3D Seismic Technology: Application to the Exploration of Sedimentary Basins
Geological Society Memoirs
Society Book Editors R . J. P A N K H U R S T ( C H I E F E D I T O R )
P. DOYLE F. J. GREGORY J. S. G R I F F I T H S A . J. H A R T L E Y
R. E. HOLDSWORTH J. A. HOWE P. T . L E A T
A. C. MORTON N. S. ROBINS J. P. T U R N E R
Society books reviewing procedures The Society makes every effort to ensure that the scientific and production quality of its books matches that of its journals. Since 1997, all book proposals have been refereed by specialist reviewers as well as by the Society's Books Editorial Committee. If the referees identify weaknesses in the proposal, these must be addressed before the proposal is accepted. Once the book is accepted, the Society has a team of Book Editors (listed above) who ensure that the volume editors follow strict guidelines on refereeing and quality control. We insist that individual papers can only be accepted after satisfactory review by two independent referees. The questions on the review forms are similar to those for Journal of the Geological Society. The referees' forms and comments must be available to the Society's Book Editors on request. Although many of the books result from meetings, the editors are expected to commission papers that were not presented at the meeting to ensure that the book provides a balanced coverage of the subject. Being accepted for presentation at the meeting does not guarantee inclusion in the book. Geological Society Publications are included in the ISI Index of Scientific Book Contents, but they do not have an impact factor, the latter being applicable only to journals. More information about submitting a proposal and producing a Society Publication can be found on the Society's web site: www.geolsoc.org.uk.
It is recommended that reference to all or part of this book should be made in one of the following ways: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 2004.3D Seismic Technology:
Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29. JONES, G., WILLIAMS, L. S. & KNIPE, R. J. 2004. Structural evolution of a complex 3D fault array in the Cretaceous and Tertiary of the Porcupine Basin, offshore Ireland. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 2004. 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 117-132.
G E O L O G I C A L SOCIETY M E M O I R NO. 29
3D Seismic Technology: Application to the Exploration of Sedimentary Basins EDITED BY
RICHARD J. DAVIES Cardiff University, UK
JOSEPH A. CARTWRIGHT Cardiff University, UK
SIMON A. STEWART BP, Azerbaijan
MARK LAPPIN ExxonMobil Exploration Company, USA and
JOHN R. UNDERHILL The University of Edinburgh, UK
2004 Published by The Geological Society London
THE GEOLOGICAL SOCIETY The Geological Society of London (GSL) was founded in 1807. It is the oldest national geological society in the world and the largest in Europe. It was incorporated under Royal Charter in 1825 and is Registered Charity 210161. The Society is the UK national learned and professional society for geology with a worldwide Fellowship (FGS) of 9000. The Society has the power to confer Chartered status on suitably qualified Fellows, and about 2000 of the Fellowship carry the title (CGeol). Chartered Geologists may also obtain the equivalent European title, European Geologist (EurGeol). One fifth of the Society's fellowship resides outside the UK. To find out more about the Society, log on to www.geolsoc.org.uk. The Geological Society Publishing House (Bath, UK) produces the Society's international journals and books, and acts as European distributor for selected publications of the American Association of Petroleum Geologists (AAPG), the American Geological Institute (AGI), the Indonesian Petroleum Association (IPA), the Geological Society of America (GSA), the Society for Sedimentary Geology (SEPM) and the Geologists' Association (GA). Joint marketing agreements ensure that GSL Fellows may purchase these societies' publications at a discount. The Society's online bookshop (accessible from www.geolsoc.org.uk ) offers secure book purchasing with your credit or debit card. To find out about joining the Society and benefiting from substantial discounts on publications of GSL and other societies worldwide, consult www.geolsoc.org.uk, or contact the Fellowship Department at: The Geological Society, Burlington House, Piccadilly, London W1J 0BG: Tel.+44 (0)20 7434 9944; Fax+44 (0)20 7439 8975; E-mail:
[email protected]. For information about the Society's meetings, consult Events on www.geolsoc.org.uk. To find out more about the Society's Corporate Affiliates Scheme, write to
[email protected].
Published by The Geological Society from: The Geological Society Publishing House Unit 7, Brassmill Enterprise Centre Brassmill Lane Bath BA1 3JN, UK Tel. +44 (0)1225 445046 Fax +44 (0)1225 442836) Online bookshop: http://bookshop.geolsoc.org.uk (Orders:
The publishers make no representation, express or implied, with regard to the accuracy of the information contained in this book and cannot accept any legal responsibility for any errors or omissions that may be made. 9 The Geological Society of London 2004. All rights reserved. No reproduction, copy or transmission of this publication may be made without written permission. No paragraph of this publication may be reproduced, copied or transmitted save with the provisions of the Copyright Licensing Agency, 90 Tottenham Court Road, London W 1P 9HE. Users registered with the Copyright Clearance Center, 27 Congress Street, Salem, MA 01970, USA: the item-fee code for this publication is 0435-4052/04/15.00.
British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. ISBN 1-86239-151-3 Typeset and Printed by The Alden Group, Oxford, UK.
Distributors USA AAPG Bookstore PO Box 979 Tulsa OK 74101-0979 USA Orders: Tel. +1 918 584-2555 Fax +1 918 560-2652 E-mail bookstore @aapg.org India Affiliated East-West Press PVT Ltd G- 1/16 Ansari Road, Daryaganj, New Delhi 110 002 India Orders: Tel. +91 11 2327-9113 Fax +91 11 2326-0538 E-mail
[email protected] Japan Kanda Book Trading Company Cityhouse Tama 204 Tsurumaki 1-3-10 Tama-shi, Tokyo 206-0034 Japan Orders: Tel. +81 (0)423 57-7650 Fax +81 (0)423 57-7651 Email geokanda @ma.kcom.ne.jp
Preface This Geological Society Memoir is the result of a highly successful conference held in November 2001 at The Geological Society, London. The 32 papers in this Memoir attempt to capture how the rapid development of 3D seismic technology has had a fundamental role in the exploration, development and production of hydrocarbons and uses movies as well as conventional figures and text to do so. The Memoir also shows that 3D seismic data are a tremendous - - but mainly underutilized - - tool for Earth
scientists involved in the exploration of sedimentary basins in the context of a diverse range of disciplines whether they be sedimentary, structural or even 'hard rock'. Given the breadth and depth of the contributions we hope that this book will become a well-thumbed reference for those working with 3D seismic technology in industry and academia, but perhaps more importantly will act as an introduction for those that are now discovering its utility.
Acknowledgements First and foremost we would to thank all the contributors to the meeting who made it such a success and then detailed their results in this book. Michele O'Callahan is thanked for helping to convene the meeting and Helen Wilson for organizing the event. ExxonMobil, Landmark, PGS Exploration Ltd, Schlumberger, Troy Ikoda, and Veritas DGC Ltd. generously sponsored the conference. This Memoir would not have been possible without the expertise of the following individuals who reviewed one or more papers. Rolf Ackerman, John Allison, John Ardill, Bryn Austin, Brian Bell, John Bingham, Stuart Bland, Ian Cloke, Pat Connelly, Rupert Dalwood, Chris Dart, Bret Dixon, Richard Dixon, Tony Dor6, Chris Elders, Duncan Erratt, Jean Christophe Faug~res, A1 Fraser, Scot Fraser, Joe Gallagher, Kerry Gallagher, Tim Garfield, Rutger Gras, Matt Grove, Jens
Peter Vind Hansen, Dan Helgeson, Peter Homonko, Howard Johnson, Hugh Kerr, Paul Knutz, Nick Kusznir, Charles Line, Lidia Lonergan, Dave Long, Richard Lovell, Steve Mathews, James Maynard, Ken McClay, Alan McInally, Steve Mitchell, Damian O'Grady, Mike Payne, Sverre Planke, Henry Posamentier, Pat Shannon, John Smallwood, Roland Smith, Gary Steffens, Martyn Stoker, Dorrik Stow and Alistair Welbon. April Newman is thanked for providing logistical and administrative assistance throughout the editorial process. ExxonMobil also provided logistic support and kindly helped fund colour pages. We are also very grateful to the Petroleum Group of The Geological Society who championed the meeting, encouraged the compilation of these papers and also generously sponsored colour pages.
Contents Preface
v
Acknowledgements
vi
3D seismic technology: are we realising its full potential?: DAVIES, R. J., STEWART, S. A., CARTWRIGHT,J. A., LAPPIN, M., JOHNSTON, R., FRASER, S. I. & BROWN, A. R.
1
Depositional systems Seismic geomorphology: imaging elements of depositional systems from shelf to deep basin using 3D seismic data: implications for exploration and development: POSAMENTIER, H. W.
11
Depositional architectures of Recent deepwater deposits in the Kutei Basin, East Kalimantan: FOWLER, J. N, GURITNO, E., SHERWOOD, P., SMITH, M. J., ALGAR, S., BUSONO, I., GOFFEY, G. & STRONG, A.
25
The use of near-seafloor 3D seismic data in deepwater exploration and production: STEFFENS, G. S., SHIPP, R.C., PRATHER, B. E., NOTT, J. A., GIBSON, J. L. & WINKER, C. D.
35
Structural controls on the positioning of submarine channels on the lower slopes of the Niger Delta: MORGAN, R.
45
Sea bed morphology of the Faroe-Shetland Channel derived from 3D seismic datasets: LONG, D., BULAT, J. & STOKER,M.S.
53
3D anatomy of late Neogene contourite drifts and associated mass flows in the Faroe-Shetland Basin: KNUTZ, P. C. & CARTWRIGHT, J. A.
63
Interactions between topography and channel development from 3D seismic analysis: an example from the Tertiary of the Flett Ridge, Faroe-Shetland Basin, UK: ROBINSON, A. M., CARTWRIGHT, J. A., BURGESS, P. M. & DAVIES, R. J.
73
3D seismic analysis reveals the origin of ambiguous erosional features at a major sequence boundary in the eastern North Sea: near top Oligocene: HANSEN, J. P. V., CLAUSEN, O. R. & HUUSE, M.
83
3D seismic interpretation of the Messinian Unconformity in the Valencia Basin, Spain: FREY MARTINEZ,J., CARTWRIGHT,J.A., BURGESS, P. M. & VICENTE BRAVO, J.
91
Structural and igneous geology 3D analogue models of rift systems: templates for 3D seismic interpretation: MCCLAY, K. R., DOOLEY, T., WHITEHOUSE, P., FULLARTON, L. & CHANTRAPRASERT,S.
101
Structural evolution of a complex 3D fault array in the Cretaceous and Tertiary of the Porcupine Basin, offshore Ireland: JONES, G., WILLIAMS, L. S. & KNIPE, R. J.
117
Three-dimensional geometry and displacement configuration of a fault array from a raft system, Lower Congo Basin, Offshore Angola: implications for the Neogene turbidite play: DUTTON, D. M., LISTER, D., TRUDGILL, B. D. & PEDRO, K.
133
Initial deformation in a subduction thrust system: polygonal normal faulting in the incoming sedimentary sequence of the Nankai subduction zone, southwestern Japan: HEFFERNAN,A. S., MOORE, J. C., BANGS, N. L., MOORE, G. F. & SHIPLEY,T. H.
143
The evolution and growth of Central Graben salt structures, Salt Dome Province, Danish North Sea: RANK-FRIEND, M. & ELDERS, C. F.
149
Integrating 3D seismic data with structural restorations to elucidate the evolution of a stepped counter-regional salt system, Eastern Louisiana shelf, Northern Gulf of Mexico: TRUDGILL, B. D. & ROWAN, M. G.
165
Exploration 3D seismic over the Gjallar Ridge, Mid-Norway: visualization of structures on the Norwegian volcanic margin from Moho to seafloor: CORFIELD, S. M., WHEELER, W., KARPUZ, R., WILSON, M. & HELLAND, R.
177
Tertiary inversion in the Faroe-Shetland Channel and the development of major erosional scarps: SMALLWOOD,J.R.
187
3D seismic analysis of the geometry of igneous sills and sill junction relationships: HANSEN, D. M., CARTWRIGHT,J. A. & THOMAS, D.
199
Kinematic indicators for shallow level igneous intrusion from 3D seismic data: evidence of flow direction and feeder location: TRUDE, K. J.
209
Application at development and production scale Visualization and interpretation of 3D seismic in the Carboniferous of the UK Southern North Sea: LYNCH, J.J.
219
Direct visualization and extraction of stratigraphic targets in complex structural settings: JAMES, H., BOND, R. & EASTWOOD,L.
227
Locating exploration and appraisal wells using predictive rock physics, seismic inversion and advanced body tracking: an example from North Africa: PICKERING, G., KNIGHT, E., BLETCHER, J., BARKER, R. & KEMPER, M.
235
Use of 3D visualization techniques to unravel complex fault patterns for production planning: Njord field, Halten Terrace, Norway: DART, C., CLOKE, I., HERDLEV/ER, ,~., GILLARD, D., RIVEN/ES, J. C., OTTERLEI, C., JOHNSEN, E. & EKERN, A.
249
Seismic characteristics of large-scale sandstone intrusions in the Paleogene of the South Viking Graben, UK and Norwegian North Sea: HUUSE M., DURANTI, D., STEINSLAND,N., GUARGENA, C. G., PRAT, P., HOLM, K., CARTWRIGHT,J. A. & HURST, A.
263
vm
CONTENTS
Integrated use of 3D seismic in field development, engineering and drilling: examples from the shallow section: AUSTJN, B.
279
4D/time-lapse seismic: examples from the Foinaven, Schiehallion and Loyal Fields, UKCS, West of Shetland: BAGLEY, G., SAXBY, I., MCGARRITY, J., PEARSE, C. & SEATER, C.
297
New applications Improved drilling performance through integration of seismic, geological and drilling data: STEWART,S. A. & HOLT, J.
303
4D seismic imaging of an injected CO2 plume at the Sleipner Field, central North Sea: CHADWICK,R. A., ARTS, R., EIKEN, O., KIRBY, G. A., LINDEBERG,E. & ZWEIGEL, P.
311
Towards an automated strategy for modelling extensional basins and margins in four dimensions: WroTE, N., HAINES, J., JONES, S. & HANNE,D.
321
Examples of multi-attribute, neural network-based seismic object detection: DE GROOT, P., LIGTENBERC, H., OLDENZIEC,T., CONNOLLY, D. & MELDAHL, P.
333
Modelling fault geometry and displacement for very large networks: LISTER, D.L.
339
Index
349
3D seismic technology: are we realising its full potential? RICHARD LAPPIN
J. D A V I E S 3, R O D N E Y
1, S I M O N JOHNSTON
A. STEWART 4, S C O T
2, J O S E P H I. F R A S E R
A. CARTWRIGHT 5 & ALISTAIR
1, M A R K
R. BROWN
6
13DLab, School of Earth, Ocean and Planetary Sciences, Cardiff University, Main Building Park Place, Cardiff CFIO 3YE, UK (e-mail:
[email protected]) 2BP Azerbaijan, C/o Chertsey Road, Sunbuo, on Thames, Middlesex TWI6 7LN, UK 3ExxonMobil Exploration Company, 233 Benmar, Houston, Texas 77060, USA 4Bp, E & P Technology Group, Chertsey Road, Sunbuo' on Thames, Middlesex TW16 7LN, UK 5Shell EP Technology Solutions, Shell International Exploration & Production Inc., 200 N Dairy Ashford, Houston, Texas 77079, USA 6Consulting Reservoir Geophysicist, 1911 Country Brook Lane, Allen, Texas 75002, USA
Abstract: Three-dimensional (3D) seismic data have had a substantial impact on the successful exploration and production of hydrocarbons. Although most commonly acquired by the oil and gas exploration industry, these data are starting to be used as a research tool in other Earth sciences disciplines. However despite some innovative new directions of academic investigation, most of the examples of how 3D seismic data have increased our understanding of the structure and stratigraphy of sedimentary basins come from the industry that acquired these data. The 3D seismic tool is also making significant inroads into other areas of Earth sciences, such as igneous and structural geology. However, there are pitfalls that parallel these advances: geoscientists need to be multidisciplined and true integrators, and at the same time have an ever-increasing knowledge of geophysical acquisition and processing. Notably the utility of the 3D seismic tool seems to have been overlooked by most of the academic community, and we would submit that academia has yet to take full advantage of this technology as a research tool. We propose that the remaining scientific potential far exceeds the advances made thus far and major opportunities, as well as challenges, lie ahead.
The age of field-based geological mapping that began with William Smith (1769-1839) started as a result of technological advances such as mining and canal building, which in turn were fuelled by basic commercial needs (Winchester 2001). In a similar way, a 'new age' of subsurface geological mapping that is just as far-ranging in scope as the early surface geological mapping campaigns is emerging. It is the direct result of the advent of 2D and subsequently 3D seismic data along with advances in seismic acquisition and processing over the past three decades. This 21st century 'quiet' revolution is driven by the increasingly sophisticated technological demands made by today's oil and gas exploitation industry but surprisingly this is going on almost without remark from less directly related sectors of the academic geological community. The 3D seismic technology revolution has its roots in the 1930s when the first 2D data were acquired. A key evolutionary stage was the advent of digital recording and processing techniques during the 1960s. This facilitated 2D subsurface imaging, followed in the 1970s by 3D imaging. The first commercial 3D survey was recorded in 1975 in the North Sea and was interpreted in the same year. 3D seismic data quickly evolved from a research idea to cost-effective methods that have substantially boosted the efficiency of finding and recovering hydrocarbons. The quality of modern 3D seismic data is so high in many cases, that the data are starting to be used as a research tool and this is just beginning to allow researchers to challenge certain paradigms of stratigraphy and structural geology. The use of seismic data in the oil and gas industry quickly led to a number of scientific advances. For example a reinvigoration in stratigraphy started in the 1950s as a direct result of the development of the common mid-point method (Liner et al. 1999) and the acquisition and interpretation of 2D seismic data (Payton 1977). Widespread dissemination of the rapidly expanding 3D database has the potential to advance many geological disciplines which, in contrast to the 'stratigraphy revolution', perhaps have less direct impact on the oil and
gas industry. In particular, the availability of surveys covering several thousand square kilometres now enables basin-scale processes to be investigated using the potential high spatial resolution of 3D seismic data. New sedimentary and structural phenomena are being imaged and explained for the first time. These advances are perhaps not surprising when one considers the scale limitations of most outcrop, which historically is the most utilized type of geological data for studying structures at similar scales. Many of our fundamental geological concepts are rooted at the outcrop scale and therefore the alternative perspective provided by 3D seismic imaging holds considerable promise for developing and challenging these concepts, as well as revealing new phenomena. Examples in this volume show new phenomena that are recognized with 3D seismic, simply because their size is such that they cannot be seen in toto for what they are at outcrop. The aim of this introductory paper is to explore the breadth of the impact of 3D seismic technology on the geological sciences and to capture the overall aims of the volume: to raise the profile of 3D seismic interpretation within the Earth science discipline. The paper will set the scene for the Memoir by reviewing the progress that has been made over the past three decades in the development and application of 3D seismic technology and exploring the future opportunities. The fundamental objective of the paper is to pose the question: are we fully realizing the scientific potential that these data and the technology could offer to Earth sciences? 2 D vs 3 D s e i s m i c d a t a The ability to acquire and process 2D seismic data was developed in the 1950s; 3D seismic data followed in the 1980s (Liner et al. 1999). 3D seismic is distinguished from 2D seismic by the acquisition of multiple closely spaced lines (e.g. 25 m) that provides regular data point spacing that feeds 3D data migration during processing. This leads to a true data volume
DAVIES,R. J., CARTWR1GHT,J. A., STEWART,S. A., LAPPIN,M. & UNDERHILL,J. R. (eds) 2004.3D SeismicTechnology:Applicationto the Exploration of SedimentaryBasins. Geological Society, London, Memoirs, 29, 1-9. 0435-40521041515 9 The Geological Society of London 2004.
2
R.J. DAVIES ET AL.
from which lines, planes, slices or 'probes' can be extracted in any orientation, with nominally consistent data processing characteristics. While it is possible to acquire a dense, 'highresolution' grid of 2D lines, or create a mesh of 2D coverage by assembling spatially coincident 2D surveys of different vintages, such grids are fundamentally different from true 3D seismic data (Lonergan & White 1999). The close line spacing of 3D seismic data means that these data do not have the problems of spatial aliasing inherent to 2D seismic data, and therefore have the potential to yield better stratigraphic resolution, better migration and imaging of structural and depositional dips. The density of subsurface reflection point coverage allows stratal reflections to be mapped using automated or semi-automated trackers to provide continuous mapped surfaces that may in turn be used to derive a range of seismic and structural attributes. These attributes feature increasingly in exploration and development workflows. Fundamentally, however, the greatest benefit of 3D resides in its spatial resolving power both in terms of absolute spatial resolution and relative accuracy in image positioning due to 3D migration techniques employed during the processing of the seismic data (Yilmaz 2001). Features such as fault systems can now be mapped in much more detail than was possible with 2D seismic data, with its inherent limitations of spatial aliasing (Freeman et al. 1990).
Commercial considerations The cost of acquiring 3D datasets is significant--tens of thousands of US dollars per square kilometre for surveys that are hundreds to thousands of square kilometres in size. Therefore academic institutions rarely acquire and process these data except over very small areas. The vast majority of existing 3D seismic data have been acquired by the hydrocarbon industry due to the key role this technology plays throughout the life cycle of oil and gas exploration, development and production. 3D seismic data and technology can often reduce exploration risk, increase the accuracy of reservoir models and at its best, enable development and production wells to be positioned within complex hydrocarbon reservoirs (Dart et aI. 2004; Pickering et al. 2004). Although it may be regarded as an expensive research tool, the cost of 3D seismic data has fallen over the past 15 years (Table 1). Evolving computer technology has facilitated the proliferation of 3D seismic data with a trend of decreasing cost but increasing data quality. Increasing computer power has allowed industry and academic groups to develop increasingly sophisticated acquisition equipment and processing algorithms, leading to advances in image quality along with increasingly
Table 1. Cost versus year of acquisition for 3D seismic data in the North Sea, UK
North Sea 3D cost over time Year
k USD (sq km)
1982
70-100
1986
30
1990
12-15
1993
8-9
1999
4
2002
10-20
large 3D datasets. The main driver behind the growth of 3D seismic data over recent years, however, is global requirement for hydrocarbon production. Although the global seismic database is expanding, the rate of exploration drilling is such that the obvious prospects are quickly tested and the inventory of prospects relies increasingly on more subtle and higher risk opportunities. This is paralleled by the movement of activity into more challenging physical and political environments. The main technical challenge today (2004) continues to reside with processing geophysicists. They commonly work in collaboration with asset teams (teams working on particular exploration acreage, development or production project) which are well grounded in the stratigraphic and structural history of a particular area. Both disciplines should work jointly to produce clearer, more accurate images that improve prospect economics or increase recovery efficiency from producing assets.
Future exploration impact--global hydrocarbon reserves
A key question for 'E and P' geoscientists is to consider the range in image quality within the proportion of global hydrocarbon reserves imaged by 3D seismic. There are many assets and basins around the world where data quality is excellent but many datasets are good to poor due to geological complexity. This is evidenced in Table 2 where the marked assets and basins demonstrate a clear variability in data quality. The hydrocarbon industry continues to make considerable investment in improving seismic acquisition and processing to improve poorly image quality within successions that have a bearing on hydrocarbon exploration, development and production. Doing so reduces risk and therefore effectively increases global oil and gas reserves. Improving seismic imaging is an ongoing, long-term and sometimes uncertain approach to improving the quality of seismic interpretation that is related primarily to economics--in most cases we know how to collect the right data--instead short-term business drivers dictate that we acquire data that we can afford. Acquisition (and reprocessing) is then commonly repeated later, perhaps several times during not only field life, but before any of this, during the exploration for economic recoverable hydrocarbon volumes.
Interpreting 3D seismic data When 3D seismic data first became available, there was no experience or tool in use to optimize workflows, and early 3D interpretations were done in a series of steps inherited from the methodology developed for 2D seismic data. For example, good understanding of the seismic wavelet, careful ties to synthetic seismograms, checking datums and positioning, followed by a methodical, grid based approach to interpretation, balancing manual and automatic picking depending on data quality at a specific reflector (Brown 1999). As the volume of 3D data has expanded and technology has advanced, new workflow options have emerged. One of the most significant developments for interpretation is the evolution of the 'voxel', which is the 3D equivalent of a 'pixel'. Pixel-based interpretation and voxel interpretation use 'steering criteria' to grow interpretations around manually inserted seed points or lines. Pixels are picked on numerous 2D lines within a 3D dataset, where as voxels can be selected within a 3D cube. This increases interpretation speed but also allows interpreters to view all of the data within a seismic data cube simultaneously, rather than on a line by line basis. An opacity function allows the interpreter to instantly remove data from view--perhaps low-amplitude reflections-leaving high amplitude bodies that may represent reservoirs or hydrocarbon accumulations. Both pixel-based autopicking and
3D SEISMIC TECHNOLOGY: REALISING ITS FULL POTENTIAL?
Table 2. List of 57 fields, assets and exploration areas subjectively ordered with respect to typical seismic resolution Angola, Block 17 West of Africa, Girassol Field (high-frequency data) North Angola Southern North Sea (Quad 43) Outer Congo Basin (with the exception of sub-salt succession) Gulf of Mexico (with the exception of sub-salt succession) Malay Basin West of Africa: Congo Nigeria Shelf Nigeria Deep Water Sakahlin Offshore Black Sea Mauritania Offshore Southern Caspian Sea West of Africa: South Angola South Texas Trinidad Shelf Mahogany Mediterannean (Spain and Italy) Beaufort Sea Argentina, Neuquen Basin, Sierra Chata Field Venezuela, Heavy Oil Belt, Cerro Negro Field Brazil Offshore (Campos Basin) India Offshore Vietnam Offshore Chad Doba and West Doba Basin Inner Moray Firth, UK West Texas, USA Alberta and B C ~ a n a d a Azerbaijan--South Caspian Russia/Kazakhstan/Azerbaij an--Middle Caspian West Siberia Foz Do Amazonas Basin Kazakhstan--PriCaspian PreSalt Australia--NW Shelf Michigan Central North Sea, UK Australia--Bass Straights McKenzie Delta--Canada Cook Inlet Alaska Moray Firth North Sea Argentina San Jorge Falklands West of Shetland--no Paleocene Basalt Cover Irish Sea US fold and thrust belt Turkey--Onshore Papua New Guinea--Onshore North Atlantic Rockall Flemish Cap---similar to Rockall off Nova Scotia United Arab Emirates thrust belt Bolivian Andes PNG fold and thrust Trinidad Onshore North Sea giant chalk sub-gas cloud Gulf of Mexico--sub-salt West of Shetland--with Paleocene Basalt Cover Gulf of Suez sub-salt
voxel autopicking are sensitive to signal variations as sensed by the steering criteria, whether those variations are actual changes in the geology or whether there is noise in the dataset. Noise level, or data quality, is an extrinsic uncertainty that masks the level of geological complexity which is an intrinsic uncertainty in the data. This framework unsurprisingly suggests that voxelbased approaches are of highest value in tackling the rapid definition of simple structures in good seismic data quality settings.
3
Pitfalls of the 3D seismic technology revolution Rapid advances in technology commonly have unforeseen pitfalls that go unnoticed in the excitement that drives the change. We identify three challenges.
The interpreter's m i n d s e t "We met the enemy and he is us'--quote by American cartoonist Walt Kelly (1970). Perhaps the most significant potential pitfall lies with us, the interpreters of 3D seismic data and it is rooted in our basic behaviour. Many modern interpreters began their professional lives without any in-depth experience of 2D seismic interpretation. The research or asset team focus--and therefore the interpreter's focus--is swiftly directed at surface mapping, attribute analysis and visualization: all of which are key tools available to interpreters of 3D seismic data. Will the present and next generation of interpreters be handicapped for the lack of an apprenticeship in the 'harder' world of 2D seismic interpretation? This world had different challenges of interpolation between widely spaced lines as a precursor to mapping but also a more holistic approach where the seismic packages and their geometries were valued as much as correlations of individual horizons. The tendency exists to pick fewer lines and interpolate between widely spaced sections and we term this 'data underutilization'. The data are valuable and we must take care not to render the data between mapped horizons as opaque or invisible through overuse of this interpretation workflow.
Geophysical grounding
While interpretation tools have become increasingly accessible it is likely that data acquisition and processing will become more complex. Difficulties are compounded in multi-dimensional seismic data types, such as 4D (Bagley et al. 2004; Chadwick et al. 2004) and 4C. An understanding of the assumptions made in the processing of 3D seismic requires an ever-deeper understanding of geophysics. For example, multiple attenuation algorithms are becoming increasingly sophisticated and need careful supervision to ensure that primary reflectors are not erroneously removed. This requirement mirrors a necessity for the data interpreter to widen their 'skillset' to encompass the range of geological architectures that are resolved on higher quality data. The impact of this on professional development--whether the specialists or generalists have the key roles--is currently a subject of debate in the oil industry. The need for geophysical understanding and careful calibration (Brown 2001) is paramount now and will become more so in the future. The view that 3D seismic volumes are 'true' realizations of geological volumes is an assumption that can result in exploration, development and production failure; and can incur commercial penalties (e.g. Stewart & Holt 2004). Basic acquisition problems along with various sources of noise, mispositioning of seismic energy and tuning (e.g. Yilmaz 2001 ) are as significant now as they have ever been. Exploration acreage can be located within complex geological settings and this forces interpreters to work right at the limits of data resolution and in many instances beyond (e.g. 'ghosting' horizons through areas of poor data). At the scale of many reservoirs the interpretation of flow units that are below seismic resolution is sometimes a necessity when projects require it--this can be termed 'data overutilization'. Arguably other examples of data overutilization come from the misuse of so-called direct
4
R.J. DAVIES ETAL.
hydrocarbon indicators (DHIs). Many experienced industry interpreters will be able to recount examples of prospects that were dramatically de-risked through the identification of a DHI that did not make geological sense. In many cases prospects have been drilled on 'overutilized' data and have been proven dry. Other common pitfalls include the tendency in clastic successions to view seismic amplitudes as a direct indicator of lithology. This of course should be cautioned against as fluid content can significantly reduce acoustic impedance and therefore seismic amplitude. To get indications of fluid or lithology from 3D seismic or impedance volumes typically requires further processing after a thorough petrophysical study of the rock involved (Whitcombe et al. 2002).
T h e e v o l v i n g role o f the s e i s m i c i n t e r p r e t e r
To meet the challenge of the pace with which the technology is advancing the interpretation community needs to include fully integrated ('quantitative') geoscientists with a general understanding of all aspects of data acquisition, processing and a specialized knowledge of interpretation. A significant development related to the growth of 3D seismic data as the 'core' of assessment of hydrocarbon occurrence, volume and distribution is the breakdown in the division of geologist and geophysicist in the hydrocarbon exploration and production industry. Until very recently it was typical for the 'geophysicist' to develop an understanding of 'the container' using 2D and 3D seismic data tied to well 'tops' while the geologist provided information on reservoir distribution and quality using wireline logs, core data and a regional understanding. This is changing: the modem interpreter must truly be a multidisciplinarian, well versed in subjects as diverse as petrophysics and sequence stratigraphy. Continued professional training is thus a priority in such a demanding environment.
3D seismic data: impact on Earth sciences Despite the inevitable pitfalls the technology is positioned to have a tremendous impact on Earth sciences. The history of research in the geosciences is populated with examples of paradigm shifts inspired by new technology, for example submarine warfare technology and its role in the recognition of marine magnetic anomalies associated with sea floor spreading. 2D reflection seismology has already played a key role in the evolution of concepts of extensional tectonics and stratigraphy during the 1970s and 1980s, perhaps in part because the academic community was fully involved in the acquisition of the data. Indeed, academic programs in deep reflection seismic were responsible for some important breakthroughs in rift tectonics and basin development (e.g. the BIRPS and COCORP projects--Klemperer & Hobbs 1991 ). The abundance of large 3D seismic surveys now represents a significant opportunity for research geoscientists from a diverse range of disciplines to benefit from this petroleum industry investment. Research is no longer spatially restricted to tens of square kilometres and the typical extent of an oil and gas field. In some areas sufficient 3D seismic data has historically been acquired so that 'megamerges' of the surveys provide coverage of entire sections of basins, for example in the North Sea Basin, where only ten years ago, there would have been incomplete coverage of variable quality 2D data with local 3D across producing fields. These aerially extensive surveys allow basin analysis at a very high spatial resolution afforded by the 3D grid spacing. This means that there is no loss of detail with increasing area and that structural and
stratigraphic elements can be placed in a basin wide context. This gives basin analysis a fundamental new tool with which to tackle diverse issues such as basin modelling, e.g. White et al. (2004), basin wide fluid flow, sub-regional tectonics or depositional systems and their stacking. A small number of academic studies have nevertheless been pursued with 3D seismic data as the principal research medium either because of their intrinsic interest, because they are based on collaborative partnerships with industry, or because of serendipitous 'discovery' whilst in pursuit of objectives of economic interest. Recognition of new geological structures is always possible on 3D surveys because of the newly available resolving power (Cartwright 1994; Davies et al. 1999, Davies in press). Recently for example a 3D seismic approach was applied to the investigation of meteor impact craters and this has raised awareness of the potential of 3D seismic amongst the specialists investigating cratering on the terrestrial planets (Stewart & Allen 2002). Every new map, whether it be a time map or seismic attribute has the potential to reveal features that we are yet to fully appreciate in the field. Such features are not identified in the geological lexicon. Due to simple economic prerogatives, some advances have been made in the effort to maximize production from discovered accumulations. For example the study of post-depositional remobilization of clastic reservoirs (Lonergan & Cartwright 1999; Huuse et al. 2004). Certain types of remobilisation structures illustrate how 3D seismic allows for the identification of features that currently have no good field analogue (Molyneux et al. 2002; Gras & Cartwright 2002). The same principle applies to the study of a diverse range of soft-sediment deformation structures from density inversion folds (Davies et al. 1999) and polygonal faults (Cartwright 1994) to giant pockmarks (Cole et al. 2000). Soft sediment deformation is likely to receive much more attention in the future, not least because it is often apparent in the highest frequency part of the seismic profile (e.g. Davies in press). There is a wealth of seismic data from the first second of two-way travel time below mud line, that has no commercial value but covers geological phenomena of significant academic interest (Knutz & Cartwright 2004; Smallwood 2004) or has important implications for offshore installation integrity (Austin 2004; Long et al. 2004) and well planning (Stewart & Holt 2004).
Stratigraphy
The concepts of seismic stratigraphy (Payton et al. 1977) were based on 2D seismic data but the advent of 3D seismic data now allows for individual depositional elements to be recognized and for the interpreter 'to go beyond the parasequence'. Studies in this volume (Fowler et al. 2004; Frey Martinez et al. 2004; Posamentier 2004; Robinson et al. 2004; Steffens et al. 2004) illustrate three-dimensional seismic facies distribution and stratigraphic architecture and demonstrate the degree to which research in these disciplines has advanced today. Recent focus has been on deepwater depositional systems. In this setting it is commonplace to use several reprocessed seismic volumes of an original dataset that are designed to exploit various rock properties calibrated to borehole petrophysics (Whitcombe et al. 2002). These datavolumes typically display the seismic differences between fluid and lithology, the presence or absence of AVO anomalies, qualitative and quantitative acoustic impedance inversion. In addition, parallel advances in software manipulation have enabled the development of spatial-stacking or optical stacking techniques for enhanced flat-spot analysis (Worrel 2001). All or some of the described techniques have been used to further
3D SEISMIC TECHNOLOGY: REALISING ITS FULL POTENTIAL? augment the seismic resolution of complex sedimentary architectures inherent in deepwater sands. Indeed visualization of ancient deepwater processes via the highest quality 3D datasets is providing the interpreter with startling images of sinuous channel complexes on deepwater slopes. What began as a primarily model-driven view of the relationship of reflection seismic data to depositional models has evolved to a point where the recognition of process-derived facies distributions can be visualized directly (Fowler et al. 2004; Steffens et al. 2004; Posamentier 2004). The understanding of clastic depositional processes has received much attention from academic researchers because of the shift towards deepwater clastic reservoirs as exploration targets (e.g. Morgan 2004). This will continue to be a major growth area in the next decade, but researchers will bridge disciplines to tackle the interactions between sedimentation and tectonics in a host of deep water settings (e.g. Hansen et al. 2004). The application of 3D seismic data to disciplines such as geomorphology, a field that has now been coined 'seismic geomorphology' (Posamentier 2004) are still being advanced by industry geoscientists.
Structural geology
The most significant advances in structural geology that have resulted from the application of 3D seismic interpretation are probably fault system geometry (e.g. Dutton et al. 2004; Jones et al. 2004; McClay et al. 2004) and kinematics and salt tectonics (e.g. Rank & Elders 2004; Trudgill & Rowan 2004). Examples include mapping distributions of displacement on fault surfaces (Nicol et al. 1996; Walsh et al. 2002; Lister 2004) and using mapped stratal terminations projected onto the fault surface plane, or Allan diagram, to map fault rock properties (Bouvier et al. 1989). 3D mapping of fault planes and intersections has allowed topological frameworks to be devised (Nicol et al. 1996) and specific 3D problems have been encountered that have caused structural fundamentals such as strain to be revisited (Cartwright & Lonergan 1996). More recently, large basement faults in rift systems have been studied using regional surveys to examine the kinematic evolution of basin-scale tectonostratigraphic architecture (Dawers & Underhill 2000; MacLeod et al. 2002). Future research will almost certainly extend the insights gained into the evolution of normal fault systems to the study of thrust and wrench fault systems. In addition to fault geometries, 3D seismic can contribute to more general strain analysis by defining the geometry of growth strata (Bouroullac 2001) and controlling 3D structural restoration that may reveal subseismic strain distribution. 3D seismic is also a powerful means for delineating small faults and fractures that can exert a major influence on field performance (Hesthammer & Fossen 1997).
Igneous geology
Although conventionally a subject restricted to field-based researchers, large acoustic impedance contrasts with surrounding sediments means that igneous phenomena easily manifest themselves on 3D seismic surveys located in the petroliferous volcanic margins of the UK, Norway, Brazil and West Africa. Complexes of igneous sills and flood basalts have been identified in these areas (Planke et al. 2000; Davies et al. 2002; Hansen et al. 2004; Trude 2004). Igneous sills in particular are a classical illustration of the optimum use of 3D seismic in a research context. They are very well imaged on 3D seismic because of their large impedance contrasts with the host sediments, and hence are relatively straightforward to interpret. This
5
has meant that the three-dimensional geometry of sills can be defined with considerable accuracy, and this has led to several novel conclusions about the interactions between sills and their host rocks as well as the fundamental mode of emplacement itself (Hansen et al. 2004; Trude 2004). These insights were not possible with field-based approaches alone, because of the incompleteness of the outcrop of even the best exposed sills.
Future potential As with any technological advance there are likely to be both predictable and unpredictable innovations as well as surprises. AVO, pre-stack depth migration, long offset 3D, 4D (e.g. Bagley et al. 2004; Chadwick et al. 2004) and other technologies such as neural network based detection systems (e.g. De Groot et al. 2004) that have emerged in the past ten years now fill geophysical journals. If investment means that the fields of research into 3D seismic technology and its application in Earth sciences are well fertilized then opportunities are significant. We can also consider the spin-off technologies such as 3D visualization (e.g. Bond et al. 2004; Corfield et al. 2004; Lynch 2004) that now allow key geological problems within a prospective region to be assessed and an efficient work direction decided within an afternoon, rather than over a period of weeks. Perhaps the application of this technology in Earth sciences may result in every Earth science department in the first World being equipped with a visionarium in 15 years time. Such a facility would be used for teaching and research but not just to look at seismic data but outcrops, core plugs and any other data that benefits from communication in an immersive 3D environment. In this paper we cannot focus on every avenue that may bear fruit but a notable absentee from the Memoir are papers devoted to what may prove to be an important area of future technological growth-4C seismic.
4C seismic 3D seismic data are essentially a discretization of the Earth in terms of the properties of sound waves of which there are three main types: surface, interface and body. Each of these is characterized by the nature of its wave propagation in Earth materials. The body waves are of most use in the seismic data context as they propagate information through the Earth, and are not confined to boundaries. There are two types of body wave: longitudinal (P-) waves, which transmit information by compressing particles in the Earth back and forth in the direction of wave travel; and transverse (S-) waves, which transmit information by shearing particles past each other in directions perpendicular to the wave direction. Seismic sources can in fact be designed to emit either wave type, and receivers designed to record either. 3D seismic images formed from marine towed streamer data typically use the properties of P-waves to remotely sense the Earth, because S-waves do not propagate in fluids. In contrast, on land 3D seismic images can be formed using either P- or S-waves, depending on the source of waves and the type of receiver at the surface. Since shear waves propagate by causing particle motion perpendicular to the direction of travel, they can only be recorded properly by an arrangement of three geophones that are sensitive to particle velocity or acceleration. It is normal to arrange the geophones in three mutually orthogonal directions, such as in the X, Y and Z directions of a Cartesian coordinate frame, thus representing the three components (3C) of a vector recording of the particle motion. 4C seismic is a method of acquiring marine seismic data that combines three orthogonal geophones from land acquisition with the
6
R.J. DAVIES ET AL.
hydrophone from towed streamers to give four-component (4C) recording. This is made possible by deploying the specially designed 4C sensor packages on the sea floor so that they are coupled to the elastic Earth to record particle motion on the geophones, but still in the water to record the pressure on the hydrophones. The marine source is typically an airgun array that creates P-waves in the water. Due to the partitioning of energy at elastic boundaries within the Earth (reflectors, reflector terminations, etc.), conversions occur from P- to S-(and vice versa) so that both P- and S-waves are recorded. It is possible, and entirely likely that these conversions occur thoughout the subsurface. However, only the strongest conversions are observed in the processed 3D volume, and these tend to occur where the highest contrast exist, such as at the sea floor (Tatham & Goolsbee 1984) or, more usually, at the reservoir refectors. There are numerous potential benefits that 4C may bring, for example: (1) imaging where towed streamer (P-wave) data cannot, for example, through gas chimneys, low P-impedance reservoirs, beneath salt, basalt or mud volcanoes; (2) reduction of water column multiple energy through 'PZ summation'; (3) flexible receiver geometries on the seafloor permit acquisition of long offsets and wide azimuths which improve illumination, fold and SNR; (4) lithology and fluid prediction, by direct measurement of shear waves for AVO, as opposed to inference from P-wave data alone; (51 fracture mapping from wide-azimuth P-wave data, and C-wave splitting analysis to give fracture orientation, fracture density and pore-fluid fill. The first commercial success of 4C seismic took place in 1994 in the North Sea (Berg et al. 1994), which showed that P-to-S conversions at deep reflectors (C-waves; Thomsen 1999) could provide an image through a gas cloud where the conventional P-wave image was obscured. This is mostly due to the pore fluid (gas) being invisible to the S-wave leg, whereas the P-waves are attenuated heavily. Although there have been very many 2D test 4C surveys, there have been relatively few 3D 4C surveys worldwide. They include Alba, Emilio, Gullfaks, Hod, Lomond, Staffjord and Valhall. There are major challenges ahead with acquisition and processing of 4C data, particularly with the X and Y components to form converted wave images.
True 4D seismic and the 'electric oilfield' 3D seismic provides a static picture of the Earth. To understand dynamic Earth processes requires observation over time. 3D seismic can provide the dynamic data in the form of repeat surveys over the same area. Over the last ten years much effort has gone into developing workflows to process multiple 3D volumes from the same area to emphasize changes due to the dynamic processes. Just as two or three 2D seismic lines would not be considered 3D seismic, so two or three 3D surveys cannot be considered 4D seismic, rather they should be termed more aptly time-lapse 3D seismic. The method of acquiring time-lapse 3D can be towed streamer, 4C or a combination of both. A principal objective in time-lapse seismic processing has been to remove acquisition differences between repeat surveys (such as variations in towed cable feathering), and to make processing as similar as possible. Installing an array of 4C seismic sensors permanently on the seafloor potentially provides very repeatable 3D seismic. 3D seismic can then be acquired with a shooting vessel as frequently as required, for example, every few months for the lifetime of an oilfield field. This is true 4D seismic since the time axis has more than a few points and dynamic effects may be observed, rather than inferred from the differences between
static 3D surveys. The first of these true 4D surveys is documented in Barkved et al. (2003). A permanent installation of sensors on the seabed coupled with instruments in wells also provide the opportunity for further monitoring of the subsurface, for example, dynamic subsidence in the overburden, and micro-earthquake events from sub-seismic faulting in the reservoir--the 'electric oilfield' vision of dynamic Earth monitoring.
Conclusions The most fundamental impact of 3D seismic was a major improvement in imaging, positioning of seismic energy and spatial frequency of data. The most closely spaced 2D seismic grids have line spacing in the order of hundreds of metres-exploration surveys were often of kilometre grid spacing. With no control on how sparsely sampled phenomena link along strike, fault patterns and displacement profiles, for example, are spatially aliased. 3D reduces the onset of aliasing by at least an order of magnitude, to around 20m and the increase in resolution is obviously more significant if factored volumetrically. So phenomena that existed at a hundreds of metre to kilometre scale were imaged in 3D for the first time. One could take the view that 2D and 3D seismic data are the first tools to directly image the subsurface in three-dimensions and that their advent represents one of the most significant new techniques available to the solid Earth sciences of any developed within the past century. The advent of this new type of data has created an opportunity to train the next generation of geoscience students in three-dimensional subsurface mapping in addition to the training they receive in traditional surface mapping techniques. By doing so this generation will be cognizant of its utility for understanding basin forming and filling processes just as the present generation of geoscientists understands the benefit of detailed geological mapping. Whilst the data stream comes mainly from the petroleum industry, the opportunities for research will be mainly in prospective basins. However, as the cost of acquisition and processing decreases, there will be increasing use of 3D surveying for primary research purposes (Heffernan et aL 2004). The major challenges facing academic exploitation of this extraordinary data resource are how to equip laboratories capable of handling large data volumes, and how to persuade the hydrocarbon industry and governmental partners and sponsors to provide the means to do so. ML, SS, SF, RJ thank ExxonMobil Exploration Company, BP and Shell for permissionto publish this paper. E. Jansen of Schlumbergerprovided information used to construct Table 1. M. Huuse made comments on an early draft of this paper. T. Dor& A. Fraser and J. Howe provided helpful reviews.
References AVSTIN, B. 2004. Intergrated use of 3D seismic in field development engineering and drilling: examples from the shallow section. In: DAVIES,R. J,, CARTWRIGHT,J. A,, STEWART,S. A., LAPPIN,M. & UNDERHILL,J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentar3' Basins. Geological Society, London, Memoirs, 29, 279-296. BAGLEY,G., SAXBY,I., MCGARRITY,J., PEARCE,C. 8~.SLATER,C. 2004. 4D/time lapse seismic: examples from the Foinhaven, Schiehallion and Loyal Fields, UKCS, West of Shetland. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART,S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 297-302.
3D SEISMIC TECHNOLOGY: REALISING ITS FULL POTENTIAL? BARKVED, O.I., BAERHEIM, A.G., HOWE, D.J., KOMMEDAL, J.H., & NICOE, G., (2003). Life of Field Seismic Implementation--"First at Valhall". Implementing 4D Seismic in the Management of Producing Reservoirs. 65th Annual Conference and Exhibition of the European Association of Geoscientists and Engineers, Workshop proceedings. BERG, E., SVENNING, B. 8`: MARTIN, J. 1994. SUMIC: Multicomponent sea bottom seismic surveying in the North Sea---data interpretation and applications. 64th Annual International Meeting of the Socie~ of Exploration Geophysocists, Expanded Abstracts, 477-480. BROWN, A. R. 1999. Interpretation of Three-Dimensional Seismic Data. 5th Edition, American Association of Petroleum Geologists, Memoir, 42. BROWN, A. R. 2001. Calibrate yourself to your data! A vital first step in seismic interpretation. Geophysical Prospecting, 49, 729-733. BOUROULLAC,R. 2001. The Kinematics of Growth Faults. PhD Thesis, University of London. BOUVIER, J. D., KAARS-SIJPESTEIJN, C. H., KLUESNER, D. F., ONYEJEKWE, C. C. 8`: VAN DER PAL, R. C. 1989. Three-dimensional seismic interpretation and fault sealing investigations, Nun River Field. AAPG Bulletin, 73, 1397-1414. CARTWRIGHT, J. A. 1994. Episodic basin-wide fluid expulsion from geopressured shale sequences in the North Sea Basin. Geology, 22, 447 -450. CARTWR1GHT, J. A. 8`: LONERGAN, L. 1996. Volumetric contraction during the compaction of mudrocks: a mechanism for the development of polygonal fault systems. Basin Research, 8, 183-193. CHADWICK, R. A., ARTS, R,, LIKEN, O., KIRBY, G., LINDEBERG, E. & ZWEIGEL, P. 2004. 4D seismic imaging of a CO2 plume at the Sleipner Field, central North Sea. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPIN, M. & UNDERH1LL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary. Basins. Geological Society, London, Memoirs, 29. 311-320. COLE, D., STEWART, S. A. 8,: CARTWRIGHT, J. A. 2000. Giant irregular pockmark craters in the Palaeogene of the Outer Moray Firth Basin, UK North Sea. Marine and Petroleum Geology, 17, 563-577. CONNOLLY, P. 1999. Elastic impedance. The Leading Edge, 18(4), 438 -452. CORFIELD, S. M., WHEELER, W., KARPUZ,R., WILSON, M. & HEI,LAND, R. 2004. Exploration 3D seismic over the Gjallar ridge, MidNorway: visualization of structures on the Norwegian volcanic margin from Moho to seafloor. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPIN, M. 8`: UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimenta O' Basins. Geological Society, London, Memoirs, 29, 177-185. DART, C., CLOKE, I., HERDLEVOER, A., GILLARD, D., R1VENOES, J., OTTERLEI, C., JOHNSEN, E. & EKERN, A. 2004. Use of 3D visualization techniques to unravel complex fault patterns for production planning: Njord field, Walten Terrace, Norway. In: DAVIES, R. J., CARTWRIGHT,J. A., STEWART, S. A., LAPPlN, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimenta~, Basins. Geological Society, London, Memoirs, 29, 249-261. DAVIES, R, J. 2004. Kilometer-scale fluidization structures formed during early burial. Geology, in press. DAVIES, R. J., CARTWRIGHT,J. 8`: RANA, J. 1999. Giant hummocks in deep-water marine sediments: Evidence for large-scale differential compaction and density inversion during early burial. Geology, 27, 907-910. DAVIES, R. J., CARTWRIGHT,J. A., BELL, B. R. & SHOULDERS,S. 2002. Three-dimensional seismic imaging of Paleogene dike-fed submarine volcanoes from the northeast Atlantic margin. Geology, 30, 223-226. DAWERS, N. H. & UNDERHILL, J. R. 2000. The role of fault interaction and linkage in controlling syn-rift stratigraphic sequences: Statfrod East area, northern North Sea. AAPG Bulletin, 84, 45-64. DE GROOT, P., L1GTENBERG, H., OLDENZIEL, T., CONNOLLY, D. & MELDAHI,, P. 2004. Examples of multi-attribute, neural networkbased seismic object detection. In: DAVIES, R. J., CARTWR1GHT,
7
J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 333-337. DVTTON, D. M,, LISTER, D., TRLDGILL, B. D. & PEDRO, K. 2004. Threedimensional geometry and displacement configuration of a fault array from a raft system: Lower Congo Basin, Offshore Angola: implications for the Neogene turbidite play. In: DAVIES, R. J., CART\VRIGHT, J. A., STEWART, S. A., LAPPIN, M. & UNDERHILI,, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 133-142. FOWLER, J. N., GURITNO, E., SHERWOOD, P., SMITH, M. J., ALGAR, S., BCSONO, I., GOFI:EY, G. & STRONG, A. 2004. Depositional architectures of recent deepwater deposits in the Kutei Basin, East Kalimantan. hT: DAVIES, R. J., CARTWRIGHT,J. A., STEWART, S. A., LAPPIN, M. & UNDERHII,L, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentar3' Basins. Geological Society, London, Memoirs, 29, 25-33. FREEMAN, B., YIELHNG, G. & BADEEY, M. E. 1990, Fault correlation during seismic interpretation. First Break, 8, 87-95. FREY M.-\RTINEZ,J., CARTVCRIGHT,J., BURGESS, P. M. & FERNANDEZ,J. 2004. 3D seismic interpretation of the Messinian unconformity in the Valencia Basin, Spain./ti: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPIN, M, ~ UNDERHILL,J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentar)" Basins. Geological Society, London, Memoirs, 29, 91-100. GRAS, R. 8`: CARTWRIGHT, J. A. 2002. Tornado faults: seismic expression on PS data from the Chestnut Field, O4th European Association of Exploration Geologists, extended abstracts, H020. HANSEN, D. M,, CART\VRIGHT,J. A. & THOMAS, D. 2004. 3D seismic analysis of the geometry of igneous sills and sill junction relationships, h~: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPIN, M. & UNDERHIEL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentar3" Basins. Geological Society, London, Memoirs, 29, 199-208. H.~NSEN, J. P. V., CI.At:SEN, O. R. & Ht:t:SE, M. 2004. 3D seismic analysis reveals the origin of ambiguous erosional features at a major sequence boundary in the eastern North Sea: near top Oligocene. hi: DAVIES, R. J., CARTWR1GHT,J. A., STEWART, S. A., L,\PPIN, M. 8`: UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins, Geological Society, London, Memoirs, 29, 83-89. HEFFERNAN, A. S., MOORE, J. C., BANGS, N. L., MOORE, G. F. & SHIPLEY, t . H. 2004. Initial deformation in a subduction thrust system: polygonal normal faulting in the incoming sedimentary sequence of the Nankai subduction zone, southwestern Japan. In: DAVIES, R. J., CART\\'RIGHT, J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 143-148. HESTHAMMER, J. 8`: FOSSEN, H. 1997. The influence of seismic noise in seismic interpretation. First Break, 15, 209-213. HURST, A., CARTWRIGHT, J. A. & DL'RNATI, D. 2003. Fluidization structures produced by upward injection of sand through a sealing lithology. In: VAN RENSBERGEN, P. ET AL. (ed.) Subsurface Sediment Mobilization. Geological Society, London, Special Publication, 216, 123-137. HUUSE, M., DURANTI, D., STEINSLAND, N., GUARGENA,C. G., PRAT, P., HOLM, K., CARTWRIGHT, J. A. & HURST, A. 2004. Seismic characteristics of large-scale sandstone intrusions in the Paleogene of the South Viking Graben. UK and Norwegian North Sea. ln: DAVIES, R. J., CARTWRIGHT,J. A., STEWART, S. A., LAPPIN, M. & UNDERHILI., J. R. (eds)3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 263-277. JAMES, H., BOND, R. & EASTWOOD, L. 2004. Direct visualization and extraction of stratigraphic targets in complex structural settings. In: DAVIES, R. J., CARTWRIGHT,J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to
8
R.J. DAVIES ETAL.
the Exploration of Sedimentary. Basins. Geological Society, London, Memoirs, 29, 227-234. JONES, G., WILLIAMS, L. & KNIPE, R. J. 2004. Structural evolution of a complex 3D fault array in the Cretaceous and Tertiary of the Porcupine Basin, offshore Ireland. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPP/N, M. & UNOERHILL,J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentao' Basins. Geological Society, London, Memoirs, 29, 117-132. KLEMPERER, S. & HOBBS, R. 1991. The BIRPS atlas: deep seismic reflection profiles around the British Isles. Cambridge University Press. KNUTZ, P. C. ~,~CARTWRIGHT,J. A. 2004. 3D anatomy of late Neogene contourite drifts and associated mass flows in the Faroe-Shetland Channel. In: DAVIES, R. J., CARTWR1GHT,J. A., STEWART, S. A., LAPPIN, M. & UNDERHII.L, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimenta O' Basins. Geological Society, London, Memoirs, 29, 63-71. LINER, C. L., HERMAN, G. C., MARFURT, K. J. & SCHUSTER, G. T. (eds.) 1999. Geophysics, Jaargang, 64. LISTER, D. L. 2004. Modelling fault geometry and displacement for very large networks. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPtN, M. & UNDERHILI., J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentar), Basins. Geological Society, London, Memoirs, 29, 339-348. LONERGAN, L. & CARTWRIGHT, J. A. 1999. Polygonal faults and their influence on reservoir geometries, Alba Field. AAPG Bulletin, 83, 410-432. LONERGAN, L. & WHITE, N. 1999. Three-dimensional imaging of a dynamic Earth. Philosophical Transactions of the Royal Socie~' q[ London Series A, 357, 3359-3375. LONERGAN, L., LEE, N., JOHNSON, H. D., CARTWRIGHT, J. A. & JOLLY, R. 2000. Remobilisation and injection in deepwater depositonal systems. In: WE~MER, P. et al. (ed.) Deep Water Reservoirs. GCSEPM Foundation, 20th annual conference, Houston, 515-532. LONG, D., BULAT,J. & STOKER, M. S. 2004. Sea bed morphology of the Faroe-Shetland Channel derived from 3D seismic datasets. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPP1N, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimenta~. Basins. Geological Society, London, Memoirs, 29, 53-61. LYNCH, J. J. 2004. Visualization and interpretation of 3D seismic in the Carboniferous of the UK Southern North Sea. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration ofSedimenta~ Basins. Geological Society, London, Memoirs, 29, 219-225. MACLEOD, A. E., UNDERHILL, J. R,, DAVIES, S. J, DAWERS, N. H. (2002). The Influence of fault array evololution on sysnrift sedimentation patterns: Controls on deposition in StrathspeyBrent-Statfjord half graben, northern North Sea. American Association of Petroleum Geologists Bulletin, 86, 6, 1061 - 1093. MCCLAY, K. R., DOOLEY, T., WHITEHOUSE, P. FULLARTON, L. & CHANTrAPRASERT, S. 2004. 3D Analogue models of rift systems: templates for 3D seismic interpretation. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPIN, M. & UNDERHII,I~, J. R. (eds) 3D Seismic Technology." Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 101-115. MOLYNEUX, S. J., CARTWRIGHT, J. A. & LONERGAN, L. 2002. Giant conical sandstone intrusions in the Tertiary of the North Sea. First Break, 20, 383-389. MORGAN, R. 2004. Structural controls on the positioning of submarine channels on the lower slopes of the Niger Delta. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 45-51. NICOL, A., WATTERSON, J., WALSH, J. J. & CH1LDS, C. 1996. The shapes, major axis orientations and displacement patterns of fault surfaces. Journal of Structural Geology, 18, 235-248.
PAYTON, C.E. 1977. Seismic Stratigraphy--Applications to Hydrocarbon Exploration, AAPG Memoir 26. PICKERING, G,, KNIGHT, E., BLETCHER, J., BARKER, R. & KEMPER, M. 2004. Locating exploration and appraisal wells using predictive rock physics, seismic inversion and advanced body tracking: an example from North Africa. In: DAVIES, R. J., CARTWRIGHT,J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL,J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 235-248. PLANKE,S., SYMONDS,A., AI.VESTAD,E. & SKOGSEID,J. 2000. Seismic volcano-stratigraphy of large volume basaltic extrusive complexes on rifted margins. Journal of Geophysical Research, 105, B8, 19335-19353. POSAMENTIER, H. W. 2004. Seismic geomorphology : imaging elements of depositional systems from shelf to deep basin using 3D seismic data: implications for exploration and development. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentao' Basins. Geological Society, London, Memoirs, 29, 11-24. RANK-FRIEND, M. & ELDERS, C. F. 2004. The evolution and growth of Central Graben salt structures, Salt Dome Province, Danish North Sea. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPP1N, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimenta O' Basins. Geological Society, London, Memoirs, 29, 149-163. ROBINSON, A. M., CARTWRIGHT,J. A., BURGESS, P. M. & DAVIES, R. J. 2004. Interactions between topography and channel development from 3D seismic analysis: an example from the Tertiary of the Flett Ridge, Faroe-Shetland Basin, UK. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 73-82. SMAI.LWOOD, J. R. 2004. Tertiary inversion in the Faroe-Shetland Channel and the development of major erosional scarps. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimenta~ Basins. Geological Society, London, Memoirs, 29, 187-198. STEFFENS, G. S., SHIPP,R. C., PRATHER,B. E., NOTT, J, L., GIBSON, J. L. & WINKER, C. D. 2004. The use of near-seafloor 3D seismic data in deepwater exploration and production. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 35-43. STEWART, S. A. & ALLEN, P. J. 2002. A 20-km-diameter multi-ringed impact structure in the North Sea. Nature, 418, 520-523. STEWART, S. A. & HOLT, J. 2004. Improved drilling performance through integration of seismic, geological and drilling data. In: DAVIES, R. J., CARTWRIGHT,J. A., STEWART,S. A., LAPPIN,M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentao' Basins. Geological Society, London, Memoirs, 29, 303-310. TATHAM, R. H. & GOOLSBEE, D. V. 1984. Separation of shear-wave and P-wave reflections offshore Western Florida. Geophysics, 49, 5, 493-508. THOMSEN, L. A. 1999. Converted-wave reflection seismology over anisotropic, inhomogeneous media. Geophysics, 64, 678-690. TRLDE, K. J. 2004. Kinematic indicators for shallow level igneous intrusion from 3D seismic data: evidence of flow direction and feeder location. In: DAVIES, R. J., CARTWRIGHT,J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 209-217. TRUDGILL, B. D. & ROWAN, M. G. 2004. Integrating 3D seismic data with structural restorations to elucidate the evolution of a stepped counter-regional salt system, Eastern Louisiana shelf, Northern Gulf of Mexico. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 3D Seismic
3D SEISMIC TECHNOLOGY: REALISING ITS FULL POTENTIAL?
Technology: Application to the Exploration of Sedimentar3, Basins. Geological Society, London, Memoirs, 29, 165-176. WALSH, J. J., NICOL, A. & CHILDS, C. 2002. An alternative model for the growth of faults. Journal of Structural Geology, 24, 1669-1675. WHITCOMBE, D. N., CONNOLLY, P. A., REAGAN, R. L. & REDSHAW. T. C. 2002. Extended elastic impedance for fluid and lithology prediction. Geophysics, 67, 63-67. WHITE, N,, I-IaINES, J., JONES, S. & HANNE, D. 2004. Towards an automated strategy for modelling extensional basins and margins in four dimensions. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPIN, M. & UNDEr~r~ILL, J. R. (eds) 3D
9
Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 321-331. WINCHESTER, S. 2001. The map that changed the world: William Smith and the birth of modern geology. London, HarperCollins. WORREL. A. 2001. Rapid and accurate 3D fault interpretation using opacity and optical stacking to reveal geologic discontinuities. The Leading Edge, 20, 1381-1385. YILMAZ, O. 2001. Seismic data analysis processing, inversion and interpretation of seismic data. SocieD' of Exploration Geophysicists, 2027,
Seismic geomorphology: imaging elements of depositional systems from shelf to deep basin using 3D seismic data: implications for exploration and development HENRY
W. P O S A M E N T I E R
Anadarko Canada Corporation, 425 1st Street SW, Calgary.', Alberta T2P 4V4, Canada (e-mail:
[email protected] )
Abstract: 3D seismic data can play a vital role in hydrocarbon exploration and development especially with regard to mitigating risk associated with presence of reservoir, source, and seal facies. Such data can afford direct imaging of depositional elements, which can then be analyzed using seismic stratigraphy and seismic geomorphology to yield predictions of lithologic distribution, insights to compartmentalization, and identification of stratigraphic trapping possibilities. Benefits can be direct, whereby depositional elements at exploration depths can be identified and interpreted. or they can be indirect, whereby shallow-buried depositional systems can be clearly imaged and provide analogues to deeper exploration or development targets. Examples of imaged depositional elements from both shallow and deep sections are presented.
Seismic data have long been used for lithologic prediction. Initially, such interpretations were based on the analysis of 2D seismic reflection profiles (Vail et al. 1977). The approach that was used involved first the identification of reflection terminations (e.g. onlap, downlap, toplap, erosional truncation) and the recognition of stratigraphic discontinuities such as unconformities. Second, the reflection geometries between discontinuity surfaces were described (e.g., oblique or sigmoidal progradation). Finally, the amplitude, continuity and frequency of reflections were described and mapped. In sum, these observations yielded insights with regard to the type of depositional systems present. This approach was referred to as seismic stratigraphy (Vail et al. 1977). With the development of 3D seismic acquisition techniques, the opportunity to image geological features in map view opened up new approaches to geological prediction (e.g. Weimer & Davis 1996). Various reflection attributes such as amplitude, dip magnitude, dip azimuth, time/depth structure and curvature, to name a few, can be observed to yield direct images of depositionally and structurally significant features. In addition, analysis of seismic intervals can lend further insight to such features. The study of depositional systems using 3Dseismic derived images has been referred to as seismic geomorphology (Posamentier 2000). This represents a significant step change in how seismic interpreters evaluate 3D seismic data. In general, depositional environments had commonly been inferred on the basis of cross-section derived stratigraphic architecture and subsequent mapping of seismic facies leading to lithologic predictions. With the advent of seismic geomorphology, discrete, detailed depositional subenvironments and depositional elements could be interpreted directly from map view images leading to much more accurate understanding of lithologic distribution patterns and enhanced prediction of the distribution of reservoir, source and seal facies. The following discussion will be divided into two parts, the first section illustrating examples of seismic images of depositional elements at exploration depths, and the second illustrating images of depositional elements at shallow depths.
Depositional elements at exploration depths Cretaceous channels--Alberta, Canada Figure 1 illustrates two views of a major channel crosscut by two lesser channels. Figure 1A is a horizon slice or flattened
time slice, whereby a reflection 32 ms above was interpreted and used as a reference horizon for the purpose of slicing through the 3D seismic volume. Figure 1B is a reflection amplitude map of reflections immediately below the reflection associated with the channel. Each images the channels in a different way, with different details brought out by the two display styles. Both show linear features within the large channel, which can be interpreted as possible point bar deposits. Both show a crosscutting and therefore younger channel in the middle of the illustration. However Figure 1A shows another smaller channel crosscutting the larger channel towards the bottom of the illustration, not apparent in Figure lB. The integration of seismic geomorphology and seismic stratigraphy is illustrated in Figure 2. Inclined reflections within the interpreted channel fill can be observed on the reflection profile oriented normal to the long axis of the large channel (Fig. 2B). These reflections can be interpreted to represent lateral accretion surfaces associated with point bar deposition within the channel (Figs 2C and D). The isopach map indicates the presence of a thicker channel fill on the southwestern side of the channel (Figs 2A and D). The seismic profile reveals that the thicker part of the channel does not correspond to a deeper channel thalweg, but rather is associated with a 'bump' across part of the channel. This 'bump' is interpreted to be associated with a substrate that is less compactible than the other part of the channel fill (Fig. 2C). This least compactible section would suggest the presence of lateral accretion sets that would be most sand-rich, sand being less compactible than silt or shale (Fig. 2D). Planning of horizontal well bore trajectories should take into account the presence of internal stratigraphic architecture comprising varying lithologies (Fig. 3). In this instance, orientation of horizontal well bores parallel to the lateral accretion deposits would allow for improved reservoir management. Lithologic variations associated with bedding parallel boreholes would be lower than those associated with bedding normal boreholes. Consequently, drilling parallel to bedding planes might better protect against gas or water breakthrough. Alternatively, if gas or water breakthrough is not a concern, then a preferred strategy might be to drill across bedding planes so as to access and drain multiple compartments with a single borehole. Several crosscutting Cretaceous-aged channels are illustrated in Figure 4. This image represents a map of the negative polarity total amplitudes within a 16 ms window that contains at
DAVIES,R. J., CARTWRIGHT,J. A., STEWART,S. A., LAPPIN,M. & UNDERHILL,J. R. (eds) 2004.3D Seismic Technology: Application to the Exploration of Sedimenta~ Basins. Geological Society, London, Memoirs, 29, 11-24. 0435-4052/04/$15 9 The Geological Society of London 2004.
12
H.W. POSAMENTIER
Fig. 1. Cretaceous fluvial/estuarine channel. (A) map of the amplitude of the reflection peak immediately underlying the top of the channel. (B) map of an amplitude extract at a level 32ms below a regional horizon (i.e., horizon slice). Two younger crosscutting channels can be observed. The linear pattern within the channel (arrow) suggests the presence of lateral accretion in association with the development of alternate or point bars.
least eight generations of crosscutting channels. Figure 5 illustrates two additional images of this same geological section; note that each image brings out different aspects of these channels. Figure 6 illustrates section views through some of these channels. Note that interpretation of such profiles alone, in the absence of 3D seismic coverage would have yielded a significantly inferior geological interpretation. The presence of lateral accretion deposits, clearly imaged on the map view image are only dimly recognizable as such on the reflection profile. Nonetheless, the integration of the map view with the section view images yields a more robust geological interpretation, which ultimately can be applied to exploration and development issues. Fluvial systems characterized by high-sinuosity channel belts are illustrated in Figures 7 and 8. The concentric arcs imaged in map view represent sections through point bar deposits and may represent scroll bars. Figure 7 illustrates an analogous modem feature from the Mississippi floodplain for comparison. Examination of the reflection profile shown in Figure 8 illustrates a stratigraphic representation of such deposits; interpretation of the correct depositional element would likely not have been possible if only the reflection profile were available. Fluvial systems overlying a major unconformity surface are illustrated in Figures 9-11, In this instance, Cretaceous fluvial channel fill deposits directly overlie Mississippian-aged car-
bonates. Figure 9 shows several co-rendered horizon attributes as well as a seismic profile illustrating the stratigraphic discontinuity between Cretaceous-aged and Mississippianaged deposits. Each image portrays the depositional elements somewhat differently. Co-rendering of different attributes also can serve to enhance the features in question (Fig. 9). In certain instances perspective views can provide a deeper appreciation for the 'lay of the land' (Figs 10 and 11). Note the apparent dendritic drainage pattern off the highland area at the fight side of Figure 10.
Shallow-marine shelf ridges--offshore northwest Java, Indonesia Numerous linear seismic reflection amplitude anomalies are observed on horizon slices in the Miocene section offshore northwest Java, Indonesia (Figs 12 and 13). These features have been interpreted as tidal current related shallow marine shelf ridges (Posamentier 2002a). The well-log cross section (Fig. 14) illustrates an abrupt sandstone pinchout towards the west, in the inferred direction of wave migration. This pinchout is expressed seismically as a sharp linear boundary (Fig. 13). In contrast the trailing edges of these shelf ridges are expressed as less welldefined amplitude changes (Fig. 13). Posamentier (2002a) has shown that these pinchouts can define a stratigraphic trapping component.
SEISMIC GEOMORPHOLOGY
13
Fig. 2. (A) Isopach map of channel fill for same channel shown in Figure 1. Blue colour indicates maximum thickness of 31 m; red colour indicates zero thickness. (B) Seismic reflection profile oriented transverse to channel axis. Irregularity of upper bounding surface indicates effects of differential compaction of channel fill. (C) Geological interpretation of seismic reflection profile shown in (A). Note presence of lateral accretion surfaces. (D) Illustrates de-compacted transverse profile of channel fill. Least compactible lateral accretion wedges are highlighted in blue.
Basement Alberta, Canada In certain instances, where basement reflections are well defined, subcrop seismic expression can provide significant insight with regard to b a s e m e n t lithologies. Figure 15
Fig. 3. Amplitude of seismic reflection at upper bounding surface of channel fill shown in Figs 1 and 2 draped onto perspective view of base channel seismic reflection. Approximate location of horizontal borehole trajectories are shown.
illustrates subcrop seismic amplitude expression indicative of likely metamorphic basement. These horizon slices and a m p l i t u d e e x t r a c t i o n s s h o w n here illustrate styles of d e f o r m a t i o n that c o m m o n l y characterize m e t a m o r p h i c terrains.
14
H.W. POSAMENTIER
Fig. 4. Negative polarity total amplitude of 16ms thick seismic interval bracketing a fluvial/deltaic depositional environment. Crosscutting channels give indication of temporal relationships; eight discrete levels of channels can be observed.
Depositional elements at shallow burial depths The study of seismic data within uppermost stratigraphic sections (i.e. within the upper 0.5 to 1.5 seconds of data) can yield significant insight to preserved depositional elements in both shallow and deep depositional environments (Posamentier
2000; Posamentier et al. 2000). These potentially well-imaged features can serve as useful analogues for deeper exploration and development targets, where similar depositional elements are known to exist. Both deep-water as well as shallow water depositional environments can be analyzed this way.
Fig. 5. (A) Horizon slice through upper part of 16 ms interval shown in Figure 4. Channel in centre of image characterized by northward-directed lateral accretion suggesting paleo-flow direction from southeast to northwest. (B) Cumulative amplitude map of same 16 ms interval shown in Figure 4. Hydrocarbon and lithological effects are accentuated.
SEISMIC GEOMORPHOLOGY
15
Fig. 6. Two transverse seismic profiles across channels shown in Figure 5A. Note the 'shingled" stratigraphic expression of point bar deposits observed in Figure 5A.
Fig. 7. Horizon slice through non-marine section illustrating seismic geomorphologic expression of meander loops. Inset illustrates Mississippi River modem analogue.
16
H.W. POSAMENTIER
Fig. 8. Horizon slice through non-marine section illustrating seismic geomorphologic and stratigraphic expression of meander loops in both cross section (A) as well as plan view (B).
Fig. 9. Several images of unconformity separating Cretaceous-aged from Mississippian aged deposits. (A), (B) and (C) illustrate seismic geomorphological expression of this surface, whereas (D) illustrates the seismic stratigraphic expression of the same surface. This unconformity surface is characterized by the presence of numerous channels evident both in map as well as section view. (A) Co-rendered dip magnitude and time structure. (B) Co-rendered dip azimuth and time structure. (C) Time structure. (D) Seismic profile.
SEISMIC GEOMORPHOLOGY
17
Fig. 10. Perspective view of unconformity separating Cretaceous-aged from Mississippian aged deposits, shown in Figure 9.
Fig. 11. Perspective view of unconformity separating Cretaceous-aged from Mississippian aged deposits. High sinuosity incised channel can be observed on this surface.
Deep-water depositional environments Shallow-buried, deep-water deposits can be readily imaged in great detail. Such detailed images provide useful analogues for more deeply buried systems. Figure 16 shows a moderate to high sinuosity channel deposited on the basin floor during the late Pleistocene (Posamentier et al. 2000). The upper bounding surface of this channel-levee system lies approximately 8 0 100 m below the sea floor. Two attributes of this surface are shown: dip azimuth (Fig. 16A) and dip magnitude (Fig. 16B). The dip azimuth map has the appearance of a shaded relief map and from it the various geomorphic elements, such as the channel, the levee crests, and overbank sediment wave fields can be interpreted. The dip magnitude map accentuates those features that are characterized by steeper slopes. In this instance, the dip magnitude map can be used to identify the larger sediment waves that lie adjacent to outer meander bends (Fig. 16B). From an exploration perspective, the sediment wave fields that are characterized by steeper flanks are inferred to be more sand prone than those with more gentle slopes. Moreover, the distribution of these sandstones will have a preferred orientation, i.e. parallel to the waves' long axes, a characteristic potentially important from a field development perspective.
Fig. 12. Amplitude extraction from seismic horizon slice illustrating linear trending shelf ridges of Miocene age, offshore northwest Java, Indonesia.
18
H.W. POSAMENTIER
Fig. 13. Amplitude extraction from seismic horizon slice illustrating shelf ridge of Miocene age, offshore northwest Java, Indonesia. Ridge migration direction is inferred to be towards the southwest. The leading edge is characterized by a sharp linear boundary whereas the trailing edge boundary is less well defined. The channel feature that appears on this image lies at a lower stratigraphic level (see Fig. 14).
The evolution of sinuous turbidity flow channels can be observed in Figure 17. This horizon slice illustrates the p r o g r e s s i v e d o w n - s y s t e m m e a n d e r loop m i g r a t i o n (i.e. 'sweep') as well as a minor degree of meander loop expansion (i.e. ' s w i n g ' ) that c o m m o n l y characterizes such systems (Peakall et al. 2000).
Another high-sinuosity channel-levee system is shown in Figure 18. The 3D perspective view (Figs 18A and C) as well as the seismic profile (Fig. 18B) illustrate the effects of differential compaction in this type of environment. The presence of relatively less compactable sand within the channel results in an inversion of topography after deposition. The top of the channel
Fig. 14. Well log cross section and seismic profiles across the shelf ridge shown in Figure 13. The sand pinches out abruptly at the leading edge, between wells 2 and 3. The locations of the well log cross section and the seismic profiles are shown in Figure 13.
SEISMIC GEOMORPHOLOGY
Fig. 15. Two images of basement in the western Canada sedimentary basin, Alberta, Canada. (A) Horizon slice and (B) amplitude extraction from basement reflection.
Fig. 16. Dip azimuth (A) and dip magnitude (B) maps of top of deep-water channel-levee complex offshore Borneo, Indonesia. The dip magnitude image illustrates a simulated relief map with lighting from the north. Numerous overbank sediment waves on either side of the channel can be observed. The dip magnitude map highlights the higher relief, steeper-flanked sediment waves.
19
20
H.W. POSAMENTIER
Fig. 17. Horizon slice amplitude extraction illustrating a high-sinuosity deep-water leveed channel. This channel is characterized by extensive meander loop expansion as well as down-system meander loop migration.
fill likely was somewhat lower than the surrounding floodplain at the time of deposition; however, because of the greater sand content of the channel fill relative to the adjacent floodplain, the floodplain compacted more and resulted in the development of a post-depositional 'channel ridge'.
A linked shelf edge and deep-water environment is illustrated in Figure 19. The dip-oriented seismic profile (Fig. 20) illustrates a shelf edge deltaic system characterized by both forced regression as well as normal regression. The transverse-oriented seismic profiles (Fig. 21) illustrate the
Fig. 18. (A) Three-dimensional perspective view of deep-water leveed channel system, eastern Gulf of Mexico. (B) Close-up of leveed channel showing effects of differential compaction expressed as channel ridge form. (C) Seismic reflection profile transverse to leveed channel flow direction. Channel evolution is characterized by aggradation coupled with lateral migration. The channel fill is characterized by high-amplitude reflections and the post-compaction expression of the channel is that of a ridge form.
SEISMIC GEOMORPHOLOGY
21
Fig. 19. Three-dimensional perspective view of the outer shelf and upper slope on a surface 80 to 100 m below the sea floor, offshore Louisiana, Gulf of Mexico. The shelf edge and upper slope are shown, The slope is traversed by a channel, which extends inboard of the shelf edge. This inboard extension becomes evident upon examination of the time slice inset. The seaward bulge of the shelf coincides with the position of the slope channel suggesting a genetic link between shelf edge depocenter (i.e. shelf edge delta) and slope channel. The grooves oriented parallel to dip likely were formed by erosion associated with mass transport processes.
Fig. 20. Dip-oriented seismic reflection profile across the shelf edge delta. The basal part of the delta is characterized by a downstepping delta plain suggesting forced regression and falling relative sea level, and overlain by an aggradational phase in the upper part of the delta suggesting normal regression and rising sea level.
22
H.W. POSAMENTIER
Fig. 21. Transverse seismic reflection profiles across leveed channel imaged in Figure 19. Levee construction can be observed. The grooves on the surface at the base of the levees are annotated on Line 2. The channel base is eroded deeply into the precursor substrate. channel-levee system that overlies the surface shown in Figure 19. The grooves at the base of these levee deposits likely represent mega-tool marks associated with the passage of slides or debris flows across this surface at the onset of a lowstand depositional episode (Posamentier 2002b). The thickness of these levees is as much as twice as great on the fight bank (facing down-system) than on the left (Fig. 22) possibly due to the Coriolis force and/or to the Gulf of Mexico loop current.
Shallow-water depositional environments Shallow-buried shelfal deposits such as shelf edge slump scars and channels, as well as incised valleys, provide useful insights as to what constitute reasonable scales for such features and also provide insights as to which depositional elements tend to be associated with each other. Figure 23 represents a horizon slice at the shelf edge offshore northwest Java, Indonesia. Welldeveloped slump scars characterize the shelf edge. In addition,
Fig. 22. Isochron map of levee associated with channel imaged in Figure 19. The levee thickness is significantly greater on the west bank, possibly caused by Coriolis force or Gulf of Mexico loop currents.
SEISMIC GEOMORPHOLOGY
23
Fig. 23. Seismic horizon slice illustrating outer shelf and upper slope depositional elements offshore Java, Indonesia. Slump scars mark the shelf edge. The inset detail of these slump scars shows incipient slumps located there as well. A small fluvial channel can be observed on the outer shelf. just inboard of the shelf edge a moderate sinuosity fluvial channel can be observed. On the mid- to inner shelf offshore Java, well developed incised valleys can be observed (Fig. 24). A distinguishing attribute of incised valleys is the presence of associated small tributary incised feeders to the principal channel (Posamentier & Allen 1999). These incised feeder channels suggest the presence of well-drained interfluves that lie above the reach of the river within the trunk valley, even when the trunk fiver is in flood.
Conclusions Depositional elements can be observed in plan view images extracted from 3D seismic volumes. The analysis of these
Fig. 24. Incised valley complex offshore northwest Java, Indonesia (Posamentier 2001). Inset details show channel bars, tributary incised valleys and fluvial terraces.
features constitutes the study of seismic geomorphology. These observations can provide direct as well as indirect benefits to exploration and field development. Where depositional elements can be observed directly at exploration depths, the presence of reservoir, reservoir source, and seal facies can be modelled more accurately. Moreover, the occurrence of stratigraphically defined compartments as well as the potential for stratigraphic trapping of hydrocarbons can be evaluated within the context of the depositional elements identified. Both exploration as well as field development can benefit directly from such analyses. The indirect benefit derived from seismic geomorphologic analyses derives from examination of shallow-buried features. Shallow-buried features commonly are significantly better imaged than their more deeply buried counterparts. Seismic
24
H.W. POSAMENTIER
geomorphological analyses of such well-imaged features provide a clearer understanding of depositional element distribution and therefore more accurate prediction of reservoir, source and seal facies. In addition, these analyses enable geoscientists to better evaluate the preservation potential of different depositional elements and therefore the likelihood of encountering such deposits in the ancient rock record. In addition, well-imaged, shallow-buried depositional elements provide 'reality checks' for scale of various such features as well as for depositional element associations. I thank Anadarko Canada Corporation for permission to publish this paper. Thanks are due Western Geco for permission to publish the seismic data shown in Figures 12-14, 16, 23-24, and to Veritas Exploration Services for permission to publish the seismic data shown in Figures 19-22. In addition I would like to acknowledge the support of R. Evans for his help with data management and interpretation, as well as the user support group at Paradigm Geophysical for their patience and support with regard to my never ending questions regarding the Stratimagic interpretation application. Reviews by T. Garfield, H. Johnson and J. Cartwright were appreciated and helped me in preparation of the final version of this paper.
References PEAKALL, J., MCCAFFREY, W. D, & KNEELER, B. 2000. A process model for the evolution, morphology and architecture of sinuous submarine channels. Journal of Sedimentary Research, 70, 434-448. POSAMENTIER, H. W. 2000. Seismic stratigraphy into the next millennium; a focus on 3D seismic data. American Association of
Petroleum Geologists Annual Conference, New Orleans, LA, April 16-19, 2000, All8. POSAMENTIER, H. W. 2001. Lowstand alluvial bypass systems: incised vs. unincised. AAPG Bulletin, 85, 1771-1793. POSAMENTIER, H. W. 2002a. Ancient shelf ridges--a potentially significant component of the transgressive systems tract: case study from offshore northwest Java. AAPG Bulletin, 86, 75-106. POSAMENTIER, H. W. 2002b. 3-D seismic geomorphology and stratigraphy of deep-water debris flows. American Association of Petroleum Geologists Annual Conference, Houston, TX, March 10-13, 2002, 142. POSAMENTIER. H. W. & ALLEN, G. P. 1999. Siliciclastic sequence stratigraphy---concepts and applications. Society of Economic Paleontologists and Mineralogists Concepts in Sedimentology and Paleontology, 7, 210p. POSAMENTIER, H. W., MEIZARWIN,WISMAN, P. S. & PLAWMAN, T. 2000. Deep water depositional systems--Ultra-deep Makassar Strait, Indonesia. In: WEIMER, P., SLATT, R. M., COLEMAN, J., ROSEN, N. C., NELSON, H., BOUMA, A. H., STYZEN, M. J. & LAWRENCE, D. T. (eds) Deep-Water Reservoirs of the World, Gulf Coast Section Socie~ of Economic Paleontologists and Mineralogists Foundation 20th Annual Research Conference, 806-816. VALE, P. R., MITCHUM, R. M. JR. & THOMPSON, S. III 1977. Seismic stratigraphy and global changes of sea level, part 3: relative changes of sea level from coastal onlap. In: PAYTON, C. E. (ed.) Seismic Stratigraphy--Applications to Hydrocarbon Exploration. American Association of Petroleum Geologists Memoir, 26, 63-81. WEIMER, P. & DAVIS, T. L. 1996. Applications of 3-D Seismic Data to Exploration and Production, American Association of Petroleum Geologists Studies in Geology, 42, SEG Geophysical Development Series, No. 5.
Depositional architectures of Recent deepwater deposits in the Kutei Basin, East Kalimantan J. N. F O W L E R
l, E . G U R I T N O ,
P. S H E R W O O D G. G O F F E Y
2, M . J. S M I T H
l, S. A L G A R ,
I. B U S O N O ,
3 & A. STRONG
tEni-LASMO Indonesia, 29th Floor Ratu Plaza Office Tower, Jl Jenederal Sudirman, Jakarta, Indonesia Currently at: Eni-London Technical Exchange, Bowater House, 68 Knightsbridge, London S W I X 7BN, UK (e-mail:
[email protected]) 20ccidental Middle East Development Co., 29th Floor, Emirates Office Tower, Sheikh Zaved Road, P.O. Box 33728, Dubai, U.A.E 3paladin Resources Plc, Princes House, 38 Jermyn Street, London SW1 Y 6DN, UK
Abstract: To aid exploration and appraisal of hydrocarbon discoveries in deepwater deposits of the Kutei Basin, a study of analogous sedimentaryarchitecturesin Recent deposits of the same basin was undertaken. High quality 3D seismic were used to develop an understandingof the external and internal geometry of slope to basin floor elements in a structured setting. Toethrust anticlines and related mud diapirs deflect slope canyons. Over slope-steps, gravity flow deposits are laterally confined with narrow facies belts. In slope mini-basins, flows are less confined resulting in deposition over a broad area. The Recent deposits of a single canyon and associated basin floor system are used to illustrate the deepwater depositional elements. Debrites at the base are followed by a slope channel complex or basin floor fan then a channel-levee complex. Large depocentres occur where gradients are low and the system switches from confined to unconfined. Erosionally confined channels feed basin floor fans at the toe-of-slope,while channels confined by levees feed fans on the "distal' basin floor. Slope channel complexes and basin floor fans are interpreted to be sand prone. From the slope to basin floor these deposits increase in width:thickness ratio and areal extent and apparent lateral connectivity increases while vertical connectivity decreases.
A number of recent hydrocarbon discoveries have been made in the deepwater Kutei Basin, in Mio-Pliocene slope to basin floor sediments. Exploration and appraisal of these discoveries is assisted by a comparative analysis of analogous sedimentary architectures in the Recent deposits of the same basin. The analysis of Recent deposits using high quality 3D seismic data enables the development of basin-specific depositional models, thus complementing the use of models based on data from basins with different tectonic settings, sediment type and other external controls. The method increases understanding of deepwater environments per se and the understanding of the basin of economic interest. Analysis and documentation of near surface deepwater deposits using high-quality 3D seismic data has increased over the last year, for example Beaubouef & Friedman (2000) and Demyttenaere et al. (2000). Posamentier (2001) and Posamentier et al. (2000) present spectacular images of Recent deposits in the deepwater environment using time slices, flattened time slices, and interval attributes combined with near surface seismic crosssections; these examples are dominated by basin floor facies, whilst this paper presents data from both the slope and basin floor. This study set out to develop a series of seismically derived architecture models for the Recent deposits of the northern Kutei Basin, East Kalimantan, Indonesia (Fig. 1). The models were used as analogues to constrain geometries, aspect ratios (width:thickness ratio) and potential net-to-gross variations for the prospective stratigraphically older sediments. However, the application of Recent models is only valid if similarities with the prospective section are demonstrated and understood. Architectural similarity between Recent and calibrated Mio-Pliocene examples from the Kutei Basin is demonstrated in Figure 2. Sherwood et al. (2001) also show analogous slope and basin floor facies in the Kutei Basin by comparing the Recent deposits with those of the Miocene and Pliocene. Seafloor images in this study are derived from 3D seismic data. Using dip attribute displays and 3D visualization, the complex interplay between structure and deepwater sedimentation is demonstrated along a slope-basin profile impinged by
toe-thrust anticlines. For example, the effect of flow impediments such as toe-thrust related anticlines and mud diapirs had a profound impact on slope canyon morphology. A series of geoseismic sections through a single slope-canyon to basin floor setting illustrate the variation in architectural elements and potential reservoir facies both vertically and down system. These observations are summarized with a series of geological models for deposits on a structured slope and basin floor during periods of relative fall and relative rise in sea level. The near-surface 3D seismic used in the study is of high quality, having a peak frequency of 45 Hz, giving a vertical resolution of approximately 11 m. The major limitation of this near-seafloor dataset is the lack of calibration. Lithology is therefore largely inferred from internal and external seismic geometries and through comparison with analogous calibrated deposits in the prospective Mio-Pliocene section (Figure 2; Sherwood et al. 2001).
Deepwater geomorphology and process Features visible on the present day seabed illustrate the complexity of the slope and basin floor. A dip attribute map of the seabed derived from 3D seismic data shows a variety of features (Fig. 1). The most dominant features are the numerous slope canyons and the surface expression of toe-thrust anticlines. The canyons vary considerably in width (0.7-2.8knl), depth (30-400 m) and geometry. This paper will show that canyon geometry is linked to the slope profile, which in turn is affected by the surface expression of active toe-thrust anticlines and associated mud diapirism. Also visible near the toe-of-slope and basin floor are sediment waves that occur both on the outer bends of channels and as 'fields' at the mouths of canyon. Meandering channels and a debrite zone are also present on the basin floor.
Flow impediments In the north of the study area there are a number of large canyons with obvious deflections to their paths (Fig. 1). The deflections
DAVIES,R. J., CARTWRIGHT,J. A., STEWART,S. A., LAPPIN,M. & UNDERHILL,J. R. (eds) 2004.3D Seismic Technology:Applicationto the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 25-33.0435-4052/04/$15 9 The Geological Society of London 2004.
26
J.N. FOWLER ET AL. The alternating pattern is a reflection of the dominant process along the slope profile; erosion and transportation dominate narrow canyons whilst deposition is dominant in broader areas.
Canyon wasting Canyon growth, by an increase in depth and width, seems to occur through a combination of sidewall wastage and removal of material by gravity flows. The inter-canyon areas are composed of interbedded mudstones, siltstones and fine-grained sandstones, where penetrated by wells. The high proportion of mud layers makes them inherently unstable in areas of steep slopes. Wastage of canyon walls, in the form of slumps, is clearly seen in map view through the identification of arcuate fault scarps running parallel to the canyon axis (Fig. 3). In cross-sectional view, down-to-canyon faults and a thickened slump-toe characterize recent slumps visible on the seabed (Fig. 3). In near-surface examples only the down-to-canyon faults are preserved. Presumably in-canyon erosional processes have removed the slump-toe. The slumping is driven by a combination of deep-seated toethrusting that drives anticlinal uplift resulting in arc stretching and failure of the overburden, and through erosion and undercutting of canyon walls by gravity flows.
Sediment waves
Fig. 1. A dip attribute map of the seabed in the study area. The image was created by merging data from two 3D seismic data sets, shown on the inset map. Highlighted are structure and interpreted depositional elements of the deepwater environment. The inset seismic section is through a sediment wave field that has formed at the mouth of a canyon. The blue arrow represents currents emanating from the canyon and the red arrow the apparent growth of the waves.
are caused by eastward-verging anticlines generated by toethrusts and by associated active mud diapirs/mud volcanoes. The deflection of flows by these structures affects both the distribution of deposits on the slope and the location of entry points onto the basin floor. Anticlines associated with mud diapirs appear to be a major cause of slope-canyon deflection and bifurcation. There are a few examples where a canyon is deflected around the nose of a simple anticline (Fig. 3). The mud diapirs are associated with underlying thrust planes, commonly best developed at their lateral tips. Toe-thrust anticlines and mud diapirs appear to increase the tortuosity of a slope canyon's path, but the canyons are ultimately through-going to the basin floor.
Slope profile and canyon geometry The dip attribute display of the seabed (Fig. 1) reveals an alternating pattern of well-defined and poorly defined canyon walls along a single canyon profile. This pattern is a function of the slope profile; where there is an increase in gradient (slopestep), the canyon has well-defined walls and the canyon is narrow. Where there is a decrease in gradient (slope mini-basin), the canyon walls are poorly defined and the canyon broadens (Fig. 3).
Sediment waves occurring on levees are observed and documented in the east of the study area by Posamentier et al. (2000). Similar sediment waves are found elsewhere in the study area around the outerbends of sinuous c h a n n e l - l e v e e complexes. Of interest is another category of sediment waves observed at the mouths of canyons (Fig. 1). These bedforms have crests arranged perpendicularly to the axis of the canyon with a wavelength of approximately 1200 m and amplitudes of up to 50m. Two sediment wave fields have been identified associated with canyons that have been active recently (activity is identified by erosion of the recent drape by channels). Like sediment waves that occur on the outer bends of channels, they are asymmetric and appear to migrate in an up-current direction (Fig. 1). The lithology of the sediment waves in the study area is unknown (those on levees and as fields proximal to active canyons), but comparisons can be drawn with published examples. A study by Migeon et al. (2001) presented calibrated examples of sediment waves on the Var sedimentary ridge. The sediment waves are asymmetric, the result of deposition on the upstream side and lower deposition rates and erosion on the downstream side, with sand preferentially deposited on the upstream side. It is possible that the sediment wave fields were generated by contour currents (Mutti 1992), but the orientations of their crests varies significantly from field to field, suggesting a different formative process. It is suggested here that currents emanating from the canyons rework material out of the canyons and onto the basin floor, resulting in wave fields with crests arranged perpendicularly to the axis of the canyon. These currents may also redistribute turbidite sand on the basin floor in the same way that contour currents are believed to transport and deposit sand in the deepwater environment (Picketing et al. 1989).
Depositional elements The depositional elements described below form the building blocks of this deepwater environment. Each element has a diagnostic seismic facies and in some cases characteristic upper
RECENT DEEPWATER DEPOSITS OF THE KUTEI BASIN
Fig. 2. Matrix displaying deepwater depositional elements from the M i o Pliocene and Recent of the Kutei Basin. Mio-Pliocene examples are calibrated using well data (displayed curves are gamma ray and induction log).
Fig. 3. 3D image showing deflection of slope-canyons by toe-thrust anticlines and associated mud diapirs (yellow arrows), and narrowing of the slopecanyon over the slope-step (blue arrow). Arcuate fault scarps running parallel to the canyon are highlighted on the inset 3D image (dashed red lines). The slumps have formed through a combination of arc stretching associated with anticlinal uplift and bank undercutting due to gravity flows. The near surface seismic data illustrates rotated packages of intercanyon sediments bound by down-tocanyon faults.
27
28
J.N. FOWLER ET AL.
and lower surfaces. The diagnostic properties of each element are presented together with an interpretation of the sedimentary facies present and their formative processes.
Debrites Debrites are recognized on seismic data by their low-amplitude, low-frequency, chaotic/mottled seismic facies and a strongly erosive base (Fig. 2). These deposits are interpreted as debris flow deposits and are interpreted not to be sand prone. The debrites in the Kutei Basin are analogous to the mass transport complexes of the Plio-Pleistocene Mississippi Fan (Weimer 1990). Where observed, on surface displays of dip attribute and amplitude, the bases of the debrites are grooved or striated. The striated bases have been interpreted by Posamentier (2001) as scours formed by blocks of semi-lithified material entrained at the base of the flow. Alternatively, the grooved base may relate to variations in flow thickness, thereby altering the erosive capacity of the flow. Debrites are ubiquitous in the study area along the basin profile where there is increased slope instability probably caused by toe-thrusting and/or frequent falls in relative sea level. Single debris flows are found to cover large areas (up to 110km 2) with considerable thickness variations parallel to the transport direction. Debrites are seen to thicken into slope mini-basins and thin over slope-steps. This thickness variation occurs over a number of toe-thrust anticlines within a single debrite unit. Similar thickness changes over pre-existing highs have been observed in mass transport complexes of the Texas continental slope, Western Gulf of Mexico (Beaubouef & Friedman 2000).
of interbedded mudstone, siltstone and fine-grained sandstone. Deposition in the inter-canyon area is interpreted to have occurred by pelagic and hemipelagic fallout and low-density turbidity current sedimentation. The high-amplitude seismic events are interpreted to be the result of density differences between different mudstones and fine-grained sandstones. Erosional surfaces are evident but they commonly disappear up-system and are interpreted as slump scars. Inter-canyon reflectors often dip into the canyon through fault block rotation resulting from canyon wasting.
Slope channel complexes Slope channels have a low width:thickness ratio, have strongly erosive bases and the channel fill has distinct lateral terminations displaying medium to high amplitudes (Fig. 2). Such deposits are interpreted as products of medium- to high-density turbidity currents, thus inferring a sand-prone nature. The scheme used by Campion et al. (2000) to describe slope channel deposits is adapted for use here. The term channel is used to describe the smallest architectural feature identifiable using 3D seismic data. The term 'channel complex' is used to describe a body of channels that generally share a common erosion surface. Within the observed channel complexes there are a number of stacking patterns, from vertically stacked to lateral offset stacking. Channels also range from amalgamated, an individual channel which is in contact with a similar channel, to isolated. The channel complex deposits are confined by the canyon's master erosion surface. The walls of a canyon and a corresponding basal erosional surface define the master erosion surface.
Inter-canyon
Passive channel fill
Layer parallel reflections of contrasting high amplitude and low amplitude events characterize the inter-canyon deposits (Fig. 4b). Well penetrations of inter-canyon deposits confirm the presence
A passive channel fill is composed of low- to mediumamplitude, laterally continuous reflections that are either fiat or drape underlying topography (Fig. 8). This phase of channel
Fig. 4. Seismic sections of near-surface deposits, corresponding interpretation and a true scale equivalent with the vertical axis approximately equal to the horizontal: (A) middle slope-step, (B) middle slope mini-basin and (C) the lower slope-step. The red line represents the master erosion surface. All diagrams displayed at the same scale and are taken from the slope canyon marked A in Figure 1.
RECENT DEEPWATER DEPOSITS OF THE KUTEI BASIN
29
Fig. 5. Seismic sections of near surface deposits, corresponding interpretation and a true scale equivalent with the vertical axis approximately equal to the horizontal: (A) lower slope mini-basin, (B) basin floor through the proximal fan complex and (C) basin floor through the median fan complex. The red line represents the master erosion surface. Note the difference in the vertical and horizontal scales of each figure. Examples are taken from the slope canyon marked A in Figure 1.
fill is interpreted to be mud prone, thus inferring deposition by pelagic and hemipelagic fallout and low-density turbidity currents. Passive channel fill is also observed within slope channel complexes taking the form of small channels or amorphous bodies, often with no internal reflectors (Fig. 4b). These features are interpreted as mud deposits 'plugging' channels.
Slope fans Slope fans have a mounded appearance on seismic data. Internally they are characterized by continuous, high amplitude, convex-up reflectors showing bi-directional downlap (Fig. 2). In the Recent example illustrated, the individual reflectors are 0.5 to 1 km in width and are interpreted as sand prone (Fig. 2). Fanlike bodies of sediment are not readily identifiable on the slope and toe-of-slope because of the complex, cross-cutting caused by prevalent channelization.
Basin floor fans Basin floor fans in the study area are composed of medium- to high-amplitude reflectors that are dominantly continuous (Fig. 2). Individual reflections display convex upward and concave upward patterns. The latter are common and can be shown to comprise the distributary channel network of the fan complex. Basin floor fans in the study area occur where the confined system switches to unconfined immediately outboard of the leading toe-thrust anticline and are interpreted as dominantly sand prone elements.
Channel-levee complexes Individual levees are wedge shaped, thinning away from the channel, and are characterized by low amplitude and often lowfrequency events that are continuous to discontinuous (Fig. 2). A
well penetration of Pliocene levee deposits indicates a mudstone dominated lithology, probably deposited by the low-density portion of turbidity currents. Occasional high-amplitude events are present on seismic data, which may represent crevasse splays. When levee pairs are visible they have a characteristic 'gull-wing' geometry. The channel deposits are defined by medium- to highamplitude reflectors that are laterally discontinuous with flat to erosional bases (Figs 2 & 5b). The leveed-channels are interpreted to be sand prone.
Drape There is a drape of sediments across much of the study area, which is up to 100m thick. It is composed of low-amplitude laterally continuous reflections that are interpreted as pelagic and hemipelagic fallout deposits (Fig. 6). The drape covers the entire study area and is associated with the Recent transgression and subsequent shut-off of the delivery system to the deepwater. Laterally continuous drape facies cannot be identified in the Mio-Piiocene prospective section due to erosion. Erosion is pronounced when movement on the toe-thrusts disrupts the equilibrium profile (Pirmez et al. 2000). This means that drape facies cannot be used as correlation tool between mini-basins (Badalini et al. 1999).
Spatial relationship of depositional elements This section documents the changes in thickness, width and spatial relationship of depositional elements along the slope to basin floor profile in the study area. A single slope canyon and related basin floor system is selected to demonstrate these architectural changes through a series of seismic cross-sections and corresponding interpretation through slope-steps, slope mini-basins and the basin floor (Fig. 1, canyon A). The upper slope facies of the selected canyon are not covered by the 3D seismic dataset available.
30
J.N. FOWLER ETAL.
Fig. 6. Schematic block diagram illustrating depositional styles in a structured slope setting during an interpreted relative sea level fall. Yellow colouration refers to sand prone deposits. Seismic sections are displayed, perpendicular to the flow direction, and are taken from near surface deposits in the northern Kutei Basin, East Kalimantan. Note variable vertical and horizontal scales.
Middle slope-step The angle of the slope over the step is 31 m/km (1.8~ The canyon is defined by a deeply incised and narrow master erosion surface resulting in confinement of the system (Fig. 4a). Resting on the basal erosion surface are amalgamated debrites. Levees 1 and 2 and associated deposits separate the slope fans from the later channel complex. The channel complex is 100-150m thick and up to 1 km wide. The control on the depositional system by a narrow canyon is reflected in the dominant stacking pattern of the slope channel complex, which is vertically amalgamated. The build-up of channel deposits between levees 3 is modest, 3 0 - 5 0 m thick and up to 800m wide. A calibrated model of erosionally confined slope channels (Mayall & Stewart 2000) illustrates an analogous vertical sequence to that observed in the study area: (1) erosional base, (2) coarse-grained lag, (3) debrites/slumps, (4) stacked channel complex and (5) a channel-levee complex.
Middle slope-mini-basin The master erosion surface in the mini-basin is wider and has not incised as deeply as the slope step (Fig. 3b). The nature of the master erosion surface reflects the decrease in slope gradient to 24 m/kin (1.4~ Consequently, the width and areal extent of the depocentre increases, when compared to the slopestep (Fig. 4a). There are thick debrites on top of the master erosion surface at the base of the system. Incised into the debrites, the channel complex is up to 2 0 0 - 2 5 0 m thick and 1.5km wide. The dominant stacking patterns are: (1) amalgamated vertically stacked, (2) amalgamated vertically offset stacked and (3) occasionally isolated vertically stacked. The leveed-channel deposits are 50-70 m thick and up to 600 m wide. The channels illustrate lateral offset stacking and are not as erosive as the channel complex below. As a consequence, amalgamation of the leveed-channels is reduced.
Lower slope-step The lower slope-step has an angle of 36 m/km (2.1~ The intercanyon areas are less imposing and the master erosion surface exhibits incision. This surface is overlain by debrites (Fig. 4c). The debrites are overlain by a unit composed of a channel complex and slope fans that is 2.5 km wide and approximately 125-200m thick. At the base of this unit are isolated offset
stacked channels, which pass upward into amalgamated vertically stacked channels. The high-amplitude convex-up reflectors within this unit are enigmatic. They may represent slope fans that have subsequently been eroded by channels. Embedded in the channel complex are channels plugged by mud-prone sediments. A phase of passive channel fill precedes the channel-levee complex. The leveed-channel complex is 5 0 - 1 5 0 m wide and stacks up to approximately 200 m thickness. The channel fill has grown concomitantly with the levees resulting in vertical offset stacking.
Lower slope-mini-basin The lower slope mini-basin studied is a major depocentre updip of the basin floor. There is a marked reduction in gradient from the lower slope-step, 36 m/km (2.1~ to the lower slope minibasin, 14.6 m/km (0.83~ Large mini-basins at the toe-of-slope capture sediment from a number of canyons. Interpretation of such deposits is therefore complicated by complex cross-cutting relationships. Figure 5a shows the same suite of depositional elements in the same order as the middle slope, but they are spread over a wider area. The debrites are followed by a laterally extensive unit of channel complexes and slope fans, which is 5 - 6 km wide and approximately 1 2 0 - 1 5 0 m thick. Channels illustrate amalgamated laterally and nested offset stacking patterns. This unit formed through the input of two canyons. Small fans at the mouths of canyons are common in the early stages of deposition. Succeeding the fans are through-going channels, ultimately debouching onto the basin floor. Channels dominate the minibasin deposits and the lack of confinement through erosional topography led to channel migration and deposition over a broad area. The channel complex and slope fan unit is succeeded by a phase of passive channel fill. Above this are channel-levee deposits that are up to 2.5km wide and 100-150m thick. At the base of the channel-levee unit are erosive channels with a low width:thickness ratio. Passing upward, the deposits have a high width:thickness ratio and do not appear to have erosional bases. The transition from channels with a low width:thickness ratio to those with a high width:thickness ratio is interpreted as a waning of the depositional system and backfilling of the channel-levee complex with fine-grained material, possibly associated with a rise in relative sea level and/or a change in the calibre of sediment on the shelf (Kolla & Macurda 1988).
RECENT DEEPWATER DEPOSITS OF THE KUTEI BASIN
31
300
Basin floor
li[;
li ',., 1[, [[ [ ., [[
IIIll
I TF
.-.. 2 5 0
The region immediately outboard of the leading toe-thrust anticline represents the toe-of-slope to basin floor transition zone. The low gradients observed (11 m/kin or 0.6 ~ and the lack of confinement though erosional topography results in a rapid loss of energy allowing sediment dispersal over a large area, precluding significant erosion into underlying bodies (Figs 5b and 5c). Debrites are thick and areally extensive in this region; individual bodies are greater than l l 0 k m 2 in area and 120m thick. Following the debrites is a large basin floor fan that forms at the toe-of-slope that is up to 23kin wide and l l 0 m thick (areal extent is greater than 380km2). The lower boundary of the fan is regionally flat but detailed inspection reveals a lower interval composed of numerous low relief and marginally erosional channels. Similar channels with a high width:thickness ratio dominate the fan complex. Interspersed between the channels are sheet-like deposits and occasional debrites. A channel-levee complex succeeds the basin floor fan, a transition that is documented in other Recent examples (Pirmez et al. 2000; Beaubouef & Friedman 2000; Posamentier 2001). The levees are part of the same channel-levee system observed on the slope, but have diminished in thickness basinward. This channel-levee system (and others in the Recent and Pliocene section) builds out beyond the toe-of-slope supplying sediment, and possibly sand, to the 'distal' basin floor (Bouma et al. 1985). Figure 4b illustrates the confined leveed-channel deposits that are 1 - 1 . 2 k m wide and 120-150m thick. The leveedchannel deposits are erosive and amalgamated vertically and laterally. Moving basinward, Figure 5c records a thinning of the levees accompanied by a widening of the leveed-channel complex where it is 3 km wide and up to 100 m thick. The lower boundary of the complex incises into the basin floor fan. Individual leveed-channels have a higher width:thickness ratio than leveed-channels illustrated in Figure 5b but still have erosional bases and stack vertically and laterally.
Discussion Reservoir distribution and connectivity It is difficult to draw conclusions regarding reservoir quality of the deposits discussed thus far due to the limited number of well penetrations in these facies, although similarity with drilled Mio-Pliocene examples is demonstrated by Figure 2. However, the size of potential reservoir bodies, maximum thickness and areal extent, and their connectivity can be estimated, assuming that the channels and fans discussed are sand prone. The size of potential reservoir bodies is in part dependent on the slope profile. On slope-steps where the canyon is narrow and erodes deeply, the channel complexes and slope fans have low width:thickness ratios and small depositional areas because bypass processes dominate (Fig. 7). In slope mini-basins the canyon is broader and the width:thickness ratio and areal extent of depositional elements increases. In addition to variations in reservoir volume caused by changes of the slope gradient there is also a dramatic increase in volume from the slope to basin floor. Basin floor fans are an order of magnitude larger than slope channel complexes and fans (Fig. 7). Leveed-channel deposits are poorly developed on the middle-slope where there are steep gradients, but are voluminous in the lower slope mini-basin and basin floor where gradients are less.
a)
200
~ 150 x,-
K lOO
a
50 o lO
100
1000
llill 10000
I 100000
Width (metres) Fig. 7. Aspect ratio data for representative sand prone depositional elements, taken from the northern Kutei basin, East Kalimantan. Black dots refer to areal extent: ( I ) Channel complex, middleslope step (l.0km2). (2) Leveed-channel complex, middle slope-step (1.8 kin2), (3) Channel complex, middle slope mini-basin (4.77km2), (4) Leveedchannel complex, middle slope mini-basin (3.2 kin2), (5) Channel complex, lower slope-step (14.8 km2), (6) Leveed-channel complex, lower slope-step (6.6 km2). (7) Channel complex, lower slope mini-basin (31 km2), (8) Leveed-channel complex, basin floor (13 kin2), (9) Fan, basin floor (370km2).
Potential reservoir bodies on the slope-steps and slope minibasins are characterized by highly discontinuous reflectors, in a strike and dip direction. The discontinuity of reflectors is the result of intense channelization. On the slope where there is topographic confinement through erosion, channels are dominantly erosive and channel complexes are therefore amalgamated. Vertical and lateral connectivity may be good at the scale observed but will be reduced by plugging of erosive channels with clay grade material. Connectivity will also be reduced by facies variations towards the channel margins, as indicated by outcrop examples from the Stony Creek Formation of northern California where erosion surfaces at the channel margins are draped by mud and beds are not amalgamated (Campion et al. 2000). Confinement of channels by levees resulted in local concentration of erosion processes and therefore amalgamation of channel deposits at the base of the leveed-channel complex. The width:thickness ratio of leveed-channel deposits generally increases vertically (Fig. 5a). This is interpreted as an increase in the fine-grained component of the system, possibly associated with a rise in relative sea level (Mutti 1992; see below). Therefore, the vertical connectivity of the leveed-channel complex may decrease up-sequence. Connectivity of this unit will also be affected by the same degrading factors as discussed for slope channel complexes. The individual reflectors of basin floor fans are continuous and lateral connectivity appears to be excellent at the scale observed. Many of the reflectors have a broadly erosional lower contact but the degree of vertical amalgamation is reduced when compared to confined systems on the slope. Apparent vertical connectivity is therefore reduced. Connectivity, both vertical and lateral, will also be affected by lithofacies variations on the margins of distributary channels and inter-channel areas. The fields of sediment waves on the basin floor are potential reservoir facies with good apparent connectivity. Analogous sediment waves on the Var sedimentary ridge are sand prone; the high net-to-gross zone extends for 23 km along the wave crest and 2 km either side of the crest, and sand bodies appear to be interconnected (Migeon et al. 2001).
Relative sea level change Mutti (1992) attempted to predict stacking patterns of deepwater sands based on the assumption that their formation is
32
J. N, FOWLER E T AL.
controlled by relative sea level fluctuations. During maximum rates of relative sea level fall large volumes of sediment are produced. Channel-levee complexes form when relative sea level starts rising slowly and large amounts of fine grained sediment are fed into the deep-water environment. Finally, draping through pelagic and hemipelagic deposition occurs during a relative sea level high. In the study area the accumulation of large volumes of sediment on a structured slope in slope-canyons and on the basin floor is probably associated with a fall of relative sea level (Fig. 6), but the development of sand prone facies will also be dependent on the nature of sediments available on the shelf (Kolla & Macurda 1988). A thick sequence of debrites formed during the early stages of relative sea level fall. Succeeding the debrites are sand prone channel complexes and fans, leading to multiple reservoir targets on the slope to basin floor profile. The volume of sediment that accumulates on the slope and basin floor during an interpreted rise of relative sea level is reduced. Channel-levee complexes are common and the system has a reduced sand content (Posamentier 2001). In the study area continuous levees are present from the middle slope to the basin floor (Fig. 8). Channel-levee complexes are poorly developed on the middle slope, possibly due to the higher gradients (approximately 2~ but well developed in the lower slope mini-basin and basin floor, where gradients are less than 1o. Levees observed in the Kutei Basin build out beyond underlying basin floor fans that form at the toe-of-slope. Channel-levee complexes therefore shift the point at which turbidity currents switch from confined to unconfined flow further out onto the basin floor. Erosional topography confines the channels that supply basin floor fans at the toe-of-slope and levees confine channels that are able to transport sediment onto the 'distal' basin floor, to be deposited as small fans at the down dip point of levee termination. Identification of channel-levee complexes therefore shifts the focus for exploration basinward in these deposits. Continuing relative sea level rise results in backfilling of the channel systems and ultimately abandonment with storage of sediment on the shelf. During a relative sea level high the area is draped through pelagic and hemipelagic deposition. Other depositional processes appear to be ongoing during drape deposition: the development of sediment wave fields at the mouths of canyons and debrite deposition. The sediment waves
are interpreted as sand prone deposits and the currents generating these fields may play an important role in redistributing turbidite sand on the basin floor. Continuing debrite deposition is associated with slope instability caused by toethrusting. A similar process occurs in the Gulf of Mexico deepwater environment where salt diapir growth and oversteepening results in debris flows during high stands of sea level (Twichell et al. 2000). Conclusions 9
9 9
9
9
9
9
Toe-thrust anticlines affect sediment distribution on the slope by impacting slope-canyon width:thickness ratio and geometry. Over slope-steps gravity flows are laterally confined and the resultant facies belts are narrow. Flows are less confined in slope mini-basins resulting in deposition over a broad area. Large depocentres occur at points of gradient decrease, especially in slope mini-basins or on the basin floor. A number of depositional elements have been recognized and characterized to aid interpretation of deepwater provinces: debrites, inter-canyon, slope channel complexes, passive channel fill, slope fans, basin floor fans, channellevee complexes and drape. Channel complexes and slope fans dominate the sand prone deposits on the slope. Slope channel complexes are often amalgamated and appear to have good connectivity both vertically and laterally, but may deteriorate towards the channel margins. Large fans and leveed-channel deposits characterize sand prone sediments on the basin floor. Apparent connectivity of basin floor fans is excellent laterally and may be reduced vertically. Basin floor leveed-channel deposits are often amalgamated and appear to have good connectivity, which may deteriorate towards the channel margins. There is an increase in reservoir volume (width:thickness ratio and areal extent) from slope-step to slope mini-basin and from the slope to basin floor. The understanding of diagnostic geometries, vertical and lateral facies progression and degree of connectivity demonstrated by this project can assist exploration, appraisal and development of these reservoirs.
The authors thank the management of PERTAMINA, Unocal and Eni for permission to publish this paper. Geco are gratefully
Fig. 8. Schematic block diagram illustrating depositional styles in a structured slope setting during an interpreted relative sea level rise. Yellow colouration refers to sand prone deposits. Seismic sections are displayed, perpendicular to the flow direction, and are taken from near surface deposits in the northern Kutei Basin, East Kalimantan. Note variable vertical and horizontal scales.
RECENT DEEPWATER DEPOSITS OF THE KUTEI BASIN acknowledged for permission to publish their data. Thanks also to B. T. Dixon and G. C. Steffens for thorough reviews of the manuscript.
References BADALINI, G., KNELLER, B. & WINKER. C. D, 1999. Late Pleistocene Trinity-Brazos turbidite system. Depositional processes and architecture in a ponded mini-basin system, Gulf of Mexico, continental slope. Extended Abstracts Volume, American Association of Petroleum Geologists International Conference, 22-25. BEAUBOUEF, R. T. & FRIEDMAN, S. J. 2000. High resolution seismic/sequence stratigraphic framework for the evolution of Pleistocene intra slope basins, Western Gulf of Mexico: depositional models and reservoir analogs. GCSSEPM Foundation 20th Annual Research Conference, Deepwater Resen,oirs oft he WorM. 40-60. BOUMA, A. H., NORMARK, W. H. & BARNES, N. E, 1985. Mississippi fan, Gulf of Mexico. In: BOUMA, A. H., NORMARK, W. H. & BARNES, N. E. (eds) Submarine Fans and Related Turbidite Systems. Springer, Berlin, 143-156. CAMPION, K. M., SPRAGUE, A. R., MOHR1G, D., LOVELL, R. W., DRZEWIECKI,P. A., SULLIVAN,M. D., ARD1LL,J. A., JENSEN,G. N. & SICKAFOOSE, D. K. 2000. Outcrop expression of confined channel complexes. GCSSEPM Foundation 20th Annual Research Conference, Deepwater Reservoirs of the WorM, 128-150. DEMYTTENAERE, R., TROMP, J. P., IBRAHIM, A., ALLMAN-WARD.P. & MECKEL, T. 2000. Brunei deep water exploration: From sea floor images and shallow seismic analogues to depositional models in a slope turbidite setting. GCSSEPM Foundation 20th Annual Research Conference, Deepwater Reservoirs of the World. 304-318. KOLLA, V. 8z MACURDA, D. B. JR. 1988. Sea level changes and timing of turbidity---currents events in deep-sea fan systems. In: WILGUS, C. K., HASTINGS, B. S,, KENDALL, C. G. St. C., POSAMENTIER, H. W., ROSS, C. A. d~ WAGONER, J. C. (eds) Sea Level Changes: an Integrated Approach. S.E.P.M. Special Publication. 42, 381-392.
33
MAYALL, M. & STEWART, I. 2000, The architecture of turbidite slope channels. GCSSEPM Foundation 20th Annual Research Conference, Deepwater Reservoirs of the World, 578-586. M1GEON, S., SAVOYE, B,, ZANELLA, E., MULDER,T., FAUGERES,J.-C. & WEBER. O. 2001. Detailed seismic-reflection and sedimentary study of turbidite sediment waves on the Var sedimentary ridge (SE France): significance for sediment transport and deposition and for the mechanisms of sediment-wave construction. Marine and Petroleum Geology, 18, 179-208. MUTTI, E. 1992. Turbidite Sandstones. Agip publication. PICKERING, K. T., HISCOTT, R. N. & HEIN, F. J. 1989. Deep Marine Environnwnts: Clastic Sedimentation and Tectonics. Unwin Hyman, London. PIRMEZ, C., BEAUBOEEF, R. T., FRIEDMAN, S, J. & MOnRIG, D. C. 2000. Equilibrium profile and baselevel in submarine channels: Examples from late Pleistocene systems and implications for the architecture of deepwater reservoirs. GCSSEPM Foundation 20th Annual Research Conference, Deepwater Reservoirs of the World, 782-805. POSAMENTIER, H. P. 2001. Seismic geomorphology and depositional systems of deep-water environments; observations from offshore Nigeria, Gulf of Mexico and Indonesia. AAPG Annual Meeting, Program with Abstracts, A 160. POSAMENTIER, H. P., MEIZARW1N. P. S. W. & PLAWMAN, T. 2000. Deepwater depositional systems--Ultra-deep Makassar Strait, Indonesia. GCSSEPM Foundation 20th Annual Research Conference, Deepwater Resen'oirs of the World, 806- 816. SHERWOOD, P., ALGAR, S., BUSONO, I., FOWLER, J. N., FRANCOIS, J., GOFEEY, G., SMITH, M. J. & STRONG, A. G. 2001. Comparison of Recent and Mio-Pliocene deepwater deposits of the Kutei Basin, East Kalimantan. Proceedings of the 28th hldonesian Petroleum Association, 1,423-438. TWlCHELL, D. C., NELSON, H. & DAMUTH, J. E. 2000. Late stage development of the Bryant Canyon turbidite pathway on the Louisiana continental slope. GCSSEPM Foundation 20th Annual Research Conference. Deepwater Reservoirs of the World, 1032-1044. WEIMER, P. 1990. Sequence stratigraphy, facies geometries, and depositional history of the Mississippi fan, Gulf of Mexico. AAPG Bulletin, 74, 425-453.
The use of near-seafloor 3D seismic data in deepwater exploration and production G.S.
STEFFENS,
R.C.
SHIPP,
B.E.
PRATHER,
J.A.
NOTT,
J.L.
GIBSON
& C.D.
WINKER
Shell International Exploration a n d Production, Inc., 3 7 3 7 Bellaire Blvd, Houston, Texas 77025, USA (e-mail: Ga~'. Steffens @ shell.com)
Abstract: The analysis of 3D seismic data in the near-seafloor interval is a useful speciality in deepwater exploration and production. In addition to the well-established benefits of 3D seismic data, the higher frequency content of near-seafloor data has a variety of applications throughout the life cycle of deepwater plays. These benefits include: (l) depositional process modelling, (2) stratal architectural information for building reservoir models, and (3) drilling hazard assessment. Detailed mapping of well-imaged 3D seismic intervals in the near-seafloor interval is providing new insights to deepwater depositional processes and architectures. Depositional patterns are more confidently identified in near-seafloor settings, enabling the investigation of testable relationships between stratal stacking patterns, gradient changes and accommodation. These relationships as well as spatial and geometric information from these data are useful for building and constraining reservoir models, linking key observations from subsurface data at prospective levels with fine-scale outcrop analogue data. In particular, near-seafloor 3D data can image surfaces related to episodes of aggradation, starvation, bypass, and/or erosion that are typically hard to recognize or map at exploration depths, but are critical in controlling reservoir bed-length and connectivity in three dimensions. Near-seafloor 3D seismic data can supplement or even replace traditional 2D-based site surveys for assessing potential drilling hazards. Although usually lower in vertical resolution than 2D site survey data, 3D data have the distinct advantage of better imaging of 3D geometric bodies, providing insight into complex stratai stacking patterns, and allowing data volume manipulation and perspective.
The high-frequency content of near-seafloor 3D seismic data in deepwater settings permits high-resolution imaging of the seafloor as well as near-surface geological features (usually up to !.5 seconds below the mud line on average). This increased 3D resolution currently has three primary uses in deepwater exploration and exploitation: (1) discerning depositional environments and more accurately inferring sedimentation processes, (2) providing architectural information for building improved reservoir models, and (3) assessing shallow drilling hazards. For nearly two decades, Shell and others from the oil industry and academia have taken a multi-faceted approach to enhance detection and characterization of deepwater reservoirs (Steffens 1993). This approach involves the integration of subsurface studies with outcrop data, near-seafloor features, and physical modelling. Through this approach, significant advances in deepwater facies models have been made. For example, detailed, fine-scale architectures are found in several outcrop studies such as Chapin et al. (1994), Cook et al. (1994), DeVries & Lindholm (1994), and Martinsen et al. (2000). Some of the more notable high-resolution near-seafloor studies include: Winker (1993), Hackbarth & Shew (1994), Winker (1996), Beaubouef et al. (1998), Badalini et al. (2000), Beaubouef & Friedmann (2000), Brami et al. (2000), Posamentier et al. (2000), Posamentier (2001, 2004) and Babonneau et al. (2002). Integration of subsurface reservoir data with some of these analogues are illustrated in Mahaffie (1994), Shew et al. (1994), Blikeng & Fugelli (2000), Dean et al. (2000), Demyttenaere et aL (2000), Moraes et al. (2000), Sullivan et al. (2000) and Winker & Booth (2000). What has emerged from these studies is that 3D seismic imagery of near-seafloor features play an increasingly key role in linking different scaled data sets and providing valuable information to develop sophisticated deepwater reservoir and basin-fill models (Fig. 1). From a shallow drilling hazard perspective, recent trends show that 3D seismic data often supplement or replace traditional 2D-based site surveys in many deepwater basins around the world. Although usually lower resolution than 2D site survey data, 3D data has the distinct advantage of imaging geometric bodies and their intricate spatial stacking patterns in three dimensions. An added benefit is that this is done at reduced
cost, if acquired for the dual purpose of exploration and site evaluation. This paper summarizes the growing impact near-seafloor 3D seismic analysis is having on these deepwater activities. Seismic resolution and data scale issues are briefly reviewed. Examples follow, illustrating how near-seafloor 3D seismic analysis is starting to provide new insights in deepwater depositional processes and resolve fine-scale architectural features normally below resolution on conventional multi-channel seismic data at prospective intervals. The industry-wide use of 3D seismic data in deepwater drilling hazard assessment is also examined.
A matter of scale: seismic resolution and links to analogues The frequency content of conventional 3D seismic at exploration depths is usually 30 to 40 Hz at best, providing a minimum vertical seismic resolution (tuning thickness) that ranges from 12 to 25 m (assuming interval velocities of 2000-3000 m s-~). The frequency content of the near-seafloor section in conventional 3D data is often much higher, ranging from 60 to 70 Hz, yielding 6 to 8 m resolution (assuming interval velocities of 1675 to 1 8 0 0 m s - l ) . Therefore, conventional 3D seismic data near the seafloor typically has two to four times the vertical resolution that conventional seismic provides at exploration depths. To illustrate this point, Figure 2 is a 3D seismic example of a 200m deep channel near the seafloor in the deepwater Gulf of Mexico whose average velocity of the near seafloor section is approximately 1675 m s -I. Solving for the seismic wavelength: ?t-=V/f where k is wavelength, V is velocity and f is frequency: ~. = 1675 m s - 1 / 7 0 H z = - - 24m. The vertical resolution or tuning thickness is one quarter the wavelength: vertical resolution : ~./4 = 24/4 = -
6 m.
DAVIES, R. J., CARTWRIGHT,J. A., STEWART,S. A., LAPPIN,M. & UNDERH1LL,J, R. (eds) 2004.3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 35-43. 0435-4052/04/$15 9 The Geological Society of London 2004.
36
G.S. STEFFENS E T AL. control both reservoir bed-length and connectivity. These surfaces may extend tens of metres laterally in any one direction with vertical relief of 8 metres or less. Outcrops show that these surfaces are often too large to be easily detected with wireline tools, but aerially too small to be confidently mapped with conventional seismic data at objective levels. High-resolution imaging of the near-seafloor using 3D data can therefore provide the crucial link between detailed 2D outcrop observations and lower resolution 3D seismic geometries at exploration depths.
Fig. 1. Multi-faceted approach to deepwater facies analysis. Highresolution near-seafloor features provide a crucial link between outcrop and physical models with subsurface reservoir data. The integration of these data improves basin-fill fan models and reservoir characterization. With this vertical resolution, near-seafloor 3D seismic provides stratigraphical and geomorphological insights into the channelfill architecture that would normally be sub-seismic scale in the deeper subsurface. Near-seafloor data also has the distinct advantage of imaging these geometries and their intricate stacking patterns spatially in 3D. Reprocessing of these data can often achieve 80Hz frequencies, and if pushed through the ghost notch, can even achieve up to 125 Hz (the ghost notch in deepwater acquisition is the energy which travels upward from the shot and then is refected downward as a separate wave from the main wave train, lowering the frequency of the source as well as creating a separate spurious reflection event on the seismic and decreasing the frequency content of the data below it). At a much greater cost, acquiring short offset high-resolution 3D seismic data ( - - 2 0 0 - 6 0 0 m cables) can yield even higher frequency content up to 300Hz in the near-surface section, enhancing vertical resolution to 2 m or less. Regardless of reprocessing, 3D seismic near-seafloor analogues are lower resolution than outcrops data. However, they provide 3D information typically lacking from outcrops and greater detail than conventional 3D at exploration depths. Near-seafloor seismic data often image surfaces related to episodes of aggradation, starvation, bypass, and/or erosion that
New insights into deepwater depositional processes Detailed mapping of well-imaged 3D seismic sequences in the near-seafloor interval improves our understanding of deepwater depositional processes (Pirmez et al. 2000; Posamentier 2001, 2004; Kolla et al. 2001). Depositional patterns are more confidently identified in near-seafloor settings, enabling the critical controls and parameters on sedimentation to be tested, such as slope gradient, entry points, and accommodation. For example, fill and spill processes that commonly occur in partitioned slope settings are easily visualized and understood in terms of the type of accommodation the sediments are filling. Likewise, relationships between channel equilibrium profiles, thalweg development and overall geomorphology are confidently recognized and better understood for predicting potential sand distribution and architectural styles in various channel settings (Pirmez et al. 2000). Understanding these processes at the seafloor can make useful analogues for deeper, prospective sequences. With large 3D surveys, rendering of large regions of the sea floor event can illuminate depositional processes associated with an entire fan system such as on the continental slope area of West Africa (Fig. 3). The inset in the lower left hand corner shows an entry point channel in the up-dip portion of the basin, with a channel knick point at the spill point area, leading into the next down-dip basin. Two different seismic fill patterns are discernable within the basin (Fig. 4). Initially, the basin filled with sediments that were deposited in the 3D closures at the base of the basin (ponded sediments). Gently inclined sediments followed, 'healing' the local topography to the lowest breach
Fig. 2. Near-seafloor resolution of conventional 3D data. Example is an erosive 200 m deep canyon in deepwater Gulf of Mexico. 3D seismic is zero phase, VAR display; frequency roll-off is 65 to 70 Hz at - 20 db (see inset) with a vertical resolution of 6 to 7 m.
NEAR-SEAFLOOR 3D SEISMIC DATA
37
Fig. 3. Rendered perspective view of a seaflooor deepwater fan system on the upper to mid slope of offshore West Africa, with a close-up view of an intraslope basin fill. Seismic transect line located on inset is displayed in Figure 4. Colour shading accents depths: green and orange (shallower) to silver/ grey (deeper). point in the basin (termed healed slope deposits). Eventual down cutting and bypass occurs in the healed slope sediments. If the process had continued, the channel knick point at the spill point area on the western portion of the basin would have migrated updip, thus connecting to the entry point channel, which enters the basin on the eastern end of the basin. Lessons learned from deepwater fan systems in the Gulf of Mexico show that different types of accommodation (along with sediment flux and the channel system's equilibrium profile) exert a significant influence on sand body distribution, architectural styles and aerial extent (Prather et al. 1998; Prather 2000). The sediments that accumulate in the ponded accommodation of the intraslope basin in Figure 4 will probably have distributive channel and lobe sand geometries, whereas healed slope accommodation may have a higher proportion of complex channels. A broad spectrum of channel morphologies is encountered in the deepwater setting. Discerning sand distribution and architectures within this spectrum is a major challenge for the oil industry, which is currently exploring and exploiting these geometries in the deepwater. The coherency map on a nearseafloor deepwater channel in Figure 5a illustrates some of the channel morphology issues; the channel changes its plan view morphology several times as it crosses two faults (alternating straight and sinuous patterns). Further examination of the channel shows that the channel probably experienced changes in its gradient profile as it crossed each fault, resulting in dramatic changes in sinuosity and associated fill pattern. Posamentier
Fig. 4. Seismic transect over the intraslope basin, offshore West Africa. This section runs down the thalweg of the channel and through the spill-point connecting to the next basin down-dip. Ponded sediments occur at the base of the fill to the level of the breach-point, overlain by gently inclined seismic reflectors, 'healing' the local topography to the lowest breach-point in the basin (termed 'healed-slope deposits'). The seismic data is a variable density display, zero phase, a 90~ phase roll has been applied to the data to mimic acoustic impedance. See Figure 3 for location of transect.
et al. (2000), illustrates a similar example in the ultra-deep waters of Makassar Strait, Indonesia. Pirmez et al. (2000) in
their investigation of modern submarine channels, demonstrate that various channel processes are associated with equilibrium disruption as well as equilibrium re-establishment (such as illustrated in Fig. 5a). Some of these processes include thalweg down cutting and meander cut-offs up-dip of knick points, distributary channel development, and channel damming and redirection associated with normal faults and folds. Defining the relationship between these processes and the equilibrium profile is fundamental to predicting the type and spatial distribution of depositional elements within deepwater channels (Pirmez et al. 2000). Linking these near-seafloor relationships to theoretical and experimental models (e.g. Kneller 1995: Kneller & McCaffrey 1995) could lead to important breakthroughs in understanding the physical mechanisms responsible for sand distribution in a variety of channel morphologies. Further emphasizing the need to understand channel processes in the deepwater setting is the example where two channels occupy the same portion of the slope (Fig. 5b). The channel on the right exhibits tortuous sinuosity while the channel on the left a few kilometres away, is dominated by straight and very low sinuosity segments. Recent examination of 1800 km of various deepwater channels imaged on the seafloor, suggests that duration and frequency of flows in a channel system may have an important impact on the overall morphology of a channel system (Elliott & Edwards 2001). In particular, channel bends appear to grow with time and often
38
G.S. STEFFENS ET AL. Fig. 5. Various channel morphologiesin the near-seafloorinterval on the Nile Cone, deepwater Egypt. Images are flattened coherency slices, which highlight edges (i.e. fault and geological features). (A) In this coherency map, the deepwater channel changes its plan view morphology several times as it crosses two faults (alternating straight and sinuous patterns). The channel probably experienced changes in its gradient profile as it crossed each fault, resulting in dramatic changes in sinuosity and associated fill pattern. (B) Two channels occupy the same portion of the slope with very different plan view morphologies. The channel on the fight exhibits tortuous sinuosity while straight and very low sinuosity segments dominate the channel on the left. A possible reason for this may be that the sinuous channel may have had a longer, more complex history of establishing its equilibrium with repeated flows than the younger channel on the left.
have a downslope component. Elliott & Edwards (2001 ) suggest that channels may develop initially as over-deepened linear scours and evolve into sinuous channel-forms as initial defects along the flow path are amplified into bends. If this is true, the sinuous channel in Figure 5b may have had a longer, more complex history of repeated flows than the younger channel on the left, establishing equilibrium between its sediment load and graded profile. In both of these examples, the near-seafloor 3D seismic features give a unique opportunity to visualize changes in morphology along a channel-form and the associated differences in gradient profiles that control them. With well control calibration, these examples would have great predictive value in the deeper subsurface, providing important relationships between local controls on channel morphology and sedimentation, as well as lithofacies distribution.
Architectural data for reservoir models High-resolution near-seafloor features are an important source of architectural and depositional surfaces for constructing and constraining reservoir models. As discussed in the seismic resolution section, these surfaces are often at a scale that are hard to recognize and map with subsurface data at exploration depths and are usually limited to 2D in outcrop exposures. Some of these surfaces are crucial to recognize and map, because they impact reservoir continuity and subsequent flow behaviour. These surfaces along with dimensional data (e.g. channel width, thickness, sinuosity), and overall stacking patterns can be applied to reservoir models where appropriate. For example, a repeatable, complex stacking pattern is often seen in slope channel and canyon fills in the near-seafloor section that can be linked to the deeper subsurface (Mayall & Stewart 2000; McHargue 2001; Sikkema & Wojcik 2000). The succession often starts with canyon formation, erosion, and sediment bypass. This is followed by basal debrite deposition and highdensity sandy gravity flows across large portions of the canyon floor. Early construction of outer levees on the flanks of the canyon walls is common. An aggradational phase follows, dominated by mixed sand- and mud-rich deposition in moderate to highly sinuous channels (typically with inner levees), all confined within the canyon. The abandonment phase is usually
marked by vertically aggrading, highly sinuous leveed-channels, which eventually encroach and spill over the confines of the canyon and capped by hemi-pelagic drape. Variations to this stacking pattern are common, where architectural elements may be missing from the vertical succession or where canyon erosion is low relief or non-existent. The near-seafloor canyon complex in Figure 6 demonstrates many of the architectural elements described above. This strikeoriented seismic section through the canyon feature shows a chaotic seismic facies confined to the basal portion of the canyon; mass transport complexes are typical for this fill pattern. This is followed by a back-stepping succession of sinuous amalgamated channel complexes, capped by a channel levee complex that overtops and extends beyond the canyon confinement. Flattened time-slices through this interval show sinuous amalgamated channels (Fig. 7a). Proprietary bodychecking techniques highlight numerous levels of seismic discontinuity based on different amplitude thresholds (Fig. 7b and 7c). Useful for building static reservoir models, these different levels of seismic discontinuity may constitute a proxy for the general range of possible reservoir geo-body connectivity within the sinuous channel section. The challenge is to discern which of these scenarios represent the appropriate range of actual connectivity for reservoir modelling and field development planning relevant to a specific oil or gas field. While there are similarities in stacking patterns seen in many 3D seismic examples, there is considerable debate about deepwater sinuous channel morphologies and their comparison to fluvial systems (recent 3D examples include Roberts & Compani 1996; Kolla et al. 2001; theoretical considerations by Peakall et al. 2000). While some sinuous channels show primarily aggradation with little lateral migratory patterns, others display striking similarities to fluvial channel morphologies. Choosing the right analogue for building static reservoir models is problematic for most operators in the deepwater arena where many development projects currently are struggling with connectivity issues associated with sinuous channel reservoirs. This points to the need for greater calibration of these near-seafloor channel systems. Once calibration is achieved, depositional processes will be better understood as well as the implications for lithofacies distribution and connectivity within and between channel complexes.
NEAR-SEAFLOOR 3D SEISMIC DATA
39
Fig. 6. An example of an upper slope canyon fill, offshore West Africa, exhibiting many of the fill patterns seen in many deepwater canyon and channel fills. Several phases of erosion are present in the basal section of the canyon. Above the shallowest erosional surface (A), a chaotic seismic facies is confined to the basal portion of this cut, interpreted as a mass transport complex (B), followed by a discontinuous, variable to high amplitude package (C) that displays sinuous channels on flattened time-slices (see Fig. 7). Above this package is a channel-levee complex (D) that extends beyond the canyon confinement. This succession is approximately 100 ms to 400 ms below the seafloor (E). Seismic is reflection coefficient data with a frequency roll off of 65-70 Hz at - 20 db.
Fig. 7. Flattened seismic time slice and 'geo-body' extraction maps delineating external reservoir architecture in the sinuous channel fill interval in the slope canyon shown in Figure 6. (A) Flattened seismic time slice of sinuous channel fill interval. A proprietary body checking technique shows numerous levels of seismic discontinuity based on different amplitude thresholds (B and C). Each colour represents wavelets forming a 'geobody'. The light blue colour shows patterns of coherent seismic within which the geobodies occur. (B) The high threshold scenario where many geobodies are disconnected. (C) The low threshold scenario where most geobodies are connected. These two scenarios may constitute a proxy for the range of possible external reservoir connectivity within this variable to high amplitude package of sinuous channels (see also Fig. 6).
Drilling hazard assessment Currently there are two existing approaches to drilling hazard evaluation: (I) traditional 2D site surveys, and (2) conventional 3D assessment. The traditional 2D site survey was originally developed in the North Sea, where trapped shallow gas pockets could destabilize the seafloor, which were a major concern for bottom-founded drilling rigs. Analogue and digital 2D data is normally collected by dedicated vessels and interpreted by a third party.
In the past decade, the deepwater industry developed a methodology for evaluating deepwater drilling hazards using conventional 3D seismic data. In Shell, this methodology has been applied to over 200 sites in greater than 500 m of water depth in both mature and frontier basinal settings. On the continental slope and beyond, 3D data have better resolution than in shallower, shelf settings, due to near vertical angle of incidence. Hazard assessment in deepwater involves regional compilation of structural and depositional trends, derived from 3D seismic mapping and offset well control. These data are
40
G.S. STEFFENS ETAL.
then integrated with a detailed well site assessment, using the appropriate seismic dataset and a broad array of tools on the seismic interpretation workstation. The use of flattened and unflattened time slices, coherency volumes, novel visualization imagery, volumetric data analysis, and seismic attributes greatly enhance the interpretation and understanding of the near-surface, hazard-prone section. After the existing conventional and/or high-resolution reprocessed seismic data are reviewed, a decision on additional data needs is more easily determined, based on geological complexity and data resolution required to successfully implement the well plans. The most common drilling hazards that impact deepwater drilling operations are slope instability, unfavourable seafloor conditions, seafloor and subsurface faulting, fluid expulsion, gas accumulations, and shallow water flow related to nearsurface geopressures. The benefits of evaluating deepwater drilling hazards using 3D seismic data are: (1) a more rapid and less expensive project turnaround, (2) a better grasp of safety and risk concerns, and (3) an improved understanding of geologic setting. Use of 3Dbased assessment utilizes existing data, so little additional time and expense is invested to generate the necessary seismic data sets. A 3D-based assessment more easily identifies all potential safety issues and aids in determining the degree of risk associated with each potential drill site. Likewise, 3D-based assessment allows for more complete well planning by better visualization of the well path, which translates directly into cost savings. A 3D-based hazard assessment permits development of near-seafloor depositional models that improve future well planning and can serve as an analogue for understanding the deeper, objective-level geology in the area. Seafloor rendering from 3D seismic surveys, convolved with various attributes, often reveal areas of recent down slope activity and/or sediment deposition. For example, a rendered seafloor perspective view in the deepwater Gulf of Mexico reveals significant channelling between two salt massifs in the vicinity of a proposed well site
(Fig. 8). The pre-drill concern was whether these conduits were still active and would affect drilling operations. A seafloor amplitudes extraction overlain on a rendered seafloor perspective view reveals areas of high amplitudes (shown in red). These high amplitudes are interpreted as recent downslope movement, but not in the immediate vicinity of the proposed well site. Therefore, recent sedimentation appears not to be active at the proposed drill site, and the well was safely positioned.
Path forward A reoccurring theme in all of the near-seafloor examples shown in this paper is the need for calibration; it is essential for proper translation and integration with deeper subsurface objectives. Synergy is building again amongst the oil Industry, academia, and the IODP (International Ocean Drilling Program) for conducting high-resolution calibration programs on near-seafloor deepwater features. Such investigations would provide much-needed calibration for building finescaled reservoir architectural models as well as for drilling hazard assessment. Conducting a calibration program on a channel system that exhibits architectural diversity (such as the example in Fig. 9), would offer global applicability to a number of channel reservoir styles currently being explored and appraised in deepwater basins around the world. Acquiring short offset, high-resolution 3D seismic data with carefully positioned logging and coring sites on such a channel feature, would provide valuable 3D characterization and quantification of channel 'flow unit' architectures. Experience has shown, however, that the nearseafloor sections of some deepwater basins have highly variable acoustic rock properties. To maximize their lateral extrapolation and inversion requires carefully positioned lithological calibration, convolved with optimal seismic acquisition, processing, integration, and visualization technology. These and other
Fig. 8. Example of a drilling hazard assessment of a well site location in the deepwater Gulf of Mexico. The image is a 3D rendered seafloor perspective view, illuminated from the southeast, using the seafloor event and seafloor amplitude from the 3D conventional survey. Rendering uses azimuth data combined with angle, light, vertical exaggeration, and colour to enhance surface topography. The image reveals channelling between two salt massifs. Seafloor amplitudes overlain on the rendered seafloor perspective view reveal areas of high amplitudes (red areas), where recent downslope movement or activity may have occurred, but not in the immediate vicinity of the proposed well site. The red zones are thought to be acoustically hard zones or deposits, which may represent more compacted, previously buried material that was remobilized by recent mass movement. It was concluded that the proposed well site was in a relatively stable slope setting and could be safely positioned. Original image produced by Jay Cole, formerly of Shell Exploration & Production Company.
NEAR-SEAFLOOR 3D SEISMIC DATA
41
performance prediction, and e n h a n c e d drilling hazard detection. These advances will have a large impact on reducing the uncertainties associated with billion-dollar investment decisions in future deepwater developments. The manuscript was improved with the helpful reviews of H. Posamentier, S. Fraser and R. Davies. The contributions of J. Maggard, K. Wall, M. Stovall and M. Deptuck are also appreciated. The authors thank Shell International Exploration and Production for permission to publish this paper and ExxonMobil, Nigerian National Petroleum Company, ENI Exploration and Production, and WesternGeco for permission to release seismic data and images.
References
Fig. 9. High resolution of a sinuous channel at the seafloor, offshore Nigeria. (A) Seismic profile from 3D seismic reflection coefficient data. Bin size is 12.5 m x 25 m, Red events = decrease in acoustic impedance, and black events = increase in acoustic impedance. Roll-off frequency for the data is 65 to 70 Hz (at - 2 0 db), giving approximately 6 - 7 m vertical resolution. The overall channel complex is approximately 500 metres thick from seafloor to the deepest portion of the erosional thalweg. The channel complex offers significant architectural diversity in its seismic facies fill pattern and is characterized by a large erosional base (arrows), overlain by a broad zone of anastomizing high-amplitude reflectors (HARs) (X), followed by vertically aggrading narrow HARs (Y) and adjacent parallel reflectors. The favoured interpretation in this uncalibrated example is that the channel evolved from laterally accreting, anastomosing sinuous channels at the base to predominantly vertically aggrading highly sinuous channels (with inner levee development) in the upper section, all contained within a meanderbelt plain approximately 4 - 5 km wide and bounded by outer levee geometries (Z). (B) Illuminated water bottom dip map with location of A - A ; seismic line. Note the highly sinuous channel expressed on the seafloor (yellow ellipses), representing the last stage of the vertically aggrading narrow HAR section below it.
challenges remain w h e n considering future near-seafloor deepwater calibration programs.
Conclusions The use of near-seafloor 3D seismic is playing an increasing role in deepwater exploration and production. High-resolution 3D i m a g e r y of near-seafloor features provides valuable insights into deepwater depositional processes, reservoir architectural styles, and drilling hazard assessment. Continued use and i m p r o v e m e n t of near-seafloor imaging will provide the crucial 3D linkage b e t w e e n different-scaled subsurface and analogue data. If calibrated, they also can provide recognition criteria and predictive capabilities for facies stacking patterns in various d e e p w a t e r depositional environments. Its use in all o f these facets o f deepwater activities is leading to better effective properties assignment in reservoir models, improved reservoir
BABONNEAU, N., SAVOYE, B., CREMER, M. & KLEIN, B. 2002. Morphology and architecture of the present canyon and channel system of the Zaire deep-sea fan. Marine and Petroleum Geology, 19, 445-467. BADALINI, G., KNELLER, B. • WINKER, C. D. 2000. Architecture and processes in the Late Pleistocene Brazos-Trinity turbidite System. Gulf of Mexico Continental Slope. In: WEIMER, P., SLATT, R. M., COLEMAN, J., ROSEN, N. C., NELSON, H., BOUMA, A. H., STYZEN, M. J. & LAWRENCE, D. T. (eds) Deep-Water Reservoirs of the World, GCS-SEPM Foundation, 20th Annual Bob F. Perkins Research Conference, 16-24. BEAUBOUEF, R. T. & FR1EDMANN, S. J. 2000. High resolution seismic/sequence stratigraphic framework for the evolution of Pleistocene intra-slope basins, Western Gulf of Mexico: depositional models and reservoir analogs. In: WEIMER, P., SLATT, R. M., COLEMAN, J., ROSEN, N. C., NELSON, H., BOUMA, A. H., STYZEN, M. J. & LAWRENCE. D. T. (eds) Deep-Water Reservoirs of the World, GCS-SEPM Foundation, 20th Annual Bob F. Perkins Research Conference, 40-60. BEAUBOUEF, R. T., FRIEDMANN. S. J. & ALWIN, B. W. 1998. High resolution seismic/sequence stratigraphy of intra-slope basins, Western Gulf of Mexico. In: ROCHA MELLO, M. & OYA Y1LMAZ, P. (eds) (1998) AAPG b~ternational Cot!ference and Exhibition Extended Abstracts, Rio de Janeiro, Brazil, 404-405. BLIKENG, B. & FUGELLI, E. 2000. Application of results from outcrops of the deep water Brushy Canyon Formation, Delaware Basin, as analogues for the deep water exploration targets on the Norwegian shelf, hi: WEIMER, P., SLATT, R. M., COLEMAN, J., ROSEN, N. C., NELSON, H., BOUMA, A. H., STYZEN, M. J. & LAWRENCE, D. T. (eds) Deep-Water Reservoirs of the World, GCS-SEPM Foundation, 20th Annual Bob F. Perkins Research Conference, 61-81. BRAMI, T. R.. PIRMEZ, C., ARCHIE, C., HEERALAL,S. & HOLMAN, K. L. 2000. Late Pleistocene deep-water stratigraphy and depositional processes, Offshore Trinidad and Tobago. hi: WEIMER, P., SLATT, R. M., COLEMAN, J., ROSEN, N. C., NELSON, H., BOUMA, A. H., STYZEN, M. J. & LAWRENCE. D. T. (eds) Deep-Water Reservoirs of the World, GCS-SEPM Foundation. 20th Annual Bob F. Perkins Research Conference, 104- I 15. CHAPIN, M. A., DAVIES, P., GIBSON, J. L. & PETTINGILL, H. S. 1994. Reservoir architecture of turbidite sheet sandstones in laterally extensive outcrops, Ross Formation, Western Ireland. hi: WEIMER, P., BOUMA, A. H. & PERKINS, B. F. (eds) Submarine Fans and Turbidite Systems, Sequence Stratigraphy, Reservoir Architecture and Production Characteristics, GCS-SEPM Foundation, 15th Annual Research Conference, 53-68. COOK, T. W., BOUMA, A. H. & CHAPIN, M. A. 1994. Facies architecture and reservoir characterization of a submarine fan channel complex, Jackfork Formation, Arkansas. In: WEIMER, P., BOUMA, A. H. & PERKINS, B. F. (eds) Submarine Fans and Turbidite Systems, Sequence Stratigraphy, Reservoir Architecture and Production Characteristics, GCS-SEPM Foundation, 15th Annual Research Conference, 69-81. DEAN, M. C., KING, K. M. & BOOTh, J. R. 2000. The present is the key to the Plio-Pleistocene: seafloor analogues at SheWs GB 516 Serrano discovery, bt: WEIMER, P.. SLATT, R. M., COLEMAN. J., ROSEN, N. C., NELSON, H., BOUMA, A. H., STYZEN, M. J. &
42
G. S. STEFFENS ET AL.
LAWRENCE, D. T. reds) Deep-water Reservoirs of the World, GCSSEPM Foundation, 20th Annual Bob F. Perkins Research Conference, 293-303. DEMYTTENAERE, R., TROMP, J. P., IBRAH1M, A., ALLMAN-WARD,P. 8r MECKEL, T. 2000. Brunei deep water exploration: from sea floor images and shallow seismic analogues to depositional models in a slope turbidite setting. In: WEIMER, P., SLATT,R. M., COLEMAN,J., ROSEN, N. C., NELSON, H., BOUMA, A. H., STYZEN, M. J. & LAWRENCE, D. T. (eds) Deep-Water Reservoirs of the World, GCSSEPM Foundation, 20th Annual Bob F. Perkins Research Conference, 304-317. DEVRIES, M. B. & LINDHOLM, R. M. 1994. Internal architecture of a channel-levee complex, Cerro Toro Formation, southern Chile. In: WEIMER, P., BOUMA,A. H. & PERKINS, B. F. (eds) Submarine Fans and Turbidite Systems, Sequence Stratigraphy, Reservoir Architecture and Production Characteristics, GCS-SEPM Foundation, 15th Annual Research Conference, 105-114. ELLIOTT, T. & EDWARDS, C. M. 2001. Morphometric insights into the dynamics and behaviour of deep sea sinuous channels. In: FRASER, S. I., FRASER, A. J., JOHNSON, H. D. & EVANS, A. M. (eds) Petroleum Geology of Deepwater Depositional Systems, Advances in Understanding 3D Architecture. The Geological Society Conference 2001 proceedings, The Geological Society, London. HACKBARTH,C. J. & SHEW, R. D. 1994. Morphology and stratigraphy of a mid-Pleistocene turbidite leveed channel from seismic, core and log data, North-eastern Gulf of Mexico. In: WEIMER, P., BOUMA, A. H. & PERKINS, B. F. (eds) Submarine Fans and Turbidite Systems, Sequence Stratigraphy, Reservoir Architecture and Production Characteristics, GCS-SEPM Foundation, 15th Annual Research Conference, 127-133. KNELLER, B. C. 1995. Beyond the turbidite paradigm: physical models for deposition of turbidites and their implications for reservoir prediction. In: MARLEY, A. J. & PROSSER, D. J. (eds) Characterization of Deep Marine Clastic Systems. Geological Society, London, Special Publication, 94, 31-49. KNELLER, B. C. & MCCAFFREY, D. W. 1995. Modeling the effects of salt-induced topography on deposition from turbidity currents. GCS-SEPM, Houston, 137-t45. KOLLA, V., BOURGES, PH., URRUTY, J. M. & SAFA, P. 2001. Evolution of deep-water Tertiary sinuous channels offshore Angola (West Africa) and implications for reservoir architecture. AAPG Bulletin, 85, 1373-1405. MAHAFFIE, M. J. 1994. Reservoir classification for turbidite intervals at the Mars discovery, Mississippi Canyon 807, Gulf of Mexico. In: WE1MER,P., BOUMA,A. H. & PERKINS,B. F. (eds) Submarine Fans and Turbidite Systems, Sequence Stratigraphy, Reservoir Architecture and Production Characteristics, GCS-SEPM Foundation, 15th Annual Research Conference, 233-244. MARTINSEN, O. J., LIEN, T. & WALKER, R. G. 2000. Upper Carboniferous deep water sediments, Western Ireland: analogues for passive margin turbidite plays. In: WEIMER, P., SLATT, R. M., COLEMAN, J., ROSEN, N. C., NELSON, H., BOUMA, A. H., STYZEN, M. J. & LAWRENCE, D. T. (eds) Deep-Water Reservoirs of the World, GCS-SEPM Foundation, 20th Annual Bob F. Perkins Research Conference, 533-555. MAYALL, M. & STEWART, I. 2000. The architecture of turbidite slope channels. In: WEIMER, P., SLATT, R. M., COLEMAN, J., ROSEN. N. C., NELSON, H., BOUMA, A. H., STYZEN, M. J. & LAWRENCE. D. T. (eds) Deep-Water Reservoirs of the World, GCS-SEPM Foundation, 20th Annual Bob F. Perkins Research Conference. 578-586. MCHARGUE, T. 2001. Recurring stacking pattern of reservoir elements in erosional slope valleys, Niger Delta, Nigeria. In: FRASER, S. I., FRASER, A. J., JOHNSON, H. D. & EVANS, A. M. (eds) Petroleum Geology of Deepwater Depositional Systems, Advances in Understanding 3D Architecture. The Geological Society Conference (2001) proceedings, The Geological Society, London. MORALS, M. A. S., BECKER, M. R., MONTEIRO, M. C. & ALMEIDA NETTO, S. L. 2000. Using outcrop analogs to improve 3D
heterogeneity modelling of Brazilian sand-rich turbidite reservoirs. In: WEIMER, P., SLATT, R. M., COLEMAN, J., ROSEN, N. C., NELSON, H., BOUMA, A. H., STYZEN, M. J. & LAWRENCE, D. T. (eds) Deep-Water Resen,oirs of the World, GCS-SEPM Foundation, 20th Annual Bob F. Perkins Research Conference, 587-605. PEAKALL, J., MCCAFFREY, B. & KNELLER, B. 2000. A process model for the evolution, morphology, and architecture of sinuous submarine channels. Journal of Sedimentar)' Research, 70, 434-448. PIRMEZ, C., BEAUBOUEF,R. T., FRIEDMAN,S. J. & MOHR1G,D. C. 2000. Equilibrium profile and baselevel in submarine channels: examples from Late Pleistocene systems and implications for the architecture of deep water reservoirs. In: WEIMER, P., SLATT, R. M, COLEMAN, J., ROSEN, N. C., NELSON, H., BOUMA, A. H., STYZEN, M. J. & LAWRENCE, D. T. (eds) Deep-Water Reservoirs of the WorM, GCSSEPM Foundation, 20th Annual Bob F. Perkins Research Conference, 782-805. POSAMENTIER, H. W. 2001. Depositional elements and processes that characterize deepwater environments: evidence from 3D seismic data. In: FRASER, S. I., FRASER, A. J., JOHNSON, H. D. & EVANS, A. M. (eds) Petroleum Geology of Deepwater Depositional Systems, Advances in Understanding 3D Architecture. The Geological Society Conference 2001 Proceedings, The Geological Society, London. POSAMENTIER, H. W. 2004. Seismic geomorphology: imaging elements of depositional systems from shelf to deep basin using 3D seismic data: implications for exploration and development. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29. 11-24. POSAMENT1ER. H. W., MEIZARWIN, WISMAN, P. S. & PLAWMAN, T. 2000. Deep water depositional systems--ultra-deep Makassar Strait, Indonesia. In: WEIMER, P., SLATT, R. M., COLEMAN, J., ROSEN, N. C., NELSON, H., BOUMA, A. H., STYZEN, M. J. & LAWRENCE, D. T. (eds) Deep-water Reservoirs of the WorM, GCSSEPM Foundation, 20th Annual Bob F. Perkins Research Conference, 806-816. PRATHER, B. E. 2000. Calibration and visualization of depositional process models for above-grade slopes: a case study from the Gulf of Mexico. Marine and Petroleum Geology, 17, 619-638. PRATHER, B. E., BOOTH, J. R., STEFFENS, G. S. & CRAIG, P. A. 1998. Classification, lithologic calibration, and stratigraphic succession of seismic facies of intraslope basins, deep-water Gulf of Mexico. AAPG Bulletin, 82, 701-728. ROBERTS, M. T. & COMPANI, B. 1996. Miocene example of a meandering submarine channel-levee system from 3D seismic reflection data, Gulf of Mexico Basin. GCS-SEPM Foundation, 17th Annual Research Conference, 241-254. SHEW, R. D., ROLLINS, D. R., TILLER, G. M., HACKBARTH, C. J. & WHITE, C. D. 1994. Characterization and modelling of thin-bedded turbidite deposits from the Gulf of Mexico using detailed subsurface and analog data. In: WEIMER, P., BOUMA, A. H. & PERKINS. B. F. (eds) Submarine Fans and Turbidite Systems, Sequence Stratigraphy. Reservoir Architecture and Production Characteristics, GCS-SEPM Foundation, 15th Annual Research Conference, 327-334. SIKKEMA, W. ~a WOJCIK, K. M. 2000. 3D visualization of turbidite systems, Lower Congo Basin, Offshore Angola. In: WEIMER, P., SLATT, R. M., COLEMAN, J., ROSEN, N. C., NELSON, H., BOUMA, A. H., STYZEN, M. J. & LAWRENCE, D. T. (eds) Deep-Water Reservoirs of the World, GCS-SEPM Foundation, 20th Annual Bob F. Perkins Research Conference, 928-939. STEFFENS, G. S. 1993. Gulf of Mexico deepwater seismic stratigraphy. AAPG Program with Abstracts, New Orleans, 186. SULLIVAN, M., JENSEN, G., GOULDING,F., JENNETTE, D., FOREMAN,L. & STERN, D. 2000. Architectural analysis of deep-water outcrops: implication for exploration and development of the Diana SubBasin, Western Gulf of Mexico. In: WEIMER, P., SLATT, R. M.,
NEAR-SEAFLOOR 3D SEISMIC DATA COLEMAN, J., ROSEN, N. C., NELSON, H., BOUMA, A. H., STYZEN, M. J. & LAWRENCE, D. T. (eds) Deep-Water Reservoirs of the World, GCS-SEPM Foundation, 20th Annual Bob F. Perkins Research Conference, 1010-1031. WINKER, C. D. 1993. Leveed slope channels and shelf-margin deltas of the Late Pliocene to Middle Pleistocene Mobile River, NE Gulf of Mexico: comparison with sequence stratigraphic models. AAPG Program with Abstracts, New Orleans, 201. WINKER, C. D. 1996. High-resolution seismic stratigraphy of a Late Pleistocene submarine fan ponded by salt-withdrawal mini-basins
43
on the Gulf of Mexico continental slope. Offshore Technology Conference, Paper 8024, Houston, 619-628. WINKER, C. D. & BOOTH, J. R. 2000. Sedimentary dynamics of the saltdominated continental slope, Gulf of Mexico: integration of observations from the seafloor, near-surface and deep subsurface. In: WEIMER, P., SLATT, R. M., COLEMAN, J., ROSEN, N. C., NELSON, H., BOUMA, A. H., STYZEN, M. J. & LAWRENCE, D. T. (eds) Deep-Water Reservoirs of the World, GCS-SEPM Foundation, 20th Annual Bob F. Perkins Research Conference, 1059-1086.
Structural controls on the positioning of submarine channels on the lower slopes of the Niger Delta RICHARD
MORGAN
Veritas D G C Limited, Crompton Way, Crawley, W. Sussex RHIO 9QN, UK (e-mail:
[email protected])
Abstract: Recently acquired 2D seismic data shot over the western Niger Delta have enabled a pre-delta rift framework to be delineated inshore of a transform fault dominated continental margin which lies beneath the later, delta sediment apron. The delta apron has been deformed by toe-of-slope thrusting where faults have climbed from a detachment surface at or near the top of the over-pressured Akata Formation mudstones. The overlying mixed clastic succession of the Agbada Formation has been faulted by a broadly oceanward stepping series of NW-SE trending thrusts climbing from this detachment level. The principal thrusts have been offset by NE-SW trending transfer zones, the positions of which have been inherited from trends within a pre-delta rift framework that underlies part of the western delta slope. 3D seismic data partly covering the 2D grid show turbidite channel complexes at numerous stratigraphic levels within the Agbada Formation and clustered in particular areas of the slope. Commonly, submarine channels can be seen to have cut through the relief caused by folding at the positions of intersection with transfer fault zones. These data show the relationship between structure and channel formation and highlight the importance of transfer fault zones in localizing channel systems on the lower slope. Nevertheless, the 2D seismic data has provided an explanation for the location of the transfer zones within the toe-thrust belt in the form of an underlying structural framework, and both data types have contributed to the understanding of controls on reservoir distribution in an area where the principal sand delivery systems are perpendicular to the main structural trend.
The discovery of a succession of major oil accumulations in the deep water parts of the Niger Delta, e.g. Bonga, Erha, Agbami and Akpo, have made this region one of the most prospective deep water provinces in the world. Consequently, the search for large fields has progressed down slope and stimulated the acquisition of an extensive seismic dataset (Fig. 1). These data have provided an opportunity to examine in detail the lower slopes of the delta apron between 1500 m and 4000 m present water depth. Prior to 1998 very little seismic data existed in the deep and ultra deep-water areas covering the lower slope of the Niger Delta and models describing the structure of the sediment apron could only draw on individual widely spaced lines of varying vintages (Whiteman 1982; Knox & Omatsola 1989; Damuth 1994; Cohen & McClay 1996). These interpretations incorporated the mega regressional cycle model to describe the progressive build out of the delta from Eocene times to present. The model predicts a diachronous contact between fine-grained, distal or basin floor sediments, broadly describing the Akata Formation, underlying more sand-rich, slope sediments, broadly describing the Agbada Formation (Fig. 2). In the present day shelf and upper slope areas, this boundary is taken as the top of the mobile shale section, although the movement of overpressured mud and associated faulting have led to considerable topography on this surface and complex relationships with overlying sediments render this boundary a problematic seismic stratigraphic correlation. The base of the sediment apron could not be determined with confidence in the legacy data (where the record length was sufficient), as the top basement reflector varies in character and visibility (Damuth 1994). Also, the existence of any distinct seismic facies beneath the presumed Eocene to recent delta slope deposits remained unconfirmed until the recent data acquisition. The database available to this study comprised an extensive grid of 2D seismic data acquired in 1998 and 1999 covering all of the lower slope region. The dip line spacing of data over the western slope area is typically 4 km with strike line spacing of 10kin. These 120 fold data were acquired with a 6 k m cable length with a 12s recording interval and processed using Kirchhoff bent ray pre-stack time migration. Additionally, a 3D
dataset acquired in 1999, was available covering 3100kin 2 of the western lower slope (Fig. 1). These data also have a 6 km offset length and 12s record interval and share a similar processing sequence. All seismic data are displayed with a reverse (European convention) polarity where an increase in impedance is represented by a trough. The aim of this paper is to use evidence derived from these data to show the relationship between the toe-of-slope thrust structures and sediment pathways in the setting of the lower slope.
Structural and stratigraphic setting The Niger Delta is a regressive clastic succession 1 0 - 1 2 k m in thickness, comprising a shelf, broad slope area and basin floor. The lower slope area upon which this study focuses, can be readily divided into the Agbada and Akata Formations, due to the occurrence of a regionally consistent seismic reflection event dividing sections of differing seismic character (Fig. 3). This division is believed to reflect the stratigraphic transition from the lower, mud-prone Akata Formation into the upper, mixed clastic, Agbada Formation as recognized onshore and on the shelf. Recent descriptions of the nearby, deep-water Bonga Field, have given Upper Oligocene ages for the lowermost parts of the Agbada interval, inferring the Akata interval to be pre-Miocene in this part of the slope (Chapin et al. 2002). The existence of older sediments underlying the Akata Formation in the deep and ultra deep water (Fig. 3) is apparent in the 1998, 1999 data, where sections in excess of l km in thickness are seen preserved in half-graben rift elements, overlain by a large sediment wedge (Morgan 2003). This older section is presumed to include Albian to Palaeogene age sediments as a late Aptian to late Albian age is given for the onset of continental separation in the Gulf of Guinea (Gradstein et al. 1995; Wagner & Pletsch 1999; Macgregor et al. 2003). The rift elements themselves may contain older deposits as the onset of rifting in Benue Trough is given as Aptian (Burke et al. 1971; Petters & Ekweozor 1982), while a Berriasian-Hauterivian age is recognized for the syn-rift Ise Formation in the Benin section of the Dahomey Trough.
DAVIES,R. J., CARTWRIGHT,J. A., STEWART,S. A., LAPPIN,M. & UNDERHILL,J. R. (eds) 2004.3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 45-51. 0435-4052/04/$15 9 The Geological Society of London 2004.
46
R. MORGAN
Fig. 1. Seismic data coverage over the deep/ultra deep waters of the Niger Delta slope. Well locations show the positions of major oil accumulations in the upper and mid-slope areas.
The pre-Akata sedimentary section is thickest beneath the northwestern area of the Niger lower slope, where it comprises a major southward building sedimentary apron. The centre of this apron is not coincident with the present Niger Delta cone and is located to the west, offshore Lagos (Fig. 1). The Akata Formation onlaps this earlier sediment apron (Fig. 3) with the basal sequence boundary marking an important change in drainage and sediment dispersal on this part of the margin at this time. The base of the Agbada Formation marks another
change in depositional style across the region with the appearance of a major progradational succession continuing through until the present and forming the main body of the Niger sediment apron (Fig. 3). The lower part of this succession is represented by a distinct section, the Dahomey wedge, socalled because the channel complexes within this part of the Agbada Formation are predominantly southward directed, sourced apparently from the Dahomey Trough region of the Nigeria margin.
Fig. 2. Tri-partite subdivision of the main components of the Niger Delta sediment cone: The fluvio-deltaic Benin Fm., the marine shelf and slope sand and muds of the Agbada Fm. and upwards of 6 km of marine slope muds of the Akata Fm. Over-pressuring in the Akata Fm. has rendered the Akata structurally weak and the entire sediment cone has collapsed on intra Akata detachment faults creating extensional, faulted-diapric and compressionai belts within the apron.
SUBMARINE CHANNELS, NIGER DELTA
47
Fig. 3. The above section represents a highly compressed, 180 km, NW-SE profile through the lower slope region of the W Niger Delta area. The boundary between the Agabda and Akata seismic divisions is marked a major downlap surface and the arrival of sediments that show greater seismic reflectivity. The Dahomey wedge, which forms the lower part of the Agbada Formation comprises large channel complexes and amalgamated basin floor fans that have prograded over the mud-dominated Akata succession. The Akata section onlaps a substantial sediment apron of presumed Albian-Palaeocene age. This apron in turn sits partially on rifted crust of unknown affinity and extends out onto presumed oceanic crust.
In the lower slope region, the Niger Delta sediment apron has been deformed by toe-of-slope thrusting and in the western area the detachment surface for this faulting occurs at the top of over-pressured, Akata Formation mudstones. The causes of faulting in the Niger Delta have been extensively discussed (Whiteman 1982; Knox & Omatsola 1989; Damuth 1994) and are generally attributed to the extensional collapse of the thickest parts of the apron above the over-pressured Akata muds. Transport of the detached section comprising the entire Agbada interval has been ocean-ward and directed radially around a broadly arcuate trend. In the lower parts of the slope, displacement passes back up through the section in the form of a toe-thrust complex (Fig. 4). On close examination the toethrust complex takes the form of a number of zones of thrusting with differing characteristics such as, trend, level of basal detachment, size of imbricated blocks and incidence of backthrusts. The western parts of the toe-thrust belt are discussed here and the role of the resulting structures on sediment dispersal on the lower slope examined.
Fault framework and depositional architecture from 2D seismic data In the western lower slope the mixed clastic succession of the Agbada seismic facies has been faulted by a series of broadly ocean-ward stepping, N W - S E trending thrust faults, climbing from an uppermost Akata detachment level (Fig. 4). The 2D seismic database has enabled the principal thrusts to be mapped (Fig. 5). These N W - S E trending faults are extremely linear in strike, a characteristic that serves to emphasise abrupt lateral terminations and a number of step-wise offsets that occur along
their length. The outermost expression of thrusting in the western delta area is marked by a complex fore-thrust/backthrust zone, and in contrast to the faulting higher up the slope this zone appears to be spatially related to deeper fault antecedents (Fig. 4). The fore-thrust/backthrust zone also displays step-wise offsets typically of the order of 5 - 1 0 k m and these structures, when linked to offsets and fault terminations up-slope, express the positions of N E - S W trending transfer zones (Fig. 5). In section, the transfer zones are characterized by steep to vertical, N E - S W trending faults with evidence of both extensional and compression movements (Fig. 6). These faults compartmentalize the toe-of-slope thrust zone and appear to have a c c o m m o d a t e d differences in displacement within the detached sediment pile. Changes in seismic character and growth of section across the transfer faults at depth within the pre-Akata succession demonstrate that a number of these fault zones were active prior to the formation of the delta apron (Fig. 6). The fact that some transfer faults pass through the semi-regional detachment surface and are linked to deeper structures reveals the tram-line like control these faults have had in the partitioning of the thrust belt. Similarly, the outer-most thrust/backthrust zone is linked spatially if not implicitly to parallel trending rift elements in the basement such that the pre-delta structural framework has had an overriding influence on the position and trend of the outermost expression of thrusting. The fact that thrusting due to the extensional collapse of the sediment apron up-slope, has inherited elements of an earlier structural fabric created by the underlying rift framework helps to explain the linear form, abrupt terminations and changes in trend seen in the geometry of the toe-thrust belt. Depositional processes within the upper parts of the Agbada seismic facies are dominated by large c h a n n e l - l e v e e systems
48
R. MORGAN
Fig. 4. Profile through the toe-of-slope thrust zone in the W Niger Delta area. A difference in general seismic reflectivity describes the characterization of the Agbada and Akata seismic divisions, while older sediments lie in a normal faulted setting beneath the Akata section. The detachment of the Agbada section towards the top of the Akata seismic interval and is demonstrated by thrust geometries inboard of the outermost thrust zone. However, the outermost thrust zone is a complex backthrust structure formed by the underthrusting of the detached sediment apron. The backthrust zone directly overlies a parallel zone of faulting affecting the basement and older section. The older faults may have acted passively as seed points off which backthrust ramps formed from the detachment surface.
Fig. 5. Time structure at base Agbada Formation, contour interval is 100 ms, red high, blue low. The positions of the main thrust faults can be seen in this surface, and demonstrate the linearity of strike of the toe-of-slope thrust belt and the positions of offset of the faults along strike. Despite the continued progradation of the sediment apron, with the deposition of over 3 km of sediment above this horizon, recently formed deepwater channels have followed the transfer zones through the cross-cutting sea-bed topography created by the thrusting.
SUBMARINE CHANNELS. NIGER DELTA
49
Fig. 6. NE-SW trending transfer/tear faults within a transfer zone showing steep to vertical geomet~, fault termination at the semi-regional detachment level at base Agbada Formation and also through-going linkage to deeper levels. The deep-rooted transfer faults offsetting the detachment level demonstrate the influence of basement structure on faulting within the collapsed delta apron. Note the clustering of high-amplitude, channelized packages around the transfer fault. and debris-flow units (Fig. 6) and clustered or amalgamated channel-fill packets are visible throughout the Agbada interval. The lower parts of the interval include large base-of-slope fan, lobe forms and these are particularly well developed immediately above the Akata-Agbada seismic event boundary (Fig. 4). The correlation of subsurface channel packets line to line is problematic without well-based stratigraphic control. However, the most recent channels expressed at sea bed can be mapped from 2D seismic data down the slope and out onto the rise, outboard of the outer-most thrust zone (Fig. 5). These are composed of discrete, multiple-phase channel corridors bounded by large levees. Evidence of channels branching and overbank splay events are not seen inboard of the outer thrust zone on the 2D data. The position of these channel systems on the slope correlates closely with the position of the transfer zones as determined by offsets in the trend of the toe-of-slope thrusts (Fig. 5), and suggests a causal link. Whilst the 2D seismic grid has been sufficient to determine the general down-slope orientation of the recent channel systems, the complexity of the subsurface, stacked, channel complexes is beyond the resolution of these data. Consequently the interaction between the depositional systems and the palaeo-sea bed relief formed by folding above thrust ramps is unclear from the 2D seismic alone.
The relationship of structure to deep-water channel formation in 3D seismic data 3D seismic data (Fig. 1) have enabled a more detailed interpretation of surface and sub-surface channel morphology, and
the controls exerted on these features by faulting and folding within the toe-of-slope thrust zone. Stacked channels with sinuous geometries are contained within larger, low sinuosity, channel complex corridors flanked by large levees. Profiles through recent channel corridors (Fig. 7) show levee walls in excess of 300 m in height, flanking aggradational channel fills with relatively little erosion of underlying units (Kolla et al. 2002). Although these channel complexes appear to be selfconfining through erosion, the actual mode of confinement has been the development of levees that have aggraded in unison with channel formation, building far more rapidly than deposition of the channel fill. The 3D data show the channel complexes to occur at numerous stratigraphic levels within the Agbada Formation, and in the upper half of the Agbada section the channels also tend to be clustered in particular areas of the slope. Channel complexes have been deflected by sea floor relief caused by the growth of hangingwall anticlines above toe-thrusts (Fig. 8) but, equally channels are seen to have cut through the relief caused by folding, commonly at the positions of intersection with transfer fault zones where the fold axis is offset and relief is reduced. Ponding of sediment brought down in the channels on the upslope or the down-slope sides of the relief created folding is not evident in this region of the slope. This may have resulted due to the effectiveness of the levees in containing the sediment flow within the channels and the fact that, although the thrusting has created sea bed relief, this has not led to the development of mini-basins within the lower slope area. Rather the slope environment depositional setting has been maintained through the toe-of-slope thrust belt. The apparent ease with which the
50
R, MORGAN
Fig. 7. Channel-levee relationships in the near surface. Note the difference in size of the channel fill packets and the levees. The amount of erosion caused by the formation of the channels is relatively small and was followed by aggradation and lateral migration during the early part of levee growth.
Fig. 8. A perspective view of the seabed in the lower slope region. The linear ridges created by underlying thrusts are clearly visible, as is the offset created by a transfer zone. The recently active channel processes have partially tracked the transfer zone down-slope and exploited the offset in the ridge to reach the outer slope/rise area. A partly buried thrust anticline further up-slope terminates abruptly against the transfer zone to create a lateral ramp (see Fig. 9). Offsets of channel corridors against the up-slope sides of the ridges created by the thrust anticlines can be seen, but these deflections do not appear to have led to sediment ponding at this level.
Fig. 9. A chair cut or box section (looking inside the box), showing the superimposition of a recent channel complex above an older, sub-surface example, influenced by the same transfer fault. The effect of the fault can be seen in the termination of steeply dipping events on the SE side. Also note that the fault continues across the central thrust anticline, over which the sub-surface channel complex is folded. Therefore the transfer fault pre-dates the thrust and shows evidence of continued movement following cessation of growth on this thrust.
SUBMARINE CHANNELS, NIGER DELTA submarine channel processes have dissected the sea-floor relief is somewhat misleading as channel and levee formation were coeval with fold growth and the development of the channel/levee systems have to a large degree kept pace with uplift of the fold axis. The general o c e a n - w a r d stepping sequence of faults climbing from the semi-regional detachment horizon can be seen in the greater degree of burial of the resulting hangingwall folds in more up-slope positions (Fig. 8). These folds became dormant as displacement was transferred via the ocean-ward propagating detachment horizon, onto faults further down-slope. In the example shown, although the up-slope fault ceased to move first and translation of the sediment pile continued on the detachment surface with this displacement taken up on a thrust fault climbing up through the section further down the slope. The transfer faults are seen to cut to sea-floor or near the seafloor across the slope (Fig. 9), demonstrating that these faults continued to accommodate displacement differences in parts of the slope where the thrusts had become dormant. This is because the transfer fault zones continued to accommodate differences in displacement between different sections of the detachment surface after the thrust front had m o v e d down-slope, It appears likely that this repeated activity on the transfer fault zones rather than c u m u l a t i v e d i s p l a c e m e n t has m a d e these structures influential in the positioning of submarine channels in this part of the slope. Both recent and sub-surface channel complexes follow transfer faults down the slope and have exploited the offsets created in the thrust related sea-floor relief to reach the base of the slope (Fig. 9). Repeated channel development in the vicinity of the transfer fault zones have led to the clustering of channel complexes around these structures seen in both 2D and 3D seismic data. It is this association that makes the transfer fault zones important to hydrocarbon exploration as the channel fill sediments are expected to contain the highest quality reservoirs in this part of the slope.
Conclusions A combination of 2D and 3D seismic data has provided important insights into the tectono-sedimentary evolution of the western lower slope of the Niger Delta. The structure of the toeof-slope thrust belt has been shown to contain transfer zone/tear fault elements that partition belt. The investigation into the effects of thrust and transfer zone structures on contemporaneous deposition, specifically the formation of slope channel/ levee complexes, has highlighted the relative importance of transfer zones to slope channel localization. Only one significant transfer zone occurs within the 3000 km 2 footprint of the 3D survey area available to this study and the detection of the location and frequency of the transfer zones has required a more regional perspective than even a large 3D dataset is able to provide. The 2D seismic data has provided sufficient resolution to locate the main transfer zones within the toe-thrust belt and has allowed these faults to be placed in context with underlying rift elements that occur along the margin. Both 2D and 3D seismic data have contributed to the understanding of controls on reservoir distribution in an area where the principal sand delivery systems are orthogonal to the main structural trend, The observation of structure and depositional systems at a number of different scales has been necessary and provides a good example of the important and
51
cost-effective role of 2D seismic data in placing the detail available in 3D data in a regional context. Access to the Veritas, Nigerian seismic database is gratefully acknowledged. Thanks are also extended to S. Thompson and T. Zaki for help with seismic imaging and to referees M. Grove and S. Mitchell for improvements to the manuscript.
References BURKE, K., DESSAUVAGIE,T. F. J. & WHITEMAN, A. J. 1971, Opening of the Gulf of Guinea and geological history of the Benue depression and Niger Delta. Nature Phys. Sci., 233, 51-55. CHAPIN, M., SHIPP, C. & WINKER, C. 2002. Bonga Field. deep water Nigeria: Comparison of near-surface, well-calibrated submarine channels with reservoir channel sands. (Abstract) AAPG Annual Meeting. Houston. COHEN. H. A. & MCCLAY, K. 1996. Sedimentation and shale tectonics of the northwestern Niger Delta front. Marine and Petroleum Geology, 13, 313-328. DAMUTH,J. 1994. Neogene gravity tectonics and depositional processes on the deep Niger Delta continental margin. Marine and Petroleum Geology, 11,320-346. DOUST, H. & OMATSOLA,E. 1989. Niger Delta. bl: EDWARDS,J. D. & SANTOGROSSI, P. A. (eds) Divergent/passin margin basins. American Association of Petroleum Geologists, Memoir, 48, 201-238. GRADSTEIN, F. M., AGTERBERG,F. P., OGG, J. G., HARDENBOL, J., VAN VEEN, P., THIERRY, J. & HUANG, Z. 1995. A Triassic, Jurassic and Cretaceous time scale, h~: BERGGREN, W. A., KENT. D. V., AUBREY,
M. P. & HARDENBOL, J. (eds) Geochronology, Time Scales and Global Stratigraphic Correlation, Society of Economic Palaeontologists and Mineralogists, Special Publication, 54, 95- i 26. KNOX, G. J. & OMATSOLA.E. M. 1989. Development of the Cenozoic Niger Delta in terms of the "escalator regression' model and impact on hydrocarbon distribution, h~: VAN DER LINDEN, W. J. M., CLOETINGH, S. A. P. L., KAASSCHEITER, J. P. K., VAN DER GRAAF, W. J. E., VANDENBERGLIE, J. & VAN DER GUN. J. A. M. (eds) Proceedings of the KNGMG Symposium Coastal Lowlands, Geology and Geotechnology The Hague. 1987. Kluwer, Dordrecht, 181-202. KOLLA, V., POSAMENT1ER,H. W. & IMRAN,J. 2002. Deepwater sinuous channels and reservoir architecture. (Abstract) AAPG Annual Meeting, Houston. MACGREGOR, D. S., ROBINSON, J. & SPEAR, G. 2003. Play fairways of the Gulf of Guinea transform margin. In: ARTHUR, J. J., MACGREGOR, D. S. & CAMERON. N. R. (eds) Petroleum Geology of Africa: New Themes and Developing Technologies. Geological Society, London. Special Publication, 207, 131-150. MORGAN, R. K. in press. Prospectivity in ultradeep water: the case for petroleum generation and migration within the outer parts of the Niger Delta apron, bl: ARTHUR, J. J., MACGREGOR, D. S. & CAMERON, N. R. (eds) Petroleum Geology of Afi4ca: New Themes and Developing Technologies. Geological Society, London, Special Publication, 307, 151-164. PETTERS, S. W. & EKWEOZOR, C. M. 1982. Petroleum geology of the Benue Trough and southeastern Chad Basin, Nigeria. AAPG Bulletin, 66, 1141 - 1149. WAGNER, T. & PLETSCH, T. 1999. Tectono-sedimentary controls on Cretaceous black shale deposition along the opening of the Equatorial Atlantic Gateway (ODP 159). In: CAMERON, N. R.. BATE, R. H. & CLURE, V. S. (eds) The Oil and Gas Habitats of the South Atlantic. Geological Society, London, Special Publications, 153, 241-265. WHITEMAN, A. J. 1982. Nigeria: Its Petroleum Geology. Resources and Potential. Graham and Trotman, London.
Sea bed morphology of the Faroe-Shetland Channel derived from 3D seismic datasets D. LONG,
J. B U L A T
& M. S. STOKER
British Geological Survey, West Mains Road, Edinburgh EH9 3LA, UK (e-mail:
[email protected])
Abstract: First returns from 3D exploration surveys have been utilized to display seafloor morphology of the Faroe-Shetland Channel between the UK and the Faroes. The image combines 32 datasets creating a regional perspective of Quaternary sedimentary processes. Geomorphic information is of significance for sea bed geohazard evaluation, environmental studies and as an analogy for former sedimentary environments. The image covers more than 25000 km 2 extending from the shelf (water depth - 120 m) to the basin floor (water depth up to - 1600 m). On any margin knowledge of the sea bed morphology is essential for understanding the environmental setting and for safe operations in deepwater. Under favourable circumstances, the sea bed can be picked from 3D exploration seismic surveys in a similar manner to any other horizon to provide detailed images of the seafloor, thereby negating the need for dedicated sea bed surveys. Combining first returns from several surveys creates a regional perspective, essential when considering importance of features e.g. the rarity of a certain seafloor environment or the presence of a potential landslide upslope from an operations area. The Faroe-Shetland Channel displays a wide range of sea bed features including, sediment waves, contourite deposits, polygonal cracking, landslides, debris flows, turbidity current channels, glacial moraines and iceberg ploughmarks. Resolving the spatial aspects of these features greatly assists the interpretation of shallow profile data for geohazard and environmental studies and provides a backdrop onto which biologists, oceanographers, sedimentologists and engineers can overlay their data sets and thus their interpretations.
The emergence of 3D seismic acquisition as a tool for regional reconnaissance as well as a tool for field development during the 1990s has resulted in near complete coverage in areas of active oil exploration. The Faroe-Shetland Channel (FSC), between the UK and the Faroes, has been one such area, being the subject of more than 35 surveys. These include exclusive and speculative surveys. These surveys were designed primarily to image depths in excess of 4 km, use low frequency sources and are recorded with low temporal sample rates (e.g. 2 or 4 ms). What is often unknown and frequently not considered is the level of detail that can be obtained of the sea bed from data with such characteristics. The sea bed can be considered as an horizon comparable to those studied in great detail at depth. The results of examining the sea bed can be applied in a range of uses e.g. rig site surveys, environmental assessments, site development investigations. Also the sea bed may be analogous of lower horizons. The advantage the sea bed horizon has over studies of other horizons is that there should be no difficulties in identification, a range of alternative seismic datasets to compare with and a greater abundance of physical samples for ground truthing. Such sea bed images can contribute to seismic geomorphology studies providing a link between subsurface features resolved within 3D cubes and modern day sediment processes (e.g. Posamentier 2001, 2002). Thus identifying what sedimentary bodies are resolvable and can be considered at depth.
Regional background The Faroe-Shetland region has been the subject of intensive oil exploration over the last decade. However operating in water depths in excess of 200 m and the intemperate climes of the North Atlantic is expensive, The British Geological Survey (BGS) has been involved in regional geological mapping of the area for the last twenty years and has, as a consequence, developed a regional understanding of the region's geology (Stoker et al. 1993). The Faroe-Shetland Channel has probably been a depocentre since the late Palaeozoic, however the present morphological expression of the basin is essentially a late Cenozoic phenomenon. The W y v i l l e - T h o m s o n Ridge, which separates the basin from the Rockall Trough to the south, is interpreted to be a mid-Tertiary inversion structure (Tate et al. 1999). Moreover, there is evidence for late Neogene seaward
tilting of the West Shetland margin (Stoker 2002). These structurations combined with the developing oceanographic regime and deteriorating climatic conditions have strongly influenced the late Cenozoic development of the region. The slope aprons, bordering the FSC, have migrated seawards during the Plio-Pleistocene by the growth of prograding wedges, which include glacigenic strata. These deposits interdigitate and/or overlap with basinal sediment-drift deposits. Thus the continental margin has developed through the interaction of both down-slope and along-slope processes (Stoker 2002).
Technical summary The sea bed image has had a long development, growing over five years for an oil industry Joint Industry Programme called Western Frontiers Association (WFA). The aim of the study was to create a regional image of the seafloor to help in identifying sea bed hazards in UK waters. The initial study has subsequently incorporated sea bed picks from the Faroese sector of the FSC, new data in UK waters and also reworking of some data sets to reduce data artefacts (Bulat & Long 2001). The present image contains data from 32 3D exploration surveys acquired in the FSC between 1990 and 2000. A fuller description of the methodology used in combining these data, data artefacts found and comparisons with high resolution seismic profiles is presented in Bulat & Long (2001). Essentially, these two-way time datasets were combined into a mosaic grid with a 100 m node spacing, and depth converted assuming a water velocity of 1500m/s. This mosaic was itself then patched into a regional bathymetry grid generated from the 100 m contour dataset in the General Bathymetric Compilation GEBCO97 (IOC, IHO & BODC 1997) to provide greater regional context. Most of the grid manipulation and the final visualization was performed using ERMapper, an industry standard grid mosaic, classification and visualization tool. The final bathymetry grid was imaged using ERMapper's 'shiny' algorithm that uses the Hue, Saturation and Intensity (HSI) colour model to produce Figure 2. The HSI model provides reflection highlights as well as shadow areas and is particularly effective in bringing out detailed surface texture. The two most common artefacts seen within these data sets are linear corrugations and survey edge effects. Systematic
DAVIES,R. J., CARTWRIGHT,J. A., STEWART,S. A., LAPPIN,M. & UNDERHILL.J. R. (eds) 2004.3D Seismic Technology:Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 53-61. 0435-4052/04/$15 9 The Geological Society of London 2004.
54
D. LONG E T A L .
noise associated with acquisition direction is often observed with 3D seismic surveys and has been described as survey footprint noise by Marfurt et al. (1998). In the image this manifests itself as linear corrugations that are broadly parallel to the line acquisition direction. With the exception of one 3D data set on the shelf, all of the surveys in this area were shot with N E - S W trending lines and consequently this is also the direction of survey footprint. To minimize the overall effect of this artefact, the bathymetric surface was illuminated from the northeast. Despite static shifts between surveys and data artefacts the overall picture is remarkably good and shows many morphological features of interest to geologists.
6~
5~
/
One difficulty with this type of image comes when attempting to map individual features, because the image is as much a product of the chosen illumination direction and elevation as of the underlying topography. The choice of different illumination parameters will result in subtly different areas of shadow and light. To overcome this problem local dip magnitude and azimuth maps have been generated using digital terrain model filters. These parameters are independent of illumination direction and so are appropriate for sea bed morphology mapping. The generalized interpretation of the sea bed morphology given in Figure 1 is based on plots of local dip direction and magnitude (Fig. 3).
4~
.~Shetland~-~
3~W
2~
//~h /~ ~-:'~
I
1~
/"~/ ,~'.... 62~
';
.
0@~/
,
"I
/
/"
/" 61 ~
I
,
,
60~
'9 f E~
Terminal moraines of the last glacial episode
Downslope channels probably due to turbidite flows during glacial times
Area of intense iceberg scouring and reworking
Base of slope fans
./~. 9 .J
Escarpment lines of Tertiary erosion hollows- Judd Deeps
I~1
Areas where rockhead is at or close to seabed
-~
ediment waves from along slope currents of glacial age
A draped topography over buried Quaternary debris flows
.-i , .
"c1
pslope prograding mounds resultant from contour currents
Seafloor showing polygonal cracking close to seabed
L~.~ Area boundary of seabed image (Fig.2)
Debris flows of the last glacial episode
Fig. 1. Morphological interpretation of Figure 2.
FR']
Recent landslides - the Afen Slide and an as yet unnamed slide
Relict topography from deepwater scour hollows
8W
7W
6W
5W
4W
3W
2W
IW
6204
62N 62N
0
20
4O
Kilometres
6IN HSI Colour Scale 6IN
Hue
'Intensity Hue Scale B ~
1800
1350
900
50N
60N
450
0 metres water depth 7W
6W
5W
4W
3W
2W
IW
Fig. 2. Sea bed relief image illuminated from the northeast over the Faroe-Shetland Channel and adjacent areas. The image was generated from a merged bathymetry grid ( 100 m node spacing) created from the General Bathymetric Compilation GEBCO97 100 m regional contour data set, and a mosaic of depth converted two-way time horizons produced from 3D seismic surveys. Water velocity was assumed to be 1500 m/s. The bathymetry grid was rendered using ERMapper's 'shiny" algorithm that uses the Hue, Saturation and Intensity (HSI) colour model. The HSI model provides reflection highlights as well as shadow areas and is particularly effective in bringing out detailed surface texture. The water depth scale is calibrated to the Hue, not the final colour which is shown in the HSI ternary diagram.
SEA BED MORPHOLOGY DERIVED FROM 3D DATA
55
Fig. 3. Dip azimuth/magnitude image of the seabed pick. Dip azimuth direction is indicated by colour. Dip magnitude is indicated by greytone with dips of 4 ~ and greater in black.
Sea bed morphology General comments The sea bed image (Fig. 2) exhibits a range of morphological characteristics highlighted by the illumination. More detailed studies are possible and warrant correlation with other datasets such as site surveys, regional shallow seismic and sidescan sonar records and recent TOBI sonar surveys (Masson 1997, 2001). Although it is tempting to consider the image a bathymetric map, no accurate depth values should be extracted as the seismic frequency is inappropriate and survey boundaries show that errors are inevitable. It should be considered solely as relative changes in sea bed topography or morphology. The principal area covered includes the slopes on either side of the F a r o e Shetland Channel that extend from about 200 m water depth to 1000 m at the southern end and 1600 m at approximately 62~ The image also includes a small fraction of the shelf ( < 200 m) west of Shetland but none of the Faroese shelf. Images of the sea bed dip magnitude and azimuth (Fig. 3) are additional useful products of the sea bed image to be used in sea bed morphology interpretation. Such images show that the mean dip magnitude for the shelf area is less than one degree, and increases to two and a half on the slope, reaching five to six degrees locally.
Interpretation of selected morphological features Moraines On the shelf, large arcuate ridges are clearly evident in the vicinity of the Clair Field (60 ~ 45/N, 2 ~ 30/W) (Fig. 4). These are interpreted as glacial terminal moraines up to 1.3 km wide and are likely to be sites of poorly sorted, stony over-consolidated material. Three major moraines occur but, between the central and eastern moraines, similar but smaller features occur which may represent short-term stages in the retreat of ice from this part of the shelf, that may even be annual events. The ridges are
irregular in form and so other sedimentary processes such as sand waves are considered an unlikely explanation of these smaller features. Examination of high-resolution profiles (BGS regional surveys and site surveys for BP) show topographic features typically of less than 2 m amplitude. Sample evidence (BGS regional sampling) indicates hard diamictons locally with undrained shear strengths > 5 0 0 k P a in the uppermost 5 m. Interpretation on single profiles (Fig. 5) would not imply continuity, but the sea bed image (Fig. 4) supports such a geological explanation for them as minor moraines. A second area of moraines is partly imaged at the SW end of the image, on the shelf (60~ 4~ Although these moraines are less well defined on the sea bed image, in profile they are much larger than those to the north (Stoker & Holmes 1991).
Iceberg ploughmarks At the southern end of the study area the images indicate a chaotic sea bed with highly variable dip directions (Area A, Fig. 1). This may be a consequence of a weak sea bed reflector due to top mute being applied to suppress refraction events or it could be evidence of extensive sea bed scouring by icebergs on the outer shelf and topmost slope. Iceberg scouring becomes more 'organized' with increasing water depth evidenced by increasing length and more uniform orientation, sub-parallel to bathymetric contours.
Sediment waves and contourites At least two separate expressions of bottom-current activity are revealed by the image. On the slope, at the northern end of the survey area, there is extensive evidence of along slope sediment migration manifest as sediment waves, concentrated on the upper-middle slope, at about 4 0 0 - 6 0 0 m water depth (Area B, Fig. 1). These are of - - 1 . 5 - 2 k m wavelength but of low amplitude (about 5 m) with crests trending down slope indicating contouritic flow. Commonly, the w a v e f o r m breaks down into a more patch-like geometry. On seismic profiles, these bedforms are not very well expressed, and without the image their geometry is
56
D+ LONG ETAL.
Fig. 4. Sea bed relief image of the outer shelf illuminated from the northwest illustrating longitudinal topographic rises of various scales interpreted as glacial moraines.
difficult to discern. On the lower slope, between about 800 and 1000m water depth, a second area of along-slope bedforms is preserved (Area C, Fig. 1) (Fig. 6) (see also Knutz & Cartwright 2004). These appear as elongate mounds on the image, with distinct bifurcation of crests. These are larger (up to 30m) than the upper-middle slope features, are slightly oblique-to-slope, and display a fairly consistent trend of 037 ~. On shallow seismic profiles, they form a discrete package of long-lived contourite mounds that have migrated upslope (up to 1.5km in 300ms) throughout late Neogene time (Bulat & Long 2001). These features pass southwestwards into an area of smooth sea bed, which is identified by Masson (2001) from backscatter response on TOBI data, between 7 0 0 - 8 5 0 m water depth (61~ 2~ as a sheeted contourite.
Debris flows, fans and gullies Downslope processes have played a major role in the shaping of the margin. The dominant expression of downslope activity is debris flows (Area D, Fig. 1) (Fig. 7). Two main areas of debris flows occur on the West Shetland margin: (1) at the SW end of the margin, debris-flow deposits extend the length of the slope, partly infilling the Judd Deeps: and (2) on the upper-middle slope NW of Shetland. It is no coincidence that the debris-flow deposits lie immediately downslope from the moraines (described above), as they are linked to former icestream activity during intervals of ice-sheet expansion onto the shelf. Such ice sheets deposited large amounts of sediment directly onto the slope during stages of peak glaciation (Stoker 1995: Davison & Stoker 2002). The debris flows are typically
Fig. 5. Seismic (1 kJ sparker) section across large glacial moraines (M1, M2 and M3) and small glacial moraines (ml, m2 and m3). For location see Figure 4.
SEA BED MORPHOLOGY DERIVED FROM 3D DATA
Fig. 6. Sea bed relief image of the northern end of the West Shetland Slope illuminated from the west highlighting contourite mounds sub paralleling the sea bed contours and buried debris flows.
Fig. 7. Sea bed relief image illuminated from the northeast over the southern end of the Faroe-Shetland Channel illustrating the Judd Deeps and debris flows extending from the upper slope to the basin floor.
57
58
D. LONG ET AL.
Fig. 8. Seismic (Deep Tow Sparker) section through debris flows showing chaotically stacked packages (5-10 m thick. 2 km wide). For location see Figure 7.
Fig. 9. Sea bed relief image illuminated from the northeast over the middle West Shetland Slope illustrating debris flows extending to mid-slope and transforming into turbidity channels that extend to the base of slope, forming base of slope fans.
SEA BED MORPHOLOGY DERIVED FROM 3D DATA elongate and sinuous, and form a stacked association of lobes, which on seismic reflection profiles form distinct seismostratigraphic packages (Fig. 8). Borehole data have proved that the debris-flow deposits are of glacigenic origin (Davison & Stoker 2002). In the middle of the West Shetland margin ( - 6 0 ~ 40~N 3 ~ 40~W), the area of debris flow accumulation is linked by a series of sub-parallel, linear gullies on the m i d d l e - l o w e r slope (Fig. 9) to base-of-slope fans (Area E, Fig. 1) that include small units of flow deposits (Fig. 10). The origin of these gullies is uncertain, but almost certainly reflects a different style of meltwater and sediment delivery to the margin than is associated with the major debris-flow deposits described above. Comparable gullies have been described from the northern California margin (Spinelli & Field 2001). A base-of-slope fan development is also identified on the Faroese margin, although the limited coverage of the image restricts its interpretation.
Irregular patterns on the floor of the Faroe-Shetland Channel The northern end of the Faroe-Shetland Channel shows irregular patterns indicative of slight topographic rises (Area F, Fig. 1) (Fig. 6). These correlate with the location of buried debris-flow deposits, 100 to 200ms below sea bed, causing subsequent hemipelagic sediments to be raised.
Polygonal cracking On the floor of the Faroe-Shetland Channel in 1000 to 1200 m of water the seafloor exhibits a mottled surface resolved as a polygonal pattern (Area G, Fig. 1). This occurs below both the Faroese and West Shetland slopes. These features are typically 1 to 2 km across. Their geometry is comparable with features reported in this area but at depth (Davies et al. 1999). Examination of high-resolution seismic profiles indicates that fine scale faulting occurs close to sea bed. Examination of these profiles indicates that they are growth faults with vertical displacements of up to 4 m within the uppermost lOOms of sediment. Their presence on the sea bed image may indicate that these processes are on-going.
Landslide The most conspicuous features indicative of recent sedimentary processes is a single submarine landslide at 61~
59
2~ This feature, first identified on sonar (Masson 1997) is known as the Afen Slide (Area H, Fig. 1). It is approximately 3kin across and 13km in length with an excavated depth up to 20 m (Wilson et al. 2003). High-resolution profiles show it to have failed along several reflectors. Detailed examination of the SEG-Y datasets covering this feature together with processing techniques to reduce the marine static effects has produced a high-resolution sea bed image (Fig. 11) that shows that this feature is a multistage event, suggestive of retrogressive failure of the backscarp upslope and block failure on the northeastern flank. There is also clear evidence of sidewall failure on the southwestern flank. A smaller (1 km by 1.5 kin) slide occurs about 2 0 k m to the northeast alongslope.
Judd Deeps The Judd Deeps are one of the most dramatic features in the area, defined on the image by the scarpline of a 'waterfall' extending northwestwards for 17kin in the Faroese licensed area (see also Smallwood 2004), and was in the past even longer for its southeastern end is buried beneath debris-flow and contourite deposits. The scarp is evident by an area in shadow on the shaded relief map and locally is too steep for the sea bed to be resolved on seismic reflection data (Stoker et al. 2003). Southwest of these scarps the seafloor is smooth, rising gently upwards to water depths comparable with those upstream of the waterfall. In contrast the seafloor northeast of the scarp is uneven and this is seen more clearly when gridded at 25 m (Fig. 7). This suggests that rock head (Middle to Lower Eocene) is at or very close to sea bed (Area J, Fig. 1). The cuspate form of the waterfall appears to be associated with the areas of probable rock outcrop suggesting differential susceptibility to erosion. These deeps were probably formed in early Miocene time in response to vigorous bottom-current activity (Stoker et al. 2003). Further partly infilled scour hollows are evident to the northeast (Fig. 1).
Conclusions Sea bed features can be resolved from 3D exploration seismic data with tremendous detail (see also Austin 2004; Smallwood 2004). However, by combining the first return from several 3D exploration surveys the regional context of the sea bed morphology can be understood and potential
Fig. 10. Seismic (1 kJ sparker) section illustrating sea bed gullies (fixes 14-17) and base of slope fan sediment package (fixes 2-11). For location see Figure 9.
60
D. LONG ET AL.
Fig. 11. Sea bed relief image illuminated from the northeast over the Afen slide with (50 m) water depth contours superimposed. The image was generated from a 25 m two-way time surface that had additional processing applied to attenuate survey footprint artefacts while retaining image detail. The image was rendered using ER-Mapper's 'shiny' algorithm but with only the intensity layer active.
geohazards evaluated. The wider regional assessment is very important when considering the significance of features, e.g. the rarity of a certain seafloor environment or the presence of a potential landslide upslope from an operations area. Assessing sea bed morphology is essential to understanding the environmental setting and for safe operations in deepwater. Under favourable circumstances, the sea bed can be picked from 3D exploration seismic surveys in a similar manner to any other horizon of interest to provide detailed images of the seafloor, thereby negating the need for dedicated sea bed surveys. This work has been supported by various funding sources. BGS's science vote, the Western Frontiers Association (membership: Agip, Amerada Hess, BP, Conoco, Enterprise, ExxonMobil, Norsk Hydro, Shell, Statoil, Texaco, TotalFinaElf), the former Faroese GEM Network (membership: Agip, Amerada Hess, Anadarko, BPAmoco, Conoco, DONG, Elf, Enterprise, ExxonMobil, Marathon, Murphy. Phillips, Saga Petroleum F~royar, Shell, Statoil, Texaco, TotalFina and Veba Oil & Gas) and the Department of Trade and Industry (DTI). Sea bed data was supplied by member companies of the Western Frontiers Association and the former GEM Network or by geophysical contractors (Fugro Multi-Client Services, Horizon, PGS, Veritas and WesternGeco). All of whom are gratefully acknowledged, in particular, the geophysical contractors for the use of information from speculative surveys. The authors publish with permission of the Executive Director, British Geological Survey. NERC.
References AUSTIN, B. 2004. Integrated use of 3D seismic in field development, engineering and drilling: examples from the shallow section. bl: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S. A., LAPP1N, M. & UNDERHILL,J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 279-296. BULAT, J. & LONG, D. 2001. Images of the seabed in the FaroeShetland Channel from commercial 3D seismic data. Marine Geophysical Researches, 22, 345-367. DAVIES, R., CARTWRIGHT, J. & RANA, J. 1999. Giant hummocks in deep-water marine sediments - evidence for large scale differential compaction and density inversion during early burial. Geology, 27, 907-910. DAVISON, S. & STOKER, M. S. 2002. Late Pleistocene glaciallyinfluenced deep-marine sedimentation off NW Britain: implications for the rock record. In: O'COFAIGH, C. & DOWDESWELL, J. A. (eds) Glacier-Influenced Sedimentation on High-Latitude Continental Margins. Geological Society, London, Special Publications, 203, 129-147. IOC, IHO & BODC, 1997. 'GEBCO-97: The 1997 Edition of the GEBCO Digital Atlas'. published on behalf of the Intergovernmental Oceanographic Commission (of UNESCO) and the International Hydrographic Organization as part of the General Bathymetric Chart of the Oceans (GEBCO); British Oceanographic Data Centre, Birkenhead.
SEA BED MORPHOLOGY DERIVED FROM 3D DATA KNUTZ, P. C. & CARTWRIGHT, J. A. 2004. 3D anatomy of Neogene contourite drifts and associated mass flows in the Faroe-Shetland Channel. In: DAVIES, R. J., CARTWRIGHT,J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 63-71. MASSON, D, G. 1997. RRS Charles Darwin Cruise 101C Leg !, 05 Jun13 Jul 1996. TOBI surveys of the continental slope west of Shetland. Southampton Oceanography Centre, Cruise Report No. 6. MASSON, D. G. 2001. Sedimentary processes shaping the eastern of the Faroe-Shetland Channel. Continental Shelf Research, 21, 825 -857. MARFURT, K. J., SCHEET, R. M., SHARP, J. A. & HARPER, M. G. 1998. Suppression of the acquisition footprint for seismic sequence attribute mapping. Geophysics, 62, 1774-1778. POSAMENTIER, H. W. 2001. Lowstand alluvial bypass systems: incised vs. unicised. AAPG Bulletin, 85, 1771 - 1793. POSAMENT1ER, H. W. 2002. Ancient shelf ridges - a potentially significant component of the transgressive systems tract: Case study from offshore northwest Java. AAPG Bulletin, 86. 75-106. SMALLWOOD, J. R. 2004. Tertiary inversion in the Faroe-Shetland Channel and the development of major erosional scarps. In: DAVIES, R. J., CARTWRIGHT,J. A., STEWART, S. A., LAPPIN, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London. Memoirs, 29, 187-198. SPINELLI, G. A. & FIELD, M. E. 2001. Evolution of continental slope gullies on the northern California margin. Journal of Sedimentary Research, 71, 237-245. STOKER, M. S. 1995. The influence of glacigenic sedimentation on slope-apron development on the continental margin off NW Britain. In: SCRUTTON,R. A., STOKER,M. S., SHIMM1ELD.G. B. &
61
TL'DHOPE, A. W. (eds) The Tectonics, Sedimentation and Palaeoceanography of the North Atlantic Region. Geological Society. London, Special Publications, 90, 159-177. STOKER, M. S. 2002. Late Neogene development of the UK Atlantic margin. M: DORE, A. G., CARTWRIGHT. J., STOKER, M. S., TURNER, J. P. & WHITE, N. (eds) Exhumation of the North Atlantic Margin: Timing, Mechanisms and hnplications for Petroleum Exploration. Geological Society, London, Special Publications, 196, 311-329. STOKER. M. S. & HOLMES, R. 1991. Submarine end-moraines as indicators of Pleistocene ice limits off NW Britain. Journal of the Geological Socieo', London, 148, 431-434. STOKER, M. S., HITCHEN. K. & GRAHAM, C. C. 1993. United Kingdom Offshore Regional Report: The Geology of the Hebrides and West Shetland Sheh'es, and Adjacent Deep-Water Areas. HMSO for the British Geological Survey, London. STOKER, M. S., Lo.~o, D. & BL'LAT, J. 2003. A record of mid-Cenozoic strong deep-water erosion in the Faroe-Shetland Channel. hi: MIENERT, J. t~: WEAVER, P. (eds) European Continental Margin Sedimentary Processes: An Atlas of'Side-Scan Sonar and Seismic hnages. Springer, Berlin, 145-148. TATE. M. P.. DODD, C. D. & GRA.~T, N. T. 1999. The Northeast Rockall Basin and its significance in the evolution of the RockallFaroes/East Greenland rift system. In: FLEET, A. J. & BOLDY, S. A. R. (eds) Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference. Geological Society, London, 391-4O6. WILSON, C. K,, LONG. D. & BULAT, J. 2003. The Afen Slide - a multistage slope failure in the Faroe-Shetland Channel. In: LOCAT, J. & Mienert, J. (eds) Submarine Mass Movements and their Consequences. Advances in Natural and Technological Hazards Research Series. Kluwer, Dordrecht, 317-324.
3D anatomy of late Neogene contourite drifts and associated mass flows in the Faroe-Shetland Basin PAUL
C. KNUTZ
1"2 & J O S E P H
A. C A R T W R I G H T
2
t Geological Institute, Universi O, of Copenhagen, Oster Voldgade 10, DK-1350, Copenhagen, Denmark (e-mail: kn utz @geol. ku. dk) 23DLab, School of Earth, Ocean and Planetary Sciences, Cardiff Universit3', Main Building, Park" Place, Cardiff CFIO 3YE, UK
Abstract: We have combined 3D mapping of key reflectors with seismic profiles to describe the geometry and anatomy of contourite drifts formed by deep waters passing through the oceanic gateway of the Faroe-Shetland Channel. The West Shetland Drift complex is characterized by sheeted-mounded drift units, and upslope migrating sediment waves accreting over an early Pliocene unconformity. The basin section is constructed by a series of asymmetric depositional units of early Pliocene-Pleistocene age, interlayered by three mega-debrite sequences that extend into the basin. The Pliocene drift surface display an enhanced topography of bifurcating moat-channels that tend to branch out in a southwest direction. Along the lower slope a succession of upslope migrating sediment waves has accumulated from the Pliocene drift topography. These features extend to the present sea bed at water depths of 700-1000 m where they appear as a series of linear, bifurcating ridges. The high accumulation rates of the West Shetland Drift since the early Pliocene transition and the formation of upslope migrating sediment waves is related to a sustained flow of Norv,egian Sea deep waters and cross-slope transport of finegrained sediments from the NW European shelf.
Contourite drifts deposited by deep ocean currents are a common feature of the North Atlantic margins. Large elongate contourite drifts have built up along the pathway of bottom currents entering the North Atlantic through the narrow conduits across the Greenland-Scotland Ridge (Heezen et al. 1966; Kidd & Hill 1986; McCave & Tucholke 1986). Accumulation of contourites occurs preferentially along the fringe of the western boundary currents that convey North Atlantic Deep Water toward the Southern Oceans as part of the global thermohaline circulation. Geostrophic bottom currents capable of mobilizing and transporting silt size sediments (velocities > 10-15cm/s) are commonly observed on modern slope setting while movement of fine sand-size material requires extreme flow conditions ( > 30cm/s) (McCave et aL 1995). The value of understanding the structure and depositional process of deep sea contourites lies mainly in their application as high-resolution palaeoclimatic recorders. Commercial interest in these predominantly finegrained deposits has been limited although this may change as the ocean margins are being increasingly explored for natural resources. Despite the common occurrence of contourite drifts in the Cenozoic marine record, relatively little is known about their depositional mechanism in comparison to gravity driven sedimentary processes. Most evidence of contourite drift formation and alongslope-downslope process interaction is based on seismic data because direct sedimentological approaches are hindered by the immense sizes of modern contourite systems and the poor representation of ancient contourite deposits in outcrops (Stow et al. 1998). The seismic expression of contourites includes a range of geometries and depositional patterns (McCave & Tucholke 1986; Faugeres et al. 1999). On the basin scale they are classified according to their external morphology as sheeted (abyssal, plastered and patch), elongated (detached and separated) and channel related drifts. Internal stacking of depositional units (progradation-aggradation) and seismic facies characteristics (reflector amplitude, continuity and configuration) can provide information on drift accumulation, migration and downslope-alongslope process interaction. The seismic characterization of contourite drifts has up until recently been limited by the lack of detail in conventional 2D
surveys. The implications are that important spatial depositional patterns can be missed and that depositional models may fail to resolve the complex interaction between sea bed topography, alongslope currents, and downslope sedimentary processes. We present results from a 3D seismic survey in the Faroe-Shetland Basin, which illustrates the detailed anatomy of contourite drifts deposited on a major late Neogene unconformity and their relationship with intervening mass flow units. An underlying aim of this paper is to demonstrate how the enhanced resolution of 3D seismic data has the potential to advance our understanding of alongslope-downslope process interaction on continental margins.
Regional background and approach The Faroe-Shetland basin has formed a sediment trap since the onset of rifting during the Late Cretaceous (Dean et al. 1999). The main phase of basin subsidence followed a compressional event at 5 6 - 5 7 Ma BP associated with the onset of rifting in the North Atlantic. The Cenozoic evolution of the basin involves several phases of intraplate subsidence and tectonic compression which led to the fonnation of the W y v i l l e - T h o m s o n Ridge during the late Palaeocene-Miocene interval (Boldreel & Andersen 1993). The ridge forms an oceanographic sill between the Faroe-Shetland basin and the Rockall Trough as part of the Scotland-Greenland ridge structure that separates the North Atlantic and the Nordic Seas (Vogt 1972). The present morphology of the Faroe-Shetland basin (Figs 1 & 2) was essentially completed during a late Neogene tectonic phase, which caused uplift on the NW European margin and subsidence along the axis of the basin (Andersen et al. 2000: Stoker in press). The physical barrier of the Greenland-Scotland sill allows the establishment of a permanent thermohaline gradient between the Nordic Seas and the North Atlantic that is balanced by a geostrophic flow of Arctic deep waters through the Denmark Strait and Faroe-Shetland Channel (Vogt 1972: Dickson & Brown 1994). Overflow waters derived from Norwegian Sea Deep Water enter the North East Atlantic basin through the F a r o e - B a n k Channel (sill depth - 8 0 0 mJ where it contributes to about a third of the total flux of North Atlantic Deep Water
DAVIES,R. J., CARTWRIGHT,J. A., STEWART,S. A., LAPPIN,M. & UNDERHILL.J. R. (eds) 2004.3D Seismic Technology:Application to the Exploration of Sedimentar)' Basins. Geological Society, London, Memoirs, 29. 63-71. 0435-4052/(14/$15 9 The Geological Society of London 2004.
64
P.C. KNUTZ & J. A. CARTWRIGHT
Fig. 1. Isochron map of the West Shetland drift (WSD) based on the two-way time difference between the glacial unconformity and the Intra-Neogene unconformity (Fig. 2). Basin and slope components of the drift body are indicated. Contour interval is 50 ms. In the far NE comer the WSD intercalates with the North Sea Fan. Grey box demarcates the area of the 3D seismic survey. White box indicates the position of 3D map and seismic profiles in Figure 7 (expanded in the lower left comer). Inserted map show the location of the seismic study in the context of the bathymetry and deep water flow (blue arrows) in the FaroeShetland Channel (FSC). Contours represent 200 m depth intervals with 800 m contour depth hatched. WTR, WyvilleThomson Ridge. (Fig. 1). Seismic stratigraphic analyses of sediment drift deposits suggest that the Faroe-Shetland deep water gateway has been active since the early Oligocene (Davies et al. 2001). The study area is located in the northeast sector of the Faroe-Shetland Basin (Fig. 1). The bathymetry of this interfan region is characterized by a well defined shelf break on the Shetland margin at about 200m water depth from where the slope dips at an average of 2 - 3 ~ to the floor of the Faroe-Shetland Channel at 1 2 0 0 - 1 5 0 0 m water depth.
The gentle inclination of the West Shetland slope margin contrasts with the steep slope (up to 5 - 6 " ) of the Faroese margin. The present hydrographic regime in the channel is characterized by oppositely flowing surface and bottom water masses separated by a strong pycnocline at water depths of 5 0 0 - 7 0 0 m . Deposition of fine-grained sediments is presently hindered by a strong southward flow of Norwegian Sea Deep Water with weakly mean near-bottom current speeds recorded between of 2 0 - 4 0 c m / s (Akhurst 1991). The modern conditions are in contrast to the last glacial when periodic reductions in deep water circulation and expansion of ice sheets to the shelf margins, allowed widespread deposition of glacial-marine sediments across the F a r o e Shetland channel and adjoining slopes (Akhurst 1991; Rasmussen et al. 2002). Commercial 2D and 3D seismic data from the northern Faroe-Shetland Basin were interpreted to reveal the structure of Neogene drift deposits in the upper 500ms TWT interval. Detailed mapping and seismic imaging within a 3D volume was carried out using Schlumberger IESX interpretation, mapping and visualization tools. The vertical seismic resolution within the 3D area is estimated to be 10-15 m while the horizontal resolution displayed in the surface maps is on the order of 100-200m. An average acoustic velocity of 1800m/s was applied to convert TWT into depth.
The West Shetland Drift
Fig, 2. (a) Seismic cross-section (Line 1 in Fig. 1) of the northern Faroe-Shetland basin with biostratigraphy derived from well 214/4-1 (Davies & Cartwright 2002). (b). Detailed insert illustrates the slope and basin section of the West Shetland Drift resting on the Intra-Neogene Unconformity (INU). GU, Glacial Unconformity. The vertical scale is two-way travel time in seconds.
The West Shetland Drift (WSD) complex forms a sedimentary prism of two asymmetric bodies that trail the axis and southeastern flank of the Faroe-Shetland Basin (Figs 1-3). Internally the drift package is characterized by smooth continuous reflectors forming mounded, asymmetrical depositional units that are typically associated with deposition by alongsiope bottom currents (Faugeres et al. 1999). The drift package is bounded at the base by a regional unconformity of Late M i o c e n e - E a r l y Pliocene age (Intra-Neogene Unconformity, INU) and at the top by a prominent glacial unconformity
NEOGENE CONTOURITES, FAROE-SHETLAND BASIN
Fig. 3. Seismic profile (Line 2 in Fig. 1) across the southern section of the West Shetland Drift. Here the slope and basin sections of the drift are separated by a broad zone of reduced deposition, The WSD slope section builds up from an alongslope trending depression that has incised the underlying Palaeogene sedimentary succession. INU, IntraNeogene Unconformity; TPU, Top Palaeogene Unconformity; BSR, Bottom Simulating Reflector. The vertical scale is two-way travel time in seconds.
(GU) (Fig. 2). The glacial unconformity is developed within a glacimarine progradational wedge that extends from the shelf margin (Stoker 1997; Stratagem-partners 2002) while downslope it fades into conformity with the basinal sedimentary succession. In the basin the WSD forms a succession of sheeted to mounded depositional units that have accreted along the steep base of the slope (Fig. 4). Updip the basin section of the WSD thins into a condensed succession of climbing sediment waves that onlap onto a distinct, convex sedimentary body that itself show downlap and onlap onto the basal unconformity (Fig. 2b, details in Figs 10 & 11). This mid-slope section of the WSD can be traced on seismic profiles from the outlet of the North Sea Fan and for about 250 km to the SW along the West Shetland margin, with a thickness of 2 0 0 - 4 0 0 m along its axis (Fig. 1). Detailed mapping of key horizons performed on the available 2D and 3D data allows a seismic-stratigraphic correlation between the upper and lower limb of the WSD (Figs 10 & 11). The stratal relationships and inferred unit ages of the WSD complex are schematically illustrated in Figure 5. Five seismic units A, B 1, B2, C and D showing the characteristic asymmetric and lenticular geometries of contourite drifts have been
65
Fig. 5. Stratigraphic summary and cross-slope stratai configuration of the West Shetland Drift based on 2D and 3D seismic profiles (compare with Figs 1~4 & 10). INU, Intra-Neogene Unconformity: TPU, Top Palaeogene Unconformity; GU, Glacial Unconformity. Grey interbedded units represent debris flow packages 1-3. Detail of the onlap relation between W S D b a s i n and WSDsiop~ is shown in the inserted box. Unit A and B of the WSD slope section show onlap onto the INU while unit C is interdispersed with shelf progradational units.
identified above the INU reflector. The drift units are intersected by three mega-debrite packages characterized by stacked toethrusts and internal chaotic seismic facies. Based on correlation of seismic reflectors to well 214/4-1, which was published by Davies et al. (2001), these units are of Pliocene-Pleistocene age. The most pronounced drift accumulation is represented by units A - B which, according to the biostratigraphy from well 214/4-1, were deposited during the early Pliocene-early Pleistocene (4-1 Ma). We emphasize that the late Neogene stratigraphic framework proposed in Figure 5 is preliminary and should be tested by future drilling through the expanded sections of the WSD. The seismic architecture of the WSD provides a mean of inferring the structure and flow direction of the water masses that prevailed during the late Neogene (Knutz & Cartwright 2003). The elongated depositional trend of the mid-slope drift section, in particular the narrowing and thinning of the sedimentary body toward southwest (Fig. 1), suggests a flow of deep alongslope currents originating from the Nordic Seas. This interpretation is supported by the southwestward migration of drift units B1 - B 2 (Fig. 9) and the bifurcating pattern of moat channels observed on
Fig. 4. Seismic cross-section (Line 3 in Fig. 1) showing the seismic-stratigraphic relationship between the basin and the lower section of the Shetland slope, The late Neogene succession, resting on the Intra-Neogene Unconformity (INU), is constructed by contourite drifts, characterized by highamplitude reflectors with sheeted to mounded geometries, intercalated by three major debris flow units (DF I-DF 3) revealed by low-amplitude hummocky/chaotic seismic facies. Upslope accreting drift deposits on have been accommodated by a large slump structure within the underlying Oligocene-Miocene sediment pile. Updip the INU merges with two underlying onlap surfaces: the Mid-Miocene Unconformity (MMU) and the Top Palaeogene Unconformity (TPU). BSR, Bottom Simulating Reflector. Scale bars in metres.
66
P.C. KNUTZ & J. A. CARTWRIGHT
the 3D seismic (Fig. 7), The WSD is likely to have formed in a water mass structure similar to the modern thermohaline circulation regime where northward directed Atlantic surface waters overlie a southern counterflow of Norwegian Sea Deep Water (Dooley & Meincke 1981 ; Tun-ell et al. 1999), The top of the mid-slope drift section may (at any given time) correspond to the boundary between oppositely flowing surface/intermediate and bottom waters (Knutz & Cartwright in press). In the modern oceanographic scenario this water mass boundary intersects the Shetland slope at water depths of 400 - 600 m (Dooley & Meincke 1981) which corresponds to a decrease in seafloor gradient at a midslope position above the upper flank of the WSD (Fig. 2),
3D seismic mapping The I n t r a - N e o g e n e U n c o n f o r m i t y The Intra-Neogene Unconformity appears as a distinct horizon that can be traced below the WSD across the northern FaroeShetland basin and upslope onto the Shetland margin (Figs 2 & 3). On the slope and margin the INU forms an angular unconformity but the erosional expression disappears in the northern part of the basin. Here the overlying sequence appears to be conformable except were the INU intersects locally with dome structures of the underlying Miocene sediment pile. An early Pliocene age of the basal unconformity has been proposed based on seismic correlation to well 214/4-1 in the basin (Davies et al. 2001) and to British Geological Survey boreholes on the shelf margin west of Shetland (Stoker 2002). The topography of the INU imaged within the 3D survey shows a pronounced change from a low hummocky relief in the basin to a blockfaulted and heavily incised surface along the slope-basin transition, generated by slumping of the underlying sediment pile (Figs 4 & 6). The hummocky surface in the basin has previously been related to large-scale differential compaction and density inversion formed during the early burial stage (Davies et al. 1999). The slumped section in the central part of the survey area appears to have accreted downslope as a large rotational mass movement with an underlying unconformity of late Palaeogene age (Top Palaeogene Unconformity, TPU) forming a glide plane (Figs 4 & 6). Internally the slumped section displays a mass of contorted, but essentially preserved, stratified units that are intersected by a bottom simulating reflector (Fig. 4) related to diagenetic precipitation of opal C/T (Davies et al. 1999). The INU is observed to merge with two underlying unconformities in an updip direction: the MidMiocene Unconformity which onlaps the Palaeogene-Early Neogene sediment package in the basin, and the TPU which intersects the INU at the slope base (Fig. 4). Further upslope the composite I N U - T P U horizon is marked by slope parallel depressions related to truncation of the NW dipping E o c e n e Oligocene strata (Figs 3 & 4). The enhanced expression of this relief is related to winnowing and erosion by alongslope bottom currents.
Contourite drifts The initial depositional phase (unit A) of the WSD-basin section mainly occurred as infilling of topographic irregularities of the INU surface (Figs 6 & 9), In the central part of the basin drift unit A is also observed as lenticular drift bodies developed locally in the front of the steep slump scarps in the central region of the 3D area (Fig. 4). This depositional pattern contrasts the initial development of the WSD-slope section where unit A appear to form a substantial part of the thick sedimentary bulge (Figs 5 & 11).
Fig. 6. Details of the topographic relationship between drift units (top B 1 and B2 marked by arrows) and the Intra-Neogene Unconformity. The relief is enhanced by illumination from SE. Section (a) shows a view from the basin toward the escarpment formed by the slump structure in the central part of 3D area. The vertical bar represents -- 120 m while the alongslope oriented profile is - 18 km long (see Fig. 7 for line orientations). Section (b) illustrates a view along the base of the slope featuring the build up of contourite drift units from slum generated depressions, Note the southwestward migration of reflectors in units BI-B2. The scale bar to the right is --230m.
The B 1 surface forms an accentuated topographic relief of mounded contourite drifts and intersecting moat-channels, as well as depositional ridges that have built up from the underlying erosional surface (Figs 6 & 7). The morphology of the B I drift indicates a directional change in relief from mounded drift bodies at the base of the slope to elongated moatridge systems in the upslope section. The most pronounced moat-channels at the base of slope display channel widths up to i 500 m and depths up to about 80 m and tend to bifurcate toward SW in an oblique downslope direction. The maximum thickness of the mounded drift bodies comprising units A - B 1 is 180m. Upslope from the moat-channel complex a series of four ridgemoat systems can be identified as part of the succession of sediment waves that connect the basin and slope section of the WSD (Figs. 7 & 10). Moat widths are here observed at 3 0 0 600 m, with depths of 10-40 m. The moat-channel systems are related to zones of reduced sediment accumulation where bottom currents were strongest (inferred bottom current pathways are marked by arrows in Fig, 7a). These are flanked by regions of aggrading strata where waning bottom current energy favoured deposition. Unit B2 forms a prograding-aggrading unit that tends to attenuate the relief of the B I surface. The internal reflector configuration of the B 1 - B 2 succession indicates that the system has migrated ups]ope toward the
NEOGENE CONTOURITES, FAROE-SHETLAND BASIN
67
Fig. 8. Dipmap of present sea bed surface. Notations and area coverage as in Figure 7. to the low-relief m o d e m seafloor the sedimentary succession represents a gradual smoothing of the enhanced topography of the e a r l y - m i d d l e Pliocene (Fig. 9).
Sediment
waves
The present sea bed above the B I surface is characterized by a series of ridge-channel systems that trend alongslope in a pattern of parallel iinearity (Fig. 8). The spacing between wave crests is irregular but generally in the order of 5 0 0 - 1 0 0 0 m . Individual r i d g e - c h a n n e l segments vary in length between a few kilometres to > 2 0 k i n . Wave heights are likewise variable with m a x i m u m values of 3 0 - 4 0 m. The wave field is observed along the mid-lower slope at 7 0 0 - 1 0 0 0 m water depth where the average seafloor gradient reaches ~ 3 ~ while locally, on upslope facing levees, dips of up to 10 ~ are observed (Fig. 11). See Long et al. (this volume) for a description of the seafloor topography of the entire F a r o e - S h e t l a n d Channel. Seismic cross sections show that the m o d e m sea bed features form part of the succession of upslope accreting sediment waves that have built up from the Pliocene contourite topography (Figs 10 & 11). Although some of the m o d e m sediment waves appear to have accumulated in continuity with the Pliocene depositionai surface, it is evident that the number of wave set increases upward through the strata. The most prominent of the m o d e m ridges appear to have aggraded from the B1 ridge structures while other linear features have evolved during deposition of units C and D (Fig. 11). The continuity of unit bounding reflectors suggests that the sediment waves are laterally coherent except where they onlap onto an erosional discontinuity. An example is shown in Figure 11 where
Fig. 7. (a) 3D image of the of the B 1 surface mapped within the 3D volume (see location in Fig. 1). The colour scale ranging from red to purple represents two way travel times between 1.4 and 2.3 s. The topography varies from a mounded relief with bifurcating moats at the base of slope to linear ridge-moat features on the updip section. (b) Dipmap of the B I surface. The grey scale shows relative changes in gradient (s m 1) with black representing dip angles > 7~ The dip attribute is highly sensitive to minor changes in relief and allows recognition of the ridge -moat features that cover the steepest part of the slope. The basinward border of the main moat-channel is shown by the white hatched line. White arrows denote inferred bottom current pathways. The positions of seismic profiles shown in Figures 4, 6 & 9-11 are indicated by orange lines. Other profiles shown in Figure 6 are marked by black hatched lines. Bold numbers refers to prograding ridges profiled in Figure 11. The striped area at the northern boundary demarcates the limit of a slide that truncates the B ! reflector. southeast as well as alongslope toward the southwest (Fig. 9). Units C and D have generally accumulated in continuity with units B 1 - B 2 except at areas along the base of slope where the top of unit B2 has been disrupted by incision and infill associated with DF 2 (Fig. 12). From the B 1 horizon and upward
SW migration of contourite units ,-
Fig. 9. Seismic profile showing the seismic-stratigraphic structure of the mounded drift deposits at the base of slope (line position is indicated in Figs 6 & 7). Notations are as in Figure 5. Scale bars in metres.
~"
9, ~ " . . , .
;,,..4.
~
-
~
,,,,/i.;,-~::~
DF2
~
,.,
~"
.- ~
,31
61~ N
r 35 Cl, 10s
7 0 degree dip) features such as the feeder dykes described by Francis (1982) and Chevallier et al. (2003). This is compounded by the
9 The sills were emplaced generally along stratigraphic bedding planes, resulting in a saucer shape in the Midland and Whin examples (also noted in the Karoo basin by Du Toit 1905 a, b). 9 Maximum sill thickness typically coincided with the bottom of the sedimentary basin. 9 At the upper edges of the saucers where the sills become thin and terminate, the sills typically climb up section and to dissociate into lenses. 9 Interestingly, Francis noted that the feeder dykes were located on the flanks of the saucers or basins, and absent in the areas of greatest thickness and depth. In the Midland Valley example, the tholeiitic feeder dykes are 10 km from the basin depocentre and the area of greatest sill thickness (120m). 9 Francis concluded that the principal emplacement mechanism was by gravitational flow downdip from feeder dykes on the margins of the sills. Sill geometries in the Jurassic Karoo basin in South Africa (Du Toit 1905 a, b, 1920; Chevallier et al. 2003) are strikingly similar to those seen in the Gjallar 3D seismic survey, with equivalent age and depth relationships to the basin sediments relative to the time of intrusion. The Karoo basin is a foreland basin and therefore does not contain tilted fault blocks, thus intrusions along faults and dipping strata are not represented. Karoo sill geometries found at intrusive depths similar to the Gjallar examples include saucer-shaped sills, stacked saucers, sills climbing up-section, and 'feeder' dykes interconnecting sills emplaced at different stratigraphic levels, particularly near the edges of saucers. Sills typically exceed 50 m in thickness, while dykes are typically thinner than 50 m. The Karoo sills and dykes form major barriers and conduits in the hydrological system,
Fig. 7. The relationship between sills, tilted fault blocks and diapirs. A sill in the deep Cretaceous section at (a) has been interpreted using autotracking techniques. Note the hom-like projections along its upper margins. An uninterpreted sill occurs at (b). A diapir emerges from the crest of the tilted fault block and extends upwards into the Miocene section (c).
182
S.M. CORFIELD E T A L .
constraints imposed by the increasing wavelength at depth and the consequent increase in the width of the Fresnel zone. At a depth of 4 - 5 s TWT in the Gjallar area ( - 6 kin), and assuming that the migration of the data is optimized, features less than 60 m in lateral extent (i.e. feeder dykes) are unlikely to be imaged.
Shallow structure: diapirs and polygonal faults The emplacement of oceanic crust in the Early Eocene (Chron 24b, Dor6 et al. 1999) in the proto-North Atlantic resulted in the cessation of rifting in the Gjallar area and the onset of relatively uniform subsidence. Deep water conditions with predominantly fine-grained sedimentation prevailed throughout the Eocene and continue to the present day. A number of mounded features occur at the level of the top Palaeocene reflector (Fig. 6). These are preferentially developed above the crests of tilted fault blocks and are 5 0 - 1 0 0 ms TWT high and 2 0 0 - 4 0 0 m wide. The mounds are commonly draped by the overlying Eocene reflectors and can be interpreted as either volcanic vents or mud volcanoes. Similar features were described by Gevers (1928) in the Karoo Basin, South Africa. In addition to vents composed of varying amounts of basaltic lavas, agglomerate and tufts, he noted that fine-grained sandstones and mudstones also occurred. There also appeared to be a close genetic relationship between the vents and the underlying igneous intrusions, both of which occurred in the area of thickest sediments in the centre of the basin. Similar mounds have been observed on 3D seismic data by Davies et at. (2002) who describe the architecture of a shallow magmatic sill-dyke-mound complex in the FaroeShetlands Basin. They interpret the conical mounds as volcanoes accreted on the seabed, directly above the tips of the basaltic dykes, with an age of 5 4 . 9 - 5 4 . 6 M a . This interpretation is based on the association with the sills/dykes, a vertical chimney of disturbed seismic data linking the two. the well-organized, lobate, 'onion-skin' internal geometry of the mounds, and their high acoustic impedance relative to overlying shales. It is impossible to determine the nature of the vent-filling materials in the Gjallar area. However, there appears to be a lack of a dramatic velocity pull-up beneath the mounds that would indicate dense, high-velocity basaltic material. This observation, combined with the fact that the Palaeocene sills are intruded into ductile Cretaceous mudstones, points to conclusion that the majority of the mounds are probably composed of mudstones
mobilized by hot fluids and gases generated during sill emplacement. This conclusion is further supported by the presence of mud diapirs in the overlying Tertiary section. At least eleven diapirs occur in the area of the 3D data and, as with the mounds, all are located above the crestal areas of tilted fault blocks defined at the Palaeocene level. In the area of the Gjallar Ridge, these chimney-like features are typically associated with a dense network of normal faults, some of which extend to the sea bed (Fig. 8). The most impressive diapiric feature is located above the crestal area of the structurally highest Palaeocene tilted fault block on the Gjallar Ridge (Figs 2 & 8). This feature consists of two diapirs or chimneys, spaced about 4 km apart, each characterized by a zone of discontinuous to chaotic reflectors about 500 m in diameter. At the sea bed (Figs 9 & 10), the diapirs form a NE-elongate mound approximately 10km long and 5 km wide with a maximum vertical relief of 20 ms TWT (c. 15 m). A central, NE elongated depression is flanked by a discontinuous, similarly oriented high (Figs 10 & 11). The N E SW elongation of the feature appears to be related an underlying fault, the sea bed expression of which can be seen to the NE of the diapir in Figure 10. Seismic-stratigraphic relations along the diapir/chimney are characterized by a vertically repeating sequence of packages that alternately thicken towards the chimney, and thin towards the chimney with onlapping reflector terminations (Fig. 8). This 'Christmas tree' pattern indicates a long history of episodic activity. Many of the diapirs and faults have evidence of gas migration in the form of localized, very high-amplitude reflections (Figs 2, 8 &l 1). The zone immediately beneath the sea bed is characterized by unusually weak reflectors. Therefore, the localized, high-amplitude gas anomalies are easily visualized by manipulating the opacity of a 3D seismic volume converted to voxels (Fig. l 1). The antiformal reflector geometry can not be attributed to a velocity effect from gas or low-density mud; indeed, if the chimney is characterized by low seismic velocities induced by the presence of gas the mound-like geometries are more acute than shown on the seismic data. The faults cutting the Tertiary section are relatively steep and planar in cross-section (Figs 2 & 8). In map view they form an approximately polygonal network of faults similar to those described by Cartwright (1994) in the Tertiary of the North Sea (Figs 9 & 10). Cartwright proposed that the North Sea examples are not the result of horizontal extension but are the result of volume loss due to catastrophic dewatering of the Tertiary section.
Fig. 8. Detail of the Tertiary section in the NE part of the survey. A diapir extends from the Palaeocene section to the sea bed in the NW part of the image and is characterized by locally discontinuous, high-amplitude reflectors interpreted as gas. Relatively recent uplift of the Ridge is indicated by an onlapping wedge of reflectors of probable Lower Pliocene age and the progressive westerly truncation of reflectors beneath the base Upper Pliocene unconformity.
3D SEISMIC VISUALIZATION OF THE GJALLAR RIDGE
183
Fig. 9. Time-dip maps of the sea bed, box encloses area shown in Figure 10. (a) raw map; (b) illuminated from NW; (c) illuminated from SE. Spectacular sea-bed deformation has also been observed above the Vema Dome to the east of the Gjallar Ridge, and it has been also attributed to mud diapirism by Hovland et al. (1998). Shallow coring of a diapir located over the Vema Dome encountered Eocene diatomaceous ooze below a 0.6 m cover of Pleistocene glacio-marine sediments, confirming the deep origin of the diapirs. DSDP (Deep-Sea Drilling Project) Leg 38 drilled similar diapiric Eocene ooze associated with methane on the landward side of the Outer VOting Plateau (Hovland et al. 1998). Hovland et al. (1998) proposed three mechanisms for the growth of the diapirs in the Vema area: 9 a buried, low density, high-porosity layer of deformable material (Eocene ooze); 9 a doming substratum which causes extension and faulting: 9 migration of light hydrocarbons focused into the diapir. The periodic 'growth' of the Gjallar diapirs began near the time of the breakup unconformity, i.e. in the Eocene or Oligocene. Hence we infer from the seismic stratigraphy that the mobile material is of a similar age to Hovland's Vema example, and that it also contains mobile diatomaceous ooze. Hovland et al. (1998) concluded that it was the onset of basin inversion that resulted in the formation of the Vema Dome that triggered the growth of the diapirs. In contrast, along the Gjallar Ridge, the diapirs correspond closely to the crests of tilted fault blocks, as well as to the up-dip terminations of sills, strata and faults (Fig. 12). These dipping structures are significant in that they would have focused migrating hydrocarbons and pore
fluids towards the tilt-block crests thus localizing diapir formation and evolution. This is particularly true of sealing faults and facies, such as the sills, and dykes intruded along faults. We propose a model that, during the Palaeocene, pore fluid was expelled from the sediments adjacent to the newlyemplaced sills, resulting in the formation of epithermal to hydrothermal fluid chimneys and vents at the palaeosurface, generally on the crests of the fault blocks (Fig. 6). Commencing with the Eocene deposition of deep marine mudstones and the cooling of the magmatic system, these vents were transformed into cooler-water fluid-escape chimneys or diapirs. We infer that the mobility of the Eocene ooze was probably enhanced by the migration of light hydrocarbons through the chimneys, a process that appears to be continuing at the present day. The original high-permeability pathways through otherwise sealing sedimentary facies or faults may have utilized the fractured margins of magmatic intrusions (e.g. Chevallier et al. 2003) or may have remained propped open by sand grains carried by the escaping pore fluids, in a manner analogous to clastic dykes, or by minerals deposited by the epi- to hydrothermal fluids. Considering the heat source, these propping mechanisms are not likely to affect strata younger than Late Eocene. We infer the formation of the thick ( > 1 km) package of 'polygonally' faulted post-rift sediment to be intimately linked to the chimneys/diapirs in several ways. The chimneys provide clear loci for the venting of laterally migrating pore fluids throughout the Tertiary and Upper Cretaceous section, obviating migration along otherwise sealing faults. Secondly, with increasing loading, the chimneys could control compaction and de-watering of deeper sediment. Although we see no evidence for the migration of significant volumes of deeper sediment up through the diapir system, we postulate that this may have occurred, with the products being widely redistributed by sea-floor currents. If so, migration of deep mobile sediment could also be responsible for a significant part of the 'polygonal' fault pattern observed. The link between lateral migration of fluids and sediment at depth and the overlying polygonal fault systems is indicated by the fact that the only diapir to have reached the present-day sea bed is also associated with recently active polygonal faults that deform the sea bed. It is probable that relatively recent compaction and fluid flow in the immediate vicinity of the diapir is responsible for the continued fault activity.
Summary Fig. 10. Detail of Figure 9. A NNW-trending group of pockmarks occurs at (a). The pockmarks, inferred to result from gas and light hydrocarbon escape, are typically about 1500 m in diameter and 15 m in relief. The feature at location (b) is interpreted as the sea bed expression of a mud diapir. Note the NE-SW elongation of the feature, parallel to the zone of faults to the NE (c). Networks of polygonal faults (d) extend from the sea bed to depths of over 1.5 kin. Note that the faults are most well developed in the immediate vicinity of the diapir.
There are a number of major points to emphasize from this study: 1. The uplift of the Gjallar ridge occurred in several phases. The first phase was in the Lower Cretaceous, well before the onset of Late Cretaceous extension, and may have resulted from local differences in thermal subsidence after the Jurassic rifting events. The next phase was Campanian-Maastrichtian in age
184
S.M. CORFIELD E T AL.
Fig. 11. (a) 3D view of the diapir in Figures 9 and 10. A NW-SE oriented inline illustrates the sea-bed expression of the diapir. The high positive amplitudes are displayed from a 3D volume between the OpaI-CT reflector and the sea bed. The shallow high-amplitude reflectors are underlain by a predominantly transparent zone which allows a 3D view of the plume of high-amplitude gas anomalies associated with the diapir. (b) 3D view from the south of the diapir. Note the NE-SW elongation of the sea bed mound and underlying gas anomaly. and corresponded to the rotation of tilted fault blocks and crustal thinning, but must have been driven by either (a) greater thinning in the mantle lithosphere, or (b) an early emplacement of a magmatic underplate. The final uplift phase was in the Palaeocene and was coeval with local upper-crustal extension.
Fig. 12. An integrated model that illustrates the genetic link between sills, vents, mounds and diapirs in the study area.
This uplift can be attributed to one or a combination of the following: intrusions ( - 3 0 0 m in thickness) in the upper crust; magmatic underplating at the base of the crust; local thinning of the mantle lithosphere; or related flexural effects. 2. The presence of gas-charged mud diapirs and faults deforming the sea bed has implications for the integrity of the Tertiary top seal in the Gjallar and Vema structures. The well on the adjacent Nyk High encountered gas in the Cretaceous section while the Gjallar and Vema wells were unsuccessful. While details of seal integrity immediately above the reservoir can be debated, we note that the Nyk High differs from the Gjallar and Vema structures in that it did not suffer major Tertiary uplift, it lacks mud diapirs, and the faults in the Tertiary section do not cut upwards to depths shallower than the mid-Oligocene (Kittilsen et al. 1999). Of the three structures, the Vema Dome suffered the most uplift and, perhaps as a consequence, it has the most spectacular and extensive sea bed diapir field. Above, we suggest that the polygonal faulting is causally related to the post-breakup chimney/diapir systems. We also inferred that pathways initially created by fluids expelled around the sills were re-used for tens of millions of years, and inferred some remained propped open, either by sand grains or mineralization. Hence the Gjallar and Vema diapir fields, while not directly indicating the lack of good top- and fault-seals at reservoir levels, do indicate that these must be evaluated with great caution. 3. We infer a genetic link between the location of Palaeocene sills, the crests of the Palaeocene tilted fault blocks and the location of the diapirs. This relates to both chimney initiation near the time of breakup, and later focusing of deep compactionrelated pore fluids, and light hydrocarbons related to generation or reservoir overfilling, towards the diapirs. 4. Although we have no direct evidence of gas escape, we infer its occurrence based on the seismic characteristics. Combined with the 'Christmas tree' stratigraphy around the diapirs, this suggests periodic release of hydrocarbons from a reservoir in the Cretaceous section. We interpret the depositional loading to be to be relatively constant and inconsistent with the periodic expulsion represented by the stratal geometries around the diapirs. Diapirism is largely driven by loading, therefore the
3D SEISMIC VISUALIZATION OF THE GJALLAR RIDGE
lower geometry of the chimneys may be pre-diapir formation, and the upper geometry characteristic of the loading-driven mobilization of near-fluid sediment. Similarly, diapiric contribution from the Cretaceous section should wane up-section, as the Cretaceous sediments become near fully compacted. 5. Fluid flux through the diapirs has significant implications for thermal models in basin analysis, which typically assume purely conductive flow. The authors would like to thank Norsk Hydro AS and their PL215 licence partners for permission to publish this paper. The manuscript was considerably improved by the reviews of J. Allison and J. Cartwright. Software donations by Schlumberger and Paradigm to the University of Manchester are gratefully acknowledged.
References BREKKE, H., DAHLGREN, S., NYLAND, B. & MAGNUS, C. 1999. The prospectivity of the VOting and More basins on the Norwegian Sea continental margin. In: FLEET, A. J. & BOLDY, S. A. R. (eds)
Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference. Geological Society, London, 261-274. BLYSTAD, P., BREKKE, H., FAERSETH, R. B., LARSEN, B, T., SKOGSEID, J. & TORUDBAKKEN, B. 1995. Structural elements of the Norwegian continental shelf, part II. The Norwegian sea region.
Norwegian Petroleum Directorate Bulletin, 8. BONINI, M., SOKOUTIS, O,, MULUGETA, G., BOCCALETTI, M., CORTI, G., INNOCENTI, F., MANETTI, P. & MAZZARINI, F. 2001. Dynamics of magma emplacement in centrifuge models of continental extension with implications for flank volcanism, Tectonics. 20, 1053-1065. CARTWRIGHT, J. A. 1994, Episodic basin-wide hydrofracturing of overpressured Early Cenozoic mudrock sequences in the North Sea Basin. Marine and Petroleum Geology, 11,587-607. CHEVALLIER, L., GOEDHART, M. & WOODFORD, A. C. 2003. The
Influences of Dolerite Sill Complexes on the Occurrence of Groundwater in Karoo Fractured Aquifers: a Morpho-tectonic Approach. Water Research Commission, Pretoria, South Africa, Report 937/1/01. DAVIES, R., BELL, B. R., CARTWRIGHT,J. A. & SHOULDERS, S. 2002. Three dimensional seismic imaging of Paleogene dike-fed submarine volcanoes from the northeast Atlantic margin. Geology. 30, 223-226. DORI~, A. G., LUNDIN, E. R., JENSEN, L. N., B1RKELAND,~., ELIASSEN. P. E. & FICHLER,C. 1999. Principal tectonic events in the evolution of the northwest European Atlantic margin. In: FLEET. A. J. & BOLDY, S. A. R. (eds) Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference. Geological Society, London. 41-61. Du TOIT, A. L. 1905a. Geological Survey of Glen Grey and parts of Queenstown and Woodhouse, including the lndwe area. Geological
185
Commission of the Cape of Good Hope, tenth annual report, 95-140. Du TOIT. A. L. 1905b. The Karoo Dolerites - a study in hypabyssal intrusion. Transactions of the Geological Society of South Africa, 23, 1-42. ELDHOLM, O. & MUTTER, J. 1986. Basin Structure on the Norwegian margin from analysis of digitally recorded sonobuoys. Journal of Geophysical Research, 91. 3763-3777. FRANClS, E. H. 1982. Emplacement mechanism of Late Carboniferous tholeiitic sills in northern Britain. Journal of the Geological Society. London. 139, 1-20. GERNIGON, L., RINGENBACH, J. C., PLANKE, S., LE GALL, B. & JONQUET-KOLSTO. H. 2003. Extension, crustal structure and magmatism at the outer Voting Basin, Norwegian margin. Journal of the Geological SocieO', London, 160, 197-208. GEVERS, T. W. 1928. The volcanic vents of the western Stormberg. Transactions of the Geological Society of South Africa, 31, 43-62. HOVLAND, M., NYGAARD, E. & THORBJORNSEN, S. 1998, Piercement shale diapirism in the deep-water Vema Dome area, V0ring Basin, offshore Norway. Marine and Petroleum Geology, 15, 191-201. KITTILSEN, J. E., OLSEN, R. R., MARTEN, R. F., HANSEN, E. K. & HOLLINGSWORTH,R. R. 1999. The first deepwater well in Norway and its implications for the Upper Cretaceous play, V0ring Basin. In: FLEET, A. J. & BOLDY, S. A. R. (eds) Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference. Geological Society, London, 275-280. LUNDIN, E. & DORI~, A. G. 2002. Mid-Cenozoic post-breakup deformation in the 'passive' margins bordering the NorwegianGreenland Sea. Marine and Petroleum Geology, 19, 79-93. MJELDE, R., DIGRANES, P., vANSCHAAK, M. & SHIMAMURA,H. 2001. Crustal structure of the outer VOting Plateau, offshore Norway, from ocean bottom seismic and gravity data. Journal of Geophysical Research, 106(B4), 6769-6791. REN, S.. SKOGSEID,J. & ELDHOLM,O. 1998. Late Cretaceous-Paleocene extension on the V0ring volcanic margin. Marine Geophysical Research, 20, 343-369. RITCmE, J. D.. GATLWF, R. W. & RICHARDS, P. C. 1999. Early Tertiary magmatism in the offshore NW UK margin and surrounds. In: FLEET, A. J. & BOLDY, S. A. R. (eds) Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference. Geological Society, London, 573-584. SANCHEZ-FERRER. F., JAMES, S. D., LAK, B. & EVANS, A. M. 1999. Techniques used in the exploration of turbidite reservoirs in a frontier setting-Heiland Hansen licence, Vcring Basin, offshire mid Norway. In: FLEET, A. J. & BOLDY, S. A. R. (eds) Petroleum
Geology of Northwest Europe: Proceedings of the 5th Conference. Geological Society, London, 281-292. TORNE, M., FERNANDEZ, M., WHEELER, W. & KARPUZ, R. 2003. 3D crustal structure of the Voring Margin (NE Atlantic): a combined seismic and gravity image. Journal of Geophysical Research, 108(B2) ETG 16-1-1 I.
Tertiary inversion in the Faroe-Shetland Channel and the development of major erosional scarps JOHN
R. SMALLWOOD
A m e r a d a H e s s Ltd., 33 G r o s v e n o r Place, L o n d o n S W l X
7HY, U K (e-mail: j o h n . s m a l l w o o d @ h e s s . c o m )
Abstract: At the shallowest point of the Faroe-Shetland Channel, between the Faroe Islands and the Shetland Isles, the sea bed is deformed into a series of major scarps and hollows. The cuspate scarps, or "Judd Falls', are up to 15 km in length and are over 200 m high. Interpretation of 3D seismic data and high resolution 2D seismic data shows that the scarps are part of a larger series of structures that are partly buried. A second series of buried asymmetric hollows has been mapped 50 km to the northwest. Both sets of hollows are interpreted to have a deep-water erosional origin, postulated to be associated with the initiation of the high-energy bottom currents of the south-flowing Northern Component Water from the NorwegianGreenland Sea into the North Atlantic. Present-day measurements presented here show that deep-water current velocity can peak at over 0.8 m s- I. Both erosional complexes are positioned directly above Tertiary inversion structures, and this study has identified two periods of compressional deformation, latest Ypresian and late Lutetian, in addition to previously documented phases. Compression in the area has been linked to changes in the interaction between the Mid-Atlantic Ridge and the Iceland mantle plume. Enhanced plume activity also concentrated deep-water flow in the Faroe-Shetland Channel by physically impeding deep-water currents elsewhere. Where enhanced deep-water flow encountered the partial barriers of the inversion structures, accelerated turbulent erosional currents carved the scarps into the sea bed.
The Faroe-Shetland Channel lies between the Shetland Isles and the Faroe Islands, forming a section of the northwest European continental margin (Fig. 1). The development of the area has been punctuated by a series of tectonic events including major rifting during the Permo-Triassic, possibly the Jurassic, the Cretaceous and the Paleocene (Dean et al. 1999). During the Paleocene, the region was strongly influenced by the protoIceland mantle plume: major extrusive magmatism (Naylor et al. 1999; Smallwood et al. 2001) and transient uplift (Clift & Turner 1995) resulted from the elevated mantle temperatures beneath the lithosphere. From Eocene to Recent times, the dominant process affecting the basin has been one of post-rift thermal subsidence (Turner & Scrutton 1993). However, several compressional episodes between the late Paleocene and the Miocene have affected the development of the region (Boldreei & Andersen 1993). Hydrocarbon exploration activity has provided a significant database of 2D and 3D seismic data across the Faroe-Shetland Channel area, and approximately 300 exploration wells and shallow boreholes. At the present day, the Faroe-Shetland Channel is an important conduit for cold, low-salinity North Atlantic Deep Water (NADW) flow from the Norwegian-Greenland Sea into the North Atlantic (Stoker et al. 1998). Presently the NADW flows southwest through the Channel and swings northeast through the Faroe Bank Channel, north of the W y v i l l e Thompson Ridge, a Miocene inversion feature (Andersen & Boldreel 1995; Boldreel & Andersen 1995; Fig. 2). Strong deepwater currents were encountered in the autumns of 2001 and 2002 during the drilling of wells 6004/16-1 and 204/16-1, which are located near the centre of the southern part of the FaroeShetland Channel (Fig. 2). Example current data measured using an acoustic Doppler current profiler for a 48-h period during the drilling of well 204/16-1 is shown in Figure 3. Speed and direction measurements were made from the West Navion drillship, at four depths: 74 m, 234 m, 650 m and 890 m below sea level. The peak current measurement during the period shown was 1.59 knots (2.95kmh -1 or 0 . 8 2 m s - j ) in a WSW direction at 650 m depth. Previous publications have suggested peak current velocities from 0.33 m s -~ (Stoker et al. 1998) to 0 . 6 m s -~ (Masson 2001), well below the measurements recorded here. The shallowest measurement of current (74 m) showed an irregular, relatively low speed (th 650m
~ir
N ~
-,
~ f
I
I
|
N18:00
I I
I
I
J
I
I
00:00
J
I
i
J
J
06:00
i
I
J l
I
12:00
1
I
I
I
~
18:00
i
i
I
00:00
07-0ct Current Direction (de,
i
I
I
i
08-0ct
E
I
06:00
1
1
|
12:00
N
,th 890m
s ~
S E i
N1
W
S
E
12:00
2'
N
s
~.o~ol & Y I&. If IX,. If Ix, I / olool ,I,, ,'~ 35oo
km ~ Mle~cat~P r o j , ,,,
200~
'~ 4OE
0 ~
8~
52~ 16 ~E
12~
lead to an increase in the number of seismically detected sandstone intrusions both in the North Sea and elsewhere 9 However, no matter how good the seismic data, they will always be band limited, i.e. with a frequency range of the order of 1 0 - 6 0 H z typical of streamer 3D seismic data at 2kin depth below the North Sea. Hence, only intrusions or intrusion complexes thicker than a few metres will be detected by seismic data, and the thickness of even large-scale sandstone intrusions is usually less than or close to the seismic resolution (a quarter wavelength or tuning thickness ~ 15-25 m), complicating the quantification of injected volumes in the subsurface 9Since subseismic scale intrusions appear to be far more common than seismic scale intrusions, injected volumes will tend to be greater than it appears from the seismic alone. The 3D seismic data used in this article are displayed as normal European polarity with red-yellow-white representing negative amplitudes (increase in acoustic impedance) and blackblue representing positive amplitudes (decrease in acoustic impedance). The data were processed to yield zero phase, but it is uncertain whether true zero phase was achieved for the entire dataset, and we estimate that phase rotation up to about 90degrees may be present. Hence, caution is needed when interpreting certain events for their porosity or pore fluid 9
seen in borehole or outcrop and those detected using seismic data. From boreholes and outcrops it appears that cm to a few m-thick intrusions are most common, although large intrusions or intrusion complexes may be some tens of metres thick and thus of a scale possible to image using seismic data (Figs 4c, d; Thompson et al. 1999; Duranti et al. 2002b).
Seismic imaging
The seismic expression of large-scale sandstone intrusions is determined by the interplay between intrusion geometry, acoustic properties of the sandstone and encasing mudstones, and the quality of the seismic processing (MacLeod et al. 1999; Mikhailov et al. 2001; Luchford 2002). Recently, it has been shown that careful re-processing of seismic data using pre-stack time or depth migration greatly enhances the definition of crosscutting events interpreted as sandstone intrusions (Luchford 2002). Moreover, it has been shown that honouring the anisotropy of the subsurface in the seismic processing may lead to significant imaging improvements of reservoirs associated with large-scale remobilization and injection in the North Sea Paleogene (Mikhailov et al. 2001). It is thus likely that future improvements in seismic imaging and processing will
ENE
WSW
~ ii 5oo! ~.: ~.'g,g
Fig. 1. Large-scale remobilized and injected sandstone bodies occur within the Paleogene of the South Viking Graben and adjacent areas of the Utsira High, Outer Moray Firth, Crawford Ridge, Beryl Embayment, northern Central Trough and SW Norwegian-Danish Basin (within white line).
0 __. Norwegian Trench ,] r . ~ - z + , - C t . ~ ' z . " . ~ ' < ~ . ' . : , ~ ' ~ : ~ - ' ~ -
-'-' ~
- =+
~.
~
'
~
~
bUU
1000 9 --~A---,~
",
~--
"
.~
.i_
9. . . . . . .
.~ " ~:--. ~.-.. z,~,
~, ,.+ ,
9
.~_,, ~
..........
+
-, .~ Z ~
. . ~
l
~ - - . ~ , r -
,. . . .
1
1500
In
o
o
=
~~
=
SANDSTONE INTRUSIONS, NORTH SEA
Fig. 15 (inset) 5km
,]
Fig. 15 Balder Fm Time-thickness
(ms TWT)
200
40
Fig. 5. Balder Formation time-thickness map covering the northern 2/3 of the study area in the South Viking Graben. The map shows the locations of steep-sided sand bodies within the Balder Formation as areas of anomalous thickness (> 120 ms TWT). The locations of seismic lines (Figs 6, 7, 9,10 & 15) and localized study areas (Figs 8, 11 & 13-15) are indicated.
267
Low-amplitude reflections are seen along the water-wet, poorly cemented sand body (Fig. 7); except from a minor section with high-amplitude response at the SE tip (Fig. 8) whilst a highamplitude wing-reflection is observed on three sides of the partly cemented, partly oil saturated sand body (Figs 9-11). The only borehole calibration of wing-reflections available in the area is from the low-amplitude feature, which correlates with a 41.5 m thick, poorly cemented sandstone detected by borehole cuttings and the gamma-ray log (Fig. 7). MacLeod et al. (1999) showed that wing-like reflections along the edges of the main Alba reservoir represent large-scale sandstone intrusions (low-angle dykes), some 20+ m thick, with excellent reservoir properties and well connected to the main Alba sandstone body. The wing-like reflections encountered at Alba and elsewhere in the North Sea Paleogene typically cross 100-200+ ms TWT (120-250+ m) of the overlying section at angles in the range 20-40 ~ (e.g. Figs 7, 10 & 11; Lonergan & Cartwright 1999; Lonergan et al. 2000). The present-day values result from compaction due to subsequent burial and, depending on the depth of intrusion, the low-angle dykes may originally have crosscut several hundred metres of less compacted section at angles of the order of 45-60 ~ (Figs 6, 7, 9 & 10; Lonergan & Cartwright 1999; Lonergan et al. 2000). The wing-like reflections illustrated here appear to terminate close to the top of the Frigg interval (Figs 7, 9 & 10) and, in one case, a discordant reflection is seen to continue as a conformable event at top Frigg (Figs 9-11). The conformable event extends several hundred metres away from the discordant reflection (Figs 10 & 11). The termination of the wings at an unconformity corresponding to the top Frigg level could be interpreted as sand extruded onto the palaeo-seabed, or as if the wing injection turned into a sill at the unconformity. Lonergan et al. (2000) suggested that dykes turn into sills at some shallow depth governed by the local stress state, reservoir fluid pressure, and bedding anisotropy. It is thus likely that the transformation of a dyke into a sill would be facilitated by the presence of an unconformity such as the top of the Frigg interval in our area, or the Eocene-Oligocene boundary in the Alba area (Lonergan & Cartwright 1999). However, the simple relation of dykes turning into sills at shallow levels does not account for crosscutting patterns of dykes and sills or for 'zig-zag'-style dyke-sill complexes (cf. Figs 4b-d) commonly seen in outcrop.
Sw
NE S
N TWT (ms) IOO0
1200
1400
Fig. 6. Depositional strike line showing steep-sided sand bodies in the Balder Formation and crosscutting (wing-like) reflections interpreted as low-angle sandstone intrusions emanating from the edges of the concordant sand bodies. Note the difference in amplitude response between the two mounds. This reflects both pore-fluid and cementation differences. The westernmost sand body is brine saturated whereas the easternmost sand body is partly oil saturated and partly cemented. For location see Figure 5.
1600
1800
2000
-1281
~
+127
1:1
268
M. HUUSE ET AL. GR ,~ w TWT (ms)
TWT (ms)
1400
1400
1500
1500
1600
1600
1700
1700
1800
1800
1900
1900
2000
2000
2100
2100
2200
2200
....
I!
. . ~ + 1 2 7
1:1
Fig. 7. Depositional dip line showing a steep-sided, brine-saturated sand body in the Balder Formation and crosscutting wing-like reflections emanating from its margins. The eastern 'wing' is calibrated by a borehole as indicated by the GR log. The well encountered 41.5 m of massive, water-bearing sandstone in the lower part of the Frigg Formation. The crosscutting reflection and the corresponding sandstone are interpreted as a low-angle sandstone intrusion. The dip of the reflection is about 35 degrees, and the real thickness of the dyke is thus of the order 30-35 m. 'Ratty' Frigg sandstones indicated by well logs about 200 m above the Balder sandstone in the centre of the structure have been calibrated to core, which contains only injected sandstones and Frigg mudstones. Both the source sandstone and the injected sand body produce low-amplitude reflections indicating that they are similar in acoustic impedance to the encasing mudstones. Well data indicate that the mounding is entirely due to the presence of a thick Balder-age sand body within the mound with no Balder sandstone present immediately to the east and south. Well and seismic data indicate that there is negligible erosion at the base of the sand body and that the encasing mudstones on either side correlate with those on top of the sand body. Note that the wings appear to terminate at the top Frigg level where there is a marked onlap from the west onto the mound. Upward termination at an unconformity is common for seismic-scale sandstone intrusions. For location see Figure 5.
For the following reasons we think the conformable event is more likely to represent an extrusion: 9
The high-amplitude event is conformable all along the tip of the feeder dyke, unlike sills, which often display (low-angle) discordant relations with the host rock. 9 The event only extends away from the mound, not across the top of it, indicating a control of mound topography on the emplacement of the feature; such topographic control would be more effective on an extrusion that flowed in response to gravity, whereas an intrusion would flow according to pressure gradients and strength of the host rock. 9 Emplacement of the intrusion by fluidized flow would require a connection between an overpressured source sand and a large aquifer or the seabed; since there is no aquifer at top Frigg in this location, it seems likely that the intrusion reached seabed.
The importance of the distinction between extrusions and sills lies in the fact that the stratigraphic level of an extrusion provides the timing of remobilization and injection of sand from the source body. Moreover, the height of the extrusion above the source sandstone allows an estimation of the depth of burial of the source sand body when the intrusion formed, given the present-day porosity and assuming a porosity-depth curve for the Frigg mudstones at 'top Frigg' time (earliest Eocene). For the smectite-rich Frigg mudstones, we refer to the porositydepth curve established by Velde (1996) for smectite-rich muds
and claystones based on ODP (Ocean Drilling Programme) and DSDP (Deep Sea Drilling Programme) measurements. For a rough, first-order estimate of the burial depth at the time of intrusion we estimate the height of the wing seen in Fig. 10 as c. 200 ms TWT, assume an interval velocity of c. 2.5 km/s, a present-day porosity of the Frigg mudstones of about 25%, an average porosity of the mudstones (muds) at top Frigg (earliest Eocene) time of about 65%. These values, which are consistent with the available well calibrations and the porositydepth curve established by Velde (1996), give a burial depth of the order of 4 0 0 - 5 0 0 m at the time of intrusion. This rough estimate is in the lower end of the range ( 5 5 0 - 8 0 0 m) estimated for the same feature by Jolly & Lonergan (2002) who assumed that dykes turn into sills at some depth below the surface.
Class 2: Conical amplitude anomalies. The seismic signature of the upper Paleogene in the northern North Sea is often rather chaotic (Fig. 6), probably owing to a combination of softsediment deformation and poor acoustic impedance contrasts in the shale prone intervals of the Horda Formation (cf, L0seth et al. 2003). In the South Viking Graben the chaotic pattern is frequently broken by a distinct level of crosscutting, highamplitude seismic reflections (Figs 6 & 9). These are commonly V-shaped in cross section at typical seismic display scales ( - 5 times vertical exaggeration) with a near-circular to angular plan geometry. The structures are thus conical in three dimensions (Figs 13-15). Similar structures in the Outer Moray
SANDSTONE INTRUSIONS. NORTH SEA
Fig. 8. (a) Three-D visualization showing the top Balder time-structure intersected by an acoustic impedance slice at 1848 ms TWT and a vertical acoustic impedance section (blue is low and orange-grey is high impedance). A local high-amplitude crosscutting event is shown as a voxel body of high acoustic impedance emanating from the SE-tip of the mound. This is interpreted as a locally cemented equivalent of the less reflective wings shown in Figure 7. (b) Vertical profile along the steep-sided mound showing the crosscutting anomaly at the tip of the mound. The anomaly appears to be both high and low acoustic impedance, suggestive of cemented and porous, hydrocarbon-filled intervals, respectively. However, this could also be an effect of seismic tuning of thin beds, which may cause problems for seismic inversion processing. For location, see Figure 5.
Firth have been interpreted as conical sandstone intrusions intruded along polygonal faults (Lonergan et al. 2000: Gras & Cartwright 2002; Molyneux et al. 2002). Similar features in the North Viking Graben have been interpreted as injected tufts or seismic artefacts (LOseth et al. 2003). The discordant amplitude anomalies seen within the upper Paleogene of the South Viking Graben are typically 5 0 0 1000m across and 5 0 - 2 0 0 + m high. The discordant elements typically dip inward at about 2 0 - 4 0 d e g r e e s , terminating downward at distinct apexes (Figs 9 & 13). The anomalies terminate upward at an unconformity below the thick sandstones of the Belton and Grid sandstones (Figs 9 & 13-15). The apexes of the cones sometimes occur within a few hundred metres of underlying sand bodies belonging to the Sele, Balder and Frigg Formations, but are more commonly located several hundred metres above these (Fig. 9). The most striking anomalies are typically bright crosscutting events that, in the present study area, appear to be mainly positive amplitude anomalies (Figs 9 & 13). These represent a decrease in acoustic impedance, which can be tied to the base of a 6 2 m thick sandstone in the calibration well (Fig. 15). The sandstones encountered at this level by other wells display varying degrees of calcite cementation, and a bright amplitude response appear to be associated with the presence of a thin ( 1 - 2 m ) well-cemented basal part (Fig. 15). The thickness of the cemented base is
269
"Bright" injection complex
usually well below the seismic tuning thickness ( 1 5 - 2 5 m) and seismic modelling indicates that it therefore enhances the positive response (decrease in acoustic impedance) of the sandshale contact at the base of the sand. When the sandstone is poorly cemented and lacks a fully cemented base, the reflections are usually less conspicuous, but can often still be seen as faint crosscutting events. In general the reflection from the top of the sandstone is less conspicuous, sometimes non-existent, apparently because the overlying section is more heterogeneous than the Horda mudstones, and the upper part of the sandstone is poorly cemented, resulting in a poor impedance contrast at the upper shale/sandstone contact. Up to three levels of discordant reflections may occur within one section, but usually only one level is associated with anomalous amplitudes. The available well control does not allow direct calibration of all discordant anomalies, but the few anomalies penetrated by boreholes in our study area are associated with the base of a thick sandstone above relatively homogeneous mudstones (Fig. 16). A high-amplitude positive reflection is also commonly seen at the base of the Belton and Grid sandstones above the unconformity. It is thus possible to trace a continuous 'reflection' through the discordant anomalies and along the overlying unconformity in large parts of the study area. A map of the resulting 'V-horizon' in a subset of the study area gives an impression of a surface with numerous near circular
270
M. HUUSE E T A L . Fig. 10
West
East
TWT (ms)
TWT (ms)
1000
1000
1200
1200
1400
1400
1600
1600
1800
18oo
2000
2000
2200
2200
,28 I1
~
+1~
t:1
North
Fig. 9. Depositional dip line showing a steep-sided sand body in the Balder Fm and an associated discordant (wing-like) reflection, terminating in a concordant amplitude anomaly at top Frigg level. A series of V-shaped amplitude anomalies occur 500-600 ms TWT above the top Balder at the base of the Belton and Grid sandstones. The distance from the bases of the V-shaped anomalies to the shallowest underlying sand body is about 400-500 ms TWT. None of the anomalies seen along this section have been penetrated by boreholes, but a well further to the SW penetrated 62 m thick sandstone coinciding with similar V-shaped anomalies (Fig. 15). For location see Figure 5.
Fig. 9
South
1600
600
1900
900
2000
~000
-128D
"
~
*127
1:1
Fig. 10. Depositional strike line showing a steep-sided sand body in the Balder Fm and associated discordant (wing-like) reflections, terminating at a high-amplitude concordant event at top Frigg level. The concordant event may be interpreted as a sand extrusion or a sill. The concordant event can be mapped around the sand body at a consistent stratigraphic level (top Frigg) and only extends away from the sand body, which may have caused a slight dome on the palaeo-seabed. Based on these observations it is inferred that the concordant anomaly represents a sandstone extrusion. If this is correct, the timing of dyke intrusion is constrained by the age of the unconformity at top Frigg, i.e. earliest Eocene. The bright positive (blue) amplitudes are indicative of highly porous sandstones whereas the bright negative (red) less porous and/or brine-filled sandstones. The wing-like reflection has as a moderate- to bright negative amplitude corresponding to (partly) cemented sand, if the data is true zero phase. Seismic modelling indicates that a sandstone of less than about 30% porosity will result in a bright negative amplitude when it is encased in Balder and Frigg mudstones, both when water and oil wet. Amplitude tuning due to interference between top and base of the sandstone will enhance the reflection response when the sandstone is of the order of 10-40 m thick, depending on the velocity of the sandstone and the frequency content of the data. Sandstone intrusions typically fall within this range and thus it is generally very difficult to quantify the thickness, porosity and pore fluid of sandstone intrusions that are un-calibrated by well data. For location see Figure 5.
SANDSTONE INTRUSIONS. NORTH SEA
Fig. 11. (a) Semi-transparent top Balder time-structure map and interpreted surface corresponding to the top of the amplitude anomaly shown in Figures 9 & 10. The depth range of the amplitude anomaly represented by the colour coding is about 200 ms TWT ( - 2 4 0 m). The geometry of the polygonal fault cells at top Balder are comparable to the geometry of the Balder sand body and match the somewhat angular outline of the wing-like reflection along its periphery. (b) Semitransparent map of the top of the amplitude anomaly intersecting a structurally flattened semblance slice close to top Balder, showing a well-developed polygonal fault pattern (semblance is a seismic attribute expressing the similarity of neighbouring traces with areas of least similarity (black) denoting abrupt changes such as faults). The sand body and the flanking intrusion appear to fit within the polygonal fault pattern. suggesting a control of their geometry by polygonal faulting. However. as seen on Figures 9-10, and outlined in (a) the wing also closely follows the periphery of an anomalously thick Balder interval. This indicates that the location of the wing-like intrusion is controlled by the sand body geometry, as also suggested for the Alba "wings" by Cosgrove & Hillier (2000).
'depressions' at the base of the sandstone (Fig. 15, inset). However, the well calibration reveals that the depressions are not filled with sandstone, rather they correspond to the bases of conical sandstone sheets, some tens of metres thick encased in m u d stone (Fig. 15). The interval containing the discordant anomalies is generally characterized by poor signal/noise ratio making imaging and mapping of features other than the bright anomalies difficult. In cases where the conical anomalies are associated with cemented sandstones thicker than a few metres or with anomalous sandstone thickness, there may be significant pull-up effects and disturbance of seismic ray paths underneath. This causes problems for seismic imaging, which is further hampered by
271
Fig. 12. Voxel-interpretation of the largest positive amplitudes of the Balder sandstone and peripheral intrusion superimposed on standard surface interpretation of the anomalous amplitudes (cf. Figs 9-1 l ). Voxel-interpretation allows several levels of voxels to be picked at any one location and thus provides more realistic (complicated) geometries of injection features than standard surface mapping. This example also highlights that surface mapping is well suited for mapping events of varying amplitude, whilst, to be efficient, voxel-picking is limited to events of anomalous amplitudes. Voxel picking was thus also attempted for the feature shown in Figure 7. but failed due to lack of amplitude, making it impossible to constrain the voxel tracker. It appears that a combination of the two methods is optimal to benefit from the additional detail inherent in the volume-based interpretation whilst preserving lowamplitude information via conventional surface mapping. Surface-based attribute extraction may provide some information on thickness and continuity, but is not a complete substitute for volume-based interpretation. strong seabed multiples beneath the brightest anomalies (Fig. 9; cf. Lc~seth et al. 2003). W h e n seen in vertical profiles the conical anomalies resemble V-shaped channel cross sections, but the near circular to angular plan geometry rules out an origin as channel scours. The conical geometry is, however, reminiscent of at least three p h e n o m e n a recently described from the northern North Sea and adjacent areas: 9 9 9
Giant pockmarks (Cole et al. 2000) Large-scale density-inversion structures (Davies et al. 1999) Conical sandstone intrusions (Molyneux et al. 2002)
Pockmarks are craters, typically up to a few hundred metres wide and several metres deep formed by gas or fluid expulsion at the seabed. They occur in fine-grained sediments on continental shelves worldwide (Hovland & Judd 1988). Recently, Cole et al. (2000) reported the occurrence of (inferred) giant pockmarks, a few kilometres in diameter and many tens to a few hundred metres deep, in the Paleogene of the Outer Moray Firth. The conical features reported here are a m a x i m u m of 1 - 1 . 5 k i l o m e t r e s across and a m a x i m u m of 1 5 0 - 3 0 0 m deep and thus within the range of the dimensions of the inferred giant pockmarks. However, the conical anomalies have a pointed base and are thus morphologically different from the flat-based structures described by Cole et al. (2000). Borehole calibration of the conical anomalies yield several tens of metres thick sandstone, but this is far less sandstone than would be expected if the structures were pockmarks filled during Belton/Grid
272
M. HUUSE ET AL.
TWT (ms)
rWT
1200
200
1200
1200
1300
300
1300
1300
1400
400
1400
1400
1500
500
1500
1500
1600
600
1600
1600
1200
1200
1200
200
1300
1300
1300
300
1400
1400
1400
400
1500
1500
1500
,500
1600
1800
1600
600
:ms)
True dip (degrees): 90 60 45
30
15
0
qlm~+"~_ ~+
~ '+ ,
+~
+,._I
"
"+-~
+'~
-
,ll+
.......
"+ , + ~
+'+""~) '~,, ~'ml~,++~
+
,+
+.. - _ . , +
A
:+++-
"
-r+,~++,+++"~+oo~~
SANDSTONE INTRUSIONS. NORTH SEA
273
Fig. 13. Time-structure map, vertical cross-sectionsat 100 m spacing, and timeslices at 60 ms spacing, through a large conical amplitude anomaly. The amplitude anomaly reaches an unconformity overlain by a 74 m thick Belton/Grid sandstone in a nearby well. The reflections dip about 30-40 degrees (interval velocity --2200 m/s). The elongation of the basal apex seems to be controlled by an intraformational (polygonal) fault. Overall, about 50-60% of this and other conical structures in the area appear to coincide with polygonal faults. Compare this map with the volume-based interpretation of the anomaly shown in Figure 14. For location see Figure 5.
sand deposition in the Middle/Late Eocene. A pockmark origin is further complicated by the apparently even sandstone thickness throughout the anomalies (Fig. 15, cf Molyneux et al. 2002). An origin as pockmarks is thus unlikely. Density inversion would have occurred when the Belton and Grid sandstones were deposited onto the highly porous clays of the Horda Fm. It is thus conceivable that the conical anomalies could represent giant load structures, although they do not seem to be arranged in the characteristic polygonal pattern recognized by Davies et al. (1999). An interpretation of the anomalies as the bases of giant load structures would be in accordance with the chaotic structure of the underlying shale, suggesting that the mudstones had experienced extensive soft-sediment deformation. However, the origin as load structures would require the structures to be filled with sand and thus does not agree with the tabular sandstone geometries indicated by. The discordant nature of the conical anomalies, the even thickness of sandstone along the anomalies and the termination at an overlying unconformity are compatible with an origin as conical sandstone intrusions spreading upward from a central feeder point. This conclusion is in agreement with previous interpretations of conical amplitude anomalies in the Outer Moray Firth (Lonergan et al. 2000; Gras & Cartwright 2002; Molyneux et al. 2002). As the conical anomalies are very widespread and because there is no potential source sandstone body for some hundred metres below the apex of the anomalies, we speculate that the sand was supplied through near-vertical feeder ('blow-out') pipes (Lcseth et al. 2001), possibly partly following polygonal fault intersections in the Horda mudstones.
Controls on large-scale remobilization and injection General controls
The mechanisms controlling the processes of large-scale remobilization and injection of deep-water sands are not
Fig. 14. (a) Seismic amplitude cube used for amplitude- and volume-constrained voxel-picking of the conical amplitude anomaly seen in Figure 13. (b) Seed pick and resulting voxel interpretation; turquoise dots represent automatically picked voxels. (c-d) Three-D visualizations of the resulting voxel body composed of both positive and negative voxels. Two seed-points were used to pick positive and negative voxels (one each). The resulting structure is more complicated (realistic?) than the one generated by surface mapping (Fig. 13). However, the surface-based interpretation is more readily interpreted in terms of structure.
completely understood, but from core and outcrop observations it appears that the most common mode of sand transport during sand injection is by fluidized flow (e.g. Duranti et al. 2002a: Jolly & Lonergan 2002). Fluidized flow can be initiated when an overpressured and unconsolidated sand body is connected to a less overpressured environment such as the sediment surface or a shallower aquifer, Remobilization of deep-water sand bodies has been related to combinations of several different processes such as stress state, earthquake activity, rheology and polygonal faulting of the encasing mudstones, overpressure development during burial and overpressure caused by hydrocarbon charge (Duranti et al. 2002c; Jolly & Lonergan 2002; Molyneux et al. 2002). Except for polygonal faulting, the controlling factors do not lend themselves to investigation by 3D seismic data and the following section will thus only deal with the relation between polygonal faults and sand injection.
Polygonal faults
Layer-bound polygonal faulting occurs in response to the de-watering (contraction) of very fine-grained, usually smectitic sediments. The faults are purely extensional with relatively small throws ( < 5 0 m ) and oriented in all directions, often attaining hexagonal to quadratic geometries when seen in plan view (Cartwright & Lonergan 1996). The dips of polygonal faults typically average 45" with fault lengths and spacings of 100-1000m (Cartwright & Lonergan 1996: Lonergan & Cartwright 1999). The dips and lateral extents are thus quite similar to those of clastic intrusions imaged on seismic data (cf. Lonergan & Cartwright 1999; Molyneux et al. 2002). Virtually all published subsurface examples of large-scale sand remobilization are from the North Sea Paleogene where the encasing mudstones are heavily affected by polygonal faulting (Lonergan & Cartwright 1999; Lonergan et al. 2000; Molyneux et al. 2002). Discordant amplitude anomalies interpreted as sandstone intrusions may have similar dimensions and can
274
M. HUUSE ETAL.
Fig. 15. Borehole calibration and map view (inset) of the "V-horizon" ( - discordant amplitude anomaly and unconformity at the base of Grid sandstones). The seismic data are acoustic impedance and the borehole data is the sonic log. At the well location the anomaly coincides with the base of a 62 m thick sandstone with a thin (1-2 m), cemented base. Borehole logs show that the sandstone is characterized by low GR and high velocity and density values, especially in the cemented basal layer. The "Grid" sandstone and its cemented base are seen as a light blue (medium acoustic impedance) interval overlying a thin orange-grey (high acoustic impedance) event in an overall dark blue (low acoustic impedance) succession. The high-amplitude discordant reflections thus coincide with the cemented base of a tens of metres thick sandstone unit, whilst the gradational impedance increase at the top of the sandstone is poorly defined. The inset shows that the structures are more or less circular in plan view. The vertical relief of the map is about 200 ms TWT (-200 m) from white-yellow to purple colours. For location see Figure 5.
often be seen to coincide with polygonal fault planes, and it thus seems straightforward to infer that large-scale sand injection generally occurs along polygonal faults (Lonergan & Cartwright 1999; Lonergan et al. 2000; Gras & Cartwright 2002: Molyneux et al. 2002).
periphery of the source sand body, and, it is possible that forced folding due to differential compaction may be a more important control on wing localization than polygonal faulting in this case.
Conical sandstone intrusions and polygonal faults Tabular ('wing-like') sandstone intrusions and polygonal faults Seismically resolvable tabular sandstone intrusions emanate upand outward from the margins of the main Alba sand body. These 'wing-like' intrusions appear to be related to layer-bound polygonal faulting of the encasing mudstones (Lonergan & Cartwright 1999). A similar relation between faults and intrusions may be inferred for one of the examples presented here (Fig. 11). Differential compaction across a thick sandstone body encased in mud will result in forced folding of the overlying mudstones and may result in fracturing of the sealing mudstones at the shale/sandstone interface, especially upward propagating fractures at the edges of the sand body and downward propagating fractures over the crest of the sand body (Cosgrove & Hillier 2000). This distribution of fractures favours the localization of large-scale intrusions along the edges of the sand body (Figs 7 & 10), whilst smaller-scale dykes and sills may be more abundant over the crest of the sand body (cf. Dixon et al. 1995; Cosgrove & Hillier 2000), as also observed in cores above the mounds described here. The interaction between polygonal fault systems and forced folds and fractures is poorly understood at present. However, it is evident that the wing-intrusion closely follows the
It has been suggested that the conical amplitude anomalies represent sandstones intruded upward along polygonal faults in the encasing mudstones (Lonergan et al. 2000; Gras & Cartwright 2002: Molyneux et al. 2002). This inference is based partly on observed coincidences between polygonal faults and amplitude anomalies, on similar plan geometries and on similar dip angles and length scales of the features (e.g. Molyneux et al. 2002). However, careful analysis of the relations between conical intrusions and polygonal faults in our study area indicates that only parts of the intrusions follow polygonal fault planes whereas others do not (e.g. Fig. 13). This is not surprising as polygonal fault cells do not form pointed conical geometries (Cartwright & Longergan 1996). Moreover, there are several cases where polygonal faults have been crosscut by conical amplitude anomalies and vice versa, e.g. Huuse et ah 2001, Fig. 4, indicating that the conical shape is a fundamental property of this class of sandstone intrusions. It thus appears that, in order to achieve their basic cone shape, conical sand intrusions will exploit polygonal faults when these are favourably oriented while they will force their own way when no suitable pre-existing weaknesses exist. Rather than interdependence between polygonal faulting and sand injection we speculate that the geometrical similarities between polygonal faults and sand intrusions may be attributed
SANDSTONE INTRUSIONS, NORTH SEA West
275 East
.,
Approx.
TWT
(ms)
depth
(m) 1000
1 oo
~
elton/Grid sands (..iddle-Upper Eocene)
-
--
1000
-
,
v\
1400
Conical sandstone intrusions
1500
Feeder pipes(?)
1600 _
~
.
Extrusion
1800
2000
2000
9
.
.
.
.
.
2200
2 km 2500
Fig. 16. Summary diagram based on the seismic section shown in Figure 9. For simplicity, the main depositional sand bodies are shown without shale intercalations seen in boreholes. There are two main types of sandstone intrusions in the South Viking Graben: Class 1 are tabular sandstone intrusions seen as 'wing-like' reflections emanating some hundred metres upward along the periphery of steep-sided sand bodies. Class 2 comprise conical sandstone intrusions seen as V-shaped amplitude anomalies in cross section emanating some hundred metres upward from central apexes, which appear detached from underlying sandstones, perhaps connected by clastic feeder ('blow-out') pipes (cf. L0seth et al. 2001). Note that the feeder pipes sketched here cannot be observed in the available data and may have a more tortuous path than indicated, similar to those observed by L0seth et al. (2001) offshore Nigeria. Assuming that both types of intrusions reached palaeo-seabed, suggests that there were at least two phases of intrusion in the South Viking Graben: (i) an earliest Eocene (top Frigg) phase of intrusion caused by a combination of forced folding along the edges and over-pressuring of Balder sand bodies, and (ii) a middle or late Eocene phase of intrusion caused by over-pressuring of deeper aquifers and upward intrusion of sand through cylindrical(?) feeder pipes into conical sandstone intrusions, partly exploiting polygonal faults in the encasing mudstones.
to a c o m m o n control on their formation by the rheological properties of the encasing highly smectitic lower Eocene mudstones.
Conclusions Large-scale remobilization and injection of sand can severely alter the geometries and reservoir properties of deepwater sandstones. Features associated with these processes are observed as steep-sided m o u n d s and discordant seismic anomalies in seismic data from the Paleogene of the South Viking Graben. Two main classes of seismic-scale clastic intrusions are defined based on well calibration of lithology, seismically defined geometry and relation to the inferred source sandstone. Class 1: Crosscutting wing-like reflections that emanate from the edges of the steep-sided mounds are interpreted as large-scale tabular sandstone intrusions. The consistent occurrence of the 'wings' at the edges of the source sand bodies and inconsistent relations between the wings and polygonal faults indicate that the location of the intrusions is controlled mainly by forced folding of the overlying mudstones due to differential compaction, whilst polygonal faulting played a secondary role. It is inferred that the intrusions reached the seabed in the earliest Eocene at an estimated burial depth of the source sand body of about 500 m. Class 2: Conical amplitude anomalies that are not visibly in contact with underlying sandstones are interpreted as conical sandstone intrusions emanating upward from a central feeder
pipe. The inferred feeder pipe may connect the intrusions to a source sand body several hundreds of metres below the intrusion apex. It is argued that the cone shape is a fundamental property of this class of intrusions and that polygonal faults are only exploited when favourably oriented. The intrusions terminate at an unconformity underlying the Belton and Grid sandstones and it is thus uncertain whether they reached palaeo-seabed or simply intruded at the base of the overlying sandstones which, due to their extensive thickness and lateral continuity, could have dissipated the excess fluid pressure driving the upward intrusion of sand. It is shown that remobilized and injected sandstones interpreted on seismic data may take on vastly different appearances, depending on the method of interpretation and visualization (cf. Figs 12-14). Finally, it should be emphasized that sandstone intrusions may be highly porous and permeable and thus constitute significant reservoirs and/or plumbing systems within thick mudstone sequences. The present study was carried out as part of a two-year research project funded by ChevronTexaco UK, Enterprise Oil Norge, Kerr-McGee UK, Norsk Hydro, Shell Expro UK, Statoil and TotalFinaElf (GRC) UK. We gratefully acknowledge the support and cooperation of all sponsors. We thank Enterprise Oil, Statoil. TotalFinaEIf, and their partners for providing access and permission to publish data. MH received support from the Danish Natural Science Research Council (Grant nos. 51-000429 and 21-01-0430). Finally, we would like to thank S. Shoulders and the reviewers D. Erratt and R. J. Dixon for their very helpful comments and suggestions. The ideas and interpretations presented herein are those of individuals and thus do not necessarily reflect those of the mentioned companies or their partners.
276
M. HUUSE ET AL.
References BROOKE, C. M., TRIMBLE, T. J. & MACKAY, T. A. 1995. Mounded shallow gas sands from the Quaternary of the North Sea: analogues for the deformation of sand mounds in deep water Tertiary sediments? In: HARTLEY,A. J. & PROSSER, D. J. (eds) Characterisation of Deep-marine Clastic Systems. Geological Society, London, Special Publications, 94, 95-101. CARTWRIGHT, J. A. ~; LONERGAN, L. 1996. Volumetric contraction during the compaction of mudrocks: a mechanism for the development of regional-scale polygonal fault systems. Basin Research, 8, 183-193. COLE, D., STEWART,S. A. & CARTWRIGHT,J. A. 2000. Giant irregular pockmark craters in the Palaeogene of the Outer Moray Firth Basin, UK North Sea. Marine and Petroleum Geology, 17, 563-577. COSGROVE, J. W. & HILLIER, R. D. 2000. Forced-fold development within Tertiary sediments of the Alba Field, UKCS: evidence of differential compaction and post-depositional sandstone remobilization. In: COSGROVE, J. W. & AMEEN, M. S. (eds) Forced Folds and Fractures. Geological Society, London, Special Publications, 169, 61-71. DAVIES, R., RANA, J. & CARTWR1GHT,J. A. 1999. Giant hummocks in deep-water marine sediments: Evidence for large-scale density inversion during burial. Geology, 27, 907-910. DIXON, R. J., SCHOFIELD, K., ANDERTON, R., REYNOLDS. A. D., ALEXANDER, R. W. S., WILLIAMS, M. C. & DAVIES, K. G. 1995. Sandstone diapirism and clastic intrusion in the Tertiary submarine fans of the Bruce-Beryl Embayment, Quadrant 9, UKCS. b~: HARTLEY, A. J. & PROSSER, D. J. (eds) Characterisation of DeepMarine Clastic Systems. Geological Society, London, Special Publications, 94, 77-94. DURANTI, D., HURST, A., BELL, C., GOVES, S. & HANSON, R. 2002a. Injected and remobilized sandstones from the Alba Field (Eocene, UKCS): core and wireline log characteristics. Petroleum Geoscience, 8, 99-107. DURANTI, D., HURST, A., HUUSE, M. & CARTWRIGHT.J. A. 2002b. Sand diapiric structures and poly-phase sand remobilization in the Santa Cruz area (Central Coastal California). 64th EAGE Conference & Exhibition, Florence, Extended Abstracts, H024. DURANTI,D., HUUSE, M., CARTWR1GHT,J. A., HURST, A., CRONIN, B., MAZZINI, A. & FLANAGAN, K. 2002c. Unusual facies and geometries of the Paleogene deep-water systems in the North Sea: effects of sand remobilization. 64th EAGE Conference & Exhibition, Florence, Extended Abstracts, P057. GRAS, R. & CARTWR1GHT, J. A. 2002. Tornado faults: the seismic expression of the Early Tertiary on PS-data, Chestnut Field, UK North Sea. 64th EAGE Conference & Exhibition, Florence, Extended Abstracts, H020. GUARGENA, C. G., SMITH, G. B., WARDELL, J., NILSEN, T. H. & HEGRE, T. M. 2002. Sand injections at Jotun Field, North Sea--Their possible impact on recoverable reserves. 64th EAGE Conference & Exhibition, Florence, Extended Abstracts, H018. HOVLAND, M. & JUDD, A. G. 1988. Seabed Pockmarks and Seepages-Impact on Geology, Biology and the Marine Environment. Graham & Trotman, London. HUUSE, M., DURANTI, D., CARTWRIGHT,J. A.. HURST, A. & CRONIN, B. 2001. Seismic expression of large-scale sand remobilisation and injection in Paleogene reservoirs of the North Sea Basin and beyond. 63rd EAGE Conference & Exhibition, Amsterdam, Extended Abstracts, L07. JENNETTE, D. C., GARFIELD,T. R., MOHRIG, D. C. & CAYLEY, G. T. 2000. The interaction of shelf accommodation, sediment supply and sea level in controlling the facies, architecture and sequence stacking patterns of the TaT and Forties/Sele basin-floor fans, central North Sea. In: WEIMER,P., SLATT, R. M., COLEMAN, J., ROSEN, N. C., NELSON, H., BOUMA, A. H., STYZEN. M. J. & LAWRENCE, D. T. (eds) Deep-Water Reservoirs of the World, GCSSEPM Foundation, 20th Annual Conference, Houston, 402-421.
JENSSEN, A. I., BERGSLIEN, D., RYE-LARSEN, M. & LINDHOLM, R. M. 1993. Origin of complex mound geometry of Paleocene submarinefan reservoirs, Balder Field, Norway. In: PARKER, J. R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 135-143. JOLLY, J. H. R. & LONERGAN, L. 2002. Mechanisms and control on the formation of sand intrusions. Journal of the Geological Socie~, London. 159, 605-617. JONES, E., JONES, R., EBDON, C., EWEN, D., MILNER, P., PLUNKETT, J., HUDSON,G. & SEATER,P. 2003. Eocene. In: EVANS,D., GRAHAM,C., ARMOUR, A. & BATHURST, P. (editors and coordinators) The Millennium Atlas: Petroleum Geology of the Central and Northern North Sea. Geological Society, London, 261-277. LONERCAN, L. & CARTWRICHT, J. A. 1999. Polygonal faults and their influence on reservoir geometries, Alba Field, United Kingdom Central North Sea. AAPG Bulletin, 83, 410-432. LONERGAN, L., LEE, N., JOHNSON, H. D., CARTWRIGHT,J. A. & JOLLY, R. J. H. 2000. Remobilization and injection in deepwater depositional systems: implications for reservoir architecture and prediction. In: WEIMER, P., SLATT, R. M., COLEMAN, J., ROSEN, N. C., NELSON, H., BOUMA, A. H., STYZEN, M. J. & LAWRENCE, D. T. (eds) Deep-Water Reservoirs of the World, GCSSEPM Foundation, 20th Annual Conference, Houston, 515-532. LOSETH, H., WENSAAS, L., ARNTSEN, B., HANKEN, N., BASIRE, C. & GRAUE, K. 2001. 1000m long gas blow-out pipes. 63rd EAGE Conference & Exhibition, Amsterdam, Extended Abstracts, P524. LOSETH, H., WENSAAS, L., ARNTSEN, B. & HOVLAND, M. 2003. Gas and fluid injection triggering shallow mud mobilization in the Hordaland Group, North Sea. Geological Society, London, Special Publications, 216, 139-157. LUCHFORD, J. 2002. The value of Pre-stack depth migration in evaluating apparent injection features. 64th EAGE Conference & Exhibition, Florence, Extended Abstracts, H022. MACLEOD, M. K., HANSON, R. A., BELL, C. R. & MCHuGO, S. 1999. The Alba Field ocean bottom cable seismic survey: Impact on development. The Leading Edge, 18, 1306-1312. MIKHAILOV, O., JOHNSON,J., SHOSHITAISHVILI,E. & FRASIER,C. 2001. Practical approach to joint imaging of multicomponent data. The Leading Edge, 20, 1016-1021. MOLYNEUX, S., CARTWRIGHT, J. & LONERGAN, L. 2002. Conical sandstone injection structures imaged by 3D seismic in the central North Sea, UK. First Break, 20, 383-393. NEWMAN, M. ST. J., REEDER, M. L., WOODRUFF, A. H. W. & HATTON, I. R. 1993. The geology of the Gryphon Oil Field. In: PARKER, J. R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 123-133. NEWSOM, J. F. 1903. Clastic dikes. Geological Society of America Bulletin, 14, 227-268. NEWTON, S. K. & FLANAGAN,K. P. 1993. The Alba field: Evolution and depositional model. In: PARKER, J. R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 161-171. NIELSEN, O. B., SORENSEN, S., THIEDE, J. & SKARBO, O. 1986. Cenozoic differential subsidence of North Sea. AAPG Bulletin, 70, 276-298. PARIZE, O. 1988. Sills et dykes grrseux srdimentaires: palOomorphologie, fracturation prrocce, injection et compaction. Ecole des Mines des Paris. Mdmoires des Sciences de la Terre, 7. SMYERS, N. B. & PETERSON, G. L. 1971. Sandstone dikes and sills in the Moreno Shales, Panoche Hills, California. Geological Society of America Bulletin, 82, 3201-3208. SURLYK, F. & NOE-NYGAARD, N. 2001. Sand remobilization and intrusion in the Upper Jurassic Hareelv Formation of East Greenland. Bulletin of the Geological Socie~' of Denmark, 48, 169-188. TEMPLETON, G., KING, P. & REEDER, M. 2002. Leadon Field-Description of Frigg reservoir sand injection structures. 64th EAGE Conference & Exhibition, Florence, Extended Abstracts, H017.
SANDSTONE INTRUSIONS, NORTH SEA THOMPSON, B. J., GARRISON, R. E. & MOORE, C. J. 1999. A late Cenozoic sandstone intrusion west of Santa Cruz, California: Fluidized flow of water- and hydrocarbon saturated sediments. In: GARRISON, R. E., AIELLO, I. W. & MOORE, J. C. (eds) Late Cenozoic Fluid Seeps and Tectonics Along the San Gregorio Fault Zone in the Monterey Bay Region, California. AAPG Pacific Section, Guide Book, GB-76, 53-74. THYBERG, B. I., JORDT, H., BJORLYKKE, K. & FALEIDE, J. I. 2000. Relationships between sequence stratigraphy, mineralogy and
277
geochemistry in Cenozoic sediments of the northern North Sea. In: NOTTVEDT, A. (ed.) Dynamics of the Norwegian Margin. Geological Society, London, Special Publications, 167, 245-272. VELDE, B. 1996. Compaction trends of clay-rich deep sea sediments. Marine and Petroleum Geology, 133, 193-2001. ZIEGLER, P. A. 1990. Geological Atlas of Western and Central Europe. Shell Intemationale Petroleum Maatschappij BV, The Hague.
Integrated use of 3D seismic in field development, engineering and drilling: examples from the shallow section BRYN
AUSTIN
Brynterpretation Ltd, Windmill Cottage, Saxlingham Nethergate, Norfolk NR15 1PB, UK (e-mail: bryn @b~nterpretation, co. uk)
Abstract: Examples from exploration acreage and field developments across the glaciated Northwest European Continental Shelf and Slope demonstrate the usefulness of conventional three-dimensional (3D) seismic data to spatially image geological features. Compared to previous grid-based two-dimensional (2D) seismic this allows fundamentally more confident identification, mapping and prediction of geotechnical conditions which is important to ensure safe, efficient engineering and drilling operations. Whilst of immense benefit, the paper argues that the 3D seismic data often do not meet the full expectations, particularly in terms of critical vertical resolution and accurate depth prediction requirements. To illustrate the limitations, direct comparison is made between conventionally acquired 3D and HiRes 2D seismic data. Whilst industry funding to support innovative HiRes 3D seismic acquisition remains sparse, much can be achieved by the careful integration and interpretative calibration of the 3D and HiRes 2D seismic datasets. Three field development case studies illustrate this. Short offset trace correction and reprocessing of the 3D seismic data followed by limited, target specific HiRes 2D seismic, calibrated where possible with drilling or other geological data. is an optimal cost-effective approach.
An understanding of the nature of the 'shallow section' impacts on the planning of many operations in the period from initial investment for an offshore exploration well through to the removal of a production facility. Operations during that period include anchor laying and chain recovery, drilling wells and attaching production pipelines or facilities to the sea bed. During the past decade an order of magnitude increase in our confidence in the interpretation and prediction of 'shallow section' soils conditions has occurred. This is due to the powerful addition of increased continuity and coherent pattern recognition visible in the spatial dimension provided by some 3D exploration seismic datasets. This paper examines the use of marine 3D seismic data in the shallow section. It describes the main considerations and background issues including vertical resolution through detailed description of data examples. These come from selected environments along the North West European Continental Shelf and Slope (NWECSS) (Fig. 1). They range from the < 100 m shelf depths of the Central North Sea (CNS), down the West Shetland and East Faroes Slopes to depths greater than 1 - 2 k i n in the Faroe-Shetland Channel or offshore MidNorway/V0ting. A larger set of 34 examples including features maps may be viewed at www.geolsoc.org.uk (Lappin et al. 2002). The paper illustrates the drawbacks of relying exclusively upon 3D seismic data bearing in mind the need for answers to the range of questions required when exploring and developing resources offshore. Three example developments from the Norwegian Sea, the West Shetland Slope, and the Central North Sea are presented (Fig. 1). They attempt to show workable solutions, through the integration of 3D seismic with HiRes 2D seismic (plus borehole or drilling data if available), where a host of cross-discipline subsurface challenges have been met. The 'shallow section' is defined as 'the interval from the sea bed down to the maximum sub-sea bed depth at which normal well killing procedures can be applied by means of a Blow Out Preventer (BOP). This depth generally corresponds to the depth at which suitable casing can be set'. This may be greater than 1.2 km below sea bed, according to the Guidelines prepared by the United Kingdom Offshore Operators Association (UKOOA 1997). The surficial sea bed with its fundamental interaction to the dynamic overlying ocean or shelf sea system is naturally considered here as an integral part of the shallow section. Several morphological features, varying in age from Miocene to Recent, that create abrupt sea bed slope gradient changes and
sediment consolidation/geotechnical property diversity are displayed in Figure 1. In the Faroe-Shetland Channel the eastern shelf margin sea bed is characterized by the Rona Apron--a series of debris flows of Late Pleistocene age. These are termed the Morrison Unit II by the British Geological Survey (BGS) and have sea bed dips of less that 0.5 ~ gently increasing to 3 ~ into the channel base. Here the very low gradients along the basinal axis of the Faroe-Shetland Channel are sharply in contrast, however, to the sinuous, scarp-like exposures of Ypresian bedrock exposed as the Judd Deeps to the southwest, Smallwood (2004). Here sea bed gradients of > 4 0 ~ are documented (Austin 2001; Long et al. (2004), an example of which is shown in profile (Fig. 2). Current scoured exposed lows are fully or only partially filled as a Palaeogene to Recent monoclinal culmination competes with climatically variable bottom current activity (Fig. 2). The region covered by Figure l overlies prime exploration targets beneath over a kilometre of water resulting in drilling conditions that require greater sensitive environmental care and consideration than elsewhere.
Engineering considerations Given the nature of oil and gas field development, operational and engineering requirements in terms of depth within the shallow section may vary considerably at any particular location. Geological variability and rate of change is also of great importance. This is the case at a site specific scale as well as for cable laying or pipeline product transport over sometimes thousands of kilometres. Consider mariners laying and recovering anchors, chains and tethers in deep water. Here current systems may oppose each other and bedrock might crop out as one end member e.g. the Faroe-Shetlands Channel. Elsewhere the soup-like consistency of the sea bed itself can be virtually indistinguishable from the water column above, e.g. Niger or Mississippi Delta. For pipeline laying and burial or cable routes, for instance, a detailed understanding of sub-sea bed soils, sediments or rock is needed in the range from 0 - 5 m. Similarly for tethered facilities or Jack-Up rig installation, knowledge of the upper sediment layers is crucial to avoid skirt buckling or punch-through. Foundation engineers likewise locating templates or other facilities equipment on the seabed need detail. Estimating stability and the geotechnical integrity of deeper strata to about 50 or 150m is fundamental for sea bed
DAVIES, R. J., CARTWRIGHT,J. A., STEWART,S. A., LAPPIN.M. & UNDERHILL.J. R. (eds) 2004.3D Seismic Technology: Application to the Exploration of Sedimentary. Basins. Geological Society, London, Memoirs, 29, 279-296. 0435-4052/04/$15 9 The Geological Society of London 2004.
280
B. AUSTIN
Fig. 1. Sea bed image of 5000 klTl 2 of the West Shetland Slope (WSS) and the Faroe-Shetland Channel (FSC) to indicate the variety of morphological features expected in glaciated margins. Image is a dip azimuth attribute display of the carefully picked sea bed horizon from five different but merged 3D seismic data volumes (blue outlines) acquired for exploration purposes. The location of important oilfields Foinaven, Schiehallion, Loyal and Suilven are annotated F, Sch, L and S respectively. Bathymetry increases northwestwards from 200 m to greater than 1000 m where the Rona Apron debris flows (derived from Scottish ice sheets) characterize the sea bed West of Block 204/20 highlighted in yellow. A rapid increase in slope gradient from 0.5 ~ to 4 ~ is preserved as the textural change between S and S'. Westwards the debris flows begin to amalgamate and form a gently dipping toe of slope wedge upon the FSC floor. Points W locate the marked slope decrease. From regional seismic correlation the flows appear similar in origin to those shown as Figures 5, 8 and 9 located along the East Faroes Slope (EFS) on the opposite flank of the FSC. The Judd Deeps (JD) are a 40 km long complex area of infilled seabed scours of ?Miocene age. These are partially exhumed by active bottom currents forming 200 m high scarps of eroded Eocene bedrock (Fig. 2). Red symbol shows location of BGS Borehole 99/03 and numbered red lines locate other text figures. Inset map shows location of the three case studies documented in the paper.
~--~ Z ' ~
"~.x'~4~Ik~..,~
--
~
ii.lillIC 4dmlsllmw aml ammi11~ ,
~
/
.....
~ r
INTEGRATED USE OF SHALLOW SECTION 3D SEISMIC
281
Table 1. Shallow section interests per discipline from the Geotechnical, Environmental and Marine (GEM) Project (Faroes Islands shelf and slope) Geohazard issue of interest or concern Slope instability
Drill* 9
Mar* 9
Pipe t
Fac ~
9
9
Slide scars & slide/slump seds
9
9
Debris flow & turbidite depos.
9
9
Creep & compression features
9
9
9
9
Env II 9
Ex&Dev I 9
9
Diapirism
9
9
Bathymetry & slope gradient change
9
9
9
9
9
Seabed features e.g. sandwaves
9
9
9
o
9
Water column current regime
9
9
9
9
9
Obstructions/ordinance
9
9
9
9
Shallow faulting
9
9
9
9
Carbonate mounds
9
9
9
9
9
Igneous intrusions/lava fields
9
9
9
9
9
9
Contourite sediments
9
9
9
9
9
Overconsolidated till
9
9
9
9
Cobble/Boulder beds
9
9
9
9
9
Coarse gravels
9
o
9
o
9
Geotechnical variation
9
9
9
9
Shallow gas presence
9
9
9
o
Hydrate presence
9
9
9
Unconsolidated/flowing sands
9
9
Fluid expulsion
9
9
9
9
9 9 9
o
9
* Drillers. * Mariner. * Pipeline Engineers. wFacilities Engineers. rlEnvironmental Scientists. I Expln. & Development Geoscientists.
installations when piling jackets, emplacing gravity bases or fixing Tension Leg Platforms or other moorings with suction or anchor piles. Pipeline engineers are also involved early in a development to design, lay and link subsea flowlines for the product export line. Like reservoir engineers, their involvement often extends from initial d e v e l o p m e n t throughout field life that during tertiary recovery can cover three decades or more. Those engineering considerations of particular relevance to the Faroese offshore are listed in Table 1 ( A t l a n t i c o n - G E M 2001). Here the shallow section stratigraphic f r a m e w o r k o f the Faroese offshore is the result o f regional integrated seismic investigation (Table 2). Major N e o g e n e facies differentiation occurs due to
b a t h y m e t r i c and climatic factors that strongly influenced syn- and post-depositional erosion by bottom currents (Austin 2001). Sediments were plastered upon regionally extensive a n g u l a r u n c o n f o r m i t i e s , resulting in a variable s e d i m e n t distribution covering bedrock. Significant geotechnical extremes have resulted that are worthy o f engineering interest (Figs 1 - 5 , 7 - 9 , 12 & 14). M a n y of the geological aspects and engineering considerations alluded to above are important elsewhere over the N W E C S S and also other margins affected by Plio-Pleistocene climatic changes. Furthermore, integrated use o f 3D seismic data in field development, engineering and drilling has added c o n s i d e r a b l e i n p u t to the s t u d y o f r e c e n t n e a r - s u r f a c e
Fig. 2. Composite HiRes 2D profile located across the Judd Deeps in axis of Faroe-Shetland Channel (Fig. 1) showing contrasting sea bed conditions and much variation in stratigraphy within the shallow section. Palaeocene seismic stratigraphy is dominated by a series of onlaps and depositional wedges (DW) resulting from late Palaeogene and Neogene monoclinal folding. Note velocity pushdown apparent at near top Palaeocene (TP light blue) event beneath Judd Deeps scour hollow. Here the incised Late Oligocene to Early Miocene Unconformity (LOEMU) preserves assumed Miocene sediments beneath Late Pleistocene cover overlying the Intra-Neogene Unconformity (INU) in yellow. These are interpreted to be deposited by contourite current processes given their downlapping, compound mounded internal reflection patterns. Accurate prediction of rock and sediment geotechnical contrasts is of critical importance to engineering and drilling in this vicinity where a 70 m deep contourite moat (CM) lies at the base of a 200 m high scarp sloping at 40 degrees. The complicated shallow section conditions, where known, have soft-firm late Pleistocene Unit 1CI soft-firm silty clays and dipping, cemented sandstones interpreted to be of Ypresian (early Eocene) age as end members. The latter are exposed as cobble/gravel strewn pavements on the rugged, upstanding eroded dip slope (DS). A sporadically strong bottom current, ( 1.5 m/s), flows towards the SW opposing the N. Atlantic Drift in the remainder of the water column. Seismic stratigraphical framework terms such as the GU, INU, LOEMU (unconformities) and Unit I C! are described in Table 2.
282
B. AUSTIN
Table 2. Regional Stratigraphic Framework, Faroes offshore Age
West shelf
Slope
Basin east
Holocene-Recent
0- l 0 cm lag
0- 20 cm lag
0- 30 cm lag
Unit IA
Unit ICI
lgu Unit 1CI
Late Pleistocene
lgu
Unit IB
Unit 1CII
Pleistocene
gu Unit 1B
Unit IB
Unit 1CIII
Unit 1B
gu
gu
Pliocene
INU
INU
INU
?Miocene
Unit 2
Unit 2
Unit 2
LOEMU
LOEMU
LOEMU
?Eocene-Oligocene
Unit 3
Unit 3
Unit 3
?Palaeocene - Late Eocene
Unit 4 Plateau Basalts
Unit 4 Basalt margin
Unit 4
Key: lgu, Late Glacial Unconformity; gu, Glacial Unconformity: INU, Intra Neogene Unconformity; LOEMU, Late Oligocene, Early Miocene Unconformity. LOEMU is equivalent to Top Palaeogene Unconformity, (STRATAGEM Partners 2003).
sedimentary processes and environments or relict environments. Geologists, geophysicists and environmental scientists alike can benefit from the essential snapshots that may provide useful analogues to several important depositional and erosional systems applicable to the geological record (Fig. 6 and Salisbury et al. 1996). One further example of the requirement to understand the shallow section is the fundamental assessment of the stability of the sea bed itself. What is the likelihood of sediments sliding downslope either naturally or in response to excess loading? This can be caused by geostrophic current erosion, internal fracturing of contourite sediments under increased loading, pore pressure fluctuation and seismic forces or simply excessive local sedimentation for example. Could the wellhead/riser be swept away or buried by catastrophic debris flows or slumps initiated kilometres away upslope? Such questions are not only important during the relatively short lifespan of an exploration well. Risk of local and regional instability must be carefully considered for the 30+ year lifespan of a large producing field, e.g. Troll, Ekofisk or, currently, the Ormen Lange gas development located proximal to the head of the Storegga Slide (Fig. 6).
leads to widening or relaxing of such specifications. For one deep-water Gulf of Mexico field, a simple compromise on site survey data quality, (a supposedly fit-for-purpose acquisition) resulted in drilling overspends and lost opportunities costing $30M. Worldwide estimates show a total annual cost of $1.5bn attributable to poor local drilling performance (Stewart & Holt 2004). Much of this spend is due to experiences within the shallow section and deeper overburden. Indeed industry spend on wells experiencing shallow-water flow problems was an average remedial $1.6M per well and importantly only 20% of this was spent on predrill prediction (1999 estimates, Dutta 2002). Consequently, for interpreters of the shallow section, 3D exploration seismic data with uniform dense sampling is a powerful tool to have at one's disposal. This is especially the case as industry develops new and under-explored areas of the ocean. Beyond the marked increases in slope approaching and beyond the shelf break, quite different and often unexpectedly complex geological conditions can be found in contrast to those previously dealt with in shallower waters of the shelf (Fig. 2).
Vertical resolution Historical background As many of the following examples given in this paper demonstrate, most 3D seismic data is regarded as superior over most 2D seismic data. This is primarily because of the good spatial continuity and reasonable resolution inherent in 3D datasets, which gives far greater continuity and connectivity and, allied with modern interpretation software, allows creation of superb slice images (Figs 3, 4 & 5). Such sampling of most larger- or medium-scale natural features of interest allows them to be positively identified and mapped. Indeed in deeper water environments the regional coverage can be better than most (e.g. TOBI) sidescan sonar results, often reducing this requirement considerably apart from final clearance inspection (Figs 6 & 8). Shallow section investigation has long been an underfunded part of the exploration and development process. As a result, acquisition of 2D seismic data for site-specific surveys has used the traditional survey grid approach. The increased confidence in interpretation gained from utilizing 3D seismic is hardly surprising given the aliasing effect of 2D data. Spatial aliasing of features is the case even under tight 2D grid specifications of 12.5-25m spaced lines. Weather dependency or misplaced financial constraint commonly
Unfortunately, the good lateral resolution of 3D seismic data does not mean the offshore industry can rely upon it solely to meet all shallow engineering requirements. Enhanced vertical resolution is also required. Some obvious visual or qualitative differences in near vertical inline resolution between HiRes 2D (SAG--Single Air Gun source) and exploration 2D/3D seismic (source arrays) are shown (Fig. 7). With 3D coverage the powerful factor of 'detectability', similar to the 3D 'limit of separability' (Brown 1991) is crucial to utilize the full benefits of digital amplitude measurement. Detectability may be estimated by comparison of Figures 5 & 8 plus the associated 3D seismic profile (Fig. 9). Note that the 3D seismic data used in the example is clipped just 0.300 s TWT below sea bed. This illustrates the conflict between (a) the commercially sensitive nature of much exploration and development 3D seismic data and (b) the ability to properly meet drilling risk reduction objectives. The vertical component of resolution is generally equal if not more important in the interpretation of a location or site during field development. It is here that conventional exploration and development 3D seismic data is clearly lacking. Four millisecond processed 3D seismic cannot hope to match the vertical resolution of single or sub-millisecond sampled and
INTEGRATED USE OF SHALLOW SECTION 3D SEISMIC
Fig. 3. Example of the superb spatial imaging possible from conventional exploration 3D seismic data in deep water from the base of the East Faroes Slope in the Faroes-Shetland, Channel (EFS located on Fig. 1). (A) Two-way Time (TWT) mapped surface shows Pleistocene contourite drift sediment wave crests (W) and troughs (T) where wavelengths are 1 km with amplitudes of I 1-12 m. Crests trend perpendicular to present day bottom current for at least 10-15 km. (B) Seismic profile line A-A' is low vertical resolution 2D exploration seismic data. It shows the waves buried beneath 60 m of Unit 1CIlI late Pleistocene overburden (P). The deposits rest upon a composite regional unconformity surface, (INU/LOEMU-see Table 2), above Eo-Oligocene deposits of the Suderoy High (Austin 2001).
.
B
.
~
.
.
.
.
.
.
processed High Resolution (HiRes) and Ultra High Resolution (UHiRes) seismic acquisition (Figs 3 & 10). The higher frequency and broader bandwidths are essential to supply safe but realistic engineering predictions to engineer and driller customers. This is clearly shown on the seismic data comparison examples but is a factor too often undervalued in the world of deeper towed streamers and shaved budgets (Fig. 11). Industry studies in deep water have shown that there is less difference in the vertical resolutions of HiRes 2D seismic and conventional 3D seismic data as source-signal stretch is reduced. This is due to increasingly near-normal incidence of the reflected wave fronts with increasing water depth. There is, however, a difference between the in-line vertical resolution of HiRes 2D seismic and conventional 3D seismic data over most continental shelves where normally only a rudimentary or muted sea bed is observable on 3D seismic. Even in deep water, much important information would never be obtained if one adopted the 'we've shot and paid for expensive 3D seismic and data processing, so why should we shoot even more seismic data?' approach. A classic example where reliance upon 3D seismic data alone would have been insufficient comes from the Faroe-Shetland Channel. The first Faroese offshore exploration well was to be spudded close to the locality shown in Figure 12 only weeks after these HiRes 2D seismic data were first interpreted. Here previously unidentified low-relief mounded features are observed in 1000m of water some 4 0 m below sea bed. Mapping of the 2 x 4 km spaced reconnaissance seismic data grid shows the features to be isolated as individual patches and therefore unlikely to consist of downslope flow sediments. Analogy with similar features recognized and researched in the Irish Rockall for example (Kenyon et al. 2003) suggests that
.
283
wavcl~r~
.
.....
-
- -
they are most likely chemosynthetic/cold water coral build-ups lying upon the Late Oligocene-Early Miocene Unconformity (LOEMU) or Top Palaeogene Unconformity as it has recently been termed. This is an unconformity of regional proportions extending from offshore Norway into the southern Rockall Basin, (Stoker 1999; STRATAGEM Partners 2002, 2003). Carbonates will have considerable geotechnical contrasts with the surrounding clastics and pose a significant challenge to drilling and large diameter casing emplacement if encountered unexpectedly. Even with hindsight, such features, which are estimated as up to 2 0 - 3 0 m thick, are barely recognizable on the exploration 3D seismic data.
A calibrated 2D/3D approach Ideally there is a justifiable need for HiRes 3D seismic data acquisition. Intrinsic to this statement, however, is the increased financial outlay due to a variety of physical factors. Notably these include poor signal-to-noise ratio and multiple suppression plus the mechanical difficulties of towing longer, closely spaced, multi-streamer and source arrays at the shallow tow depths required for HiRes acquisition. Such financial and physical barriers are beginning to be broken in the deep water Gulf of Mexico. This has partially come about by the realization that to understand the nature of and hence predict a particular major geohazard, (slightly overpressured flowing sands), is to actually save money in the longer term. This is a non-glaciated environment, however, and thus acoustically far more benign than the NWECSS. Until such commercial and physical factors are surmounted here a calibrated 2D/3D seismic approach may be adopted.
NE
Off'set TWT
tt
""
" " ~" a l p p e d h n # ~r
UnIt .t
,
- - 1I I I I I 0
S krn
:
| :
f r u i t X i, 0.015 (! Inn) at
aec
Fig. 4. Recent faulting interpreted to cut the sea bed near foot of East Faroes Slope (see Figs 1 and 5 for location). (A) Seabed bathymetry in TWT from 3D seismic data shows subtle NNE/SSW trending lineaments (F) noticeable as glitches whilst autopicking. Note subtle difference to inline acquisition/processing artefacts (AP) that similarly limit autopicking. Regular grid is 3D Line/ Trace; other lines are coiour-coded database of 2D Exploration, HiRes 2D or UHiRes 2D seismic data. (B) Down to the basin faulting is clearly observed on the HiRes 2D seismic profile A - A ' and the corresponding lower resolution exploration 3D seismic random track shown below in (C) that is similarly located. The HiRes 2D seismic was acquired specifically to confirm the faulting suspected from interpretation of the 3D data in order to formulate risk of slope failure as identified more than 150 km further north. Apparent offset of the seabed reflection at fault X is 11 m. See Table 2 for unit stratigraphy and unconformity nomenclature.
Fig. 5. 3D seismic data images of the seabed from Sandoy Fan (SF) region of the Faroe-Shetland Channel showing zero requirement for regional sonar to delineate major geological features. The 3D seafloor imagery covers some 2200 km 2. (A) Display of Seabed Dip attribute defines heavily iceberg scoured Faroe Embayment Shelf (FES) and steep E. Faroes Slope (see Fig. 4) with NNE/SSW trending contourite moat feature at base (M). The SF amalgamated debris flows have moved over 40 km downslope and overlie older Pleistocene basinal fill that exhibits polygonal faulting exposed through bottom current scour/non-deposition. The SF featheredge extends slightly further southeast than that of a featureless contourite sheet outlined with dashed line. The sheet was deposited in the lee of the Sandoy High faulted anticlinal axis (SH) and thickens into the moat feature. Interfingering between contourite and debris flow deposits is suggested by age dated piston core samples (red triangles) and gravity cores, (blue triangles), collected along the HiRes and UHiRes 2D database shown in red and blue. (B) Reflection Intensity attribute display of the seabed pick clearly shows most westerly (younger) debris flow, (MWDF), riding down into, up and over the contour current generated moat with little deflection (along line PSAT98-44). See also the calibrated 3D seismic data profile (Fig. 9) highlighted in red.
INTEGRATED USE OF SHALLOW SECTION 3D SEISMIC
Fig. 6. (A) Colour coded swathe bathymetry image from the E. Norwegian Sea showing the Storegga Slide slope instability feature. The location of the giant Ormen Lange gas accumulation (in red) with the footprint of the Havsule 3D seismic data acquisition (yellow box) are dwarved by the massive scale of Storegga illustrating the requirement for regional awareness. (B) Sea bed amplitude display from the Havsule 3D seismic dataset after reprocessing to a low fold, short offset volume. The display shows a small portion of the massive Storegga Slide, covering approx. 25 x 50 km, located 50 km from the base of the steep headwall slopes. Specially processed 3D seismic data such as these allow phenomenal increases in our ability to visualize and research the catastrophic gravity driven mass movement processes important for understanding risk to slope instability. Numbering refers to relative timing of the multiphase mass flows occurring at seabed where 5 is oldest. Lettering is the order of discrete sub-flows interpreted from local cut offs and broader geometrical relationships now visible at this sub-regional scale. Knowledge of the rate at which mass wasting processes occur is critical to estimate risk throughout the lifetime of a producing field.
i
285
-loo .~,
N Havsule 3D seismic coverage (yellow) and Ormen Lange field (red)
|
There is tremendous benefit in first calibrating 3D seismic with HiRes 2D seismic data and then using the greater 3D seismic spatial coverage to explore the shallow section for differences and anomalies (Figs 5, 8 & 9 or Fig. 12). With due
NW IZi(m) Vp'm/sec) I 18.50 ~'* 1485 18.75
1500
20.00
1600
|
c a r e and attention this t e c h n i q u e has r e c e n t l y b e e n successfully e m p l o y e d , during large regional investigations like the Norwegian Seabed, Western Frontiers Association (WFA) or Faroese Geotechnical, Environmental and Marine
ESE/NW [ SAGSeabedpulse [ length = I 1.25msec
[
" ~ - ~
.
N
"
~IC!
SE/NE
abea =
SW
I I
'
~'~.~ : ~
0
Fig. 7. Data comparison along the same line showing aspects of inline vertical seismic resolution. The velocity profile shown is typical of late Oiigocene through Holocene sediments along the NE Atlantic Margin at least from S. Rockall-N. Faroes. Limit of Separability (L of S) is derived from Brown (1991). (A) Relatively poor quality 2D HiRes seismic data, sourced by a single air gun (SAG); (B) Exploration 2D seismic data; (C) 3D seismic data conventionally acquired to image exploration targets beneath the sloping Faroe-Shetland Channel seabed.
21.25 25.00
1700
,~" "',~.
000
~
0-
:1
LGU
_
!
.~'t~nil 2
Unit 3
k o f S = 2.78m
k o f S = 7.75m
k o f S = 8.98m
286
B. AUSTIN
Fig. 8. Depositional architecture of the Sandoy Fan amalgamated debris flows through time from detailed mapping of 3D seismic data calibrated by HiRes 2D seismic data tied to surficial core samples. (A) Reflection Intensity (R.I.) display shows surficial expression at present day sea bed--youngest debris flow is approx. 7000 years BP; (B) R.I. display of distinctive erosive base to larger flow elements; (C) Enlarged R.I. display of amalgamated debris flows (3rd order) located at base of late Pleistocene Unit I CI. See Figure 9 for 3D seismic profile highlighted in red in (A).
WNWz79 5 u 6 T~,~,-I"
~
(ms) 800
~
SE S W s 7 ~
4826
4865
4903
4942
4990
501':t
5057
5095
5133
5171
5209
524'7
5.-'85 ('1)1"~ N ~ - .
;
Line Intersect~on
EFS
2nd
SB
900 Contourite moat axis
1000
1t 0 0
1200
l
3rd Enlargement
Sandoy Fan
Conto~tes
C)~der. dlscrele debris flow lenses can be rnal:~:c,d on 3D sesn~c 4green Irough ;4ck~
Limit
1300
J of fan
MWDF
ind~clual d~lxis Iow~ at seabed show internal bl~ridl=eedmnal down
1400
lap and mounded external g e o w ~
I
approx. ]
~ .
.
.
.
.
~
.-~.-O~-:'*~~-'~? ~'~ -
~.7"~.
1500 .- .....
~._
~
~.
"Sub-Unit 1 O I
1600 Acoustically layered sediments of Unit ICIII
Same internal refleclocs
I
,
1700 0
1800
10 km
Z0 km ~nJctureless , chaoQc ~po~bon
i
Fig. 9. 3D seismic profile across distal portion of Sandoy Fan showing hummocky sea bed relief and interfingering relationship with contourite sedimentation. For location see Figures 5 and 8. Recovered piston core samples to 2.4 m record Late PleistoceneHolocene ages. Seismic interpretation, however, shows the depositional system forms uppermost part of developing Sandoy and Suderoy Basin fill considered to have initiated in late Oligo-early Miocene times. This cannot be seen here due to commercial 3D data cut-off. Note gross vertical exaggeration: the East Faroes Slope (EFS) grades at a maximum of 4 degrees. MWDF is most westerly debris flow observed to cross the contourite moat identified in Figures 5 and 8. SB is the sea bed pick whilst 2nd and 3rd refer to sub-sea bed reflections picked to generate the attribute maps of Figures 5 and 8. Seismic stratigraphical framework terms such as the LGU, INU and Sub-Unit I CII are described in Table 2.
I N T E G R A T E D USE OF S H A L L O W SECTION 3D SEISMIC
YWTIB
m s
287
SW
NE
( w a t e r c o l u m n c l i p p e d )
300
.wF 7-:
....
~-'-
.......
:._
~
~
;-~:-:~Y.
9
"'~
, "''""
+:1"50
--
.
:+'+ ='+.
los "
" r
500
-
t o,
,
- .: ~"
-"-
,-
" +
.
:t+
"-- +. . . .
-
~ _ , -+"+
.
:
.
.
.
.
.
......
." " - -
.
f
'
-"
" -~+~,
.'
.
"
-
+-+.'
"- :
;
.
,i~ ~
-.
" - "++ I ~
" . . . . .
.
.
.
"": ..... +-
-
,"
-':
" ..... .
+.+.~,~
.
.
'
-_:,
+r.-l,,+~. ~ ,~ .,~+,+ . . . . . . . . . . . . ,~~+ 6.0 which occurred between 1911 and 1992. Vector at bottom left represents 20 mm/year.
V e l o c i ~ fields In two dimensions, we assume that material does not flow out of the plane of section (i.e. Vl = u, v2 = 0, v3 = v, Xl = x, x2 = 0, and x3 = z; Fig. 3). The problem is further simplified if the horizontal velocity does not vary with depth (i.e. Ou/Oz = 0). These assumptions mean that Equation (1) simplifies to
Finite deformation in section In extinct basins and margins, strain rate data and velocity fields must be indirectly determined. At present, the most promising approach uses the history of the sedimentary pile itself to extract information about the strain rate tensor. Over the last ten years, we have developed simple 1D and 2D methods which estimate strain rate by inverting subsidence data. Three features characterise the simplest 2D inverse model (Fig. 3). First, strain rate is allowed to vary through space and time. Secondly, the evolving temperature structure of the lithosphere is solved using
~F
--+
~t
OF OF --+v--=LF
u Ox
(6)
Oz
where
/o,,
Veloci~
"-" t
0
0
~ ~v 0
~v
~
)
ll (X ]
f
Z =:a 9
Crust
node points """~i
Lithosphere 9 9 9 9 9 9
It:
X
u , 13Ax ,
Axn
9
/ /
9
p
9
~ r "
mate lithosphericbase
I
- 0
T~
~b
r 7;
Asthenosphere Fig. 3. Cartoon corresponding to yellow panel in Figure 1 and illustrating principles which underlie 2D strain rate inverse modelling. For a given strain rate distribution, G(x, t), the vertical and horizontal velocities, u(x. t) and v(x. z. t). are calculated (note that notation has changed slightly from Figure 1). v decreases from v0 at the base of the lithosphere to zero at the surface. This two-dimensional velocity field is used to solve the heatflow equation on a finite-difference grid. Boundary conditions for the linear temperature structure are T -----T1 at z = 0 (base of lithosphere) and T = 0 at z = a (top of lithosphere and reference level). Vertical, horizontal and temporal node spacing is governed by Von Neumann stability criteria (Press et al. 1992). Horizontal node spacing increases as a function of 13(x.t). The resultant temperature field varies through space and time. Other parameters listed in Table 1.
325
MODELLING BASINS IN FOUR DIMENSIONS Table 1. Definitions and values of model parameters Symbol
Parameter
Value
Units
a
Lithospheric thickness
120-125
km
tc
Pre-rift thickness of continental crust
km
G
Lithospheric strain rate
Ga-
u
Horizontal advective velocity for 2D modelling
km s-1
v
Vertical advective velocity for 2D modelling
km s-
13
Stretching factor
none
%
Lithospheric elastic thickness
km
T
Temperature
~
7"1
Temperature (real) at base of lithosphere
K
1333
~
Lithospheric thermal expansion coefficient
3.28 x 10 -5
~
Thermal diffusivity of the lithosphere
8.04 •
m 2 s -1
Pa
= pro(1 - aT0, asthenospheric density
10 - 7
3.20
g cm
-3 -3
Pc
Density of continental crustal material at STP
2.78
g cm
Pm
Density of mantle material at STP
3.35
g cm
pw
Density of seawater
1.03
g cm-3
(r
Poisson's ratio
0.25
E
Young' s modulus
g
Gravitational acceleration
Thus any vector p(0) is deformed to p(t) where
(8) By definition, the spatial and temporal variation of the stretching factor is [3(x, t) = F l l . F is initially a unit matrix and Equation (8) reduces to
013 a13 Ou a~- + u a-~ = ax [3.
(9)
Thus we are exploiting just one of the nine components of the deformation gradient tensor which is directly related to the vertical strain rate and to the strain. Equation (9) describes the two-dimensional finite deformation of the lithosphere. For any velocity field, subsidence of the Earth's surface is calculated by solving this equation in conjunction with the appropriate heatflow equation. Given a horizontal strain rate distribution, G(x,t), we must calculate the velocity field which governs lithospheric deformation. By definition ~Ju
G(x, t) --- - - . Ox
(11)
Since the amount of material which flows sideways must be balanced by an equal amount of material which flows across z --- 0, the compatability condition 3u
0v
3x
Oz
(12)
applies and so
v(x, z, t) = G(x, t)(a - z).
GPa
9.8
ms
-2
This velocity field, (u, v), prescribes the spatial and temporal variation of crustal and lithospheric stretching, 13(x, t), which is obtained by solving Equation (9). We solve for [3(x, t) on a deforming grid and details of the methodology are given in White & Bellingham (2002). Once the strain rate pattern has been defined, extension across the basin can be calculated at any time. [3(x, t) will grow through time and space, reflecting the horizontal advection of lithospheric material. The ability to calculate the structural and thermal development of a basin as a function of time and space has considerable commercial potential.
Temperature structure The second part of the algorithm solves the temperature history as a function of x, z and t. T(x, z, t) is calculated by solving the heatflow equation aT+
o-7
OT
OT
[02T
oz
l oz +o--J]
+v-------K ~
02T~
(14)
(1 O)
Therefore, the horizontal velocity is given by
u(x,t)=I~G(x,t)dx.
70.0
-3
(13)
The boundary conditions are T = 0 at z = a and T = T 1 at z = 0. Other parameters are listed in Table 1. The initial thermal structure is-given by T = Tl (a - z). This second order partial differential equation has horizontal and vertical advective terms which vary as a function of space and time. It is coupled and not amenable to analytical attack. We have solved it on a finitedifference grid using a combination of the Forward TimeCentred Space and Lax methods (Press et al. 1992). Numerical stability is ensured by choosing the time step according to the von Neumann stability criteria. Finite-difference schemes often use a grid of node points which remains undeformed throughout the calculation. Here we allow the grid to deform according to u(x, t), the horizontal advective velocity (i.e. a Lagrangian formulation). The main
N. WHITE ET AL.
326
advantage of a deforming grid is that the horizontal advective terms of Equations (9) and (14) reduce to zero when there is no horizontal motion with respect to the grid itself. A deforming grid also simplifies the subsidence calculation since an increment of subsidence at a given time step is simply added to the accumulated subsidence because material is being tracked as it moves horizontally. The vertical grid spacing is fixed throughout. Solving for T(x, z, t) is the slowest part of the algorithm. We minimise the time spent calculating temperature history by using the smallest possible number of node points consistent with finite-difference stability criteria.
Lithospheric loading The temperature structure is used to calculate the density structure of the crust and lithosphere through space and time. We assume that density varies linearly with temperature PT = P0( 1 - o~T)
(15)
where Pr is the density at temperature T and P0 is the density at 0~ This changing density structure generates a series of lithospheric loads. There are two important sources of loading, which evolve as functions of space and time. First, the base of the lithosphere rises and lithospheric mantle is replaced by asthenosphere which slowly cools to become lithospheric mantle. This load is usually negative during extension because litbospheric mantle is being replaced by less dense asthenosphere. When extension stops, the sign of this load changes. The second source of loading is generated at the Moho where crust is replaced by lithospheric mantle. This load grows during extension and is always positive. We have not included the effect of melt generated by decompression of asthenosphere. Crustal and lithospheric loads act in concert to deflect the Earth's surface. The total load, L(x, t), is given by L(x, t) = a(1 - l/f3(x, t)) - BQ(x, t)
where k is the wavenumber. If D ---* 0. Equation (18) reduces to the Airy isostatic form. The deflection of the Earth's surface is calculated as a function of time and space and is identical to the subsidence history, S(x, t) (White 1994).
Search engines There are many ways to search for optimal solutions and only a brief discussion is given in this contribution. One important issue is whether to linearize the problem (Parker 1994). Here we summarize the simple and pragmatic approach advocated by White (1994). We note that Faulkner (2000) has developed a linearized inversion scheme but the Fr~chet derivatives must be calculated numerically which considerably slows down the algorithm. G(x, t) is parameterized by using M discrete values of G in the space direction, sampled at intervals of ~x, and N discrete values of G in the time direction, sampled at intervals of ~t. As before, it is necessary to impose smoothing and positivity on G/j in order to stabilize the inversion. Thus we have chosen to minimize a trial function, H, such that
H=
~
t~{i=1 \ ~ l
]J
+,#
(20)
where ~ and ~ are the observed and calculated water-loaded subsidence, respectively.j is the number of stratigraphic horizons (varying from 1 to K), i is the number of points on each horizon (varying from 1 to L), and ~; is the palaeobathymetric uncertainty. H is minimized by varying G(x, t) which controls 5~i).~ is a set of weighting factors which ensure that the first and second derivatives of G are smooth and that G is positive. Smoothing criteria are necessarily applied to G rather than to S. Thus
,,--P,
,21 j
,logc,j,
(16)
where A = (Pro - - P~)gtc, B = Otpmg, and
Q(x, t) =
i o[T(x, z, t) -
T(x, z, ~)]dz.
i=1 j = l
(17)
I/2
0
A and B are constants, which are calculated using the parameters listed in Table 1, but Q(x,t) is a measure of the difference between the perturbed and equilibrium temperature structure and is necessarily a function of G(x, t).
Stratigraphic evolution Finally, the loading history is used to calculate the subsidence history, S(x,t). The relationship between L(x.t) and S(x.t) depends upon D, the flexural rigidity of the lithosphere. D is often expressed in terms of %, the equivalent or effective elastic thickness (Watts et al. 1982). For simplicity, we assume that %, does not vary through space and time although this assumption could be relaxed. At a time t~, the deflection of the Earth's surface, w(x, tl), is obtained by solving d4w D~--g +
(Pa
--
Pfill)gw = L(x, tl)
(18)
where ptiu is the density of the material which is deposited when the Earth's surface is deflected downwards. If SO(k) and ~r (k) are the Fourier transforms of L(x, tl) and w(x, tl), respectively, then ~/t'(k) =
~(k) (Pa - Pfut)g + Dk4
(19)
+
CS;/j
(21)
i=lj--t
where A = M • N and P1-5 are weighting coefficients. The PI term ensures that G o stays positive since this term tends to oo as G o tends to zero. The P2-5 terms cause G o- to be smooth with respect to the first and second derivatives through space and time. The results presented here were obtained using P~ - 10 -2, and P2-5 - 10-'*. H is an ad hoc function and it is important to check how inversion results vary when different values of the weighting coefficients are used. In our experience, varying Pj -5 by several orders of magnitude has negligible effects; the main purpose of the weighting coefficients is to ensure that the observed subsidence is not overfitted. A more sophisticated approach would optimize the weighting coefficients during inversion. We use Powell's algorithm to minimize H (Press et al. 1992). It is a direction-set method which performs successive line minimizations to try and locate the global minima of a misfit function. Consider point P in N-dimensional space. A vector direction u j is chosen and the function of N variables, f (P), can be minimized in that given direction using a one-dimensional search engine. Then one chooses a different direction and minimizes along it, repeating the process until the global minimum is found. Obviously, by changing directions it is possible to 'spoil' any previous minimization by searching along in a subsequent direction (i.e. w h e n f ( P ) is minimized along the second vector, u2, the function may no longer be at a minimum
MODELLING BASINS IN FOUR DIMENSIONS with respect to the first vector, Ul). Powell's method overcomes this problem by choosing a set of conjugate directions, n, which are superior to the co-ordinate vectors e~, e2 ....... This noninterfering direction set defines one-dimensional search vectors which yield large decreases in functional value.
327
feather edge. The resultant strain rate history, which was generated for re = 0 k m , is complex with two phases of extension. An earlier intense phase is confined to the centre of the basin and lasts from 140 to 120Ma. A second phase of extension, lasting from 110 to 90 Ma, has lower strain rates and is more spatially diffuse. Between 150 and 200 km, this second phase is continuous with the earlier event. We have carried out a range of tests with different compaction models which suggest that this second phase of weak extension is real. After - 80 Ma, strain rate reduces automatically to negligible values indicating that the gradient of subsidence is consistent with thermal subsidence driven by the preceding extensional event. As before, the offset between peak strain rate and peak cumulative stretching factor is a logical consequence of the horizontal advection of lithosphere away from the fixed left-hand boundary. Elsewhere, we have shown that the smallest misfit is obtained when the elastic thickness of the lithosphere, %, is less than - 3 km (Bellingham & White 2002). Independent evidence for the duration and number of rift periods can be obtained from the history of normal faulting and from the timing of volcanic activity (Faulkner 2000). The accepted view is that rifting commenced during the Tithonian (151-144 Ma) with activity reaching a peak during the Early Cretaceous (142-99 Ma). There are two main phases of normal faulting. The first phase is concentrated towards the centre of the basin where there is excellent evidence for substantial stratigraphic growth across major faults. The second phase of faulting is concentrated between 80 and 140 km range. The existence and distribution of these phases corroborate the strain rate patterns
Application of 2D scheme We have applied the 2D strain rate inversion algorithm to - 4 0 sedimentary basins and margins located worldwide. Here, we show how this algorithm can be applied to three different examples. Our purpose is to illustrate strengths and weaknesses of the existing 2D approach. Further details about each basin and its location are given by Bellingham & White (2002).
San Jorgd basin This small basin is located off the east coast of Argentina and it is our most straightforward application (Fig. 4). The basin formed by multiple extensional episodes prior to, and coeval with, the break-up of Gondwanaland (Fitzgerald et al. 1990). It is filled with predominantly shallow marine sedimentary rocks which makes it ideal for modelling since uncertainties in palaeobathymetry are negligible ( 0 - 5 0 m: Faulkner 2000). Inverse modelling shows that the observed stratigraphic record can be matched for most of the basin's history. Significant misfit occurs at the northern end of the basin where later uplift and denudation have modified the basin's
I
I
i
[
i
0
i 84 I
i '
I
0 Fig. 4. San Jorg6 basin, offshore Argentina (see Bellingham & White (2002) for further details and location). (a) Depthconverted and interpreted cross-section. Yellow zone, modelled basin; thin lines, decompacted and water-loaded stratigraphic horizons (intra-Callovian (162 Ma), Top Valanginian (132 Ma), Top Barremian (121 Ma), intra-Cenomanian (96 Ma), Top Coniacian (86 Ma), intra-Maastrichtian (69 Ma), intraPalaeocene (58 Ma), and sea bed). Dashed lines, best-fitting synthetic horizons generated by inverse modelling. Errors in palaeobathymetry were included in the inversion but they are generally small (_+50 m) and have been omitted from the profile for clarity. Thick lines = normal faults. (b) Spatial and temporal variation of strain rate for % = 0 km, which yields the synthetic horizons shown in (a). See Bellingham & White (2002) for discussion of elastic thickness. Note localized primary event at - 130 Ma and more diffuse secondary event with much lower strain rate at 100 Ma.
'
Distance
295
290 0
300
50
.vml
150 1 O0
200
300
Distance (km) 0
4
Strain Rate (Ga -1)
I
200
1O0
8
(km)
300
N. WHITE ETAL.
328
shown in Figure 4. There does not appear to be any need for strain rate to vary significantly with depth (i.e. depth-dependent stretching, lower crustal flow).
North Sea basin The evolution of the northern North Sea is well understood and this basin has been an excellent testing ground for rifting models over the last 30 years. Here, we apply the inversion algorithm to a profile which crosses the basin at 61~ (Fig. 5). The western half of this profile between 0 and 80 km shows the classic tilted block geometry of the East Shetland basin. Maximum subsidence occurs further east in the Viking Graben between 100 and 130kin. The structural development and subsidence history of the North Sea basin have been thoroughly investigated using a variety of one- and two-dimensional forward modelling techniques (e.g. Barton & Wood 1984; Marsden et al. 1990). The latest phase of rifting occurred in the Late Jurassic (155145Ma) and its spatial and temporal distribution is well understood. Interpretational difficulties and a lack of well penetration into the pre-Jurassic section mean that the preceding Triassic extensional episode is less well constrained. For simplicity, we concentrate on modelling the Late Jurassic extensional event. The cross-section is based on a seismic reflection profile which was calibrated with well-log information. Bellingham & White (2002) describe how this profile was converted from two-way travel time to depth. We present water-loaded results to allow easy comparison with published models. There are three important sources of error in calculating water-loaded
~D
subsidence. First, the compactional history of a basin is poorly known. Fortunately, synthetic testing shows that a large range of initial porosities and compaction decay lengths can be tolerated without seriously affecting calculated strain rate distributions. Secondly, decompacted sediment is converted into water, which requires assumptions about isostasic response. Like many others, we have replaced sediment loads by assuming that the isostatic response of the lithosphere can be approximated either by Airy isostasy or by flexure which assumes some elastic thickness. When a flexural response is used, it is essential that the same elastic thickness is used to unload and load the observed or predicted subsidence. Note that we do not explicitly invert for "re, the elastic thickness, although the residual misfit can be plotted as a function of "re to identify the optimal value (Bellingham & White 2002). Thirdly, the largest source of uncertainty arises from poor knowledge of palaeobathymetry. In the Cretaceous, water depths are particularly poorly known and could range from 200 to 800 m. We formally include palaeobathymetric errors during either one-dimensional or two-dimensional inversion (e.g. White 1994). For clarity, these error bars have not been plotted but the necessary details are given in Bellingham (1999). In general, we have tried to assign conservative estimates of palaeowater depth. The calculated strain rate distribution in Figure 4b has picked out the principal rifting episode in the Late Jurassic (155145 Ma). Strain rate is almost uniform over the East Shetland basin where domino-style normal faulting occurs. Peak strain rates occur in the Viking graben proper where extension continued into the Early Cretaceous. The total amount of
1
i
1
,
I
I
~
I
'
r
2 a
0
100
200
Distance, km
50
a; .~_ I00 [--, 150 100
Distance, km
0
4
Strain Rate, Ga-1
200
Fig. 5. Regional profile which crosses East Shetland Basin and Viking Graben of northern North Sea (see Bellingham & White 2002 for further details and for location). (a) Depth-converted and waterloaded profile. Yellow zone, modelled basin; thin lines, principal stratigraphic horizons (Top Callovian (159 Ma); Top Jurassic (142 Ma), Top Albian (99 Ma), Top Campanian (71 Ma), Top Cretaceous (65 Ma), Top Palaeocene (55 Ma), Top Eocene (34 Ma), seabed at present-day). Dashed lines, best-fitting synthetic horizons generated by inverse modelling. As before, palaeobathymetric errors have been omitted for clarity. Thick lines, normal faults. (b) Spatial and temporal variation of strain rate for % = 0 km, which yields the synthetic horizons shown in (a). Note Jurassic phase of extension which is more protracted within the Viking Graben (~ lOOkin).
MODELLING BASINS IN FOUR DIMENSIONS extension across the basin is - 25 km which compares well with Marsden et al.'s (1990) estimate of 22 km (see also Roberts et al. 1993). During the Cretaceous and Cenozoic, strain rates are generally negligible as might be expected. However, two minor events have been highlighted by inversion: during the Late Cretaceous/Palaeocene (80-60Ma) and during the Neogene (20-10Ma). The inversion algorithm is evidently picking up small increases in the subsidence gradient which cannot be accounted for by thermal subsidence following Late Jurassic rifting. Neither episode can easily be attributed to rifting
329
although there is some evidence for a mild extension during the Late Cretaceous: localized extension occurred around major basin-bounding faults on the western of the East Shetland basin and in the Viking Graben. In both cases, a small number of faults cut through the Cretaceous section. This faulting has not previously been attributed to rifting. There is excellent evidence for Mid-Late Cretaceous rifting on the Atlantic margin several hundred kilometres further north and associated thermal effects appear to have affected the North Sea basin to the south. The subsidence anomaly continues into the Paleocene and there is
150 Ma
160 Ma
OO
2o
4~
0o
2~
1~
1~
i0 o
i0 ~
59 ~
59 ~
130 Ma
140 Ma
OO
2o
4~
4~
0o
2~
4o
n
59 ~
59 ~
in Fig. 6. Four strain rate maps for northern North Sea calculated by inverting a set of approximately east-west profiles.
-1
3
_ll. 5
7
9
11
S t r a i n R a t e s ( G a -~)
13
15
N. WHITE ET AL.
330
excellent evidence throughout the northern North Sea for anomalous Paleocene subsidence which is usually linked to the evolution of the Iceland plume. The Neogene strain rate event occurs further east and it is much more localized. Once again, an extensional origin cannot be justified. It may also be associated with the Iceland plume. Thus inverse modelling can be used to extract the detailed pattern of Late Jurassic rifting and to identify the temporal and spatial distribution of anomalous events. Differences in basin deflection for Airy isostasy and for elastic thicknesses of 1 or 2 km are small. However, if "re is greater than 5 km, the residual misfit is significant and calculated strain rate distributions are geologically less plausible. For a given value of're, strain rate is varied during inversion to achieve the smallest misfit. There is clearly some trade-off between % and strain rate but it is relatively weak. We infer that the elastic thickness during syn-rift and post-rift phases is less than about 2 - 4 k m although variation of residual misfit with elastic thickness is so small that we cannot distinguish between 0 and 2 kin. The relationship between free-air gravity anomalies and load topography in the frequency domain also shows that % < 5 km (Barton & Wood 1984). We have inverted a set of east-west regional profiles which cross the northern North Sea and used the results to construct maps of the spatial distribution of strain rate (Fig. 6). These
400 .
,
I
350 ,
,
=
,
|
Section Length (km) 250 200 150
300
j
,
,
,
I
,
,
,
.
I
,
.
,
I
I
i
i
i
I
I
. . . .
strain rate maps are only valid if extension occurred parallel to the profiles (i.e. east-west). During the early stages of rifting, deformation is distributed across all of the basin (Fig. 6a). Over a 10 Ma period, strain rate decreases rapidly, especially over the East Shetland basin. The Viking Graben itself remains active for the longest period. Strain rate maps can be easily converted into maps of heatflux by converting vertical strain rates into temperature gradients.
Norwegian margin Our final example is from the Voring Basin on the Norwegian margin. The only purpose of this example is to show how the strain rate tensor can be used to construct an integrated basin animation (Fig. 7). In contrast to the other examples, there are considerable uncertainties in the stratigraphic interpretation and we do discuss these uncertainties here. In such cases, the purpose of inverse modelling is to test the validity of differing interpretations--the best ones are likely to be be those which result in the smallest residual misfit of the stratigraphical data. Once strain rate has been determined as a function of time and space, the thermal and structural evolution of a basin or margin can be calculated for any time. Successive snapshots can
100
50
I
1
. . . .
0 . . . .
I
o 5 lO 15 20 o 5 lO 15 20 ._.. o E v-
5 >, r,r~
10
20 N 0 5 10 15
Temp. ~ 400 350 300 250 200 150 100 50
Base Cretaceous
20 0 5 10 15
Base Jurassic
20 0
Base Triassic 9
"
I
v'
400
,
,
,
I
. . . .
350
I
. . . .
300
I
. . . .
250
I
I
. . . .
200
I
. . . .
150
Section Length (km)
I
. . . .
100
I
50
. . . .
!
0
Fig. 7. Set of five snapshots which illustrate structural and thermal evolution of Voring basin offshore Norway. At each time interval, an image is constructed using integrated strain rate history. Black lines, water-loaded subsidence layers; dotted lines and colour scheme, temperature profiles calculated from basal heat flux which in turn was calculated from strain rate. For simplicity, we assumed that thermal conductivity of sedimentary pile is constant.
MODELLING BASINS IN FOUR DIMENSIONS be combined to construct an animated representation of the basin's evolution. In Figure 7, the five snapshots clearly show how crust is advected sideways during rifting and how the smoothed structure of the basin evolves. The strain rate pattern can also be used to calculate the temporal and spatial variation of heatflow into the bottom of the sedimentary pile. Basal heatflow constrains the temperature and maturation history of the basin. Note how the evolving temperature history has been superimposed on the basin structure.
Conclusions We have outlined the main features of a general 3D inverse algorithm for modelling extensional sedimentary basins and passive margins. This model is a generalization of 1D and 2D models developed by White (1994) and Bellingham & White (2000). At present, only 2D planform and section implementations exist. In the second half of this contribution, we show how the 2D strain rate inversion of Bellingham & White f2000) can be used to model stratigraphic cross-sections. This approach can be used to test, and therefore risk, different interpretations. The resultant strain rate patterns can also be used to calculate the structural and thermal evolution with time. We have not discussed model resolution but emphasize that this issue forms a central part of the application of inverse theory. We are conscious that a variety of different forward models could have been chosen as the basis for an inversion algorithm. Our general approach concentrates on extracting information from the subsidence record since it is probably the best constrained observation in many extensional sedimentary basins. The essential features of the original stretching model have been incorporated but we ignore the detailed, short wavelength deformation of the brittle upper crust. Rougher models, which are more structurally complex and which include the effects of normal faulting, can be developed by permitting strain rate to vary as a function of depth. A complete dynamical description of lithospheric extension relies on assumptions about driving forces and about lithospheric rheology, which together determine the spatial and temporal patterns of strain and strain rate. We hope to refine the dynamical constraints by measuring these patterns in a large number of basins.
We are very grateful to the organisers of the 4D seismic imaging conference for the opportunity to present our ideas. Figures were prepared using Generic Mapping Tools (Wessel & Smith 1995). N. Kusznir and K. Gallagher provided most constructive reviews. Department of Earth Sciences Contribution Number 7476.
References BARTON, P. & WOOD, R. 1984. Tectonic evolution of the North Sea basin: crustal stretching and subsidence. Geophysical Journal of the Royal Astronomical Societx, 79, 987-1022. BEAVAN, J. & HAINES, J, 2001, Contemporary horizontal velocity and strain rate fields of the Pacific-Australian plate boundary zone through New Zealand. J. Geophys. Res., 106, 741-770. BELLINGHAM, P. 1999. Extension and subsidence in one and two dimensions, north of 60 ~ N. PhD Dissertation, University of Cambridge. BELLINGHAM, P. & WHITE, N. 2000. A general inverse method for modelling extensional sedimentary basins. Basin Research, 12, 219-226. BELLINGHAM,P. & WHITE, N. 2002. A two-dimensional inverse model for extensional sedimentary basins: 2. Application. J. Geophys. Res., 107, ETG 19, DOI 10.1029/2001JB000174.
331
FAULKNER, P. 2000. Basin formation in the South Atlantic Ocean. PhD Dissertation, University of Cambridge. FITZGERALD, M. G., MITCHUM,R. M., ULIANA,M. A. & BIDDLE, K. T. 1990. Evolution of the San Jorge basin, Argentina. AAPG Bulletin, 74. 879-920. FLESCH, L. M., HOLT, W. E., HAINES, A. J. & SHEN-TU, B. 2000. Dynamics of the Pacific-North American plate boundary in the western United States. Science, 287, 834-836. GEMMER, L. & NIELSEN, S. B. 2000. SVD analysis of a 3D inverse thermal model. In: HANSEN, P. C., JACOBSEN, B. H. & MOSEGAARD, K. (eds) Methods and Application of hn'ersion. Lecture Notes in Earth Sciences, Springer, Berlin, 92, 142-154. GEMMER, L. & NIELSEN, S. B. 2001. Three-dimensional inverse modelling of the thermal structure and implications for lithospheric strength in Denmark and adjacent areas of Northwest Europe. Geophysical Journal International, 147, 141 - 154. HAINES. A. J. 1982. Calculating velocity fields across plate boundaries from observed shear rates. Geophys. J. R. Astron. Soc., 68,203-209. HAINES. A. J. & HOLT, W. E. 1993. A procedure for obtaining the complete horizontal motions within zones of distributed deformation from the inversion of strain rate data. Z Geophys. Res., 98, 12 057- ! 2 082. HOLT, W. E., CHAMOT-ROOKE,N., LE PICHON, X., HAINES,A. J., SHENTU, B. & REN, J. 2000. The velocity field in Asia inferred from Quaternary fault slip rates and GPS observations. J. Geophys. Res., 105, 19 185-19 210. JACKSON,J. A., HAINES,A. J, & HOLT,W. E. 1992. The horizontal velocity field in the deforming Aegean Sea region determined from the moment tensors of earthquakes. J. Geophys, Res., 97, 17 657-17 684. KREEMER, C.. HAINES,J., HOLT, W. E., BLEWITT, G. & LAVALLEE,D. 2000. On the determination of a global strain rate model. Earth Planets Space, 52, 765-770. MCKENZIE, D, P. 1978. Some remarks on the development of sedimentary basins. Earth and Planetary Science Letters, 40, 25-32. MCKENZIE, D. 1981. The variation of temperature with time and hydrocarbon maturation in sedimentary basins formed by extension. Earth Planet. Sci. Letts., 55, 87-98. MALVERN, L. E. 1969. hltroduction to the Mechanics of a Continuous Medium. Prentice-Hall, Old Tappan, N.J. MARSDEN, G., YIELDING.G.. ROBERTS, A. M. & KUSZNIR,N. J. 1990. Application of a flexurai cantilever simple-shear/pure-shear model of continental lithosphere extension to the formation of the northern North Sea basin, bT: BLUNDELL,D. J. & GIBBS, A. D. (eds) Tectonic Evolution of the North Sea Rifts. Clarendon, Oxford, 240-261. NEWMAN, R. & WHITE, N. J. 1999. The dynamics of extensional sedimentary basins: constraints from subsidence inversion. Philosophical Transactions of the Royal Society, London, 357, 805-830. PARKER, R. L. 1994. Geophysical Inverse Theory. Princeton University Press, Princeton. PRESS, W. H., TEUKOLSKY,S. A., VETTERLING, W. T. & FLANNERY, B. P. 1992. Numerical Recipes in Fortran 77: The A rt of Scientific Computing. 2nd Edition. Cambridge University Press. ROBERTS, A. M., YIELDING, G., KUSZNIR, N. J.. WALKER, I. ~ DORNLOPEZ, D. 1993. Mesozoic extension in the North Sea: constraints from flexural backstripping, forward modelling and fault populations. In: PARKER. J. R. (ed.) Petroleum Geology of Northwest Europe. Proceedings of the 4th Conference. Geological Society, London, i 123-1136. WATTS, A. B., KARNER, G. D. & STECKLER, M, S. 1982. Lithospheric flexure and the evolution of sedimentary basins. Philosophical Transactions of the Royal SocieO', I,zmdon, 305, 249-281. WESSEL, P. & SMITH, W. H. F, 1995. New version of the Generic Mapping Tools released. EOS Transactions of the American Geophysical Union, 76, 329. WHITE, N. J. 1994. An inverse method for determining lithospheric strain rate variation on geological timescales. Earth and Planetary Science Letters, 122, 351-371. WHITE, N. & BELLINGHAM,P. 2002. A two-dimensional inverse model for extensional sedimentary basins: 1. Theory. Journal of Geophysical Research, 107, ETG 18, DOI 10.1029/2001JB000173.
Examples of multi-attribute, neural network-based seismic object detection P. DE GROOT
j, H . L I G T E N B E R G
j, T . O L D E N Z I E L
2, D . C O N N O L L Y
3 & P. MELDAHL
4
tdGB Earth Sciences BV, 7511 AE Enschede, The Netherlands (e-mail:
[email protected]) 2dGB Rotterdam BV, Taborstraat 12, The Netherlands 3dGB-USA LLC, One Sugar Creek Center Boulevard, Suite 935, Sugar Land, TX 77478, USA 4Statoil AS, N-4001 Stavanger, Norway
Abstract: Certain seismic objects, like faults and gas chimneys, are often difficult to delineate using conventional attribute analysis. Many attributes contain useful information about the target object but each new attribute provides a new and different view of the data. The challenge is to find the optimal attribute for a specific interpretation. In this paper the optimal attribute is found with a pattern recognition approach based on multi-dimensional/multi-attributes and neural network modelling. Multi-dimensional attributes, as opposed to point attributes, can provide the spatial information on the seismic objects. The role of the neural network is to classify the input attributes into two or more output classes. Neural networks are trained on seismic attributes extracted at representative example locations that are manually picked by a seismic interpreter. This approach is a form of supervised learning in which the network learns to recognize certain seismic responses associated with the identified target objects. Application of the trained network yields an "object probability' cube for the target object. Essentially, the neural network can target any seismic or geological feature requiring detailed analysis. In this paper the method is described and examples are shown of gas chimneys, faults, salt domes and 4D anomalies. Some interpretation aspects are discussed.
Seismic objects such as gas chimneys, salt bodies and stratigraphic features are defined here as spatial elements with an observable size and orientation and with a different seismic response with respect to their surroundings. Although they are often straightforward to recognize, their spatial boundaries and distribution are often difficult to map. Objects can be solid in which case the internal texture differs, or they are twodimensional features characterized by a break in the response. Many workers use attributes to better visualize and interpret objects. Often the interpreter extracts multiple 'point' attributes, which immediately causes two interpretation problems: 9 9
the object is not uniquely defined by any of the extracted attributes and attributes on their own may not discriminate between objects of different geological origin.
The method described, based on Statoil's seismic object detection technology (Meldahl et al. 1999) addresses both problems by calculating the multi-dimensional attributes in subcubes that contain spatial information and by re-combining extracted attributes into one or more new attributes using neural network technology. The new attributes correspond to the output nodes of the neural network and can represent different meanings depending on what the neural network has learned to recognize. Two learning approaches are used: supervised and unsupervised (e.g. de Groot 1999). This paper describes a supervised methodology, where a neural network is trained on data points selected by the user to classify the response into two or more classes. In the simplest case the network has two output nodes. It learns to classify the seismic response into object or non-object, represented by vectors (1,0) and (0,1 ) respectively. The two output nodes mirror each other and it is thus sufficient to output the 'object' node only when we apply the trained network to generate an 'object probability' volume. Values close to 1 in this volume indicate a high 'probability' of finding the object at these positions. Figure 1 shows a seismic line from the Gulf of Mexico. A seismic 'cloud' of incoherent noise, which may be related to hydrocarbons migrating upwards, is located above a salt dome. Next to the seismic line four different single attribute displays
are shown (energy, similarity, dip variance and polar dip). The two right-most displays show the results of supervised neural network classifications. The networks were targeted at recognizing salt and chimneys respectively. It can be observed that several single attributes pick up the anomalous responses associated with the two geological features of interest but none shows a clear image of either object. The outline of these features is much better defined in the output from the neural network and it is clear that the networks were able to discriminate between two objects of different geological origin. The latter is achieved by choosing suitable input attributes per object and by careful picking of example locations.
Attribute sets, neural networks and 'dip-steering' Attribute sets are assemblies of single-trace and multi-trace (i.e. volume) attributes calculated from one or more seismic input cubes. Attributes in a particular set are chosen to be sensitive to a particular object, e.g. they pick up faults. Some attributes are more sensitive than others are but none is expected to be perfect. To get an optimum fault image we have to use the information from all attributes simultaneously. This is where neural network modelling comes in. The supervised neural network is trained on attributes extracted at example locations picked by the interpreter. In the example case the network learns to classify the input attributes into two classes: faults or non-faults. Neural networks belong to a group of computing techniques that are inspired by the so-called 'brain metaphor', which means that these are algorithms that aim to mimic the human brain (e.g. de Groot 1999). Many different types of neural networks exist. The type used in this paper for the supervised learning of object classes is the popular Multi-Layer-Perceptron (MLP) network (Fig. 2). It consists of a large number of connected processing nodes that are organized in layers. The information in an MLP network is passed from left to right: from input layer via hidden layer to output layer. Each node is connected to all nodes in the next layer (often referred to as a 'fully connected MLP') and each connection has a weight assigned to it. Training starts with a random set of connection weights. The learning algorithm updates the weights during the training phase such that the error between neural network predicted output and
DAVIES, R. J., CARTWRIGHT,J. A., STEWART, S. A., LAPPIN,M. & UNDERHILL,J. R. (eds) 2004.3D Seismic Technology:Applicationto the E.~ploration of Sedimentary Basins. Geological Society, London, Memoirs, 29+ 333-337. 0435-4052/04/$15 9 The Geological Society of London 2004.
334
DE GROOT ET AL. ............
d
. ""
~!u
"~ . . . .
,
~
",
Fig. 1. Comparison between single attribute displays and multi-attribute neural network detections of salt and chimneys, respectively. Each section is about 10km wide and 2.5 s deep.
Dip information opens a whole category of powerful dip-steered attributes and filters that are calculated in data-driven shapes such as 'warped' disks, cubes or slices. The concept of attribute sets makes it possible to create defaults for different objects. Non-experts can detect objects on other seismic surveys using such default sets. In practice for each seismic survey new example locations are picked and the neural network is re-trained to calibrate the object detection method.
Fig. 2. Fully connected Multi-Layer-Perceptron (MLP) neural network with ten input nodes, five nodes in the hidden layer and two output nodes.
Examples
(known) actual output is minimized. This type of mapping between input and desired output is a form of multiple, nonlinear regression that can be used to find complex relationships. Attribute selection for a particular attribute set is based on experience, visual inspection and using statistical support tools. Analysis of the neural network weighting function is a simple and effective way to determine the discriminative power of individual attributes. The higher the weights of a node in the input layer, the more important the associated input attribute is for solving the problem. By colour coding the nodes according to the normalized sum of their weights, the relative importance of each attribute can be assessed visually. In Figure 2 attributes with red nodes are more important than attributes with yellow nodes, which in turn are more important than attributes with white nodes. The detection power of attributes and attribute sets is greatly improved if the calculations are 'dip-steered', i.e. local dip information is utilized. For example the similarity attribute, which calculates the normalized Euclidean distance between two or more trace segments, is much better defined if the trace segments belong to the same seismic event (Fig. 3). This requires knowledge of the local dip and azimuth, which can be calculated a/o with a sliding 3D kf-transform (Tingdahl 2003).
Chimneys T h e C h i m n e v C u b e is a new concept that uses a 3D volume of stacked seismic data with other prior information such as the interpreter's insight and other geological data, to highlight vertical chaotic seismic character that are often associated with gas chimneys. Through this process, a seismic volume (and corresponding attributes) is provided as input to a neural network and a chimney cube is generated as its output. High values in this cube indicate a high 'probability' of belonging to a chimney. Initially chimney cubes were used in geo-hazard interpretation, e.g. to avoid drilling shallow gas pockets and to identify regions of sea floor instability. In recent years chimney interpretation has also proven to be very useful for exploration of hydrocarbon targets both in ranking prospects and to improve our understanding of the petroleum system. Chimney cubes can reveal where hydrocarbons originated, how they migrated into a prospect, and how they spilled or leaked from this prospect and created shallow gas anomalies, mud volcanoes or pockmarks at the sea bottom (e.g. Heggland et al. 1999; Aminzadeh et al. 2001). Current applications of T h e C h i m n e v C u b e include unravelling a basin's migration
im
i No steenng
(a)
Central Steering
(b)
. _~, .....~--4D.i Full Steering
(c)
Fig. 3. Dip steering is the process of following the local dip and azimuth to find the input segments for a filtering or attribute calculation process. On the left no steering is applied: the trace segments are aligned horizontally. In central steering (middle figure) the local dip and azimuth of the evaluation point is used to find the requested trace segments. On the right full steering is used: the local dip and azimuth is updated at every trace to follow the seismic event. The figures show the principle in 2D but the actual process of dip steering is 3D.
NEURAL NETWORK-BASED SEISMlC OBJECT DETECTION
335
Fig. 4. Stratigraphic pinchout offshore Nigeria. The horizontal tick marks are 500 m apart, the vertical scale is in ms. The orange-yellow sand-body pinches out against a shale diapir. The onset of a seismic chimney coincides with the interpreted sand '0' line. Apparently the stratigraphic trap is leaking hydrocarbons. With the aid of geochemistry and analogues it is feasible to predict the type of fluids that have leaked and that may still be trapped.
history, distinguishing between charged and non-charged prospects or sealing versus non-sealing faults, determining vertical migration of gas, identifying potential for over-pressure, and detecting shallow gas and geo-hazards. Other potential applications of the chimney cube data are predicting hydrocarbon phase and charge efficiency, which are commercially interesting objectives especially in multiphase petroleum systems. The following example is from offshore Nigeria. Figure 4 shows a sand body pinching out against a shale diapir. The mapped '0' sand line coincides with the onset of a seismic chimney (shown in yellow on one cross-line only). Apparently hydrocarbons are leaking from the stratigraphic trap at the highest position, which is also the position of highest strain. It has been observed frequently that gas chimneys are located in
Fig. 5. Seismic chimneys are associated with a shale diapir in a data set offshore Nigeria. Chimneys are often located in areas of high strain. These types of chimneys are believed to be more often associated with oil rather than gas seeps. The horizontal tick marks are 500 m apart, the vertical scale is in ms.
areas of high strain. Thus many strong chimneys are located over shale diapirs. For source rocks to be efficient in charging a reservoir, they not only need to be organically enriched and thermally mature, but they also need to have a mechanism for being expelled from the source rock. This is crudely measured in basin models as the hydrocarbon expulsion efficiency. Areas of high strain act as vertical pressure valves to release hydrocarbon saturated fluids from the source rock into shallow reservoir intervals. Areas of intense vertical migration, detected in our method as chimneys, may be more oil-prone than areas of less intense chimney development. Further data is needed to support this hypothesis. However, many oil fields have been observed to be in close proximity to shale diapirs. Figure 5 shows the same shale diapir as in Figure 4. The stratigraphic trap is to the left of the chimney. TheChinmevCube data (yellow) illuminates the
DE GROOT ETAL.
336
(a)
(b)
Fig. 6. Comparison between a time-slice (8 • 7 kin) through a chimney cube (a) and through a fault cube (b). Faults that exhibit a characteristic pockmark pattern on the chimney cube slice are thought to be leaking. Faults that do not show up in the chimney cube but are visible in the fault cube are interpreted as sealing faults. sealing faults. Chimney and fault cube data then need to be integrated with other regional and prospect specific information. Chimneys and seepage related features might be interpreted in different ways depending on geological setting and geographic location. For example, in some areas of the North Sea a high correlation between chimneys and known hydrocarbon discoveries has been observed. Dry wells in these areas coincide with areas without chimney activity. Chimneys are thus interpreted as positive features that may upgrade a prospect. In contrast chimneys in the East Timor Sea are often interpreted as features indicating seal breach, hence downgrading prospects.
extensive expulsion of hydrocarbons related to the diapiric shale and its subsequent seafloor expression.
Fault sealing Hydrocarbon seepage is often associated with features such as carbonate mounds, mud volcanoes, seabed depressions and pockmarks. The latter are small circular features that are often aligned along fault planes, which can be seen on sea floor maps from around the world (e.g. Heggland 2003). Similar circular features can often be observed along fault planes on time-slices through the chimney cube. Figure 6 shows a data set from Nigeria. On the left a time-slice through the chimney data is shown. Apart from the larger circular features that correspond to major seismic chimneys we also observe smaller circular features that are organized along fault trends. These are interpreted as leaking faults. The amount of circular features in the chimney cube is a qualitative measure for the amount of leakage. A comparison with a time-slice through the fault cube on the right confirms that the circular features are aligned along the faults. However, some faults in the fault cube data do not show up in the chimney cube. These faults are interpreted as
Salt Salt bodies often exhibit a very characteristic response of low reflectivity, low energy and a high degree of chaos, Nevertheless it is in general quite difficult to map the exact outline of a salt structure. For an optimal detection of a salt body we can again make use of a supervised neural network approach. The attribute set comprises a/o various curvature attributes, dip-steered similarities, energy, and the variance of the dip. Figure 7 shows an example of salt detection from the North Sea.
-
_
(a)
(b)
Fig. 7. Section (21 km across) of a salt dome detection using multi-attributes and a supervised neural network.
~
- -
.~i.-
f - -
NEURAL NETWORK-BASED SEISMIC OBJECT DETECTION
337
learn that there may exist subtle differences in the attribute sets of repeatable noise and true 4D objects. Due to the enhanced visualization, the interpretation of 4D anomalies is facilitated (Fig. 8). Furthermore, being able to study the time-lapse anomalies in three dimensions allows for better integration within reservoir engineering, thus increased benefit of time-lapse seismic. The procedure is simple and fast. The interpreter does not have to be an expert on all available seismic attributes or advanced filters to be able to visualize 4D objects in a sophisticated manner. The picking of 4D anomaly example locations is however, a crucial step where the user can steer the process and influence the result. The technology is not limited to two time-lapse data sets: any number of seismic input cubes such as pre-stack and inverted volumes can be used simultaneously to improve the analysis.
Fig. 8. Neural network predicted 4D anomalies and mapped intrareservoir horizon (5 x 4.4 kin). 4D anomalies
Compared to conventional single attribute analysis, time-lapse visual inspection can be improved considerably by analysing multiple attributes simultaneously and by visual comparison of the resulting 4D anomalies in three dimensions. Depending on the reservoir, several attributes may exhibit time-lapse behaviour. Of these, each attribute may yield different time-lapse responses. Studying attributes in isolation is not only timeconsuming but may also lead to confusing results. For an interpreter it is impossible to study and compare several cubes quickly and in great detail. A simple, yet sophisticated, time-lapse object detection procedure is presented and illustrated on a North Sea field (Meldahl et al. 2002). The procedure comprises two parts: (1) an analysis phase in which representative examples of 4D anomalies have to be found, and (2) a phase to train a neural network on these examples. In the analysis phase, both single and multi-attribute analyses are used to explore the time-lapse data set and find examples of 4D anomalies. Reservoir and well data is analysed simultaneously to ensure the 4D anomalies are related to production changes rather than to acquisition and processing artefacts. In the second phase, the example locations of the analysis phase are used to train a supervised neural network in distinguishing between 4D anomalies and background. A variety of different attributes can be used as input to the neural network. The trained network is applied to the entire data set yielding a 4D-anomaly cube. When analysing time-lapse seismic, (non-)repeatable noise must be reduced as much as possible, because the signals of interest are usually weak and may be completely obscured by the noise. Our method reduces the non-repeatable noise considerably by tackling it in different ways. Firstly we apply robust statistical filters to all attributes that we extract. Nonrepeatable noise is further reduced when we apply neural networks to detect 4D objects. A well-known feature of supervised neural networks is their capability to 'see' through noise to capture the general trend in the data. The supervised approach also has the potential to reduce remnant repeatable noise through careful selection of example locations. The user selects example locations in areas with large 4D differences that are attributed to repeatable noise. Classifying these example locations as non-4D anomalies gives the network a chance to
Conclusions Seismic objects such as faults, gas chimneys, salt domes and 4D anomalies can be delineated in greater detail using a pattern recognition approach, which is based on multiple attributes and neural networks. The examples shown in this paper are based on a supervised learning approach in which a seismic interpreter picks example locations of object and background. At these locations single- and multi-trace attributes are extracted for training the neural network. Application of the trained neural network to a 3D volume results in an "object probability' volume for the target object. This method can in principle be used to enhance the visibility of any geological/seismic feature that is worth studying in detail. Statoil and sponsors of the d-Tect seismic object detection project are thanked for financial and intellectual contributions and for permission to publish the examples shown in this paper.
References AMINZADEH.F., DE GROOT. P., BERGE, T. & VALENTI,G. 2001. Using gas chinmeys as an exploration tool (part 1 & 2). World Oil Magazine, May 2001, 50-56 (part 1) and June 2001, 69-72. DE GROOT. P. f. M. 1999. Seismic reservoir characterisation using artificial neural networks. 19th Mintrop Seminar. MOnster, 16-18 May, 1999. HEGGLAND, R. 2003. Vertical hydrocarbon migration at the Nigerian continental slope: application of seismic mapping technologies. AAPG Conference, Salt l_xlke Cio', I 1-14 May 2003. HEGGLAND, R., MELDAHL. P., BRIL, B. & DE GROOT, P. 1999. The chimney cube. an example of semi-automated detection of seismic objects by directive attributes and neural networks: Part II; interpretation. 69th SEG conference, Houston, 1999. HEGGLAND, R., MELDAHL, P., DE GROOT, P. & AMINZADEH,F. 2000. Chimney Cube unravels subsurface. The American Oil & Gas Reporter, Feb. 2000. MELDAHL, P., HEGGLAND, R., BRIL, B. & DE GROOT, P. 1999. The chimney cube, an example of semi-automated detection of seismic objects by directive attributes and neural networks: Part 1; methodology. 69th SEG conference, Houston, 1999. MELDAHL, P., NAJIAR, M., OLDENZIEL,T. • LIGTENBERG, H. 2002. Semi-automated detection of 4D objects. 64th EAGE conference, Florence, 2002. TINGDAHL, K. M. 2003. Improving seismic chimney detection using directional attributes, hi: NIKRAVESH, M., AMINZADEH, F. ZADEH. L. A. (eds) Soft Computing and Intelligent Data Analysis in Oil Exploration, Developments in Petroleum Science, 51, 157-173.
Modelling fault geometry and displacement for very large networks DUSTIN
L. LISTER
Department of Earth Sciences and Engineering, hnperial College, RSM Building, Prince Consort Road, South Kensington, London, SW7 2BP (e-mail:
[email protected]) Present address: Schlumberger House, Buckingham Gate, Gatwick Airport, West Sussex, RH6 0NZ, UK
Abstract: Traditional methods for building fault models are time-consuming when applied to a complex fault network or
where many faults exist since the workflows typically rely on manual intervention at several stages. Structural detail is often simplified to reduce cycle times and consequently, the workflow favours large-scale and simplistic fault systems. There is generally no integrated assessment of kinematic information that would be useful in guiding fault interpretation. A new methodology for constructing a complex fault network with small offset is presented. The method recognizes that interpretation of large numbers of interconnected low displacement faults, is most efficiently done using map based interpretations. A novel semi-automated skeletonization algorithm is used to extract fault traces from horizon maps providing a polyline data set for subsequent use in 3D surface creation. Displacement information is derived automatically during or after the skeletonization providing kinematic information for guiding further interpretation. The new method is validated against manual interpretations of fault geometry and displacement before application to a region of the Central North Sea exposing polygonal faults. The new technique allows for the first time, a rapid and accurate appraisal of complex near-seismic scale fault geometry and displacement from interpretations of 3D seismic data across a large survey area.
Recent improvements in the vertical and lateral resolution of seismic-reflection datasets has allowed the mapping of geological strata and faults with throw of the order of 10 m over areas of tens of square kilometres. Observations made from detailed interpretations of this data have revealed that sub-surface fault geometries are often complex and form highly inter-connected planar and non-planar geometries. Polygonal style faults were recognized from horizon maps that describe trace geometries from minor extensional faults arranged in polygonal cells with an approximately equal distribution in fault strike (Cartwright 1994). This fault style has been documented in many basins worldwide and is ubiquitous throughout the Cenozoic succession in the Central North Sea being closely related to sedimentary grain size and clay mineralogy (Dewhurst et al. 1999) such that a method of formation based on volumetric contraction during compactional dewatering has been proposed (Cartwright & Lonergan 1996; Dewhurst et al. 1999). To date. the trace geometry of this complex fault style has been well described from two-dimensional maps (Lonergan et al. 1998: Watterson et al. 2000) and Lonergan et al. (1998) have made limited three-dimensional descriptions of key fault geometries. No attempt has been made to describe the linked temporal and spatial inter-relationships exhibited by these fault systems in part because of complexities involved with interpretation and because of time constraints imposed by using traditional interpretation tools. This class of faulting has been used here, as a case study for investigating an alternative approach over traditional methods for building complex fault networks that exhibit low displacement. The existing mapping techniques implemented by current interpretation software (SeisWorks & Geographix--Landmark, Charisma, IESX & Petrel--SiS, The Kingdom Suite--SMT amongst others) used to map sub-surface faults and horizons from seismic datasets are based on line interpretations digitized from vertical sections and time-slices through the 3D seismic volume. The procedures can be very time-consuming especially where a high number of interconnected fault surfaces exist and the 3D surface geometries produced are subject to limitations imposed by the interpretation methodology and imposed by the algorithms used to create surfaces through the digitized polyline data. The 2D visual representation of seismic data used during the mapping procedures restricts the amount of information available to the interpreter that can lead to errors
during the mapping process, resulting in models that lack threedimensional geometrical consistency. The resolution of 3D models is also not detailed enough to elucidate intersection points, branch lines or minor lateral offsets that may occur across small intersecting faults such as would be available from field mapping, and therefore detailed information about the temporal distribution of a fault network is not available. Fault displacement is often analysed as a proxy for fault growth so that a temporal and spatial evolution for the fault network can be derived but there are few existing approaches (one software example FAPS--Badley Earth Sciences) to allow such information to be derived efficiently from the interpreted data and integrated into the modelling workflow to help guide future interpretation. Manual sampling of displacement information might be attempted for isolated faults with simplistic geometry but would be inconceivable on the scale of typical seismic surveys especially where the faults exhibit complex interactions. Manual displacement mapping is also subject to the same limitations and human error associated with fault structure mapping. A detailed understanding of the geometry and kinematics of complex 3D fault systems at fine scales has rarely been addressed using realistic models captured from sub-surface data. Fault and displacement interpretations (Needham et al. 1996) have lacked the geometrical detail and displacement sampling resolution necessary to understand how faults grow and interact with each other at the scale of polygonal faults where fault lengths and horizon offsets are small. Manual sampling of displacement information (Mansfield & Cartwright 1996) has been attempted but only for a limited number of well-constrained faults that suffer from a lack of geometrical accuracy at their intersection imposed by the fault interpretation strategies. Hypothetical simulations of fault geometry and modelled fault displacement have been used on single or simplistic networks (Maerten 1999) but the results and conclusions that can be drawn from such models may only be applicable to the respective model in isolation and not generalized for other datasets. Detailed models of geometry and displacement for real data would greatly assist in our ability to identify controls on the development of complex inter-related fault systems we know to exist at fine scales and improve our understanding of factors affecting for example, fluid migration in the reservoir.
DAVIES,R. J., CARTWRIGHT,J. A., STEWART,S. A,, LAPPIN,M. & UNDERHILL,J. R. (eds) 2004.3D Seismic Technology:Application to the Exploration of Sedimentar3.,Basins. Geological Society, London, Memoirs, 29, 339-348. 0435-4052/04/S!5 ~ The Geological Society of London 2004.
340
D.L. LISTER
3D visualization and modelling The rapid and accurate generation and visualization of 3D fault and horizon models is of paramount importance in the search for oil and gas reserves (Needham et al. 1996), locating sub-surface ore bodies (Stuart et al. 2000), structural analysis of local anomalies such as salt diapirs (Guglielmo et al. 1999) and providing data for structural restoration (Egan et al. 1999), analyses of stress (Maerten 1999), and fault growth models. The model building procedures available through current computational software allow increasingly complex structural models to be created using a variety of algorithmic constructions and manipulations. However, there will always be limitations in generic modelling algorithms due to the huge variety and complexity of structural features, and often many different procedures are iteratively applied before an end model can satisfy geometrical and geological constraints. The tools required to construct an accurate model are largely dictated by the geometry of the model data and may require completely new algorithmic techniques to be designed and employed. 3D visualization affords a better understanding of the geometrical relationships within the model and when combined with flexible editing tools provides a powerful means to impose geological and structural principles and increase the coherency and validity of the model. Automatic estimation of displacement information during the modelling procedures would provide valuable kinematic data that can be used interactively by the interpreter during the fault modelling process and improve our understanding of the kinematics of complex fault networks.
Scope This work is focused at (but not limited to) modelling large numbers of small-scale extensional faults in 3D that exhibit near-isotropic distributions of fault strike in plan view. Such faults are almost impossible to model accurately using traditional modelling approaches. Included in these new techniques is the ability to model detailed displacement information across large numbers of interacting faults. The techniques allow for the first time, detailed analysis and quantitative appraisal of the extent of fault segment interaction. The new techniques are not restricted to this complex style of faulting and may be of significant benefit in modelling other faulting styles. A full description of the techniques described in this paper can be found in Lister (2001).
Limitations in traditional fault modelling Traditional approaches to building structural models from seismic datasets have concentrated around a section-based interpretation. The 3D seismic dataset is sliced into vertical sections and horizontal time-slices, usually parallel to lines and traces in the seismic survey shot-point grid, where the interpreter is able to digitize picks on the section that represent either the surface of a reflection or the mid-point of dislocation in a reflection that corresponds to the surface of a fault. Horizon mapping creates a regular grid of data-points that represent the elevation of a bedding surface. Faults are mapped by digitizing points on vertical sections to create fault sticks or polylines that represent the intersection of the fault surface with the vertical seismic section. Automated techniques exist to assist with horizon interpretation by auto-tracking along reflectors using the seismic amplitude or other parameter as a guide, whilst more advanced neural network approaches aim to learn about common patterns in the seismic which can be used to match reflections over the whole dataset. Thus interpretation for
horizons can often be performed relatively quickly and with little user intervention. A robust automated procedure for mapping faults is much more difficult to implement with no tangible event to correlate so accurate fault models are currently very time consuming to create. Where there are many intersecting faults with complex surface geometry such as with polygonal fault systems, the traditional interpretation approach will be prohibitive and subject to many potential problems that compromise the accuracy and reproducibility of the fault model. Such problems include: c o r r e l a t i o n - - w h e r e the interpreter inadvertently changes the criteria used to digitize the fault between sections (largely due to practical limitations in the number of sections that can be viewed and the lack of kinematic information to act as a guide), where dislocations are simply not imaged or at some oblique angle to the section, or where the grouping of fault sticks in plan view is not correct (the correlation is decoupled from interpretation in the workflow)-aliasing is likely to occur with separate faults mapped as one; simplification--segmentation, internal holes, lateral and vertical tears are ignored by the interpreter due to time constraints or because of limitations in the surface creation algorithms; geometr3'--intersection points and branch lines cannot be resolved during the interpretation procedure, they are deferred to a geometrical calculation step later in the workflow which may require complex parameters and rules to be defined that have more potential for introducing errors. The fault stick interpretations require a surface creation procedure to produce a geometrical structure representing the fault plane. The traditional approaches to fitting a surface to the 'fault sticks' tend to simplify or ignore the tipline geometry and significantly smooth irregularities in the surface that result from changes in the spatial position of the fault sticks relative to their neighbours. Re-sampling the fault picks to a smoothed regular grid and triangulation of picked points are common surface forming techniques both of which do not allow for holes and ignore segmentation, lateral and vertical tears in the fault.
Fault geometry modelling by skeletonization The fault modelling approach implemented here recognizes that detailed horizon interpretations inherently contain the spatial position of faults as minor offsets in elevation and when a combination of seismic attribute and surface attributes are applied to the horizon, one can delineate and infer with some accuracy and expediency, the trend and location of fault traces even when displacement approaches zero due to, for example, fault shadow effects. Leaving gaps in the horizon grid to represent the areas of fault heave provides areas into which the skeletonization procedure can be applied whereby the gaps are reduced to a line approximation for subsequent use in fault surface creation and displacement calculation. Defining the areas of fault heave can be very time consuming if done by manual picking alone, especially in areas that exhibit complex fault intersection. Automated approaches to identifying such areas can be employed based on artificially created attributes or attributes of the seismic volume (Tanner & Sheriff 1977; Bahorich & Farmer 1995) and are often useful in aiding horizon interpretation by revealing the spatial position of discontinuities. Alternatively, properties of the surface can be used to define the faulted regions, for example the dip of an element in the surface may be used to distinguish the spatial position for a fault. Once the area of fault heave is identified, the skeletonization procedure can be performed to produce a fault trace pattern.
MODELLING LARGE FAULT NETWORKS
Skeletonization The implementation of the skeletonization procedure used here requires a mesh representing the area of fault heave (alternative approaches could be used that do not involve a mesh). Inverting a horizon grid that contains gaps representing the faults and then triangulating this new grid generates such a mesh. Figure l a, shows the mesh as lines together with black dots representing grid edge nodes (horizon terminations). The internal nodes of the new mesh are ignored and all edge nodes (black dots in Figs la & lb) are considered to create a voronoi tessellation (Tipper 1991) in two-dimensional space (Fig. lb). This tessellation subdivides the region such that polygonal cells are created around individual nodes with all interior cell space (shaded region of Fig. lb) being associated with the closest node. The voronoi skeleton is the collection of cell boundaries represented by a set of connected line segments. The lines in the skeleton are clipped such that only lines falling entirely within the area defined by the fault mesh are preserved (Fig. lc). The result is a set of medial axis lines that approximate the position of all fault traces intersecting the horizon. The 2D skeleton is then 'dropped' onto the fault mesh so that the lines truly lie along the 'medial axis' of the fault surface mesh and represent accurately the position of fault and horizon intersections (Fig. l d). The mapping of the skeleton from 2D to 3D also eliminates artefacts in spatial position introduced by regionally dipping horizons. The whole skeletonization procedure is automated and does not require user intervention. The skeleton may contain some unwanted line segments that were not removed during the automated clipping procedure. Such segments occur where the width of heave across the fault is broad and the heave boundary is irregular. The voronoi tessellation produces boundary intersection points that fall inside the area of fault heave and consequentially line segments that also lie completely within the fault mesh. Such line
341
segments are assumed to be a valid skeleton line segment during the clipping procedure. These unwanted segments can be removed manually or using another automated algorithm that identifies short, terminating line segments. The skeleton may also be further manipulated by smoothing, simplification, cutting and joining to produce a fault trace pattern that meets geometrical and geological constraints. User intervention at this stage, although not required, is beneficial as it allows inconsistencies not detected in the automated procedures to be corrected and some coherent assembly of fault traces to be refined. The resolution of the mesh defined by the horizon grid affects the spacing between vertices of the skeleton. The grid dictates that voronoi regions often intersect at a similar spacing to the grid. Very short line lengths between close intersection points can also result that cause problems for surface creation algorithms so some simplification options are available to alter the skeleton removing line segments below a certain threshold value. The technique is also used to help maintain the skeleton structure when points are removed or line segments smoothed.
Surface creation from skeletons To build a 3D fault model, multiple skeletons are created at different elevations through the volume (Fig. 2a). Visualization is important at this stage of the model building process since it allows all fault-horizon intersections (the segments of the skeleton) to be seen in three dimensions for assessment of fault continuity (Fig. 2b). Several skeleton segments from different elevations often approximate a planar structure in 3D space that represents a fault surface. The skeletons contain useful information for checking structural coherency in the model such as intersection points, the trend and length of connected line segments and dip information used to guide the segment correlation. The intersection points should form branch lines in
voronoi tessellation
Fig. 1. (a) A mesh shown as lines representing the fault network is defined from gaps in a horizon interpretation or by some specified attribute of the interpretation. The mesh boundary points highlighted as black dots are identified and used within a voronoi tessellation to produce (b) a set of voronoi regions surrounding each point. The shaded cell shows the region of space associated with the internal point. The region boundaries form the initial skeleton. The line segments are selectively removed based on co-incidence with the mesh triangles in b, to give (e) the medial axis skeleton for the fault network. Finally the 2D skeleton is mapped onto the mesh to produce (d) a 3D medial axis skeleton for the fault network.
(a)
2d medial axis skeleton
(c)
(d)
342
D.L. LISTER
9
ior 0ge
Dust_2700
~ l,t ~ 4 "~~ ~- ~/ ~
~
: \\
skeletons
-----~._>
d) 3D that can be used to help build proper fault surfaces from the skeleton segments. The skeletons (vertices and connecting line segments) can also be efficiently manipulated (e.g. extended to make intersections, cut into separate pieces to create fault segments, merged etc.) to ensure that geometrical constraints are met and valid when compared to the 3D seismic. A fully automated surface creation algorithm might be employed taking advantage of the information made available through the skeletonization procedure but is beyond the scope of this paper. Skeleton segments are defined to be any contiguous collection of line segments that are terminated by intersection points or tip lines. Skeleton segments are selected for fault surface creation based on their geometrical relationship to each other so as to approximate a fault surface. The selection is arbitrary and can include multiple segments from the same skeleton (to define a vertical tear) with coincident intersections or with no intersection, and also segments from different elevations. A novel surface creation routine is invoked to build the fault surface from the segments based on triangulating points from neighbouring segment pairs. Depending on the distribution of points along the segments a choice of closest point (Fig. 2c) and distributed point methods (Fig. 2d) is available for accurate triangulation that honours lateral fault tiplines and internal holes (the hole could form by segmentation, or misinterpreted tearing but is shown here simply to demonstrate the triangulation methodology).
Modelling test case The case study area offshore Louisiana (Fig. 3b), Gulf of Mexico was chosen to demonstrate the new fault modelling technique. The high-resolution, shallow-water survey exhibits an extensive series of growth faults developed as a result of gravitational collapse during progradation of deltaic sequences. The fault structure is transitional between two major trends along the northern coast of the Gulf of Mexico. There is good lateral continuity of the highly reflective sequence of deltaic sands, silts and clays expressed in the seismic data (Fig. 3a) that are clearly offset by a number of faults. A total of seven horizons were mapped and used to build a 3D fault model using the skeletonization procedure described above. The final model (Fig. 4) is consistent with dislocations in the seismic dataset and the fault geometries are structurally coherent, correctly portraying branch line intersection and conjugate faulting structure. The lateral sampling resolution is very high (at the scale of the interpretation grid), leading to the corrugated surface appearance. The sampling density is
Fig. 2. (a) A vertical stack of fault skeletons defined from seven horizon interpretations through a 3D seismic dataset, the colours represent different skeletons. (b) A perspective view showing distinct curvilinear, planar and intersecting fault surfaces approximated by the skeleton segments. (c) A simple triangulation by a closest point method through selected skeleton segments (not taken from the model in b) and (d) the same segments triangulated using a distributed point method to create a surface approximating the fault.
important for detailed displacement calculations but could be reduced if surface smoothness is desired without affecting the methodology. The model was built from horizon interpretations in a matter of hours.
Fault displacement modelling Mapping fault displacement distributions in any structural model is a very important procedure to help identify mechanisms that cause slip distributions to deviate from an ideal elliptical shape so that growth models based on the observation of displacement patterns might be proposed. Displacement mapping may also act as an aid in solving problems of sub-surface fault correlation (Freeman et al. 1990; Maerten 1999), and when displacement information is applied to trace maps may also provide useful data for fracture analysis (Gillespie et al. 1993). Automated techniques for calculating displacement values for fault surfaces have tended to require at least two pieces of information, the fault surface geometry and horizon surfaces at one or more elevations through the model representing strata that have been displaced as a result of slip along the fault surface. Intersection points or lines are calculated where the fault surface intersects horizon surfaces (which may need to be truncated or extended to ensure proper intersection). The intersection of one horizon surface with the fault would ideally be represented with an ellipse lying on the fault surface that represents the intersection with the upper horizon surface on the footwall side of the fault, and the lower horizon surface on the hangingwall. The new fault modelling technique above describes a method for creating fault surfaces based on multiple horizon surfaces. The traditional displacement mapping techniques require both horizon and fault interpretations, but the new map based skeletonization technique can obtain elevation information recorded at locations close to fault surfaces on both the hangingwall and footwall sides of the fault so that an estimate of throw, heave and displacement information can be derived across nodes of the skeleton. An algorithm is presented that allows displacement information to be mapped from horizon interpretations onto fault trace skeletons and consequentially included as part of any 3D fault surfaces so that displacement contours across a whole fault surface can be generated. Horizon cut-offs (the points along the hangingwall and footwall horizon terminations) are used together with the points of the fault skeleton to produce a point set for delauney triangulation. After triangulation, triangles that do not contain
MODELLING LARGE FAULT NETWORKS
343
N 500 1000
1500 E i--
2000
2500
3000
Fig. 3. (a) Vertical seismic section showing seven horizon interpretations from a 3D seismic dataset (WestCam-same as Fig. 2 but different colours) located offshore Louisiana (b). (c) Time contour map for the yellow surface exhibiting many intersecting faults, (d) dark green surface shows the spatial position of the fault intersections change with elevation through the data.
3500
_f
'"-"9.
a point in the skeleton are removed (Fig. 5a) and some internal adaptation is automatically applied at intersection points (Fig. 5b) to create a mesh that represents the fault network with all triangles containing at least one point from the boundary. Vertex loops are then identified around each skeleton point that contain only edge points from the horizon (Fig. 5c). The vertex loops are used to decide which points from hangingwall and footwall should be used for calculating heave, throw and displacement for any given node in the skeleton. The method assumes dip-slip displacement for all
WestCam
~
Oul!of Mexico faults. The sense of throw can be maintained by considering the order of vertices in the skeleton segment relative to the downthrown side of the fault (this feature identifies the dip direction of any segment and is used to assist earlier fault trace correlation). The throw value can then be signed so that reverse faults can be distinguished from normal faults. The surface creation algorithms are adapted to accommodate the sense of throw so that throw values are consistent across a dataset. An example of the method and the application of signed throw can be seen in Figure 6 where a throw profile has been generated for
~
Fig. 4. 3D fault model (in time) built using the new methodology for the same WestCam dataset coloured by elevation. The seismic section validates the model that includes proper branch-lines and intersection points.
c
section
3Dfaultmodel
/
344
D.L. LISTER
~./Skeleton
~
~~~i~Sequential
edge-flip
No interna " after edge-flip
(b) (a)
~r ~
Internal diagonals " where all vertices are from skeleton
~ Skelet~ vertex (Sn) 9 Surrounding vertice7
/J
o Trace vertex(S) 9Boundary vertex
t2
t3 !Vie
(
%;
/v2",o._~
v~ v~
v16 v17
],~_I 1-v,5 v4~ %4
Counter-clockwise vertex definition
(c)
selected segments of one horizon of the West Cameron dataset. The throw profile shows faults of opposing dip direction as positive and negative along-strike with intersection points highlighted as vertical dashed lines. The throw profile tends to zero at the terminations of the faults which in this case is an artefact of the method where all segment terminations are assumed to be fault terminations, in reality, the faults are truncated by the edge of the survey. The method can be used to guide fault interpretation especially for complex and highly faulted regions where the sense of throw is subtle or unclear from map view.
Manual comparison of the method The displacement derivation technique is applied here to the example dataset from West Cameron. Values for heave, throw
b) 72.0 57.1 42.2 -~- 27.3 E 12.4 ~" -2.5 o -17.4( -32.3 -47.2 -62.1 -77.0
9
"'~,
v22.'" r
I I
~~ V21,'" t4
D Intersectionpoint
..... ~, ...w-~r~../~v~,v., ~ ^ _
tlength (m)
,/
Fig. 5. (a) A fault surface mesh created from delauney tessellation of skeleton nodes and footwall and hangingwall termination points of the horizon interpretation. Triangles that do not connect to a skeleton node are removed. (b) Some adaptation of the mesh is performed near to intersection points to ensure vertex loops can be easily defined. (c) Vertex half-loops are defined around the skeleton nodes on the hangingwall and footwall sides of the skeleton segment. Displacement values are calculated from half-loop mid-points and assigned to the skeleton nodes.
and dip-vector displacement were calculated for skeletons of the model using the displacement derivation methodology and consequently incorporated onto 3D fault surfaces. The automated displacement maps generated for several key faults are shown in Figure 7. Mansfield & Cartwright (1996) provide an independent study of fault growth for the same fault data. He performed manual sampling of throw on strike-normal profiles (of the same key faults as the automated study) at a regular spacing of 50 m and with approximately 20 measurements down each fault plane. Sample data was projected onto strike-parallel vertical planes for contouring, the results of which are shown in Figure 8. The throw values and displacement patterns of the automated method are checked against this independent manual interpretation to validate the automated approach. All throw distributions are shown from west to east. An overall comparison of throw distribution contours for the
0~='lkm Artifactof method (edgeof surveytracetermination)
Fig. 6. (a) Horizon map coloured and contoured in time. The fault skeleton is visible through the medial axes of fault heave for the horizon (b) and dip direction is calculated from the sign of throw calculated during displacement analysis. Fault segments highlighted in grey are depicted in (c) as a throw profile. Intersection points are clearly marked and a distinction is made between positive and negative fault throw based on skeleton segment orientation with respect to the plane of the throw profile.
M O D E L L I N G L A R G E FAULT NETWORKS
345
West (~) Throw contour projection I ntersection .... with fault _2
,.
--
"9
East -"--"
--
150e
0
P e _,. . . . . . . . . .
._
Distance (m)
South
-
West (~) Throw contour projection
5000
North
East
W
1
" , ~,------~o~~
|
@
Faults coloured by throw South West
~
342C
!
Intersection'-< with fault 1
0
Distance (m)
(~) Throw contour projection
North East
~
West
@
6900
Throwcontour projection (~) Fault 3
0-
4 /rj
.
~"/~107
0
Distance (m)
East
3000
0
9
Distance (m)
5000
Fig. 7. Fault displacement for three faults of the WestCam dataset calculated using the new displacement derivation technique. Throw contours determined on the fault surfaces are shown in two-way travel time (milliseconds).
manual sampling method and the new automated method shows a good correlation between the distribution of and approximate values for throw. As with the a u t o m a t e d m e t h o d , the distributions of throw are characterized by a general increase
with depth modified by frequently steep local changes in throw gradient. The location of increases in throw gradient that represent a local reduction or increase in displacement appear consistent with the intersections from other faults.
346
D.L. LISTER
I Fault 1 I West
Line of intersection with Fault 2
East
, ,
lOOO
15oo
~
~o a
2000 i Q
| | | I I I L L ~ I I J | | I I i I I I I ] I I I i t I l i | ' I I I I I , I T I I I T I I I I I l l I ~
1000
200o
3000
4000
50oo Distance ~ )
I Fault 2 I o-
1000 ~
2000~, ~
3000
-
t~
4000 --
0
1000
2000
3000
4000
5000
6000
7000
Distance (m) I Fault 61
0
1000
2000
3000
4000
Fig, 8. Fault displacement calculated from manual sampling of the seismic survey projected onto a plane and contoured in one-way travel time (milliseconds) (after Mansfield & Cartwright fig 5.6, 1996). Fault 1 and Fault 2 correspond to those in Figure 7.
5000
Distance (m)
Branch lines do not always /jJ/Complexdisplacement patterns correspond to displacement minima 7 ~ ~ . with many examples of non-idealised j _.j.7-- j_ j r - ' - ...... concentricslip distribution j
J
Highest displacement on NW dippingfaults
,,\
jJ.z.--Z /-/"
Displacement 'bullseyes; not restricted to one level
Fig. 9. A perspective view of a polygonal fault model coloured and contoured by absolute throw. The model is built using the methodology previously described. The complex fault network shows many throw perturbations coinciding with the interaction of other faults.
MODELLING LARGE FAULT NETWORKS Two contrasting phenomena can be found between the manual sampling and automated displacement mapping methods. Firstly, there appears to be a much higher frequency of small, localized throw anomalies with the manually sampled results. Secondly, the throw gradient near to the intersection with other faults is much less than the automated method, in other words, the intersecting faults appear to have less influence on the displacement profile under the manual sampling method. The higher number of localized perturbations of throw under the manual sampling method might be accounted for by changes in the contouring methodology or due to minor sampling errors by the interpreter. The contours generated by the automated method honour the sampled data points exactly: no re-sampling or aliasing of data takes place. The throw contours for the manually sampled results may be distorted by projection onto a strike-parallel plane prior to contouring and may also be re-sampled or aliased during the contouring procedure. The contouring interval and method will also affect the number of apparent 'bullseyes' in the throw distribution. Manual picking of displacement information from vertical sections may also be prone to human error and errors associated with mapping consistent horizons although it is recognized that the latter would also affect the automated approach. For the area of intersection overlap of two faults, we should expect an abrupt increase or decrease in throw in the vertical sections as the horizon is subjected to displacement across two faults. The manually sampled dataset does not show such marked throw deviations which may be due to interpretation error where the wrong horizon pair (hangingwall and footwall terminations) is chosen at the intersection overlap or alternatively, there may simply be no sample data at the intersection point. Measuring displacement accurately at intersection points is addressed by the automated methods described above.
Application of the new fault modelling techniques Three-dimensional shape and slip variations are closely related to localized stress/strain perturbations (Maerten 1999) providing an understanding of fault interaction, linkage and the localization of secondary structures below seismic resolution. The geometrical and displacement modelling methodologies outlined above are applied to a 3D survey over the Alba field. Central North Sea. The detailed analysis of the 3D survey encompassing the field has revealed a complex polyhedral network of small extensional faults within the mudrock dominated Eocene-Lower Miocene succession that surrounds the Alba sandbody.
Polygonal faults The Alba fault model is built from three horizon interpretations and contains more than 100 faults that intersect and abut each other with high angles of incidence (Fig. 9). The average length of the faults in the complex network is approximately 200 m with a maximum of 576m. Analysis of fault strike patterns reveals a bimodal distribution with most faults oriented N W - S E and a less pronounced set trending NE-SW. The regional dip direction of the sedimentary package is SE. All N E - S W striking faults dip up-slope whereas the N W - S E faults show no preference of dip direction. Displacement analysis across the faults also highlights the importance of the up-slope dipping faults despite being fewer in number and extent compared to faults striking parallel to the dip of slope. Throw across the up-slope dipping faults is on average twice that of the N W - S E striking fault set.
347
In general, displacements are highest at the vertical centre of each fault and can be seen in Figure 9 as concentric maxima in throw (highlighted with a number 3). There are many instances of non-idealized slip distributions in the fault network with slip maxima changing elevation along strike (highlighted with 1 in Fig. 9), or displaying localized slip maxima and minima that alternate along strike. The majority of slip maxima appear to be unrelated to sites of fault intersection, lying at the lateral centre of fault segments. The polygonal faulting is assumed to be coeval suggesting that the faults nucleated in a number of isolated locations and coalesced over time toward the presentday geometry. The preferred alignment and dip direction of the faults suggest some outside influence on geometry during fault growth. The sedimentary package appears to be of uniform thickness throughout the Tertiary indicating little or no regional or localized slope during the early Tertiary. A hiatus in sediment deposition co-incident with some regional tilting during the middle Miocene however, may have triggered fault initiation and controlled fault orientation. The intersecting three-dimensional geometry combined with displacement information represents the first real attempt to describe the intricate and complex interactions of this faulting style.
Conclusions The methodology developed for this research provides an alternative approach for assessing complex fault geometry and displacement in regions where a high number of faults exist that would normally require significant time to interpret. The technique allows detailed structural geometry to be created and assessed via automated displacement mapping so that an understanding of the kinematics of faulting in complex regions can be attempted. The work has implications for structural modelling at all scales in the oil and gas industry and research into fault growth and fault population statistics. I would like to thank both referees and S. Stewart for valuable contributions that have greatly improved the focus of this paper.
References BAHORICH, M. & FARMER, S. 1995.3-D seismic discontinuity for faults and stratigraphic features: The coherence cube. The Leading Edge, 14, 1053-1058. CAR-rV,'RIGHT, J. A. 1994. Episodic basin-wide fluid expulsion from geopressured shale sequences in the North Sea basin. Geology, 22, 447-450. CARTWR1GHT, J. A. 8~, LONERGAN, L. 1996. Volumetric contraction during the compaction of mudrocks. A mechanism for the development of regional-scale polygonal fault systems. Basin Research. 8, 183-193. DE~,HL'RST, D. N., CARTWRIGHT, J. A. & LONERGAN, L. 1999. The development of polygonal fault systems by syneresis of colloidal sediments. Marine and Pettvleum Geology, 16. 793-810. EGAN, S. S., KANE, S., BUDDIN, T. S., WILLIAMS, G. D. & HODGETTS, D. 1999. Computer modelling and visualisation of the structural deformation caused by movement along geological faults. Computers & Geosciences. 25, 283-297. FREEMAN, B., YIELDING, G. & BRADLEY, M. 1990. Fault correlation during seismic interpretation. First Break, 8, 87-95. GILLESPIE, P. A., HOWARD.C. B., WALSH, J. J. & WATTERSON,J. 1993. Measurement and characterisation of spatial distributions of fractures. Tectonophysics, 226, 113-141. GUOLIELMO, G. Jr, VENDEVlLLE, B. C. Jr & JACKSON, M. P. A. Jr 1999. Isochores and 3-D visualization of rising and falling salt diapirs. Marine and Petroleum Geology, 16, 849-861.
348
D.L. LISTER
LISTER, D. L. 2001. Computer Modelling and Characterisation of Intersecting 3D Fault Networks PhD Thesis, Imperial College. London, UK. LONERGAN, L., CARTWRIGHT,J. & JOLLY, R. J. H. 1998. The geometry of polygonal fault systems in Tertiary mudrocks of the North Sea. Journal of Structural Geology, 20, 529-548. MAERTEN, L. 1999. Mechanical Interaction of Intersecting Normal Faults: Theory, Field Examples and Applications PhD Thesis, Stanford University, California, U.S.. MANSFIELD, C. & CARTWRIGHT, J. A. 1996. High resolution fault displacement mapping from three-dimensional seismic data: Evidence for dip linkage during fault growth. Journal of Structural Geology, 18, 249-265. NEEDHAM, T., YIELDING, G. & FOX, R. 1996. Fault population description and prediction using examples from the offshore U.K. Journal of Structural Geology, 18, 155-167.
STUART, G. W., JOLLEY, S. J., POLOME, L. G. B. T. & TUCKER, R. F. 2000. Application of 3-D seismic attributes analysis to mine planning: Target gold deposit, South Africa. The Leading Edge, 9, 736-742. TANNER, M. T. ~ SHERIFF, R. E. 1977. Application of Amplitude, Frequency and other Attributes to Stratigraphic and Hydrocarbon Determination. In: PAYTON, C. E. (ed.) Seismic Stratigraphy-Applications to ttydrocarbon Exploration, American Association of Petroleum Geologists Memoir, 26, 301-327. TIPPER, J. C. 1991. Fortran programs to construct the planar Voronoi diagram. Computers & Geosciences. 17, 597-632. WATTERSON, J., WALSH, J., NICOL, A., NELL, P. A. R. & BRETAN, P. G. 2000. Geometry and origin of a polygonal fault system. Journal of the Geological Society, London, 157, 151-162.
Index Page numbers in italic, e.g. 153, refer to figures. Page numbers in bold, e.g. 321, signify entries in tables. 2D seismic acquisition 1-2, 282 3D seismic acquisition 2, 282 displacement mapping 135 history 1-2 impact on earth sciences 4 - 5 interpretation 2-3 long-offset 5 short-offset, high-resolution 40 versus 2D 1-2 4C seismic 3, 5 - 6 4D seismic 3, 5-6, 297-302, 311-320 accommodation space 36, 76, 88, 94, 101-113, 149, 155, 173, 173 accretionary prism 143-148 acquisition 284 wide azimuths 6 long offsets 6 acoustic impedance 236-238, 239, 313-314 inversion 4 acoustic rock properties 40 Aegean Sea 323, 324 Afen Slide 54, 59, 60, 191 Agbami Field, Nigeria 45 aggradational reflection configurations 93, 93-95, 99, 121 Akpo Field, Nigeria 45 Alba Field, North Sea 6, 346, 347 Allan diagram 5 amplitude anomalies 228, 228-229, 312 conical-shaped 263, 268-273,273 extraction maps 79 inversion 234 mapping 227 V-shaped 263, 268, 270, 274-275, 311-320 analogue models 157 of flow of melt 178 of reservoir architecture 25-33, 35-43 salt structure formation 159 sandbox 101-115, 129, 130 analogue outcrops 204, 207, 209, 251,266 Asgard Field development, offshore Norway 288-289, 290 aspect ratios 25, 28, 30-32, 38 asset teams 2 Atlantic Ocean, opening 119, 120 attenuation algorithms, multiple 3 attribute analysis 40, 219, 305, 333-334, 340 edge enhancement 250, 252 reflection index 286-287, 288-289, 290, 295 see also dip attribute displays autopicking area tracking 255-261 seed points 2 sensitivity to signal variations 3, 284 shape-based 227-230 steering criteria 2 trace difference 293 autotracking see autopicking AVO 5-6, 230, 237, 237-238 anomalies 4 integrity 302
back-arc spreading 144 back-thrust conjugate 144 ramps 48, 50 barriers to fluid flow, lateral 123-124 basalt 5-6, 75 basement imaging 13, 19 basin floor fan 27, 29-32 basin modelling 4, 321-331 basin-scale processes 1, 5, 25 bathymetry 171 - 172 Berkine Basin, Algeria 235-248 BIRPS deep reflection seismic projects 4 Bonga Field, Nigeria 45 borehole stability analysis 303 see also wellbore stability bottom simulating reflector 65, 67-69 Boulton Fields, Southern North Sea 220-221, 222 breach-point 37 breccias explosion 209 injection 265 Brendan's Dome igneous complex 200, 200 British Tertiary Volcanic Complex 205 Caister Field, Southern North Sea 219-221 calibration, of seismic 40 canyon, submarine 95, 98-99 formation 38, 39 wasting 26, 28 Canyonlands Grabens, USA, rift system 109-110 carapace doming, salt 150 carbonate mounds 295-296, 336 see also coral build-ups; mounds Cascadia accretionary wedge 144 Caspian Sea 306 Castellon Field, offshore Spain 92-100 channel 86-88, 228, 228-231,231 development, interaction with topography 73-81 equilibrium profiles 36 fill 28-29, 32, 49 see also channel plugging geomorphology 36-38, 40-41, 49, 49, 54 imaging 11-33 incision 76, 78, 80, 81, 87, 88, 124 knick point 36-37, 37 plugging 29-31 ridge 18, 20, 20 structural controls on 45-51 see also channel-levee system; meanders; moat-channel system; palaeochannel; turbidity flows channel-levee system 17, 19-20, 26-32, 38, 39. 41, 49-51, 76-79, 95 Chao dacitic coulee, northern Chile 215-216 Charlie Gibbs Fracture Zone 118, 119 Clare Lineament, offshore Ireland 118, 119 Clark Field, Southern North Sea 220-221,222 clinoforms breakpoint 83 geometry 216 prograding and downlapping 76, 95 CO2 injection 311-320
350
INDEX
COCORP deep reflection seismic projects 4 Coffee Soil Fault 149-150, 153-159, 163 common mid-point method 1 compactional history 328 see also differential compaction compensation faults 141 completion engineering 308 compressional deformation 120, 135, 146. 162, 193-197. 219 ridge-pull stress 180 conjugate margins 321 Connemara Field, offshore Ireland 118 continuity loss 219 contour currents 26, 86, 88, 88 contourites 55-56, 57, 63-71, 83, 87, 88, 120, 210. 280. 284. 286, 291,295 seismic characteristics 63,283 see also moat-channel system; moat-drift complex coral build-ups, chemosynthetic/cold water 283. 289 Coriolis Force 22 Corona Sill, Faroes-Shetlands basin 209- 216 crevasse splays 29, 229 crustal melting 73 crustal thinning 321,323
differential compaction 18.20, 66.95, 99, 124, 129, 13 l, 147-148, 162, 213. 216. 265, 274, 289 differential loading 171 digital terrain model, filters 54 dip attribute displays 25-26 direct hydrocarbon indicators 4 discontinuity analysis 250-261 anomalies 25 l, 2 5 2 - 2 5 7 , 253 data. dip-steered 250 surface 38 disconformity surfaces 11 see also unconformity surfaces down-warping, flexural 120 drainage patterns 12 drapes, sediment 29, 32, 38, 54 see also pelagic and hemipelagic sedimentation drilling hazards 35, 39-40. 287, 288-293 incidents 305, 305 performance 303-310 steering 294 use of 3D seismic in 279-296
Dan Field, offshore Denmark 149 Dan salt structure, offshore Denmark 149-150, 152-157, 159-160, 161 data, seismic overutilization 4 underutilization 3 debris flows 22, 54, 56-59, 65, 68, 69, 210, 265. 280. 282.
Earth model 303-308 earthquake activity, cause of sediment remobilization 273 Ebro continental margin, offshore Spain 91, 92, 95, 99 Ekofisk Field. offshore Norway 282 elastic impedance 236-242. 244-248, 244, 297 elastic inversion technique 230 environmental assessments 53 Erha Field, Nigeria 45 Erland igneous complex 200 erosional features 192, 196, 242, 265 as a sequence boundary 83-89 Messinian 91 Ethiopian Rift. Northern, rift system 109 extension 179, 184 basin modelling 321-331 regional 129 thick-skinned 162 thin-skinned 133-142, 156, 160, 323
284, 286
hummocky 291 see also debrites; Rona Apron; Sandoy fan debrites 25-28, 30, 30-32, 38, 65, 69, 69 d~collement surface 136-138, 140, 145 deep reflection seismic 4 deepwater depositional facies analysis 36 depositional systems 4-5, 17-18.20-22, 25-33 processes 36-38 deformation brittle 323 styles 209 velocity 322-325 see also compressional deformation; fault; fold-and-thrust style deformation; thrust faulting Delauney tessellation 342, 344 delta 121,123, 129, 131,191 development 91 - 100 progradation 75-77 density inversion 4, 66, 199, 271,273 depositional architecture 25-33, 35-43.47, 79, 263. 286 depositional environments, discerning 35 depositional systems deepwater see deepwater depositional systems shallow water see shallow water depositional systems depth conversion techniques 188, 250, 289, 295 depth imaging 233 detachment faults 84, 85 detachment zone 4 8 - 4 9 , 104, 123, 129, 136, 141, 156 see also d~collement surface detectability 282 development, integrated use of 3D seismic in 279-296 dewatering, compactive 146, 183-184, 273 DHI see direct hydrocarbon indicators diapirs see mud diapir; salt; shale diapir
Faroes-Iceland Ridge 187, 188 Faroes-Shetland Basin 73-82, 182, 199-217,289, 289-293 see also Faroes-Iceland Ridge; Faroes-Shetland Channel; Judd Deep; Wyville-Thompson Ridge Faroes-Shetland Channel (FSC) 283 seabed morphology 53-61, 63-71,279, 280, 2 8 3 - 2 8 5 Tertiary inversion 187-198 fault 249-261,284, 287, 295 array evolution 117-142 block 236. 328 rotation 28, 129, 179, 181, 183-184 compaction, layer-bound 146-148 concentric 158 conjugate arrays 106, 106. 123, 124 counter-regional 165, 166 cut-off lines 258, 259 displacement analysis 138-139, 339-348 extensional 112, 221,293, 308 offset 104, 105. 107, 123, 126, 131 geometry 5, 339-348 birds-foot 127 growth 140, 156 hydraulic properties 308
INDEX
351
linkage 105, 112, 129, 133, 137, 138, 141,250 listric 135-141, 179, 221,293 master 151 models 108, 150, 340-348 normal 144-148, 321,327 polarity change 101, 107, 110, 113, 136, 138-140 polygon maps 220 radial 158-159, 158, 162 reactivation 122, 129-131, 150 recognition 250 rock properties 5 roller 166, 168, 170, 175 rollover 167 sealing 183, 307, 335-336 slip rate 321 strike-slip 104, 109, 119 model 150 surface mapping 5 tip 109, i10, 125, 126, 129, 139, 257, 287, 340 rotation 106 trace kinked 102 splay 102-103, 129, 136, 138, 141 welds 165, 168, 170, 172, 174, 175 see also compensation faults; discontinuity; fracture; graben geometries; horizon cut-offs; horst; hydro-fracturing; Murdoch Fault; polygonal faults; rift basins, fault reactivation feeder ('blow-out') pipes 273, 275 fill and spill processes 36, 39 flat-spot analysis 4 Flett Ridge 73-82 flowing sands 283, 290, 291,295 see also shallow water flows fluid content, impact on acoustic impedance 4 flow models 321 prediction 6 fluidization 209, 214, 216, 263-277 fluidized flow 273 fluvial system, stacked braided 235 Foinaven Active Reservoir Management (FARM) experiment 297, 298 Foinaven Field development, West of Shetland 280, 289-293, 297-302 fold-and-thrust style deformation 144, 147 footwall traps 294 fracture analysis 342 gradient 306 mapping 6, 308 FSC see Faroe-Shetland Channel fuzzy logic analysis 238, 239-241,244-248
geomechanics 305, 307-308 Geotechnical, Environmental and Marine (GEM) Project regional investigation, Faroes Islands 281, 285,288 geotechnical problems 279-296 ghost notch 36 ghosting horizons 3 Gjallar Ridge, offshore Norway 177-185 glacial tills 290, 295 graben geometries 106, 110, 135, 146 collapse 158, 162 raft-graben framework 135-141 see also Viking Graben gravitational collapse 129-131 gravitational stability 323 gravity flows 26, 32, 38, 181,285 see also debris flows; debrites; turbidity flows modelling 177-178 sliding 135, 219 grid-based interpretation 2 grounding, salt 150 Gulf of Mexico 20-22, 35-37, 40, 282-283, 290, 291,306, 342-346 counter-regional salt system 165-176 growth-fault array 133, 141 loop current 22 Gulf of Suez, Egypt, rift system 109, 110 Gullfaks Field, Norway 6 gullies 59, 59
gas accumulations 40 shallow anomalies 84, 288, 290, 293-295, 307 gas chimneys 6, 152, 182-183, 183-185, 191,287, 293, 333-337 CO2 312- 320 see also gas accumulations; seep communities gas hydrates 307 geo-body identification 232-234 tracking 245, 247 geohazards 60, 147, 281, 283, 288, 303, 304, 334-335 see also drilling hazards geological model 253, 257-258 depicting uncertainty 307-308, 309
iceberg scouring 54, 55,284, 288, 290 Icelandic mantle plume 69. 187, 188. 194-196, 196, 199, 330 igneous geology 5, 75- 76, 192 depth of intrusion 213 rare earth geochemistry 188 see also sills image quality 2 imbricate thrust zone 144, 145 increased oil recovery project 249 injectites 263-277 inter-canyon sediments 28, 32 International Ocean Drilling Programme 40 leg 38 184 leg 131 143
hangingwall 287 folds 51, 144, 153, 165 Hawksley Field, Southern North Sea 219, 220-221,222 healed-slope deposits 37, 37 heat flow 325-326, 330-331 High Velocity Body 177 highstand 81, 94-95 Hod Field, Norway 6 horizon cut-offs 342 horizontal wells 11, 13 horst 146, 253 hydrocarbon maturation 321 migration 184, 273, 293, 334-335, 339 see also gas chimneys; hydrothermal fluid chimneys production 2 reserves 2 hydrofracturing 129-130, 147-148 hydrothermal fluid chimneys 183-185 systems associated with sills 199 hyperpychnal plumes 70
352
INDEX
leg 190 143 leg 196 143 well 808i 144, 146-147 well 1173b 144 well 1174b 144, 146-147 interpreter, seismic evolving role of 4 mindset of 3 inverse modelling 321-331 inversion 74, 112, 119, 119, 159, 184, 187-198 see also density inversion; Westray inversion complex IODP see International Ocean Drilling Programme Judd Deeps see Judd Falls Judd Falls (previously known as Judd Deeps) 54, 56, 57. 59, 188, 190-191, 191-197, 200, 279, 280 Judd High 73, 74 Karoo Basin, South Africa 181,204, 209-210, 216 kinematics 5 Kraka Field, offshore Denmark 149 Kraka salt structure, offshore Denmark 149, 152-156. 158-160 Kutei Basin, Indonesia 25-33 lacustrine basin 134 landslides 54, 59 see also slides lateral accretion surfaces 11, 12-16 limit of separability 282, 285 lithofacies 239 lithological distribution patterns 11 lithology direct indicators of 4 prediction 6, 11 lithospheric loading 326 load structures 273 Lomond Field, UK 6 Lower Congo Basin, offshore Angola 133-142 lowstand deposition 22, 311 systems tracks 121 Loyal Field, West of Shetland 280, 297-302 magmatic underplating 177-179 see also High Velocity Body magnetic anomalies, recognition of 4 mapping, subsurface geological, new age 1 mass transport deposits 37, 68, 69, 77, 78-80, 95, 192 processes 21, 28, 66, 285 wasting 84 see also Afen Slide; contourites; debris flows; debrites; landslides; slides; slump scars; Storegga Slide: talus deposits; turbidity flows master erosion surface 28, 30 McAdam Field, Southern North Sea 220-222, 221,224 meanders, channel 14-17, 17, 25, 37, 41, 99, 228-231 loop migration 18, 20 megamerges 4 Messinian 'salinity crisis' 91, 98-99 meteor impact craters 4 micro-earthquake events 6 Mid-Atlantic Ridge 196 Mid-North Sea High 149 migration depth, post-stack 220, 224 depth, pre-stack 5, 230, 264 over 211
smiles 202 time 220-221,223-225 time, pre-stack 264 Mississippi 12, 15 Delta 279 Fan 28 moat-channel system 66-68, 69 moat-drift complex 69, 284, 286 moraines, glacial 55, 56 mounds 181, 184, 263, 265,267-268, 269, 275 mud diapirs 25-26, 27, 179, 181,181-184, 183-184 mud volcanoes 6. 307, 334, 336 multiple suppression 283 Murdoch Fault 220 Murdoch Field, Southern North Sea 219-221 Murdoch K Field, Southern North Sea 219-226 NADW see North Atlantic Deep Water Nankai subduction zone, SW Japan 143-148 Navier-Coulomb behaviour 101 near-seafloor seismic studies 25, 35-43 net-to-gross variations 25, 31 neural network detection system 5, 219, 333-337 Niger Delta. Nigeria 41, 45-51,279, 336. 336 Njord Field, offshore Norway 249-261 noise levels 3, 236, 271,283,302 see also seismic data quality North Atlantic Deep Water (NADW) 63, 69, 187, 189-190 see also Northern Component Water North Atlantic Drift 280 North Sea 83-89, 101, 146-163, 263-277, 279, 288-289, 311-320, 328-330, 336-337, 339, 346, 347 Southern 110, 112, 219-226 Northern Component Water 187, 196, 196 southern gateway 195 Norwegian Sea Deep Water 64, 66, 70 offshore installation integrity 4 onlap 75, 79, 84, 95, 182, 184, 242 opacity function 2, 228 opal C/T precipitation 66 reflector 179, 288 optical stacking 4 Ormen Lange Field, offshore Norway 282, 285 outcrop scale limitations 1 overpressure 40, 129-130, 147, 158, 184, 273, 283, 290, 335 palaeo-flow direction 14 palaeo-sea bed 213-214, 267-268 palaeo-shelf-break 93 palaeo-topography 79-80, 88, 98, 99, 184, 194, 235 palaeobathymetry 88, 327-328 palaeochannel 98 palaeoclimatic records 63 palaeoslope 83, 147 Patanni, Gulf of Thailand, rift system 110, 111 pelagic and hemipelagic sedimentation 28-29, 38, 143-144, 263, 265 peperites 209 petrophysics 4 corrections 236- 237 for automatic geo-body identification 231-234 see also rock properties; synthetic seismograms phreatic eruptions 209 pipeline and cable routes 275, 296
INDEX pockmarks 4, 183, 191, 199, 271,334, 336, 336 point bar deposition 11-12, 13-16 polygonal faults 4, 54, 59, 126, 129, 146-147, 158, 158-159, 162, 184, 183-185, 199, 269, 271-272, 272-275,284, 339, 346, 347 ponded sediments 36-37, 49, 92 Porcupine Basin, offshore Ireland 117-132 pore pressure 302, 304, 305,306, 307 pressure prediction 298, 299 see also pore pressure principal components analysis 229 progradation 75-76, 84, 88, 92-93, 94-95, 121 protothrust zone 144, 145 pull apart basin 227
raft system 133-142, 219 ramp system 84, 85, 249 recording and processing, digital 1 regression 20, 21 regressional cycle model 45 relative sea-level change 21, 28, 30, 30-32, 76-77, 80. 81, 94-95, 120 relay ramp structures 101 - 113, 249 soft-linked 109 remobilization of clastic sediments, post-depositional 4, 263 -277 reservoir caprock 311 see also seal characterization 236, 301 connectivity 31, 39 distribution controls on 51 prediction 11, 31 management improvement 11 models 35, 38, 39 resolution 219 Reykjanes Ridge 188 ridge-channel systems 67, 67-68 ridge jumps 195, 196 rift basins 101-113, 119, 210, 235, 321,329 architecture 113, 134 fault reactivation 122, 129-131 margin fault system 105, 107-109, 112-113, 122 rift model offset 110 orthogonal 110 rig site surveys 53 Ringkc~bing-Fyn High 83, 84, 149 risk reduction 2, 282, 284-285, 287 Rita Field, Southern North Sea 220-221,222 rock properties calibrated to petrophysics 4, 235-248 strength 307 roho salt system 165, 168, 175 Rona Apron debris flow, Faroes-Shetland Channel 279, 280, 289
Saline Aquifer CO... Storage (SACS) methods 311-320 salt 40, 333 counter-regional salt system 165-176 diapirs 32, 135, 140-141, 150, 293-295, 340 evacuation 141, 149, 157-160, 162, 166, 167, 173-175 flow rate 172 imaging below 6, 101 'keel' structure 167, 171 pillows 136, 138, 141, 150, 152, 153, 154-155, 158, 160 rise and fall model of raft tectonics 133, 136, 140-141
353
roller-type structure 151 stock canopies 165 structures, evolution and growth 149-163, 165-176 tectonics 5 tongue canopy 172-175 wedges 159 welds 166-172, 174. 175 San Jorge Basin. Argentina 327-328 Sandoy fan. Faroes-Shetland Channel 284. 286 sandstone intrusions 263-277 saturation prediction 302 Schiehallion Field development, West of Shetland 280, 289-293,297-302 sea-level change 74, 83, 99 eustatic 73, 80, 134 glacio-eustatic 69 see also relative sea-level change seabed morphology 53-61 sensors, permanent 6 stability 281,282 seafloor conditions, unfavourable 40 spreading 119. 146, 147 see also ridge jump seal 236 breach 336 distribution prediction 11 fracturing of 274 integrity 183-184 resolution 219. 223 sediment apron 45, 95 see also Rona Apron extensional collapse of 47.49 bypass zone 173 drape see drape mobilization 263-277 supply, directional switch 83 waves 17. 19, 25-26, 31-32, 55-56, 67, 69-70 climbing 64, 68 sedimentation processes, discerning 35 seed points, for autopicking 2 seep communities 268 seismic 2D see 2D seismic 3D see 3D seismic band limitation 264 data quality 302 geomorphology 5, 11-24 imaging 264. 283 inversion 241-248 processing, challenges 2, 5, 269. 284, 301 refraction 177 resolution 2, 25, 35-36, 279, 283, 283, 288, 301,302, 339 signal/noise levels 3, 236, 271 stratigraphy 4-5, 11, 92-95, 183, 280 integration with seismic geomorphology 11, 12 thin-bed effects 312-314, 318 trace difference technique 293 valving models for permeability 308 wavelet, understanding of 2, 243 wide-angle reflection studies expanding-spread profiles 177 ocean-bottom seismometer 177 see also 2D seismic: 3D seismic; 4C seismic: 4D seismic; acoustic impedance; amplitude anomalies; attribute analysis: autopicking; AVO; bottom simulating reflector; calibration; clinoforms; data; deep reflection seismic; depth conversion techniques; depth imaging;
354
INDEX
detectability; dip attribute displays; discontinuity; elastic impedance; elastic inversion technique; fault; flat-spot analysis; geo-body; geological modelling; grid-based interpretation; interpreter; inversion; limit of separability; migration; multiple suppression; near-seafloor seismic studies; neural network detection system; noise levels; opacity function; shallow section 3D seismic; shear velocity; skeletonization algorithm; synthetic seismograms; trace shape extraction techniques; velocity; visualization; volume; voxel; V-shaped profile; workflow sequence restoration 170-173 sequence stratigraphy 73 - 81, 120-121,292 boundary 83-89 maximum flooding surfaces 166, 169 see also seismic stratigraphy shale diapir 335, 335 shallow section 3D seismic 35-43, 53-61,279-296 shallow water depositional systems 22-23 flow 40 shear velocity 236-238, 247 shelf ridges, shallow marine, imaging 12, 17-18 signal-to-noise ratio see noise levels sills 5, 180-181,209-217 geometry of 199-202, 207-208 junctions 202-207 simulation model building 258-260, 300, 301, 319 history matching 308 site development investigations 53 skeletonization algorithm 135,340- 342 Skjold Field, offshore Denmark 149 Skjold salt structure, offshore Denmark 149-150, 152-160, 162 Sleipner Field, North Sea 311-320 slides 83-86, 86, 88, 131,282 scars 68, 68 see also Afen Slide; gravity sliding; Storegga Slide slope fan 27, 29, 30, 32, 54, 58-59, 59 instability 40, 129, 131,284-285 slope-canyon morphology 25-33 slope-channel complexes 28-30, 30, 32 slump 86, 88, 120 scars 22-23, 23, 28, 65-67, 70, 191 see also slide scars Slyne Basin, offshore Ireland 118, 118 Smith, William (1769-1839) 1 soft-sediment deformation 4 source rock distribution prediction 11 maturation of 133, 141 South Pass 89 Field, offshore Gulf of Mexico 165 Southern Salt Dome Basin, North Sea 149-163 spatial aliasing 2 spatial stacking 4, 28, 35 spill-point 37, 37 stacking patterns 30-32, 38, 41, 49, 58, 121,235 see also spatial stacking Statfjord Field, Norway 6 steering criteria, for autopicking 2 Storegga Slide, offshore Norway 282, 285 strain analysis 5 rate history 321-331 stratigraphic discontinuities 12 stratigraphic growth patterns 136-138 stratigraphic targets, extraction of 227-234 stratigraphic trap 23-24, 333, 335
stratigraphy 'Christmas tree' 183, 185 reinvigoration of 1 see also seismic stratigraphy; stratigraphic targets; stratigraphic trap stress, regional 307 stretching, lithospheric 322, 327, 331 structural aliasing 6 structural complexity 219 structural restoration 340 structure, relationship to deep-water channels 49-51 subduction thrust system 143-148 subsidence 174, 181,321,323 differential 99 gradient 329 history 326 in overburden 6 regional 131 thermal 73, 76, 83, 119-121, 162, 195, 199, 327, 329-330 Suilven Field, West of Shetland 280 supercritical fluid 311 survey footprint noise 54 syneresis 147 synthetic seismograms 2 t-test 239, 240-241 talus deposits 98, 99 template horizon 227-230 thalweg development 36-37, 98 thermohaline currents 70 thrust faulting 47, 48 imbricate 144 see also back-thrust ramps; protothrust zone; subduction thrust system; toe-thrusting time-lapse 3D seismic see 4D seismic toe-thrusting 28, 32, 47, 48, 65 anticlines 25-26, 26-27, 29, 31 toplap relationship 93 trace shape extraction techniques 227 classification 229-230, 231 trackers 2 transfer faults 102, 104, 109, 113 zones 47-51, 73 Transverse Zone 150 Troll Field, offshore Norway 282 turbidite plays 134, 141 turbidity flows 18, 28-29, 54, 58, 70, 99, 143-147, 173 Tyne Fields, Southern North Sea 220-221,223 unconformity surfaces 16-17, 138, 162, 192-193 base Cretaceous, offshore Ireland 120, 131 base Permian, Southern North Sea 220, 221,222 base Upper Pliocene, offshore Norway 179 glacial, Faroes-Shetland Basin 64-65, 65, 282, 284-286, 289
Hercynian, Algeria 242, 244-245 intra Neogene, Faroes-Shetland Basin 64-69, 64-66, 190-196, 279-296 Messinian, Spain 91-100 mid Miocene, Central North Sea 293, 293-294 mid Miocene, Faroes-Shetland Basin 65, 66 mid Miocene, Southern North Sea 220 near top Oligocene, North Sea 83-89 top Palaeogene (formerly termed the latest Oligocene-early Miocene or LOEMU), Faroes-Shetland Basin 65, 66, 188, 190-197, 279-296
INDEX uplift Permian 210, 219 pre-Cenomanian 179-185 regional 120, 129, 130, 187 V-shaped profile 3 l 1-320 amplitude anomalies 263 ridges 188, 195-196, 196 Valencia trough, offshore Spain 91, 92 Valhatl Field, Norway 6 valleys buried 84, 94-95 incised 23, 124 Messinian dendritic system 94-98 Var sedimentary ridge 26, 31 velocity pull-up 182, 271 push-down 152, 191,312-320 push-down-amplitude ratio 319, 319 factor 316-319 Total Integrated Time Delay 316, 318-319, 318 shear 236-238, 247 structure 219 Viking Graben, offshore UK 263-277 virtual reality 260 visualization 5, 25, 40, 177-185,219-234, 249-261. 271. 273, 340 of fault systems 138 of production characteristics 297-302 volcanoes, submarine 199, 213
355
volume interpretation methods 217-234, 238 object probability 327 porosity 231,232 rendering 219 sculpturing 254-256 V!3ring Basin. offshore Norway 177-185, 279, 287-288. 330-331 Voronoi tessellation 341. 341 voxel, evolution 2-3. 271 welding, pre-salt/post-salt 140-141 well performance prediction 297-298, 300 planning 4, 260, 303, 307-308 wellbore stability 303, 304, 306-307 West Delta 133 Field, offshore Gulf of Mexico 165 West Shetland Drift (WSD) 64-70 Western Frontiers Association (WFA) regional investigation, Norway 53, 285 Westray inversion complex, Faroes-Shetland Basin 190, 191, 195. 195, 197, 200 workflow for production forecasting 249 for reservoir prediction 236, 238 for seismic interpretation and processing 250, 251,339 for time migration interpretation 223 for well planning 249, 303-309 optimisation 2 Wyville-Thompson Ridge 63, 64, 73, 74, 187, 188, 194-195
3D SeismicTechnology Application to the Exploration of Sedimentary Basins Edited by R. J. Davies, J. A. Cartwright, S. A. Stewart, M. Lappin and J. R. Underhill
;..
A 'new age' of subsurface geological mapping that is just as far ranging in scope as the frontier surface geological mapping campaigns of the past two centuries is emerging. It is the direct result of the advent of 2D, and subsequently 3D, seismic data paralleled by advances in seismic acquisition and processing over the past three decades. Subsurface mapping is fuelled by the economic drive to explore and recover hydrocarbons but inevitably it will lead to major conceptual advances in Earth sciences, across a broader range of disciplines than those made during the 2D seismic revolution of the 1970s. Now that 3D seismic data coverage has increased and the technology is widely available we are poised to mine the full intellectual and economic benefits. This book illustrates how 3D seismic technology is being used to understand depositional systems and stratigraphy, structural and igneous geology, in developing and producing from hydrocarbon reservoirs and also what recent technological advances have been made. This technological journey is a fast-moving one where the remaining scientific potential still far exceeds the scope of the advances made thus far. This book explores the breadth of the opportunities that lie ahead as well as the inevitable accompanying challenges.
Visit our online bookshop: http://bookshop.geolsoc.org.uk Geological Society web site: http://www.geolsoc.org.uk
Cover illustration:
A sculpted seismic volume from offshore Indonesia in which opacity has been used to delineate a sinuous channel that has been offset by a major extensional fault. The vertical blue lines represent exploration wells. Seismic data courtesy of Clyde Petroleum, image courtesy of Rob Bond (Paradigm Geophysical,Woking, UK).
ISBN
1-86239-151-3
ll!![l!!l!Ir