as groupers, parrotfish, and wrasses (Hsieh et al., 2008). It has been suggested that this ecological disaster was caused by the exceptional intrusion of cold Mainland China Coastal Current under the influence of climate change (Hsieh et al., 2008; Chang et al., 2009). The negative effect of cold water events during a La Niña year on the composition and abundance of marine organisms was particularly significant for some reef fish because water temperature fell below their critical thermal minimum (16.3◦ C) in the eastern Pacific Ocean (Mora and Ospina, 2002). Although changes in water temperature during La Niña can cause more intense variation in hydrographic features than those during El Niño, the impact of La Niña winter on marine populations and fisheries in the TS are not well understood. The present study investigated the effect of the cold temperature event in La Niña winter on coastal fisheries by analyzing long-term SST, wind, and fishery data against the Ocean Niño Index (ONI, http://www.cpc.noaa.gov/products).

14

Satellite Observation on the Exceptional Intrusion of Cold Water

283

2 Data and Method Satellite-derived wintertime SST data measured by AVHRR (Advanced Very High Resolution Radiometer) sensors during the 13-year period from 1996 to 2008 were obtained from the regional AVHRR data library at the National Taiwan Ocean University (Lee et al., 2005a). SST images were produced by the MCSST algorithm (Multi Channel Sea Surface Temperature, McClain et al., 1985) with spatial resolution of 1.1 km. We also adopt an entropy-based edge detection method (Shimada et al., 2005) to detect SST fronts in the TS. To investigate long-term variation of SST related with climate factors, daily wind speed data measured by the gauge station on Peng-Hu Islands were used, and the Oceanic Niño Index (ONI) was also analyzed as the indicator of El Niño/La Niña events. Fishing data were collected from the Peng-Hu Fishery Association during the period of January–June, 1996–2008. There are four major fisheries in the water around Peng-Hu Islands (PHI in Fig. 14.2a), they are: pole & lines; gill net; long

Fig. 14.2 Monthly SST (a) and SST front gradient magnitude, black arrows indicated 14◦ C isotherm (b) maps of February in long-term average from 1996 to 2008, white arrows indicated the Peng-Chang front (Chang, 2009) on the same maps in 2008 (c and d). CYR is the Chang-Yuen Ridge and PHI is the Peng-Hu Islands

284

M.-A. Lee et al.

line; and set net. During the period of 2006–2008, all fish catch compositions of these four fisheries were recorded through questionnaire and then classified as nonmigratory and migratory species, for evaluating the influence of this exceptional cold water intrusion.

3 Wintertime SST Patterns Related with Climate Change The long-term monthly mean SST map (Fig. 14.2a) reveals that usually cold China Coastal Current (20◦ C) moved northward to the area around PHI in the eastern TS (Fig. 14.2a and c). The cold water was met by the warm water west of Chang-Yuen Ridge (CYR in Fig. 14.2b). A sharp frontal band (>0.3◦ C/km) was clearly formed along the Chinese coast and acted as a boundary between the cold China Coastal Current and warm Kuroshio/South China Sea Waters (Fig. 14.2b and d). The frontal pattern at the northern edge of CYR also indicated that the magnitude of the SST front gradient varied slightly (30◦ C by end May/early June. Over this region intense organised moist convection occurs heralding rapid northward advance of the summer monsoon (Gadgil et al., 1984; Joseph, 1990; Shenoi et al., 1999; Vinayachandran and Shetye, 1991; Rao and Sivakumar, 1999; Shenoi et al., 2005). The region off the southwest coast of India is one of the most biologically productive regions of the world oceans contributing to large volumes of fishery resources due to upwelling driven nutrient enrichment process during the SMS (Banse, 1987; Bauer et al., 1991; Madhupratap et al., 2001). Much of the upwelling in this region is driven by the coastal alongshore winds (Shetye et al., 1985, 1990; Gopalakrishna et al., 2008). Further, SEAS exhibits strong seasonal variability in the near-surface hydrography and circulation under the influence of seasonally reversing monsoons (Cutler and Swallow, 1984; Shetye et al., 1991; Shankar et al., 2002). Vinayachandran et al. (2007) have reviewed hydrography and circulation of the SEAS and the possible influence of the ocean on the onset of the monsoon. Using a high resolution numerical model Durand et al. (2007) have examined the processes that control the upper-ocean thermodynamics of the SEAS. Interannual variability of the upper-ocean heat budget of the north Indian Ocean is examined by de Boyer et al. (2006). Both hydrography and satellite altimeter measurements show a sea level high (low) in the sea surface topography in the SEAS during winter (summer monsoon). During winter (summer monsoon) the near-surface isothermal layer is also deeper (shallower) due to downwelling (upwelling) caused by the anticyclonic (cyclonic) eddy circulation popularly known as Lakshadweep High (low) seen in the satellite altimetry (Bruce et al., 1994, 1998; Shankar and Shetye, 1997). Formation of SST high in the SEAS has been related to the sea level high that forms due to the arrival of downwelling Kelvin waves from the BoB and the consequent westward propagating Rossby waves (Shenoi et al., 1999). The remote forcing is well known to play an important role in the dynamics of the SEAS through propagation of the coastal Kelvin waves that trigger westward propagating Rossby waves (McCreary et al., 1993; Shankar et al., 2002). During winter, both the East India Coastal Current (EICC) and Winter Monsoon Current (WMC) shown in Fig. 16.1a carry low salinity waters from the northern BoB into SEAS resulting in the drop in SSS and formation of strong salinity stratification in this region (Cutler and Swallow, 1984; Johannessen et al., 1987; Shetye et al., 1991, 1996; Rao and Sivakumar, 1999, 2003; Shenoi et al., 1999; Prasanna Kumar et al., 2004; Gopalakrishna et al., 2005). Further, intrusion of these low salinity waters results in the formation of a barrier layer (a layer embedded between top of the thermocline and bottom of the surface mixed layer) in the SEAS (Sprintall and Tomczak, 1992; Rao and Sivakumar, 1999; Durand et al., 2004; Masson et al., 2005). In addition, the SST in the southwestern BoB is cooler by ~2◦ C compared

16

Observed Interannual Variability of the Thermohaline Structure

307

Fig. 16.1 Monthly mean climatology of (a) sea surface salinity (Boyer et al., 2006) over laid with mean surace currents (Ekman and geostrophic) derived from climatology of QuikSCAT and AVISO T/P merged sea-level anomaly data for November to February. Schematic of the East India Coastal Current and Winter Monsoon current are shown by white arrows. (b) Monthly mean climatology of TMI SST for November to February

to the SST of the SEAS region (Fig. 16.1b) and accordingly these currents transport relatively cooler surface waters into this region. During their passage, the intruded low salinity waters encounter intense surface cooling south of the Indian tip due to strong winds that blow through the orographic gap between Indian tip and Sri Lanka, enhancing the turbulent heat losses resulting in lowering of SST by about 1◦ C in the region south of Gulf of Mannar (Luis and Kawamura, 2000; Rao et al., 2008). Occurrence of near-surface thermal inversions in the SEAS during winter (November–February) is a well documented phenomenon in the literature (Thadathil and Ghosh, 1992; Shankar et al., 2004; Gopalakrishna et al., 2005; Thompson et al., 2006; Nisha et al., 2008). Recently Kurian and Vinayachandran (2006) have examined the possible mechanisms of thermal inversions in a numerical model simulation in the SEAS and concluded that the haline stratification is an important prerequisite for the formation of thermal inversions. Using an ocean general circulation model, Durand et al. (2004) have shown that the heat trapped within these thermal inversions makes a significant contribution in increasing the

308

N. Kurian et al.

SST at least by 1.1◦ C during November–March contributing to the seasonal buildup of the warm pool in the SEAS. Masson et al. (2005), using a coupled general circulation model, demonstrated that the lack of heating associated with the barrier layer in SEAS results in late onset of summer monsoon. Another study by Masson et al. (2002) has shown that barrier layer enhances the spring SST warming and leads to a statistically significant increase of precipitation in May linked to an early monsoon onset. Gopalakrishna et al. (2005) have reported large differences in the life cycle and the depth of occurrence of these thermal inversions between the winters of 2002–2003 and of 2003–2004. In view of the large dynamic variability of this region, under the aegis of Indian Climate Research Program a field experiment called the Arabian Sea Monsoon Experiment was thus conceived, planned and executed to understand the coupling between the summer monsoon and the SEAS. Under this program, repeat XBT transects and sampling of SSS was carried out in the SEAS in a systematic manner for the first time during 2002–2008. In this paper, we have reviewed the work carried out using the above data sets and further an attempt is also made to describe the observed seasonal and interannual variability of SSS in the SEAS.

2 Data and Processing Repeat XBT measurement program in the SEAS is a major long-term ongoing observational initiative supported by the Ministry of Earth Sciences in India. Under this initiative near-fortnightly XBT surveys are being systematically organized for the first time since May 2002 using passenger ships which ply regularly between Kochi and Lakshadweep Island Chain. This systematically collected XBT data set is unique in several respects to examine the observed seasonal cycle and its interannual variability of near-surface thermal structure and its most important embedded feature-upwelling. During each XBT survey 10–13 vertical temperature profiles (T7 Sippican XBT probes and MK21 data acquisition system) and 20–25 sea surface water samples (bucket samples) are being collected (black dots in Fig. 16.2 depicts the XBT and SSS stations). The XBT data are processed and quality controlled following using standard techniques. These water samples are analyzed for SSS using Guild Line 8400 Autosal. The Kochi – Kavaratti is the most densely covered XBT transect (shaded strip in Fig. 16.2) utilized in this study to (a) construct snap-shot vertical thermal sections in the upper 200 m water column to characterize the nature of the observed upwelling and (b) to examine the seasonal cycle and interannual variability of near-surface thermal structure and SSS fields. The period during November 2002–February 2003 is considered to represent the winter season 2002–2003 (W23). The winter seasons of 2003–2004, 2004–2005, 2005– 2006 and 2006–2007 are referred as W34, W45, W56 and W67 respectively. The AVISO girded SSH anomaly product was utilized to characterize the nature of

16


309

Fig. 16.2 Location map showing DBT and sea surface salinity stations (dots) collected during May 2002–August 2008. Densely convered Kochi–Kavaratti transect is shown by the shaded strip.

propagating upwelling Kelvin and Rossby waves. The QuikSCAT wind data is utilized to characterize both the alongshore wind stress and WSC. The TMI SST data for the KK XBT transect are used to examine the observed cooling caused by the upwelling.

3 Analysis The Individual snap-shot thermal sections along KK XBT transect for the years 2002–2006 are shown in Fig. 16.3. The annual cycle is typically characterized by deep near-surface isothermal layer during winter caused by downwelling (Shenoi et al., 2005). Thermal inversions in the near-surface layer are also seen during November–February of these years with some differences (Thadathil and Gosh, 1992; Gopalakrishna et al., 2005). All these sections show the occurrence of a warm pool (Rao and Sivakumar, 1999; Shenoi et al., 1999) during March–May in all the years with minor differences in the intensity. With the onset and progress of the summer monsoon the warm pool collapses with some differences among the years of study. A mild secondary warming is noticed again in the near-surface layers during October–November after the withdrawal of the summer monsoon. Below the surface layer, the thermocline also shows a pronounced annual cycle. The deep thermocline seen during winter (December–February) begins to shoal from February/March reaching its shallowest depth by September due to upwelling. During the upwelling season, the near-surface isothermal layer progressively shoals with increasing magnitude towards the coast (Sharma, 1968; Shetye et al., 1990; Shankar et al., 2005; Shenoi et al., 2005). The topography of depth of 25◦ C isotherm (D25) representing the core of the thermocline is extracted from the near-fortnightly snap shot thermal sections for all the seven years. The annual cycle of D25 is presented in Fig. 16.4 for a coastal box (shown in Fig. 16.2) as upwelling is more pronounced towards the coast

310

N. Kurian et al.

Fig. 16.3 Snapshot of vertical thermal sections for the upper 150 m along Kochi-Kavaratti XBT transect from to January to December during 2002–2006. J2-D2, J3-D3, J4-D4, J5-D5, J6-D6 represents January–December individual snapshot vetical thermal sections for the years

16


311

Fig. 16.4 Annual march of D25 for the coastal box (shown in Fig. 1) for the years 2002–2008

(Shetye et al., 1990) to highlight the observed inter-annual variability in upwelling. Although the uplift of D25 has started as early as February/March, the upwelling is weaker (stronger) during the SMS of 2005 (2002) compared to all the other years. During all the other years, the D25 reached an average depth of about 37 m during September. During the SMS 2002, the D25 reached the shallowest depth of about 15 m, the lowest value recorded in the entire data set. However, during the SMS 2005, the D25 reached the shallowest depth of only 50 m during November–that is about two months latter than the normal indicating the weaker and prolonged upwelling. The observed annual cycle of SSH anomaly along the KK XBT transect for five years (Fig. 16.5) clearly shows distinct differences in the amplitude and the temporal extent of the signature caused by the westward propagating Rossby waves triggered by the northward propagating coastal Kelvin waves. During the upwelling season, the SSH anomaly is negative due to upwelling Kelvin wave. The change of sign of SSH anomaly occurs during May as seen in all the years. Interestingly, this change of sign has occurred a little later and the negative values lasted longer during 2005. In addition, the magnitudes of these negative values are also relatively weaker during 2005 compared to any other year. This implies that the amplitude of the propagating waves and upwelling is relatively weaker during the SMS of 2005 resulting in relatively weaker uplift of D25. The SSH anomaly during SMS of 2002 is relatively stronger suggesting stronger upwelling as seen in the XBT measurements. The cooling of the sea surface is usually associated with upwelling [Shetye et al., 1990]. Accordingly the annual cycle of SST along the KK XBT transect derived from TMI is examined for 5 years (Fig. 16.6). The annual cycle of the SST is typically characterized by a primary (secondary) heating maxima during pre-monsoon (post-monsoon) season. The primary (secondary) cooling maxima occur during summer (winter) monsoon season. However, there are perceptible differences in the distribution of heating and cooling cycles among these 5 years. The most noteworthy feature is the absence of the secondary warming during October–November 2005 unlike any other year. This anomalous feature is also noticed in the near-surface

312

N. Kurian et al.

Fig. 16.5 Evolution of SSH anomaly along the KK XBT transect for the years 2002–2007

thermal structure along KK transect during October–November 2005 (Fig. 16.3). During 2005 the cooling episodes continued beyond the SMS and persisted till the end of the year. This is in excellent agreement with the prolonged upwelling as inferred from both the vertical thermal sections and the SSH anomaly fields.

16


313

Fig. 16.6 Evolution of TMI mean monthly SST along the KK XBT transect for the years 2002–2006

4 Governing Mechanisms 4.1 Local Forcing It is important to understand the possible mechanisms that produce the observed anomalous nature of upwelling during the SMS of 2002 and 2005. It is well known that the local winds play an important role in driving the offshore Ekman transport and the associated divergence in the near-surface layers leading to the uplift of isotherms in the thermocline (Shetye et al., 1985). Accordingly the annual cycle of both the observed alongshore wind stress and WSC are examined for the KK XBT transect for five years. The annual cycle of the alongshore wind stress for the KK XBT transect is presented in Fig. 16.7. The alongshore wind stress is equatorward during the SMS of all the years. In addition, unlike other years the equatorward wind stress has also persisted till the end of December in a transient manner only during 2005. The present analysis has clearly revealed that the local wind forcing is distinctly different during the SMSs of 2002 and 2005 resulting in differences in the observed upwelling.

4.2 Remote Forcing The recent modeling studies have clearly shown that the winds over the equatorial Indian Ocean play an important role in modulating the circulation features of the

314

N. Kurian et al.

Fig. 16.7 Evolution of QuikSCAT along-shore wind stress component along the KK XBT transect for the years 2002–2006

north Indian Ocean (Potemra et al., 1991; Yu et al., 1991; McCreary et al., 1993, 1996; Shankar et al., 2002). The energy imparted by the surface winds propagate along the equatorial wave-guide (Sengupta et al., 2007) as upwelling/downwelling Kelvin waves and traverse around the rim of BoB and enter the SEAS. In addition, the alongshore winds in the coastal BoB also trigger and modulate the propagating Kelvin waves. The signature of this wave propagation is examined utilizing the satellite surface wind and satellite altimetry measurements along the equator. The observed zonal wind stress climatology (Figure not shown) is relatively stronger over the east central equator and shows strong intraseasonal variability with pronounced peaks during the monsoon transitions (resulting in Spring and Fall Wyrtki Jets).

5 Observed Inter-Annual Variations in the Thermal Inversions in the SEAS During 2002–2008 The XBT data are examined to describe the evolution of the near-surface thermal inversions. The spatial distribution of all the XBT stations with (red dots) and without (blue dots) thermal inversions respectively for each month of the individual winter season are shown in Fig. 16.8. The near-surface thermal inversions are only considered when their amplitude exceeds the SST value by at least 0.25◦ C and the inversion layer thickness exceeds 5 m. The distribution of these thermal inversions shows a distinct life cycle during each winter. They first appear few in

16


315

Fig. 16.8 XBT station locations for W23, W34, W45, W56, W67 and W78 in the SEAS. Red and Blue dots represent XBT stations with and without temperature inversions respectively

number during November and their population increases with the progress of the season. They peak during January and disappear by March when the SST begins to increase. They also show large year-to-year variability in their characteristics and population density (Gopalakrishna et al., 2005). Interestingly relatively a reduced number of thermal inversions have occurred throughout W56. During the winters of 2002–2007 the percentage occurrence of thermal inversions observed from the XBT data have shown a large spread from a minimum of 16% in W56 to a maximum of 49% in W23. The W56 can be cited as the winter season with the least occurrence of thermal inversions among winters 2002 through 2005. In addition, it is also interesting to note that these inversions have occurred at shallower depths (~10 m) and occupied thicker water column (~35 m) in W56 compared to the other winters in the study. The plausible causative mechanisms for this unusual occurrence of reduced number of thermal inversions are examined in the following sections. In the following sections the possible mechanisms are examined to identify the actual mechanism responsible for the observed reduced number of inversions in the SEAS during W56.

6 Anomalous Background State of the Lakshadweep Sea In order to understand the role of the background state of the SEAS, several parameters are examined. Usually a secondary warming occurs in the SEAS after the summer monsoon cooling (Colborn, 1980). However, perceptible differences in the

316

N. Kurian et al.

secondary warming are seen among the years in the observed evolution of TMI SST (Fig. 16.6) along the Kochi – Kavaratti XBT transect (shaded strip in Fig. 16.2). The most noteworthy feature is the occurrence of unusually weaker secondary warming during October–November 2005 unlike any other year. This anomalous feature is also noticed in the near-surface thermal structure along Kochi–Kavaratti transect during October–November 2005 (Fig. 16.3). The cooling regime continued beyond the summer monsoon season of 2005 and persisted almost till the end of the year. The governing mechanisms responsible for this observed anomalous weaker secondary warming in 2005 are examined. The surface winds and the surface turbulent heat losses play an important role in the cooling process of the surface mixed layer of the ocean through vertical mixing. The observed mean monthly surface wind speed anomalies derived from QuikSCAT for W56 showed stronger positive anomalies during September 2005–January 2006 compared to the corresponding climatological wind filed particularly in the SEAS region. The net surface heat flux anomalies during September–December 2005 are also distinctly more negative (less heat gain by the ocean) in the SEAS compared to any other year considered in this study. Thus both the observed stronger winds and decreased heat gain by the ocean during the post-monsoon season might have also contributed to the weaker secondary warming observed during October–November, 2005 in association with the prolonged upwelling till November 2005 noticed in the snap-shot vertical thermal sections along Kochi-Kavaratti XBT transect (Fig. 16.3). This has resulted in a weaker horizontal SST gradient between the SEAS and the intruding low salinity waters from the BoB. Such a situation potentially contributes to the formation of reduced number of thermal inversions.

7 Intrusion of Low Saline Waters from the Bay of Bengal The low salinities observed in the SEAS during winter are associated with both the intrusion of low salinity waters from the BoB and unusual high precipitation. The observed variability of SSS along Kochi–Kavaratti and Kochi–Minicoy XBT transects (Fig. 16.9) is examined to understand its evolution during May 2002–May 2008. The Hovmoller field of sea surface salinity along these transects clearly shows the evolution of seasonal cycle in conformity with the earlier published climatologies. The seasonal cycle is characterized by the appearance of low (high) salinity waters during winter extending into pre-monsoon (summer monsoon and extending into post-monsoon) with decreasing (increasing) values towards southwest coast of India (Kavaratti/Amini Islands). The observed drop from November to February in the Conkright et al. (2002) sea surface salinity climatology along the Kochi– Kavaratti transect is 1.6 PSU. Interestingly the present observations show a much greater drop of 4.03 PSU during W56. As hypothesized earlier by Gopalakrishna et al. (2005), this large freshening is primarily attributed to the heavy fresh water input into the BoB and SEAS through river discharges and rainfall during the

16


317

Fig. 16.9 Hovmaoller fields showing the observed interannual variability of SSS along (a) Kochi– Kavaratti and (b) Kochi-Minicoy XBT transects

preceding summer monsoon and winter monsoon seasons. On the inter-annual time scale, the most dominant signals noticed are the occurrence of low salinity waters during W56 and W34 among which the freshening is greater during W56. The low salinity waters that intrude from the BoB through EICC and WMC determine the near-surface freshening in the SEAS region. In order to examine the year-to-year variability of river discharges in to the BoB, we have used the data on monthly total river discharges for two major rivers Mahanadi and Godavary which are situated along the east coast of India. Among the years, the total river discharge during May to November is greater during 2003 and 2005 (Fig. 16.10) lending support to the observed excessive freshening in the SEAS during W34 and W56. Further, the available GPCP satellite precipitation estimates over the southwestern BoB and the SEAS (for two boxes shown in Fig. 16.11) also clearly show occurrence of more rainfall during June 2005–February 2006 compared to the same period during the other years (Fig. 16.12). The larger rainfall signal noticed during 2005–2006 and 2003–2004 coincided with greater surface freshening during the following winter. Thus it is clearly seen that in addition to the influence of the river discharges, the cumulative rainfall and strong local surface currents have also contributed to the observed intense freshening noticed during W56 and W34.

318 Fig. 16.10 Comparison of river discharges totals (may to November period) for the rivers Mahanadi and Godavary

Fig. 16.11 Boxes considered for comparison of total rainfall (mm) during June to February

Fig. 16.12 Histrograms showing the total rainfall (June to February) over (i) box A and (ii) box B for the years 2002–2006

N. Kurian et al.

16


319

8 Intrusion of Cooler Waters (SST Gradient Between SEAS and the Intruding BoB Waters) It is well known that the low salinity waters that intrude from the BoB into the SEAS are cooler than the local ambient waters (Luis and Kawamura, 2002). The horizontal SST differences between these two regions determine the amplitude of the observed thermal inversions. These thermal inversions can only occur when the BoB surface waters are both low saline and cooler than the near-surface ambient waters in the SEAS. To test this hypothesis the difference between the SSTs for two selected boxes representing the SEAS and the intruding BoB waters shown in Fig. 16.13a is examined for all the six winters. The SST of box B is by and large lower than that of box A (positive gradients) during all the five winters thus clearly indicating that the intruding waters from the bay that enter the SEAS are cooler than the local

Fig. 16.13 Time series of SST differences (Bottom panel) between the boxes B and A (Top panel) during the winter seasons of 2002–2007

320

N. Kurian et al.

ambient waters (Fig.16.13b). However, during W56 the difference was very weak (sometimes with negative gradients) during December 2005–January/February 2006 suggesting that the SST of the intruding waters from the bay is of comparable magnitude (or warmer) of the SST of box A. This weak horizontal gradient in SST has resulted in the occurrence of reduced number of thermal inversions during January–February 2006 as shown in Fig. 16.8.

9 Summary The near-fortnightly repeat XBT measurements made in SEAS during May 2002– August 2008 have provided a unique and first of its kind time series data set to examine the evolution of near-surface thermohaline structure and to describe the observed inter-annual variability. In the summer monsoon season of 2005 (2002), the upwelling is relatively weaker (stronger) than that of any other year. The upwelling (uplift of D25) during 2005 has persisted longer by about two months when compared to the other years. The observed surface westerly winds along the equator during winter 2004–2005 are short lived and relatively weaker resulting in the occurrence of weaker positive SSH anomaly. This has triggered a weaker upwelling Kelvin wave that has propagated into the SEAS by February–March 2005. The present study provides the first of its kind observational evidence in the SEAS to support the importance of equatorial wind forcing as shown in the modelling studies. In spite of relatively stronger haline stratification observed in the SEAS region, a reduced number of thermal inversions have occurred in W56. Utilizing both in-situ and satellite measurements several processes that are responsible for the occurrence of reduced number of thermal inversions (10%) at shallower depths (~10 m) during W56 in the study region are examined. This study highlights the importance of the secondary surface warming in SEAS region for the formation of thermal inversions. It also illustrates that even the large amount of low salinity from the BoB into SEAS, may not be sufficient condition for the formation of thermal inversions. The advection of low salinity waters from the BoB may play an important role in the formation of thermal inversions only if the advected waters are relatively cooler than the ambient waters in the SEAS. Our study, based on observations suggests that the background state in the SEAS prior to the arrival of low salinity waters is an important condition for the formation of thermal inversions. Acknowledgments We thank the Lakshadweep Administrator for permitting XBT measurements onboard their passenger ships. This work was supported by the Ministry of Earth Sciences through INCOIS, Hyderabad. K. Nisha acknowledges the financial support from the Council of Scientific and Industrial Research, India. Mr. Shamkant Akerkar prepared the figures for publication. This is NIO contribution 4890.

References Anderson SP, Weller RA, Lukas RB (1996) Surface buoyancy forcing and the mixed layer of the western Pacific warmpool: observations and one- dimensional model results. J Climate 9:3056–3085

16


321

Banse K (1987) Seasonality of phytoplankton chlorophyll in the central and northern Arabian Sea. Deep Sea Res 34:713–723 Bauer S, Hitchcock GL, Olson DB (1991) Influence of monsoonally-forced Ekman dynamics upon surface layer depth and plankton biomass distribution in the Arabian Sea. Deep Sea Res 38:531–553 Boyer TP, Antonov JI, Garcia HE, Johnson DR, Locarnini RA, Mishonov AV, Pitcher MT, Baranova OK, Smolyar IV (2006) World Ocean Database 2005. In Levitus S (Ed), NOAA Atlas NESDIS 60, U.S. Government Printing Office, Washington, D.C., 190 pp, DVDs Bruce JG, Johnson DR, Kindle JC (1994) Evidence for eddy formation in the eastern Arabian Sea during the north east monsoon. J Geophys Res 99:7651–7664 Bruce JG, Kindle JC, Kantha LH, Kerling JL, Bailey JF (1998) Recent observations and modeling in the Arabian Sea Laccadive High region. J Geophys Res 103:7593–7600 de Boyer Montégut C, Vialard J, Shenoi SSC, Shankar D, Durand F, Ethé C, Madec G (2007) Simulated seasonal and interannual variability of mixed layer heat budget in the northern Indian Ocean. J Climate 20:3249–3268 Clarke AJ (1983) The reflection of equatorial waves from oceanic boundaries. J Phys Oceanogr 13:1193–1207 Colborn JG (1980) Thermal structure of the Indian Ocean. IIOE Monographs No.2. East West Center Press, Honolulu Conkright ME, Locarnini RA, Garcia HE, O’Brien T, Boyer TP, Stephens C, Antonov J (2002) World Ocean Atlas 2001, Objective Analyses, Data Statistics and Figures, CD-ROM Documentation. National Oceanographic Data Center, Silver Spring, MD Cutler A, Swallow J (1984) Surface currents of the Indian Ocean (to 25◦ S, 100◦ E) (compiled from archived historical data). Rep. 187, Meteorological Office, U.K. Institute of Oceanographic Sciences, Bracknell, 42 pp. Available from U.K. Meteorological Office, Bracknell, Berkshire RG12 252, United Kingdom Durand F, Shankar D, de Boyer Montegut C, Shenoi SSC, Blanke B, Madec G (2007) Modeling the barrier-layer formation in the southeastern Arabian Sea. J Climate 20:2109–2120 Durand F, Shetye SR, Vialard J, Shankar D, Shenoi SSC, Ethe C, Madec G (2004) Impact of temperature inversions on SST evolution in the South-Eastern Arabian Sea during the presummer monsoon season. Geophys Res Lett 31:L01305 Gadgil S, Joseph PV, Joshi NV (1984) Ocean-atmosphere coupling over monsoon regions. Nature 312:141–143 Gopalakrishna VV, Johnson Z, Salgaonkar G, Nisha K, Rajan CK, Rao RR (2005) Observed variability of sea surface salinity and thermal inversions in the Lakshadweep Sea during contrast monsoons. Geophys Res Lett 32:L18605 Gopalakrishna VV, Rao RR, Nisha K, Girishkumar MS, Pankajakshan T, Ravichandran M, Johnson Z, Girish K, Aneeshkumar N, Srinath M, Rajesh S, Rajan CK (2008) Observed anomalous upwelling in the Lakshadweep Sea during the summer monsoon season of 2005. J Geophys Res 113:C05001. doi:10.1029/2007JC004240 Han W, Webster PJ (2001) Forcing mechanisms of sea level interannual variability in the Bay of Bengal. J Phys Oceanogr 32:216–239 Johannessen OM, Subbaraju G, Blindheim J (1987) Seasonal variations of the oceanographic conditions off the southwest coast of India during 1971–1975. Fisk Dir Skr Hav Unders 18: 247–261 Joseph PV (1990) Warm pool over the Indian Ocean and monsoon onset. Trop Ocean-Atmos News Lett Winter:1–5 Kurian J, Vinayachandran PN (2006) Formation mechanisms of temperature inversions in the southeastern Arabian Sea. Geophys Res Lett 33:L17611 Luis AJ, Kawamura H (2000) Winter wind forcing and sea surface cooling near the south India tip observed using NSCAT and AVHRR. Remote Sens Environ 73:55–64 Luis AJ, Kawamura H (2002) A case study of sea surface temperature – cooling dynamics near the Indian tip during May 1997. J Geophys Res 107:1–11 Madhupratap M, Nair KNV, Gopalakrishna TC, Haridas P, Nair KKC, Venugopal P, Gauns M (2001) Arabian Sea oceanography and fisheries of the west coast of India. Curr Sci 81:355–361

322

N. Kurian et al.

Masson S, Delecluse P, Boulanger JP, Menkes C (2002) A model study of the seasonal variability and formation mechanisms of the barrier layer in the eastern equatorial Indian Ocean. J Geophys Res 107(C12):8017 Masson S, Luo JJ, Madec G, Vialard J, Durand F, Gualdi S, Guilyardi E, Behra S, Delecluse P, Navarra A, Yamaghata T (2005) Impact of barrier layer on winter-spring variability of the southeastern Arabian Sea. Geophys Res Lett 32:L07703 McCreary JP, Kundu PK, Molinari RL (1993) A numerical investigation of the dynamics, thermodynamics and mixed layer processes in the Indian Ocean. Prog Oceanogr 31: 181–244 McCreary JP, Han W, Shankar D, Shetye SR (1996) Dynamics of the East India Coastal Current, 2. Numerical solutions. J Geophys Res 101:13993–14010 Nisha K, Suryachandra Rao A, Gopalakrishna VV, Rao RR, Girishkumar MS, Pankajakshan T, Ravichandran M, Rajesh S, Girish K, Anuradha M, Gavaskar SSM, Suneel V, Krishna SM (2008) Reduced near-surface thermal inversions in 2005–2006 in the Southeastern Arabian Sea (Lakshadweep Sea). J Phys. Oceanogr 39:1184–1199 Potemra JT, Luther ME, O’Brien JJ (1991) The seasonal circulation of the upper ocean in the Bay of Bengal. J Gephys Res 96:12667–12683 Prasanna Kumar S, Jayu N, Ajoy K, Shaji C, Anand P, Sabu P, Rijomon G, Josia J, Jayaraj KA, Radhika A, K.K.C Nair (2004) Intrusion of the Bay of Bengal water into the Arabian Sea during winter monsoon and associated chemical and biological response. Geophys Res Lett 31:L15304 Rao RR, Girish Kumar MS, Ravichandran M, Gopalakrishna VV, Thadathil P (2008) A cold pool south of Indo-Sri Lanka channel and its intrusion into the southeastern Arabian Sea during winter. Deep Sea Res I: Oceanogr Res Pap 55(8):1009–1020 Rao RR, Sivakumar R (1999) On the possible mechanisms of the evolution of a mini-warm pool during the pre-summer monsoon season and the onset vortex in the southeastern Arabian Sea. Quart J Roy Meteor Soc 125:787–809 Rao RR, Sivakumar R (2003) Seasonal variability of sea surface salinity and salt budget of the mixed layer of the north Indian Ocean. J Geophys Res 108(C1):3009 Sengupta D, Senan R, Goswami BN, Vialard J (2007) Intraseasonal variability of equatorial Indian Ocean zonal currents. J Climate 20:3036–3055 Shankar D, Gopalakrishna VV, Shenoi SSC, Durand F, Shetye SR, Rajan CK, Johnson Z, Araligidad N, Michael GS (2004) Observational evidence for westward propagation of temperature inversions in the southeastern Arabian Sea. Geophys Res Lett 31:L01305 Shankar D, Shenoi SSC, Nayak RK, Vinayachandran PN, Nampoothiri G, Almeida AM, Michael GS, Rameshkumar MR, Sundar D, Sreejith OP (2005) Hydrography of the eastern Arabian Sea during summer monsoon 2002. J Earth Syst Sci 114:459–474 Shankar D, Shetye SR (1997) On the dynamics of the Lakshdweep high and low in the southeastern Arabian Sea. J Geophys Res 102:12551–12562 Shankar D, Vinayachandran PN, Unnikrishnan AS (2002) The monsoon currents in the north Indian Ocean. Prog Oceanogr 52:63–120 Sharma GS (1968) Seasonal variation of some hydrographic properties of the shelf waters off the west coast of India. Bull Nat Inst Sci India 38:263–276 Shenoi SSC, Shankar D, Michael GS, Kurian J, Rameshkumar MR, Almeida AM, Unnikrishnan AS, Fernandes W, Barreto N, Gnanaseelan C, Mathew R, Praju KV, Mahale V (2005) Hydrography and water masses in the southeastern Arabian Sea during March–June 2003. J Earth Syst Sci 114:475–491 Shenoi SSC, Shankar D, Shetye SR (1999) On the sea surface temperature high in the Lakshadweep Sea before the onset of the south-west monsoon. J Geophys Res 104:15703– 15712 Shetye SR, Gouveia AD, Shankar D, Shenoi SSC, Vinayachandran PN, Sundar D, Michael GS, Namboodiri G (1996) Hydrography and circulation in the western Bay of Bengal during northeast monsoon. J Geophys Res 101:14011–14025

16


323

Shetye SR, Gouveia AD, Shenoi SSC, Sundar D, Michael GS, Almeida AM, Santanam K (1990) Hydrography and circulation off the west coast of India during south west monsoon of 1987. J Mar RKes 48:359–378 Shetye SR, Shenoi SSC, Antony MK, Krishna Kumar V (1985) Monthly mean wind stress distribution along the coast of the North Indian Ocean. Proc Ind Acad Sci (EPS) 94:129–137 Sheyte SR, Gouveia AD, Shenoi SSC, Michael GS, Sunder D, Almeida AA, Sandanam K (1991) The coastal currents off western India during northeast monsoon. Deep-Sea Res 38A: 1517–1529 Sprintall J, Tomczak M (1992) Evidence of the barrier layer in the surface layer of the tropics. J Geophys Res 97:7305–7316 Thadathil P, Gosh A (1992) Surface layer temperature inversion in the Arabian Sea during winter. J Oceanogr 48:293–304 Thompson B, Gnanaseelan C, Salvekar PS (2006) Seasonal evolution of temperature inversions in the north Indian Ocean. Curr Sci 90:697–704 Vinayachandran PN, Shetye SR (1991) The warm pool in the Indian Ocean. Proc Indian Acad Sci (Earth Planet Sci) 100:165–175 Vinayachandran PN, Shankar D, Kuria J, Durand F, Shenoi SSC (2007) Arabian Sea mini warm pool and the monsoon onset vortex. Curr Sci 93:203–214 Yu L, O’Brien JJ, Yang J (1991) On the remote forcing of the circulation in the Bay of Bengal. J Gephys Res 96:20449–20454.

Part V

Natural Hazards

Chapter 17

Satellite Observations Defying the Long-Held Tsunami Genesis Theory Y. Tony Song and Shin-Chan Han

Abstract Using seismographs and GPS displacement measurements, we have estimated the seafloor deformation history of the December 2004 Sumatra-Andaman earthquake and the March 2005 Nias earthquake by separating their deformation period into intervals of 800-s, 1-h, and 6-months. We have then calculated their corresponding gravity changes (induced by the seafloor deformation), which are 11.3, 12.5, and 14.9 microgalileo, respectively. We show that the seismographs and GPSderived values are consistent with the known postseismic to coseismic moment ratio of 30% and the Gravity Recovery and Climate Experiment (GRACE) satellites measurements of 15 microgalileo for the same period of 6 months. However, the vertical component of the accumulated seafloor deformation during the tsunami formation period (~30 min) could only generate a potential energy of 1.2 × 1015 Joules and account for only one third of the actual tsunami height. The evidence is overwhelmingly contrary to the long-held theory that the vertical deformation of seafloor is the primary source of tsunamis. Furthermore, we have carefully examined the pioneering wave-maker experiment that initially conceived the ubiquitous tsunami genesis theory. Surprisingly, we found that the experimental ratio of the horizontal slip distance to the water depth – the non-dimensional parameter that allows comparing the experiment with reality on an apple-to-apple basis – was 200 times of realistic earthquake parameters. The experimental conclusion is problematic in conceiving the tsunami theory. By including the horizontal momentum energy transferred by the faulting continental slope in a three-dimensional tsunami model, we have re-examined the December 2004 tsunami using both seismographs and GPS measurements. Our results show that the new theory is more consistent with altimetry and tide data than the conventional theory of using the vertical force alone, suggesting that the tsunami formation mechanism is not as simple as previously thought. Keywords Tsunami genesis theory · GRACE · Vertical uplift · Horizontal displacement · Seismograph · GPS data Y.T. Song (B) Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109, USA e-mail: [email protected] D. Tang (ed.), Remote Sensing of the Changing Oceans, C Springer-Verlag Berlin Heidelberg 2011 DOI 10.1007/978-3-642-16541-2_17,

327

328

Y.T. Song and S.-C. Han

1 Introduction The tsunami life cycle can be divided into three stages: formation of initial waves, wave propagation in deep oceans, and run-ups to shallow seas and beaches. This study focuses on the first stage of tsunami excitation because it is poorly studied. In contrast, the latter two stages have been understood relatively well. For example, the 2004 Indian Ocean tsunami waves have been clearly observed over the deep ocean by several satellite altimeters (Song et al., 2005). Their run-ups to shallow seas have also been recorded by many coastal tide-gauges (Merrifield et al., 2005; Choi et al., 2006). Even using two-dimensional shallow-water-equations models, tsunami propagation and run-ups can be modeled to replicate satellite measurements and tide records fairly well (e.g., Titov et al., 2005; Hirata et al., 2006; Grilli et al., 2007). The challenging question is whether the initial sea-surface perturbation, used in the tsunami models as the only initial condition, is really the vertical displacement of the seafloor. In the other words, the conventional tsunami theory used in the model has not been substantiated by data and observations. In fact, the tsunami formation mechanism is poorly understood and its theory is still in debate. The conventional theory is that the vertical displacement of the seafloor caused by undersea earthquakes is the major force of tsunamis (Tuck and Hwang, 1972; Abe, 1973; Iwasaki, 1982; Tanioka and Satake, 1996). The initial sea-surface condition in the tsunami propagation model is often assumed to be identical to the vertical displacement of the seafloor, which can be derived by the simple formulation of Abe (1973) or more sophisticated subfault model of Okada (1985). However, it is difficult to verify the modeled vertical displacement quantitatively by data. Immediately after the December 2004 Indian Ocean tsunami, the Royal Navy survey vessel HMS Scott conducted bathymetric mapping of the Sumatra subduction zone. Surprisingly, only a small area of seafloor uplift was found, leading many to wonder how the enormous amount of tsunami energy was transformed from the earthquake (Moran et al., 2005). Because a full survey of the fault area before and after the initial earthquakes is impracticable, there is great difficulty in determining the vertical values of the seafloor deformation, therefore, the conventional theory remains unverified. Previously, we had used seismic inversion (Song et al., 2005) and GPS displacement data (Song, 2007) separately to estimate the source of the 2004 Indian Ocean tsunami and the March 2005 Nias tsunami, but we were unable to verify those seismically and GPS solutions due to lack of additional data. Since then the seismic solution has been improved significantly. GPS data have also been modified with field measurements (Chlieh et al., 2007). More recently, the space-based gravity measurements from the Gravity Recovery and Climate Experiment (GRACE) satellites have also been demonstrated capable of estimating the seafloor deformation of the 2004 Sumatra-Andaman earthquake (Han et al., 2006; Chen et al., 2007). Here we have combined these three kinds of independent datasets to gain insight into the 2004 Indian Ocean tsunami genesis. This chapter is organized as follows. Section 2 presents the seismic data and GPS displacement measurements, as well as inversion methodologies for estimating the

17

Satellite Observations Defying the Long-Held Tsunami Genesis Theory

329

earthquake source. In Sect. 3, we will verify the earthquake source by GRACE data. Section 4 is used to test the conventional tsunami theory by the earthquake source and altimetry measurements. In Sect. 5, we will examine the wave-maker experiment that initially conceived the conventional tsunami theory. Section 6 compares our inversions with previous modeling studies. In Sect. 7, we will extend our previous work (Song, 2007; Song et al., 2008) by including both seismic data and GPS measurements in a three-dimensional tsunami model. Section 8 gives a brief summary.

2 Seismographs and GPS Displacements Before discussing specific data and methods used in this study, the vertical deformation of the seafloor due to faulting must be clearly defined. Following Tanioka and Satake (1996), the equation of the vertical seafloor deformation can be written as: h = U + E · hx + N · hy

(1)

where h represents the water depth anomaly due to the seafloor deformation, hx and hy are the eastward and northward slopes of seafloor topography, and E, N, and U are the eastward, northward, and upward components of the seafloor displacement, respectively. Because E, N, and U are functions of location (x, y) and time t, so is the vertical deformation h. Integrating the seafloor deformation over the fault area gives the accumulated seafloor-deformed volume, which is the volume of seafloor collided into the ocean water, mathematically defined as Volume = A |h|dxdy, where A is the undersea fault area. This includes all contributions from both vertical and horizontal displacements of the seafloor. Using seismographs and GPS displacement measurements, we have first estimated the seafloor deformation history of the 2004 Sumatra-Andaman earthquake and the 2005 Nias Island earthquake with three different models: M1: seismic waveform inversion model that estimates the initial seafloor displacements (~800 s) of the earthquake from seismograms. Such a model has been previously used in studying the Sumatra-Andaman earthquake rupture process and the 2004 Indian Ocean tsunami (Ji, 2004; Hjorleifsdottir, 2007; Song et al., 2005). M2: GPS inversion model that uses the continuous GPS measurements to estimate the seafloor displacements. Though far-field, the continuous GPS measurements represent the ground motions within a few minutes to an hour after the initial quake (~1 h), therefore providing the closest possible verification of the seismic inversion model (Blewitt et al., 2006; Song, 2007). M3: the long-term slip model that combines a coseismic model with the campaign GPS (cGPS) data to include the postseismic deformation of the earthquake (~months) (Vigny et al., 2005), providing a constraint to both model M1 and M2.

330


It should be noted that tsunamis are triggered by the initial seafloor displacements during the tsunami formation period, which is less than 30 min because a tsunami phase seldom exceeds an hour. The slow postseismic (afterslip and viscoelastic) deformation can be significant and last for months, but does not contribute to the tsunami excitation. For tsunami generation, M1 and M2 are more relevant than M3 because they give the cumulative seafloor deformation approximately during the tsunami formation period. However, the M3 gives an upper-bound constraint and can be verified by the GRACE measurement, as will be shown later. For the December 2004 Sumatra-Andaman earthquake, these three different models give a seismic moment of 6.2 × 1022 Nm (Mw 9.1), 6.5 × 1022 Nm (Mw 9.12), and 8.8 × 1022 Nm (Mw 9.22), respectively. These estimates are fairly consistent with previous estimations (Ammon et al., 2005; Chlieh et al., 2007). Particularly, M1 and M2 agree well with previous GPS estimate of Mw 8.9–9.1, cumulative during the first 15 minutes of the initial quake (Blewitt et al., 2006). Their cumulative seafloor deformation, including the March 2005 Nias earthquake, is given in Fig. 17.1a–c, respectively.

15N

A. deform (M1)

15N

meter

EQ

5N

EQ

5N

2

1

0

0

0

−1

−1

−1

EQ

−2

95E

D. gravity (M1)

1

EQ

−2

90E

100E

15N

95E

100E

E. gravity (M2)

μGal 15

5N

μGal 15 10

5

5

EQ

5S

−15

85E

5N

95E

100E

F. gravity (M3)

0 −5

−10

85E

μGal 15

5

EQ

0

5S

−15

90E

100E

10

−5

−10

95E

10N

10

0

−2

90E

15N

10N

−5

5S

meter

2

1

10N

5N

C. deform (M3)

10N meter

2

90E

15N

15N

10N

10N

5N

B. deform (M2)

−10 −15

90E

95E

100E

85E

90E

95E

100E

Fig. 17.1 Seafloor deformation history and corresponding gravity changes of the December 26 Sumatra-Andaman earthquake and the March 2005 Nias earthquake cumulated in 800 sec (M1), 1 h (M2), and 6 months (M3), respectively

17


331

3 Gravity Verification Here we show that the vertical deformation of the seafloor can be verified by the space-based gravity measurements from GRACE. This is because the replacement of water by the corresponding seafloor deformation (in addition to intrinsic density change) results in gravity change. Particularly, the gravity change is mostly caused by the vertical component of the seafloor deformation, which is the key quantity needed for testing the tsunami generation theory. Specifically, the oceanic water density (ρo = 1, 025 kg/m3 ) differs greatly from 3 the crust density (ρ c = 2, 750 kg/m ). The volume of seafloor collided into the ocean (Volume = A |h|dxdy, where A is the undersea fault area) would cause the same volume of water to spill away, resulting in a mass change due to the density difference. Following Han et al. (2006), we have converted the mass changes derived from the three models into gravity changes, as shown in Fig. 17.1d–f. The GRACE gravity measurements include changes due to the seafloor deformation and subsurface density compression (internal mass redistribution caused by dilatation of a compressible Earth, including crust and mantle, and Moho surface changes). The subsurface mass changes are under-seafloor variations that should have no effect on tsunami genesis. To separate the subsurface values, we have used the earthquake model (M3) to estimate the internal density changes. In addition, the GRACE observations are averaged over 6 months after the earthquake to enhance spatial resolution and eliminate signals other than those associated with the earthquakes. To be consistent, our models have extended to include the March 28, 2005 Nias earthquake which affected the GRACE observations. The GRACE-observed gravity changes and corresponding model results are shown in Fig. 17.2. It can be seen that the largest positive gravity change is due to the vertical uplift of seafloor, while the negative gravity change is due to the subsidence. Though the model uplift is slightly higher than GRACE values in some area, their area-integrated total gravity are the same, as detailed in Table17.1. Table 17.1 separates the positive and negative values, showing a good agreement with GRACE within a margin of 3%. Particularly, M3 gives a gravity change of 14.9 microgalileo (1 microgalileo = 10−8 m/s2 ), consistent with the GRACE measurement of 15 microgalileo for the same period. In addition, M1 and M2 are well constrained by M3 within a range of 25–35% in deformed volume and gravity change, consistent with recent studies suggesting the postseismic geodetic moment is within the 30% range of the coseismic moment from the Sumatra-Andaman earthquake (Chlieh et al., 2007). A similar coseismic and postseismic ratio is also found in the March 2005 Nias earthquake (Hsu et al., 2006).

4 Test the Conventional Tsunami Theory Once the seafloor deformation is obtained and verified, the task to test the conventional tsunami genesis theory is straightforward. We have tested the theory by two independent approaches. First, we have calculated the tsunami energy from

332


15N

15N

A. surface (M3)

B. subsurface (M3) D26&M28

D26&M28 10N

10N

5N

EQ

μGal 15 10

5N

μGal 0 −5

5

EQ

0

−10

−5 5S

−10 −15 85E

15N

90E

95E

EQ

−20 85E

100E 15N

C. total (A+B)

10N

5N

−15

5S

90E

95E

100E

D. GRACE

10N μGal 10

5N

0

μGal 10

EQ

0 −10

−10 5S

5S −20 85E

90E

95E

−20 85E

100E

90E

95E

100E

Fig. 17.2 Gravity changes: (a) due to surface deformation only; (b) due to density change (dilatation effects) only; (c) their total (a+b, cumulated in 6 months); and (d) the corresponding GRACE measurements

Table 17.1 Volume of seafloor collided into the ocean and corresponding gravity changes (1 microgalileo = 1 μGal = 10−8 m/s2 ) for the December 26, 2004 (D26) and March 28, 2005 (M28) earthquakes D26

M1 (~800 s) M2 (~1 h) M3 (~6 months)

D26 plus M28

Volume

Gravity

Volume

Gravity

362 km3 370 km3 500 km3

10.7 μGal 10.8 μGal 13.3 μGal

406 km3 434 km3 550 km3

11.3 μGal 12.5 μGal 14.9 μGal

17


333

the vertical deformation of seafloor. Assuming the sea surface be raised identically to theseafloor deformation, the ocean would gain a potential energy (i.e., PE = 12 gp A h|2 dxdy) of 1.1 × 1015 Joules from M1 and 1.2 × 1015 Joules from M2, respectively. They are only one fifth of the energy of 4.2–6.2 × 1015 Joules that are needed to match the satellite-observed tsunami height (Song, 2007). The energy calculation is independent of tsunami models, confirming that the ocean could not have gained enough energy from the vertical deformation of seafloor alone to trigger the deadly tsunami. Second, we have used the commonly-used shallow-water equations model to simulate the tsunamis by translating the vertical seafloor deformation directly into the sea surface height as the model initial conditions (Satake, 1995; Tanioka and Satake, 1996). Noted that any tsunami model should be able to replicate this experiment because the satellite observations are very close to the source and there is no room for errors in modeling wave propagations over deep oceans. Fig. 17.3 compares the model results with three satellite altimeters. It can be seen that none of the seafloor deformations estimated from M1 and M2 explains the

Fig. 17.3 Satellite-observed (red) and model tsunamis (blue and green) from M1 and M2: (a) satellite tracks of TOPEX about 1:50 hours, Jason about 2:00 h, and Envisat 3:10 h after the first quake, respectively; (b) tsunami along TOPEX track, (c) tsunami along Jason track; and (d) tsunami along Envisat track. Straight lines at ±30 cm represent the range of background surface waves due to ocean dynamics, eddy and wind forcing

334


satellite-observed tsunami height (Note that M3 is impropriate for tsunami initial because it includes aftershocks beyond the tsunami excitation period). The model tsunamis are only one-third of the actual tsunami heights, suggesting that neither of the seismic and GPS deformation models, nor the GRACE data explains the conventional tsunami theory.

5 Discussion on Wave-Maker Experiments During the 1970s, several wave-making experiments had demonstrated that vertical uplift of seafloor could raise the sea level (e.g., Hammack, 1973). This is true because ocean water is almost incompressible; therefore, the seafloor uplift can be essentially translated into the same amount of perturbation to the sea surface. However, this does not mean the seafloor uplift is the only or major cause of tsunamis because the experiment only had a vertical motion. To our knowledge, no tsunami has ever been generated in an ocean of uniform depth. Instead, all tsunamis have been generated near the continental edges because giant earthquakes often occur where large oceanic plates underthrust continental margins and involve significant lateral displacement of slopes (Johnson, 1999). The experimental study of Iwasaki (1982) is probably the cornerstone in conceiving the “vertically-forced theory” of tsunami genesis. His wave-maker experiment was based on a movable tank with a sloping bottom to reproduce the horizontal motion of ground. Because the accuracy of the instrument used to measure the wave was about 1 cm during that time, the experiment had to generate a wave higher than 1 cm to be measurable. To generate such a wave, the slope was slipped 3–40 cm into the water with a depth of only 6 cm. Unfortunately, the key nondimensional parameter – the ratio of the horizontal displacement L to the water depth D, i.e., R = L/D > 1/ 2 does not represent the actual situation of earthquake tsunamis. The experiment actually represented a scenario that a continental slope had slipped 2 km into a 4 km depth ocean, which has not happened in any recorded earthquake on Earth. In fact, Iwasaki’s non-dimensional ratio R = L/D has been exaggerated up to 200 times of that of the 2004 Sumatra-Andaman earthquake. Table 17.2 compares Iwasaki’s experimental parameters with those representing the 2004 Sumatra-Andaman earthquake. Such a parameter range has not been examined before.

Table 17.2 Comparing Iwasaki’s experimental parameters with the 2004 Sumatra-Andaman earthquake: S, L and D represent the bottom slope near one end of the tank, the total horizontal displacement and the water depth of the tank. The non-dimensional ratio R = L /D in the experiment is more than 200 times of that of the 2004 Sumatra-Andaman earthquake

Iwasaki’s main experiment Sumatra earthquake

S

L

D

R = L/D

1/3 5%

3 cm 10 m

6 cm 4000 m

0.5 0.0025

17


335

One might ask why the scale of the experimental non-parameter matters. As we know, the ocean has to receive energy from the earthquake to generate tsunamis. When a large-scale continental section slips into the ocean, it not only adds volume into the ocean, but also transfers a huge amount of momentum to the ocean (Song et al., 2008). The former would raise ocean’s potential energy, which is proportional to the slip distance; while the latter gives the ocean kinetic energy, which is proportional to the slip speed and other parameters. Therefore, slipping too much volume of seafloor into the water in the experiments could have overshadowed other factors contributing to the tsunami excitation.

6 Compare with Previous Models As we mentioned in the introduction, previous models had replicated historical tsunami waves well (e.g., Titov et al., 2005; Hirata et al., 2006; Grilli et al., 2007). However, they did not verify their tsunami source quantitatively by data, such as the GRACE measurements, before applying the source as model initial conditions. In fact, there are significant differences in deriving the tsunami source. Our tsunami source is derived from seismic inversions and the GPS measurements, while previous models obtain the tsunami sources based on an idealized finite-fault model (Okada, 1985), as shown in Fig. 17.4. Although their finite-fault model does not give the surface displacements, their vertical seafloor deformation can be estimated from 16N

16N

5m

5m

14N

12N

cGPS offsets

14N

cGPS offsets

cGPS−derived

cGPS−derived

Seismic inversion 5m

Seismic inversion 5m

12N 2m

Model

Model

12 m 10N

10N

8N

12 m

17 m

8N

6N

6N

12 m

13 m

4N 23 m

4N 21 m

2N

2N 86E

88E

90E

92E

94E

96E

98E

86E

18 m 88E

90E

92E

94E

96E

98E

Fig. 17.4 Tsunami source model comparison: (left panel) with Titov et al. (2005) and (right panel) with Grilli et al. (2007). The heavy-black arrows represent their slip values over the corresponding subfaults. The red arrows are the campaign GPS displacements (cGPS). The green arrows are the seismic inversion (M1) and the blue arrows are the GPS estimation (M2). All arrows use the same scale

336


the slip values multiplied by a sinusoidal function of the corresponding dip angles (i.e., U = d sin ϕ, where d is the slip distance and ϕ~15◦ is the dip angle). First, it can be seen that their slip values are several times larger than both the seismic inversion and the nearby GPS measurements (Vigny et al., 2005). Secondly, their fault area represented by the subfault boxes is much narrower than that suggested by the seismic and GPS data. The narrower fault and longer slip would result in an overestimated tsunami initial condition (Imamura et al., 1993; Satake, 1994). In fact, their slip values give a range of seismic moment 8.8–9.8 × 1022 Nm (Mw 9.2–9.3), estimated by Mo = μAd with commonly used fault rigidity, μ = 3–4 × 1010 N/m2 , where A is the fault area and d is the slip distance. This is much higher than our estimates from seismographs and GPS, as well as the 1-day accumulated seismic moment (coseismic and postseismic) of 6.7 × 1022 Nm (Mw 9.15) (Chlieh et al., 2007). Fairly to say, their modeling studies were to investigate the tsunami propagation and run-ups, rather than to verify the tsunami genesis theory. However, the confusions on the tsunami source have to be clarified here.

7 Momentum Energy Transferred by Earthquakes To demonstrate the possibility of unaccounted tsunami sources existing, we have employed a three-dimensional ocean model that allows including both the seasurface perturbation due to vertical displacement of seafloor and the impulsemomentum perturbation due to the horizontal motions of continental slopes. The idea of including an earthquake-induced momentum perturbation in the tsunami model is based on the impulse-momentum principle of fluid mechanics (Vennard and Street, 1982). In that sense, it is just like throwing a rock into a pond, the wave height is not only depends on the size (potential energy) of the rock, but also the speed and shape (kinetic energy) of the rock. Both energies are equally important in generating waves (the energy principle of fluid mechanics). When a large-scale continental section is slipped into the ocean due to earthquakes, a transfer of momentum occurs and a three-dimensional force is exerted on the fluid through a distance. This idea has been successfully demonstrated by Song et al. (2008) based on a three-dimensional tsunami model. For completeness, we have extended the earlier study by including both seismic data and GPS measurements. The threedimensional tsunami model is briefly illustrated as the following. Let x, y and z be the eastward, northward and upward coordinates, the oceanic equations in homogeneous fluid can be written as: ∂η ∂ ∂u ∂u + v · ∇u − fv = −g + K ∂t ∂x ∂z ∂z ∂v ∂η ∂ ∂v + v · ∇v + fu = −g + K ∂t ∂y ∂z ∂z ∂v ∂w ∂u + + =0 ∂x ∂y ∂z

(2) (3) (4)

17


∂ ∂ ∂(η + h) + (D¯u) + (D¯v) = 0 ∂t ∂x ∂y

337

(5)

where v = (u, v, w) is the three-dimensional oceanic velocity, ∇ is the gradient operator, f is the Coriolis parameter due to Earth’s rotation, η is the sea-surface elevation, h is the bathymetry, and D = η + h is the water depth. Also, K is the η η udz and v¯ = D1 vdz are the depth-averaged velocity. vertical viscosity, u¯ = D1 −h

−h

Let (E, N, U) be the three-dimensional GPS or seismically-derived seafloor displacements, which represents a motion of a grid size of Δx = 0.125◦ by Δy = 0.125◦ (a subfault) in this study. Based on the impulse-momentum principle of fluid mechanics (Vennard and Street, 1982), the accelerated three-dimensional velocity of water particles in the vicinity of the moving seafloor can be written as:

ub (z) =

E/τ if −h ≤ z ≤ −Rx = min{h, LH |hx |} 0 othewise

N/τ if −h ≤ z ≤ −Ry = min{h, LH |hy |} 0 othewise wb (z) = U + E · hx + N · hy /τ −h ≤ z < 0 vb (z) =

(6) (7) (8)

where represents the impulse velocity only during the rise-time period τ of faulting, h is the ocean topography, and hx and hy are the eastward and northward slopes of the subfault surface, respectively. Also LH is the effective scale of the horizontal motion, z is the vertical coordinate at the undisturbed ocean surface, and ub (z) and vb (z) are the bottom-water velocity within the range of Rx and Ry , respectively. Here, wb (z) is the vertical velocity due the seafloor uplift. The velocity is actually the displacement divided by a rise-time τ (We have used 5–10 s for the 2004 Sumatra earthquake) within the water near the moving bottom. Notice that a flat bottom would have no contribution to the tsunami source because the slope is zero (i.e., Rx = Ry = 0). A parallel slip component would have no contribution either because the slope in that direction is zero (i.e., Rx = 0 or Ry = 0). It should be noted that the vertical acceleration of water particles does not contribute to the tsunami propagation, but the resultant sea-surface perturbation does. Because the vertical velocity wb (z) can be approximated by the ocean bottom/seasurface perturbation with the relationship: dh(t)/dt ≈ wb (z), our vertical velocity condition of Eq. (8) is actually equivalent to the conventional assumption of the initial sea-surface perturbation (Tanioka and Satake, 1996): η0 (x, y, t) ≈ h = U + E · hx + N · hy .

(9)

This part is essentially the conventional tsunami source, from which the ocean would gain potential energy. The finite energy can be calculated over an area of Δx·Δy by

338


PE = g · ρ ·

h2 · x · y, 2

(10)

where g is the gravity acceleration and ρ is the water density. The total accumulated potential energy is the integration of (10) over the whole faulting area. Now, we explain how the horizontal velocity can be applied to the ocean model. In the tsunami source Eqs. (6), (7), and (8), the only unknown parameter is the horizontal effective scale LH . In the deep ocean, the linear-wave theory implies that the tsunami height proportionate to the total source energy: the potential energy due to the seafloor uplift and the kinetic energy due to the horizontal motions of water. Because the potential energy can be estimated from the seafloor uplift, the kinetic energy has to complement for the total tsunami energy. It is found that LH = 1.5 hmax (the maximum of ocean depth) is a proper value for the 2004 Sumatra earthquake. This is what we have expected because LH cannot be arbitrary and should be constrained by the ocean depth. The initial forcing (within the period t < τ ) for the horizontal momentum can be written as u0 (x, y, z, t) = ub (z),

(11)

v0 (x, y, z, t) = vb (z).

(12)

It should be noted that the conditions apply to the ocean model only during the risetime and near the bottom of fault area. Based on the bottom velocity formulation, the tsunami kinetic energy gained by the ocean over an area of Δx·Δy due to the horizontal motion can be calculated as KE = ρ ·

1 2 ub + v2b · z · x · y, 2

(13)

where Δz is the vertical grid size in the bottom layer. The total accumulated kinetic energy is the integration of (13) over the whole faulting area. In summary, Eqs. (9), (11), and (12) give the three-dimensional tsunami source, and Eqs. (10) and (13) give the total tsunami-source energy. The momentum perturbation can be estimated from the GPS displacements or the seismically-inverted earthquake source, as shown in Fig. 17.4. The global ocean model with ocean dynamics and the tsunami only (after removing the ocean dynamics) is shown in Fig. 17.5. By including the seafloor-deformed velocity (horizontally) along with the seafloor-deformed volume (vertically) in the three-dimensional ocean model, we have found that the earthquake models M1 and M2 are able to replicate the 2004 Indian Ocean tsunami well (Fig. 17.6). They give a total tsunami source energy of 5.2 × 1015 Joules and 6.0 × 1015 Joules, respectively, in which the kinetic energy account for about 65% of the observed tsunami strength, while the potential energy accounts for the other 35%. In addition, the asymmetric tsunami pattern, recorded by tide gauges showing leading-elevation waves toward Sri Lanka and

17


339

Fig. 17.5 The 2004 Indian Ocean tsunami in the three-dimensional model with ocean dynamics (upper panel) and without the ocean dynamics (lower panel). Color units are in centimeters. Note that the tsunami height has to exceed the variability range of ocean dynamics significantly to become a killer tsunami. The mean variability range of the ocean dynamics is 30 cm or higher in the deep

leading-depression waves toward Thailand, is best explained by the horizontallyforced components (Fig. 17.7). Interestingly, the kinetic energy is more sensitive to the faulting direction and speed (rise-time) of the continental slope, i.e., the way of the earthquake, rather than the magnitude of earthquakes alone.

8 Summary In summary, wave-makers and tsunami models can simulate tsunami propagating processes, but they are not a proof of the vertically-forced tsunami theory. The 2004 Indian Ocean tsunami event has provided unprecedented amounts of data on the earthquake and the tsunami. For the first time, GRACE gave a synoptic verification of the vertical component of the seafloor deformation – the ‘Holy Grail’ of the long-held tsunami genesis theory. We have found that evidence from seismographs

340


25N

cm A.

Satellites

60

15N

5S

20

0 cm

0

−50

−20

TOPEX 15S

−40

ENVISAT

Jason

65E

75E

85E

95E

105E

cm 60

−60 10S

5S

Eq

5N

10N

15N

20N

5N

10N

15N

20N

cm C.

EVISAT

60

40

40

20

20

0

0

−20

−20

−40

−40

−60

TOPEX

40

50 5N

B.

20S

10S

Eq.

10N

20N

−60 10S

D.

Jason

5S

Eq

Fig. 17.6 Same as Fig. 17.3, but based on the three-dimensional tsunami theory by including the horizontal impulse-momentum due to the faulting continental slopes (green from M1 and blue from M2)

and GPS is consistent with GRACE measurements, but contradict to the verticallyforced tsunami theory. This study does not advocate any other tsunami theories under debate (Dutykh et al., 2006; Song et al., 2008). However, we emphasize the importance of searching for the true cause of tsunami genesis. Without knowing how tsunamis form from earthquakes, it would be impossible to develop a reliable warning system for saving lives and property during tsunami emergencies. In this study, we have analyzed multiple satellite data and ground-based measurements. We have also used a three-dimensional ocean model to demonstrate the significance of the earthquake-transferred oceanic momentum in exciting tsunamis. Our methodology demonstrates that combing satellite data with in-situ measurements and using advanced ocean models can be an effective approach to solving the tsunami genesis puzzle. Acknowledgements The research described here was conducted partially at the Jet Propulsion Laboratory, California Institute of Technology, under contracts with the National Aeronautics and Space Administration (NASA). We appreciate the valuable contributions from Prof. Hiroo Kanamori, who has participated in discussing the work, revising the manuscript, and providing helpful comments.

17

Satellite Observations Defying the Long-Held Tsunami Genesis Theory (a) Tide Stations (2004)

(c) Thailand side (leading depression)

25N 15N

1

Thailand

+1

2♦ 3♦ 4♦

5N 5S

+ ++5 342 + 6 +7 +8+7 +9

1♦ Sri Lanka

2

5♦

15S 65E

75E

3 85E

95E

105E

(b) Sri Lanka side (leading elevation)

1

−− GPS −− Seismic

−− Tide

6:57N, 79:51E

3m interval

3 4

PortBlaire 11:41N, 92:46E Ranong 10:00N, 98:30E Kuraburi 9:14N, 98:23E Krabi 8:05N, 99:00E Taphaonoi 7:50N, 98:25E Tarutao 6:42N, 99:39E

6:46N, 73:10E

7

Male

Tummarang 6:15N, 100:02E

4:11N, 73:32E

8

Gan 0:41S, 73:09E

9

7:17S, 72:24E

100

200

Belawan 3:55N, 98:43E Sibolga

Diego Garcia

5

5 6

Hanimaadhoo

2

3m interval

4

Colombo

341

300 Time (min)

400

1:44N, 98:48E

−200

0

200 Time (min)

400

Fig. 17.7 Validations by tide gauges for the 2004 Indian Ocean tsunami: (a) tide stations, (b) closest stations in the Sri Lanka side (recorded the leading elevation waves), and (c) closest stations in the Thai side (recorded the leading depression waves). Numbers denote the tide stations on the map. If the model grid is not on a tide location, a nearest point is used with a depth correction

References Abe K (1973) Tsunami and mechanism of great earthquakes. Phys Earth Planet Inter 7:143–153 Ammon CJ et al. (2005) Rupture process of the 2004 Sumatra-Andaman earthquake. Science 20:1133–1139 Blewitt G et al. (2006) Rapid determination of earthquake magnitude using GPS for tsunami warning systems. Geophys Res Lett 33:L11309. doi:10.1029/2006GL026145 Chen JL et al. (2007) GRACE detects coseismic and postseismic deformation from the SumatraAndaman earthquake. Geophys Res Lett 34:L13302. doi:10.1029/2007GL030356 Chlieh M et al. (2007) Coseismic slip and afterslip of the great Mw 9.15 Sumatra-Andaman earthquake of 2004. Bull Seismol Soc Am 97. doi:10.1785/0120050631 Choi BH, Hong SJ, Pelinovsky E (2006) Distribution of runup heights of the December 24, 2004 tsunami in the Indian Ocean. Geophys Res Lett 33:L13601. doi:10.1029/2006GL025867 Dutykh D, Dias F, Kervella Y (2006) Linear theory of wave generation by a moving bottom. C R Acad Sci Paris Ser I 343:499–504 Grilli T et al. (2007) Source constraints and model simulation of the December 26, 2004 Indian Ocean tsunami. J Waterway Port Coast Ocean Eng 133(6):414–428

342


Hammack JL (1973) A note on tsunamis: their generation and propagation in an ocean of uniform depth. J Fluid Mech 60(4):769–799 Han S-C et al. (2006) Crustal dilatation observed by GRACE after the 2004 Sumatra-Andaman earthquake. Science 313:658–662 Hirata K et al. (2006) The 2004 Indian Ocean tsunami: tsunami source model from satellite altimetry. Earth Planets Space 58:195–201 Hjorleifsdottir V (2007) Earthquake source characterization using 3D numerical modeling. PhD Thesis, Caltech. http://etd.caltech.edu/etd/available/etd-03212007-170259/ Hsu Y-J et al. (2006) Frictional afterslip following the 2005 Nias-Simeulue earthquake, Sumatra. Science 312:1921–1926 Imamura F, Shuto N, Ide S, Yoshida Y, Abe K (1993) Estimate of the tsunami source of the 1992 Nicaraguan earthquake from tsunami data. Geophys Res Lett 20(14):1515–1518 Iwasaki S (1982) Experimental study of a tsunami generated by a horizontal motion of a sloping bottom. Bull Earthq Res Inst Univ Tokyo 57:239–262 Ji C (2004) http://neic.usgs.gov/neis/eq_depot/2004/eq_041226/neic_slav_ff.html Johnson JM (1999) Heterogeneous coupling along Alaska-Aleutians and interred from tsunami, seismic, and geodetic inversions. Adv Geophys 39:1–116 Moran K, Austin JA, Tappin DR (2005) Survey presents broad approach to tsunami studies. Trans EOS, AGU 86:430 Merrifield MA et al. (2005) Tide gauge observations of the Indian ocean tsunami, December 26, 2004. Geophy Res Lett 32:L09603. doi:10.1029/2005GL022610 Okada Y (1985) Surface deformation due to shear and tensile faults in a half-space. Bull Scismol Soc Am 75:1135–1154 Satake K (1994) Mechanism of the 1992 Nicargua tsunami earthquake. Geophys Res Lett 21: 2519–2522 Satake K (1995) Linear and nonlinear computations of the 1992 Nicaragua earthquake tsunami. PAGEOPH 144:455–470 Song YT (2007) Detecting tsunami genesis and scales directly from coastal GPS stations. Geophy Res Lett 34. doi:10.1029/2007GL031681 Song YT et al. (2005) The 26 December 2004 tsunami source estimated from satellite radar altimetry and seismic waves. Geophys Res Lett 23. doi:10.1029/2005GL023683 Song YT et al. (2008) The role of horizontal impulses of the faulting continental slope in generating the 26 December 2004 tsunami. Ocean Modell. doi:10.1016/j.ocemod.2007.10.007 Tanioka Y, Satake K (1996) Tsunami generation by horizontal displacement of ocean bottom. Geophy Res Lett 23(8):861–864 Titov VV et al. (2005) The global reach of the 26 December 2004 Sumatra tsunami. Science 309:2045–2048 Tuck EO, L-S Hwang (1972) Long wave generation on a sloping beach. J Fluid Mech 51:449–461 Vennard JK, Street RL (1982) Elementary fluid mechanics, 6th edn. John Willey & Sons, New York, p 689 Vigny C et al. (2005) Insight into 2004 Sumatra-Andaman earthquake from GPS measurements in Southeast Asia. Nature 436:201–206

Chapter 18

Tsunami Source Reconstruction by Topex/Poseidon Data Vladimir V. Ivanov

Abstract The data of the Topex-Poseidon mission (altimetry, files MGDR), recorded on the Sea, bay or strait shows essential deflection of the Sea surface from geo-id. The amplitude of deflection is near to 1 m, the correlation radia is near to 50 km. Most deflections are viewed as the same one for different cycles of the mission, if we examine the same point of the Earth. This means that the main part of anomaly do not depend on time. The variation of the surface’s deflection on distance shows the peculiarities on the continental border. The several maximums and minimums are located on the boundary between the continent and ocean. The deep minima are observed on the variations recorded for each track which cross the trench. The tracks 34, 212, 136, 60 cross the trench between Pacific and Asia. If we join together the sequential minima (the sequential from North to the South) we receive the line of ravine on the surface of geo-id. This line is the especial line from geophysical point of view. The epicenters of the most earthquakes are placed near the line, the principal axis of tsunami sources is directly along the line. We interpret the deflections as a horizontal gravity anomaly with amplitude 10−5 g. The great earthquake (October 4, 1994, Shikotan, M = 8.3 and December 28, Sanriku, M = 7.6) creates the most notable changing of gravity deflections. The variation has an amplitude of several percent (10−6 g). The perturbation is disposed near the area of aftershocks’ epicenters. The perturbation arrived after the main shock of the Earthquake in the case of earthquake December 4, 1994. The perturbations grow during the 6 months before the earthquake December 28, 1994 and shrink after the main shock. Analogous phenomena are observed after the earthquake on December 26, 2004 (Sumatra).The epicenter of the earthquake is placed in the ravine that spreads from the point 3.37 N. 94.4 E to the point 12.7 N 92.4 E. The depth of ravine is 1.4 m. After earthquake the ravine disappears. V.V. Ivanov (B) Institute of Marine Geology & Geophysics, Yuzhno-Sakhalinsk, Russia e-mail: [email protected]


343

344

V.V. Ivanov

If we regard the geo-id’s perturbation as a result of gravity variation, then we can estimate the parameters of tsunami sources. If we use the piston model of tsunami source, then the parameters of plunger can be estimated. The location of plunger looks like the location of the aftershock’s area. The height of plunger can be estimated by amplitude of gravity variation, if we use the potential theory. It is 58 m for the 1994 earthquake and it is the 150 m for the December 26, 2004 tsunami. Keywords Geo-id · files MGDR · Topes/Poseidon mission · gravity anomaly · sea level · potential

1 Introduction The investigation of the nature of great earthquakes is a series problem of geophysics. At the last time, the quality and quantity of measurements were improved, we have a chance to reconsider the old perceptions and to complete it by new measurements. One of the concepts, connected with earthquakes, is the concept of tsunami source. Today, the piston model was regarded as a theory of tsunami generators. It is proposed that tsunami sources arise at the moment of the main shock of earthquake simultaneously in all areas, and that it is the deformation of the bottom on the area S with amplitude H. The plunger parameters (cross section S and height H) are estimated by the interpretation of different observations. The cross section of plunger is estimated by interpreting the arrival time of the tsunami in different points. The line, corresponded to time propagation of wave, is constructed for every point. The envelope line of this set is the boundary of the plunger section. The height of the plunger is estimated by seismic moment estimation. In this presentation we propose to estimate the plunger parameters by the data of Topex/Poseidon mission. The satellite measurements of Sea level permit analysis of the gravity fields of great earthquakes. The results of measurements of gravity fields contain the information about bottom perturbation connected with earthquakes. To estimate the plunger parameters we propose that the density of plunger media is the same as an average density of the Earth (ρ = 5.5 g/cm3 ). The information for the piston reconstruction is the variation of the gravity anomaly, which is connected with the great earthquake.

2 Theory In this section we write the equation, which connect the gravity variations and parameters of piston, which is described as a bottom perturbation h(x,y) (the value of depth perturbation in point with coordinates x, y).

18

Tsunami Source Reconstruction by Topex/Poseidon Data

345

By using the potential theory we can write the perturbation of gravity potential φ, connected with piton arising, as (r) = f/(4π )

(ρ − 1)/(r − r’) d’.

(1)

f– gravity constant, r –coordinates of regarded point, r’ – coordinates of perturbation. The integration is spread by volume of perturbation. If density is not dependant on r’, after integration by vertical coordinate, we find on the bottom that (x, y) = f

(ρ − 1) h(x’, y’)/((x − x’)2 + (y − y’)2 )1/2 dx’dy’

(2)

Here, it is assumed that the horizontal size of source and distances x−x’, y−y’ to the source are much more than height of the piston h. This formula can be used for estimation of value h(x,y), if we have measured the variations of potential (x,y). The measured value h(x,y) is connected with value (x,y). We suppose that the surface of sea is the equipotential surface. It means that the value ∂ (x,y)/∂x/g = ∂H(x,y)/∂x.

(3)

The last values can be estimated by data of Topex/Poseidon mission. So, we need to differentiate the Eq. (2) by x and by using (3) to write the Eq. (4), which connect the observable value ∂H(x,y)/∂ x and bottom perturbation h(x,y) ∂H(x,y)/∂x = − f/(4π g)

(ρ−1) h(x’,y’) (x − x’)/((x − x’)2 + (y − y’)2 )3/2 dx’dy’ (4)

Formula (4) is the main result of the topic. It connects the measured value H and unknown quantity h.

3 The Observation of Sea Surface by Topex/Poseidon Mission The Topex/Poseidon mission provides the information for an estimation of “sea level variations without tides and inverse barometer” (Benada, 1993). The data are received along the tracks every 10 days through the each 5.8 km. Full number of tracks 254. The even number corresponds to satellite movement from the North to the South. The odd numbers correspond to motion in direction from South to the North. The parts of tracks, which crossed the area of Sakhalin region, are shown in Fig. 18.1. The stable deflections of surface from geo-id are observed on the data, which is received on the shallow water (Ivanov, 2003). These variations are 10 times more than the surface current variation. The especially large variations are observed on the continental slope near to the trench. Two stable, deep minima are observed

346

V.V. Ivanov

Fig. 18.1 The tracks of Topex/Poseidon mission on the map of Sakhalin region. The track numbers (086, 162, 238) are indicated near the boundary. The stars show the positions of the great earthquake epicenters. The numbers near the star are earthquake data

on every track. The stable component of variation for track 212 is shown on the Fig. 18.2. There is shown the mean data for 10 cycles (100 days). We interpret the variation as a result of a horizontal gravity anomaly. The anomaly is two dimensional vectors with amplitude near to 10−5 g. The estimation succeeds from the value of slope of geo-id . = g The component along the track can be estimated by data of Fig. 18.2. Between the points 148.5◦ E and 149◦ E the level changes on the value 2.4 m (distance 60 km), so = 2.4/60 103 = 0.4 10−4 g. The picture of variation contains the characteristic peculiarities. In the area of continental slope and trench the sequence of maximums and minimums are observed. This picture is observed on the each track, which cross the trench. The stable component of variation along the tracks 34, 212, 136, 60 are shown on the Fig. 18.3. The tracks 34, 212, 136, 60 are located sequential from North to the South.

18


347

Fig. 18.2 The sea level variation recorded by Topex/Poseidon mission. Mean value for 10 cycles. The time interval between the different lines is near 100 days, track 212

We can see (Fig. 18.1) that the epicenters of the great earthquakes are located near to the track. The epicenter projections to the track are shown on the Fig. 18.3. We can see that the locations of epicenters and anomalies are correlated. We join the positions of sequential minima and display this line on the map (Fig. 18.4). On the same map are shown the epicenters of the great earthquakes for period 1949– 1986 and sources of tsunami. We can see that the epicenters are located on this line and the main axis of tsunami sources are parallel with direction of the line. It is natural to propose that the location of the anomaly is connected with ability of area to generate the great earthquake. We are looking for the anomaly variation near these points. We have a chance because of earthquakes October 4, 1994 (Shikotan, magnitude 8.3) and December 28, 1994 (Sanriku, magnitude 7.6 [5]), which occur during the action of Topex/Poseidon mission. The epicenter of the first one is located near the track 212, the epicenter of the second one is located near the track 60.

4 The Variation of the Gravity Anomaly due to the Earthquake October 4, 1994 We need to make several remarks about processing of data (files MGDR). Apparently, the value of variation is less than the stable component. It is necessary to subtract the statinary component for observation as it was done in paper for

348

V.V. Ivanov

Fig. 18.3 The stable component of sea level variations along the tracks 34, 212, 136, 60, which cross the continental slope in vicinity of Kuril trench. The lines show the projection of epicenters to the track. The numbers are the dates of earthquakes

18


Fig. 18.3 (continued)

349

350

V.V. Ivanov Table 18.1 The sample of record of camputation for hindrance deleting. Track 60

Shift

Correlation max

Correlation min

77 78 79 80 81 82 83 84 85

0 −0.9 −2 −0.3 −0.9 −1 0.8 −1.5 0.8

0.99 0.99 0.95 0.99 1.00 0.98 1.0 0.99 0.99

0.85 0.86 0.71 0.91 0.87 0.83 0.92 0.8 0.92

tsunami discoveries. It is necessary to take into account the trajectory deflections from one cycle to another. The additional hindrance arrives due to this deflection. This hindrance can be deleted if we have corrected data. The correction can be done after shifting of trajectory. Usually the shift value is less than one step of measuring. The procedure was described in paper (Ivanov, 2003). The protocol of camputation for cycles from 75 to 86 (Track 60) was shown on the Table 18.1. The number of cycle was shown in the first column. The value of shift was shown in the second column, the maximal value of correlation in the thirth column, the minimal value in

Fig. 18.4 The seismic activity and gravity anomalies on the shelf of Kurile Islands

18


351

last one. The value of shift is recorded in digital unit (5.8 km) of Topex/Poseidon mission. The results of cleaning for cycles from September 4 to November 13, 1994 for track 212, which pass by the epicenter of the Earthquake October 4, are shown on the Fig. 18.4. The epicenter is corresponded to the point 148.6◦ . The data for cycle 77 (October 14, 1994 goda) is absent in the files of MGDR. We can see that as a result of the earthquake, the gravity anomaly changes for several percent in the region, located to the Northwest of the epicenter. This is the area of maximal activity of the aftershock process. The area of aftershock’s process and trajectory of satellite are shown on the Fig. 18.5. On the Fig. 18.6a the space density of aftershocks are compared with variation of the gravity anomaly. This process was completed to October 24. The amplitude of additional anomaly is 0.07 m, the characteristic distances is 30 km. The corresponding gravity anomaly is 2.3 10−6 g. The space variation is not simple one. It reflects the complicated peculiarity of faults during this earthquake (Ivanov, 1995, 1998). In result of this earthquake the three rupture planes arrived. If we guess that the length of tsunami source is much more than width, then we can simplify the Eq. (4) for tsunami source height. In this case we can regard the two-dimensional pictures of perturbation. The gravity potential perturbation φ is described by equation 1/xd/dx x dϕ/dx + d2 ϕ/dy2 = f h(x) (ρ−1)

(5)

If the length of tsunami source is much more than width, then 1/xd/dxxdϕ/dx >>> d2 ϕ/dy2 and or by using the expression (3) 1/xd/dx x gdH/dx = f h(x)(ρ − 1) and expression (5) for g g = 4fπ/3 R, R = 6, 370 km – the radius of Earth, we receive the estimation of plunger height, h(x) = ρ(ρ − 1) 4π/3R dH/dx,

(6)

by using the value dH/dx = 2.3 10−6 we receive the estimation of h as 58 m (Fig. 18.7). We can see that the effect of earthquake is much more than we usually represent (Yoon, 1994). The great earthquake occurs at December 26, 2004 at point 3.3◦ N, 95.87◦ E near the Sumatra Island. The map of earthquake’s area is displayed on the Fig. 18.8. The epicenters of the aftershocks and the tracks of the Topex/Poseidon mission are displayed also. The broken line is the trajectory of the seismic process, It joins together the successive epicenters of aftershock.

352

V.V. Ivanov

Fig. 18.5 The trajectory of seismic process for earthquake October 4, 1994 and track 212 of Topex/Poseidon mission

The epicenter of earthquake (white square) is located outside of aftershock area. We guess, that the tsunami source is connected with aftershocks and coincide with trajectory of seismic process, which is shown on the picture as a broken line. This line connects the epicenters of sequential aftershocks. The source area are crossed by track 205 (North part) and track 27 (the central part). The sea level variations along these tracks, which are coincided with surface of geo-id, are shown on the Fig. 18.9. The points of tsunami source crossing are designated as A and B. We can see that the tsunami have been generated in vicinity of deep minima of geo-id surface, as it was in the case of earthquake October 4, 1994. The tsunami source area coincides with ravine on the geo-id surface. The measurements of Topex/Poseidon mission after earthquake have been received only for track 27. These data before and after earthquake are shown on the Fig. 18.10. We can see that the geo-id surface alters essentially as a result of the earthquake. The ravine (in which the earthquake is located) vanishes. Thus we can conclude that the gravity anomaly is the important cause of the Earthquake. Using the formula 6 we can estimate the height h of tsunami source. The value dH/dx for this case estimate as δH/δx WhereδH = 0.6 m, δx = 90 km,δH/δx = 6 10-6, h = 146 m. In this case the tsunami intensity 4 times more than in the case of earthquake October 4, 1994. We can see that the geo-id changes after the earthquake and ravine disappears.

18


353

(a)

(b)

Fig. 18.6 (a) The cleaned records for track 212 for different cycles before and after the earthquake October 4, 1994. The epicenter projection is the point 148.6◦ . (b) Comparing of gravity anomaly (upper line) with anomaly variation after the earthquake (down line)

354

V.V. Ivanov

Fig. 18.7 The comparison on aftershock’s area and area of maximal variations of gravity

Fig. 18.8 The map of earthquake 26 December 2004 at Sumatra. The white lines with number – the tracks of Topex/Poseidon mission

18


355

Fig. 18.9 The geo-id surface in earthquake area. The points indicate the cross points of track and source

Fig. 18.10 The geo-id surface before and after earthquake December 26, 2004

356

V.V. Ivanov

5 The Evolution of Gravity Anomaly During the Earthquake We can conclude that the investigation of development of gravity anomaly is the main cause of problems in earthquake investigation. In the case of December 28, 1994 earthquake, the gravity anomaly was growing during the 6 months before the main shock of the earthquake. The gravity anomaly changed as a result of the main shock of the earthquake October 4, 1994, which is located 200 km from epicenter. After the earthquake on December 28, the created part of anomaly at epicenter disappeared. The anomaly developed in vicinity of epicenter. In the point of epicenter the space dependence had peculiarity, the sign of variation changed. The maximal value of anomaly was observed after the earthquake October 4. Its value is 0.04 m per 10 km, or 4 10−6 g. The evolution of gravity anomaly is shown on the Fig. 18.11

the earthquake

The gravity anomaly and variation due to earthquake

Fig. 18.11 The gravity anomaly evolution. Left side: the anomaly records before and after earthquake. Right side: The additional anomaly before the earthquake. December 28, 1994

18


357

6 Discussion The gravity anomaly observations near the epicenter of the great earthquake show that the gravity field is the important signal, which contains the information about the process of earthquake evolution. This signal contains the information about development of dangerous anomaly. The space variations indicate the place of dangerous earthquake, the depth of ravine shows the maximal amplitude of the possible tsunami wave. This information sometimes confirms old results (conceptions about several gaps arriving due to earthquake October 4, 1994 (Ivanov et al., 1998). Sometimes we can observe the peculiarity of process, which can not be observed by other means. The new information permits estimation of the scale of phenomena. Acknowledgement The files of MGDR have been received from NASA Physical Oceanography Distributed Active Archive Center, Jet Propulsion Laboratory, California Institute of Technology. The work supported by the grant RFFI 08-05-01096.

References Benada R (1993) Merged GDR (Topex/Poseidon) Users Handbook. Physical Oceanography Distributed Active Archive Center, p 142 Ivanov VV (1995) The motion of the seismic source during the earthquake July 12, 1993. Physic Earth g N11:c. 3–17 (Russian) Ivanov VV (1998) The interpretation of geophysical data for tsunami warning. Dissertation, Institute of Oceanography, RAN, Moscow, p 343 Ivanov VV (2003) The satellite measurements of sea level interpretation. The Earth’s investigations from the space. N3, pp 1–8 Yoon In Taek (1994) Numerical Experiments on the Tsunami of the 1993 South West of Hokkaido Earthquake. Seoul National University, p 138

Chapter 19

Scientific Research Based Optimisation and Geo-information Technologies for Integrating Environmental Planning in Disaster Management Hussain Aziz Saleh and Georges Allaert

Abstract Natural and environmental disasters have profound social, economic, psychological, and demographic effects on the stricken individuals and communities. The literature of disaster management of the 21th Century has pointed out that there is a missing part in the knowledge, scientific research, and technological development that can optimise disaster risk reduction. With the improvement of dynamic optimisation and geo-information technologies, it has become very important to determine optimal solutions based on the stability and accuracy of the measurements that support disaster management and risk reduction. However, a scientific approach to the solution of these disasters requires robotic algorithms that can provide a degree of functionality for spatial representation and flexibility suitable for quickly creating optimal solution that account for the uncertainty present in the changing environment of these disasters. Moreover, the volume of data collected for these disasters is growing rapidly, and sophisticated means to optimise this volume in a consistent, dynamic and economical procedure are essential. This chapter effectively links wider strategic aims of bringing together innovative ways of thinking based on scientific research, knowledge and technology in many scientific disciplines to providing optimal solutions for disaster management and risk reduction. Real-life applications using these disciplines will be presented. Keywords Disaster management · Risk assessment · Geo-information technology · Early warning · Artificial intelligence · Dynamic optimization · Environmental planning

H.A. Saleh (B) Higher Commission for Scientific Research, P.O. Box 30151, Damascus, Syria; Institute for Sustainable Mobility, Ghent University, Gent, Belgium e-mail: [email protected]; [email protected]; [email protected]


359

360

H.A. Saleh and G. Allaert

1 Introduction In the past several years, natural disasters which have major impacts in every corner of the world have dramatically increased. They cannot be prevented and it is not always possible to completely eliminate their risks, but they can be forecasted at times, enabling people to properly deal with their consequences. Extensive experience and practice in the past few decades has demonstrated that the damage caused by any disaster can be minimised largely by careful planning, mitigation, and practical actions. For example, with the help of advanced geo-information technology, hurricanes can clearly be seen before they cause devastation and this will enable preparation and minimisation of expected damage. Early planning has saved lives, but additional planning could have further reduced destruction. Many scientific studies have considered the effects of these disasters, but few have searched for ideal solutions. Scientific research and analysis of hazard data is needed before (risk analysis, prevention, preparedness), during (emergency aid), and after a disaster (reconstruction) to understand its effect and dimensions. This will help and support determining how best to respond to existing and potential losses, and how to aid effectively with recovery activities. However, risk reduction measures have to be considered and evaluated according to several parameters and factors such as social, demographic, and environmental effects, economical cost, available technology, etc. Much more work and research is needed as there are many gaps in our knowledge and understanding of the changing behaviour of these disasters. In particular, there is a lack of efficient use of geo-information and communication technologies that are powerful sources for providing accurate information, facilitating communication, and permitting the monitoring of emergency conditions and impacts. The optimal study for disaster management and risk reduction is based on scientific research, information, knowledge and technology development such as electronic comparisons using innovative methods that help the decision makers to accurately understand the relationships between reasons and results, to differentiate between the strategic and secondary objectives, and to measure and analyse the gap in performance between the optimal and local models of the disaster. The knowledge methods used in disaster management consist of the varieties of the principals and procedures based on scientific research such as trail and error, action and reaction, simulation, modelling, and dynamic optimisation, etc. Also, these methods use performance scales based on e-benchmarking to test and analyse the developed techniques for disaster management and mitigation and to define the improvement and development domains. The information systems presented by the internet revolution provides and supports databases with all types of information about disasters and experts anywhere and anytime. While the knowledge system is presented with electronic development in the early warning and information technology through establishing and updating databases, other important functions and innovative methods are essential to achieve results. These functions and methods are: urgent services, creating information system about all disasters, constructing a site on the net for exchanging information using emails and direct communication to support the decision makers, doing the continuous improvements in disaster management, predictions, models, and indicators, etc. This research

19

Scientific Research Based Optimisation and Geo-information Technologies

361

insists on the importance and the necessity of an intelligent system based on the scientific research and technology development for disaster management and risk reduction. To achieve an efficient solution to disaster risk reduction, this chapter links wider strategic aims of bringing together innovative methods based on many scientific disciplines (e.g., geo-information technology, earth observation techniques, artificial intelligence, early waning systems, dynamic optimisation, risk analysis and environmental impact assessment, spatial and environmental planning, etc). More precisely, the purpose of this chapter is to implement robotic algorithms for providing automated data processing strategies to find optimal solutions to disaster management and risk reduction. This will provide access to a wide range of data collected at an investigated region, and combine the observational data with practical data analysis in order to improve forecasting and risk assessment. This chapter highlights the critical role of technology in disaster reduction and management and identifies a few key areas for strengthening/improving technological inputs to the operational system. Section 2 discusses the disaster management cycle and presents all the practical activities that must be carried out during all the phases of this cycle to minimise the disaster risk reduction. Also, it outlines disaster management procedures and components of the hazards analysis for risk assessment. Section 3 presents the most recent processes that have been made through advances in early warning and observing systems, computing and communications, scientific research and discoveries in earth science, and how this is helping to understand the physics of hazards and promote integrated observation and modelling of the disaster. Furthermore, it discusses the use of scientific research and technology development in supporting decision support systems using early warnings for disaster risk reduction. Section 4 outlines the disaster warning network and its real-life applications which utilises the strengths of geo-information technology, dynamic optimisation, information communication technology, the internet for providing and representing spatial data, and dynamic models for analysing temporal processes that control the disaster. Section 5 describes the geo-information technologies that support and accelerate the search process during all the phases of the disaster. Also, it presents the role of these technologies and other advanced methods during the operational process for creating digital maps for disaster management. Section 6 shows the important part of information communication technology and other supporting tools in accelerating the information flow during the phases of the disaster management cycle. Section 7 outlines the framework for developing a dynamic model of the disaster monitoring network, and it describes the structure of the central database that will be connected to this network. Also, it explains optimisation metaheuristic techniques that will be included in the dynamic model to accelerate the search process for achieving early warring that support the decision support system. Section 8 illustrates some real-life applications based on the use of the disaster warning network, and it insists on the importance of the capacity building in achieving successful use of all the above technologies for risk reduction. This chapter ends with some recommendations, conclusions and future work.

362


2 Disaster Management Disasters are tragic events to development process as they end lives, disrupt social networks, and destroy economic activities. They cut across many organizational, political, geographic, professional, topical and sociological boundaries (Turner and Pidgeon, 1997). Therefore, there is a need to integrate information and knowledge across many disciplines, organizations, and geographical regions. An effective and comprehensive disaster management system must allow access to many different kinds of information at multiple levels at many points of time. It is a continuous process by which all individuals, groups, and communities manage hazards in an effort to avoid or minimize the impact of disasters. Several exact interconnecting steps (depending on the disaster phase) are typically required to generate the type of action that is needed by the disaster management community. Disaster management involves preparing, warning, supporting, and then rebuilding society when disasters occur. More specifically, it requires response, incident mapping, establishment of priorities, and the development and implementation of action plans to protect lives, property, and environment (Cuny, 1983). The following sections present the general framework for the disaster management cycle and the phases that differ according to the type of the disaster.

2.1 The Disaster Management Phases Disaster management activities, generally, can be grouped into six main phases that are related by time and functions to all types of emergencies and disasters. These phases are also related to each other, and each involves different types of skills and data from a variety of sources. The appropriate data has to be gathered, organized, and displayed logically to determine the size and scope of disaster management programs. During an actual disaster, it is critical to have the right data displayed logically, at the right time, to respond and take appropriate action for emergencies (Mileti, 1999). Figure 19.1 depicts the framework for the disaster management cycle which consists of six phases: The Prevention and Mitigation phase includes the activities that are trying to prevent a disaster and minimize the possibility of its occurrence (e.g., legislation that requires building codes in earthquake prone areas, implementing legislation that limits building in earthquake or flood zones, target fire-safe roofing materials in wild land fire hazard areas, etc). These actions are designed to reduce the long-term effects of unavoidable disasters (e.g., land use management, building restrictions in potential flood zones, etc). When potential disaster situations are identified, mitigation actions can be determined and prioritized. For example, in the case of an earthquake, some questions must be examined: What developments are within the primary impact zone of earthquake faults? What facilities are in high hazard areas (main bridges, primary roads, hospitals, hazardous material storage facilities, etc.)? What facilities require reinforced construction or relocation? Based on the expected magnitude of an earthquake, the characteristics of soils, and other geologic data, what damage may occur? (Handmer and Choong, 2006).

19


1) Prevention & Mitigation

2) Preparedness

- Hazard prediction & modelling. - Risk assessment & mapping - Spatial planning - Structural & non-structural measures - Public Awareness & Education

-Scenarios development - Emergency Planning - Training - Food & medical supplies

6) Post Disaster - Lessons learned - Scenario update - Socio-economic & environment impact assessment - Spatial (re)planning

Information Communication Technology ICT is used during most of all the phases of the disaster cycle

363

Disaster

3) Alert -Real time monitoring & forecasting -Early warning -Secure & dependable telecom -Scenario identification -All media alarm

4) Response 5) Rehabilitation & Reconstruction - Recovery and Early damage assessment - Re-establishing life-lines transport & communication infrastructure - Reinforcement

-Dispatching of resources - Emergency telecom - Situational awareness - Command control coordination - Information dissemination - Emergency healthcare

Fig. 19.1 The disaster management cycle

The Preparedness phase includes plans or preparations made and developed by governments, organizations, and individuals to save lives, property, and minimize disaster damage (e.g., mounting training exercises, installing early warning systems, and preparing predetermined emergency response forces). In addition, these activities seek to enhance and help the disaster response and rescue service operations (e.g., providing vital food and medical supplies, mobilizing emergency response personnel, and training exercises). The Alert phase supports all early warning processes such as real time monitoring and forecasting, secure and dependable telecom, scenario identification, and all media alarms. The Response phase is the implementation of action plans that including activities for following the disaster, to provide emergency assistance for causalities and save lives (e.g., search and rescue, emergency shelter, and medical care, and mass feeding) to prevent property damage, preserving the environment (e.g., shutting off contaminated water supply sources), and speeding recovery operations (e.g., damage assessment). The Rehabilitation and Reconstruction phase starts when the disaster is over and includes activities that assist a community to recover and return to normality after a disaster was occurred. These activities are divided into main two sets: short-term recovery activities that restore vital services and return vital life support systems to minimum operating standards (e.g., clean up, ensuring injured people have medical

364


care, providing temporary housing or shelter to citizens who have lost homes in the disaster, and access to food and water), and long-term recovery activities that may continue for several years after a disaster and return life to normal or even improved levels (e.g., community planning, replacement of homes, water systems, streets, hospitals, bridges, schools, etc.) (Wisner et al., 2004). The Post Disaster phase includes analyzing lessons learned, scenario updates, socio-economic and environment impact assessment, and spatial re-planning, etc. These six phases usually overlap, as the information communication technology is being used in all these phases, but the usage is more apparent in some phases than in the others (Ramesh et al., 2007).

2.2 Disaster Management Planning and Hazards Analysis Disaster management programs are developed and implemented through the analysis of information, most of which is spatial, and therefore can be mapped. Once this information is mapped and data is linked to the map, disaster management planning activities can begin. These activities are necessary to analyze and document the possibility of a disaster and the potential consequences or impacts on life, property, and the environment. This includes assessing hazards, risks, mitigation, preparedness, response, and recovery needs. Planning disaster management starts with locating and identifying potential disasters using advanced technology. For example using a GIS, officials can pinpoint hazards (e.g., earthquake faults, fire hazard areas, flood zones, shoreline exposure, etc.) and begin to evaluate the consequences of potential emergencies or disasters. When hazards are viewed with other map data (e.g., streets, pipelines, buildings, residential areas, power lines, storage facilities, etc.), emergency management officials can begin to determine mitigation, preparedness, response, and possible recovery needs. Public safety personnel can focus on determining where mitigation efforts will be necessary, where preparedness efforts must be focused, and where response efforts must be strengthened, and the type of recovery efforts that may be necessary. Before an effective emergency management program can be implemented, thorough analysis and planning must be done. GIS facilitates this planning process by allowing planners to view the appropriate combinations of spatial data through computer-generated maps. This will be explained in more detail in Sect. 5. Once life, property, and environmental values are combined with hazards, emergency management personnel can begin to formulate all the disaster cycle: prevention and mitigation, preparedness, alert, response, rehabilitation and reconstruction, and post disaster plans needs. Important components of these plans are mapping hazardous areas, analyzing potential risks to the communities and the individuals, and estimating possible losses and damages resulting from the disasters. The quality of spatial and attribute data plays an important role in achieving a successful hazard analysis which aims to identify properties and populations within a region that are most at risk from natural disasters. The hazard analysis usually includes five components: hazard identification, profiles of hazard

19


365

events, community profile, estimating losses, and vulnerability analysis. The hazard identification is to identify which types of natural disasters that have the potential of occurring within a region, and in this case, recorded incidences of past natural disasters were used to make this determination. Profiles of hazards events identify past incidences of natural disasters within each region. In this part, the information and data presented in these profiles were obtained through review of historical data from news media sources, and discussions with community residents and officials. The community profile then compares overall county property statistics to those within the pertinent hazard area. In the last stage of the hazard analysis, individual parcels and property asset data were used in the determination of estimated losses and vulnerability analysis. Also, advanced geo-information technology (especially GIS) is an ideal tool to fulfill all the above tasks of hazard analysis as shown in Sect. 5.

3 Real-Time Early Disaster Warning Network Space technologies provide valuable tools for the solution of many real-world problems in fields such as weather forecasting, communication, and disaster management. With satellite communications, people sending or receiving information do not have to be connected to a ground network. With ground-based networks, satellite communications provide access to much of the information over the World Wide Web (Internet). However, there are weak points in operational utilisation of these technologies, such as inadequate coverage of space data, the effects of clouds on optical data, inadequate terrain models, assimilation of data in models etc. An ideal system needs to have sub-systems on vulnerability/risk assessment, early warning and monitoring, emergency communication and short/long term mitigation strategies. Therefore, in recent years, the focus of disaster management community is increasingly moving to the more effective utilization of these technologies, enabling communities at risk to prepare for, and to mitigate the potential damages likely to be caused due to the natural disasters. Using these advanced and hybrid technologies, the main application to be considered as a warning base for all the disasters is the designing of a geomatic network which implements a set of control stations spread over the whole geographic area of the hazardous region. The network provides reliable information on a continuous basis through the parallel process of coverage accuracy prediction (using Least-Squares equations) and integrity risk simulation functions (using Monte Carlo sampling). The major part of the above processes was successfully demonstrated in simulation software considering all standard ranging errors (e.g., satellite clock, ephemeris, multipath, receiver noise, troposphere and ionosphere, etc.) (Saleh, 1996). Then, this network was integrated with a dynamic model based on advanced metaheuristic algorithms (which are based on ideas of Artificial Intelligence) to find the optimal design for this network as shown in the designing geomatic networks of Malta and Seychelles (Saleh and Dare, 2001, 2002b).

366


Geo-Information Technology (GIS, GPS, RS, processing etc)

Central Database

Data processing and analytical centre

User interface (Optimisation, Forecasting, Simulation, Modelling)

Fig. 19.2 The real-time warning network and its database structure

The developed warning network utilizes the strengths of the most advanced geoinformation technologies such as geographic information systems and centralized databases, remote sensing, global navigation satellite systems, dynamic optimisation and geospatial models, data collection, hand-held GPS, internet, networking, information communication technology and service delivery mechanisms, warning dissemination, expert analysis systems, information resources etc. This will have potential to provide valuable support to decision making through providing and representing spatial data, and dynamic models in analysing and representing temporal processes that control the disasters. The combined system of the network and dynamic model will be connected to the central database that combine environmental and geophysical data from earth observation, satellite positioning systems, in-situ sensors and geo-referenced information with advanced computer simulation and graphical visualisation methods as shown Fig. 19.2. Hence, the database will provide the following internet-based services: quickly locating and ensure data availability where and when needed, detailed descriptions of the contents and limitations of the data, and presenting the data in different formats (maps, graphs, pictures, videos, etc.). In addition, the database will be designed to be searchable (by data type, data holder/owner, location, etc), and will be used in three modes: planning and design for protection, real-time emergency, and disaster recovery (Saleh, 2003).

4 Scientific Research and Technology Development for Early Warning Exchange of information and communication practices play key roles in the realization of effective disaster management and risk reduction activities. Therefore, operational use of technology, in terms of information gathering and their real

19


367

time dissemination leading to effective risk reduction at the national and local level, requires appropriate facilities, techniques and institutional systems to be in place. In general form, the disaster information system includes three subsystems: knowledge, interconnectivity, and integration. Knowledge sub-system involves observation techniques for data collection and visualization, analysis, forecasting, modelling and information management. The interconnectivity sub-system relates to the mode of communication employed to retrieve and distribute data and to the dissemination of information products. On the other hand, the integration subsystem addresses the operational system, standards and protocols, procedures for evaluation of quality and reliability and training of key personal. Data availability is crucial for ongoing research, to monitor hazards and to assess risks. Integrating new developments in information management with established and more traditional methods can help to create a better understanding about hazards and their risks. Effective information management and communication are also instrumental for Early Warning (EW) systems and effective mitigation efforts. The main objectives of EW is to be better prepared to face challenges of the risk of short/long term or sudden disasters through these steps: (1) avoiding and reducing damages and loss, (2) saving human lives, health, economic development and cultural heritage, and (3) upgrading quality of life. However, the main EW challenges can be seen when: (1) risks and warnings are not understood, (2) information is scattered, and (3) dissemination is limited. Within this context of EW, the main purposes of scientific research is to overcome these challenges which can be summarised and concluded as so: (1) bridge gaps between science and decision making communities, (2) increased warning time/quicker response, and (3) better understand disasters. Therefore, scientific research, geo-information technology, forecasting, modelling, warning systems are only valuable when they are applied and when they are put into practice for disaster management as shown in Fig. 19.3. Taking this in consideration, the optimal decision support system can be achieved through the following: (1) integrating information, science, research, and technology to improve decision support capabilities, (2) improved observation systems/data/analysis, (3) advanced algorithms, and models, (4) GNSSs, RS, GIS, visualization and display, (5) ICT and networks (EEA, 2001). More practically, an effective early warning system (which must be concentrated on the people at risk) is consisted of four main parts: risk knowledge and assessment; technical monitoring and warning service; dissemination of warnings; and public awareness and preparedness (Egeland, 2006). These main parts must be integrated in one system and failure in any one of them will cause failure of the whole early warning system. To achieve this effectiveness for this early system, significant progress and large improvements have been made in the quality, timeliness and lead time of hazard warnings and decision support system for disaster management and risk reduction. All these advances have been marked and driven by scientific research and technology development particularly through the use of the computer sciences, artificial intelligence, and operational research, etc. on the other hand, there have been continuous improvements in the accuracy and reliability of monitoring instrumentation, and in integrated observation networks particularly

368


Artificial Intelligence

Operational Research

Simulation Optimization

Best Practices

Environmental Disasters

Natural Disasters

Disaster Management & Risk Reduction

Scientific Research & Technology Development Modelling

Applications of new technologies in supporting Early Warning System & Decision Support System for disaster management & risk reduction

Remote Sensing GNSSs Geo - information & Communication Technology GISs

ICT Man-made Disasters

Internet & Intranet

All types of Surveying Methods

Fig. 19.3 The complete system of scientific research and technology development for disaster management and risk reduction

through the use of geo-information and information communication technologies, internet, and other observation methods. As shown in Fig. 19.3, these developments, in turn, have supported research on hazard phenomena, modelling, simulation, monitoring, detecting and forecasting methods and developing hazard warnings for a wide range of all types of natural and man-made disasters. However, capacities in the monitoring and prediction of these disasters vary considerably from one disaster to another and are faced by major challenges and gaps. Some of these challenges and gaps include the availability of these technical capabilities and its integration into the disaster risk reduction decision process within a sustainable procedure; the need for improvement of technical warning capabilities for many hazards. Other gaps and challenges can be seen in insufficient coverage and sustainability of observing systems for monitoring of all type of disasters and hazards, insufficient level of technical capabilities (resources, expertise and operational warning services) in the operational technical agencies that are responsible for

19


369

monitoring and forecasting of severe hazards, difficult access to information from the related teams outside of the affected areas, weak communications for providing timely, accurate and meaningful forecasting and early warning information to all the users. With regards to dissemination, telecommunication mechanisms must be operational, robust, and available every minute of every day, and tailored to the needs of a wide range of threats and users. All of these mechanisms must be based on clear protocols and procedures and supported by an adequate communications infrastructure. However, there are gaps and challenges that affect warning messages due to the underdeveloped dissemination infrastructure and systems, the incomplete coverage of systems, and the resource constraints contributing to the lack of necessary redundancy in services for information. Other gaps and challenges might include insufficient institutional structures to issue warnings due to limited understanding of the true nature of early warning, lack of clarity and completeness in warnings issued due to the lack of common standards for developing warning messages within and across countries, unclear responsibilities about who provides forecasts (of hazards) and who provides warnings (of risks), insufficient understanding of vulnerability due to the lack of better integration of risk assessment and knowledge in the authoritative, official warnings at the national level, ineffective engagement of warning authorities with the media and private sector, (6) the lack of feedback on the system and its performance and learning from previous experience. The characteristics of risk can usually be presented through scenario plans, practical exercises, risk mapping, and qualitative measures, etc. To improve the basis for collecting and analysing risk data, risk assessment requires standard indicators to measure the success and failure of early warning systems. Therefore, the development of effective warning messages must depend on relatively good data resources and the generation of accurate risk scenarios showing the potential impacts of hazards on vulnerable parts of hazardous area. In this direction, more research is needed to make qualitative data and narratives of vulnerability accessible and useable to engineers, planners, policy makers and all the other parties working in this domain. This will support the rescue teams with capabilities to analyse not only the hazards, but also the vulnerabilities to the hazards and the consequents of the risk, and thus will help them decide whether and when to warn. In addition to gathering statistics and mapping populations’ risk factors, risk assessments should involve the community to ascertain their perceived risks and concerns. To ensure the optimal decision support system for natural and environmental disaster management and risk reduction using early warning capabilities, capacity building has to be highly considered on all the aspects as follows: Academic programme and technical workshops: training of scientist and engineers during installation phase. Institutional capacity building: consulting of organizational structures and inter-institutional communication, planning and construction of new infrastructures, establishment of communication platforms and chains. Warning culture: establishment of warning mechanisms products (e.g., risk maps, evacuation plans, etc.) and information products for end users, (e.g. development of teaching units in schools, universities, and the community).

370


5 Geo-Information Technology Geo-information technologies provide real-time information that allows agencies working on disaster management and risk reduction to effectively manage the situation and to plan community evacuation and relief operations in case of emergencies. These technologies can help considerably to show vulnerable areas, enhance mapping, and ameliorate the understanding of hazards (Oosterom et al., 2005). The following sub-sections present these advanced technologies and their roles in disaster risk reduction.

5.1 Global Navigation Satellite Systems (GNSSs) It is well known that throughout the world the use of the GNSSs is dramatically increasing, demanding the optimisation of the accuracy of the measurements provided by these systems. The Global Positioning System (GPS), the GLObal NAvigation Satellite System (GLONASS), and the European Satellite Navigation (Galileo) are the most widely known satellite systems as shown in Fig. 19.4. GNSSs Satellites provide the user with a 24-h highly accurate three-dimensional position, velocity and timing system at almost any global location. However, these systems suffer from several errors that affect the accuracy of the observation and Fig. 19.5 depicts the sources of these errors. Large part of these errors can be theoretically and practically minimized and eliminated and using differential and wide area augmentation systems and other surveying methods (Elliott, 1996) and (Leick, 1995).

5.2 Remote Sensing (RS) RS satellites are used to monitor the land, the surface, the oceans and the atmosphere, and how their situations they change over time. Most RS satellites cover the

Fig. 19.4 The GPS and GLONASS constellation of navigation satellite systems

19


Fig. 19.5 Sources of errors in GNSSs

371

Satellite Clocks Ephemeris Selective Availability Atmospheric Delays Multipath Delays Receiver Clocks

whole globe, making them important for the study of large-scale phenomena such as climate changes and desertification as shown in Fig. 19.6. For example, remotely sensed imagery helps to identify the most fire-prone areas and to develop fire propagation models which allow emergency evacuation to be modeled at the level of the individual vehicle for avoiding congestion during evacuation. In addition, RS has application in the characterisation of earthquake risk through the identification of regions prone to liquefaction (river valleys and coastal areas). Seismic vulnerability from tsunamis is easily assessed by convolving digital elevations and bathymetry data with the distribution of coastal populations and economic infrastructures. However, the main limitations of RS satellite images are cloud cover and resolution.

Fig. 19.6 The remote sensing technology

372


Some of these problems may be circumvented using GNSSs satellites. Combining remotely sensed imagery with ground data reduces the cost of ground-based sampling efforts by more than 50% while substantially increasing the accuracy of collected data (Brown and Fingas, 2001).

5.3 Geographic Information Systems (GISs) Innovations in GIS technology are increasingly accepted tools for the presentation of hazard vulnerabilities and risks. The data obtained from GNSSs and RS will be used by GISs technology to produce maps that identify and analyze all applicable types of natural hazards. These maps then can be used by local governments to inform citizens within their communities of the potential risks from these hazards. GISs facilitate the integration of data obtained from various sources (e.g., topographic hardcopy maps, tables, aerial photos, satellite images, satellite navigation systems, etc). Then, this data will be analysed and processed to produce “Smart Maps” that link database to map and for every feature on this map, there is a row in a table. Figure 19.7 depicts the GIS operational cycle to process geographic information and create digital maps through these steps: data acquisition, data processing, and data dissemination. By utilizing a GIS, all related parties can share information through databases on computer-generated maps in one location. GISs provide a mechanism to centralize and visually display critical information during a disaster (Masser and Montoya, 2002). 5.3.1 The Role of GIS During the Disaster Management Phases GIS plays an important part during all the disaster management phases as explained previously in Sect. 2. The Planning phase for disaster involves predicting the area and time of a possible disaster and the impacts on human life, property, and

Sources of Geographic Information

Data Processing & Modelling

Visualising Geographic Information

Visualisation “worth a thousand words” Database “not easy to interpret”

Fig. 19.7 The operational cycle for geo-information technologies to create digital maps

19


373

environment. These factors are used to determine an effective planning procedure for the mitigation of possible disaster effects. This planning can be done effectively and quickly using the application of GIS, which is a very good tool for short and long term planning. GIS modeling allows disaster managers to view the scope and dimension of a disaster and its impacts. GIS allows disaster managers to quickly access and visually display critical information by location. This information facilitates the development of action plans that may be transmitted to disaster response personnel for the coordination and implementation of emergency initiatives. During the mitigation and prevention phase (and by utilizing existing databases linked to geographic features), GIS can be used for managing large volumes of data needed for the hazard and risk assessment as values at risk can be displayed quickly and efficiently through a GIS. For example, in case of a wildfire disaster, a GIS can identify specific slope categories in combination with certain species of flammable vegetation near homes that could be threatened by wildfire. GISs can answer these questions: Where are the fire hazard zones? What combination of features (for example, topography, vegetation, weather) constitutes a fire hazard? With regards to the other disasters such as earthquakes and floods, a GIS can identify certain soil types in and adjacent to earthquake impact zones where bridges or overpasses are at risk. GISs can identify the likely path of a flood based on topographic features or the spread of a coastal oil spill based on currents and wind. Most importantly, human life and other values (property, habitat, wildlife, etc.) at risk can be quickly identified and targeted for protective action. During the preparedness phase, GISs can be use as a tool for planning of evacuation routes, for the design of centers for emergency operations, and for integration of satellite data with other relevant data in the design of disaster warning system. They can provide answers to questions to those activities that prepare for actual emergencies: Where should fire stations be located if a short response time is expected? How many paramedic units are required and where should they be located? What evacuation routes should be selected? How will people be notified? Will the road networks handle the traffic? What facilities will provide evacuation shelters? What quantity of supplies will be required at each shelter? For Early warning purposes, GISs can display real-time monitoring for early emergency warning. Remote weather stations can provide current weather indexes based on location and surrounding areas. Wind direction, temperature, and relative humidity can be displayed by the reporting weather station. Wind information is vital in predicting the movement of a chemical cloud release or anticipating the direction of wildfire spread upon early report. Earth movements (earthquake), reservoir level at dam sights, and radiation monitors can all be monitored and displayed by location. It is now possible to deliver this type of information and geographic display over the Internet for public information or the Intranet for organizational information delivery. During the response phase, the closest (quickest) response units based at fixed and known locations can be selected, routed, and dispatched to a disaster. Depending on the kind of the disaster, GISs can provide detailed information before the first units arrive. For example, during a fire in housing area and while the rescue team

374


in the route to the emergency, it is possible to identify the closest hydrants, electrical panels, hazardous materials, and floor plan of the building. For hazardous spills or chemical cloud release, the direction and speed of movement can be modeled to determine evacuation zones and containment needs. Advanced vehicle locating can be built-in to track (in real-time) the location of incoming emergency units and then to assist in determining the closest mobile units (which are located on the map through GNSS transponders) to be dispatched to a disaster. During multiple disasters (numerous wildfires, mud slides, earthquake damage) in different locations, GISs can display the current emergency unit locations and assigned responsibilities to maintain overall situation status. In general, the response phase is divided into two phases: a short-term phase and a long-term phase. One of the most difficult tasks in the short-term recovery phase is damage assessment, but a GIS integrated with GNSSs can play important roles such locating each damaged facility, identifying the type and amount of damage, displaying the number of shelters needed and where they should be located for reasonable access, and displaying areas where services have been restored in order to quickly reallocate recovery work to priority tasks. In this phase, laptop computers can update the primary database from remote locations through a variety of methods. GISs can display (through the primary database) overall current damage assessment as it is conducted. Emergency distribution centers’ supplies (medical, food, water, clothing, etc.) can be assigned in appropriate amounts to shelters based on the amount and type of damage in each area. Action plans with maps can be printed, outlining work for each specific area. Shelters can update inventory databases allowing the primary command center to consolidate supply orders for all shelters. The immediate recovery efforts can be visually displayed and quickly updated until short term recovery is complete. This visual status map can be accessed and viewed from remote locations. This is particularly helpful for large emergencies or disasters where work is ongoing in different locations. During the long-term recovery phase, prioritization plans and progress for major restoration investments can be made and tracked utilizing GIS. In addition, response requirements, protection needs (e.g., supportive bridge in the event of floods, removing vegetation in the case of wildfire, etc) can be determined for areas at high risk. Long term recovery costs can be highly expensive for large disasters and accounting for how and where funds are allocated is demanding. In this part and after allocating the funds for repairs, accounting information can be recorded and linked to each location, then GISs can be implemented to ease the burden of accounting for these costs. In the disaster relief phase, GISs are extremely useful in combination with GNSSs in search and rescue operations in areas that have been devastated and where it is difficult to orientate. In disaster rehabilitation phase, GISs are used to organise the damage information and the post disaster census information and in the evaluation of sites for reconstruction. 5.3.2 HAZUS (Hazards U.S.) While GISs are used to capture, analyse, and display spatial data, the models provide the tools for complex and dynamic analysis. Input for spatially distributes models, as well as their output, can be treated as map overlays (Fedra, 1994). The familiar

19


375

format of maps supports the understanding of model results, but provides also a convenient interface to spatially referenced data. Expert systems, simulation and optimisation models add the possibility for complex, and dynamic analysis to the GIS. Recently, a new program based on GIS was presented called HAZUS (Hazards U.S.), which is open source, free to use, and highly responsive to end-user requirements. Users incorporate data and modelling the physical world of infrastructure, build inventory, geology, damage estimation formulas, and critical operating centre locations, and then subject HAZUS to the complex consequences of a hazard event as shown in Fig. 19.8. After that, users can implement HAZUS to prepare for disasters (pre-event), respond to the threat (during the event), analyze and estimate the potential loss of life, injuries, property damage, forecasts casualties, and to manage the critical situation (post-event). One major challenge in building effective information systems for disaster management (e.g., fault movement, river basin) is the integration of dynamic models with the capabilities of GIS technology. This can provide a common framework of reference for various tools and models addressing a range of problems in river basin management, supply distributed data to the models, and assist in the visualisation of spatial model results in the form of topical maps (Fedra, 1995). The possibility of applying HAZUS program to investigate some critical situations in Syria and neighbouring countries (e.g., the West Shaam fault as shown in Fig. 19.9) were planned for fault extends for about 1,100 km along the western part of the Shamm countries (Syria, Lebanon, Jordan, Palestine) representing the north-western Arabian African plates boundary.

Fig. 19.8 HAZUS in estimating the peak ground acceleration and source in the earthquake scenario

376


Fig. 19.9 The West Shaam fault

6 Information Communication Technology (ICT) for Disaster Management ICT is used in almost all phases of the disaster management process and can effectively be used to minimize the impacts of disasters in many ways. ICT plays a critical role in facilitating the reconstruction process and in coordinating the return of those displaced by disasters to their original homes and communities. Disaster management activities, in the immediate aftermath of a disaster, can be made more effective by the use of appropriate ICT tools. These include tools for resource management and tracking, communication under emergency situations (e.g. use of Internet communications), and collecting essential items for the victims. GISs and RS are examples of ICT tools being widely used in almost all the phases of disaster management activities. RS for early warning is made possible by various available technologies, including telecommunication satellites, radar, telemetry and meteorology. More clearly, the rule of used ICT is to accelerate the flow of information during all the stages of the disaster between the emergency and rescue teams in disaster location and the main authorities (decision makers) in central control room. Any one or a combination of the following ICT and media tools that are shown in Fig. 19.10, can be used in disaster management: radio and television, telephone

19


GPRS Station

GSM Digital Camera with GPS

Rainfall

377

Data

Dissemination to the public River stage

Internet Synoptic charts

Telemetry/ Data box/ Voice

Radio

wireless communication

Weather forecast

Satellite images

Boundary estimation Rainfall, Water level

TCP/IP for GPRS

Modem Fax Modem

Television

Fax

Bulletin

manual entry

via modem

Radio Tower

Telephone

Dissemination to various agencies Data Entry & Processing

Modelling & Mapping

GIS data Satellite dish

Fig. 19.10 The ICT system used for flood monitoring network

(fixed and mobile), short message service (SMS), cell broadcasting, satellite radio, internet/email, amateur radio and community radio (Wattegama, 2007).

7 The Dynamic Metaheuristic Model Within the concept of dynamic optimisation, these disasters can be regarded as non-differentiable and real-time Multi-objective Optimisation Problems (MOPs). These problems involve multiple, conflicting objectives in a highly complex search domain. Moreover, the volume of data collected for these problems is growing rapidly and sophisticated means to optimise this volume in a consistent and economical procedure are essential. Therefore, robotic algorithms are required to deal simultaneously with several types of processes which are concerned with the unpredictable environment of these problems (Deb, 2001). These algorithms can provide a degree of functionality for spatial representation and flexibility suitable for quickly creating real-time optimal solutions that account for the uncertainty present in the changing environment of these problems which can be formulated in a design model for the monitoring network as follow in Eq. (1): NetworkMOP = optimize : f (x) = { f1 (x), f2 (x), . . . , f2 (x)} subject to x = (x1 , x2 , . . . , xn ) ∈ X

(1)

378


where fi (x) is the model of the network that consists of ith monitoring objective functions to be optimised, x is a set of variables (i.e., decision parameters) and X is the search domain. The term “optimise” means finding the ideal network in which each objective function corresponds to the best possible value by considering the partial fulfilment of each of the objects. More specifically, this network is optimal in a way such that no other networks in the search domain are superior to it when all objectives are considered. The main innovative aspect of the developed network is the integration of the state of the art geographical and environmental data collection, and data management tools with simulation and decision tools for disaster management and risk reduction. Then, this network was integrated with the artificial intelligence optimisation algorithms to find the optimal network design. This will allow the modeller to develop a precise and unambiguous specification that can strongly help in estimating the impacts of an actual development process of the presented design. Therefore, it is almost impossible even for an experienced and higher-level designer to find an optimal design by the currently used methods which do not provide spatial representation to the whole situation and lack the ability to select “interesting” contingencies for which to optimise. Once such designs are obtained, the technical user will be able to select an acceptable design by trading off the competing objectives against each other and with further considerations. The final design of the network should be robust (i.e., performs well over a wide range of environment conditions), sustainable (i.e., not only optimal under current condition, but also considering predicted changes), and flexible (i.e., allows easy adaptation after the environment has changed) (Peng et al., 2002). Initial Network Formation

Neighbourhood (set of alternative networks derived from the initial one) Search by Move Formation (Local Search)

Provisional Neighbourhood Formation

Search by Network Formation (development and guided search techniques) New Neighbourhood (solution formation)

Acceptance Criteria

Fig. 19.11 The general framework for metaheuristic algorithms

Termination of the Search

19


379

Metaheuristic techniques (which are based on the ideas of artificial intelligence) potentially have these capabilities to produce set of high quality real-time designs that can model more closely and easily many functions and visualize the trade-offs between them and then to filter and cluster top optimal solution (Osman and Kelly, 1992). These techniques are iterative procedures that combine different operational and organizational strategies based on robustness and computerized models in order to obtain high-quality solutions to complex optimization problems. They can provide instantaneous comparisons of the achieved results of different developed designs using several procedures such as convergence, diversity, and complexity analysis, etc. Figure 19.11 depicts the general framework of the metaheuristics algorithms that has been adopted in this research. The dotted lines indicted option that can be skipped or used. In this research, several metaheuristics are proposed and implemented for optimising the scheduling activities of designing the monitoring network. The well-known metaheuristics that have been successfully applied to optimise real-life applications based on monitoring network are: simulated annealing, tabu search, ant colony optimization, and genetic algorithm (Saleh and Dare, 2002a). These metaheuristics are inspired, respectively, by the physical annealing process, the proper use of memory structures, the observation of real ant colonies and the Darwinian evolutionary process.

7.1 Simulated Annealing (SA) The SA technique is flexible, robust and capable of producing the best solution to complex real life problems (Kirkpatrick et al., 1983) and (Rene Vidal, 1993). This technique derives from physical science and is based on a randomisation mechanism in creating solutions and accepting the best one. The annealing parameters that have to be specified are; the initial temperature, the temperature update function, the length of the Markov chain and the stopping criterion. The initial temperature simulates the effect of temperature in the search process to find the best candidate of the final design. The temperature update function determines the behaviour of the cooling process, while the length of the Markov chain represents the number of iterations between the successive decreases of temperature. The optimization process is terminated at a temperature low enough to ensure that no further improvement can be expected. With a suitable annealing parameters, an optimal network design or close to it can be achieved for optimization the flooding problem (Saleh and Allaert, 2008). The basic steps for the SA, which returns a better network, are depicted in Fig. 19.12.

7.2 Tabu Search (TS) The TS technique, which is a global iterative optimisation, exploits knowledge of the system or “memory” under investigation to find better ways to save computational efforts without effecting solution quality (Glover and Laguna, 1997). The

380


Initialisation: Step 1 Select an initial Network design candidate NINT with value V(NINT) Step 2 Initialise the annealing parameters: • Set the initial temperature Ti. • Set the Markov temperature length L. • Set the cooling factor F (Fθ} where θ is a uniform random number 0

Remote Sensing of the Changing Oceans

Space Remote Sensing of Subtropical Oceans

Remote Sensing of Coastal Environments (Remote Sensing Applications Series)

Remote Sensing: Applications

Optical Remote Sensing

Laser Remote Sensing

Atmospheric ultraviolet remote sensing

Introduction to Remote Sensing

Atmospheric Ultraviolet Remote Sensing

Geoscience and Remote Sensing

The Remote Sensing Data Book

The Remote Sensing Data Book

Remote Sensing Image Processing

Remote Sensing of the European Seas

Spatial Statistics for Remote Sensing

25.Electromagnetic Subsurface Remote Sensing

Spatial statistics for remote sensing

Remote Sensing with Polarimetric Radar

Field Methods in Remote Sensing

Remote Sensing with Polarimetric Radar

Remote Sensing With Imaging Radar

Polarisation: Applications in Remote Sensing

Satellite remote sensing for archaeology

Remote Sensing and Image Interpretation

Signal Processing for Remote Sensing

Earth Science Satellite Remote Sensing

Satellite Remote Sensing for Archaeology

Field Methods in Remote Sensing

Remote Sensing in Soil Science

Remote Sensing of Urban and Suburban Areas

Remote sensing of snow and ice

Remote Sensing of the Changing Oceans

Space Remote Sensing of Subtropical Oceans

Remote Sensing of Coastal Environments (Remote Sensing Applications Series)

Remote Sensing: Applications

Optical Remote Sensing

Laser Remote Sensing

Atmospheric ultraviolet remote sensing

Introduction to Remote Sensing

Atmospheric Ultraviolet Remote Sensing

Geoscience and Remote Sensing

The Remote Sensing Data Book

The Remote Sensing Data Book

Remote Sensing Image Processing

Remote Sensing of the European Seas

Spatial Statistics for Remote Sensing

25.Electromagnetic Subsurface Remote Sensing

Spatial statistics for remote sensing

Remote Sensing with Polarimetric Radar

Field Methods in Remote Sensing

Remote Sensing with Polarimetric Radar

Remote Sensing With Imaging Radar

Polarisation: Applications in Remote Sensing

Satellite remote sensing for archaeology

Remote Sensing and Image Interpretation

Signal Processing for Remote Sensing

Earth Science Satellite Remote Sensing

Satellite Remote Sensing for Archaeology

Field Methods in Remote Sensing

Remote Sensing in Soil Science

Remote Sensing of Urban and Suburban Areas

Remote sensing of snow and ice

Recommend Documents