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
~
,.,
~"
.- ~
<"
"
i ~ . -
-
DF2
,,,I
,
w
....... <...
.
.
. . .
" -~'~2~.
=. . . . .
:z't~...-~,..-..~.-:
=,
9 ,.
.,,=.~.
,,,.
9
~ A ~ p i ~ ~ . . . . / .
j=,, 1000
68
P.C. KNUTZ & J. A. CARTWRIGHT
Fig. 10. Seismic cross-section of the lower slope (position indicated in Figs 7 & 8). Notations are as in Figure 5. reflectors of unit Bl onlap onto a slide scar developed in the lower flank of the WSD midslope section. At the present scale of vertical resolution (i.e. - 10 m) there are no indications of major offsets or thrusting between individual wave sets that could relate to slope instability.
Mass-flow deposits Debris flow packages identified within the 3D survey area form thick (up to 120m) lobate depositional features with internal chaotic acoustic expressions that extend from the slope far into the basin (Fig. 4). Three major mass flow packages (DF 1-3) can be identified within the sedimentary sequence of the WSD (Fig. 5). The most recent unit, DF 3, forms a lensoid depositional lobe with a smooth, apparently non-erosive base (Fig. 12). Unit DF 3 can be traced on 2D seismic lines to the Miller Slide located upslope to the NE of the 3D area (Knutz & Cartwright 2003), which has been related to mass failure during the midPleistocene (Stevenson 1991). In contrast, the debris flow package DF 2 that is interbedded between units B2 and C is characterized by laterally stacked units with erosive bases and intense thrusting along the margins (Fig. 12). The lateral extent and erosive base of individual debris flow lobes of the DF 2 package are revealed by the amplitude map in Figure 13. The debris flows appear to be connected updip to downslope channel systems that cut cross the ridge-moat topography. This pattern
may be entirely circumstantial or reflect a process-related link between the pathways of turbidity currents and initiation of slope failure. The high amplitudes (bright colours) within moat and downslope channels suggest the presence of more condensed, sandprone facies relative to the surrounding lowamplitude regions (Fig. 13). In the northern part of the 3D area a large section of the drift units B I - B 2 has been removed during several episodes of mass movements (white line in Fig. 13). In general the B2 surface reflects an extended phase of slope failure and downslope sediment transfer to the basin.
Discussion and conclusions The mapping and visualization of 3D seismic data reveal the anatomy of the West Shetland Drift and its basal unconformity (Figs 6 & 7). The basin section of the drift body forms an elongate-mounded sedimentary prism that has accreted upslope along the Shetland slope while forming a low-relief sheeted geometry in the Faroe-Shetland basin. This drift type is commonly observed on continental margins where strong western boundary currents, constrained against the slope by the Coriolis force, produce moat-channel systems bounded by prominent contourite levees. Numerous examples of elongatemounded drifts formed in the path of western boundary currents have been observed in the North Atlantic, e.g. Feni Drift (Kidd & Hill 1986), Faro Drift (Faugeres et al. 1984), Newfoundland
Fig. 11. Detail of seismic cross-section illustrating the succession of upslope climbing sediment waves. Bold numbers refers to ridge-channel systems shown on the dipmap in Figure 7b. The top bounding reflector of contourite drift units A, B1, B2, C and D are indicated. The hatched line demarcates truncation of unit A along the lower flank of the West Shetland Drift-slope section. Scale bars in metres.
NEOGENE CONTOURITES, FAROE-SHETLAND BASIN
69
Fig. 12. Seismicprofile from the basin (line position shown in Fig. 13). Note the large differences in seismic signature between debris flow units DF 3 and DF 2. The blue reflector defines the top of unit B2. Scale bars in metres. Notations are as in Figure 5. Drift (McCave & Tucholke 1986) and the Sackville Spur (Kennard et al. 1990). The seismic correlation of the WSD with the stratigraphy of well 214/4-1 suggests that the Pliocene units A, B1 and B2 accumulated rapidly with average sedimentation rates up to 0.1 m/ka on the thickest part of the drift (assuming that deposition occurred between 4 - 1 M a ) . The abrupt change from erosion to contourite drift accumulation around the Miocene-Pliocene transition could be the sedimentary response to a general decrease in northern source deep water formation in the Nordic Seas, or reflect an increase in deep water exported through other gateways relative to the Faroe-Shetland Channel. A tectonic control on North Atlantic Deep Water formation may have involved differential adjustments of the GreenlandScotland sill causing relative changes in the volume of overflow between the Faroe-Shetland Channel and the Denmark Strait (Vogt 1972, Wright & Miller 1996). The asymmetric distribution of Neogene sediment drifts in the Faroe-Shetland channel (Figs 1-3) presumably reflects the persistent current regime of northern source deep waters which prevent any substantial drift accumulation along the NW flank of the channel. An additional factor that would favour drift accumulation along the Shetland margin is the nearby sources of fine-grained sediments from the shelf seas of NW Europe and gravity driven sedimentation on the North Sea Fan. A general southward direction of alongslope currents is inferred from the southwestward migration of units B 1 and B2 (Fig. 9) and the tendency for moat-channels to bifurcate towards the SW (Fig. 7). On a basin scale, the development of a moat-drift complex on the Shetland margin seems to contradict conventional models of drift formation (Faugeres et al. 1999). This is because deep waters flowing southward along the SE flank of the basin would tend to deflect to the right into the basin and consequently attenuate the construction of contourite levees. The topography of the main moat channel as well as the first of the linear ridges of the B 1 surface (Figs 7a, b) suggests that bottom currents were following a slightly oblique upslope direction. The multiple moat-channels and dissected morphology of the B 1 surface may thus reflect the basinward evacuation of alongslope currents topographically confined by their own depositional products. This type of drift topography could be rather unique to the ocean gateway setting of the Faroe-Shetland Channel where deep waters below - 8 0 0 m depth are being funnelled into the North Atlantic. Phases of drift build-up were interrupted by periods of enhanced mass movements producing a series of mega-debrite units that intercalate the drift units (Figs 4, 5 & 12). Infilling and incision of the drift topography by stacked debris flows is
evident on the surface of unit B2 (Fig. 13). The predominant downslope sedimentation during this stage may be related to the glaci-eustatic fall in sea level that accompanied growth of northern hemisphere ice sheets at the Pliocene-Pleistocene transition (Raymo et al. 1992). In a more regional perspective increased stage failure and resedimentation could also relate to the late Neogene uplift of the Fennoscandia (Huuse, 2002). The ensuing succession of drift units C and D reflects the continuation of bottom current controlled sedimentation in the Faroe-Shetland Channel throughout the Pleistocene. The transition between the basin and midslope section of the WSD is characterized by a succession of sediment waves that protrude to the sea bed of the Shetland slope. The origin of alongslope trending sedimentary ridges on continental slopes has previously been related to depositional processes as well as synsedimentary gravity deformation (e.g. Lee et al. 2002). The internal configuration and external geometry of the WSD sediment waves suggests that these are formed by constructional processes (e.g. 'true' sediment waves) rather than slope instability. The multitude of sediment waves that occur at the
Fig. 13. Amplitude map of the B2 surface. The amplitude signal is strongly influenced by the footprint of mass flows and associated slide scars that form part of the overlying mega-debrite unit DF 2 (see crosssection in Fig. 12). The area is approximately similar to that covered in Figure 8. The thick lines define a series of vertically stacked debris flow lobes. Updip thin (e.g. < 200 m wide) downslope oriented channels (Ch) appear to be superimposed on the alongslope trending moat-channel systems.
70
P.C. KNUTZ & J. A. CARTWR1GHT
modern sea bed have evolved from a series of 3 on 4 ridge structures that accumulated on the early Pliocene unconformity (Figs 10 & 11). The asymmetry and lateral continuity of the wavy structures points to active migration rather than slump deformation. Moreover, rotational slumping on continental margins is likely to result in an arcuate thrust pattern rather than the linear-bifurcating structures observed in Figure 8 (e.g. Kenyon et al. 1978). Although the data presented in this study is limited to 10m scale resolution, we note that high-resolution seismic profiles collected across the same wave field show that unit D ( ~ t h e upper 30 m) contains numerous sub-metre scale reflectors that can be traced continuously between adjacent wave sets (Damuth & Olson 1993). It is possible that minor mass flow deposits, below the resolution of this study, may form part of the sediment wave succession, but these are more likely to smooth and infill the topography. Upslope migrating sediment waves have been related to (1) alongslope sediment transport driven by geostrophic ocean currents, (2) gravity driven sedimentary processes (turbidity currents or hyperpychnal plumes) and (3) interaction between these two end-member situations (Kenyon & Belderson 1973: Normark etal. 1980; Faugeres etal. 1999). Sediment wave fields formed on submarine fans or in vicinity of river mouths often show a consistent downslope decrease in wave dimensions while the spatial geometry of migrating sediment waves associated with contourite drifts is more irregular (Wynn & Stow 2002). The regional context of the WSD sediment waves, far away from fluvial sources or submarine fan environments, suggests that these are mainly a product of alongslope currents. However, this criteria does not rule out the possibility of process interaction. On the modern sea bed environment west of Shetland the development of benthic nepheloid layers has been identified as a principal agent of cross-slope sediment transfer (van Raaphorst et al. 2001). Sediment resuspension and formation of internal nepheloid layers occurs preferentially along zones where water mass boundaries and internal waves impinge on the sea bed. As sediments of the benthic boundary layer are transferred to depths of more than 6 0 0 m they become increasingly available for alongslope transport/deposition by southward-flowing Norwegian Sea deep water (Turrell et al. 1999; van Raaphorst et al. 2001). It is possible that the recent the sediment waves have principally formed during glacial periods when the availability of sediments introduced to the slope environment was much higher than at present. However, the influence of glacimarine environments on the shelf would not explain the initial growth of sediment waves during the e a r l y - m i d d l e Pliocene. The continued upslope accretion of the WSD sediment waves through the Pleistocene was probably sustained by the combination of prevailing thermohaline currents and the availability of finegrained sediments from the NW European shelf margin. We wish to thank PGS Ltd. and ExxonMobil International Ltd. for making the 3D seismic data volume accessible to this project. We acknowledge the support on Geoquest tESX applications provided by Schlumberger Systems Information. This paper benefited from helpful reviews by J.-C. Faugeres and D. Stow.
References AKHURST, M. C. 1991. Aspects of Late Quaternaly sedimentation in the Faeroe-Shetland Channel, northwest UK continental margin. British Geological Survey, 1- 118. ANDERSEN, M. S., NIELSEN, T., SORENSEN, A. B., BOLDREEL, L. O. & KUIJPERS, A. 2000. Cenozoic sediment distribution and tectonic movements in the Faroe region. Global and Planeta O' Change, 24, 239-259. BOLDREEL, L. O. & ANDERSEN,M. S. 1993. Late Paleocene to Miocene compression in the Faeroe-Rockall area. b~: PARKER, J. R. (ed.)
Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 1025-1034. DA.'qU'rH, J. E. & OLSON, H. C. 1993. Preliminary observations of Neogene-Quaternary depositional processes in the FaeroeShetland Channel revealed by high-resolution seismic facies analyses. In: PARKER,J. R. (ed.) Petroleum Geology of Northwest Europe." Proceedings of the 4th Conference. Geological Society, London, 1035-1045. DAVIES, R., CARTWRIGHT,J. & RANA,J. 1999. Giant hummocks in deepwater marine sediments: Evidence for large-scale differential compaction and density inversion during early burial. Geology, 27, 907-910. DAVIES, R., CARTWRIGHT,J., PIKE, J. & LZNE,C. 2001. Early Oligocene initiation of North Atlantic Deep Water formation. Nature, 410, 917-919. DAVIES, R. & CARTWRIGHT,J. 2002. A fossilized Opal A to Opal c,rI" transformation on the northeast Atlantic margin: support for a significantly elevated Palaeogeothermal gradient during the Neogene? Basin Research, 14, 467-486. DEAN, K., MCLACHLAN, K. & CHAMBERS, A. 1999. Rifting and development of the Faroe-Shetland Basin. Petroleum Geology of Northwest Europe. ln: FLEET, A. J. & BOLDY, S. A. (eds) Proceedings of the 5th Conference. Geological Society, London, 533-544. DICKSON, R. R. & BROWN, J. 1994. The production of North Atlantic Deep Water: Sources, rates and pathways. Journal of Geophysical Research, 99, 12 319-12 341. DOOLEY, H. & MEINCKE, J. 1981, Circulation and water masses in the Faroese channels during Overflow-'73. Deutsche Hydrographische Zeitschrift, 34, 41-55. FAt:GERES, J. C., GONTmER, E. & STOW. D. A. V. 1984. Contourite drift moulded by deep Mediterranean outflow. Geology, 12, 296- 300. FAUGERES, J. C., STOW, D. A. V., IMBERT, P. & VIANA, A. 1999. Seismic features diagnostic of contourite drifts. Marine Geology, 162, 1-38. HEEZEN, B. C., HOEEISTER,C. D. & RUDDIMAN,W. F. 1966. Shaping of the continental rise by deep geostrophic contour currents. Science, 152, 502-508. HUUSE, M. 2002. Cenozoic uplift and denudation of southern Norway: insights from the North Sea Basin. In: DORE, A. G., CARTWRIGHT, J. A., 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, 209-233. KENNARD, L., SCAFER,C. & CARTER,L. 1990. Late Cenozoic evolution of Sackville Spur: a sediment drift on the Newfoundland continental slope. Canadian Journal of Earth Science, 27, 863-878. KENYON, N. H. & BELDERSON, R. H. 1973. Bedforms of the Mediterranean undercurrent observed with sidescan sonar. Geology, 9, 77-99. KENYON, N. H.. BELDERSON, R. H. & STroDE, A. H. 1978. Channels, canyons and slump folds on the continental slope between southwest Ireland and Spain. Oceanologica Acta, 1, 369-380. KJDO, R. B. & HILL, P. R. 1986. Sedimentation on mid-ocean sediment drifts. In: SUMMERHAYES.C. P. & SHACKLETON,N. J, (eds) North Atlantic Paleoceanography. Geological Society, London, Special Publications, 21, 87-102. KNUTZ, P. C. & CARTWRIGHT,J. 2003. Seismic stratigraphy of the West Shetland Drift: implications for Late Neogene paleocirculation in the Faroe-Shetland Gateway. Paleoceanography, 18, 4. LEE, H., SYVITSKY, J. P. M., PARKER, G., ORANGE, D., LOCAT, J., HUTTON, E. W, H. & IMRAN, J. 2002. Distinguishing sediment waves from slope failure deposits: field examples, including the 'Humboldt slide', and modelling results. Marine Geology, 192, 79-104. 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., LAPPIN, M.
& UNDERHILL, J. R. (eds) 3D Seismic Technology: Application
NEOGENE CONTOURITES, FAROE-SHETLAND BASIN
to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 53-61. MCCAVE, I. N. & TUCHOLKE, B. E. 1986. Deep current-controlled sedimentation in the western North Atlantic. ln: VOGT, P. R. & TUCHOLKE, B. E. (eds) The Geology of North America. Geological Society of America M, 451-468. MCCAVE, I. N., MAMGnETTt, B. & ROBINSON, S. G. 1995. Sortable silt and fine sediment size composition slicing--parameters for palaeocurrent speed and palaeoceanography. Paleoceanography, 10, 593-610. NORMARK, W. R., HESS, G. R., STOW, D. A. V. & BOWEN, A. J. 1980. Sediment waves on the Monterey Fan levee: a preliminary physical interpretation. Marine Geology, 37, 1-18. RASMUSSEN, T. L., BACKSTROM, O., HEINEMEIER, J., KL1TGAARDKRISTENSEN, O., KNUTZ, P. C., KUIJPERS, A., LASSEN, S., THOMSEN, E., TROELSTRA, S. R. & VAN WEERING, T. C. E. 2002. The Faroe-Shetland Gateway: Late Quaternary water mass exchange between the Nordic Seas and the northeastern Atlantic. Marine Geology, 188, 165-192. RAYMO, M. E., HODELL, D. 8r JANSEN, E. 1992. Response of Deep Ocean Circulation to Initiation of Northern Hemisphere Glaciation (3-2 MA). PaIeoceanography, 7, 645-672. STEVENSON, A. G. 1991. Miller (Sheet 61~176 Quaternary Geology. 1:250,000 Offshore Map Series 1991, British Geological Survey. STOKER, M. S. 1997. Mid- to late Cenozoic sedimentation on the continental margin off NW Britain. Journal of the Geological SocieO', London, 154, 509-515.
71
STOKER. M. S. 2002. Late Neogene development of the UK Atlantic margin. In: DORE, A. G., CARTWRIrHT, J. A., STOKER, M. S., TURNER. J. P. & WHITE, N. (eds) E,xhumation of the North Atlantic
Margin: Timing, Mechanisms and bnplications for Petroleum Exph~ration. Geological Society, London, Special Publications, 196, 313-330. STOW, D. A. V., FAL'GERES,J. C., VIANA, A. & GONTHIER, E. 1998. Fossil contourites: a critical review. Sedimentary Geology, 115, 3-31. Stratagem-partners. 2002. The Neogene stratigraphy of the glaciated European margin from Lofoten to Porcupine. Svitser Ltd, Great Yarmouth, UK. TURRELL, W. R.. SLESSER, G., ADAMS, R. D., PAYNE, R. & GILL1BRAND, P. A. 1999. Decadal variability in the composition of Faroe Shetland Channel bottom water. Deep Sea Research, 46, 1-26. VAN RAAPHORST, W., MALSCHAERT, H,. VAN HAREN. H., BOER, W. ~; BRUMMER, G. J. 2001. Cross-slope zonation of erosion and deposition in the Faroe-Shetland Channel, North Atlantic Ocean, Deep Sea Research, 48, 567-591. VOGT, P. R. 1972. The Faroe-Iceland-Greenland aseismic ridge and the western boundary under current. Nature, 239, 79-81. WRIGHT, J. D. & MILLER, K, G. 1996. Control of North Atlantic Deep Water circulation by the Greenland-Scotland Ridge. Paleoceanography, 11, 157-170. WYNN, R. B. & STO~,', D. A. V. 2002. Classification and characterisation of deep-water sediment waves. Marine Geology, 192, 7-22.
Interactions between topography and channel development from 3D seismic analysis: an example from the Tertiary of the Flett Ridge, Faroe-Shetland Basin, UK ANDREW
M. ROBINSON
1 JOSEPH
A
& RICHARD
9
CARTWRIGHT J. D A V I E S
l
~
PETER
M
9
BURGESS
2
j
13DLab, School of Earth, Ocean and Planeta O, Sciences, Cardiff Universit3', Main Building, Park Place, Cardiff CFIO 3YE, UK (email:
[email protected]) 2Shell International Exploration & Production, Volmerlaan 8, PO Box 60, 2280 AB, Rijswijk, The Netherlands
Abstract: Interpretation of 3D and 2D seismic data in the Faroe-Shetland Basin (FSB) has revealed the important
role that structurally controlled bathymetry had in controlling sedimentary dispersal during Early Cenozoic thermal subsidence. The Flett Ridge was a major NE-SW structural high during some of the Palaeogene. actively growing and influencing adjacent sedimentary systems. During the Palaeogene this area of the FSB was a key entry point for siliciclastic sediment with a major deltaic system prograding towards the NW during the Middle Eocene. Prior to delta development, the Flett Ridge was onlapped during the Late Palaeocene and subsequently blanketed and drowned in Early Eocene times. Major periods of fluvial incision cutting up to 100m into the Middle Eocene strata are identified and a variety of channel networks with differing trends documented. Broad channels or valleys of earliest Middle Eocene age inherited the palaeotopography created by the Flett Ridge, whereas subsequent later Middle Eocene meandering channels trend perpendicular to the shelf edge and traverse the Flett Ridge structure. Seismic amplitude maps suggest that a complex and variable channelized drainage system developed across the coastal plain and delta top in the Middle Eocene. These channels influenced sediment supply creating an area of bypass to the more distal fan systems preserved at the base of slope. Later faulting on the ridge crest may also have affected the channel network pattern.
The objective of this paper is to show how 3D seismic data can be utilized to describe the sequence stratigraphic evolution of a sector of the basin margin in the Faroe-Shetland Basin (FSB) throughout Palaeogene times. Seismic and well data from the SE Shetland Margin allowed a detailed study of the response of the margin to local tectonics, changes in eustatic sea-level and sediment supply. A sequence stratigraphic approach is taken to subdivide the Cenozoic strata into depositional sequences, which can be traced locally, allowing documentation of the relative sea-level history for part of the SE margin of the FSB. The main focus of this paper is the interplay between local tectonics and changes in sea-level and the influence they have on sedimentary architecture. By using high quality 3D seismic reflection data we have been able to derive a high-resolution relative sea-level curve for this sector of the Shetland margin. Detailed mapping of delta top and pro-delta systems has proved successful using 3D seismic data allowing us to further understand the deltaic systems and the relative importance of their interactions with sea floor topography.
Regional geological setting The FSB has a dominant N E - S W orientation (Fig. 1), which manifests itself in the form of major rotated fault blocks and lineaments initiated in the Mesozoic (Duindam & Van Hoorn 1987; Hitchen & Ritchie 1987). Extensional tectonism initiated in the Permo-Triassic and continued episodically throughout the Jurassic. However, rifting reached its peak during the Cretaceous and continued until Palaeocene times (Dean et al. 1999). Major N W - S E transfer zones orthogonally cut this dominant N E - S W trend (Rumph et al. 1993). Major zones of transfer and late Cenozoic inversion are the Judd High and the W y v i l l e Thompson Ridge (Fig. 1). In addition to major extension, the FSB experienced a significant magmatic episode at the end of the Palaeocene. Intrusive and extrusive igneous material is preserved in the FSB as a result of Early Eocene spreading of the North Atlantic to the NW (Andersen 1988; Naylor et al. 1999: Ritchie et al. 1999). A broad sheet of lavas and tufts are found on the Faroe plateau and in the FSB and they form part of the
larger igneous province which was associated with the protoIceland hotspot (White 1988: Morton et al. 1988: Ritchie et al. 1999). The Palaeogene fill of the FSB developed as a classic post-rift sag basin with a N E - S W trending axial depocentre. Thermal subsidence occurred during the Late Palaeocene and Eocene (Cliff 1999: Cliff & Turner 1998) and a thick succession of sediments are preserved (Fig. 2). The fill onlaps onto both the Shetland and Faroe margins (e.g. Davies et al. 2004). Throughout the Palaeogene, denudation and drainage of the marginal areas like the Scottish Highlands supplied siliciclastic sediment into the basins surrounding the British Isles, particularly the shelf surrounding the Shetland Islands (Ziegler 1990: Anderton 1993: Knott et al. 1993). There are considerable thickness variations within the Palaeogene succession, which is seen to both onlap and drape the inherited palaeo-topography. Small areas of thick deposits, as in the Foinaven and Flett sub-basins developed (Lamers & Carmichael 1999) and are separated by comparatively thinner areas on the intrabasinal highs (Clair, Rona, Corona, Flett and Westray Ridges and Judd High, Fig. 1). The study area discussed in this paper is located on one of these intrabasinal high areas, the Flett Ridge. From detailed mapping of the Mesozoic structure, it can be seen that the Flett Ridge is a N E - S W trending high that has a distinct saddle-like feature on its crest (Fig. 3). A deep time-slice through the structure (Fig. 4) shows a near perfect circular feature that could indicate an igneous pluton at depth (timeslice at 4250 ms). Combining this evidence with the presence of numerous of sills in well 205/10-2b where many of which are felsic and indicate crustal re-melting (Smallwood & Maresh 2002), the presence of a deep plutonic/volcanic edifice under the Flett Ridge is viewed as highly likely.
Sequence framework An extensive progradational deltaic wedge has been mapped in the area of the Flett Ridge and Flett sub-basin (Figs 1 & 2). Biostratigraphical and lithological information is sparse, with only five nearby wells. Consequently there is limited knowledge about lithology and palaeo-environmental setting. However,
DAVIES, R. J., CARTWRIGHT,J. A., STEWART.S. A.. LAPPIN,M. & UNDERH1LL.J. R. (eds) 2004.3D Seismic Technology:Application to the Exploration of Sedimentao' Basins. Geological Society, London, Memoirs, 29. 73-82. 0435-4052/04/$15 9 The Geological Society of London 2004.
74
A.M.
ROBINSON
ETAL.
(a)
N
T
50 km
i62~
N
~
Islands
o
61 N
60 ~
N
"'
~I~
Inversion anticlines
0 L:'
Bathymetry 200m contours
_::
:~t'; ~ ~-"',: .-' ~8W
I S E
~
4W
Limit of Lava
0~W
(b)
(c) Ma Ch. BP
This study
T
I'
o
Lithology
>,31
61~ N
r 35 Cl, 10s
<,3 Limit o 3D seismic
su?
:.... -- ~ """--..l~j /
~ :/?.?
~ ~'~.
r ~PsPi;?~ ['fl~''''''~ f i g 7 1 ? s ? ' !
C1! 39 C1(
....... Leocl
C1~
41
b
-c1~ rg~
9
Approximatei positions of figures 6a,
~ 43 th , - 32(]
- - ~ e ~
~
':
0 - - 4947_ ~ r8( ........ ~e&'11~ ..........
6b, 7b &7c.
i1
110 kmi
20,,;
~
sandstone
I ...... ~
~
"
...... Eeoc1
.223 Lu 53 - :24 . . . .
55 T5C Wz ,,, 57jc25r4_o O_
~w
i
51 c22
20~
9
horizon used in this study
Mudstone/claystone volcanic tuff coal
within the area of 3D coverage (Fig. 2) a number of mappable semi-regional reflectors can be seen. These reflectors form the boundaries of depositional units or sequences that are genetically similar (Van Wagoner et el. 1988). The sedimentary response to changing global sea-level and local tectonic effects is documented by seismic reflector geometries and their stratigraphic position within the depositional system. A relative sea-level history for this particular part of the SE margin of the
~e~l _ _
Lpat2~'_'~.[.i ;.
$ J sL:2 e <,
61 __ r20 - 327- -
<~ uJ 63_ ?,2~rt e
65 32c~
Lpall Epall
Fig. 1. (a) Location map showing the main structural elements of the FarceShetland Basin (FSB). Bathymetry, SE edge of lava and inversion axis are also shown. FR, Flett Ridge; CoR, Corona Ridge; CR, Clair Ridge; RR, Rona Ridge; WR, Westray Ridge; JH, Judd High; WTR, Wyville Thompson Ridge; MR, Munkagrunnur Ridge. (b) Detailed location map of the study area showing the limit of the 3D seismic survey and positions of seismic panels and amplitude maps and key wells. (c) Stratigraphic chart showing Palaeogene stratigraphy in FSB. BP T-scheme of Ebdon et el. (1995). Magnetostratigraphy (Ch.) of Cande and Kent (1995).
FSB can be determined by mapping reflector packages in the Palaeogene and interpreting their stratigraphic significance.
Early P a l a e o c e n e 'Danian' ( E p a l l ) In the Early Palaeocene, reflector packages can be seen to onlap and thin onto the Flett Ridge structure both landward and
EVOLVING TOPOGRAPHIC CONTROL ON CHANNEL DEVELOPMENT
75
Fig. 2. Regional 2D seismic line trending SE-NW across the study area (see Fig. 1b for location). The main features seen on the seismic line are the NE-SW trending Flett Ridge structure, the limit of Palaeocene lavas and the progradational clinoforrn systems that feed base of slope fans in the Middle Eocene. All seismic markers that have been interpreted are discussed in detail in the text.
seaward of its crest (Fig. 3). This Early Palaeocene package is thin and consists of a shale/mud succession with biostratigraphic evidence suggesting an open marine, outer shelf- upper bathyal environment (Fig. 1). No deposits of Early Palaeocene age occur on the crest of the Flett Ridge. Well 205/10-2b is located near to the crest and has a much thinner Early Palaeocene section (less than 40 m) compared to the more basinal well 205/9-1 which contains over 105m of comparable section. Non-deposition continued on the ridge crest throughout the Early Palaeocene.
Late Palaeocene 'Selandian-Thanatian' (Lpall) Significant thicknesses of Upper Palaeocene sediments continued to onlap and thin onto the Flett Ridge (Figs 2 & 3). Large volumes of siliciclastic sediment fed from the Shetland Margin were deposited in the Flett sub-basin including local sands that were encountered in 205/9-1. These sands are seen to thin and pass laterally into claystone or mudstone intervals in 205/10-2b, which is located near to the crest of the Flett Ridge (Fig. 1). Smallwood & Maresh (2002) highlighted the distribution of these sands and argued that they were controlled by sill intrusions at
Fig. 3. NE-SW trending 3D seismic line (see Fig. lb for location) showing onlap of the Early Palaeocene (Epall) units onto the Flett Ridge structure. Later thinning of the Early Eocene (Eeoc 1 & 2) units are also highlighted.
depth (see Smallwood & Maresh 2002, fig. 11). There is a considerable thickness variation of Upper Palaeocene strata found between the two wells mentioned above. Over 1600 m of Upper Palaeocene sediments are recorded in 205/9-1 and this sequence drastically thins to just over 300 m in 205/10-2b. A high-amplitude reflector can be seen abutting against the ridge flank (Fig. 2). The lithology of this reflector has been defined as dolerite and interpreted as a lava flow based on information from cuttings from well 205/9-1. The lava seems to represent the feather edge of the Upper Series of the Faroe Plateau Lava Group (Ritchie et al. 1999) and is concordant with the surrounding strata suggesting these strata were extruded and onlapped the NW flank of the ridge.
Latest Palaeocene-Earliest Eocene 'Late Thanatian-Early Ypresian' (Lpal2) A distinct change in reflector geometry is seen in the uppermost Palaeocene to lowermost Eocene, with sediments being deposited over the top of the ridge crest for the first time during the Palaeogene. A dramatic thinning of depositional sequences
76
A.M. ROBINSON ET AL. to continued subsidence throughout the Early Eocene on the SE margin of the FSB.
Middle Eocene 'Early Lutetian' (Meocl)
Fig. 4. Horizontal time-slice at 4260 ms showing near circular body under the Flett Ridge at depth. This has been interpreted as an igneous intrusion. A smaller circular body is seen to the southwest of the main body and many high amplitude sills and dykes are associated with the intrusion. See Figure lb for location of the timeslice.
capped by the reflector termed Lpal2 can be observed over the ridge (Fig. 3). The palaeogeographic significance of this blanketing of sediments as opposed to the previous onlap onto the ridge is of great importance when considering the stratigraphic and depositional setting. Well calibration of this uppermost Palaeocene interval shows abundant coals, sandstones, mudstones and tufts which strongly suggests that deltaic conditions prevailed at the Palaeocene-Eocene transition. An increase in sediment supply from the margin allowing for delta progradation could explain why the crest of the Flett Ridge became blanketed with sediments at the end of the Palaeocene. Additionally, the submarine conditions that the Flett Ridge was experiencing could have been due to an increase in subsidence on the basin margin causing to the ridge crest to subside relative to the sediment pathways surrounding it. The change in thickness of uppermost Palaeocene and lowermost Eocene sediments over the crest of the Flett Ridge is coeval with the deposition of the Sele and Balder Formations, the latter of which coincides with the start of sea floor spreading in the North Atlantic during the Eocene and the concomitant widespread deposition of tuffaceous and volcaniclastic sediments (Knox & Morton 1988; Ritchie et al. 1999). The thinning of sediments of the Sele and Balder Formations indicate that the Flett Ridge still showed some positive topographic relief on the sea floor during deposition (Fig. 3).
Early Eocene 'Ypresian' (Eeocl) Lower Eocene sediments exhibit parallel reflection configurations and these suggest that the previous deltaic environment was short-lived and that fully marine conditions were resumed during deposition of the Early Eocene. Biostratigraphic data from wells 205/9-1 and 205/10-2b support a marine setting. A relative sea-level rise is therefore inferred throughout the Late Palaeocene into the Early Eocene as the Balder Formation delta was flooded and the succession became progressively more marine dominated. This relative sea-level rise can be attributed
Shelf deposition during relative highstand conditions continued into the Middle Eocene. However, a major incision surface with over 100 m of erosional relief is recognized and mapped locally on the NW flank of the Flett Ridge (Figs 5a & 5b). When correlated to well 205/9-1, the major incision event occurs at the top of a sand-rich package, locally known as the Upper Judd Sands, which are known from biostratigraphic reports to have been deposited in a shelf setting. When mapped out on 3D seismic data the erosion surface shows a broad channel-like feature that trends parallel to the shelf edge. This channel system is approximately 2.5 km wide and can be seen on the amplitude attribute map of this horizon by the distribution of higher amplitudes seen within the axis of the channel (Fig. 5b). The width of the channel narrows to less than 1 km in the north. It can be seen to be trending N - N E along the shelf and then turn abruptly downslope to the west near the northern limit of the 3D seismic survey (Fig. 5b). Smaller channel systems, which appear to be tributaries to this main channel, are seen to the east of the main channel feeding into its axis. The amount of incision within the channel system can be seen to vary along its length. Incision of approximately 80 m can be seen in the south of the survey (in a more proximal position with regard to the shelf edge) and increases up to 200 m in the north where the channel turns towards the basin centre. We interpret this erosional feature as a submarine channel which has a deeper incised canyon at its northern end. It is postulated here that either a lowering in relative sea-level over the shelf triggered the incision of this channel system, or that the channel was controlled by local faults. The stratigraphic significance of this feature will be discussed later in the discussion section.
Middle Eocene 'Mid-Lutetian' (Meocl-Meoc2) After incision during the early part of the Middle Eocene, a thick sequence of clinoforms prograded and downlapped over the channel system (Fig. 3). The direction of clinoform progradation is from the SE and is interpreted as resulting from a major period of subsidence. This created significant accommodation space which filled by aggradation and progradation of deltaic systems during highstand conditions. The clinoforms of the delta have relatively steep slope angles of approximately 1.5 degrees (not corrected for compaction), and suggest that clinoform relief is also significant and that water depths were up to 400 m at the base of slope position.
End of the Middle Eocene 'Bartonian' (Meoc2) A second phase of pronounced incision occurred into delta top sediment near the end of the Middle Eocene. This incision was not a single event, but is seen on the 3D data as a series of closely spaced incision surfaces, cut into successive delta topsets. Figure 6 shows an example of the Meoc 1 incision event. Locally, up to five separate incision surfaces can be identified over a relatively small time interval (Fig. 6b). Detailed mapping of these individual local incision events shows complex channel planforms that trend in a N W - S E direction (Figs 6a & 6c). These channels have a low sinuosity, are 500---1000 m in width with low-amplitude meanders and braided planforms. They are affected by later E N E - W S W trending faults (Fig. 6a). These channel systems are located near the slope break between the topset and the pro-delta region of the clinoform, suggesting that
EVOLVING TOPOGRAPHIC CONTROL ON CHANNEL DEVELOPMENT
77
Fig. 5. (a) 3D seismic line showing thinning of Lower Eocene depositional packages onto the Flett Ridge (see Fig. 5b for location). A major incision surface is highlighted basinward of the ridge (Meoc 1, dashed line). (b) Amplitude extraction map on the major incision surface (Meocl) showing a broad shelf-parallel channel network (light colours, high amplitudes: dark colours, low amplitudes). The channel turns sharply towards the west into a deeper canyon at the northern end of the survey. Minor drainage in the form of tributary channels can be seen draining off the Flett Ridge.
they were incised into the edge of the delta plain, close to the delta front. Well and seismic data shows that the channels cut into delta top sands interbedded with silts, coal and limestone fragments, and the incisional geometries are thus interpreted as an expression of a lowering of relative sea-level. The complex network of channels observed on the delta top and pro-delta slope provide evidence for a discrete focus for clastic input from the Shetland margin. Additionally, small-scale (up to 100m wide) confined channels are seen to break out of the main broad channel axis (Fig. 6). These smaller channels are narrow, perpendicular to the shelf and become broader distally towards their termination in lobe units. These particular channels are located on the transition between the topset and the clinoform and could represent conduits for mass flow deposits, which broke out of the main channel axis and deposited sediment on the pro-deltaic clinoform slope. However, they could alternatively represent instability at the delta front and be more indicative of collapse of unstable, uncompacted sediments at the delta front transition.
Latest Eocene-Early Oligocene 'Priabonian-Rupelian' (Leocl) Clinoforms formed during the Late Eocene, draping the incision surface are indicative of another relative sea-level rise. These clinoforms seem to have lower angles of slope ( < 1 degree) than the Middle Eocene clinoforms and probably represent progradation into shallower water depths (Figs 2 & 3). Additionally, they could represent a lower rate of sediment supply. The uppermost Eocene succession is close to the present-day seabed and only a thin post-Eocene to Recent sequence is present.
Controls on channel development Superimposed maps of the Meocl and Meoc2 channels and the Flett Ridge palaeotopography show many interesting correlations (Figs 7a, 7b & 7c). Firstly, during Meocl times the main
78
A.M. ROBINSON E T A L .
Fig. 6. Seismic profiles and amplitude extraction maps showing channels from the end of the Middle Eocene incision event (Meoc2, dark green horizon). NB: red colours show high amplitudes and black colours show low amplitudes. (a) shows a 3D seismic line through the axes of the channel system and draped by its seismic amplitude looking south east up the channel network. The characteristics of these channels are low sinuosity, braided to slightly meandering planforms and they have a width of 500-1000m. (b) shows a NE-SW seismic section with up to five incision events at the Meoc2 event. These are interpreted as small base level falls into the delta top sediments. (c) shows an amplitude extraction map of one such incision surface in the Meoc2 channel systems. Note the braided and sinuous forms of the complex channel network that trend NW and perpendicular to the shelf edge.
channel axis development is seen basin-ward of the ridge crest following the ridge contours and running parallel to the palaeoshelf edge (Fig. 7a). There is also evidence for small channel features (or tributaries) diverging off the ridge crest which link to the main channel axis to the west, indicating that the ridge crest was a minor drainage area for the main channel. At the NE limit of the ridge the M e o c l channel is seen to bend sharply basin-ward and become much more deeply incised and canyon-
like. The planform of the Meocl channel shows that it is diverted around the Flett Ridge structure and indicates that the ridge had some positive topographic relief at this time. River systems are known to cut downwards and incise into structurally elevated features during base level fall if the rate of erosion matches or exceeds the rate of any uplift (Jackson et al. 1996; Burbank et al. 1996; Humphrey & Konrad 2000). In contrast, channels in a wholly marine environment would be more likely
EVOLVING TOPOGRAPHIC CONTROL ON CHANNEL DEVELOPMENT to trace and follow the contours of a palaeo-ridge, so it is interpreted that the incision in this M e o c l example was not the result of fluvial erosion.
79
Meoc2 channel morphologies show a markedly different pattern to the Meocl channel system discussed above. By superimposing these later Meoc2 channels on the Flett Ridge structure map they can be seen to traverse the ridge crest perpendicular to the shelf edge. The ridge crest seems to have had no effect on the channel architecture and pattern (Figs 7b & 7c). The channel bifurcates on the basin-ward side of the ridge and trends to the SE over the two high crests of the ridge. An area out with the channel axis (possibly an interfluve or coaly floodplain based on its high amplitude seismic response) sits directly over the small saddle feature between the ridge crests. This architecture and spatial variation of these Meoc2 channels and their relationship to the underlying Flett Ridge suggests that the ridge was no longer a significant topographic feature, with the channels not being affected by any remnant structural relief. Instead, the channel pattern is incised as a geomorphic response to a fall in relative sea-level on the delta top, without a structural control of depositional or erosional topography. A summary of the change in channel morphology and the controls on their distribution with respect to the Flett Ridge can be seen in Figure 8.
Discussion From seismic and well data we have been able to detail the stratigraphic evolution of a marginal, shallow marine shelf environment and recognize the role of buried tectonic features on changes in relative sea-level and the effects of this on the sedimentary succession. The schematic evolution from the Early Palaeocene to the Late Eocene is displayed as a series of cartoons, illustrating the change in depositional style through time (Fig. 9). During the Early Palaeocene, the Flett Ridge was a positive topographic feature that influenced the sedimentation on its flanks (Fig. 9a). Onlap of reflectors can be mapped onto the flanks of the ridge, but it is not known whether the ridge was above or below sealevel in the Early Palaeocene. A significant thinning from 110 m of muds off structure in well 205/9-1 to less than 40 m nearer to the crest of the structure (in well 205/I0-2b) suggests a more submarine origin. No erosion can be seen on top of the ridge at this time. Additionally, there is no supporting evidence of downlap of clinoforms onto to the ridge in the Late Palaeocene (Lpall times). However, it is known that off structure in well 205/9-1 there were significant deposits of deepwater sands during the Late Palaeocene (e.g. the Vagar sands of Late Palaeocene age- Sequence-T35, Ebdon et al. 1995). At this time however, sediment was deposited over the crest of the ridge and this indicates that there was a sufficient increase in water depths by subsidence or eustacy to allow the accumulation of significant sediment on the crest. Alternatively an increase in
Fig. 7. Amplitude extraction maps of (a) earliest Middle Eocene (Meoc 1) channel and (b) & (c) late Middle Eocene (Meoc2) channels. NB: light colours are high amplitudes: dark colours are low amplitudes. (a) shows the broad channel axis (short dashed horizon of Meocl ) that parallels the shelf-edge traversing in a NNE direction and turns abruptly into a deep canyon. The channel system can be clearly seen to run around the NW margin of the palaeo-topography of the Flett Ridge (white contours in TWT-F at the Epall level). (b) & (c) show 2 individual channel events of Meoc2 age (long dashed horizon), These channel systems clearly traverse the palaeo Flett Ridge trending perpendicular to the shelf-edge in a NW-SE direction. Smaller, confined channels (up to 200 m wide) are seen to break out of the main channel axis and could be interpreted as small confined mass flow deposits which run off the delta front and down the slope. Epal 1 contours (in TWTT) of the Flett Ridge are shown in black.
80
A.M. ROBINSON E T A L . (Fig. 9e). There is a high frequency of incision events at this stage, with erosional truncation seen on virtually every seismic loop (Fig. 7). This can be interpreted as either small-scale oscillations in the eustatic sea-level over a relatively short time period, or by minor changes in the rate of discharge and sediment supply coming off the hinterland. Lastly, they could be the result of pulsed uplift of the Flett Ridge. These later channel systems are interpreted as subaerial because they cut into successive delta topsets, feed base of slope fans, and are not downlapped by clinoform systems which would have been indicative of submarine conditions, Uplift and compression of the FSB during the Middle-Late Eocene and throughout the Oligo-Miocene (Boldreel & Andersen 1993; Davies et al. 2004) could explain the fall in relative sea-level, though eustatic variation cannot be ruled out especially since global sea-level fell by 5 0 - 1 0 0 m during the Eocene-Miocene (Haq et al. 1988). From our observations and interpretations, a relative sealevel history for the Palaeogene has been erected for the SE margin of the FSB (Fig. 9f). This assumes that both channel systems caused by the incision events were a result of relative sea-level fall, the first of which was probably not sufficient to expose the shelf. The first shelf margin-parallel channel system (Meocl) is therefore interpreted here as a submarine channel which may have been additionally associated with faulting and along-shelf bottom currents. The Meoc2 channels are interpreted as subaerially incised valleys, based on the evidence highlighted above.
Conclusions Fig. 8. Schematic summary of channel morphology showing the controls on their temporal and spatial variability during the Eocene. The Meoc I incision was clearly controlled by topographic relief on the Flett Ridge. The later Middle Eocene incision (Meoc2) is seen to traverse the ridge and it is believed that by this time the Ridge had subsided.
sediment supply into the Flett sub-basin could account for this observation (Fig. 9b). Sediment continued to cover the submarine ridge until Meocl times when major incision occurred, possibly during a relative sea-level fall (Fig. 9c). This incision event resulted in a broad channel system that paralleled the shelf break and followed the contours of the N E - S W trending ridge (Figs 5b & 7a). Any relative sea-level fall could have been a result of a decrease in the rate of subsidence on the basin margin when combined with a eustatic fall or the Flett Ridge could have undergone relative uplift or faulting. Shallow level igneous intrusions of plutons and associated sills and dykes on the ridge could have locally uplifted the SE margin of the FSB relative to surrounding areas causing local base level fall and incision in the early part of the Middle Eocene. The relative sea-level fall was short lived and renewed subsidence on the margin allowed later Middle Eocene sediments to prograde out across the ridge and into the basin during a relative highstand (Fig. 9d). Large clinoform systems are seen on 3D seismic data and can be traced down dip to correlate with higher amplitude, semi-discontinuous packages that are interpreted as base of slope fans on 2D data (Fig. 2). The latest Middle Eocene channels (Meoc2) cut into the delta topsets and trend perpendicular to the shelf break traversing the Flett Ridge. These Meoc2 channels are both braided and meandering and are found across the whole of the delta top at this level
The study of 3D seismic data from the margin of the FSB demonstrates its potential use in aiding the reconstruction of complex relationships between sedimentation and tectonics, and in interpreting relatively short-term changes in relative sealevel. We have shown that during the Early Palaeocene there was a N E - S W trending structural high, the Flett Ridge; a positive topographic feature that influenced sediment dispersal patterns. During the Late Palaeocene the high was onlapped and covered during a possible relative sea-level rise, or by the increase in sediment supply from the margin. This first blanketing of the ridge could have been triggered by increased subsidence on the basin margin or merely by continued sediment supply. Sediments continued to drape the margin into the Eocene and a major incision event occurred near the start of the Middle Eocene (Meocl). It is argued that the ridge had a major impact on the position of the channel systems at this time, resulting in a shelf-parallel axis of channel belts, which follow the contours of the ridge. A major period of progradation and delta development followed the early Middle Eocene (Meocl) incision, and by the end of the Middle Eocene (Meoc2) a second period of major incision took place. This incision exhibits no spatial relationship with the buried ridge and the channel systems traverse this underlying structure in a N W - S E direction, perpendicular to the shelf edge and across the ridge structure. It is argued therefore that by the end of the Middle Eocene, the Flett Ridge had little or no impact on controlling the spatial variation of the channels caused by the earlier incision. It is believed that the channel system seen at the start of the Middle Eocene incision (Meocl) event is submarine in origin, whilst the later Meoc2 channels are subaerial. From these interpretations we have produced a relative sea-level curve for this particular part of the FSB.
EVOLVING TOPOGRAPHIC CONTROL ON CHANNEL DEVELOPMENT
81
Fig. 9. Geological evolution of the Flett Ridge and the SE margin of the Faroe-Shetland Basin. The five cartoons summarize the evolution throughout the Palaeocene-Eocene. (a) Early Palaeocene (Epall) - Onlap is seen onto the Flett Ridge and a hiatus developed on the exposed high. (b) Late Palaeocene-Early Eocene (Lpal 1/2 & Eeoc 1) - Drowning of the Flett Ridge occurred and draping stratal geometries developed during a relative sea-level rise. Subsidence occurred on the flanks of the ridge. (c) Earliest Middle Eocene (Meoc 1) - At this time there was a lowering in relative sea-level and incision into the shelf. A broad channel system trending parallel to the shelf (NE) developed. See Figure 5b for seismic amplitude map of channel system. (d) Middle Eocene (Meocl-Meoc2) - A further rise in relative sea-level allowed for major delta progradation from the Shetland Margin. (e) End of the Middle Eocene (Meoc2) A second incision in to the delta tops occurred during a period of small relative sea-level falls and created small channel networks that run perpendicular to the shelf edge (NW), traversing the Flett Ridge. See Figures 6a & 6b for seismic amplitude map. (f) Relative sea-level curve constructed from the interpretations of the seismic data highlighting the incision events during major falls throughout the Palaeogene.
The authors would like to thank Amerada Hess Ltd and Kerr McGee UK Ltd for permission to show all the 3D seismic data used in this study. TotalFinaElf are thanked for allowing permission to show the 2D seismic data shown in Figure 2. Many thanks to J. Underhill and J. Smallwood for detailed and insightful reviews. Sclumberger Information Systems are thanked for the donation of their Geoquest software that was used for the interpretation of the data. AR thanks NERC for funding of the project by grant GT04/99/ES/307.
References ANDERSEN, M. S. 1988. Late Cretaceous and early Tertiary extension and volcanism around the Faroe Islands. In: MORTON, A. C. & PARSON, L. M. (eds) Early Tertiar3' Volcanism and the Opening of the NE Atlantic. Geological Society, London, Special Publications. 39, 115-122. ANDERTON, R. 1993. Sedimentation and basin evolution in the Palaeogene of the northern North Sea and the Faroe-Shetland Basins. hi: PARKER, J. R. (ed.) Petroleunl Geology of Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 31-32. BOLDREEL, L. O. & ANDERSEN, M. S. 1993. Late Palaeocene to Miocene compression in the Faroe-Rockall area. In: PARKER,J. R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 1025-1034. BURBANK, D., MEIGS, A. & BROZOVIC,N. 1996. Interactions of growing folds and coeval depositional systems. Basin Research, 8, 199- 223. CANDE, S. C. & KENT, D. V. 1995. Revised calibration of the geomagnetic polarity time-scale for the Late Cretaceous and Cenozoic. Journal of Geophysical Research, 100, 13, 917-13, 951. CLWT, P. D. 1999. The thermal impact of Palaeocene magmatic underplating in the Faroe-Shetland-Rockall region. In: FLEET, A. J. & BOLDY, S. A. R. (eds) Petroleum Geology of Northwest
Europe: Proceedings of the 5th Conference. Geological Society, London, 533-544. CLWT, P. D. & TL'RNER. J. 1998. Palaeogene igneous underplating and subsidence anomalies in the Rockall-Faroe-Shetland area. Marine and Petroleum Geology, 15. 787-800. DAVIES, R. J., CLOKE, I., CARTWRIGHT, J. A., ROBINSON, A. M. & FERRERO, C. 2004. Post Break-Up Compression of a Passive Margin and its Impact Upon Hydrocarbon Prospectivity: An Example from the Tertiary of the Faroe-Shetland Basin, UK. DEAN, K., MCLACHLAN, K. & CHAMBERS, A. 1999. Rifting and the development of the Faroe-Shetland Basin. In: FLEET, A. J, & BOLDY, S. A. R. (eds) Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference. Geological Society, London, 533-544. DUINDAM, P. & VAN HOORN. B. 1987. Structural evolution of the West Shetland continental margin. In: BROOKS, J. & GLENNIE, K. W. (eds) Petroleum Geology of NW Europe. Heyden, London, 765-773. EBDON, C. C., GRAINGER, P. J., JOHNSON, H. D. & EVANS, A. M. 1995. Early Tertiary evolution and sequence stratigraphy of the FaroeShetland Basin: implications for hydrocarbon prospectivity. In: SCRt;TTON. R. A., STOKER, M. S., SHIMMIELD,G. B. & TUDHOPE, A. W. (eds) The Tectonics. Sedimentation and Palaeoceanography of the North Atlantic Region. Geological Society, London, Special Publications, 90, 51-69. HAQ, B. U., HARDENBOL,J. & VAIL, P. R. 1988. Mesozoic and Cenozoic cbronostratigraphy and cycles of sea-level changes, h~: WILGUS, C. R., HASTINGS, B. S., Ross. C. A., POSAMENTIER, H. W., VAN WAGONER, J. C. & KENDALL,C. G. C. (eds) Sea-Level Changes: An hztegrated Approach. Society of Economic Petrologists and Mineralogists, Special Publication. 42, 71-108. HITCHEN, K. & RITCHIE. J. D. 1987. Geological review of the West Shetland Area. bt: BROOKS, J. & GLENNIE, K. W. (eds) Petroleum Geology of NW Europe. Heyden, London, 737-749.
82
A.M. ROBINSON ETAL.
HUMPHREY, N. F. & KONRAD, S. K. 2000. River incision or diversion in response to bedrock uplift. Geology, 28, 43-46. JACKSON, J., NORRIS, R. & YOUNGSON,J. 1996. The structural evolution of active fault and fold systems in central Otago, New Zealand: evidence revealed by drainage patterns. Journal of Structural Geology, 18, 217-234. KNOTT, D. S., BURCHELL, M. T., JOLLEY, E. J. & FRASER, A. J. 1993. Mesozoic to Cenozoic plate reconstructions of the North Atlantic and the tectonostratigraphic history of the UKCS Western Margin. In: PARKER, J. R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 953-974. KNOX, R. W. O'B. & MORTON, A. C. 1988. The record of early Tertiary North Atlantic volcanism in sediments of the North Sea Basin. In: MORTON, A. C. & PARSON, L. M. (eds) Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society, London, Special Publications, 39. 407-419. LAMERS, E. & CARMICHAEL,S. M. M. 1999. The Palaeocene deepwater sandstone play west of Shetland. h~: FLEET, A. J. & BO1,DY. S. A. R. (eds) Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference. Geological Society, London, 645-659. MORTON, A. C., EVANS, D.. HARLAND, R., KING, C. & RITCHIE. D. K. 1988. Volcanic ash in a cored borehole W of the Shetland Islands: evidence for Selandian (late Palaeocene) volcanism in the Faroes region, h~: MORTON, A. C. & PARSON. L. M. (eds) Earl~, Tertiary Volcanism and the Opening t ( the NE Atlantic. Geological Society, London. Special Publications. 39, 263-269. NAYLOR, P. H., BELL, B. R., JOLLEY, D. W., DURNALL, P. & FREDSTED, R, 1999. Palaeogene magmatism in the Faroe-Shetland Basin: influences on uplift history and sedimentation. In: FLEET, A. J. &
BOLDY, S, A. R. (eds) Petroleum Geologyof Northwest Europe: Ppvceedings of the 5th Conference, Geological Society, London, 545-558. RITCHIE, J. D., GATLIFF, 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. RUMPH, B., REAVES, C. M., ORANGE, V. G. & ROBINSON, D. L. 1993. Structuring and transfer zones in the Faroe Basin. In: PARKER,J. R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 999-1009. SMALLWOOO. J. R. & MARESH, J. 2002, The properties, morphology and distribution of igneous sills: modelling, borehole data and 3D seismic from the Faroe-Shetland area. b~: JOLLEY, D, W. & BELL, B. R. (eds) The North Atlantic Igneous Province: Stratigraphy, Tectonic, Volcanic and Magmatic Processes. Geological Society, London, Special Publications, 197, 271-306, VAN WAGONER,J. C., POSAMENTIER,H. W., MITCHUM,J. R., VAIL, P. R., SANG, J. F,, LOUTIT, T, S. & HARDENBOL,J. 1988. An overview of the fundamentals of sequence stratigraphy and key definitions. In: WILGUS, C. R., HASTINGS, B. S., ROSS, C. A., POSAMENTIER,H. W., VAN WAGONER. J. C, & KENDALL, C. G. C. (eds) Sea-Level Changes: An bltegrated Approach. Society of Economic Petrologists and Mineralogists, Special Publication, 42, 39-45. WHITE. R. S. 1988. A hot-spot model for early Tertiary volcanism in the NE Atlantic. hi: MORTON, A. C. & PARSON, L. M. (eds) Early Tertiary" Volcanism and the Opening of the NE Atlantic. Geological Society, London. Special Publications, 39, 3-13. ZIEGLER. P. A. 1990. Geological Atlas of Western and Central Europe. Geological Society. London.
3D seismic analysis reveals the origin of ambiguous erosional features at a major sequence boundary in the eastern North Sea: near top Oiigocene J. P. V . H A N S E N
1, O . R . C L A U S E N
l & M. HUUSE
2
1Department o f Earth Sciences, University o f Aarhus, 8000 ]trhus C, Denmark (e-mail: [email protected]) 23D-Lab, School o f Earth, Ocean and Planetary Sciences, Cardiff University, Main Building, Park Place, Cardiff" CFIO 3YE, UK
Abstract: The near top Oligocene unconformity is a major sequence boundary in the eastern North Sea Basin. It is characterized by erosional scarps below the boundary and a pronounced basinward shift in onlap above. The shift in onlap has previously been interpreted as caused by a major fall in sea level. Detailed 3D seismic analysis of a 20 by 20km area at and basinward of the uppermost Oligocene clinoform breakpoint reveals that the erosional scarps were caused by undercutting of steep clinoforms by contour-parallel currents and resulting mass wasting whilst the lowermost onlap package consists of a contour-parallel drift deposited as the erosive currents waned. The 3D seismic analysis corroborates a recent analysis based on regional 2D seismic data, which found that the erosional scarps and the geometry of the onlap sequence were indicative of a major shift in sediment input directions and not necessarily associated with any change of sea level. The paper thus demonstrates the utility of local 3D seismic analyses as a form of "ground truthing" regional basin analyses based on widely spaced 2D seismic grids.
During the Cenozoic the North Sea constituted an epicontinental sea where up to 3000 m of siliciclastic sediments were deposited contemporaneous with a net uplift of the margins (Nielsen et al. 1986; Ziegler 1990; Jensen & Schmidt 1992; Jensen & Michelsen 1992; Japsen 1998; Michelsen et al. 1998; Clausen et al. 1999). No major faulting took place during the Cenozoic, but the subsidence observed cannot be explained by thermal post rift subsidence only, and several tectonic models have been introduced in order to explain both the subsidence and the marginal uplift (e.g. Cloetingh et al. 1992; Vejba~k 1992; Japsen 1998; Gemmer et al. 2002). The post-Danian sediments in the eastern North Sea consist of marls, clays and silt with a general upward increasing grain size during Oligocene and Miocene (Kristoffersen & Bang 1982). The majority of the sediments originate from the Fennoscandian Shield (Larsen & Dinesen 1959; Ziegler 1990; Friis 1995; Clausen et al. 1999, 2000; Huuse et al. 2001; Huuse 2002). The Cenozoic succession has been subdivided into seven major sequence stratigraphic units separated by major unconformities (Michelsen et al. 1998). The most pronounced unconformities are the top Chalk (midPaleocene), top Eocene, near top Oligocene, the mid-Miocene unconformity and the base Quaternary (Huuse et al. 2001). Previous studies have shown that the clinoform breakpoint migrated through time indicating changes in sediment input directions in a clockwise direction from N to E through the latest Paleogene and Neogene (Ziegler 1990; Scrensen et al. 1997: Clausen et al. 1999, 2000; Huuse et al. 2001). Sequence stratigraphic studies have shown that the Oligocene was characterized by south and southwestward progradation (Michelsen & Danielsen 1996; Danielsen et al. 1997, Michelsen et al. 1998). In the eastern North Sea area the progradational units generated an E - W striking palaeosiope with an overall dip towards the south during the Oligocene as outlined by the time-structure map in Figure 1. Based on 2D seismic data, Huuse & Clausen (2001) mapped the morphology of the near top Oligocene and observed erosional truncations at the near top Oligocene surface. It was suggested that the surface is a submarine erosional surface with slides on the upper and middle part of the slope with onlap of possible contourites and that the major basinward shift in onlap previously interpreted as evidence for a major sea-level fall (Vail et al. 1977; Michelsen et al. 1998) was largely due to a major switch in sediment input directions (Huuse & Clausen 2001; Huuse et al. 2001: Huuse 2002).
The aim of the present study is to test previous interpretations of the origin of the erosional features and onlap units by taking advantage of the enhanced spatial resolution provided by 3D seismic data and the additional interpretational functionalities provided by workstation technology. The main functionalities used here are time/horizon slicing, attribute extraction and detailed 3D visualization of the interpreted surface. In this manner, we use 3D seismic as a local test of the validity of interpretations based on regional 2D seismic investigations and place the enhanced local understanding in the context of the regional study, thus enhancing our understanding of the Cenozoic development of the eastern North Sea Basin.
Data used The present study focused on the approximately 20 by 2 0 k m extent of the AG9801 3D survey, which is located in the Norwegian-Danish Basin at the northern rim of the Ringkc~bing-Fyn High (Fig. 1). The data were truncated at top Chalk (between 0.98 and 1.27 seconds) so the study of the deeper part of the section (primarily fault analysis) was restricted to the use of 2D seismic sections (both conventional and high resolution), which were also used for well correlation and for regional mapping. The interpreted horizons are calibrated by a number of wells located just outside the 3D survey area (Ibenholt-1, R-1, Ida- 1 and Inez- 1: Fig. 1). The vertical resolution, defined as a quarter of the dominant wavelength (M4) of the 3D survey is approximately 10m assuming an average velocity of the Cenozoic of 2 km/s (cf. Huuse et al. 2001 ). The resolution is thus higher than that of the conventional 2D seismic data (k/4 - 15-20 m), but less than the resolution the high-resolution 2D seismic data ( - - 5 - 1 0 m ) (Huuse et al. 2001) (Fig. 2). The main advantage of the 3D seismic data lies in their lateral resolution which is equivalent to a few tens of metres, i.e. much higher than that of even the densest 2D seismic grids regardless of the nominal vertical resolution.
Seismic characteristics of the near top Oligocene surface In most of the survey area, the near top Oligocene surface is characterized by a continuous, high-amplitude positive reflection (Fig. 3). A positive reflection is here defined as a reflection
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. 83-89. 0435-4052/04/$15 9 The Geological Society of London 2004.
84
J.P.V. HANSEN ETAL.
Fig. 1. The location of the AG9801 3D seismic survey at the northern rim of the Ringk~bing-Fyn High superimposed on a pre-Zechstein structure map (from Vejba~k & Britze 1994). The survey is located in the Norwegian-Danish Basin, 25 km south of the Norwegian Danish border. The close-up of the study area is a time/structure map of the near top Oligocene unconformity (from Huuse & Clausen 2001) showing the sub-regional context of the 3D seismic survey area. The extent of the AG9801survey is approximately 20 by 20 km
from a downward increase in acoustic impedance between two layers and is shown as a white reflection (trough) on vertical sections. The 3D seismic data were processed to zero phase. The data quality is generally high, although the presence of a buried Quaternary valley and shallow gas anomalies in the overburden locally disturb the seismic image. The near top Oligocene surface is recognized as a marked onlap surface separating rather parallel downlapping clinoforms below from sub-parallel and clinoform geometries above. The surface is characterized by erosional truncations and a major basinward shift in onlap. The underlying reflection pattern is
caused by the presence of a major progradational system, which built out from NNE to SSW during the Oligocene (Danielsen et al. 1997: Huuse & Clausen 2001). The overlying onlap reflections are oblique, distal onlaps resulting from the distal pinch-out of a SW prograding system and should not be mistaken for proximal or coastal onlap (cf. Vail et al. 1977: their fig. 12: Michelsen et al. 1998: their fig. 4; Huuse & Clausen 2001: their figs 9, 10).
Observations The following geological features have been seen in connection with the near top Oligocene surface: complex faulting, sliding, submarine incision, submarine channels and contour-parallel deposits.
Faulting
The near top Oligocene surface is offset by a number of faults within the study area, most of them detaching in the underlying Zechstein salt. The faults roughly strike N W - S E , dipping to the SW and NE, creating a complex fault pattern with ramp systems and interaction of antithetic faults (Fig. 4). In order to determine the timing of fault activity, we interpreted two reflections, one above (H1) and one below (top Chalk) the near top Oligocene surface (Fig. 5). These were used to construct time-thickness maps of the over- and underlying successions. Thickness variations across the faults on these maps are used to indicate fault movement during the equivalent time interval. The maps and seismic back stripping show that only a few of the faults were active prior to the development of the near top Oligocene unconformity. The active faults are located in the western- and the central part of the survey (Fig. 5). On a larger scale, the major salt detachment faults of the region were active in pulses throughout the Cenozoic.
Slides and incisions
Fig. 2. Vertical resolution of the different seismic data used illustrated by wiggle trace displays. The figure shows conventional 2D seismic lines ('Conv. seis.'), AG9801, 3D seismic ('3D seis.') and highresolution 2D seismic ('H.R. seis.').
The near top Oligocene surface is characterized by two distinct slopes: slope l (SL l) to the north striking W S W - E N E and slope 2 (SL2) in the centre of the survey striking W N W - E S E (Fig. 4). The northern slope (SLl) rises to a height of 7 0 - 1 0 0 m with a dip of about 2 - 3 ~ locally reaching angles up to 5 ~ Seismic lines perpendicular to the slope-contours of SL1 show well-defined
ORIGIN OF EROSIONAL FEATURES
85
Fig. 3. Near top Oligocene reflection is characterized by a continuous reflection from a downward increase in acoustic impedance. In the study area it is defined by truncations below and onlap above. Location is shown on Figure 4.
evidence for mass wasting, such as erosional scarps and slide blocks (Figs 3 & 6). The slides at SL1 are seen as mass m o v e m e n t along a concave upward sliding plane (Fig. 6) similar to slide planes defined by W o o d c o c k (1979) and Stow et al. (1996). These slides are most well defined in the western part of the survey where they are seen as steps in down-slope direction on 3D displays (Fig. 4).
Fig. 4. Time-structure map of the near top Oligocene unconformity as plan view and 3D perspective view from the south. The two maps illustrate the complex fault pattern in the study area showing graben and ramp systems of salt detaching faults in the northern part of the survey area. Two slopes are seen, one to the north (SL1) striking WSW-ENE and one in the centre (SL2) striking WNW-ESE. Slide scarps are seen as steps on SL1 on the 3D display.
Fault analysis indicates that the fault adjacent to the slope was less active during generation of the unconformity. However, fault activity appears to have increased in the period following the generation of the near top Oligocene and it is hard to define the precise onset of fault activity. Even though fault activity cannot be excluded as a possible trigger of mass wasting, the main factors causing the slide events are more likely connected with other factors such as oversteepened upper Oligocene
86
J. P, V. HANSEN ETAL.
Fig. 5. (a)(b) Isochore maps of the successions above and below near top Oligocene, used to indicate the timing of fault activity. The black lines illustrate the active faults. (c) The two horizons (top Chalk and H1), which are used for constructing isochore maps are indicated on the seismic line. The maps show that most of the fault activity can be dated to post-date the generation of the near top Oligocene. clinoforms and undercutting of these by contour currents. The near top Oligocene clinoform breakpoint was exposed for a period exceeding 1 Ma (Huuse & Clausen 2001) and it is thus likely that instabilities caused by the above-mentioned factors could have been affected by changes in sea level. Just to the west of the centre of the survey erosional truncations (incisions) are observed at the near top Oligocene surface. The erosional scarps constitute a minor slope (SL2) striking W N W to ESE. The incisions are approximately 1 - 2 km wide and 4 0 - 5 0 m deep and are very distinct on seismic sections
with marked truncations and onlaps (Fig. 7). The seismic data give no direct explanation to whether the incisions are eroded by bottom-currents or mass-wasting processes such as sliding. The SL2 is situated just to the east of a fault that seems to be active both in the. period before and after the generation of the near top Oligocene, and the outline of the incisions and the thult strike seem to be parallel, indicating connection between the slides and the fault. We suggest that the incisions outline an overall area of sliding and redistribution of sediment to the southwest of the underlying succession triggered by fault activity on a background of intense contour parallel bottom current activity
Channels
Fig. 6. Slide block with concave upwards glide plane beneath SL1. Line location shown on Figure 4.
At the foot of SL 1, several semi-circular features are seen on the near top Oligocene time/structure map (Fig. 4). These features are also seen at on an RMS amplitude map made in a 3 0 m s window around the near top Oligocene (_+ 15 ms) as sinuous features in a seismically disturbed area at the foot of SL 1 (Fig. 8). The features are interpreted as a channel system running parallel to the slope contours. The channel outline is often very dim on seismic lines and it is disrupted by noise caused by a Quaternary valley in the overburden. When the channel is observed on seismic, it is almost U-shaped, around 500 m wide and cut 40 m into the underlying succession (Fig. 8). The slump features seen on SLI might have disturbed the channel outline/pattern by erosion and redistribution of the channel sediments. This reorganization of the channel bed could explain why it is seen as semi circular features on the time structure map and not as a clear channel system. There is no clear evidence for subaerial exposure, like major valley incision perpendicular to the clinoform contours of the
Fig. 7. Incisions into the near top Oligocene surface with marked truncations and onlaps. The incisions are approximately 1 - 2 km wide and 4 0 - 5 0 m deep.
Fig. 8. (a) Channel system seen on amplitude attribute map in a 30 ms window around the near top Oligocene (_+ 15 ms). (b) Seismic cross section illustrating the geometry of the channel (vertical exaggeration - 30:1).
Fig. 9. Deposits onlapping the near top Oligocene at the base of SL1 constitute contour parallel deposits. The seismic line is flattened to H1 in order to enhance the interpreted original geometry.
88
J . P . v . HANSEN ETAL.
near top Oligocene (Huuse & Clausen 2001) and the channel system is located at the base of the slope (SL1), 1 0 0 - 2 0 0 m below the clinoform break point. This suggests that the channel was generated by a contour parallel bottom current, powerful enough to create a meander-like channel pattern at the foot of the slope.
Contourites
The overlying reflections onlap the near top Oligocene surface from the SSE. The first 4 - 5 reflections bow down in a downward concave pattern towards the near top Oligocene reflection (Fig. 9). The accommodation space between SL1 and the concave geometry of the overlying succession indicates that the onlap units were deposited by a waning contour current. The geometry of the onlap unit is similar to contour parallel sedimentary bodies (contourites) deposited by contour parallel bottom currents (Faug~res et al. 1999: Stow et al, 1996). Whether it is a contourite drift (sensu Stow et al. 1996: Stow et al. 1998; Stow & Mayall 2000) is difficult to determine, since there are no precise estimates of palaeo-water depth, which could range from tens to several hundred metres (e.g. Huuse & Clausen 2001). Also, there is no borehole control on the sediments that make up the drift unit. The geometry of the sedimentary volume is, however, very similar to those described as giant elongated contourite drifts some tens to hundreds of kilometres long, tens of kilometres wide and 0.1-1 km in relief (Faug~res et al. 1999: Stow et al. 1996; van Weering et al. 1998).
Discussion and conclusions Palaeotopographic reconstructions (Fig. 10) show the outbuilding of clinoforms during the Oligocene approximately from north to south with only minor fault activity and no indications of a pronounced palaeo-current, sliding or incision prior to the generation of the near top Oligocene unconformity (Fig. 10a) Towards the end of the Oligocene (c. 25Ma) a major decrease in sediment input to the study area (Fig. 10b) is associated with a major shift in sediment input direction from north to northeast and the major influence on the morphology of the surface seems to be the development of a contour parallel bottom current. The current is cutting into the slope (SL1), thus facilitating sliding, and also cuts into the fore-slope area generating a channel system that runs parallel to the contours. Sliding and incision of the second major slope (SL2) was most likely related to the interaction of fault activity in association with the contour-parallel bottom current. Following the erosional sculpting of the near top Oligocene renewed sedimentation of the area began with the development of an elongated contourite drift striking N E - S W . The drift is finally covered with passive infill of the moat between the elongated drift and the SL1 slope. There is no evidence for undercutting, slumping or incision in the area in this period. The 3D seismic analysis of the near top Oligocene within the AG9801 survey supports previous interpretations of its origin based on regional 2D data. In addition the 3D study has given a much higher degree of detail and understanding of the erosional and depositional processes in the system. It has produced observations impossible to obtain from only 2D coverage like the geometry and morphology of the geographically localized SL2 slope and the contour parallel channel system. It is also evident that the marked downward shift in onlap seen at the near top Oligocene is an effect of a major change in the direction of sediment input and not indicative of a major fall in sea level as previously hypothesized. The results gained via 3D seismic
Fig. 10. (a) Palaeotopographic reconstruction prior to the development of the erosional features of near top Oligocene, showing the overall direction of progradation. (b) Palaeotopographic reconstruction at the time of the generation of near top Oligocene unconformity showing undercutting of SL1 resulting in mass wasting at the slope. Fault activity and bottom current incision at SL2. (c) The onlap of near top Oligocene starts with bottom current controlled deposition parallel to the slope contours and passive infilling of the depression between SL2 and the contourite mound. analysis thus has direct implications for our understanding of regional basin development. We thank the University of Aarhus and The Danish Energy Agency for financing, DONG and Norsk Agip for access to 3D data, and GEUS, Danpec. TGS-Nopec, Geoteam, Fugro and Ma~rsk Oil Gas AS for access to 2D seismic data. Reviewers J. Cartwright and J. Smallwood are thanked for very helpful reviews.
ORIGIN OF EROSIONAL FEATURES
References CLAUSEN, O. R., GREGERSEN, U., MICHELSEN, O. & SORENSEN,J. 1999. Factors controlling the Cenozoic sequence development in the eastern parts of the North Sea. Journal of the Geological Societ)'. London, 156, 809-816. CLAUSEN, O. R., NIELSEN, O. B., HUUSE, M. & MICHELSEN, O. 2000. Geological indications for a Palaeogene onset of the "Neogene uplift" in the eastern North Sea area. Global and Planetar).' Change. 24, 175-187. CLOETJNGH, S., REEMST, P., Kool, H. & FANAVOOL, S, 1992. Intraplate stresses and the post-Cretaceous uplift and subsidence in northern Atlantic basins. Norsk Geologisk Tidsskrift, 72, 229-235. DANIELSEN, M., MICHELSEN, O. & CLAUSEN, O, R. 1997. Oligocene sequence stratigraphy and basin development in the Danish North Sea sector based on log interpretations. Marine and Petroleum Geology, 14, 931-951. FAUGERES, J. C., STOW, D. A. V., IMBERT, P. & VIANA, A. 1999. Seismic features diagnostic of contourite drifts. Marine Geology. 162, 1-38. FRnS, H. 1995. Neogene aflejringer. In: NIELSEN, O. B. (ed.) Danmarks geologi fra Kridt' til i dag. Aarkus Geokompendier, 1, 115-127. GEMMER, L., NIELSEN, S. B. & LYKKE-ANDERSEN,H. 2002. Differential vertical movements in the eastern North Sea area from 3-D thermomechanical finite element modelling. Bulletin of the Geological Society of Denmark, 49, 119-144. HUUSE, M. 2002. Late Cenozoic palaeogeography of the eastern North Sea Basin: climatic vs. tectonic forcing of basin margin uplift and deltaic progradation. Bulletin of the Geological Society of Denmark, 49, 145-169. HUUSE, M. & CLAUSEN, O. R. 2001. Morphology and origin of major Cenozoic sequence boundaries in the eastern Danish North Sea Basin: top Eocene, near top Oligocene and the mid-Miocene unconformity. Basin Research, 13, 17-41. HUUSE, M., LYKKE-ANDERSEN, H. & MJCHELSEN, O. 2001. Cenozoic evolution of the eastern Danish North Sea. Marine Geology, 177, 243-269. JAPSEN, P. 1998. Regional velocity-depth anomalies, North Sea Chalk: a record of overpressure and Neogene uplift and erosion. AAPG Bulletin, 82, 2031 - 2074. JENSEN, L. N. & MICHELSEN, O. 1992. Terti~er h~evning og erosion i Skagerrak, Nordjylland og Kattegat. Dansk Geologisk Forening, Arsskrift for 1990-91, 159-168. JENSEN, L. N. & SCHMIDT,B. J. 1992. Late Tertiary uplift and erosion in the Skagerrak area: magnitude and consequences. Norsk Geologisk Tidsskrift, 72, 275-279. KRISTOFFERSEN, F. N. & BANG, I. 1982. Cenozioc excl. Danian limestone. In: MICHELSEN, O. (ed.) Geology of the Danish Central Graben. Danmarks Geologiske Unders0gelse, Series B. 8, 61-70.
89
LARSEN. G. & DINESEN. A. 1959. Vejle Fjord Formationen ved Brejning. Danmarks Geologiske Unders0gelse, R~ekke II 82. MICHELSEN, O. & DANIELSEN, M. 1996. Sequence and systems tract interpretation of the epicontinental Oligocene deposits in the Danish North sea Sector. In: DE BATIST, M. & JACOBS, P. (eds) Geology of Siliciclasitc Shelf Seas. Geological Society, London, Special Publications, 117, 1-13. MICHELSEN, O., THOMSEN, E., DANIELSEN, M., HEILMANN-CLAUSEN, C., JORDT, H. & LAUERSEN, G. V. 1998. Cenozoic sequence stratigraphy in eastern North sea. In: GRACIANSKY, P. C., HARDENBOL, J., JACQUIN, T. & VAIL, P. R. (eds) Mesozoic-Cenozoic Sequence Stratigraphy of Western European Basins, SEPM Special Publications, 60. 91 - 118. NmLSEN, O. B,, SOREYSEN,S., THIEDE, J. & SKARBO,O. 1986. Cenozoic differential subsidence of North Sea. AAPG Bulletin, 70, 276-298. SORENSEN, J. C., GREGERSEN, U., BREINER, M. & MICHELSEN, O. 1997. High-frequency sequence stratigraphy of Upper Cenozoic deposits in the central and southeastern North Sea areas. Marine and Petroleum Geology, 14, 99-123. STOW, D. A. V. & MAYALL, M. 2000. Deep-water sedimentary systems: New models for the 21 st century. Marine and Petroleum geology, 17, 125-135. STOW, D. A. V., READING, H. G. & COLLINSON,J. D. 1996. Deep seas. In: READING, H. G. (ed.) Sedimentary Environments. Processes, Facies and Stratigraphy. Blackwell, London, 395-453. STOW, D. A. V., FAUGI~RES,J. C., VIANA, A. & GONTHIER, E. G. 1998. Fossil contourites: a critical review. Sedimentary Geology, 115, 3-31. VAIL, 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. VAN WEERING, T. C. E., NIELSEN. T.. KENYON, N. H., AKENTIEVAD. K. & KU1JPERS. A. H. 1998. Sediments and sedimentation at the NE Faroe continental margin; contourites and large-scale sliding. Marine Geology, 152, 159-176. VEJBa~K, O. V. 1992. Geodynamic modelling of the Danish Central Trough. In: LARSEN. R. M.. BREKKE, H., LARSEN, B. T. & TALLERAAS, E. (eds) Structural and Tectonic Modelling and its Application to Petroleum Geology. Elsevier, Amsterdam, Norwegian Petroleum Society. Special Publication No. 1, 1-17. VEJB,,EK. O. V. & BRITZE, P. 1994. Geological map of Denmark; Top pre-Zechstein (two-way travel time and depth). Map series no. 45. Geological Survey of Denmark (GEUS). Copenhagen, Denmark. WOODCOCK, N. H. 1979. The use of slump structures as palaeoslope orientation estimators. Sedimentology, 26, 83-99. ZIEGLER, P. a. 1990. Geological Atlas of Western and Central Europe. Shell Internationale Petroleum Maatschappij BV, The Hague.
3D seismic interpretation of the Messinian Unconformity in the Valencia Basin, Spain JOSE
FREY
MARTINEZ
l, J O S E P H JOSE
A. C A R T W R I G H T VICENTE
BRAVO
l, P E T E R
M. BURGESS
1 &
2
13DLab, School o f Earth, Ocean and Planeta 9 Sciences, Cardiff Universi~.', Main Building, Park Place, Cardiff CFIO 3YE, UK (e-mail: j o s e @ o c e a n . c f a c . u k ) 2Repsol-YPF. Paseo de la Castellana 280, 4 ~ 28046, Madrid, Spain
Abstract: This paper documents the complex three-dimensional geometry of the Messinian Unconformity in the Tarraco concession area on the Ebro Shelf, based on the Tortuga 3D seismic survey. A detailed map of the Messinian Unconformity Surface has been constructed in the survey area, and shows a dendritic drainage pattern with valleys that are around 1km wide, and 6-7 km in length. The overall morphology of the surface is strongly reminiscent of badlands topography, from semi-arid zone erosion of sandstone-shale sequences. The depth of incision of the valleys into the pre-Messinian clastic sequences is around 400 m. The results obtained confirm previous models of sub-aerial exposure creating erosion of the preMessinian shelf and slope sequences that generated the spectacular Messinian Unconformity.
The first drilling campaign (Leg 13) of the Glomar Challenger in 1970 revealed the existence of an extensive Late Miocene erosional surface over most of the Mediterranean continental margins, and a more than 1.5km thick sequence of Upper Miocene evaporites including salts in the deeper basins (Ryan 1973). It is now well-established that the Mediterranean Sea became isolated from the Atlantic Ocean during the Messinian Stage of the Late Miocene between about six and five million years ago. Lack of marine recharge combined with high evaporation rates produced a lowering of water level below that of the open ocean leading in turn to increased aridity in the region. This resulted in accumulation of salt within deep depressions that dried out from excess evaporation and in the formation of extensive erosive subaerial surfaces over most of the continental margin (Maldonado & Nelson 1990). Deeper areas of the Mediterranean and marginal basins (e.g. Sorbas Basin) were transformed into a series of large hyper-saline lakes in which precipitated a thick and extensive sequence of evaporites including gypsum, halite and other salts, that exceeded one million cubic kilometres in volume (Hsfi et al. 1978; Stampfli & Hrcker 1989; Maldonado & Nelson 1990). The history of desiccation of the isolated Mediterranean basins is complicated and major evaporitic deposition occurred in two main phases with distinct paleoenvironmental significances (Hsfi et al. 1978). Whether these evaporites were deposited in a deep Mediterranean basin with a bathymetry that strongly resembles that of today, or were deposited in a shallow basin, and whether the sea level was maintained at or dropped below of the world ocean, it required a restricted, shallow portal between the Mediterranean and the Atlantic Ocean during Messinian time, perhaps influenced by the tectonic closure of the Betic and Rif straits during the Late Miocene (Ryan 1973). Geological evidence such as Messinian outcrops, distribution of erosional surfaces imaged in the subsurface, the nature of evaporitic strata and biological response to progressive isolation, have been investigated in numerous studies to allow a reconstruction of the Mediterranean basin paleogeography during Messinian times. Late Miocene erosional unconformities are visible in seismic reflection profiles of numerous Mediterranean basins. They have been described by Ryan & Cita (1978) and have been interpreted as the product of subaerial erosion. In addition, more subaerial erosional features in deep marginal areas (desiccation cracks, stromatolite layering and fossil drainage systems) have been described over the entire Mediterranean basin by Ryan & Cita (1978), reinforcing the
theory of subaerial conditions in some areas of the Mediterranean during the Messinian. In this paper 3D seismic and well data from the Tortuga 3D survey, offshore Tarragona (Spanish Mediterranean), have been used to investigate the Messinian unconformity on the Ebro continental margin (Fig. 1). The Tortuga 3D Survey was acquired and processed in 1997 by Repsol. This is the most recent and also the largest of the 3D surveys in the area and has an inline and crossline interval of 12.5 m. This study builds on the earlier interpretation of another 3D seismic survey by Stampfli & Hrcker (1989) located 50km further north on the margin of the Ebro Delta. By means of seismic interpretation calibrated with well log data, we provide more detailed description of the key structures of the sequence stratigraphy that permit the evaluation of both the impact of the sea-level drop on the Ebro continental margin and the paleogeography that resulted from the erosion of the pre-Messinian Castellon shelf.
Geological setting The Ebro continental margin, offshore from the present Ebro Delta, was developed within the southern (Mesozoic-Cenozoic) Alpine megasuture system (Fontbot6 et al. 1986), and is related to the development of the Valencia Trough. The Valencia Trough has been defined as an aborted rift system that became active at the end of the Oligocene and the beginning of the Miocene following the Alpine orogeny (Biju-Duval et al. 1978; Mauffret et al. 1981: Dafiobeitia et aL 1990). The rift underwent thermal and tectonic subsidence, but failed before oceanic crust was emplaced. The opening of the trough produced a series of tectonic grabens bounded by faults that strike parallel to the Iberian continental margin (Fig. 2). Stratal patterns on the northwestern margin in the Valencia trough were controlled by this tectonic setting (Julivert et al. 1972; Stoeckinger 1976; Soler et al. 1983: Medialdea et al. 1986). The general evolution of the Valencia trough continental margin during the Miocene post-rift period was characterized by the subsidence of the basin and the deposition of a thick sedimentary cover. Lower to Middle Miocene marine deposits filled structural depressions, draping the pre-existing topography (Dafiobeitia et al. 1990) (Fig. 3). During the Late Miocene, the Messinian 'salinity crisis" resulted in an extensive unconformity in the margins and in deposition of evaporites over the former abyssal plain (Montadert et al. 1970; Hsfi et al. 1978). The deposition of the post-Messinian marine sequence
DAVIES,R. J., CARTWR1GHT,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, 91 - 100. 0435-4052/04/$15 9 The Geological Society of London 2004.
92
J. FREY-MARTINEZ E T AL.
Fig. 1. Location map of study area. Well locations inset. Note locations for Figures 4 and 5.
took place since the Early Pliocene, with up to 80km of progradation and deposition of 1200-2500 m of sediment over the Messinian surface (Alonso et al. 1990) (Fig. 3).
Seismic stratigraphy
Fig. 2. Sketch of the northeastern Iberian Peninsula and surrounding offshore areas in which main Cainozoic structures and basins are reported. After Roca et aL (1999).
Fig. 3. Generalized cross section of the tectonic setting and post-Messinian growth patterns of the Ebro margin. After Nelson (1990).
The stratigraphy of the study area is described with reference to two key seismic profiles (Figs 4 & 5). A seismic stratigraphic framework has been established by tying the 3D seismic interpretation with fourteen interpreted well-logs suites from wells in the area (Fig. 1). Two megasequences have been identified and mapped over the Tortuga 3D Area: Megasequences A and B (they correspond to MA and MB respectively in Figs 4 & 5). Megasequence A and B are separated by a major erosional unconformity, characterized by truncations of the reflectors beneath, and onlap onto this surface from above. This unconformity is interpreted as the Messinian Unconformity Surface (MUS) (Figs 4 & 5). The geometry of Messinian Unconformity Surface is described in detail in a later section.
3D SEISMIC INTERPRETATION OF THE MESSINIAN UNCONFORMITY
93
Fig. 4. NW-SE seismic section trough the Tortuga 3D Area. This seismic profile illustrates Megasequence A (MA) and Megasequence B (MB). Megasequence A shows clinoform geometry with medium-angle and oblique-tangential progradational reflectors that become steeper throughout the foreset and the bottomset (A2). Topset strata consist of parallel, continuous and horizontal high amplitudes reflections (A 1). For section line see Figures 4a and 4b. Megasequence B shows aggradational configurations in the lower part and progradational configurations in the upper part. For seismic section see Figures 4c and 4d. B2 and B 1 correspond to B2 Unit and B [ Unit respectively. The A Surface marks their boundary. Near the palaeo-shelfbreak, clinoforms are interbedded with packages of low and transparent with erosive concave-up planes in the basal surfaces. For seismic section see Figure 4d. Wells mark the seismic profile location.
Megasequence A Megasequence A is severely truncated at the Messinian Unconformity Surface, but where it is best preserved it is characterized by a progradational reflection configuration. Clinoform geometry shows medium-angle and oblique-tangential progradational reflectors that become steeper throughout the foresets and the bottomsets where the reflection amplitudes are generally lower (Fig. 4a). Throughout the study area, the topset strata consist of parallel, continuous, horizontal and high amplitudes reflections
(Figs 4a & 4b). These reflectors exhibit aggradational reflection configurations. The uppermost reflections in this aggradational unit are truncated by the overlying Messinian Unconformity Surface (Figs 4a & 4b). The lowermost set of reflections in this unit form a toplap relationship with the underlying clinoforms (Fig, 4a). All the wells of the study area penetrated Megasequence A. However, none of them penetrated the entire clinoform system. From petrophysical data and completion reports of the wells, two lithostratigraphic formations are distinguished: the Castellon Shale Formation and the Castellon Sandstone
94
J. FREY-MARTINEZ ET AL.
Fig. 5. NNE-SSE seismic section across the study area, MA and MB correspond to Megasequence A and Megasequence B respectively. Note the markedly erosional character of the Messinian Unconformity Surface (MUS) separating them. A2 and A I correspond to the Castellon Shale Formation and the Castellon Sandstone Formation respectively. Surface B marks the pass from undulated and laterally limited high amplitude reflections (A2) to parallel, continuous and horizontal reflections (A 1). B2 and B 1 correspond to B2 Unit and B ! Unit respectively. The A Surface marks their boundary. Formation. They correspond to A2 and A1, respectively, in Figures 4 and 5. The Castellon Shale Formation comprises mainly grey-brown clays and firmer shales interbedded with locally glauconitic argillaceous siltstones and sandstones. The Castellon Sandstone Formation consists of interbedded marls and sandstones. The Castellon Shale Formation yielded several samples of rich and diverse foraminiferal assemblages, containing both planktonic and benthonic species. With regards to the Castellon Sandstone Formation, the biostratigraphy is based on the highest occurrence of the planktonic marker GIoborotalia merotumida and common specimens of the benthonic species Ammonia pinctatogranosa. No zonal markers are encountered until Globorotalia continuosa, which is taken to indicate the top of the Tortonian. The Castellon Shale Formation has been interpreted as the product of deposition in a low-energy deep marine environment. During the Middle Miocene, active uplift in compressional fold belts around the western Mediterranean provided a large siliciclastic sediment supply to the basin, favouring the development of coarse-grained deposits in nearshore environments and coral reef platforms in areas of lower siliciclastic input (Mauffret et al. 1981; Rehault et al. 1985). Increasing input of coarse-grained sediment and consequent delta progradation is reflected in the gradational boundary between the Castellon Shale and the Castellon Sandstone. This coarsening-upwards trend is continued in the Castellon Sandstone Formation, interpreted as deposited in shallow marine and deltaic conditions. Delta progradation reflected a balance between sediment supply and creation of topset accommodation space during relative sea-level highstand. In summary, Megasequence A is interpreted as a prograding and aggrading deltaic depositional system. The onset of this progradation system is not constrained by well data, but it is at least Tortonian in age, and this progradation most probably continued up to the time when the development of the Miocene unconformity began.
Megasequence B
Megasequence B represents the entire post-Messinian succession of the Ebro margin. It appears as a largely progradational and aggradational slope wedge of predominantly fine-grained sediment reaching up to 1500 m in thickness. Megasequence B is divided into two main units based on their seismicstratigraphic expression: the B I Unit and the B2 Unit (Figs 4 & 5). The B I Unit is characterized by strongly progradational reflection configurations. The B2 Unit exhibits clearly aggradational seismic patterns within the survey area. According to petrophysical data and completion reports, two lithostratigraphic formations are distinguished within Megasequence B: the Ebro Shale Formation and the Ebro Sandstone Formation. However, lack of accurate dating does not allow clear recognition of the boundary between them, and makes correlation of the lithostratigraphic markers with the seismic stratigraphy highly problematic.
The B2 Unit. The B2 Unit of Megasequence B consists of an approximately 700 m thick accumulation of very fine clastic sediments according to well log data. The B2 Unit is found between 725ms (TWT) and 1400ms (TWT) in the northwestern part of the study area, and between 800ms (TWT) and 1650ms (TWT) in the southeastern part of the survey set (Fig. 4). Unit B2 exhibits a relatively high frequency of continuous, medium amplitude reflections alternating with more transparent or opaque, low-amplitude packages. The reflections within the lower part of the B2 Unit infill the Messinian valleys onlapping the Messinian Unconformity Surface (Figs 4a & 4b). Outside the valleys, the reflections of the lower part of Unit B2 are continuous and exhibit an undulating geometry where lows are coincident with the Messinian valleys and highs are coincident with the positive relief between valleys (Fig. 4b). The bottomsets of the B2 Unit
3D SEISMIC INTERPRETATION OF THE MESSINIAN UNCONFORMITY consist of continuous reflections with low and medium amplitudes, which are parallel to the Messinian Unconformity Surface (Fig. 4c). These reflections are characterized by medium frequency and dominantly aggradational and progradational configurations. Profiles in the dip direction clearly exhibit clinoform geometries in the upper parts of the B2 Unit. These clinoforms are characterized by high-angle sigmoid-oblique reflection configurations (Figs 4 & 4d). Downdip of the depositional shelf break in the upper slope position of the clinoforms, the continuity of the clinoform reflections are disturbed by intervals with a chaotic seismic facies of low to moderate amplitudes that have erosive concave-up contacts with the more continuous clinoform reflections. These features are arrowed in Figure 4d. Profiles in the strike direction display laterally discontinuous undulating reflectors (Fig. 5). These reflectors are parallel to subparallel and are grouped in lenticular reflector packages that create an upward aggrading system of erosive features and their partially excavated fills that are interpreted as submarine canyons (Fig. 5). The B2 Unit is interpreted as a dominantly progradational stage in the evolution of the Ebro Delta, occurring between the Messinian and the late Pliocene. The first stages of deposition of the B2 Unit took place throughout an extensive relative sea level highstand that is presumed to have occurred during the refilling of the Mediterranean after the Messinian event. A substantial flux of sediment was transported into the basin in the early Pliocene (Maldonado & Nelson 1990). This resulted in deepwater clay and mass-flow deposition that onlapped and locally draped the Messinian Unconformity Surface. During this stage, deposition is thought to have been controlled by tectonic activity and the topography of the Messinian Unconformity Surface (Field & Gardner 1990). Deposition of this deep-water sediment over the Messinian valleys indicates a rapid sea-level rise after the post-Messinian opening of the Strait of Gibraltar (Ryan & Cita 1978; Stampfli & Hrcker 1989). Towards the late Pliocene, sediment supply was sufficiently high to allow the Ebro deltaic system to prograde to a position immediately west of the study area. Consequently, a dominantly mud-prone sediment load was delivered to the pro-deltaic position occupied by the study area. Sediment was distributed through several poorly developed submarine canyons that crossed the continental margin and fed the more distal areas. This probably resulted in the creation of slope channel-levee complex wedges and base-of-slope aprons in the deep water system outside the limits of the Tortuga 3D survey area. The presence of chaotic packages in the upper foreset position of many of the clinoforms is interpreted as the result of slope instability processes (delta front instabilities) due to differential compaction of the underlying sediments over the highly irregular topography of the Messinian Unconformity. These factors and the high input of sediment into the basin might have resulted in oversteepening of the slopes causing cyclic, repeated episodes of local collapses mainly focused on the inner areas, where the topography of the Messinian valleys is greatest, and, consequently, the degree of differential compaction maximized.
The B1 Unit, The B 1 Unit of Megasequence B consists of a 600m thick accumulation of fine clastic sediments, which include carbonate-rich plastic clays interbedded with sand and silt layers. The B 1 Unit is best developed at 200-650 ms (TWT) in the western part of the survey area and at 200- 800 ms (TWT) in the eastern part of the study area (Fig. 4). It consists of numerous moderate amplitude reflections alternating with continuous, high-amplitude reflections and locally developed chaotic packages (Fig. 4). The B1 Unit is characterized by strongly aggradational reflection configurations. Clinoform geometries cannot be seen within the confines of the survey
95
area, except to a limited extent on its SE limit, where a series of clinoform break points are visible (Fig. 4). The B1 Unit is interpreted as a dominantly aggradational stage in the evolution of the Ebro Delta, which occurred between late Pliocene and present day. During this stage, a large influx of sediments extended the Ebro Delta across the study area. The amount of aggradation during this interval is approximately 200m, suggesting a modest relative sea level rise during the progradation (20km across the study area). This increase in sediment supply, which resulted in a doubling of sediment yield from the Ebro drainage area, has been attributed to climatic changes (glaciation and deforestation) by Nelson (1990). It has, however, been suggested that the Castellon delta system was fed by smaller rivers (including a proto-Ebro) draining the Catalan Coastal Ranges (Elders, pers. comm. 2003). In this alternative view, the proto-Ebro is considered to be separated from the Ebro Basin, which was an isolated basin of internal drainage unconnected from the sea by the Catalan Coastal Ranges (Coney et al. 1996). It is thought that the Messinian Sea level drop allowed the Ebro to cut rapidly headwards through the Catalan Coastal Ranges, creating the present day antecedent drainage pattern that was imposed on the buried Pyrenees and Ebro Basin, dramatically increasing the sediment supply to the Ebro Delta (Elders, pets. comm. 2003).
The Messinian Unconformity Surface The seismic expression of the Messinian Unconformity Surface in the study area is described with reference to three seismic sections (Figs 4, 5 & 6) and three key maps (Figs 7, 8 & 9). A seismic framework of the Messinian Unconformity Surface has been established for this study by using two different approaches. The first approach consisted of a combination of seismic and lithostratigraphic analysis. This was undertaken in order to differentiate the erosional and depositional features generated during the Messinian event. The second approach consisted of a comparison between the morphology of the Messinian Unconformity Surface and the distribution of amplitude anomalies associated with the surface, either in its immediate subcrop or in the earliest infill units. This was done primarily in order to investigate the nature and distribution of the valley-filling strata of the Messinian valleys. The vertical seismic sections show that the Messinian Unconformity Surface is not a continuous reflection of unique
Fig. 6. Seismic section through a valley on the Messinian Unconformity. MUS corresponds to the Messinian Unconformity Surface and the l-Reflector represents the first infill of the Messinian valleys. Note the different seismic response of both reflectors. Well control shows a decrease in clay content in the I-Reflector.
96
J. FREY-MARTINEZ ET AL.
Fig. 7. A structural contour map (TWT) of the Messinian Unconformity Surface. Note the intense erosive patterns and the dendritic valleys dissecting interttuve areas and merging into larger channels in a N W - S E direction. Two or possibly three orders of stream branching can be distinguished.
Fig. 8. A display of the Messinian Unconformity Surface time/structure map on GeoViz (Schlumberger). Does not indicate absolute depth. Note the intense erosive patterns and the dendritic character of the Messinian Unconformity Surface.
3D SEISMIC INTERPRETATION OF THE MESSINIAN UNCONFORMITY
97
Fig. 9. Amplitude extraction and isochrones map of the Messinian Unconformity Surface. Note the coincidence between the location of the Castellon Sandstone Formation and the higher seismic amplitudes. Seismic amplitude reflections are also higher in the interfluve areas between Messinian valleys. They have been interpreted as high-amplitude valley fills (A), talus deposits (B) and meandering forms (C). phase and polarity. Instead, it is a conspicuous discontinuity surface at which it is possible to distinguish clear reflection terminations. Throughout the study area, the Messinian Unconformity Surface exhibits a remarkable lateral variation in both seismic amplitudes and geometries, as might be expected
for a major unconformity surface juxtaposing different combinations of interbedded sands and shales. One of the most important observations of the unconformity surface is that in the area where the Messinian Unconformity Surface is topographically higher, it is seen to cut into the
98
J. FREY-MARTINEZ ET AL.
Castellon Sandstone Formation in the position of the preMessinian shelf, i.e. the topset region of the clinoforms of Megasequence A. Here, the Messinian Unconformity Surface is well defined because of its markedly erosional character and the strong seismic reflections of the subcropping units of Megasequence A (Fig. 4b). In this part of the study area, a distinctive dendritic topography, with abundant erosive features including numerous V-shape valleys, was established (Figs 4, 5, 7 & 8). Two or possibly three orders of stream branching can be distinguished (Fig. 7). Valleys dissect interfluve areas and merge into larger channels crossing the study area in a N W - S E direction (Figs 7 & 8). Maximum incision of these valleys into the Castellon Sandstone Formation is to be of the order of 300-400 m. In the deeper parts of the survey area, the Messinian Unconformity Surface has a subcrop consisting mainly of the Castellon Shale Formation. This contrast with the region updip to the northwest is also expressed in a much different palaeotopography (Figs 4 & 4c). In these parts, the Messinian Erosional Surface consists of a flatter-lying surface that slopes gently down to the southeast and erodes to a level that is subparallel with the topset and upper slope positions of a deeper set of Miocene clinoforms (Fig. 4c). It is clear that Unit A1 has been almost completely eroded from this sector of the survey area. From a comparison of these two areas, it is thus evident that the topography reflects both the level of erosion, and the differing lithologies of the subcrop, with the dendritic valley system hosted by the alternating sands and shales of Unit A1. Mapping within the fills of the dendritic valley system of the Messinian Unconformity Surface is complicated by the generally highly discontinuous nature of the seismic facies comprising the lowermost valley fills. However, locally at least, it is possible to trace a high-amplitude continuous reflection (termed the I Reflector) on seismic sections through the thalwegs of the main dendritic valleys (Fig. 6). The I Reflector has been tied to a strong gamma-ray kick in the Castellon B-6 well, and is interpreted as coarse sandstone on the completion log (Fig. 6). By comparison with Stampfli & H6cker (1989), this I-Reflector is interpreted as the top of the first stage of infill of the Messinian valleys prior to the main phase of deep marine sedimentary infill. The gross distribution of coarser lower fills of the Messinian valleys can be seen from the amplitude extraction of the Messinian Unconformity Surface (Figs 9a & 9b). The amplitude distribution is plotted along with the structural contours of the Messinian Erosional Surface (Figs 9a & 9b). From this map, it is evident that high amplitudes at this surface occur in two distinct topographic contexts: (1)
(2)
Along the intervalley ridges where Castellon Sandstone Formation forms the subcrop to the unconformity surface, and where curvilinear stripes of high amplitude track parallel to the structural contours. Along the valley bases where the high amplitudes correspond to the coarse fills of the thalwegs.
The high-amplitude values along the thalwegs are particularly clear where several minor Messinian valleys merge into a trunk valley (Figs 9a & 9b). Some of these valleys are associated with ribbon-like bands of high amplitude with gently sinuous meandering patterns (Fig. 10), which could be interpreted as evidence for fluvial infills, Further basinwards to the southeast, the amplitude response is more patchy and amplitudes gradually decrease (Figs 9a & 9b). Minor zones of high amplitude are also located on steeply dipping areas of the Messinian Unconformity Surface where slopes were eroded into the underlying Castellon Sandstone Formation (Fig. 9b). These small zones are evident with long axes perpendicular to the structural contours and are interpreted
as talus deposits of reworked Castellon Sandstone mantling the lower slopes of the valley margins (Fig. 10). Some of these possible talus deposits exhibit fan-shaped geometries widening downslope, and are approximately 1500 m long and 1000 wide. Well-log data shows that the Castellon Sandstone Formation is often highly cemented, and coarse grained, and due to the relatively high carbonate content; these sandstones may have eroded mainly through rockfalls possibly enhanced by the instability of the shales. The amplitude map presented in Figures 9a and 9b, and interpreted in Figure 10, is intended as a simple guide to the complex amplitude response of the Messinian Unconformity, and to show the critical link between subcrop lithology and topography of the unconformity surface. As would be expected for a major unconformity, the amplitude distribution has contributions both from the subcrop and the supercrop, which need to be unravelled further with additional slicing methods.
Discussion The morphological characterization of the Messinian Unconformity mapped with high spatial resolution from the Tortuga 3D seismic survey is highly reminiscent of the complex sub-aerial erosional networks seen in areas of badlands topography, for example in southern Spain or the USA (Gonzalez 2000; Canton et al. 2001). This direct comparison leads us to suggest that the Messinian Unconformity Surface was formed by a phase of sub-aerial erosion during the Late Miocene. Some previous studies of the Messinian Unconformity Surface have concluded that the presence of different generations of paleochannels, showing different stages of erosion and deposition under sub-aerial conditions during the Messinian, indicate that there was more than one phase of erosion (Cita & Ryan 1978; Hs/,i et al. 1978). However, this does not seem to be the case in the study area, where the morphology of the drainage network and the lack of evidence of infilling and subsequent excavation of the sub-aerial deposits indicate that this area was a topographically elevated region exposed to sub-aerial conditions during the whole period of the salinity crisis, and erosion of the Castellon Shelf was the main process developed during the Messinian lowstand. The fact that we do not see evidence of sub-aerial terraces may result from the rapid lowering and subsequent rise of the base level that did not permit development of valley terraces. During the rapid rise of the sea-level and after the reestablishment of the full basin sea level, the main valleys formed during the previous sub-aerial exposure became preferential loci for incision of submarine canyons during subsequent construction of the margin, and these may have formed pathways for turbiditic units transported to deeper areas of the basin. The Messinian valley fill has been interpreted as a coarsegrained fluvial channel floor deposits that resulted from the deposition of reworked materials from the Castellon Sandstone Formation. Both amplitude maps and well log analysis supported this interpretation as both caliper and gamma-ray values decreased in the reflector above the channel floor, showing a coarsening of facies at this horizon. This interpretation is consistent with the interpretation of the Messinian valley fills by Stampfli & H6cker (1989) who affirmed that two wells positioned in Messinian thalwegs encountered tight conglomerates overlying grey clays with irregular fractures filled with calcite. These authors assumed that the high amplitudes at the thalwegs reflected cemented siliciclastics (i.e. conglomerates coming from Mesozoic formations of the onshore Tarragona area) and deposited as Messinian fluvial deposits in Messinian thalweg positions. Strong cementation by calcite may be
3D SEISMIC INTERPRETATION OF THE MESSINIAN UNCONFORMITY
99
Fig. 10. Sketch of the interpretation of the Messinian Unconformity Surface in the Tortuga 3D Area. See Figures 9a and 9b for reference. understood as being related to surface processes in an arid climate. This is reinforced by the overall morphology of the Messinian Unconformity Surface that has a close similarity to recent badlands suggesting that the vegetative cover was poor or absent. Stampfli & H6cker (1989) estimated a minimum value of nearly 2000 m for the Messinian sea level fall on the Castellon Shelf. The Tortuga 3D Area covers only a rather limited area, in which no Messinian evaporites have been observed. Therefore, only a minimum range of sea level drop has been inferred, amounting to around 400 m, equivalent to the largest relief from valley floor to inter-valley crest (Figs 9a & 9b). The fact that the whole of the Castellon Sandstone Formation was eroded in the deepest parts of the study area proves that the sea level drop was at least as large as the original thickness of the Miocene shelf sequence. However, to obtain an estimate of the true local
Messinian paleorelieL the actual relief measured on seismic data has to be corrected for the effects of post-Messinian compaction and differential subsidence (Stampfli & H6cker 1989). Since the amount of post-Messinian tilting over a distance between ridge and valley floor positions of 5 km is small, and there are no major post-Messinian hinges visible on the seismic sections, our estimate of 4 0 0 m palaeorelief is probably a good close approximation to the true relief. The configuration of the Messinian Unconformity Surface in the study area presented in Figures 9a and 9b illustrates the difference in response between shelf and deeper basinal positions during the Messinian salinity crisis. The presence of topographic breaks of slope on the Messinian Unconformity Surface in the study area reveals that fluvially controlled subaerial erosion was restricted to the most topographically elevated part of the study area and it represents the highest
J. FREY-MARTINEZ ET AL.
100
part o f the pre-Messinian marine sediments that were exposed to sub-aerial conditions during the Messinian salinity crisis. The question o f whether deeper parts o f the western Mediterranean Basin dried up will require further investigation o f the detailed topography o f the Messinian U n c o n f o r m i t y Surface in more basinal settings than possible with the current study.
Conclusions 9
9
9
9
9
9
The pre-Messinian Ebro Delta is characterized by progradational and aggradational sequences developed within a deltaic depositional system. These conditions remained, at least, until the Tortonian. During the Messinian, a m a j o r u n c o n f o r m i t y surface developed in the study area due to erosion of the preMessinian Castellon Shelf under sub-aerial conditions. The presence o f topographic breaks-of-slope on the Messinian U n c o n f o r m i t y surface on the Castellon Shelf reveals that the sub-aerial erosion was restricted to highest part o f the study area representing the edge o f the pre-Messinian marine sediments that were exposed to sub-aerial conditions during the Messinian salinity crisis. The relief on the Messinian U n c o n f o r m i t y Surface is characterized by a dendritic valley network in the northwest o f the study area, and a gently dipping smoother surface to the southeast. The contrast in topography is explained by differences in the lithology o f the subcropping units, and these are related to the m o r p h o l o g y of the pre-Messinian depositional system. The m a x i m u m lowering o f sea level during the Messinian Salinity Crisis that can be determined from the palaeorelief in the study area is around 400 m. Post-Messinian submergence o f the Messinian Unconformity and the Pliocene regime of high sea levels resulted in deep water sedimentation o f fine-grained units over the highly irregular topography. During the Late Pliocene to Pleistocene, a major input of sediment into the basin during o v e r a l l h i g h s t a n d c o n d i t i o n s a l l o w e d the m a r g i n to prograde.
We are indebted to Repsol-YPF for making the 3D seismic and well data available and for their permission to publish the interpretation of the data. We are also grateful to Schlumberger GeoQuest for use of IESX and GeoViz software. C, Elders, J. P. Hansen, J. F. Pefia, A. Echanove and M. Matzo are also acknowledged for their comments and suggestions.
References ALONSO, B., FIELD, M., GARDNER, J. V. & MALDONADO, A. 1990. Sedimentary evolution of the Pliocene and Pleistocene Ebro margin, northeastern Spain. In: NELSON, C. H. 8~ MALDONADO, A. (eds) The Ebro Margin. Marine Geology, 95, 313-331. BIJU-DUVAL, B., LETOUZEY,J., MONTADERT, L. ETAL. 1978. Structure and evolution of the Mediterranean basins, ht: Hst), K. J., (ed.) Deep-Sea Drilling Project. 42, 951-984. CANTON, Y., DOMINGO, F., SOLI~-BENET, A. & PUIGDEF,g,BREGAS, J. 2001. Hydrological and erosion response of a badlands system in semiarid SE Spain. Journal afHydrology, 252, 65-84. CITA, M. B. & RYAN, W. B. F. (eds) 1978. Messinian erosional surfaces in the Mediterranean. Marine Geology, 27, 193-363.
CONEY, P. J., Mu~oz, I. A., MCCLAY, K. R. & EVENCHICK,C.A. 1996. Syntectonic burial and post-tectonic exhumation of an active foreland thrust belt, southern Pyrenees, Spain. Journal of the Geological Society of London. 153, 9-16. DAI~OBEITIA, J. J., ALONSO, B., 8r MALDONADO, A. 1990. Geological framework of Ebro continental margin and surrounding areas. hi: NELSON, C. H. ~ MALDONADO, A. (eds). The Ebro Margin. Marine Geology, 95, 265-288. FIELD, M. E. ~ GARDNER, J. V. 1990. Prograding slope model: PlioPleistocene growth of the Ebro margin, NE Spain. Geological Society of American Bulletin, 102, 721-733. FONTBOTt~, J. M., MUlqOZ, J. A. ~,~ SANTANACH, P. 1986. On the consistency of proposed models for the Pyrenees with the structure of the eastern parts of the belt. Tectonophysics, 129, 291-301. GONZALEZ, M. A. 2000. Recent formations of arroyos in the Little Missouri Badlands of southern North Dakota. Geomorphology, 38, 63-84. HS0, K. J., MONTADERT, L. E T A L . 1978. History of the Mediterranean salinity crisis. In: Hsu, K. J., ETAL. (eds) Deep Sea Drilling Project, 42(1), 1053-1978. JULIVERT, M., FONTBOTE, J. M., RIBEIRO, A. & CONDE, L. M. 1972. Memoria del Mapa Tect6nico de la Penfnsula y Baleares. Instituto Geol6gico y Mineral6gico Espafiol, Madrid. MALDONADO, A. & NELSON, C.H. 1990. The Ebro margin study, northwestern Mediterranean Sea--an introduction. In: NELSON, C. H. & MALDONADO, A. (eds) The Ebro Continental Margin, Northwestern Mediterranean Sea. Marine Geology, 95, 157-163. MAUFFRET, A., LABARBARIE, M. & MONTADER, L. 1981. Les afleurements de series s~dimentaires pr&Plioc~ne dans le bassin M~diterranren nord-occidental. Marine Geology, 45, 159-175. MEDIALDEA.J.. MALDONADO,A., ALONSO, B., DiAZ, J. I., FARR,~N, M., GIRO, S., VAZQUEZ, A., SAINZ-AMOR, E., MART[NEZ. A. & MED1ALDEA,T. 1986. Mapa geol6gico de la plataforma continental espafiola y zones adyacentes. E 1:200.000. Tarragona. Memoria. Hojas 41 and 42 Instituto Geol6gico y Minero, Madrid. MONTADERT, L., SANCHO, J., FAIL, J. P., DEBYSER, J. & WINNOCK, E. 1970. De l'age tertiare de la s6rie salif~re responsible des structures diapiriques en M~diterran~e occidental (nord-est des Bal~ares). Acaddmie des Sciences, Paris, 271, 812-815. NELSON, C. H. 1990. Estimated post-Messinian sediment supply and sedimentation rates on the Ebro continental margin, Spain. In: NELSON, C. H. & MALDONADO,A. (eds) The Ebro Margin. Marine Geology, 95, 395-418. REHAULT, J. P,, BOILLOT, G. & MAUFFRET, A. 1985. The western Mediterranean basin. In: STANLEY, D. J. & WEZEL, F. C. (eds) Geological Evolution of the Mediterranean Basin. Springer, New York, I 01 - 129. ROCA, E., SANS, M., CABRERA, L. & MARZO, M. 1999. Oligocene to Middle Miocene evolution of the central Catalan margin (northwestern Mediterranean). Tectonophysics, 315, 209-233. RYAN, W. B. F. 1973. Time scale and general synthesis. In: RYAN,W. B. F., HSO K. J. ETAL, (eds) Deep-Sea Drilling Project, 13(2), 1405-1515. RYAN, W, B. F. & CITA, M. B. 1978. The nature and distribution of Messinian erosional surfaces--indicators of a several-kilometresdeep Mediterranean in the Miocene. Marine Geology, 27, 193-230. SOLER, R., MARTINEZ, W., MEGIAS, A. G. & ABEGER, J. A. 1983. Rasgos bfisicos del Ne6geno del Mediterrfmeo espafiol. Servicio Geol6gica, 71 - 82. STAMPFLI, G. M. ~ HOCKER, C. F. W. 1989. Messinian palerelief from a 3D seismic survey in the Tarraco concession area (Spanish Mediterranean Sea). Gealogie en Mijnbauw, 68, 201-210. STOECKINGER, W. T. 1976. Valencia Gulf offer deadline near. Oil Gass Journal, 74(13) 197-204 (part I ) and 74 (14), 181 - 183 (part 2).
3D analogue models of rift systems: templates for 3D seismic interpretation K.R. M C C L A Y , T. D O O L E Y , P. W H I T E H O U S E ,
L. F U L L A R T O N
& S. C H A N T R A P R A S E R T
Fault Dynamics Research Group, Geology Department, Royal Hollowav University o f London, Egham, Surrey TW20 OEX, UK
Abstract: 3D visualizations of modem, high-resolution seismic data have provided valuable insights into the finite geometries and spatial extent of extensional fault systems, but their evolution in time is poorly understood. Scaled 3D analogue models of rift basin evolution provide kinematic templates for understanding the 4D evolution of extensional fault systems. This paper reviews the development of extensional fault systems in analogue models of orthogonal, oblique and offset rifts. In orthogonal and oblique models, stretching above a zone of ductile deformation at the base of the model initially produced segmented rift border faults whose orientations were strongly controlled by the underlying baseplate configuration. In contrast, the intra-rift faults generally initiated at high angles to the extension direction. With increased extension both the rift border faults and the intra-rift faults propagated along strike, first producing segmented fault systems separated by relay ramps, which, with increased extension, became breached as fault linkage occurred, Kinks in the fault traces indicate linkage points. Within the models, asymmetric intra-rift sub-basins were formed where the extensional fault arrays had a dominant dip polarity. Intra-basin accommodation zones, separating individual sub-basins along the rift axis, were formed by interlocking oppositely dipping fault systems. Offset oblique rift models, formed above a zone of ductile stretching with basement offsets, generated intra-basin accommodation zones whose orientation was controlled by the underlying basement fabric. The results of the analogue models can be directly compared with fault systems in the Northern Ethiopian rift system, with the accommodation zones in the Gulf of Suez, Egypt, with extensional fault arrays in Canyonlands, Utah. and with rift fault systems in the Gulf of Thailand and the southern North Sea.
Many intracontinental and marine rift basins such as the Gulf of Suez (e.g. Patton et aL 1994; Morley 1994; Bosworth & McClay 2001; Moustafa 2002, Younes & McClay 2002), the East African rift (Bosworth 1985, 1994; Ebinger 1989a, b; Rosendahl et al. 1986; Rosendahl 1987; Morley, 1999; Morley et al. 1990), the Rio Grande rift system (Kelly 1982), and the North Sea are characterized by half-graben sub-basin systems with the dominant fault systems changing polarity along the rift axis (cf. Bally 1981; Gibbs 1983, 1984; Etheridge et al. 1987: Morley 1994; Nelson et al. 1992; Lister et al. 1986; Faulds & Varga 1998). Extensional fault systems within these rifts are commonly segmented with long rift border faults ( > 10kin) separated by relay ramp structures (Kelly 1982) that may provide pathways for sediment input into the basin (e.g. Gawthorpe & Leeder 2000). This segmentation in rift systems characteristically occurs every 5 0 - 1 5 0 k m along the rift axis (Rosendahl et al. 1986; Hayward & Ebinger 1996) and may or may not involve polarity reversals of the sub-basin systems. However the detailed 3D geometries and the evolution of the segmentation zones in exposed natural rift systems are comparatively poorly understood. Similarly, although structural geometries are wellimaged in some rift basins (e.g. Gulf of Thailand, Kornsawan & Morley 2002), 3D seismic datasets covering many rifts such as the Gulf of Suez (Patton et al. 1994) are of poor quality due to sub-salt imaging problems and do not provide sufficient detailed structural information on fault architectures. Physical simulations using scaled sandbox analogues have given valuable insights into the nature of extensional faulting in rift systems (e.g. Cloos 1968; Faugere & Brun 1984; Withjack & Jamison 1986; Serra & Nelson 1989; Tron & Brun 1991; McClay & White 1995; McClay 1990a, b; McClay et aL 2001, 2002). These analogue experiments show the initiation, growth and linkage of extensional fault systems and provide models for the development of natural rift fault systems. In particular the development of accommodation zones formed by interlocking opposing dip fault systems, where changes in rift sub-basin polarities occur and where rift sub-basins are stepped or offset, has been successfully simulated. This paper reviews the results of 3D analogue rift models and compares them with natural
examples of well-exposed rift systems (e.g. Northern Ethiopian rift) and interpretations of 3D seismic surveys across rift systems (e.g. Pattani basin, Gulf of Thailand; North Sea). Conceptual models for the 3D geometries and evolution of rift basin fault systems are developed and these may provide templates for seismic interpretation in poorly imaged rift basins.
Analogue modelling Experimental method The analogue modelling experiments were carried out in a deformation rig 120 cm • 60 X 7.5 cm in size (Fig. 1). A 7.5 cm thick pre-extension sandpack was formed by mechanically sieving 2 to 3 mm thick layers of white and coloured, dry quartz sand (average grain size 90 Ixm) into the deformation rig on top of a basal detachment/baseplate set-up (Fig. la). The basal detachment was formed by a 10cm wide rubber sheet fixed between two aluminium end sheets (Fig. l b) and simulates a zone of ductile stretching at the base of the brittle upper crust. Two models were run with parallel-sided rubber sheets, one at 90 ~ to the extension direction (orthogonal rift), and one with the rubber baseplate oriented at 60 ~ to the extension direction (60" oblique rift). A third model was run where the rubber baseplate was 60" oblique to the extension direction but offset by three extension parallel faults (Fig. la). Deformation was achieved by moving both of the end walls with a motor driven worm screw (Fig. 1) at a constant displacement rate of 4.16 x 1 0 - 3 c m s -1. The models were extended in 0.2 cm increments to a maximum of 100% extension of the zone of basal stretching. The top surfaces were recorded by 35 mm digital photography. After each 2 cm of deformation the accommodation space generated by the extension was infilled with alternating layers of white and coloured sand to simulate syn-rift sedimentation. The quartz sand has a linear Navier-Coulomb behaviour with an angle of friction of 31 ~ (McClay 1990b). The models described in this paper are scaled such that they simulate brittle deformation of a sedimentary sequence between 1 and 7.5 km thick (cf. McClay 1990a) (Fig. 2). Models using each baseplate geometry were
DAVIES,R. J., CARTWRIGHT,J. A., STEWART,S. A., LAPPIN,M. & UNDERHILL,J. R. (eds) 2004.3D Seismic Technology: Application to the Exploration of Sedimentao' Basins. Geological Society, London, Memoirs, 29, 101-115.0435-4052/04/$15 9 The Geological Society of London 2004.
102
K.R. MCCLAY ETAL.
Fig. 1. Analoguemodelling rig. (a) Plan view showing baseplate geometry and orientation with respect to the extension direction (arrows). The baseplate shown is for the offset oblique rift models. (b) Cross-sectional view showing the 7.5 cm thick layered sandpack above the central stretching baseplate.
repeated at least twice to ensure reproducibility. At the end of the deformation run the models were sectioned both horizontally and vertically in order to analyse the internal deformation patterns.
Analogue models of rift fault systems--results The results of the three representative rift basin analogue models are briefly summarized in this paper.
Model 1--orthogonal rift.
In the orthogonal rift model the underlying zone of stretching at the base of the model was oriented at 90" to the extension direction. Initial extension produced isolated, short strike length faults perpendicular to the extension direction. Even after 2cm extension some of these segmented faults had linked to produce slightly sinuous, longer faults particularly along the margins of the rift zone (Fig. 3a). Numerous kinks observed along the rift boundary fault system are sites of fault segmentation and linkage forming high-angle, small-scale relays and breached relays (Fig. 3a). In the centre of the basin, major, low-angle relay ramps are clearly observed between parallel extensional fault segments with large overlap distances and relatively low displacements (Fig. 3a).
With increased extension individual fault segments propagated along strike. Non-co-linear segments overlapped to form relay ramps. Increased extension produced long, linear to slightly sinuous faults formed by linkage of the originally smaller fault segments (Fig. 3b). Linkage typically occurred by footwall or hangingwall breaching at relay ramps. Positions of fault linkage are marked by kinks in the fault trace and in many places splays are also developed. Two long rift border faults were well developed by 4cm of extension together with an intra-rifl fault system that dominantly dipped towards the left-hand side of the model except near the left-hand border fault. Antithetic right-dipping faults formed near the right-hand border fault (Fig. 3b). At 6 cm of extension, most of the faults had become linked and extended across most of the model. Breached relay ramps, splays and kinked fault traces are common (Fig. 3c), The tips of the individual faults were typically curved. Individual extensional faults increased their displacement and extension tended to focus inwards to the centre of the model (Fig. 3c). After 10cm of extension (approximately 100% stretching of the basal detachment sheet), highly linked, sinuous faults extend across the whole model (Fig. 3d). Distinct kinks in the fault traces show where segments have linked. Local accommodation zones are formed where oppositely dipping fault arrays interfere (Fig. 3d). The line diagram (Fig. 4a) shows the dominant fault systems with a characteristic switch in fault dip (polarity) across the model producing the overall graben geometry. Local intra-basin accommodation zones are highlighted. All faults developed at high angles (near 90 ~) to the extension direction and remained so. Deviations from this mainly occurred where relay ramps became breached and at fault tips where splays propagated at an angle to the main fault (Fig. 4a). Overlap zones between likedipping faults form relay ramps and only minor accommodation zones were developed reflecting local fault geometries and interactions (Fig. 4a). No strike-slip transfer faults were seen to develop in this model. Serial cross-sections through the central section of the completed model show the moderate asymmetric nature of the rift system (Fig. 4b). Crosscutting faults developed on the righthand portion of Sections 5 and 6 reflect the relative importance of left or right dipping faults as the rift evolves (Figs 3 & 4b). As the model rift evolved and fault systems propagated along strike some older faults were crosscut by new, more active faults. In cross-section the model is the nearly cylindrical along
Fig. 2. Scaling of the analogue models showing the angle of friction of the sandpack with depth in the analogue models versus that for natural rocks in the brittle upper crust (after McClay 1990a).
3D ANALOGUE MODELS OF RIFT SYSTEMS
103
Fig. 3. Analogue model 1--orthogonal rift. (a) Overhead view of the analogue model after 2 cm extension. Illumination is from the left. Dark bands are faults dipping to the left and light bands are faults dipping to the fight. The linkage of the fight-hand border fault system is highlighted. (b) Overhead view of the analogue model after 4 cm extension. Note the increased displacement on the fault systems with fault overlaps producing relay ramps and breaching of relay ramps (highlighted). (r Overhead view of the analogue model after 6.0 cm extension. Relay ramps have become breached by along-strike propagation of overlapping faults. (d) Overhead view of the analogue model after 10 cm extension. Local accommodation zones are highlighted.
the rift axis. Local graben systems are developed adjacent to each border fault (Fig. 4b). In Sections 5 - 8 the rift is symmetric and composed of two sub-basins separated by a weakly developed central horst block (Fig. 4b).
M o d e l 2 60 ~ oblique rift. The 60 ~ oblique rift model shows a marked change in fault patterns compared to the orthogonal rift model described above. In this model the rift borders were characterized by en-echelon offset fault segments oblique to the extension direction whereas extension normal faults formed in the central axial section of the rift (Fig. 5). After only 2 cm of extension, long faults composed of numerous segments had formed along the model rift borders--parallel to the boundaries of the underlying basal detachment (Fig. 5a). The tips of these fault segments overlap to form relay ramps. Along the rift axis, a series of parallel, overlapping intra-rift faults formed perpendicular to the extension direction (Fig. 5a). The overlapping intra-rift faults form numerous relay ramps in the central section of the model (Fig. 5a). After 4 cm of extension the fault segments along the rift borders had propagated and linked (Fig. 5b). Short overlap, steep, relay ramps are found
between longer linked segments (Fig. 5b). Within the central section of the rift new faults nucleated perpendicular to the extension direction and existing faults underwent marked alongstrike propagation (Fig. 5b). Numerous gently dipping relay ramps were formed between the long overlaps of intra-rift faults. Two localized intra-basin accommodation zones formed separating oppositely dipping fault arrays (Fig. 5b). These are oriented oblique to the extension direction and are formed by the overlap of oppositely dipping fault arrays. The lower accommodation zone in Figure 5b is moderately oblique to the extension direction and within it the overlapping fault tips are strongly curved. In contrast the upper accommodation zone is only slightly oblique and the overlapping fault tips show only slight tip curvatures (Fig. 5b). With increased extension to 6 c m the rift margin fault systems displays increased hard linkage although some relay structures remained extant on both margins (Fig. 5c). Kinks and splays along the traces of linked fault segments indicate where the original segment boundaries were. The central intra-basin accommodation zone (highlighted in Fig. 5c) became more pronounced. The tips of overlapping faults in the accommodation zone are strongly rotated (Fig. 5c). In contrast, the upper
104
K.R. MCCLAY ETAL.
Fig. 4. Analogue model 1-----orthogonalrift. (a) Line diagram interpretation of the surface fault pattern at the end of 10cm extension. Dark bands are faults dipping to the left and light bands are faults dipping to the right. Intra-basin accommodation zones are highlighted. (b) Serial sections through the orthogonal rift model. Syn-kinematic strata are the red and white layers infilling the upper part of the graben system whereas the pre-kinematic strata are the blue, white and black layers. Note the relative homogeneity of cross-sections along strike.
intra-basin accommodation zone remains near parallel to the extension direction with only minor along-strike propagation of the overlapping fault arrays. At this amount of extension the intra-rift fault systems are well developed, and strongly linked. They define local sub-basins within the rift (Fig. 5c). At the end of extension (11.5 cm ~ 100% extension of the underlying basal detachment sheet), the rift borders are defined by long, linked fault segments forming near-continuous structures along the margins (Fig. 5d). The patterns of intra-rift faulting display significant changes with deformation focused into the centre of the rift (Fig. 5d). In the upper half of the model, the accommodation zones are well developed but are limited at their ends by segmented fault systems parallel to the rift borders (Fig. 5d). This rift border parallel fault system is formed in part by strongly curved tips of pre-existing intra-rift faults as well as the development of new oblique fault segments parallel to the rift margins. As a result, the main intra-basin accommodation zone is breached and a graben formed adjacent to, and parallel to, the left-hand rift margin of the model (Fig. 5d). Within the axial zone of the rift, the dominant trend of the intra-rift faults was at high angles to the extension direction. Individual faults, however, commonly display sinuous and kinked traces due to the linkage of original offset en-6chelon fault segments. A complex pattern of local sub-basins developed within the central section of the model. Figure 6a shows the interpretation of the fault patterns at the end of extension. Fault linkages, relays and
breached relays are highlighted along the fight-hand border fault system as well as oblique, intra-basin accommodation zones. No strike-slip faults or transfer faults were seen to develop in this model although it is possible that major accommodation zones with associated fault tip rotation (Fig. 6a) could be misinterpreted as strike-slip structures. This is discussed further below. Serial cross-sections through the completed model (Fig. 6b) show the essential asymmetry of the rift architecture as well as displaying changes in fault polarities across accommodation zones (e.g. Sections 2 & 3, Fig. 6b). These polarity changes are also associated with a marked increase in fault density, reflecting the interlocking nature of these zones and associated damage zones (see Sections 3, 4 & 5, Fig. 6b). Local graben systems are developed adjacent to the border faults on either side of the model.
Model 3 60" offset rift.
Offset rift models were run in order to investigate the effects of along strike changes in orientation and position of the zone of basal stretching on the internal rift fault patterns, as well as to investigate whether strike-slip offsets in the basement could generate through-going 'transfer faults' in the overlying cover sequence. In Model 3 the dominant orientation of the rift axis was 60 ~ to the extension direction. Three, extension-parallel, sinistral strike-slip faults were cut into the basal detachment sheet, thus offsetting the zone of ductile
3D ANALOGUE MODELS OF RIFT SYSTEMS
105
Fig. 5. Analogue model 2 ~ 0 ~ oblique rift. (a) Overhead view of the analogue model after 2 cm extension. Illumination is from the left. Dark bands are faults dipping to the left and light bands are faults dipping to the right. Note the strong en-echelon offset extensional faults along the border fault system. (b) Overhead view of the analogue model after 4 cm of extension. The border fault system has become linked with longer fault traces and breached relay ramps. Note the development of intra-basin accommodation zones (highlighted). (c) Overhead view of analogue model alter 6.0 cm of extension. The intra-basin accommodation zones are more prominently developed. (d) Overhead view of the analogue model after i 1.5 cm of extension. The border fault system is now strongly linked and the central intra-basin accommodation zone is breached by a left-dipping segmented fault system.
stretching at the base of the model (Fig. la). As in the simple 60 ~ oblique rift model described above (Fig. 3), initial extension produced two dominant fault populations. The rift margin fault set formed at 60 ~ to the extension direction, parallel to the boundaries of the underlying detachment sheet whereas the intra-rift fault segments tended to be at high angles to the extension direction (Fig. 7a). Across the central zone of the rift the intra-rift faults were slightly sigmoidal in shape as a result of their interaction with the rift margin as well as with oppositely dipping intra-rift faults (Fig. 7a). Overlapping fault tips formed relay ramps and even at this low extension stage of the model, separate, well-developed arrays of like-dipping faults had formed (Fig. 7a). With more extension individual faults increased fault displacement and propagated along strike along the rift margins (Fig. 7b). The rift margin fault systems were parallel to the edge of the basal rubber sheet (i.e. the boundary of the zone of stretching), but were offset across the basement steps generating well-developed relay ramps at these positions (Fig. 7b). Within the model, arrays of convex to sigmoidal, like-dipping intra-rift faults formed strongly asymmetric sub-basins. These oppositely dipping fault arrays were separated by intra-basin accommodation zones subparallel to slightly oblique to the offsets in the basal detachment sheet (Fig. 7b). Within these accommodation zones individual fault tips are strongly rotated to form a zone
of overlapping and interlocking faults at angles as much as 30 ~ to the extension direction. Outside of these intra-basin accommodation zones, overlapping like-dipping faults form relay ramp structures (Fig. 7b). After 6 c m of extension, individual sub-basins formed by arrays of like-dipping extensional faults are well developed and separated by well formed intra-basin accommodation zones (Fig. 7c). Within these accommodation zones, fault tips are strongly curved and form complex interlocking arrays. Along-strike propagation of these domino fault arrays was inhibited by the interlocking nature of oppositely dipping fault tips in the accommodation zones (Fig. 7c). At this stage of extension, the rift border fault systems display well-formed relay ramp structures, some of which have begun to be breached (Fig. 7c). With increased extension from 6 to 11.5 cm (Fig. 7d), the degree of fault linkage and complexity increased but the general fault architecture within the model remained largely unchanged. Relay ramps along the rift border fault systems became breached (Fig. 7d). At the end of extension (approximately 100% at the base of the rift zone), the model contained four distinct arrays of left-dipping faults and three distinct arrays of right-dipping faults (Fig. 7d). These oppositely dipping fault arrays interfere in three well-developed intra-basin accommodation zones (Figs 7d & 8a). The accommodation zones developed in this
106
K.R. MCCLAY E T A L .
Fig. 6. Analogue model 2--60 ~ oblique rift. (a) Line diagram interpretation of the surface fault pattern at the end of extension. Dark bands are faults dipping to the left and light bands are faults dipping to the right. Segmentation and linkages along the right hand border fault system are highlighted together with the intra-basin accommodation zones. (b) Serial sections through the oblique rift model. Syn-kinematic strata are the red and white layers that infill the graben system whereas pre-kinematic strata are the blue, black and white layers. Intra-basin accommodation zones are characterized by conjugate fault arrays--Sections 2, 3, 5 and 7.
offset oblique-rift model are long-lived and relatively constant in position. Fault tip rotation in these zones results in the generation of convex-to-the-hangingwall fault segments and sigmoidal fault traces when traced from one accommodation zone to the next (Figs 7d & 8a). Overlapping tips of oppositely dipping faults produce complex conjugate fault arrays in the intra-basin accommodation zones. These are commonly characterized by a dense zone of small-scale extensional faults immediately adjacent to the conjugate fault array (see below: Figs 7d & 8a). Figure 8a shows the interpreted fault pattern at the end of extension. Relay ramps and kinked linkage zones along the border fault system are highlighted as well as the dominant intra-basin accommodation zones. Serial cross-sections through
the completed models shows distinct changes in dip polarities with three prominent accommodation zones characterized by graben geometries and conjugate fault arrays (Sections 3, 6 and 9: Fig. 8b). Dense arrays of small-scale faults are seen to characterize parts of the rift directly adjacent to the accommodation zones (Figs 8a & b). As in the other models described above, no strike-slip or oblique-slip transfer faults were developed in the offset rift models despite the presence of discrete sinistral strike-slip offsets in the basal detachment sheet. However, as in the 60 ~ rift, the accommodation is typically characterized by fault tip rotation and elongation into the accommodation structure, forming an oblique fault zone that could be misconstrued as a strike-slip system (Figs 7d & 8a).
3D ANALOGUE MODELS OF RIFT SYSTEMS
107
Fig. 7. Analogue model 3--60 ~ offset oblique rift. (a) Overhead view of the analogue model after 2 cm extension. Illumination is from the left. Dark bands are faults dipping to the left and light bands are faults dipping to the right. Intra-basin accommodation zones (highlighted) develop above the basement offsets. (b) Overhead view of the analogue model after 4 cm of extension, Intra-basin accommodation zones (highlighted) are well developed and consist of overlapping and interlocking fault arrays. Overlapping faults at the rift borders produce well-developedrelay ramps. (c) Overhead view of the analogue model after 6.0 cm extension. The intra-basin accommodation zones are highlighted. At the borders of the rift relay ramps are breached by propagation of overlapping faults. (d) Overhead view of the analogue model after 11.5 cm extension. Three well-developed accommodation zones (highlighted separate different domains of like-dipping faults.
Discussion
Summary of the analogue models Analogue models of orthogonal rifts produce simple, symmetric to asymmetric graben systems that are relatively uniform along strike (Fig. 3). The model rifts are defined by long rift border faults. These formed by the along strike propagation of initially shorter segments which are then linked by the breaching of the intervening relay ramps (Fig. 3b). The border fault segments are typically orthogonal to the extension direction. Intra-rift faults also are typically at high angles to the extension direction but many are slightly sinuous or kinked along strike as the result of linkage of originally en-6chelon offset fault segments (Fig. 4a). Oblique rift models produce segmented rift basins with rift border faults that are controlled by the boundaries of the underlying zone of basal stretching in the models (Figs 5 & 6a). Intra-rifl fault systems formed at a high angle to the extension vector (Figs 5 & 6a). Oblique rift models display significant structural variation along the length of the rift zone. Localized fault polarity switching produces sub-basins separated by intrabasin accommodation zones (Figs 5 & 6a). These intra-basin accommodation zones are formed by interlocking arrays of
oppositely dipping faults and are commonly the sites of dense networks of small-scale faults (Figs 5 & 6). These fundamental patterns were established very early in the extension history of the model and thus had a profound effect on the evolution of the rift interior, although increased extension resulted in some modifications of these accommodation zones (Fig. 5a-d). The interlocking nature of the oppositely-dipping fault systems that form the accommodation zones inhibited further along strike propagation of the intra-rift faults, and may have lead to abnormal displacement/length relationships (Fig. 5; see below). Analogue models of offset 60 ~ oblique-rifts generated a segmented rift zone, bordered by rift-margin faults parallel to the main 60 ~ basement detachment sheet (Fig. 7). Individual sub-basins display a dominant dip polarity of slightly convex fault arrays. Intra-basin accommodation zones parallel to the extension direction separate each sub-basin and are located above the offset basement fabrics (Figs 7 & 8a). Opposite polarity fault arrays interlock with overlapping fault tips that show a high degree of rotation into the accommodation zones (Figs 7 & 8a). Sub-basins in these models are separated by softlinked accommodation structures rather than by 'hard-linked' sinistral strike-slip transfer faults (cf. Figs 7 & 8a). The strikeslip faults that offset the basal rubber sheet do not propagate
108
K.R. MCCLAY E T A L .
Fig. 8. Analogue model 3--60 ~ offset oblique rift. (a) Line diagram interpretation of the surface fault pattern at the end of extension. Dark bands are faults dipping to the left and light bands are faults dipping to the right. The intra-basin accommodation zones above the offset basement are highlighted. Linkages of the right-hand border fault system are also shown. (b) Serial sections through the offset oblique rift model. Syn-kinematic strata are the red and white layers that infill the graben system whereas pre-kinematic strata are the blue, black and white layers. The well developed intra-basin accommodation zones are marked by conjugate fault arrays in Sections 4, 6, 7 and 10.
upwards into the overlying sandpack. Serial sections through the model illustrate the switch in polarity of the intra-rift structures across the changes in basement fabric and demonstrate a marked increase in small-scale fault density as these accommodation structures are approached (Figs 8a & b). These orthogonal and oblique rift models show similar fault growth and linkage characteristics to those found by Cowie and co-workers in numerical stress feedback fault models (Cowie 1998; Gupta e t al. 1998). In these models initially isolated, optimally positioned, fault segments rapidly link to form long,
continuous high displacement fault zones, or soft-linked high displacement segmented rift border faults (cf. Cartwright et al. 1995). Along strike propagation and linkage is promoted initially by this stress feedback mechanism and associated stress shadows (cf. Gupta et al. 1998) and is illustrated by the rift margin fault systems that develop above the linear discontinuities in the basement fabric. In the offset oblique rift model, rift border fault zones are initially not continuous but consist of like-dipping fault segments linked by major relay ramps parallel to the rift axis (Fig. 7). Further extension in this offset rift system
3D ANALOGUE MODELS OF RIFT SYSTEMS produced breaching and linkage of the relay ramps between offsets in the border fault system (Figs 7 & 8a). In all of the rift models, changes in displacement along the strike of individual fault systems were usually accommodated by overlapping fault segments that produced 'soft-linked' relay ramps. Where faults had joined (linked) along strike, this generally occurred by breaching of relay structures (cf. Childs et at. 1995), such that distinct kinks in the fault trace were formed at these linkage points. Changes in the dip polarity of arrays of extensional faults were accomplished across complex intra-basin accommodation zones that were controlled by the basement discontinuities parallel to the extension direction. Discrete strike-slip transfer faults were not developed in any of the analogue models carried out during this research programme.
Comparisons with natural rift architectures The results from the analogue models are compared to fault architectures in natural rift systems including outcrop examples from the northern Ethiopian rift, the Gulf of Suez rift, and the Canyonland extensional fault systems in Utah, as well as 3D seismic examples from the Jupiter Fields, Southern North Sea and the Southern Pattani basin, Gulf of Thailand.
Northern Ethiopian rift, Afar. The northern Ethiopian rift that extends into Afar developed during the Early-Mid Miocene by dominantly W N W - E S E regional extension (Hayward & Ebinger 1996). Along the axis of the rift a number of distinct rift segments formed with basins typically 30-50 km wide and 80-100kin long. Two well-developed basins are shown in
Fig. 9. Natural Example. Northern Ethiopian rift, Afar. (a) Landsat TM image of the northern Ethiopian rift system. Evaporitic lake basin sediments are white to light coloured. Recent volcanic cones are red. Note the elliptical (i.e. extended) caldera of the Ayelu Abidu volcano in the lower left-hand corner of the image. (b) Fault map interpretation of (a) above showing the two major sub-basins with dominant border faults on the eastern side. A central accommodation zone separating them is characterised by abundant volcanic cones and flows.
109
Figure 9a, a Landsat TM image of part of the northern Ethiopian rift. The southern basin was named the Southern Adda-do segment by Hayward & Ebinger (1996). Both the northern and southern basins are characterized by large rift border faults on their eastern margin (Fig. 9). Basin geometries are outlined by light-coloured lake and evaporitic sediments deposited in the hangingwalls of the rift border faults. The terminations of each of these main segments show intense faulting and a marked concentration of Recent volcanic cones and flows (Fig. 9a) developed in the accommodation zones between segments. In this example both the border faults and the intra-rift faults are orthogonal to the extension direction. Individual faults, in particular the border faults, show well-developed relay ramps between overlapping fault tips (Fig. 9b). The geometries in this rift setting display many similarities to those generated in the orthogonai rift model as well as to offset orthogonal rift models presented in McClay et al. (2002; Figs 3 & 4). These Type-1 accommodation zones (McClay et aL 2002) are perpendicular to the rift axis and parallel to the extension direction. The high density of small faults in the vicinity of these accommodation zones is also a feature observed in the analogue models (Fig. 7d). Hard-linked transfer faults are not commonly found in the wellexposed rift systems of northern Ethiopia and Afar (Kronberg 1991; Hayward & Ebinger 1996).
G u l f o f Suez, Egypt The Northwestern Red Sea-Gulf of Suez rift system was initiated during the Late Oligocene and underwent significant N65~ extension during the Early Miocene (Bosworth 1994; Patton et al. 1994; Moustafa 1997; McClay et al. 1998; Bosworth & McClay 2001: Younes & McClay 2002). The Gulf of Suez rift is characterized by dominant NNW and N - S trending extensional faults that form a distinct rhomboidal map pattern (Fig. 10). McClay & Khalil (1998) and Younes & McClay (2002) showed that these fault orientations were controlled by pre-existing Precambrian basement fabrics in the crystalline basement of the rift system. The Gulf of Suez has long been recognized as one of the best examples of along-axis segmentation into sub-basins with different dip polarities (Moustafa 1976; Colletta et al. 1988; Moustafa 1993, Bosworth 1994; Patton et al. 1994; McClay et al. 1998; Bosworth & McClay 2001; Younes & McClay 2002). In the north the dominant rift border fault occurs on the SW rift margin (Fig. 11 ). In the central Gulf of Suez, the dominant rift faults occur on the NE rift margin whereas further south they switch to the SW margin (Fig. 11). Diffuse, poorly-defined, accommodation zones separate these different dip domains--sub-basins (Fig. 11). They are oblique to the extension direction and are characterized by overlapping fault arrays similar to the accommodation zones developed in the analogue models (Fig. 7). Individual faults were initially strongly segmented and offset across 'soft-linked' relay structures (McClay & Khalil 1998: Khalil 1998). With increased extension these faults became linked by breaching of relay structures with the development of local 'hard-linked' transfer faults, thus giving rise to the rhomboidal fault pattern of the rift system. No crossbasin strike-slip transfer faults have been identified in the Gulf of Suez.
Canvonlands Grabens. Utah. The spectacular grabens of Canyonlands National Park, southeast Utah provide unique mesoscale outcrop examples of extensional fault systems, fault linkages and graben structures in plan view (Trudgill & Cartwright 1994: Fig. 11). Spectacular faults and graben structures are developed in 450-500 m of shale and sandstone of the Pennsylvanian Hermosa Formation detached above
I l0
K. R. MCCLAY ET AL,
Fig. 11. Natural Example. Canyonlands, Utah. (a) Enlarged aerial photograph of part of the Canyonlands graben system. Illumination is from the east and down-dropped grabens are grey areas of Quaternary infill. High areas are light shaded sandstones outcrops, in places strongly jointed. (b) Fault map interpretation of (a) above showing accommodation zones (AZ) between offset graben systems as well as individual relay ramps (R). Fig, 10. Natural Example. Gulf of Suez rift, Egypt. Map of the principal extensional faults within the Cenozoic Gulf of Suez rift showing the major sub-basins and changes in fault polarities across oblique, poorly defined accommodation zones (Galala-Abu Zenima AZ and the Morgan AZ).
ductile evaporite units (McGill & Stromquist 1979: Trudgill & Cartwright 1994). Gravitationally driven S E - N W directed extension produced steep brittle faults in the sandstones (Fig. l la). The map patterns of the faults display many similar characteristics to those in the orthogonal rift models (cf. Fig. 3). Within individual grabens strongly segmented faults are observed with linkage points marked by distinct kinks in the fault traces (Fig. l lb). Offsets between like-dipping faults produce characteristic relay ramps (marked R on Fig. lib). Offsets or along strike changes between different graben systems produces complex accommodation zones (AZ) characterized by complex conjugate fault arrays and an increase in fault density similar to marked fault density increases in the models presented in this paper (Figs 3-8). The overall patterns observed in the aerial photographs and from detailed mapping by Trudgill & Cartwright (1994) are directly comparable to those in the orthogonal rift model (Fig. 3) and to offset rift models presented in McClay et al. (2002).
Southern Patanni Basin, Gulf of Thailand. The Tertiary Southern Patanni basin in the Gulf of Thailand forms part of a N - S trending intra-continental rift that continues southward from northern Thailand (Chantraprasert 2000; Kornsawan & Morley 2002). The basin is a N - S trending, half-graben
bounded by the N - S trending, easterly dipping, South Pattani extensional fault (Chantraprasert 2000). The basin exhibits complex fault patterns in the Oligo-Miocene syn-rift and Mid Miocene post rift sequences that make up the fill of the hangingwall of this major half-graben structure (cf. Chantraprasert 2000; Kornsawan & Morley 2002). In particular within the Mid Miocene strata distinct changes in graben positions and fault polarities are found (locally known as graben shifts--Kornsawan & Morley 2002), One such example is shown in Figures 11 and 12a where there is a sinistral shift in the main graben axis. Individual extensional faults show welldeveloped relay ramp structures (Fig. 12) whereas the offset graben shows a well-developed accommodation zone formed by interlocking fault tips (Fig. 12). In this basin, E - W extension is interpreted to have been the dominant force in the generation of the structures, orthogonal to a strong N - S trending basement fabric (Chantraprasert 2000). Along-strike segmentation of the post-rift grabens in the Southern pattani basin may have been related to the growth mechanisms of the associated faults, similar to accommodation zones formed in the uniformly stretching analogue models (Figs 3-6). However, the offset between the grabens may also be attributed to underlying oblique fabrics that define the lateral terminations of N - S trending faults (Chantraprasert 2000). These accommodation zones are comparable to those formed in offset rift experiments whereby basement fabrics controls the orientations and locations of these structures (Figs 7, 8 & 12). Southern North Sea-Jupiter Fields, Sole Pit Basin. The Southern North Sea has experienced a complex tectonic history
Fig. 12. Natural Example. Patanni Basin, Gulf of Thailand. (a) 3D visualization of an accommodation zone showing a sinistral offset to the depocentre and change in dominant fault polarity (location shown in the inset diagram). (b) i--Time structure map of the middle Miocene horizon Mm 1 and the location of three seismic lines in ii-iv; ii-iv Seismic lines across the accommodation zone showing the conjugate fault arrays and the switch of the depocentre to the right.
3D ANALOGUE MODELS OF RIFT SYSTEMS
! 11
112
K.R. MCCLAY ET AL.
during the Mesozoic, with extension occurring in the JurassicEarly Cretaceous and inversion from Late Cretaceous through to the Tertiary. The NW trending Sole Pit Basin is approximately 150km in length and 50kin wide (van Hoorn 1987) and forms one of the major basins in the southern North Sea separated from other basins by relative highs. A high quality 3D seismic survey across the Jupiter Fields, Sole Pit Basin, provides excellent examples of rift-related extensional fault systems (Fig. 13; Fullarton 2000). The Jupiter graben system (Fig. 13) was formed by regional extensional extension from the end Jurassic to the Tertiary. Fullarton (2000) carried out a detailed analysis of the fault systems within the Jupiter Fields 3D survey and clearly demonstrated that the majority of the extensional faults evolved from initially offset segments that propagated both laterally and vertically with increased extension. Figure 14a shows an oblique visualization of one of these segmented systems on the southwestern margin of the central graben in the Jupiter Fields 3D survey. Detailed analysis of the fault patterns (Fig. 14b) shows that two relay ramps are developed between overlapping fault segments. Fault displacement analyses and hangingwallfootwall diagrams show that each relay structure is a displacement low and that the displacement profiles show elliptical patterns typical of segmented extensional fault systems (Figs 14c & d). These patterns are also seen in the analogue models where offset and overlapping fault segments form relay systems with similar displacement profiles (Fig. 5).
Conclusions Analogue models of orthogonal, oblique and offset rifts characteristically produce linked rift border fault systems, the
orientation of which was controlled by the boundaries of the underlying ductile basal detachment sheet (cf. Figs 3-8; also see McClay & White 1995). Figure 15 summarizes the progressive evolution of individual faults, linkages and fault array interactions as deduced from the analogue model experiments. Classic, synthetic, relay ramps (Kelly 1982; Larsen 1988; Peacock & Sanderson 1991, 1994; Walsh & Watterson 1991; Childs et al. 1995) were formed between overlapping, likedipping extensional fault segments (e.g. Figs 3-8). In the analogue models, however, different domains of like dipping faults were separated by intra-basin accommodation zones. These were both parallel (basement controlled) and oblique to the regional extension direction (Figs 5, 7 & 12c). In crosssection these intra-basin accommodation zones consist of conjugate fault arrays formed by interlocking tips of dominostyle fault systems that dip in opposite directions. Such interlocking extensional fault arrays have also been observed in seismic data (Nicol et al. 1996; Kornsawan & Morley 2002) with the conjugate fault zones being regions of significant fault damage necessary to accommodate the displacements on the oppositely dipping faults. In the physical models of oblique rifts, accommodation zones form early in the evolution of the rift and persist throughout much of the evolution of the rift. However, these accommodation zones may be modified as laterally propagating faults cut or by-pass the structure. Basement offsets in models generate intra-basin accommodation zones that remain active throughout the model run and control sub-basin development along the length of the segmented rift. Accommodation zones in the models strongly influence the ability of faults to propagate along strike due to the interlocking nature of the conjugate fault systems within them. Anomalous fault
Fig. 13. Natural Example. Jupiter Fields, Sole Pit basin. Southern North Sea. (a) Location of the Jupiter Fields, Sole Pit Basin. (b) 3D visualization of extensional faults in the Jupiter 3D survey--Base Tertiary horizon. Illumination from the SW. The area of Figure 14 is highlighted.
3D ANALOGUE MODELS OF RIFT SYSTEMS
113
Fig, 14. Example of segmented extensional fault system at the Base Tertiary horizon, Jupiter Fields, Southern North Sea. (a) 3D oblique visualisation of offset extensional fault system. (b) Map pattern of northdipping extensional faults shown in (a) above. Two prominent relay systems 1 and 2 are developed between overlapping fault segments. (c) Fault length-displacement diagram for the fault map shown in (b) above, Faults are colour coded as in the map. Note the displacement minima at the relay ramps. (d) Along-strike hangingwall-footwall map (in the fault plane) showing the elliptical fault displacement patterns (Base Tertiary horizon) for individual fault segments. displacement/length relationships may be expected for these fault systems in accommodation zones. The analogue models and the natural examples show that the architectures of rifts and in particular along strike switches in basin polarities and dominant fault dips across accommodation zones are more complicated that the simple overlapping or interlocking rift border fault models of Rosendahl et al. (1986) (Fig. 15). No extension parallel, strike-slip or oblique slip transfer faults were developed in the models. This is in direct contrast to the extension parallel strike-slip or oblique-slip transfer fault model c o m m o n l y used to account for depocentre changes and polarity switches in rift systems (cf. Gibbs 1983, 1984; Lister et al. 1986; Etheridge et al. 1987: Faulds & Varga 1998). The analogue model fault geometries of orthogonal, oblique and offset rift systems make powerful templates for the interpretation of rift fault systems in the subsurface, particularly where seismic data is sparse or of poor quality due to imaging problems such as sub-salt sections. K. McClay gratefully acknowledges support from BP Exploration. Howard Moore and Mike Creager constructed the deformation apparatus. L. Fullarton was supported by a Royal Holioway Scholarship and seismic data for the Jupiter study was kindly supplied by Conoco and Mobil North Sea Ltd. S. Chantraprasert was supported by a Thai Government scholarship. Seismic data used for the Patanni basin study was kindly supplied by Texaco Thailand. I. Cloke and J. Underhill are thanked for constructive reviews of this manuscript.
Fig. 15. Conceptual patterns of fault evolution derived from analogue model results. (a) Sequential development of a fault system in an orthogonal rift experiment from 2-6 cm of extension. Note initial segmentation and development of relay ramps and breached relay ramps with increased extension. (b) Sequential development of a fault system in a 60 ~ oblique rift experiment from 2 - 6 cm extension. Note the strong en-echelon offset of fault segments and linkage with increased extension. (c) Sequential development of an accommodation zone in an offset 60 ~ oblique rift model. Note the complex overlapping fault tips.
References BALLY, A. W. 1981. Atlantic type margins, bT: Geology of Passive Continental Margins. Histor3", Structure and Sedimentologie Record, American Association of Petroleum Geologists Education Course Notes Series, 19, 1-48. BOSWORTH, W. 1985. Geometry of propagating continental rifts. Nature. 316, 625-627. BOSWORTH,W. 1994. A high-strain rift model for the southern Gulf of Suez (Egypt). In: LAMBIASE.J. J. (ed.) Hydrocarbon Habitat in Rift Basins. Geolo~cal Society. London, Special Publication, 80, 75-102.
114
K.R. MCCLAY ETAL.
BOSWORTH, W. & MCCLAY, K. R. 2001. Structural and stratigraphic evolution of the Gulf of Suez rift: A synthesis. In: ZEIOLER, P. A., CAVAZZA, W., ROBERTSON, A, H. F. R. & CRASQUIN-SOLEAU,S. (eds) Peritethvian Rift~Wrench basins and passive margins. Memoirs du Museum National d'Historie Naturelle de Paris, 186. 567-606. CARTWRIGHT, J. A., TRUDGILL, B. D, & MANSFIELD, C. S. 1995. Fault growth by segment linkage: an explanation for scatter in maximum displacement and trace length data from the Canyonlands Graben of SE Utah. Journal of Structural Geology, 17. 1319-1326. CHANTRAPRASERT, S. 2000. Extensional Faults and Fault Linkages, Southern Pattani Basin, Gulf of Thailand. PhD thesis, University of London. CHILDS, (~., WATTERSON, J. & WALSH, J. J. 1995. Fault overlap zones within developing normal fault systems. Journal of the Geological Socie~, of London, 152, 535-549. CLOOS, E. 1968. Experimental analysis of Gulf Coast fracture patterns. AAPG Bulletin, 52, 420-444. COLLETTA, B., LE QUELLEC, P., LETOUZY, J. & MORETTI, I. 1988. Longitudinal evolution of the Suez rift structure (Egypt). Tectonophysics, 153, 221-233. COWIE, P. A. 1998. A healing-reloading feedback control on the growth rate of seismogenic faults. Journal of Structural Geology. 20, 1075-1087. EBINGER, C. J. 1989a. Geometric and kinematic development of border faults and accommodation zones, Kivu-Rusizi Rift, Africa. Tectonics, 8, 117-137. EBINGER, C. J. 1989b. Tectonic development of the western branch of the East African rift system. Geological Society of America Bulletin, 101, 885-903. ETHER1DGE, M. A., BRANSON, J. C. & STUART-SMITH,P. G. 1987. The Bass, Gippsland and Otway basins, southeast Australia: A branched rift system formed by continental extension, hT: BEAUMONT, C. & TANKARD, A. J. (eds) Sedimentary Basins and Basin-Forming Mechanisms. Canadian Society of Petroleum Geologists Memoir, 12, 147-162. FAUGERE, E. & BRUN, J, P. 1984. Modelisation experimentale de la distention continentale. C.R. Acad. Sc. Paris, 229, Set. II, 365-370. FAULDS,J. E. & VARGA,R. J. 1998. The role of accommodation zones and transfer zones in the regional segmentation of extended terranes. In: FAULDS,J. E. & STEWART,J. H. (eds) Accommodation zones and transfer zones: the regional segmentation of the basin and range province. Geological Society of America Special Paper, 323, 1-46. FULLARTON, L. D. 2000. Structural Analysis of the Jupiter and Camelot Areas, Southern North Sea. PhD Thesis, University of London. GAWTHORPE, R. L. & LEEDER, M. R. 2000. Tectono-sedimentary evolution of active extensional basins. Basin Research, 12, 195-218. GIBBS, A. D. 1983. Balanced section constructions from seismic sections in areas of extensional tectonics. Journal of Structural Geology, 5, 153-160. GIBBS, A. D. 1984. Structural evolution of extensional basin margins. Journal of the Geological Socieo', London, 141,609-620. GUPTA, S., COWIE, P. A., DAWERS, N. H. & UNDERHILL.J. R. 1998. A mechanism to explain rift-basin subsidence and stratigraphic patterns through fault array evolution. Geology, 26, 595-598. HAYWARD, N. & EBtNGZR, C. J. 1996. Variations in the along-axis segmentation of the Afar rift system. Tectonics, 15, 244-257. KELLY, V. C. 1982. The right-relayed Rio Grande Rift, Taos to Hatch, New Mexico. New Mexico geological Society Guidebook, 33rd Field Conference, 147-151. KHALIL, S. M. 1998. Tectonic and Structural Development of the Eastern Gulf of Suez Margin. PhD thesis, Royal Holloway University of London. KORNSAWAN, A. & MORLEY, C. K. 2002. The origin and evolution of complex transfer zones (graben shifts) in conjugate fault systems around the Funan Field, Pattani Basin, Gulf of Thailand. Journal of Structural Geology, 24, 435-450. KRONBERG, P. 1991. Geometries of extensional fault systems, observed and mapped on aerial and satellite photographs of
Central Afar (Ethiopia/Djibouti). Geologie en Mijnbouw, 70, 145-161. LARSEN, P. 1988. Relay structures in a Lower Permian basementinvolved extension system. East Greenland. Journal of Structural Geology, 10, 3-8. LISTER, G, S., ETHERIDGE, M. A. & SYMONDS,P. A. 1986. Detachment faulting and the evolution of passive continental margins. Geology, 14, 246-250. MCCLAY, K. R. 1990a. Extensional fault systems in sedimentary basins: a review of analogue model studies, Marine and Petroleum Geology, 7. 206-233. MCCLAY, K. R. 1990b. Deformation mechanics in analogue models of extensional fault systems. In: KNIPE, R. J. & RUTTER, E. H. (eds) Deformation Mechanisms, Rheology and Tectonics. Geological Society, London, Special Publications, 54, 445-453. MCGILL, G. E. & STROMQUIST, A. W. 1979. The grabens of Canyonlands National Park, Utah: Geometry, mechanics and kinematics. Journal of Geophysical Research, 84(B9), 4547-4563. Mr K. R. & KHALIL. S. 1998. Extensional hard linkages, Eastern Gulf of Suez, Egypt. Geology, 26, 771-774. MCCLAY, K. R. & WHITE, M. 1995. Analogue models of orthogonal and oblique rifting. Marine and Petroleum Geology, 12, 137-151. MCCLAY, K. R., NICHOLS. G. J., KHALIL, S. M., DARWISH, M. & BOSWORTH, W. 1998. Extensional tectonics and sedimentation, eastern Gulf of Suez, Egypt. h~: PURSER, B. & BOSENCE, D. W. J. B. (eds) Sedimentation and Tectonics of the Gulf of Aden--Red Sea Rift System, Chapman & Hall, London. MCCLAY, K. R., DOOLEY, T., GLOAGUEN, R., WHITEHOUSE, P. & KHALIL, S. 2001. Analogue modeling of extensional fault architectures: comparisons with natural rift fault systems. In: HILL, K. C. & BERNECKER, T. (eds) Eastern Australian Basins Symposium (2001). A refocused energy perspective for the future, Petroleum Exploration SocieO' of Australia Special Publications, 573-584. MCCLAY, K. R., DOOLEY, T., WHITEHOUSE, P. & MILLS, M. 2002.4D evolution of rift systems: Insights from scaled physical models. AAPG Bulletin, 86, 935-959. MORLEY, C. K. 1994. Structural geology of rifts. In: LAMBIASE, J. J. (ed.) Hydrocarbon Habitat in Rift Basins. Geological Society, London, Special Publications, 80, 75-102. MORLEY, C. K. 1999. (ed.) Geoscience of rift systems--Evolution of east Africa. American Association of Petroleum Geologists Studies in Geology, 44. MORLEY, C. K., NELSON, R. A., PATTON, T. L. & MUNN, S. G. 1990. Transfer zones in the East African Rift system and their relevance to hydrocarbon exploration in rifts. APPG Bulletin, 74, 1234-1253. MOUSTAFA, A. R. 1976. Block faulting in the Gulf of Suez. Egyptian General Petroleum Corporation, Unpublished report presented at the 5 th Exploration Seminar, Cairo, 1976, 1-19. MOUSTAFA, A. R. 1997. Controls on the development and evolution of transfer zones: the influence of basement structure and sedimentary thickness in the Suez rift and Red Sea. Journal of Structural Geology, 19, 755-768. MOUSTAFA, A. R. 1993. Structural characteristics and tectonic evolution of the east-margin blocks of the Suez rift. Tectonophysics, 223, 381-399. MOUSTAFA, A. R. 2002. Controls on the geometry of transfer zones in the Suez rift and northwest Red Sea: Implications for the structural geometry of rift systems. AAPG Bulletin, 86, 970-1002, NELSON, R. A., PATTON, T. L. & MORLEY, C. K. 1992. Rift-segment interaction and its relation to hydrocarbon exploration in continental rift systems. AAPG Bulletin, 76, 1153-1169. N1r A., WATTERSON, J., WALSH, J. J. & CHILDS, C. 1996. The shapes, major axis orientations and displacement patterns of fault surfaces. Journal of Structural Geology, 18, 235-248. PATTON, T. L., MOUSTAFA,A. R., NELSON, R. A. & ABDINE, S. A. 1994. Tectonic evolution and structural setting of the Suez rift. In: LANDON, S. M. (ed.) Interior Rift Basins. The American Association of Petroleum Geologists Memoir, 59, 9-56.
3D ANALOGUE MODELS OF RIFT SYSTEMS
PEACOCK, D. C. P. & SANDERSON, D. J. 1991. Displacement and segment linkage and relay ramps in normal fault zones. Journal of Structural Geology, 13, 721-733. PEACOCK, D. C. P. & SANDERSON, D. J. 1994. Displacements, segment linkage and relay ramps in normal fault zones: AAPG Bulletin, 13, 721-733. ROSENDAHL, B. R. 1987. Architecture of continental rifts with special reference to East Africa. Annual Review o[ Earth and Planetary Sciences, 15, 445-503. ROSENDAHL, B. R., REYNOLDS, D., LORBER, P., BURGESS, C., MCG1LL, J., SCOTT, O., LAMBIASE, J., DERKSEN, S. 1986. Structural expressions of rifting: lessons from Lake Tanganyika. In: FROSTICK, L. E. ET AL. (eds.) Sedimentation in the East African Rifts. Geological Society, London, Special Publications, 25, 29-43. SERRA,S. & NELSON, R. A. ! 989. Clay modelling of rift asymmetry and associated structures. Teetonophysies, 153, 307-312. TRON, V. & BRUN, J-P. 1991. Experiments on oblique rifting in brittleductile systems. Tectonophysics, 188, 71-84.
115
TRUDGILL. B. & CARTWRIGHT, J. 1994. Relay-ramp forms and normal-fault linkages, Canyonlands National Park. Utah. Geological Society c~f America Bulletin, 106. 1143-1157. VAN HOORN, B.1987. Structural evolution, timing and tectonic style of the Sole Pit inversion. In: ZIEGLER, P. A. (ed.) Symposium on Inverted Mesozoic" sedimentat3" basins in the Alpine Foreland, a pan of the Third meeting of the European Union of Geosciences, Elsevier, 137, 239-284. WALSH, J. J. & WATTERSOt~, J. 1991. Geometric and kinematic coherence and scale effects in normal fault systems. In: ROBERTS, A. M.. YIELDING, G, & FREEMAN, B. (eds) The Geomett 3" qf Normal Faults. Geological Society, London, Special Publications. 56, 193-203. W1THJACK, M. O. & JAM1SON, W. R. 1986. Deformation produced by oblique rifting. Tectonophysics, 126, 99-124. YOUNES, A. I. & MCCLAY, K. R. 2002. Development of accommodation zones in the Gulf of Suez-Red Sea rift, Egypt. AAPG Bulletin, 86, 1O03 - 1026,
Structural evolution of a complex 3D fault array in the Cretaceous and Tertiary of the Porcupine Basin, offshore Ireland GREG
JONES
t, L . S . W I L L I A M S "
& R . J. K N I P E
1
fRock Deformation Research Ltd, School of Earth Sciences, University of Leeds, Leeds LS2 9JT, UK (e-mail: g.jon es @ rdr. leeds, ac. uk) 2Chevron Europe, 93, Wigmore Street, London W1U 1HH, UK (current address." PanAfrican Energy UK Ltd, Sheridan House, 40-43 Jewry St., Winchester, Hampshire S023 8RY, UK)
Abstract: A high-quality 3D seismic survey, located in the northwest Porcupine Basin (Irish Atlantic Margin), has been used to investigate the geometry and origin of pervasively developed and complexly distributed linked extensional fault arrays, present within Late Cretaceous and Early Tertiary sequences. The faults show a downwards transition from relatively simple, planar fault segment geometries (-N-S-trending) within younger Early Eocene sand-dominated clastic sequences, into complex conjugate arrays in the underlying older Early Eocene to Late Cretaceous shale-dominated sequences. Rectilinear to polygonal structural configurations are developed at the deeper levels. Most of the fault array ultimately terminates downwards into the Late Cretaceous, where structural accommodation may have taken place by localized or more regional bedding plane slip and/or by volume changes resulting from compaction of fine-grained sequences. Locally. reactivated Jurassic syn-rift extensional faults are locally seen to link upwards into the shallow fault array and appear to have controlled both the intensity and facing direction of the shallower faults on a km scale. The seismic data also clearly show that early upslope-throwing faults are cross-cut by later, downslope-throwing faults. Such geometries are comparable to those formed in sandbox models where gravitational collapse of a tilted sequence is the dominant process controlling fault development. Overall, the fault array geometries seen in the Cretaceous and lower Tertiary successions in this area are interpreted to have resulted from gravitational collapse processes during basin subsidence and sediment compaction, and where the main deformation mechanism was non-rigid block rotation. Differential compaction of Cretaceous and lower Tertiary sediments over pre-Cretaceous rift topography and selective reactivation of the Jurassic fault array are also considered important influences on the resultant fault distribution in 3D.
Structural geology is amongst the disciplines to have benefited significantly in recent years from the widespread acquisition of high-quality 3D seismic datasets during hydrocarbon exploration and production. Seismic data are contributing widely to the understanding of structural processes and providing 'real world' comparisons and examples for other methods of structural investigation, for example sandbox modelling (McClay et al. 2004). 3D datasets are often able to provide very detailed information on fault geometries and linkages, fault activity and even the relationships between faulting and fluid flow. In turn, this information can be used during hydrocarbon exploration programmes, for example to reduce risk during prospect definition and also to help constrain potential trapping mechanisms. In this paper, a 3D seismic dataset is used to describe pervasive and complexly distributed extensional fault arrays present primarily within the Tertiary stratigraphic section present along the northwestern flank of the Porcupine Basin, offshore Ireland (blocks 34/15, 34/20 and 35/16; Figs 1 & 2). The data lie within exploration acreage operated by Chevron Europe Ltd on behalf of the Irish Offshore Licence 5/95 partnership (Dana Petroleum, Conoco UK Ltd, Enterprise Energy Ireland Ltd and Statoil Exploration (Ireland) Ltd). The primary objective of this study was to characterize the distribution, geometrical characteristics and timing of the faulting and then to consider the potential controls on fault array development. Previous studies of structure in the Porcupine Basin (e.g. Tate 1993; McCann et al. 1995) were based on 2D seismic data, thus the present analysis provided a new opportunity to focus in detail on structural development within the basin.
Datasets A 3D survey was obtained over the licence by PGS Ltd and subsequently processed by Western Geophysical Ltd in 1998
as part of the ongoing exploration programme in this area. A number of 2D regional seismic lines were also available to provide a wider overview of the basin. Interpretation of the 3D data was initially undertaken by Chevron during 1999. Seismic volumes utilized during interpretation in this study included near and far-offset time-migrated datasets, a depth-converted dataset (Landmark TDQ) and a Chevron 'Edge Technology' (i.e. coherency or semblance equivalent) time dataset. Detailed seismic line locations cannot be revealed due to data confidentiality issues. Overall seismic data quality is considered to be good within the Tertiary section. Seismic noise is nevertheless noted within the dataset, resulting from poor migration due to rapid velocity changes at interpreted major lithological interfaces and across faults. A feature of these seismic data are the prominent dipping reflectors often seen within the low-amplitude reflective packages. These dipping reflectors can sometimes be ascribed to geophysical noise resulting from poor migration of the dataset, especially where the features are regularly spaced (i.e. coherent noise is present). However, elsewhere these dipping features can be interpreted as fault plane reflectors with some confidence, as a result of the demonstrable vertical continuity from well-defined (faulted) offset horizons within overlying high-amplitude reflective packages, downwards into the dipping reflections within the underlying low-amplitude packages. A limited number of hydrocarbon exploration wells exist in this part of the basin to provide lithological calibration of the seismic stratigraphy. However, the only well that is tied directly to the 3D survey (well 34/19-1; see Fig. 2) lies in a marginal position, where the Tertiary sequences are not fully represented. To the east and north of the survey area, other wells (e.g. 35/8-2, 35/13-1, 35/18-1, etc.; see Fig. 2) provide a more complete record of the Early Tertiary basin-fill sequences (see below).
DAVIES,R. J., CARTWRIGHT,J. A., STEWART,S. A., LAPP1N,M. & UNDERHILL,J. R. (eds) 2004.3D Seismic Technology:Applicationto the Exploration of Sedimenta~ Basins. Geological Society, London, Memoirs, 29, 117-132. 0435-4052/04/$15 9 The Geological Society of London 2004.
118
G. JONES ETAL.
Geological setting of the Porcupine Basin Basin development
Fig. 1. Regional location of the Porcupine Basin and the 3D seismic survey.
and regional structural evolution
The Porcupine Basin is a north-south trending basin that lies on the eastern passive margin of the North Atlantic Ocean (Fig. 1). The main Porcupine Basin is separated from the smaller North Porcupine Basin (and the adjacent Slyne Basin) by the WSW-ENE-trending North Porcupine wrench fault system (Fig. 1; Tate 1993; Dancer et al. 1999). A continental basement feature, the Porcupine High, forms a major topographic ridge on the western margin of the basin, separating it from Atlantic oceanic crust (Naylor et al. 1999, 2002). The basin is normally pressured at the present day at Tertiary levels. The tectonic evolution of aspects of the basin has previously been described in detail in a number of publications (e.g. Croker & Shannon 1987; Naylor & Anstey 1987; White et al. 1992; Tate 1993; McCann et al. 1995). Several recent regional tectonic synthesis papers published in Fleet & Boldy (1999) and Shannon et al. (2001) have since contributed significantly to the understanding of this part of the North Atlantic margin, notably that of Dor6 et aL (1999) and Roberts et al. (1999), who have reviewed the regional tectonic setting and evolution of the North Atlantic. A summary of the main regional events is presented in Figure 3. The Porcupine Basin contains a number of major preMesozoic tectonic features, including the probable western
Fig. 2. Map of the northern Porcupine Basin showing the major structural and topographic features and exploration wells.
FAULT ARRAY EVOLUTION. PORCUPINE BASIN
119
uSW
Atlantic Margin & B o r d e r s
Hatton N. ~ I S. Rockail Hebrides Porcupine,
Plate Tectonic=
NE
Fame Mere Shetland North Sea W 9Shetlan(
Vering Halten
Quaternary 9
1.6
25 -3 O ~ 3 5
13
6
~1
HR
24
.---~
54
~..,..
i
- ~ -
q
qp
? : Js,ws~
'
,,s,.,--'v-
..,..
$
Opening of NE Atlantic Ocean
]
,~.....g.~
25
.
AD,R08
i'i !
" 131
| ~52
i
._
3t'
It.
NR
P
FS
FS MTF2
WSB
C
179
//s. ,
[t" !
"~,ET
~
[t"
231
- -
240
250
~
19
,
I
I
Imprecise Dating for Permian and Triassic Rifting
VG
-----T 2~ Fig. 3. Summary of main regional events along the Atlantic Margins (from Dor~ et al. 1999) 9
:
Sea Ftoo~S p r u d ~ g ~
P
~
In~abas~nal
or
Uplift
extensions of both the Highland Boundary and Southern Uplands faults (Caledonian orogenic sutures) present in the north (i.e. the North Porcupine wrench fault; Fig. 1) and the Variscan front (Hercynian suture) lying in the south of the basin (see Dor6 et al. 1999, their fig. 1). The central part of the basin is transected by the Clare lineament, which links into the Charlie Gibbs Fracture Zone of the North Atlantic Ridge system to the west and the Late Palaeozoic Variscan Front to the east (Fig. 1). Potential field data indicates that the Clare lineament separates areas of significantly different crustal rheology to the north and south (Johnson et al. 2001). These structures have strongly influenced subsequent basin development, as have Early Mesozoic rifting events of Permo-Triassic, Late Triassic and Middle Jurassic age (see Fig. 3; Tate & Dobson 1989, Dor~ et al. 1999; Dancer et al. 1999). The key basin generating event was the major phase of lithospheric stretching and volcanism associated with northwards propagation of the Atlantic rift. Rifting was initiated in the late Callovian/Kimmeridgian and continued into the Early Cretaceous ('Barremian rift' of Roberts et al. 1999), forming the 'paired' Jeanne d'Arc and Porcupine Basins, together with other - N - S - o r i e n t e d rifts in the North Atlantic province. A medial
, ~
Orogeny ~ip, ~
Rifling
o
with orientelion
or~nl,~on
volcanic ridge is present in the basin to the south of the study area (Tate 1993). which together with a narrowing of the basin to the north, indicates rift propagation in that direction. During rifting, the North Porcupine wrench fault (Fig. 1) bounded the Porcupine Basin in the north, whilst to the west, the Porcupine High (Fig. 1) became an isolated feature. Basin development was terminated by continued northwards propagation of the Atlantic rift further to the west of the Porcupine Basin (AlbianCenomanian) and subsequent establishment of seafloor spreading (Roberts et al. 1999). Following this Mesozoic rifting event, the basin underwent thermal subsidence and significant sediment infilling. Limited evidence for post-Jurassic tectonic activity is noted, due to the basin being located on the fringes of both the developing North Atlantic and contracting Tethyan Ocean systems. However, the 'far-field' effects of a number of major regional tectonic events are considered likely to have interrupted this situation, with potentially important implications for Mesozoic and Tertiary evolution 9 The main Late Cretaceous to present day regional events considered to have influenced the Porcupine Basin (Fig. 3; Dor~ et al. 1999) are: (1) possible inversion associated with major phases of Alpine collision in the early Late Cretaceous,
120
G. JONES E T AL.
the Paleocene and especially the Oligo-Miocene (Tate 1993; McCann et al. 1995); (2) regional uplift of the entire Atlantic margin (Paleocene) due to thermal effects related to incipient NE Atlantic opening in the Early Eocene ( - 53 Ma; Dor6 et al. 1999; Jones et al. 2001) and (3) compressional deformation due to North Atlantic ridge-push effects (possible from the ?Late Cretaceous onwards in Porcupine area) (Shannon et al. 1993; Dor6 et al. 1999). Minor extension (e.g. White et al. 1992) a degree of transcurrent faulting (Tate 1993) as well as volcanic activity (Brendan Igneous centre; NE margin of the basin) have been noted over the Late Cretaceous to Oligocene interval, which may be attributable to these regional events. Jones et al. (2001) noted that in the Early Eocene, subsidence within the basin did not fit the expected regional pattern. These authors documented transient regional uplift of 3 0 0 - 6 0 0 m at the Paleocene-Eocene boundary, followed by subsidence of 500-800 m after Early Eocene time (over a time interval of between 25 and 55 Ma). Although the cause of this uplift is not certain, it was n o t considered to have been due to renewed rifting, as rifting cannot account for the observed uplift and also because significant normal faults of Paleogene age were absent. A fall in relative sea level at the Paleocene-Eocene boundary has been ascribed to basin margin uplift associated with opening of the northeast Atlantic (Dor6 et al. 1999; Jones et aL 2001) and in the Porcupine Basin, may have caused erosion and transportation of sediment across the shelf, to deeper water environments. The most detailed studies of structural features undertaken in the basin to date are those of Tare (1993), McCann et al. (1995) and Naylor et al. (2002). In particular, McCann et al. have defined a set of structures in the basin from 2D seismic, that they have termed 'tectonic non-growth faults', which appear comparable to those structures described in this study. These faults affect the Base Cretaceous to Oligocene interval and were attributed by these authors to stress build-up resulting from thermal subsidence, regional flexural down-warping and to sediment compaction.
Stratigraphy
Late Jurassic to earliest Cretaceous rifting was followed by thermal subsidence within the Porcupine Basin, with resultant Cretaceous deposition occurring in shelf to deep marine environments. Thick Early Cretaceous clastic sequences and thin Late Cretaceous and Paleocene chalk sequences are present. Subsequentlyl up to 4km of Cenozoic clastic sediments were deposited in the basin (Tare 1993; McDonnell & Shannon 2001) above the Jurassic and Cretaceous (Fig. 4). Faults, contourite deposits and slump structures (Moore & Shannon 1991; Tate 1993; Stoker 1997: Stoker et al. 2001) are the dominant features seen within the Tertiary sequences (Fig. 4). A seismic sequence stratigraphic interpretation (Van Wagoner et al. 1990) has been used to define the range of possible depositional environments within the Tertiary sequences. The overall lack of well data in the study area means that the lithological characteristics of the Tertiary stratigraphy in the area covered by the 3D survey have largely been inferred. However, the interpretations are constrained by data from the available regional wells (Fig. 2), as well the geometries and the reflective characteristics of the major packages on seismic, which together allow definition of a consistent depositional model. The major units are now outlined. Following the cessation of chalk deposition in the early Paleocene, the upper Paleocene was deposited in shelf, slope and basin floor depositional settings at a time of low to rising sea level, and includes submarine fan sequences in the basin centre (Shannon et al. 1993; McDonnell & Shannon 2001). In the study area, and elsewhere across the basin (McDonnell & Shannon 2001), the Paleocene is considered to be dominated by shales, although sands are associated with the submarine fan sequences present to the east. Following this, at least three main phases of Early Eocene deltaic sedimentation (Moore & Shannon 1992; Shannon et al. 1993: Tate 1993) are recognized in the basin (Figs 4 & 5). In the
Fig. 4. Sketch cross-section based on a regional seismic line, illustrating the tectono-stratigraphy of the western margin of the Porcupine Basin. BCU is the base Cretaceous unconformity.
FAULT ARRAY EVOLUTION. PORCUPINE BASIN
121
Fig. 5. Sketch cross-section of the main facies and stratal geometries present in the upper Paleocene and Early Eocene on the northwestern margin of the Porcupine Basin. study area, the Early Eocene is developed as a stacked sequence of lowstand systems tracts, which initially show progradation and aggradation and later only progradation, expanding to in-fill the basin (Fig. 5). These prograding sequences reflect deposition in shallow marine shoreline, shelf and deltaic shelf to slope environments (Figs 5 & 6; McDonnell & Shannon 2001 ). Stratal geometries characteristic of delta foresets and topsets can be interpreted on seismic in the study area (see Fig. 5). Progradation of these deltaic sequences was largely from the west and northwest (Jones et al. 2001), from the Porcupine High. For the purposes of description, the Early Eocene can be split into two main packages. As shown in Figures 7 & 8, the lower part of the Early Eocene, where progradational and aggradational sequences are well developed, is mostly seismically transparent and lithologies are interpreted to be dominated by shales and siltstones deposited in deltaic shelf to slope environments. This transparent zone is overlain by strongly reflective and predominantly parallel-bedded sequences with a high net to gross (proven by well 34/19-1 and others; Fig. 5), deposited in shallow marine shelf (e.g. Nummulitic sands) and delta top (e.g. clastic sequences with coals) environments. The fault arrays described below occur both within the low net to gross (seismically transparent) lower sequences and in the higher net to gross (seismically reflective) upper Early Eocene sequences (Fig. 7).
Within the Lower Tertiary packages, some lines (e.g. Fig. 8) indicate that the up-dip margin of the faulting coincides approximately to the point where the shale-dominated package thickens significantly basinwards. This location of this boundary is variable along strike with respect to the shale package
Fault array review The main geometrical features of the Tertiary fault array in the area are now briefly reviewed. A regional 3D seismic inline across the area of interest on the western flank of the basin (Fig. 7) provides a good overview of the Porcupine High, the faulted Mesozoic rift sequence and the Cretaceous to Tertiary thermal subsidence-related sedimentary infill. Figure 8 illustrates a number of key features pertinent to the overall interpretation of the extensional fault array, which is very well developed in and around the Lower Tertiary section. Firstly, the significant thickening of the Cretaceous section (mainly the Early Cretaceous) towards the east and the regional tilt from NW to SE (approx. 5 ~ is apparent, relating to burial and subsidence along the basin axis. Secondly, partial tectonic reactivation of the Jurassic syn-rifl extensional fault array is also apparent, mainly at the Mesozoic f i r shelf break (Fig. 8). The shelf margin fault system penetrates upwards into the Lower Tertiary section, which itself is widely deformed by extensional faulting. Similar evidence for reactivation of the Porcupine High eastern bounding fault system (see Fig. 9), together indicate that the largest Jurassic rift faults in this part of the Porcupine Basin were reactivated during the Mid- to Late Tertiary.
Fig. 6. Stratigraphy, lithology and facies in well 35/13-1 (see Fig. 2 for location). From Jones et al. (2001).
122
G. JONES E T AL.
Fig. 7. Regional NW-SE 3D seismic line showing the main structural and stratigraphic features of the study area. Note the Porcupine High in the NW, the major shelf break in the Mesozoic section in the centre of the line and the thick Early Eocene package, which contains the fault arrays described. (A) High amplitude sand-dominated younger Early Eocene package. (B) Low amplitude shale-dominated older Early Eocene package. thickness and is bounded by a complex fault system. The lower (downslope) boundary of the fault system lies mostly outside the 3D survey, but 2D lines clearly show that the faulting terminates near the base of slope, where it displays similar geometries to those seen on the shelf.
The Tertiary fault array displays considerable overall variability in the distribution and throw direction of the faults seen along strike on a semi-regional scale. The areas above (and adjacent to) the reactivated Mesozoic faults are characterized by graben systems that cut through much of the Tertiary section
Fig. 8. NW-SE seismic line showing the complex fault array developed in the Lower Tertiary sequence. (A) Reactivated Jurassic rift fault at Mesozoic basin margin. (B) Note the - 5~ tilt present on the top Chalk. (C) Complex fault array developed in the Early Eocene. Locally, as on this line, a degree of symmetry is seen in the facing directions of the shallower fault array around the reactivated Jurassic rift fault, but this is not always observed. (D) Approximate top of the sand-dominated younger Early Eocene package. (E) Note that on this section, faulting appears to initiate spatially where the lower Early Eocene (shale-dominated) package thickens.
FAULT ARRAY EVOLUTION. PORCUPINE BASIN
123
Fig. 9. NW-SE seismic line showing late-stage fault activity (probably Miocene) on the eastern margin of the Porcupine High. (Fig. 8), within which complex antithetic and synthetic fault arrays occur. On some lines, there is a notable symmetry in the dip direction of faults which are formed either in the footwall (mostly basinwards-facing faults) or hangingwatl (mostly landward-facing faults) slopes of the reactivated Jurassic syn-rift extensional faults, over a distance of several kilometres (Fig. 9). On other semi-regional lines (Fig. 10), it can clearly be demonstrated that the offset direction of cross-cutting faults within the Eocene section is (at least locally) consistent. In the latter case, the down to the east (i.e. basinward) dipping faults regularly offset the west (landward) dipping faults, the former of which therefore appear to have been active latest (Fig. 10). The significance of this observation for fault array evolution is discussed further below. Other more detailed lines illustrate the extent of the complexity present in the Tertiary fault array. Normal faults are very clearly imaged within the generally high-amplitude (i.e. sand-rich) sequences of the younger Early Eocene delta top sequences in both trace and inline orientations (Figs 11 & 12). Mapping reveals that many of these well-imaged faults are approximately N - S to NNW-SSE-trending at this level (see below). When traced downwards into the underlying older Early Eocene and upper Paleocene (shale-dominated) intervals, these faults are often represented as high-amplitude, but somewhat variably imaged dipping reflections within these overall lowamplitude reflective packages (Figs 10-12). The faults at this deeper level also have more complex geometries and orientation distributions (see below). Sub-horizontal reflections representing sedimentary and/or diagenetic layering in the shaledominated package are commonly offset by these dipping features, supporting their origin as fault related reflections (e.g. Figs 9 & 12). In other places, the imaging is weaker and a fault interpretation is less conclusive, especially where vertical continuity cannot be proven into the shallower sand-dominated packages. The high-amplitude dipping reflectors often remain
relatively planar down through the shale rich package and into the upper part of the Cretaceous (Fig. 12), although more curvilinear features are also seen locally (Fig. 11), which may represent shallowing of fault dips into the lower tip lines or detachment surfaces. The fault array appears to generally converge with depth and the frequency of conjugate/offsetting relationships consequently increases downwards into the lower Tertiary section (Figs 11 8: 12). Fault displacement also generally decreases downwards and many of the faults observed appear to terminate near or below the top Chalk horizon level (Fig. 12). The largest faults ultimately tip out or detach near the base Chalk / top of the Early Cretaceous sequence. It is not clear from seismic data whether the faults link into one or more semi-regional detachment surface(s), or whether more local tip zone strain accommodation at depth has instead occurred by bedding plane slip and volume contraction processes that are typically associated with faulting in under-compacted sediments. The general lack of listric fault geometries at depth and the absence of an observed contractional toe to the fault system (confirmed with 2D seismic lines), suggests that relatively local accommodation of strain has taken place and that major regional detachment surfaces are not necessarily present. Upwards, the faults do not often penetrate far beyond a tow amplitude (i.e. probably shale-dominated) package overlying the younger Early Eocene delta top sequences (Fig. 12). A further feature of these data are the complex relationships developed between interpreted faults and relatively localized high-amplitude, moderately-dipping reflectors of variable continuity seen within the shale-dominated Early Eocene package (Fig. 13). A full discussion of these features is beyond the scope of this work. Nevertheless, the variable continuity of highamplitude reflections on the seismic data (Fig. 13) suggests strongly that some faults have behaved as lateral barriers to fluid flow (at least in 2D), whereas other faults have not, and the same
124
G. JONES ETAL.
Fig. 10. NW-SE seismic line showing early-formed dipping (fault plane) reflections (often restricted to the older Eocene shale-rich package) which face up-dip (i.e. towards the basin margin) and that are cross-cut (offset) by later down to the basin faults, over a 10km length section. (A) There is a possible link in the initiation of this down to the basin fault activity to the underlying Mesozoic shelf margin step (arrowed). (B) Note the consistent truncation of upslope-facing faults by downsiope-facing faults. This line appears to preserve a significant amount of early stage fault activity (i.e. within the shale-rich package); many faults in the centre of the line do not penetrate far into the sand-dominated sequences. (C) Base of the delta top sand-dominated sequences. (D) Eocene graben system above the reactivated Jurassic rift fault. fault can display variable barrier/non-barrier behaviour with depth. Also of note is the apparent expansion of some of the highamplitude packages across faults (Fig. 13). It is not known whether the enhanced reflectivity is due to variation in rock properties (e.g. cementation) or to the presence of a fluid phase (or phases) in the system, or perhaps due to differential compaction. A high probability of fault-related hydrocarbon trapping by juxtaposition of sand bodies against mudstones (and by possibly clay smearing of sandstone units on faults) can be predicted in the older Early Eocene of sequences. In addition to fault-related trapping, stratigraphic trapping of hydrocarbons may occur in a
number of ways in environments such as those interpreted for the Tertiary of the area, e.g. by stratigraphic pinchout (up-dip and down-dip), onlap surfaces, erosional truncation and within isolated channel bodies or incised valleys (Van Wagoner et al. 1990). Considerable complexity in the distribution of such trapping mechanisms can also be expected in 3D.
Fault geometries, orientations and linkages The fault arrays have been mapped at several key stratigraphic levels and a variety of techniques have been utilized to aid
Fig. 11. NW-SE seismic section showing faulting in the Lower Tertiary sequences. The clearly imaged extensional faults within the high-amplitude package provide a significant degree of confidence in fault interpretations downwards into the older Early Eocene shale-rich packages. Similar geometries are seen on seismic traces (i.e. strike orientations) and the fault system is wider than the survey area in a NE-SW orientation (i.e. is laterally extensive parallel to the basin margin). (A) Complex conjugate fault intersections in the older Early Eocene. (B) N-S-oriented planar fault arrays in the younger Early Eocene.
FAULT ARRAY EVOLUTION, PORCUPINE BASIN
Fig. 12. S W - N E oriented (trace) section showing the Tertiary faulting. (A) Note the increased complexity of fault intersection geometries downwards into the shale-dominated package and the downwards termination of many of the dipping reflectors at or just below the top Chalk horizon. (B) Probable coherent seismic noise within the dataset. (C) Note many faults tip out in transparent package of Late Eocene-Oligocene age.
Fig. 13. Seismic line (depth converted in m) showing discontinuous high amplitude reflections within the shalerich package. (A) Note that some of these bright reflectors appear to terminate at the interpreted location of faults penetrating downwards from the sandrich package, where they are wellimaged. The interpreted location of faults is further evidenced within the shales by fault plane reflections (see text for discussion).
125
126
G. JONES E T A L .
Fig. 14. Time slices of the Edge Technology dataset, showing the general transition from linear to rectilinear/polygonal fault patterns which takes place downwards through the lower Tertiary section. Note that the NE-SW oriented elements of the deeper fault array trend sub-parallel to the shelf margin. (A) N-S and NE-SW faults in older Early Eocene sequences. (B) N-S faults in younger Early Eocene sequences.
understanding of 3D fault geometries and linkages. The Chevron Edge Technology volume in GoCad TM has proven particularly useful in visualizing the relationships between faults, especially in a vertical sense. Fault mapping confirms the general observation that the Tertiary fault array becomes increasingly complex with depth. An overview of the structure with depth is provided by a timeslices of the Edge Technology data (Fig. 14) across part of the survey, which show the complexity increasing down through the Eocene sequences. In the younger Early Eocene, the fault array is relatively easily characterized within this package of high-amplitude reflectors. Most larger fault segments at this level are oriented in an approximately N - S direction (Fig. 14). Occasionally, shorter, sigmoidal - N - S fault segments become aligned along both N E - S W and N W - S E trends at these levels, to form what appear to be larger-scale conjugate fault arrays. These - N - S faults link downwards in a complex fashion into the older Early Eocene and upper Paleocene structures. Both rhomboidal and polygonal fault patterns, defined largely by intersecting N - S and N E - S W fault segments, are present at deeper Tertiary levels on time slices (Fig. 14). Note that the NE-SW-trending fault elements in the lower Early Eocene section trend essentially parallel to the strike of the shelf, suggesting that slope may be exerting a greater influence on fault trend with depth. Although the overall pattern is relatively consistent with depth, the distribution of structures is 'domainal' across the 3D survey area at a given time interval, indicating there is a degree of along-strike variability in the factors controlling fault array development. The interpreted seismic volumes (including faults) have been imported into GoCad TM visualization software to interpret the complex 3D distributions and linkages between these features. Figure 15 demonstrates the nature of the complex fault linkages and segmentation at depth. In the younger sand-dominated sequences, fault orientations and geometries are relatively simple, comprising linear to sigmoidal individual fault segments. In the shale-dominated package, fault linkage is more common and both the fault orientations and geometries present
are more complex, generating the rectilinear to polygonal structural patterns in 3D (Fig. 15).
Fault activity history Although the majority of fault activity is interpreted to have occurred during the later Eocene to Oligocene interval (i.e. following deposition of the main Early Eocene progradationai packages), it is clear from the 3D seismic that fault growth and propagation was ongoing throughout the Tertiary. For example, in section, the upper tip lines of the faults in this area terminate at several different stratigraphic levels (e.g. Figs 16 & 17). This may be related to variable fault offset along strike, to faults linking in 3D, to spatially variable timing of fault activity, or possibly to factors such as lithological controls on propagation (i.e. some fault arrays may appear restricted to certain lithological packages). Accordingly, it can be difficult to date the timing of the last phase of activity on many faults. Activity is best evidenced by erosional truncation of faults (especially those which show significant offset below the erosion surface), or by the rare development of syn-sedimentary growth packages. Whilst emphasizing the progressive, ongoing nature of the faulting in this area, three main phases of Tertiary fault activity have been informally distinguished, mainly to aid understanding of structural development. These are now outlined below.
Phase 1 faults (active mostly in the ?upper Paleocene to older Early Eocene interval) A significant number of faults terminate near the base of the younger Early Eocene delta top sequences ('Phase 1' faults; Fig. 16), where they are considered likely to be erosionally truncated (Fig. 16). The erosional interpretation is evidenced by the anomalously rapid termination of fault offset below the point of truncation, instead of a gradual offset decline (Fig. 17). Faulting was therefore active during the initial influx of the younger Early Eocene clastic sequences. These earlier faults can be either landward or basinward-dipping, but are often seen to
FAULT ARRAY EVOLUTION, PORCUPINE BASIN
127
Fig. 15. Models to show the 3D distribution of the fault array with depth using the Edge Technology Icoherency) dataset in the GoCAD visualization package. (1) Time-slice and intersecting panels within the shallower Early Eocene sand-rich sequences. (A) The generally simple - N N W - S S E oriented fault segments in the shallow section (i.e. SE part of the image), pass locally downwards into "birds-foot" geometry linkage patterns (NW part of image). (2) Time-slice and panels transecting the deeper shale-dominated Early Eocene section. (B) Note the complex intersecting NE/SW and N/S faults forming intersecting lozenge-shaped fault intersection geometries in 3D. TM
Fig. 16. NW-SE seismic line showing faulting restricted mainly to the older Early Eocene shale-dominated interval ('Phase 1' faults). (A) Fault contained largely within the older Early Eocene low-amplitude package. (B) Faults that appear to be erosionally truncated by the earliest units of the younger Early Eocene sand-rich sequence.
128
G. JONES E T A L .
Fig. 17. SW-NE seismic trace showing the main 'Phase 2' faults penetrating through the younger Early Eocene sandrich sequences (e.g. centre of image). This line also shows variation in fault activity evidenced by termination of some faults at different points within the delta-top package. (A) Probable erosionally truncated fault, evidenced by the rapid vertical displacement decrease evident.
be landward-dipping on a regional scale (Fig. 10). Many such "early-formed' faults have presumably been later re-utilized during ongoing extension. Other faults are restricted to the older Early Eocene package because conjugate fault development terminates individual segments (Fig. 16); the age of activity on these is uncertain.
Phase 2 faults (active mostly in the younger Early Eocene to Oligocene interval) The main phase of faulting occurred during the late Early Eocene to Oligocene interval and generated numerous normal faults that propagated throughout the sand-rich younger Early Eocene delta top sequences (Figs 11 & 12). Although the 'tectonic non-growth' classification based on 2D seismic (McCann et al. 1995) is broadly correct, in detail on the 3D dataset there is variable activity seen on individual faults. For example, evidence for erosional truncation of some faults is
noted high in the sand-rich sequences (Fig. 17), indicating there was interaction between fault growth and deposition.
Phase 3 faults (active in the Oligocene-Miocene interval) Rare structures persist higher in the sequence, above a major regional unconformity (Oligo-Miocene) to shallower stratigraphic levels ('Phase 3' faults), the most prominent of which is the eastern bounding fault system of the Porcupine High. This fault forms a complex zone of deformation within the later Tertiary sequences (Fig. 9), with several phases of movement indicated. Other shallower fault arrays present in this area appear to be vertically restricted within lithological packages, several examples of which are seen on seismic in the shallower part of the Tertiary (Oligo-Miocene). On the outer part of the shell basin-scale erosional scour may have removed local evidence of faulting of this age (see Figs 7 & 8).
FAULT ARRAY EVOLUTION. PORCUPINE BASIN
Controls on fault array evolution The processes contributing to the generation and development of faulting in the study area are now considered, before an evolutionary model for the fault array is proposed. Firstly, it is appropriate to consider the general setting of the Porcupine Basin in the Tertiary. It is clear from the preceding discussion that both regional tectonic processes (e.g. uplift, extension) and non-tectonic processes (e.g, gravitational collapse, compaction over underlying topography) could potentially have influenced the initiation, trends and development history of the fault array in this area during the Early Eocene to Oligocene. The following key observations are considered important to understanding the generation and evolution of the fault array: 9 There is significant regional tilt (5 ~ towards the basin centre (i.e. from NW to SE). This slope could promote gravitydriven slip and fault initiation. 9 The trend of some of the deeper faults parallels the basin flank, also suggesting basin margin tilt/slope is important in controlling the generation of faulting. 9 2D regional lines prove that the fault arrays do not extend entirely across basin; instead 'mirror image' fault arrays occur on the opposite flank of the basin, separated by undeformed sequences in the basin centre. This indicates instability on the basin flanks was important to the development of faulting. 9 Increased complexity in fault linkages/orientations is seen with depth, prior to relatively rapid disappearance of the faulting. This indicates that changes in the volume and shape of the fault blocks has taken place to accommodate the strain. 9 Earlier faults throw up-dip towards the basin margin, late faults mainly throw down to the basin centre. 9 Local reactivation of Mesozoic syn-rift extensional faults influences the shallower Tertiary array (potentially generating tectonic instability in the cover sequences). The observations listed above suggest that downslope gravitational collapse of the sediment pile was the dominant driving mechanism for the Tertiary faulting. There is a striking resemblance between analogue sandbox models for gravitational sliding created by Higgs & McClay (1993) and fault geometries and activity observed on seismic in the Porcupine Basin (compare Fig. l0 and Fig. 18). Higgs & McClay (1993) modelled extensional faulting above a tapered wedge to research the evolution of a fault array interpreted on 2D seismic data in the Outer Moray Firth. In that area, faults both face and throw predominantly in an up-dip direction. The faults were planar rather than listric in nature, and there was a decrease in brittle deformation observed with depth, although there was no common detachment horizon. These authors also observed a relationship between the trend of the faulting and the underlying sequences, suggesting a gravitational mechanism. A regional Middle Miocene tilting event is considered to have triggered fault development. In contrast to the Porcupine Basin, the Moray Firth stratigraphic sequence changes from lower sand-dominated packages to overlying more shale-dominated sequences. The analogue modelling predicts landward-dipping planar faults for small to moderate degrees of extension (10-30%). As extension increases (50%), the landwards-facing faults become more listric in nature and basinward-facing faults develop (Higgs & McClay 1993; their figs 9-11 ). Later faults generated by further extension were antithetic or downdip facing (i.e. towards the basin centre). In the model,
129
faults evolve from planar to listric as it develops (i.e. the top of model moves faster than the base), which is likely enhanced by increased compaction with depth. Faults can however also steepen and splay downwards in the model. In the Porcupine area, major variations in controls such as lithology, degree of compaction and lithification, amount of basal friction in the system and other factors such as synchronous reactivation of Mesozoic rift faults, have generated a more complex system in detail, but the mechanism described by Higgs & McCiay (1993) appears to fit very well with seismic observations. An important element in the formation of the Porcupine fault system is considered to be selective tectonic reactivation of the Jurassic syn-rift extensional fault array. The basement Jurassic rift faults are intimately linked with the evolution of the shallower structures, as is demonstrated clearly in Figures 8 & 9. These faults control the intensity, the local symmetry in throw direction and also the overall geometry of the fault arrays, notably at the reactivated Mesozoic shelf edge. It is considered likely that tectonic reactivation of Jurassic faults contributed to the generation of slope instability within the Tertiary sequences. Sub-regional or more localized detachment levels, along which slip has been accommodated, could potentially be present near the top Chalk/Paleocene and possibly at the top of the Early Cretaceous package. If so, these surfaces could hard-link into the reactivated Jurassic rift faults. Alternatively, deformation may be accommodated by other mechanisms. Higgs & McClay (1993) have described a model of non-rigid fault block rotation for the under-compacted sediments of the Moray Firth. In this model, pure shear at the base of the section and simple shear at the top (due to a vertical compaction gradient) means the fault blocks undergo an angular shear rotation, and change both shape and volume. This change can be accommodated by flexural slip, bulk flexurai shear, or compaction at the downslope limit of the deforming sequence. A regional basal detachment is therefore not a requirement of this model and instead the faults reduce in displacement downwards and terminate onto a common tip line. Such a model would describe well the variable geometry (trend) of faults seen with depth down through the Porcupine array and removes the need for an extensive regional detachment surface. Hydro-fracturing within overpressured mud-dominated sequences is a further important process, sometimes generating (less common) larger faults that can link vertically though intervening sand packages (e.g. Cartwright 1994: Lonergan & Cartwright 1999). Observations that suggest hydro-fracturing was unlikeh' to be the critical driving mechanism in the study area include the fact that the faults are currently not strongly layer-bound and also that they are not truly polygonal in geometry (i,e. relative to the fault arrays described by Cartwright 1994). Furthermore. the Porcupine faults do not extend right across the deepest part of the basin (as is observed in the central North Sea), as might be expected to occur if the hydro fracture mechanism was dominant in controlling fault initiation and development. Nevertheless, loading and differential compaction (and thus possibly overpressuring) of the dominantly fine-grained sequences during basin subsidence and tilt could still have contributed to the initiation and development of the fault array in this area. Although there is currently no fluid overpressuring to act as a drive mechanism for hydro-fracturing in the Tertiary of the Porcupine Basin, this cannot be discounted as a secondary mechanism contributing to faulting, especially during the early stages of fault nucleation and growth and prior to widespread linkage. Indeed. layer-bound polygonal fault arrays are observed in the Oligocene in parts of the study area that may have been generated by hydro-fracturing processes.
130
G. JONES ETAL.
SANDBOX EXPERIMENT 1 0% EXTENSION
15% EXTENSION
30% EXTENSION
50% EXTENSION
50% EXTENSION INCREMENTS
FAULT
f
,i
NUMBER],
7-12
19-24 --' --
13-18 ----
2 5 - 3 0 ...........
34 32 312524
1328
1618
26 1 4 1 1 2 7 1 2 0
29 2 1 9 1 7
It is difficult to correlate the timing of faulting in the study area (i.e. mainly Late Eocene-Oligocene) specifically to the regional geological evolution described earlier. Several of the documented regional events could have been linked directly or indirectly to the evolution of the fault array: for example the transient regional uplift and subsidence described by Jones et al. (2001) in the Paleocene and Early Eocene may initially have generated instability on the margin of the basin. More work is required to investigate potential driving mechanisms on a semiregional scale.
Model of fault array evolution A suggested model for fault array development is shown in Figure 19a-c.
'Phase 1' (end lower shale~base delta top sequence times; Fig. 19a) In the Early Tertiary, a basinwards slope was present in the area, as a result of ongoing subsidence and/or a lack of sediment
5
3
6
7
22
Fig. 18. Sandbox models of gravitational faulting from Higgs & McClay (1993). Compare to seismic line in Fig. ! 0. Sandbox model with basal dip of 5~ initially produced faults that mostly face up-dip (lower diagram). Later faults are antithetic/down-dip facing (and initially more commonly located in the up-dip area).
supply to infill the significant Early Cretaceous rift topography. Of primary importance is the clear evidence of initial faulting which facing directions and throws that are down to the basin margin (i.e. up-dip), which is locally preserved, as described above (Fig. 10). At this time, reactivation of the Mesozoic fault array may also have initiated primary instability in the sediment pile. The notable symmetry of fault facing direction in the lower Tertiary package, centred around the reactivated Mesozoic fault system on the basin margin (e.g. Fig. 8), suggests this Mesozoic fault may have been active and already have penetrated into the Cretaceous and lowest Tertiary by this stage. In addition, down to the basin throwing faults were probably becoming important, forming as conjugates or simply cross-cutting the earlier opposed structures. Initial differential compaction of the Cretaceous sequence may also have played a role in fault development along the basin margin at this stage. Regional tilting of late Eocene age could also have occurred, correlating with a deepening of the basin to the south (Moore & Shannon 1992) and preceding the more obvious tilting observed in the basin in Oligocene and Miocene times (McDonnell & Shannon 2001 ).
FAULT ARRAY EVOLUTION. PORCUPINE BASIN
131
conjugate fault relationships exist where both sets of faults were active, or where new antithetic faults formed (i.e. faults antithetic to the propagating basinwards-dipping faults). Gravitational slip processes were dominant at this time, as well as somewhat increased differential compaction due to loading of the mostly fine-grained Cretaceous and Early Tertiary sequences by younger Early Eocene delta-top clastic sediments. The differential compaction results from major depositional changes in Cretaceous sediment thickness from shelf to basin, leading to structural accommodation in the overlying Eocene sequences. Overall, the sediments probably remained relatively under-compacted during development of the faulting, especially if overpressures were able to develop within the deeper section. The fault array in this area simplifies upwards, suggesting that there is a stronger pre-Teniary influence (increased friction, increased sediment compaction and/or deep faulting) on structural development in the chalk and lowermost Tertiary section. The fault array was able to develop more freely at shallower levels (where simple geometries are seen) but was "pinned' at depth at Cretaceous levels. The more complex fault geometries and orientation distributions seen with depth have developed as a result of this vertical change in the degree of freedom for downslope movement present within the sediment package. It is further evident that variations in the nature and timing of fault activity are present across the 3D area, related to individual structural domains. The structural domains are in turn controlled by factors such as the geometry and distribution of the reactivated Jurassic rift faults, the topography on the shelf at various stratigraphic levels, major lithological changes and variations in the degree of compaction.
'Phase 3' (minor fault reactivation from MiocenePliocene; Fig. 19c)
Fig. 19,
Evolutionary model for fault array development in Tertiary sequences of the Porcupine Basin. Active faults are shown in different colours per time interval. (A) Suggested model for initial configuration of faulting (Phase 1; immediately after lower shale-dominated unit deposition). Early up-slope facing faults are cross-cut by later down to the basin features, especially around reactivated Mesozoic syn-rift faults. The latter could have initiated the instability in the Tertiary sequence, then penetrated upwards to become hard-linked into the developing shallower array. (B) The major phase of gravitational faulting ('Phase 2') shown in light green, mainly down to the basin structures and associated conjugates, which have offset and re-utilized the early-formed structures. (C) A limited number of structures ('Phase 3') penetrate through a Miocene unconformity to shallow Tertiary levels (dark green).
'Phase 2' (mainly end delta top sequence times; Fig. 19b) Following initial development of the array, some faults were active during deposition of the younger Early Eocene deltaic sequences (see Fig. 17). However, the end of this interval of clastic influx was the main phase of fault activity in this part of the Porcupine Basin (i.e. Eocene-Oligocene; Fig. 19b). Many of the faults active at this stage are basinwards-facing and can be demonstrated to cross-cut the early landwards-facing faults, as was clearly illustrated in Figure 10. More complex
Faulting continued into the Oligo-Miocene (Fig. 19c). This activity is mostly associated with structures that link to older Mesozoic rift faults. Vertically extensive faults are seen in parts of the 3D area, which penetrate to very shallow levels, including the Porcupine High bounding fault (Fig, 9), In the OligoMiocene, faulting (probably associated with sediment compaction) has also generated mainly layer-bound arrays of smalloffset normal faults, which are comparable in geometry to those described by Cartwright (1994) in sequences of the same age in the central North Sea.
Summary Fault array geometries seen in the Cretaceous and lower Tertiary in the northwest Porcupine Basin are interpreted to have resulted primarily from gravitational collapse of the sediment pile to the east. Downslope movement may have been initiated by a number of regional and/or more local driving mechanisms. The most likely of these are slope instability during basin subsidence and tilting, together with selective local reactivation of the Jurassic fault array. Differential compaction of Cretaceous and Lower Tertiary sediments over pre-Cretaceous rift-related topography also resulted in decreased freedom for downslope rnovement with depth in the sediment package, which could explain the increased downwards complexity observed in the fault array. Major detachments are not likely to be present at the base of the system: instead deformation is considered to have been accommodated by flexural slip, bulk flexural shear, or compaction at the downslope limit of the deforming sequence (non-rigid fault block rotation), with faults merging at common tip lines.
132
G. JONES ET AL.
The authors would like to express their thanks to Chevron Europe and the Irish Offshore 5/95 licence partners (Dana Petroleum PIc., Conoco UK Ltd, Enterprise Energy Ireland Ltd, Statoil Exploration (Ireland)), to the Irish offshore 8/95 licence partners and also to PGS Ltd, for permission to present this data. The interpretations presented here are those of the authors and do not necessarily represent the views of the 5/95 partnership. J. Vermeulen and D. Lewis of ChevronTexaco Upstream Europe have also helped greatly during the work. Finally, P. Shannon and K. McClay are thanked for their careful reviews of the manuscript.
References CARTWR1GHT, J. A. 1994. Episodic basin-wide hydro-fracturing of overpressured early Cenozoic mudrock sequences in the North Sea Basin. Marine and Petroleum Geology, 11(5), 587-607. CROKER, P. F. & SHANNON,P. M. 1987. The evolution and hydrocarbon prospectivity of the Porcupine Basin, offshore Ireland. h~: BROOKS, J. & GLENNIE, K. W. (eds) Petroleum Geology of Northwest Europe. Graham and Trotman, London, 633-642, DANCER, P. N., ALGAR, S, T. & WILSON, I. R. ] 999. Structural evolution of the Slyne Trough. In: FLEET, A, J. & BOLDY, S. A. R. (eds) Petroleum Geology of Northwest Europe, Pivceedings of the 5th Conference. Geological Society, London, 445-454. DORI~, A. G., LUNDIN, E. R., JENSEN, L. N., BIRKELAND.O., 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, 41-62. FLEET, A. J. & BOLDY, S. A. R. (eds) 1999. Petroleum Geologyof Northwest Europe. Proceedings of the 5th Conference. Geological Society, London, 41-62. HIGGS, W. G. & MCCLAY, K. R. 1993. Analogue sandbox modelling of Miocene extensional faulting in the Outer Moray Firth. In: WILLIAMS, G. D. & DOBB, A. (eds) Tectonics and Seismic Sequence Stratigraphy. Geological Society, London, Special Publications, 71, 141 - 162. JOHNSON, H., RITCHIE, J. D., GATLIFF, R. W., WILLIAMSON. J, P., CAViLL, J. & BULAT, J. 2001. Aspects of the structure of the Porcupine and Porcupine Seabight basins as revealed by gravity modelling. In: SHANNON, P. M., HAUGHTON, P. D. W. & CORCORAN, D. V. (eds) The Petroleum Erploration of h'ehmd's Offshore Basins. Geological Society, London, Special Publications, 188, 265-274. JONES, S. M., WHITE, N. & LOVELL, B. 2001. Cenozoic and Cretaceous transient uplift in the Porcupine Basin and its relationship to a mantle plume. In: SHANNON, P. M., HAUGHTON, P. D. W. & CORCORAN, D. V. (eds) The Petroleum Exploration of lrehmd's Offshore Basins. Geological Society, London, Special Publication, 188, 345-360. LONERGAN, L. & CARTWRIGHT, J. A. 1999. Polygonal faults and their influence on deep water sandstone reservoir geometries, Alba Field, United Kingdom Central North Sea. AAPG Bulletin, 83(3), 410-432. MCCANN, T., SHANNON,P. M. & MOORE, J. G. 1995. Fault styles in the Porcupine Basin, offshore Ireland: tectonic and sedimentary controls. In: CROKER, P. F. & SHANNON,P. M. (eds) The Petroleum Geology of Ireland's Offshore Basins. Geological Society, London, Special Publications, 93, 371-383. McCLAY. K. R., DOOLEY, T., WHITEHOUSE, P., FULLARTON, L. & CHANTRA PRASERT, S. 2004. 3D analogue models of rift systems: templates for 3D seismic interpretation. In: DAVIES, R. J., CARTWRIGHT, J. A., STEWART, S, A., LAPP1N, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exl~loration
of Sedimenmr)" Basins. Geological Society, London, Memoirs, 29, 101-115. MCDONNELL, A. & SHANNON, P. M. 2001. Comparative Tertiary stratigraphic evolution of the Porcupine and Rockall Basins. In: SHANNON, P. M., HAUGHTON, P. D. W. & CORCORAN, D. V. (eds) The Petroleum Exploration of Ireland's Offshore Basins. Geological Society, London, Special Publication, 188, 323-344. MOORE, J. G. & SHANNON, P. M. 1991. Slump structures in the Late Tertiary of the Porcupine Basin, offshore Ireland. Marine and Petroleum Geology, 8, 184-197. MOORE, J. G. & SHANNON, P. M. 1992. Palaeocene-Eocene deltaic sedimentation, Porcupine Basin, Offshore Ireland. First Break, 10, 461-469. NAYLOR, D. & ANSTEY, N. A. 1987. A reflection seismic study of the Porcupine Basin, offshore West Ireland. Irish Journal of Earth Sciences, 8, 187- 210. NAYLOR, D.. SHANNON, P. & MURPHY, N. 1999. Irish Rockall Region-a Standard Structural Nomenclature. Petroleum Affairs Division, Special Publication. 1/99. NAYLOR, D., SHANNON, P. & MURPHY, N. 2002. Porcupine-Goban Region--a Standard Structural Nomenclature System. Petroleum Affairs Division (Ireland), Special Publication, 1/02, 65pp and 2 Enclosures. ROBERTS. D. G., THOMPSON, M., MITCHENER, B., HOSSACK, J., CARMICHAEL, S. & BJORNSETH, H.-M. 1999. Palaeozoic to Tertiary rift and basin dynamics, mid Norway to the Bay of Biscay--a new context for hydrocarbon prospectivity in deep water frontier, bl: FLEET, A. J. & BOLDY, S. A. R. (eds) Petroleum Geology of Northwest Europe, Proceedings of the 5th Conference. Geological Society, London, 7-40. SHANNON, P. M.. MOORE, J. G., JACOB, A. W. B. & MAKRIS, J. 1993. Cretaceous and Tertiary basin development, west of Ireland. In: PARKER, J. R. (ed.) Petroleum Geology of NW Europe: Ppvceedings of the 4th Conference. Geological Society, London, 1057-1066. SHANNON, P. M., HAUGHTON, P. D. W., CORCORAN, D. V. (eds) 2001. The Petroleum Erploration of Ireland's Offshore Basins. Geological Society, London, Special Publications, 188. STOKER. M. S. 1997. Mid-Late Cenozoic sedimentation on the continental margin off NW Britain. Journal of the Geological Societ)', London, 154, 509-515. STOKER, M. S., VAN WEERING, T. C. E. & SVAERDBORG, T. 2001. A Mid-Late Cenozoic tectonostratigraphic framework for the Rockall Trough. In: SHANNON, P. M., HAUGHTON, P. D. W. & CORCORAN, D. V. (eds) The Petroleum Erploration of Ireland's Offshore Basins. Geological Society, London, Special Publications, 188. TATE, M. P. 1993. Structural framework and tectonic-stratigraphic evolution of the Porcupine Seabight Basin, offshore Western Ireland. Marine and Petroleum Geology, 10, 95-123. TATE, M. P. & DOBSON, M. R. 1989. Late Permian to early Mesozoic rifting and sedimentation, offshore NW Ireland. Marine and Petroleum Geology, 6, 49-59. VAN WAGONER, J. C., MITCHUM, R. M. JR., CAMPION, K. M. JR. & RAHMANIAN, V. D, 1990. Siliciclastic sequence stratigraphy in well logs, cores and outcrop: concepts for high resolution correlation of time and facies. AAPG Methods in Exploration Series 7, Tulsa, WHITE, N., TATE, M. P. & CONROY, J. J. 1992. Lithospheric stretching in the Porcupine Basin, West of Ireland. In: PARNELL, J. (ed.) Basins on the Atlantic Seaboard: Petroleum Geology, Sedimentology attd Basin Evolution. Geological Society, London, Special Publications. 62, 327-333.
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 DAVID
M. DUTTON
1"3, D U S T I N
LISTER
1"4, B R U C E
D. T R U D G I L L
1"5 & K A P E L A
PEDRO 2
IDepartment of Earth Sciences and Engineering, hnperial College, RSM Building, Prince Consort Road, South Kensington, London SW7 2BP, UK (e-mail: david.dutton @ ic.ac, uk) 2Sonangol DPP, Rua 1 ~ Congresso do MPLA, N. ~ 8-16, Caixa Postal 1316, Luanda, Reptiblica de Angola 3present address: CNR International UK Ltd, St Magnus House, Guild Street, Aberdeen AB11 6N J, UK (e-mail: david, dutton @cn rinte rnational, com ) 4present address: WesternGeco, Schlumberger House, Buckingham Gate, Gatwick Airport, West Sussex RH6 0NZ, UK 5Present address." Colorado School of Mines, 1500 Illinois Street, Golden, CO 80401-1887, USA
Abstract: We investigate fault growth and linkage during developmentof a rafted terrain in the Lower Congo Basin. offshore
Angola. Miocene thin-skinned extension has led to the development of isolated raft blocks separated by a graben filled with syn-deformational strata. Angular unconformities together with thinning and onlapping of intra-rafi strata onto salt bodies suggest that thick salt was mobile during thin-skinned extension. 3D fault array geometries and displacement patterns record the subsequent deformation history of the graben during further thin-skinned extension. The mode of thin-skinned extension has important consequences for the Neogene turbidite hydrocarbon play associated with the rafted province of the Lower Congo Basin. The presence of thick mobile salt will influence pre-salt source rock maturation and the development of pre-salt/post-salt hydrocarbon migration windows.
Better understanding of the structural evolution and structural styles associated with thin-skinned extension is crucial for continued exploration success on the Angolan margin. Gravityinduced, thin-skinned extension in the form of rafting has exerted an important control on sediment dispersal and turbidity flow patterns of the Neogene turbidite sand play within the extensional domain of the Lower Congo Basin, offshore northern Angola (Anderson et al. 2000). Current evolutionary models used to explain present-day structural geometries associated with raft tectonics fall into two end-member categories: (1) an extensional raft model (Fig. la); (2) a salt rise and fall model (Fig. lb). The principal distinction between the two models is the amount of salt present in the system and its behaviour during thin-skinned extension. This distinction has important implications for the hydrocarbon system associated with the post-salt Neogene turbidite play due to the requirement of a pre-salt source rock charging a post-salt reservoir (Fig. 2). Large amounts of salt will significantly reduce the maturation of a pre-salt source rock due to the high thermal conductivity properties of salt (6.0 W/m K at surface, 4.0 W/m K at 130~ In addition, the ability of salt to act as impermeable barrier to vertical hydrocarbon migration requires the generation of a direct pre-salt/post-salt interface, usually as a result of extensive withdrawal of the salt layer (Rowan et al. 1999). Understanding the mode of rafting is therefore important for a critical analysis of the Neogene hydrocarbon play. Recent insights into raft tectonics from 3D geometrical restorations address the 3D structural framework and movement history of a raft/listric fault system from offshore West Africa (Rouby et al. 2002). Commercial structural restoration software, although crucial in understanding the evolution of structurally complex terrains, requires simplification and/or limiting of fault populations. This drawback can ignore crucial information associated with rafting that complex faulting records. 3D seismic interpretation and analysis of displacement patterns of
a salt related growth-fault array from the Gulf of Mexico shows that 3D fault array geometries can be complex with both lateral and vertical branching and linkage of fault segments (Rowan et al. 1998). The evolution of fault segments has implications for understanding the history of hydrocarbon migration and entrapment. The past 15 years have seen rapid and fundamental advances in our understanding of the geometry and evolution of normal fault arrays. Field observations show that fault displacement varies considerably along segmented faults and suggest that a strong relationship exists between fault slip, segment geometry and arrangement of the array of segments (Peacock & Sanderson 1991, 1994; Trudgill & Cartwright 1994; Cartwright et al. 1996). Three-dimensional, mechanical studies of fault arrays, rather than individual faults, have led to the recognition of the kinematic coherence of an array, in which the sum of individual displacement/length profile of a fault array generally represents that of an idealized isolated fault segment (Willemse et al. 1996). Here, we present a detailed 3D geometrical and kinematical analysis of a salt-related fault array that has developed in response to rafting during gravity-driven thin-skinned extension. Our approach is to investigate fault growth and linkage in the cover sequence above a raft system, together with analysis of the internal structural/stratigraphic geometries of rafts in order to assess the extent of salt mobility during the development of a rafted terrain. We use a 3D seismic dataset acquired in 1996 by Ranger Oil Ltd (now CNR International UK Ltd). Recent techniques in 3D fault displacement mapping allow a rapid and accurate appraisal of fault displacement from interpretations of 3D seismic data across a large area (Lister 2004). We present an application of these techniques with the generation of a 3D displacement model of a raft-related fault array in order to provide insights into the kinematics of fault growth associated with thin-skinned extension.
DAVIES,R. J., CARTWR1GHT,J. A., STEWART,S. A., LAPPIN,M. & UNDERHILL,J. R. (eds) 2004.3D Seismic Technology: Application to the E.wIoration of Sedimentars' Basins. Geological Society, London, Memoirs, 29, 133-142. 0435-4052/04/$15 9 The Geological Society of London 2004.
134
D.M. DUTTON ETAL.
Fig. 1. Two end-member evolutionary models for raft tectonics (adapted from Jackson & Cramez 1989): (a) extensional mode (thin salt). (b) salt diapir rise and fall mode (thick salt).
Geological setting
1966: Burke 1975; Roberts 1975: Burke 1996; De Matos 1999; Karner & Driscoll 1999). These three phases are associated with pre-salt, salt and post-salt stratigraphic units respectively and record the transition from the development of a terrestrial rift basin to open marine conditions (Fig. 4). The rift phase represents generation of deep under-filled lacustrine basins, predominantly filled with Neocomian fluvialdeltaic sandstones, conglomerates and lacustrine shales. The cessation of rifting in the Early Aptian marked the onset of a series of marine transgressions. Evaporitic sediments of the Late Aptian Loeme Formation define the rift-drift transition phase. The drift phase represented a major marine transgression following a Late Cretaceous eustatic sea-level rise. A carbonate ramp depositional setting developed during the Albian but from the Late Cenomanian clastic sedimentation became dominant over carbonate production as sea level started to fall. This relationship continues to the present day with deposition of slope facies muds. The Congo River supplies the majority of clastic sediment to the basin today.
The study area is located within the southern part of the Lower Congo Basin, offshore Angola in water depths ranging from 500 to 1000 m (Figs 3a & 3b). The Lower Congo Basin extends from northern Angola to Gabon and is one of a series of north-south trending sub-basins, located on the passive Atlantic margin of West Africa. Initiation of the break-up of Gondwana during the Early Cretaceous represents the commencement of Lower Congo Basin formation (Brice et al. 1982: Moore 1988; Standlee et al. 1991).
Tectono-stratigraphic setting Three tectono-stratigraphic phases characterize the evolution of the Lower Congo Basin: (1) Early Cretaceous (Neocomian to Lower Aptian) rift phase; (2) an Upper Aptian rift-drift phase: (3) an Albian-present day post-rift phase (Brognon & Verrier
J Compressional Province J W
Diapiric Province Clashc Reservoir
=
]
Extensional Province
I Post-soltOligo-Miocene]
]
J
I Post-saltAlbian Corbonate I and Clastic Reservoir /
4_
Fig. 2. W - E geo-section across the Lower Congo Basin showing variation in post-rift structural styles. Reservoir distributions and source rock requirements associated with the Neogene Turbidite sand play are also shown. The extensional province requires a pre-salt source rock to charge a postsalt reservoir. Hydrocarbon exploration plays in the diapir and compressional provinces utilize a post-salt source rock charging a post-salt reservoir. Figure courtesy of CNR UK International Ltd.
3D GEOMETRY OF A RAFT-RELATED FAULT ARRAY
135
Variation in the structural style of post-salt/post-rift sediments within the Lower Congo Basin allows identification of distinctive structural provinces (Fig. 2). Diapiric salt structures (diapir province) and thin-skinned compressional structures (compressional province) represent basinward accommodation of up-dip thin-skinned extension associated with the raft province (Spathopoulos 1996: Cramez & Jackson 2000). Thin-skinned compression and extension are both detached above an evaporitic level. Valle et al. (2001) propose further subdivision of the eastern Lower Congo Basin into structural domains characterized by different ages and styles of structural elements. From east to west these are: (1) Cretaceous raft blocks/grabens: (2) Tertiary raft blocks/grabens: (3) Tertiary pre-rafts: (4) salt diapirs. In the Lower Congo and Kwanza Basins, thin-skinned extension initiates during the Albian in response to an increase in westward tilting of the detachment surface during thermal relaxation and subsidence of the basins (Duval et al. 1992: Lundin 1992: Spathopoulos 1996: Eichenseer et al. 1999). Further episodes of westward directed tilting of the dficollement surface occur during the Miocene because of uplift of the Southwest African craton (Lunde et al. 1992). This later basin-tilting phase results in renewed and accelerated gravity sliding of the post-salt following additional Miocene sedimentary loading and progradation (Lundin 1992: Anderson et al. 2000).
Database and interpretation methodology
Fig, 3. (a) Location of the Lower Congo Basin and the study area. (b) Map showing distribution of rafts and grabens within the study area. The dashed box defines the areal extent of 3D seismic data used in this study. Locations of displayed seismic profiles are shown.
The seismic dataset for this study, from CNR International UK Ltd (formerly Ranger Oil UK Ltd), consists of 650 square kilometres of 3D seismic data acquired in July and August of 1996 (Fig. 3b) with bin dimensions of 12.5m in the inline (NW-SE) direction and 25.0 m in the trace (NE-SW) direction. Horizon and fault interpretation exists on every 10th trace/line (i.e. every 250/125 m) with more closely spaced traces, lines and arbitrary lines utilized in areas of structural complexity. The use of coherency slices, time slices and amplitude extractions aided the correlation of faults and improved fault tip-line definition. Well data exist for pre-deformational and syn-deformational strata from five wells drilled by Ranger Oil Ltd (now CNR International UK Ltd). Biostratigraphic data exists for sequence boundary correlation across the study area. Vertical seismic profiles and calibrated velocity data allow matching of biostratigraphical and lithological tops to the seismic. In addition, further interpretation of correlateable horizons delineates fault frameworks for additional structural levels. The seismic data quality is generally very good with excellent definition of fault plane reflections within the grabens.
Structural setting
3D displacement mapping
The study area is characterized by two Miocene cored grabens separated by an isolated fault block comprising AlbianOligocene strata (Fig. 5). The structural style of the study area is associated with an extreme form of thin-skinned extensional tectonics known as raft tectonics. Raft tectonics refers to the generation of isolated allochthonous fault blocks/rafts that separate to the extent that they are no longer in mutual contact (Duval et al. 1992; Lundin 1992; Spathopoulos 1996: Anderson et al. 2000; Rouby et al. 2002). The lateral movement of rafts result in the generation of grabens containing syn-deformational strata. With less extension, fault blocks that are still in mutual contact are termed pre-rafts (Duval et al. 1992).
Mapping fault displacement distributions helps identify mechanisms that can cause slip distributions to deviate from an ideal elliptical shape, so that growth models based on the observation of displacement patterns can be proposed. Our fault displacement analysis uses a new automated skeletonization algorithm that allows mapping of displacement information from horizon interpretations onto fault trace skeletons (Lister 2004). Generation of displacement contours across a whole fault surface can then be undertaken following integration of this dataset with 3D fault surfaces. Lister (2001) provides a full discussion of the method (procedures, validation and applicability).
136
D.M. DUTI'ON ETAL.
Fig. 4. Post-tiff stratigraphic framework for the study area.
Structural analysis
Rafi-graben structural framework A N E - S W seismic profile demonstrates the post-salt structural configuration of the study area (Figs 5a & 5b). The profile shows the existence of two rafts separated by a graben of Miocene to Recent syn-deformational strata. These rafts have glided along a westward dipping detachment surface of evaporitic sediments. The crossover of two master raft-bounding listric faults within the Pliocene-Recent stratigraphy defines a conjugate fault system above the eastern raft block (Fig. 5). The development of this conjugate fault system also results in the generation of a triangular-shaped graben with internal faulting. Identification of such a feature in a seismic volume may be a characteristic signature of a raft block present within the stratigraphy. The eastern raft has started to separate/raft internally, resulting in the development of pre-rafts. The principal observation from this seismic profile is that the basinwarddipping (master listric fault) at the eastern margin and two antithetic faults at the western margin define the graben. Both of these fault-sets detach at depth along the flanks of relict salt pillows before flattening onto an evaporitic d~collemerit surface. Only the graben defining faults link at depth onto the d~collement surface whereas the inner graben faults do not. A well-developed, mixed antithetic and synthetic fault system characterizes the inner graben fault array. It is clear from the
seismic profile that a complex relationship exists between the inner graben and graben-bounding fault types. The style of inner-graben faulting shown in Fig. 6 exemplifies this relationship. Two synthetic splay faults link with the antithetic Fault F progressively with depth. Fault F has a lower tip point (point of zero displacement) within in the high-amplitude seismic package defined by the Upper Miocene marker 2 horizon and the 10.5Ma S.B. Below this tip point, synthetic Fault G undercuts antithetic Fault F and links with an adjacent antithetic Fault H with depth. This relationship repeats itself again with Fault H undercut by another synthetic fault (Fault J) that then also links with another antithetic fault with depth (Fault K). This pattern of undercutting and relaying of linkage points (Fig. 6) reflects the internal deformation history of the graben block with both the graben-defining fault sets competing for control of graben subsidence during the Miocene. The source of this competition may be the nature of salt withdrawal/migration patterns associated with reduction of a salt wall/diapir during the Miocene (see Fig. l b). A characteristic feature of the salt rise and fall mode of raft tectonics is the switching of fault polarities and stratigraphic thickening trends as salt migrates out of the province with continued extension (Lundin 1992).
Stratigraphic growth patterns Thickening trends within the study area indicate a contrasting style of Tertiary sedimentary graben fill through time
3D GEOMETRY OF A RAFT-RELATED FAULT ARRAY
Fig. 5. (a) Post-rift structural style: un-interpreted NE-SW seismic profile. (b) Post-rift structural style: interpreted NE-SW seismic profile. Location of Figure 6 is also shown. Note development of two isolated rafts separated by a graben filled with syn-deformational strata of Miocene to Recent age. The eastern margin of the graben is defined by a basinward dipping listric fault (black) and the western margin is defined by a landward dipping antithetic fault. Both these faults detach at depth along the flank of relict salt pillows before flattening out onto a dEcollement surface.
Fig. 6. Seismic profile showing the variation in fault displacements and geometries of inner graben faults with increasing depth. This figure also shows the complex relationship that exists between antithetic and synthetic inner graben faults. Synthetic faults progressively undercut antithetic faults and relay antithetic/synthetic linkage points. Variations in stratigraphic thickening trends and patterns can also be clearly observed.
137
138
D.M. DUTTON ETAL.
Fig. 7. Internal structural/stratigraphic geometries of rafted strata:(a) un-interpreted seismic profile of raft blocks; location of line is shown in Figure 3b; (b) Interpretation of intra-raft structural and stratigraphic geometries. Note the presence of strong angular unconformities within the Albian-Eocene strata of the SW raft. Additionally, the AlbianEocene strata of the NE raft shows dramatic thinning towards the salt pillow. These observations may indicate significant salt growth during Albian-Eocene sediment deposition.
(Figs 5 & 6). In particular, the middle Miocene stratigraphic unit defined by the 10.5Ma SB and the 16.5Ma SB shows the largest amount of thickness variation across eastern and western margins of the graben. The middle Miocene package is stratigraphically thinner in the centre of the graben but has a thickening trend directed equally towards both grabenbounding faults. Conversely, the upper Miocene stratigraphic package defined by the Base Pliocene event and 10.5 Ma SB shows a gross thickness trend directed towards the master basinward dipping fault. Examination of the internal stratigraphic units that collectively define the upper Miocene package indicate that the 10.5Ma SB to IUM 1 section shows little thickness variation within the graben and only minor stratigraphic growth across the graben bounding faults. The IUM1 to Base Pliocene unit shows the greatest amount of landward expanding growth. The Pliocene to Intra-Pliocene strata show a switch in the direction of thickening trend with development of a basinward expanding growth sequence. The polarity switches observed within the graben are not as abrupt as other examples shown from similar grabens from the Angolan Margin such as the Gaivota graben (Lundin 1992). Our interpretation of graben stratigraphic growth patterns suggests a more gradual shift from a lower to middle Miocene landward and basinward expanding growth sequence to a late Miocene landward expanding sequence to a final Pliocene basinward expanding sequence.
Timing and evolution of rafting Structurally, the time of graben core development marks the onset of significant raft development (Lundin 1992). As a result, the stratigraphic age of the strata within the core of a raft-related graben indicates the time at which full rafting (as opposed to pre-rafting) initiated. Well data from the study area indicate that the core of the graben comprises lower Miocene strata suggesting development of full-scale rafting and graben generation occurred during this period ( - 20 Ma).
Internal stratigraphic/structural geometries of rafted strata. The rafted strata comprise Albian to Oligocene sediments characterized by the presence of unconformities and variations in stratigraphic thicknesses. Figure 7 shows two rafts of Albian to Oligocene strata separated by approximately 10km. The NE raft rests above a relict salt body and is characterized by strata that thin dramatically towards and above the salt body. In particular, Albian-Eocene strata show a strong degree of stratigraphic thinning with an approximate 75% reduction towards the SW flank of the raft, whereas the Oligocene strata show less extreme thinning ( - 3 0 % ) . The SW raft is characterized by the presence of a welldeveloped angular unconformity. The lowermost AlbianEocene strata are strongly dipping onto the d~collement surface, whereas the overlying raft strata have much shallower dips. Downfaulting of Oligocene strata onto the d~collement surface characterizes the NE flank of this raft.
3D structural framework of the graben fault system A depth-converted 3D structural model of the study area (see Fig. 8 and Movie 1 on supplementary CD) allows a threedimensional analysis of the seismic interpretation and facilitates division of the system into three structural elements; (1) large westward dipping (basinward) listric faults; (2) eastward dipping (landward) antithetic faults; (3) an inner graben array of smaller synthetic faults. A 3D fault model of only the inner graben and antithetic faults from different perspective views and angles allows a more unobstructed visualization of the graben fault system (Figs 9a & 9b). This closer examination of 3D fault geometries indicates that along-strike changes in fault frequencies and fault polarities characterize the inner-graben/antithetic fault array system. With northwards progression along the graben axis, antithetic faults die out and synthetic faults define the dominant fault style (Fig. 9a). Inter-fingering fault geometries, a change in the structural level of antithetic-synthetic splay linkage and the development of a left stepping en echelon array characterize the junction at which there is a change in fault polarity. The stratigraphic level of the antithetic-synthetic fault linkage varies from approximately the 16.5 Ma S.B. in the southeast of
3D GEOMETRY OF A RAFI'-RELATED FAULT ARRAY
139
Fig. 8. 3D depth-converted fault model of the study area depicting the threedimensional geometries of faulting associated with deformation of a cover sequence above a raft system. Fault system comprises a laterally bifurcating basinward-dipping listric fault, a set of antithetic faults and an inner graben array of small synthetic faults. Location of seismic profiles (Figs 5 & 6) are also shown. See Movie 1 on supplementary CD for a movie of this fault model.
the graben to the upper Miocene IUM 1 marker horizon in the northwest of the graben (Fig. 9b). These changes in fault polarities, fault geometries and the stratigraphic level of fault linkages may be reflecting a variation in the deformation style or driving m e c h a n i s m within the graben block. Additionally, these observations may be indicative of changing salt evacuation patterns along the axis of the graben. A s y m m e t r i c salt withdrawal of the salt wall/diapir associated with a salt rise and fall m o d e of rafting would generate c o m p l e x fault relationships and changes in fault polarities.
Kinematic analysis
Fault displacement analysis from seismic interpretation A characteristic feature of raft tectonics is the ability to accrue a large amount of sedimentary growth across graben bounding faults, as seen in Figs 5a & 5b. However, there appears to be little sedimentary growth across inner graben faults (Figs 5a, 5b & 6). Fault F for example has an upper tip point located within the Pliocene stratigraphic package. From this upper tip
Fig. 9. (a) View facing westwards and looking obliquely downwards into a 3D fault model of only the intra-graben faults. Lighting is applied from the bottom right of the model. This view shows more clearly the along-strike changes in fault polarities along the graben axis. The development of a leftstepping en echelon array of small synthetic faults characterizes the location at which significant antithetic faulting dies out. Location of seismic profiles (Figs 5 & 6) are also shown. (b) 3D fault model of the intra-graben fault system from a view facing southwards. This visualization shows how the stratigraphic level of antithetic-synthetic fault linkage varies along the axis of the graben. Location of seismic profiles (Figs 5 & 6) are shown.
140
D.M. DUTTON ETAL.
Fig. 10. (a) 3D displacement model of the laterally bifurcating and basinward-dipping listric fault. 3D displacement mapping shows that two styles of displacement patterns are present on a single fault surface. The lower portion of the fault plane is characterized by displacement values that increase with increasing depth whereas the upper potion of the fault displays discrete and laterally separated displacement maxima. Location of Figure 10b is shown. (b) Zoom-in of a lower portion of the fault plane showing displacement broadly increasing with increasing depth. Note that minor splays link with the master fault and have been made transparent in order to allow clearer visualization.
point there is an increase in displacement with depth, shown by horizon juxtapositions at the Base Pliocene and IUM1 stratigraphic levels. However, below the IUM2 horizon there is a decrease in displacement towards a lower tip point within the graben core. Only the Intra-Pliocene stratigraphic unit (Intra-Pliocene event to Base Pliocene event) displays synsedimentary growth across Fault F. This observation suggests initiation of Fault F during deposition of the Intra-Pliocene stratigraphic unit with subsequent downward propagation into the graben core.
Displacement patterns of a basinward-dipping master
fault Utilizing 3D displacement mapping techniques (Lister 2001 ) the displacement distribution across fault surfaces can be analysed. The master basinward-dipping listric fault shows a pattern of displacement that broadly increases (in excess of 900 m) with depth (Figs 10a & 10b), indicative of a growth fault history (Walsh et al. 2003). With further examination, there is a marked change in the vertical displacement gradient approximately halfway down the fault plane at the IUM1 stratigraphic level. The upper portion of the fault is characterized by widely spaced displacement contours, with discrete elliptical areas of increased displacement. In contrast, a downward increase in displacement, with closely spaced horizontal displacement contours defines the lower section of the fault. We interpret these faults as composite faults that comprise two contrasting patterns of displacement on a single fault surface possibly as a result of differing salt evacuation rates through time.
Discussion Salt diapirism versus extensional rafting Anderson et al. (2000) propose that the present day post-salt structural geometries of the study area result from an extensional mode of rafting because of the lack of unconformities and other
salt-related features within the rafted strata. Our examination of fault growth and linkage in the cover sequence above a raft system, together with analysis of the internal structural/stratigraphic geometries of the rafts suggests that mobile salt may play a more significant role in the post-rift structural evolution of the study area than previously envisaged. In addition, application of a purely extensional mode of rafting to this study area is unable to account for the generation of large antithetic faults that presently define the western margin of the graben and detach at depth onto a dEcollement surface. These antithetic faults were active during the early stages of rafting (indicated by stratigraphic growth trends) and developed from the collapse of a salt diapir (see Fig. lb) as opposed to antithetic fault deformation associated with the rollover of a hangingwall above a listric fault.
Evidence for salt growth. Internal stratigraphic geometries within the raft blocks are consistent with the rise or growth of a salt diapir/pillow during the early stages of thin-skinned extension. A number of angular unconformities together with extreme thinning and onlap of intra-raft units onto salt bodies, collectively suggest that thick salt was mobile during the thinskinned extension phase (Fig. 7). This may conform to the evolutionary model proposed by Rouby et al. (2002), in which salt diapir growth is initially reactive (Vendeville & Jackson 1992a,b) following thin-skinned extension of the overlying strata, before quickly becoming more passive (Vendeville & Jackson 1992a,b) in response to continued sediment deposition. Evidence for salt fall~evacuation from stratigraphic thickness patterns. Variation in the polarity of stratigraphic thickening trends characterize the syn-deformational strata of the raftrelated graben. Seismic profiles and isopach maps show switches in the polarity sense of thickening trends within the upper Miocene to middle Pliocene stratigraphic units (Figs 5 & 6). The largest amount of stratigraphic growth across grabenbounding faults occurs within the lower and middle Miocene units, with both landward and basinward expanding growth
3D GEOMETRY OF A RAFT-RELATED FAULT ARRAY sequences (Fig. 5). This suggests that the major faults that define the graben were both active during the lower/middle Miocene and competed with one another for control of the graben subsidence. Evacuation of a large salt diapir can create this competition, as a graben filled with syn-salt evacuation strata develops above the subsiding salt body. Asymmetric reduction/evacuation through time of such a salt diapir will control the polarity changes of thickening trends within the overlying graben. Additionally, progressive rotation of syndeformational strata associated with both pure extensional rafting and salt diapir rise and fall modes can result in the development of welded contacts between post and pre-salt sediments that can alter the polarity of graben stratigraphic thickening trends (Lundin 1992). The observation that the 10.5 Ma SB to IUMI stratigraphic package is characterized by uniform thickness within the graben may be related to the development of welding at the presalt/post-salt interface, impeding the ability of the graben bounding faults to control subsidence (Fig. 5). Thickness variations across the basinward-dipping listric fault observed in the IUM1 to Intra-Pliocene unit may represent renewed activity of the fault. A mechanism for renewal of fault activity following a previous welding event is possible because well data indicate that strata at the pre-salt/post-salt weld are currently overpressured (Anderson et al. 2000). Overpressure may contribute to the continued ability of the graben bounding faults to accommodate further subsidence by maintaining the required drcollement conditions for thin-skinned extension.
141
subsequently resulted in the sporadic welding of faults with direct juxtaposition of pre-salt/post-salt strata. 3D structural restorations of the fault system would be required to investigate this further.
Implications for Neogene turbidite play system. Uncertainty with respect to the amount and behaviour of mobile salt during thin-skinned extension will impart a higher degree of risk to certain elements of the Neogene turbidite sand play outlined earlier (Fig. 2). The key play risks that result from a salt rise and fall mode of rafting instead of a purely extensional mode of rafting are associated with source rock maturation history and hydrocarbon migration issues. The presence of thick salt will inhibit maturation of a pre-salt source rock during the period that salt is present within the system. This is due to the high thermal conductivity properties of salt (6.0W/m K at surface, 4.0 W/m K at 130~ In addition, charging a post-salt reservoir with a pre-salt source rock requires a welded pre-salt/post-salt contact (i.e. very thin salt), due to the impermeable properties of salt. Hydrocarbons may pool beneath salt early in the history and then re-migrate into post-salt strata once welds form (Rowan et al. 1999). Understanding the structural style of rafting (i.e. purely extensional versus salt diapirism) has significant impact on source rock maturation/migration modelling requirements associated with the Neogene hydrocarbon play.
Conclusions Evidence for salt evacuation from analysis of 3D fault geometries and displacement patterns. 3D geometric relationships of the inner graben fault array and the antithetic faults record the deformation history of the graben. Sequential undercutting and relaying of antithetic-synthetic linkage points with depth, together with fault polarity changes along the graben axis, inter-fingering fault geometries and left-stepping en echelon arrays graben reflect the history of salt evacuation and migration (Figs 6 & 9). In particular, these observations are indicative of asymmetric salt reduction/evacuation of a salt diapir with fault polarity changes representing spatial variation in salt evacuation rates. The 'younging' direction of the antithetic-synthetic splay linkage towards the northwest and along the graben axis indicates that possible salt evacuation and diapir collapse started in the southern part of the study area and migrated northwards. 3D displacement analysis suggests that the basinwarddipping graben bounding faults are composite faults with two styles of fault displacement patterns on a single fault surface (Figs 10a & 10b). Examples of similar composite throw patterns are observed on faults from a salt-related array, offshore Louisiana in the northern Gulf of Mexico (Rowan et al. 1998). In that study, faults that generally display a closed loop style of throw were termed compensation faults and interpreted to represent fault growth in response to deformation on nearby larger faults. Composite faults within that array comprise compensation and growth displacement patterns that separate laterally. The composite faults identified in this study from the Lower Congo Basin differ from the Gulf of Mexico composite faults in that they comprise a vertical separation of different displacement patterns as opposed to a lateral separation. The junction between the two types of displacement patterns within the Lower Congo Basin study occurs at the IUM1 structural level. We suggest that the vertical separation of displacement patterns on the graben bounding faults reflects a change in the ability of the faults to accrue slip after the IUM1 structural level. This change is the consequence of significant withdrawal and migration of salt away from the fault system that
Detailed three-dimensional structural analysis of a raft-related fault array that developed in response to Miocene thin-skinned extension reveals the following: 9 A raft-related graben, composed of three structural elements: (a) landward dipping antithetic faults; (b) inner graben set of small synthetic splays: (c) a basinward dipping master synthetic fault that detaches at depth along the flank of a relict salt pillow. 9 A complex relationship exists between the antithetic faults and the inner graben synthetic splays, characterized by a varying level of structural linkage along the graben axis, inter-fingering fault geometries, undercutting and relaying of linkage points with depth and along-strike changes in fault polarity. These relationships record the internal deformation history of the graben block during MioceneRecent evacuation of a salt diapir. 9 3D displacement analysis suggests that the basinwarddipping graben bounding faults are composite faults with two styles of fault displacement patterns on a single fault surface. A growth fault style in which displacement increases with depth (lower portions of fault plane) and an elliptical style in which displacement decreases outwardly from a localized maximum (upper portion of fault plane). 9 Analysis of fault linkages and displacement configurations, syn-deformational stratigraphic thickening patterns, together with the internal stratigraphic geometries of raft blocks suggest that mobile salt may play a more significant role than previously envisaged within the study area during thin-skinned extension. Asymmetric reduction/evacuation of an initially reactive/passively growing salt pillow is consistent with the observed structural/stratigraphic geometries. 9 The mode of thin-skinned extension can have important consequences for the Neogene turbidite hydrocarbon play associated with the rafted province of the Lower Congo Basin. The presence of thick mobile salt will influence
D.M. DUTFON ETAL.
142
the extent o f pre-salt source rock maturation and development of pre-salt/post-salt hydrocarbon migration windows. The authors would like to express their gratitude to Sonangol for supporting the publication of this work. We thank CNR UK International Ltd for access to this dataset and for contributing to the costs of the colour pages. In particular, we would like to thank J. Anderson (CNR) for his thoughtful comments and discussions about raft tectonics in the Lower Congo Basin and J. Chessell (CNR) who has been instrumental in facilitating the use of this dataset. This work has benefited from thorough and thought-provoking reviews from D. Helgeson and an anonymous reviewer. This project is part of ongoing work for a PhD (NERC award reference NER/S/A/2000103770).
References ANDERSON, J. E., CARTWRIGHT,J., DRYSDALL,S. J. & VIVIAN, N. 2000. Controls on turbidite sand deposition during gravity-driven extension of a passive margin: examples from Miocene sediments in Block 4, Angola. Marine and Petroleum Geology, 17, I 165-1203. BRICE, S. E., COCHRAN, M. D., PARDO, G. 8z. EDWARDS, A. D. 1982. Tectonics and sedimentation of the South Atlantic Rift Sequence: Cabinda, Angola. American Association of Petroleum Geologists Memoir, 34, 5-18. BROGNON, G. P. & VERRIER, G. R, 1966. Oil and geology in Cuanza Basin of Angola. AAPG Bulletin, 50, 108-158. BURKE, K. 1975. Atlantic evaporates formed by evaporation of water spilled from Pacific, Tethyan and southern oceans. Geology, 3, 613-616. BURKE, K. 1996. The African Plate. South African Journal of Geology, 99, 341-409. CARTWRIGHT, J., MANSFIELD, C. • TRUDGILL, B. 1996. The growth of normal faults by segment linkage. In: BUCHANAN, B. G. 8,: NIEUWLAND, D. A. (eds) Modern Developments in Structural Interpretation, Validation and Modelling. Geological Society, London, Special Publications, 99, i 63-177. CRAMEZ, C. ~ JACKSON, M. P. A. 2000. Superposed deformation straddling the continental-oceanic transition in deep-water Angola. Marine and Petroleum Geology, 17, 1095-1109. DE MATOS, R. M. 1999. History of northeast Brazilian rift system: kinematic implications for the break-up between Brazil and West Africa. 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, 55-73. DUVAL, B., CRAMEZ,C. & JACKSON, M. P. A. 1992. Raft tectonics in the Kwanza Basin, Angola. Marine and Petroleum Geology, 9, 389-404. EICHENSEER, H. T., WALGENWITZ, F. R. & BIONDI, P. J. 1999. Stratigraphic control on facies and diagenesis of dolomitised oolitic siliciclastic ramp sequences (Pinda Group, Albian, offshore Angola). AAPG Bulletin, 83, 1729-1758. JACKSON, M. P. & CRAMEZ, C. 1989. Seismic recognition of salt welds in salt tectonic regimes. Gulf Coast section of the Society of Economic Palaeontologists and Mineralogists. Tenth Annual Research Council, Houston, Texas, 66-71. KARNER, G. D. & DRISCOLL, N. W. 1999. Tectonic and stratigraphic development of the west African and eastern Brazilian margins: insights from quantitative basin modelling. 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, 11-40. LISTER, L. D. 2001. Computer Modelling and Characterisation of Intersecting 3D Fault Networks. PhD Thesis, Imperial College London, UK.
LISTER, L. D. 2004. Modelling fault geometry and displacement for very large networks. 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, 339-348. LUNDE, G., AUBERT, K., LAURITZEN.O. & LORANGE, E. 1992. Tertiary uplift of the Kwanza Basin in Angola. In: CURNELLE, R. (ed.) Geologic Africaine Compte Rendu des colloques de Geologic de Libreville. 13, 6-8. LUNDIN, E. R. 1992. Thin-skinned extensional tectonics on a salt detachment, northern Kwanza Basin, Angola. Marine and Petroleum Geology, 9, 405-411. MOORE, T. 1988. Atlantic margin basins of Africa. AAPG Bulletin, 72, 224-225. PEACOCK, D. C. P. & SANDERSON, D. J. 1991. Displacements, segment linkage and relay ramps in normal fault zones. Journal of Structural Geology, 13, 721-733. PEACOCK, D. C. P. & SANDERSON, D. J. 1994. Geometry and development of relay ramps in normal fault systems. AAPG Bulletin, 78. 147-165. ROBERTS, D. G. 1975. Evaporite deposition in the Aptian South Atlantic Ocean. Marine Geology, 18, 65-72. ROUBY, D., RAILLARD, S., GUILLOCHEAU, F,, BOUROULLEC, R, & NALPAS, T. 2002. Kinematics of a growth fault/raft system on the West African margin using 3D restoration. Journal of Structural Geology, 24. 783-796. ROWAN, M. G., HART, B. S., NELSON, S., FLEMINGS, P. B. & TRUDGILL, B. D. 1998. Three-dimensional geometry and evolution of a saltrelated growth-fault array: Eugene Island 330 field, offshore Louisiana, Gulf of Mexico. Marine and Petroleum Geology, 17, 309-328. ROWAN, M. G., JACKSON, M. P. & TRUDGILL, B. D. 1999. Salt-related fault families and fault welds in the northern Gulf of Mexico. AAPG Bulletin, 83, 1454-1484. SPATHOPOULOS, F. 1996. An insight into salt tectonics in the Angola Basin, South Atlantic. In: ALSOP, G. I., BLUNDELL, O. J. 8r DAVISON, I. (eds) Salt Tectonics. Geological Society, London, Special Publications, 100, 153-174. STANDLEE, L. A., BRUMBAUGH, W. D. & CAMERON, N. R. 1991. Controlling factors in the initiation of the south Atlantic rift system. Geologic Africaine, Elf Acquitaine Memoir, 13, 141 - 152. TRUDGILL, B. & CARTWR1GHT,J. 1994. Relay-ramp forms and normalfault linkages, Canyonlands National Park, Utah. Geological Society of America Bulletin, 106, 1143-1157. VALLE, P. J,, GJELBERG, J. G. & HELLAND-HANSEN, W. 2001. Tectonostratigraphic development in the eastern Lower Congo Basin, offshore Angola, West Africa. Marine and Petroleum Geology, 18, 909-927. VENDEVILLE, B. C. & JACKSON, M. P. A. 1992a. The rise of diapirs during thin-skinned extension. Marine and Petroleum Geology, 9, 331-353. VENDEVILLE, B. C. & JACKSON, M. P. A. 1992b. The fall of diapirs during thin-skinned extension. Marine and Petroleum Geology, 9, 354-371. WALSH, J. J., CHILDS, C.. IMBER, J., MANZOCCHI,T., WATTERSON,J. & NELL. P. A. R. 2003. Strain localisation and population changes during fault system growth within the Inner Moray Firth, Northern North Sea. Journal of Structural Geology, 25, 307-315. WILLEMSE, E. J. M., POLLARD, D. D. & AYDIN, A. 1996. Threedimensional analyses of slip distributions on normal fault arrays with consequences for fault scaling. Journal of Structural Geology, 18, 265-279.
Initial deformation in a subduction thrust system: polygonal normal faulting in the incoming sedimentary sequence of the Nankai subduction zone, southwestern Japan A. S. HEFFERNAN
I , J. C . M O O R E
I , N. L. BANGS
2, G . F . M O O R E
3 & T. H. SHIPLEY
2
tDepartment of Earth Sciences, University of California Santa Cruz, 1156 High St., Santa Cruz, CA 95064, USA (e-mail: [email protected]) 2University of Texas Institute for Geophysics, 4412 Spicewood Springs Rd., Austin, TX 78759, USA 3Department of Geology and Geophysics, 1680 East West Rd., UniversiO" of Hawaii, Honolulu, HI 96822, USA
Abstract: 3D seismic data from the Nankai margin provide detailed imagery documenting the onset of deformation at an active sediment-dominated accretionary prism, including a previously unmapped network of normal faults. The Nankai margin off southwest Japan is characterized by active subduction, seismogenesis, and a large accretionary prism with foldand-thrust belt structure. Imbricate thrusting is the dominant structural style of the outer 20 km of the prism. This structural domain develops at the prism toe, where an incipient imbricate thrust displays significant along-strike variability in dip, offset, and development of hangingwall anticlines. Compressional deformation is preceded by normal faulting that initiates seaward of the trench axis. Seismic data in this area reveal a complex, intersecting pattern of normal faults within the incoming hemipelagic sediments. Underlying the faulted section is a high-amplitude reflector interpreted as representing oceanic basement. This reflector contains elongate horsts and grabens oriented perpendicular to the margin interpreted as relict spreading centre fabric. Analysis of the orientation of normal faults within the Shikoku basin sequence shows a correlation between fault geometry and basement structure. This faulting is notably similar to layer-bound compaction faults, documented in the North Sea and elsewhere, attributed to both hydrofracturing and volumetric contraction of fine-grained sediments. Mapped normal faults may thus be the result of a combination of differential compaction of sediments above irregular, dipping oceanic basement and compactional dewatering seaward of the toe of the accretionary prism.
The Nankai Trough is the bathymetric expression of the convergent margin formed where the Philippine Sea Plate subducts beneath the southwest Japan arc. Convergence rates here are between 2 and 4 c m / y r (Karig 1986; Seno et al. 1993) (Fig. 1). As subduction occurs a thick sequence of hemipelagic sediments and turbidites is accreted to the overriding plate. The resulting accretionary prism provides a record of subduction zone processes, including the development of classic foldand-thrust belt structure, and mechanical and diagenetic processes associated with the accretion of young marine sediments. The Nankai margin has been a focus of extensive, multinational scientific investigation in recent years. These investigations have included Ocean Drilling Program Legs 131, 190, and 196 (Shipboard Scientific Party 1991, 2001, 2002: Mikada et al. 2002), multiple 2D and 3D seismic reflection surveys (Aoki et aI. 1982; Moore et al. 1990), OBS surveys (Kodaira et al. 2000), and various seafloor mapping surveys. Historical records describe frequent, massive earthquakes in the Nankai area from 684 AD to modern times. The 1946 Nankaido (Ms = 8.2) and 1944 Tonankai earthquakes (Ms = 8.0) are the most recent large earthquakes to occur in this area (Ando 1975, 1991) and underscore the ongoing nature of seismic and tsunamigenic hazards here. This study utilizes a 3D multichannel seismic (MCS) reflection survey collected off Muroto peninsula comprising an 8 by 80 km transect oriented perpendicular to the margin. This survey images the stratigraphy of the Shikoku Basin, the onset of compressional deformation, and landward approximately 60 km into the accretionary prism (Moore et al. 2001a). Structural characteristics of this accretionary system have been described in detail by previous work (Moore et al. 1990, 2001c). This study utilizes 3D seismic data to document fine-scale structural variability in the earliest stages of deformation, a previously unmapped pattern of normal faults in the incoming sedimentary section, and underlying basement morphology. Normal fault geometry could not be determined from previously existing 2D
seismic data, and may be characteristic of deformation found in other sediment-dominated subduction zones worldwide.
Data acquisition and processing The desire to understand the upper aseismic to seismic transition in subduction zones and associated earthquake hazard motivated the collection of the data used in this study. The effort was funded by the US National Science Foundation and equivalent agencies in Japan. Data were acquired aboard the LamontDoherty Earth Observatory's R/V Ewing during cruise EW990708, from 18 June 1999 to 18 August 1999. During this period 81 individual lines were shot, covering an 8 x 80km transect oriented N W - S E across the margin with a line spacing of 100 m. A 6 km streamer was used, equipped with 240 channels with a group spacing of 25 m. The seismic source used was an array of 14 airguns with a total volume of 4276 in 3. Shot spacing was 50m, with data being recorded for 12s with a sampling interval of 2 ms (Moore et al. 2001a). Initial data processing was conducted on the Ewing and included decimation to 4 m s samples, trace editing, and initial 2D stack and migration. Subsequent data processing was conducted at both the University of Texas and the University of Hawaii. 3D processing consisted of velocity analysis, normal moveout correction, inside and top mute, 3D stack and 3D post-stack migration. Bin size for the 3D volume is 25 m (inline) by 50 m (crossline). All interpretations used in this study are from 3D time migrated data.
Structural overview Tectonic and stratigraphic setting The Nankai accretionary system has been described as an 'endmember' convergent margin, due in part to the relative simplicity of the margin's structure and stratigraphy, and for
DAVIES,R. J., CARTWRIGHT,J. A., STEWART,S. A., LAPPIN,M. & UNDERH1LL.J. R. (eds) 2004.3D Seismic Technology:Applicationto the Exploration of Sedimentar3, Basins. Geological Society, London, Memoirs, 29. 143-148.0435-4052/04/$15 9 The Geological Society of London 2004.
144
A.S. HEFFERNAN ET AL.
Fig. 1. Regional setting of the Nankai Trough area, including Shikoku island, southwest Japan. Location of study area is outlined in grey (from Shipboard Scientific Party 2001). the almost complete accretion of the incoming sedimentary section (Shipboard Scientific Party 2001). Subducting oceanic crust in the Moroto area is about 15 Ma in age, and has been interpreted to have formed by back-arc spreading resulting in the creation of the Shikoku Basin (Hall et al. 1995; Okino et al. 1994). The fossil spreading ridge associated with this process is now oriented roughly perpendicular to the Nankai margin (Fig. 2). Relatively young oceanic crust is overlain by approximately 700 m of hemipelagic sediments (Fig. 3). A well at ODP site 1173 (Fig. 2) penetrates the entire sedimentary section within the outer trench margin, providing abundant geological data and allowing correlation between seismic data and well logs and cores (Moore et al. 2001b; Mikada et al. 2002). The middle Miocene-Quaternary Shikoku Basin Sequence consists of 688 m of hemipelagic mud and claystone, and has been divided into Upper and Lower Shikoku Basin facies (Moore et al. 2001b), Upper Shikoku Basin facies are represented by moderate~amplitude, parallel reflectors correlating to interbedded volcanic ash and tuff layers. Diagenesis has altered
these ash and tuff layers to acoustically transparent siliceous claystones in the Lower Shikoku Basin facies (Moore et al. 2001b). Overlying this basin fill sequence is a - 5 0 0 m thick trench fill sequence imaged as an asymmetrical wedge of highamplitude reflectors onlapping onto the underlying unconformity (Fig. 3). This trench-fill sequence was deposited primarily by turbidites transporting terrigenous material derived primarily from uplift and erosion associated with the collision of the IzuBonin arc with Honshu to the north (Fig, 1), and transported southwest along the margin by trench-axial turbidity flows (Taira & Niitsuma 1985; Underwood et al. 1993). A basal d~collement exists within the lower Shikoku Basin sequence. Sediments above this fault are accreted, while lower sediments are initially underthrust. The resulting accretionary prism displays fold-and-thrust style deformation analogous to structures observed in accretionary prisms and mountain belts worldwide (Davis et al. 1983). The Cascadia accretionary wedge is another sediment-dominated subduction thrust system with highly analogous structures (MacKay 1995).
Onset o f imbricate thrusting. Detailed interpretation of reflections at the deformation front document the initiation of fold-and-thrust belt style deformation in a relatively simple tectonic and stratigraphic setting (Shipboard Scientific Party 2001). Regularly spaced, seaward-verging imbricate thrusts that sole into a basal d~collement are the dominant structural features of the 'imbricate thrust zone' that comprises the outermost 20kin of the prism (Fig. 2) (Shipboard Scientific Party 2001). Smaller scale, conjugate back-thrusts are common in this zone, and fault-bend fold hangingwall anticlines develop above imbricate thrusts (Taira et al. 1991). The onset of this style of deformation occurs in a 'protothrust zone' that exists seaward of the frontal thrust, and is characterized by a prominent protothrust, as well as horizontal shortening and structural thickening (Moore et ai. 1990) (Figs 2 & 3). In the protothrust zone several incipient thrusts can be identified in portions of the Shikoku Basin sequence and overlying trench fill (Figs 3 & 4). The most prominent of these thrusts (Fault 1 on Figs 3 & 4) is similar in gross geometry to the older thrusts characteristic of the imbricate thrust zone, but exhibits less throw and is discontinuous across the survey area. A low-amplitude hangingwall anticline develops in association with this protothrust, and is expressed on the sea floor as a subtle bathymetric high (Fig. 2). This protothrust, and the deformation associated with it, display significant along-strike structural variability across the survey area. The fault is well defined by truncated reflectors in the southwest portion of the survey area. Moving to the northeast the amount of offset varies, and the fault splays into three distinct slip surfaces (Fig. 4). Further northeast, offset on all three splays decreases to undetectable amounts. This fault system steps landward and forms a seismically resolvable fault surface at the northeastern-most portion of the survey area. Hangingwall anticline development also varies spatially, with areas of more pronounced folding separated by a zone of no detectable folding (Figs 2 & 4). Spatial distribution of folding correlates with variations in fault displacement.
Normal faulting F a u l t geomet~.
Fig. 2. Shaded-relief seafloor depth map from 1999 3D seismic data illustrating the bathymetric expression of various structural domains at the toe of the Nankai accretionary prism. Locations of ODP wells 808i, 1174b and 1173b are shown.
Reflectors of portion of the small-offset by truncated
the Upper Shikoku Basin facies and the upper Lower Shikoku Basin facies are cut by numerous ( 2 0 - 5 0 m throw) normal faults, represented and offset reflectors in seismic cross sections
POLYGONAL FAULTING. NANKAI SUBDUCTION ZONE
145
Fig. 3. Seismic and interpretive sections illustrating stratigraphy and structure of the prism toe area. Sections are perpendicular to the trench axis and parallel to the direction of convergence (location shown on Fig. 2). Fault 1 is the prominent protothrust. Fault 2 is the frontal thrust. Faults 3 and 4 are older insequence imbricate thrusts. Numerous smalloffset normal faults are visible in the upper and middle portion of the incoming Shikoku Basin sequence.
(Figs 3 - 5 ) . As the mean frequency in this interval is about 35 Hz and the velocity is about 1800 m/s the observed offsets are well above the quarter wavelength of seismic resolution. A horizon representing these faulted sediments was mapped across the survey area. A dip map of this horizon shows numerous discrete, curvilinear zones of high dip (shown in black) separated by nearly flat-lying areas (Fig. 6). Zones of high dip correlate to normal faults consisting of 1 0 0 - 1 0 0 0 m segments with complex, intersecting geometries. Other seismic attributes, including
Fig. 4. Perspective view looking NNE of seismic data and interpreted faults and horizons at the prism toe. The vertical dimension is two-way travel time, vertical exaggeration at the seafloor is approximately 5 x . The upper surface represents the seafloor and is shaded by reflection amplitude (yellow/orange, high amplitude; blue/violet, low amplitude). Lower surface represents subducting oceanic basement, shaded by two-way travel time. Fault 1 is the prominent protothrust that bifurcates to form the green and light blue fault splays. Fault 2 is the frontal thrust. Faults 3 and 4 are older in-sequences imbricate thrusts. Normal faults are seen crosscutting the yellow horizon in the incoming sedimentary sequence.
amplitude and similarity, were utilized in fault mapping but did not resolve fault geometry better than a simple dip map. Fault density varies across the survey, with zones on higher fault density. Numerous short fault segments with high-angle ( - 9 0 ~ intersections characterize these high fault-density zones. The contact between the Shikoku basin sequence and overlying trench contact between the Shikoku basin sequence and overlying trench fill turbidites is represented by a continuous, high-amplitude, wavy reflector (Fig. 3).
146
A.S. HEFFERNAN ETAL.
Fig. 5. Detailedseismic section of upper and middle Shikoku basin sequence, seaward of the onset of compressional deformation, showing numerous smalloffset normal faults. The dashed line is the horizon shown in map view in Figure 6.
Overlying reflectors representing trench-fill turbidites onlap onto this reflector and do not appear to be cut by faults (Fig. 4). Underlying sediments at the base of the Lower Shikoku Basin facies are imaged as low-amplitude, continuous reflectors that do not appear to be cut by faults (Moore et al. 2001a). Thus, normal faults are stratigraphically confined to the upper and middle portions of the Shikoku basin sequence.
Oceanic crust morphology and overlying fault orientations Oceanic basement underlying the faulted sections is imaged as a high-amplitude, semicontinuous reflector in the seawardmost - 4 0 k i n of the survey area. This reflector displays elongate horst and graben morphology interpreted as a primary sea floor spreading center fabric (Fig. 7). These northwest trending, linear structural highs and lows are on the order of l km wide and display up to 200ms (approximately 175 m) of vertical relief. These features share the same trend as survey shiptracks. Their scale, however, makes it unlikely (but possible) that they are an artifact of data collection or processing. Furthermore, the orientation of this crustal fabric is consistent with the expected orientation of the palaeo-spreading centre active during the formation of the Shikoku basin (Le Pichon et al. 1987). This oceanic basement morphology influences normal fault geometry seen in overlying sediments. Line segments representing 226 km of normal fault traces were digitized within an area of 55 km 2, based on the faulted horizon dip map (Fig. 7). An orientation was recorded for every 50 m of these mapped fault traces, so that longer segments could be proportionally weighted in subsequent orientation analysis. Analysis of fault trends reveals a preferential orientation parallel to the margin, or approximately perpendicular to the oceanic basement fabric (Fig. 7). Subdivision of digitized fault segments into areas overlying basement highs and lows highlights basement control on fault orientation. Faults overlying basement lows are oriented strongly parallel to the margin trend. Areas overlying basement structural highs, in contrast, have a bimodal distribution of fault trace orientations. Mapped fault traces in these areas show preferential orientations both parallel to, and perpendicular to, the margin (Fig. 7). Areas overlying basement structural highs also have a greater density of faulting, with an average of 6.44km of fault traces per km 2 mapped, versus 2.70km/km 2 in areas underlain by basement lows. The correlation between fault orientation and fault density with basement morphology strongly suggests basement control on overlying fault geometry.
Layer-bound compaction faults The observed pattern of normal faulting is notably similar to polygonal fault systems in the North Sea Tertiary interval and elsewhere (Cartwright & Lonergan 1996). This newly described style of deformation has been attributed to compactive dewatering processes and has been previously documented in 28 localities worldwide, all of which are characterized by very fine-grained sediments (Cartwright 8,: Dewhurst 1998). The faults observed at Nankai meet several of the criteria Cartwright & Dewhurst use to identify these layer-bound compaction faults: they are polygonal in map view, occur within specific stratigraphic boundaries, are all normal faults with throws of 10-100 m. and are closely spaced. The Nankai faults differ from other layer-bound compaction fault systems in that they are not organized into 'tiers' occupying multiple stratigraphic levels. North Sea polygonal faults are characterized by typical fault trace lengths (100-1000 m) and average throws (30-50m) (Lonergan et al. 1998) that are nearly identical to faults mapped at Nankai. Both sets of faults have complex, intersecting geometries that are polygonal in plan view. The development of layer-bound compaction faults has been linked to volumetric contraction of fine-grained sediments during early compaction (Cartwright & Lonergan 1996). Recent work suggests that the colloidal nature of these sediments allows them to behave as gels which contract without
Fig. 6. Dip map of faulted horizon shown in Figure 4. Dark zones are areas of high dip and correspond to normal fault traces. Heavy black line denotes the frontal thrust trace, with teeth on the overriding accretionary wedge.
POLYGONAL FAULTING. NANKAI SUBDUCTION ZONE
147
Fig. 7. Shaded-relieftime-structure map of subducting oceanic basement underlying the normal faulted sequence. Heavy red line denotes the frontal thrust trace, with teeth on the overriding accretionary wedge. Thin red lines denote fault segments mapped from dip map (Fig. 6). Black lines outline prominent structural trends interpreted to be relict seafloor spreading fabric. Fault orientations are shown for all faults, faults overlying basement structural highs, and faults overlying basement structural lows.
evaporation of pore fluid through a process known as syneresis (Dewhurst et al. 1999). High pore pressures may also play a critical role in the development of layer-bound faulting. Hydrofracturing due to episodic expulsion of fluids from overpressured shales was an early explanation used to explain the lower Tertiary North Sea polygonal faults (Cartwright 1994). The presence of polygonal faults has been further interpreted as a potential indicator of overpressure in geological hazard studies (Haskel et al. 1999). At Nankai the incoming sedimentary section is loaded by deposition of overlying trench turbidites. This process may generate overpressure within the incoming section and facilitate faulting. P o s s i b l e c o n t r o l s on f a u l t o r i e n t a t i o n
Faults described in the North Sea generally lack the obvious preferred orientations seen at Nankai. Orientations in at least one locality, however, are biased towards paleoslope strike (Cartwright 1994). The strong margin-parallel preferred orientation observed at Nankai may similarly be related to the landward dip of the Shikoku Basin sequence as it approaches the trench. The process of differential compaction of fine-grained sediments over irregular topography is known to produce both folding and faulting (Brown 1969; Labute & Gretener 1969). At Nankai normal faults overlying basement highs display basement
fabric-parallel (margin-perpendicular) orientations in addition to margin-parallel preferred orientations. Differential compaction of sediments above basement horsts likely drives the development of faults with the same orientation as basement fabric. The combination of trenchward basement dip and trench-normal basement fabric may thus result in the bimodal orientation distribution of the overlying layer-bound compaction faults.
Conclusions This paper describes the seaward limit of deformation at the toe of the Nankai accretionary prism based on interpretation of a 3D MCS survey collected off Shikoku island, southwest Japan. Data document the initiation of fold-and-thrust belt style deformation, and detailed mapping of the protothrust reveals significant along-strike structural variability. The 3D geometry of a pattern of normal faults existing seaward of the onset of compressional deformation is described for the first time. Normal faults cut hemipelagic sediments of the incoming Shikoku Basin sequence, and are characterized by a complex, intersecting geometry, which is stratigraphically restricted to a - 5 0 0 m thick layer. Faults are preferentially oriented parallel to the trench axis. Sediments overlying basement ridges, however, also contain trench-perpendicular faults that correlate spatially to underlying basement structure. Fault network geometry and basement morphology is resolvable only with quality three-dimensional subsurface imagery.
148
A.S. HEFFERNAN ETAL.
Observed normal faults are notably similar to layer-bound compaction faults described in lower Tertiary o f the North Sea basin, and at least 27 other basins worldwide. These faults have been linked to both volumetric contraction o f fine-grained sediments and natural hydrofracturing. Fault development at Nankai is likely influenced by early compactional dewatering processes characteristic o f fine-grained sediments, sedimentary loading and burial, b a s e m e n t dip, and b a s e m e n t fabric. The bimodal distribution o f preferred orientations observed in layer-bound normal faults seaward o f the toe of the Nankai accretionary prism m a y result from differential compaction above a trenchward dipping basement surface with pronounced trench-normal fabric. Normal faults similar to those described here could explain the earliest tectonic structures formed in sedimentary sequences o f some ancient accretionary prisms (Fisher & Byrne 1987). Support for this work was provided by the National Science Foundation (NSF grant # OCE9802264) and USSP Grant F001431F001464. Acquisition and processing of 3D MCS data was made possible through the collaborative effort of many, including researchers at the University of Texas, the University of Hawaii, the University of Tokyo, and the Geological Survey of Japan. Seismic interpretation and visualization software was provided by Landmark Graphics Corporation through their university partnership program. J. Cartwright, L. Lonergan and R. Nelson made insightful comments regarding fault mapping and structural geology phenomena that proved very helpful. We thank J.R. Underhill and 1. Cloke for their helpful reviews of the manuscript.
References ANDO, M. 1975, Source mechanisms and tectonic significance of historical earthquakes along the Nankai Trough, Japan. Tectonophysics, 27, 119-140. ANDO, M. 1991. A fault model of the 1946 Nankaido earthquake derived from tsunami data. Physics of Earth and Planetary btteriors. 28, 320-336. AOKI, Y., TAMANO, T. & KATO, S. 1982. Detailed structure of the Nankai Trough from migrated seismic sections, h~: WATKINS. J. S. & DRAKE, C. L. (eds) Studies in Continental Margin Geology. American Association of Petroleum Geologists, Memoir. 34, 309-322. BROWN, P. 1969. Compaction of fine-grained terrigenous and carbonate sediments--a review. Bulletin of Canadian Petroleum Geology, 17, 486-495. CARTWRIGHT, J. A. 1994. Episodic basin-wide fluid expulsion from geopressured shale sequences in the North Sea basin. Geology. 22, 447-450. CARTWRIGHT,J. A. & DEWHURST,D. N. 1998. Layer-bound compaction faults in fine-grained sediments. Geological Society of America Bulletin, 110, 1242-1257. 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. DAVIS, D. J., SUPPE, J. & DAHLEN, F. A. 1983. Mechanics of foldand-thrust belts and accretionary wedges. Journal of Geophysical Research, 88, 1153-1172. DEWHURST, D. N., CARTWRIGHT, J. A. ~,~ LONERGAN, L. 1999. The development of polygonal fault systems by syneresis of colloidal sediments. Marine and Petroleum Geology, 16, 793-810. FISHER, D. & BYRNE, T. 1987. Structural evolution of underthrusted sediments, Kodiak Islands, Alaska. Tectonics. 6, 775-794. HALL, R., FULLER, M., ALl, J. R. & ANDERSON, C. D. 1995. Active margins and marginal basins of the western Pacific. American Geophysical Union Geophysical Monographs, 88, 371-404. HASKEL, N. ETAL. 1999. Delineation of geologic drilling hazards using 3-D seismic attributes. The Leading Edge, 18, 373-382.
KARIG. D. E. 1986. Physical properties and mechanical state of accreted sediments in the Nankai Trough. S.W. Japan. In: MOORE, J. C. (ed.) Strucn,ral Fabrics in Deep Sea Drilling Project Cores from Forearcs, Geological Society of America Memoirs, 166, il7-133. KODAIRA, S., TAKAHASHI, N., PARK, J. O., MOCHIZUKI, K., SHINOHARA, M. & KIMURA, S. 2000. Western Nankai Trough seismogenic zone: Results from a wide-angle ocean bottom seismic survey. Journal of Geophysical Research, 105, 5887-5906. LABUTE, G. J. ,~. GRETENER, P. E. 1969. Differential compactions around a leduc reef--Wizard Lake area, Alberta. Bulletin of Canadian Petroleum Geology, 17, 304-325. LE PICHON, X. ETAL. 1987. Nankai Trough and the fossil Shikoku Ridge: Results of box 6 Kaiko survey. Earth and Planetary Science Letters', 83. 186-198. LONERGAN, L., CARTWRI~HT, J, & JOLLY, R. 1998. The geometry of polygonal fault systems in Tertiary mudrocks of the North Sea. Journal of Structural Geology, 20, 529-548. MACKAY, M. 1995. Structural variation and landward vergence at the toe of the Oregon accretionary prism. Tectonics, 14, 1309-1320. MIKADA, H., BECKER, K., MOORE, J. C., KLAUS, A. Er AL 2002. Proceedings of the Ocean Drilling Program, Initial Reports, 196 (CD ROM). MOORE, G. F., SHIPLEY, T. H.. STOFFA, P. L., KARIG. D. E., TAIRA, A., KL'RAMOTO, S., TOKUYAMA. H. & SUYEHIRO, K. 1990. Structure of the Nankai Trough accretionary zone from multichannel seismic reflection data. Jounral of Geophysical Research, 95, 8753-8765. MOORE, G. F., TAIRA, A., BANGS, N., KURAMOTO,S., SHIPLEY,T. EraL, 2001a. Structural Setting of the Leg 190 Muroto Transect. hr: MOORE, G. F., TAIRA, A., KLAUS, A. ET AL. Proceedings of tire Ocean Drilling Program, hfftial Reports, 190, Chapter 2 (CD ROM). MOORE, G. F., TAIRA, A., KLAUS, A., BECKER, K. ET AL 200lb. Proceedings of the Ocean Drilling Program. Initial Reports. 190 (CD ROM). MOORE, G. F.. TAIRA, A., KLAUS, A., BECKER, L. ET AL, 2001C. New insights into deformation and fluid flow processes in the Nankai Trough accretionary prism: Results of Ocean Drilling Program Leg 190. Geochemistry, Geophysics and Geosvstems 2, 10,129/2001GC000166. OKINO, K., SHIMAKAWA, Y. & NAGAOKA. S. 1994. Evolution of the Shikoku Basin. Journal of Geomagnetism and Geoelectricity, 46, 463-479. SENO, T., STEIN, S. & GRIPP, A. E. 1993. A model for the motion of the Philippine Sea Plate consistent with NUVEL-1 and geological data. Journal of Geophysical Resealz'h, B, Solid Earth and Planets, 98, 17941 - 17948. SmPt~OARD SCIENTIFIC PARTY 1991. Site 808. hr: TAIRA, A., HILL, I., FIRTH. J. V. ET AL., Proceedings of tire Ocean Drilling Program: hritial Reports. (Ocean Drilling Program), College Station, TX, 71 - 269. SHIPBOARD SCIENTIFIC PARTY, Leg 190 2001 Summary. ht: MOORE, G. F., TAIRA, A.. KLAUS. A., Proceedings of the Ocean Drilling Program: Initial Reports. 190: tOcean Drilling Program), College Station. TX (CD ROM), 1-87. SHIPBOARD SCIENTIFIC PARTY, Leg 196 2002 Summary. hr: M1KADA, H., BECKER, K., MOORE, J. C., KLAUS, A., Proceedings of the Ocean Drilling Program: hfftial Reports, 196: (Ocean Drilling Program), College Station, TX (CD ROM), 1-29. TAIRA, A. & NIITSUMA, N. 1985. Turbidite sedimentation in the Nankai Trough as interpreted from magnetic fabric, grain size, and detrital modal analysis. Initial Reports Deep Sea Drilling Project, 87, 611-632. UNDERWOOD, M., ORR, R., PICKERING, K. & TAIRA, A. 1993, Provenance and dispersal patterns of sediments in the turbidite wedge of Nankai Trough, hr: HILL, I. A., TAIRA, A., FIRTH, J. V. ETAL. (eds) Proceedings of the Ocean Drilling Program: Scientific Results. College Station, TX, 15-33.
The evolution and growth of Central Graben salt structures, Salt Dome Province, Danish North Sea MALENE
RANK-FRIEND
1"2
& C H R I S T O P H E R F. E L D E R S
1
lDepartment of Geology, Royal Holloway, University of London, Egham, Surrey TW20 OEX, UK "-'Present address: WesternGeco, Schlumberger, Schlumberger House, Buckingham Gate, Gatwick Airport, West Sussex RH6 0NZ, UK (e-mail: MFriend2 @gatwick, westerngeco.slb, corn)
Abstract: The remarkable spatial resolution of 3D seismic data is particularly important in the study of salt structures, where changes in the distribution of accommodation space resulting from salt withdrawal can be mapped and related to the evolution of the individual salt structures. The Salt Dome Province of the Danish Central Graben provides an interesting example of this approach. Three adjacent salt structures (Skjold, Dan and Kraka) exhibit very different geometries, and their evolution has been the subject of some debate. The present study suggests that the Dan structure is composed almost entirely of Zechstein salt, which, facilitated by de-coupled extension during the Mid-Late Jurassic, has been intruded into and along weak planes of Triassic salt, resulting in an overall domal, circular outline. Skjold is located to the NW of Dan. It consists of Zechstein salt and forms a mature salt stock, which terminates within Upper Cretaceous strata. Skjold evolved from a linear NW-SE trending salt wall and became a point source structure during the Early Cretaceous, in common with the many other Central Graben diapers and coincident with the commencement of thermal subsidence. By contrast, Kraka, a NW-SE trending pillow structure located to the south of the Dan-Skjold alignment never became diapiric. Gravitational downbuilding of the mature Skjold diapir during rapid Cenozoic deposition was punctuated by rejuvenated (active) growth induced by regional compression, most significantly during the Mid-Miocene. This event affected also the Dan and Kraka structures, which otherwise experienced very limited growth during the Cenozoic.
The Danish Salt Dome Province (Andersen et al. 1982), which forms the southernmost part of the Danish Central Graben, is characterized by numerous salt structures (Fig. 1). Most hydrocarbon exploration in this area is related to such structures with hydrocarbon production coming from Upper Cretaceous and Danian chalk reservoirs (Megson 1992). The main objectives of this paper are to describe the Triassic to presentday evolution of the three adjacent salt structures, Skjold, Dan and Kraka, all of which are associated with producing hydrocarbon fields, and to discuss the mechanisms that have determined their different evolutionary paths. In order to achieve this, a 3D seismic data set has been used to enable detailed structural mapping and direct comparisons of structures at different stratigraphic levels. This is extremely difficult to achieve with conventional 2D seismic data, especially in situations where spatially restricted or highly complex structures are covered by a sparse 2D seismic grid. In terms of salt structures, 3D seismic imaging enables changes in accommodation space induced by salt withdrawal to be mapped and related to the evolution and growth of the salt structures. The initial results of the study presented in this paper form the basis for further 3D modelling and analysis of the development of salt structures in the Danish Salt Dome Province, as well as a better appreciation of the style and development of traps in similar structural settings.
Regional geological setting The Danish Salt Dome Province is situated on a relatively undisturbed, highstanding shelf that forms the flank of the main Central Graben (Fig. 1). The Danish part of the Central Graben is located within the Mid-North S e a - R i n g k 0 b i n g - F y n High, where the rift system changes trend from N W - S E in the UK and Norwegian sectors, to N - S in the Dutch sector. It is an overall east-dipping graben system, which is subdivided into sub-basins separated by intra-basinal highs. The existence of an initial fracture system during the late Permian in the area of the Mesozoic rift probably enabled connection between the Northern and Southern Permian Basins. Although salt structures are
absent from the deeper parts of the Tail End Graben and Sogne Basin of the Norwegian and Danish sectors (Fig. l ), the seismic expression of disharmonic relations between Triassic and Zechstein reflectors suggests the presence of mobile evaporites to the north of the Salt Dome Province (Gowers & S~eboe 1985). It is therelore plausible that a pathway through the Tail End Graben and S0gne Basin connected these basins to the Salt Dome Province at the northernmost edge of the Southern Permian Basin. The presence of mobile salt and mature salt structures in the Horn Graben (Vejba~k 1990) suggests that this rift system may also have facilitated a similar passage from the Southern Permian Basin. E - W oriented extension, with subordinate N W - S E trending faults, was prevalent in the Danish Central Graben during the Triassic. Although this resulted in the transection of the North Sea H i g h - R i n g k c b i n g - F y n High system by N - S trending basins and half-grabens, a complete through-going graben system was not established at this stage (Sundsb0 & Megson 1993). This was facilitated by Jurassic extension, during which time the Danish Central Graben evolved under the influence of a combination of normal faulting and halokinesis. A three-phase model has been applied to the Mid-Jurassic-Early Cretaceous evolution of the Danish Central Graben, comprising an initial extensional phase in the Callovian to Early Oxfordian, succeeded by a second phase during the Late Kimmeridgian to Early Volgian, and a final, phase in the mid-Ryazanian (Gregersen & Rasmussen 2000). The first phase was possibly associated with a reactivation of Triassic faults, and was characterized by subsidence along N - S trending faults, including segments of the eastern graben-bounding fault, the Coffee Soil Fault and faults within the Salt Dome Province (Frandsen et al. 1987; Damtoft et al. 1992). Displacement associated with the second tectonic phase, during the Late Kimmeridgian-Early Volgian, took place along new N W - S E trending faults, with the development of NE-dipping halfgrabens in the central and northern part of the Danish Central Graben. However, the Salt Dome Province was less affected by both this and the final rift phase in the mid-Ryazanian, which was associated predominantly with reactivation of the N W - S E
DAVIES,R. J., CARTWR1GHT,J. A., STEWART,S. A., LAPPIN.M. & UNDERHILL.J. R. (eds) 2004.3D Seismic Technology:Applicationto the Exploration of Sedimentar3' Basins. Geological Society, London, Memoirs, 29, 149-163. 0435-4052/04/$15 9 The Geological Society of London 2004.
150
M. RANK-FRIEND & C. F. ELDERS
Fig. 1. Location of the Central Graben (after Errat et al. 1999. Davison et al. 2000: Buchanan et al. 1996, Warner 2001; Damtoft et al. 1992). The frame indicates the location of the study area in the Southern Salt Dome Province. trending faults (Gregersen & Rasmussen 2000), thus maintaining a relatively flat pre-Zechstein basement (Sundsbo & Megson 1993). Mapping by Vejb~ek & Britze (1994) reveals that the Top pre-Zechstein basement in the southern part of the Danish Central Graben is characterized by a N W - S E trending fault grain. Gentle inversion in the Late Cretaceous-Early Cenozoic modified and further amplified the existing structures. Locally this was associated with reverse faulting along the Coffee Soil Fault and, to a smaller degree, along faults in the overburden above salt structures, resulting in deposition of a thinner Chalk Group in these areas.
Data set and methodology The seismic database used for this study covers an area of approximately 750kin 2 and comprises part of a larger highresolution 3D seismic data set compiled from a number of individual surveys acquired by M~ersk Oil and Gas A/S during the 1980s and 1990s for exploration, appraisal and production purposes. The data are time-migrated, and characterized by a 12.5 m CDP spacing in the inline direction, with a 25 m crossline interval. Although the image quality deteriorates significantly adjacent to and below the salt structures, the resolution of the 3D seismic data is in general very good, allowing detailed interpretation and structural mapping down to at least Base Zechstein level. Twelve horizons were interpreted (Fig. 2) and correlated to nine wells within the 3D survey area (Fig. 3). Time structure maps and associated vertical time thickness maps were produced, the latter providing clear, time specific indications of the location and outline of areas of salt withdrawal and their relation to the salt structures. Systematic comparison of these time thickness maps illustrates changes in the location of withdrawal areas and their relationship to the evolution of the respective salt structures. An important aspect of seismic interpretation is the reliability of time-migrated data and thickness maps derived from them. Although pre-stack, depth-migrated data position reflectors more accurately and hence provide more accurate thickness maps, such data were not available for this study. Given the relatively simple overburden structure away from the immediate vicinity of the salt structures, time-migrated data are considered a reliable basis for the interpretation of the geological evolution of the salt structures and study area in
general. In addition seismic reflector patterns (onlap, offlap etc.) have been used to check the validity of interpretations derived from time thickness maps.
Geometry of the present-day salt structures The time structure map of the Base Chalk horizon provides a clear, present-day outline and location of the salt structures (Fig. 3). Skjold, which is associated with a sub-circular outline at this level, is a mature diapir with a N W - S E trending base or 'stem root' situated on the southern footwall side of a similarly trending extensional ridge system (Fig. 4). Grounding is observed between the diapir and the U-1 pillow structure, located to the south, which is only partially within the study area. The bulbous head of Skjold penetrates and terminates within the Upper Cretaceous Chalk Group, which, along with Palaeocene Danian chalk successions, form the hydrocarbon reservoir (Megson 1992). However, significant carapace doming is observed up to Mid-Miocene level (Fig. 4). The Dan structure is located approximately 10 km to the SE, directly along strike from the Skjold diapir (Fig. 3). Despite being associated with a relatively simple sub-circular outline at Base Chalk level, Dan is characterized by a more complex geometry. It can be defined as consisting of an upper and lower part, where the latter is associated with a central, asymmetric Zechstein salt core, whilst the upper part is composed of salt at a minimum of two different stratigraphic levels within MiddleUpper Triassic strata (Fig. 5). Based on 2D seismic data, JCrgensen (1992) interpreted the upper part of the structure as consisting of entirely Triassic salt (Dudgeon Saliferous Formation). On the other hand, Sundsbr & Megson (1993) interpreted the intra-Triassic structure to be composed of intruded Zechstein salt. The latter suggestion was based on interpretation of 3D seismic data, including the data set on which the present study is based. Adopting the strike-slip model of Cartwright (1987) for the Danish Central Graben, Sundsbr & Megson (1993) argue that episodic dextral and sinistral strikeslip motion and associated extension within the Dan Transverse Zone would have permitted Zechstein salt to escape and intrude overlying Triassic strata. Graversen (1994) also suggested that the location of piercements was governed by intersections between graben-parallel faults and crosscutting, transverse faults at basement level. The present study suggests that the
EVOLUTION OF CENTRAL GRABEN SALT STRUCTURES
151 Seismic units & horizons in this study
DCG
System
Stage
lithostratigraphic units
Lithology
Cen~ drl[
--
Base Q u a t e r n a r y
C*.m-5 Url~
C e n c z c
M ccene
.4~ Oligocene
I ntra Ofiqocene
c
_
G r
Ecce-'e
. .'., ,',: .-., ;~-
."
o ,J D -
Paleocene Dam, an
_ .
f~'eaT Base Eocene
Cen-1
o q a k-6 O,l~
m n m m m m m n l i ~ U IIIn~l~ In n n n l m u n l u m rumnun~mmnn n m m n m , m F~ i ~ , ~ ~ m mm m ~ n l m m u mi~ I m n n ~ u ~ m HUUl n u n n m U ~ i n m u ~ u l ~ mm m m m n n m m i ~ u ~ m l l m m m u n n , m ~ l m I m m m ~ U ~ l l m u n n n m n n ~ i~ a n ~ nnmna~m IN
Upper
-
,..
"i'~'--'-----'-"
~-'."~"~'-'~
- ~
9
Near To;] Chalk
C n a I,:-5 Un~ Cqalk-4 Ur)~f Chd~k-2 Un:l C h a l k - 2 Ullll
._ Turemap Sh c h a i n . 1 Un,1 Rodtl'r Fin
Lower
-,y
........
,,"
K
k, alh~ II Fm ~ . ,., /
..
'
: ":--:-
Upper
":-~-
Base Crerlaceotls
,~t
!qha Kn~rq~,lidglan
.1t
Base C a l l o { l a n
Loia Fm Mm,,k. Graben Shat.P Fr'n
: j
Middle
8w,,~ I
~
Graberl S a n d Frq
II
- :~ "
': >"L
<:?.."
:
,
.:
-
~.~--
Lower
,, , .
-'J'.
,."
.
. . . . . .
F er':lsle, ~m
'.. 9 '-
Upper
S 9
9, "
".
MiddIe
9'
-" " ,' ,.
9
.~d nl~rl(,r" F n ,
h
"
Lower
- - - qlt
....
B ": a ,: t, k " ,~ F m ~
Do,-.~,ng Dc)iomlt,c F m '~
-
T~)~J Lo,,,,er Triassic
Baclon Gr
..
9-,
%
%
",,
,,,
u
_
_
%
". "" ". "" ". "" ". "" ". "" " - ' ....,.....,.... ,.-., 9. .. ., .. .. ,,
Upper
R M I A N
- "~ ~ r , a J p p c r - n a s s , c "~
: J?'~.
...,.., ,........:.....,: P E
Base M~Jdle Jtlrasslc
..
Zechslell} Gr
"~'.~ " v z , . ~
l
n
I
u
u
I.:
Lower -~
" 9 .......
Fig. 2. General stratigraphic column of the Danish Central Graben, showing the interpreted seismic horizons and units of the present study (modified from Errat 1993; Errat et aL 1999; Frandsen et al. 1987; Jc~rgensen 1992).
-
Sandszcne ~Cor'tmenta/fuvla[,
S~-ae ~Cortinenta
7 9; ;7 . -. 9 9 ~[[.I. ".'[.'. ".'[1".
~
Sandsto,~e_marine ,Sqallow & deepi
asymmetric salt core is a roller type salt structure with a N N W SSE trending fault which detaches along its western flank. Whilst it is difficult to clarify whether there is any present-day salt up-dip along this relatively steep master fault between the lower and upper part of the structure, it is clear that the intra-Triassic salt bodies and the underlying Zechstein salt are only slightly juxtaposed in places along the fault. It is
v'oicanlcs
L"-:'I;"
"..''..''e"
Evapcrltes
Dmestone/chalK
;, ; " i
' "." ' " . " " "." '
"-~,'..'_ 2 9 -"
Shabe ~M a r i n e )
suggested that while Triassic salt is indeed present, it has primarily provided weak zones or surfaces along which Zechstein salt has intruded (see discussion). Whilst there is no penetration of salt beyond Triassic age strata, deformation and updoming of the overburden is prevalent up to Eocene level, most significantly affecting Jurassic-Cretaceous successions (Fig. 5).
152
M. RANK-FRIEND & C. F. ELDERS
Fig. 3, Shaded relief time structure map (ms TWT) of the Base Chalk Group (Base Middle Cretaceous), from which the location and present-day outline of the three salt structures, Skjold, Dan and Kraka, is clearly observed. Notice also the U-I pillow structure located to the south of the Skjold diapir. The locations of the nine wells used to calibrate the seismic interpretation are indicated. Illumination from NE.
The present-day Kraka structure is an elongated, slightly asymmetrical pillow (swell), which is located south of and parallel to the Skjold-Dan alignment (Figs 3 & 5). As in the area of the Dan structure, the Base Zechstein horizon is relatively fiat. The exception is a larger, N N W - S S E trending normal fault, which transects the SW part of the study area just to the west of the Kraka pillow itself (Fig. 5).
Evolution and growth of the salt structures Triassic Uppermost Triassic strata thicken into active faults (Fig. 5), suggesting that extensional faulting was a central factor in the Late Triassic evolution of this part of the Central Graben. In the
Fig. 4. NE-SW orientated seismic traverse through the Skjold diapir, and associated interpretation. The insert indicates location of the traverse. Notice the gas cloud above Skjold. The associated velocity push down may be partly responsible for the apparent crestal graben at Mid-Miocene level.
EVOLUTION OF CENTRAL GRABEN SALT STRUCTURES
153
Fig. 5. NE-SW orientated traverse through the Dan and Kraka salt structures with a small part of the Coffee Soil Fault to the NE of the study area. The insert within the associated interpretation indicates the location of the traverse.
study area this was manifested by early thick-skinned extension along the graben-bounding fault, the Coffee Soil Fault, and the formation of a N - S trending depocentre in its hangingwall (Fig. 6a). N - S and N N W - S S E orientated faults dominate the east of the study area (Fig. 7). Whilst there is no clear indication that the Dan structure itself had begun to form at this point, there is evidence of a high immediately to the NE of its present-day location. A number of the Triassic faults detach on the side of this structure, suggesting that initial salt development was occurring in the immediate vicinity (Figs 5 & 6a). Similarly, further to the west, in the area of the present Skjold structure,
Late Triassic syn-kinematic strata suggest that active extensional faulting was associated with the N W - S E trending fault system along which the present-day Skjold diapir is aligned, resulting in the formation of a depositional low to the north (Figs 4 & 6a). While it is impossible to deduce whether there was any salt migration associated with this extension, an initial Kraka structure clearly appears to have been established by the end of the Triassic, with a slightly thinner overburden suggesting the outline of an early N W - S E trending pillow structure (Figs 5 & 6a). Although the immediately underlying fault segment at Rotliegend level is trending N N W - S S E , there
154
M. RANK-FRIEND & C. F. ELDERS
EVOLUTION OF CENTRAL GRABEN SALT STRUCTURES
155
156
M. RANK-FRIEND & C. F. ELDERS
Fig. 6, Vertical time thickness maps (ms TWT). (a) Upper Triassic-B.M. Jurassic: Extensional faulting is associated with the NW-SE trending fault system in the area of present-day Skjold (note: salt wall intrusion accounts for the interpretation gap in the area of Skjold, with the white dotted lines indicating the estimated width of the underlying extensional fault system). Notice how the apparent fault system interconnects with the NNW-SSE trending faults in the area of the present-day Dan structure. The thinner overburden in the area of Kraka indicates the outline and growth of a NW-SE trending pillow structure. (b) Upper Jurassic: Kraka and U-1 salt structures form part of a larger NW-SE trending salt ridge system. (c) Lower Cretaceous: Skjold has undergone a transformation from salt wall to point source structure, while Dan has developed a similar circular outline. This is associated with the formation of distinctive fault patterns. Notice radial and concentric faults in the Skjold overburden. (d) Upper Cretaceous: a change in the general accommodation pattern to NW-SE trending thickness isolines is associated with similar trending salt structures. Skjold has become diapiric with a distinctive, thick withdrawal cell to the south. Notice the clearly defined feeding channel. (e) Paleocene: passive growth associated with Skjold. (f) Mid-Oligocene-Mid-Miocene: passive growth associated with Skjold, while Kraka and Dan are dormant. (g) Mid-Miocene-Base Quaternary: rejuvenation of the three salt structures. (h) Base Quaternary time map = Quaternary thickness.
seems to be a correlation between the orientation of Kraka and the trend of the overall N W - S E trending pre-salt fault grain present in the study area and commonly observed at Base Zechstein level throughout the Danish Central Graben (Fig. 1: Vejb~ek & Britze 1994).
Mid-Late Jurassic The structural evolution of the study area during M i d - L a t e Jurassic was associated with renewed thick-skinned extension along the eastern margin of the Central Graben, resulting in the development of two separate depocentres adjacent to the Coffee Soil Fault (Fig. 6b). Activity along new N - S trending faults, antithetic to the main graben fault, and immediately west of the depocentres was probably induced by graben formation (Fig. 8). Renewed halokinesis was a significant characteristic of this period, probably responding to decoupled, extension-induced stretching and faulting of the overburden caused by graben formation and activity along the Coffee Soil Fault. Thickness contours and a significantly thinner unit in the area of Kraka suggest that the structure formed part of a larger N W - S E trending salt ridge (swell) during the Late Jurassic (Fig. 6b). As well as activity along N W - S E trending faults, the M i d - L a t e Jurassic Kraka overburden is characterized by extensional growth faults, orientated at a high angle to the elongated structure (Figs 5 & 6b). These relatively small crestal faults are related to the growth and shape of the salt structure which exhibited relatively advanced relief at this time. The area in
which Dan is currently located was characterized by similar decoupled extensional faulting with which salt structure growth was clearly associated. Extension was particularly prominent along the two faults orientated N N W - S S E and N E - S W respectively. In this area the intra-Triassic salt also provides a surface along which the N E - S W trending fault detached, whilst renewed activity along the N N W - S S E trending fault was facilitated by detachment along the underlying Zechstein salt core (Fig. 9). Similar multiple detachment surfaces are commonly observed in the southern North Sea (Fullarton 2000). However, it was not until the very late Jurassic that a vaguely expressed outline, indicated by a thinner overburden, suggests the evolution of an initial salt structure in the Dan area. Growth associated with the salt structure to the east of Dan became less significant during the Late Jurassic (Fig. 6b). The period was associated with renewed activity along the N W - S E trending fault system in the area of the Skjold structure. Beyond further accentuating the general depocentre to the north, this most likely facilitated initiation or amplification of a linear salt wall.
Lower Cretaceous In addition to marking a significant change in the fault and depositional patterns, the post-rift Cretaceous period defined a major transition in the morphological evolution of the salt structures. As a result of along-axis salt migration, Skjold evolved from a linear salt wall to a sub-circular point source structure with an upward-doming central segment during the
Fig. 7. 3D time structure map (ms TWT) of the Base Middle Triassic, which is characterized by an overall N-S fault trend towards the Coffee Soil Fault, while NW-SE and NNW-SSE trending faults dominate further to the west. A clearly defined fault of the latter group is observed in the present-day area of Dan
EVOLUTION OF CENTRAL GRABEN SALT STRUCTURES
157
Fig. 8. 3D time structure map (ms TWT) of the base Upper Jurassic+ where the NNW-SSE trending fault in the area of Dan appears less significant.
very latest Jurassic-Early Cretaceous (Fig. 6c) associated with the formation of radial and concentric faults at Base Cretaceous level (Figs 6c & 10). Most prominent to the north of the structure, the longest radial faults exceed a length of 5 km from the crest, propagating towards the area of greatest subsidence. Concentric faults, on the other hand, developed only over the outer edge of the diapir crest itself, and did not extend beyond the position of the vertical walls of the bulbous head. Similar fault patterns, local to mature salt diapirs, have been observed in the overburden of other Central Graben salt structures, where they have been interpreted to mark or be associated with the salt wall to point source transition. The timing of the Skjold transition is contemporaneous with that of many other Central Graben diapirs (Davison et al. 2000; Warner et al. 2002) and corresponds to the transition from active rifting to passive subsidence in the graben system. The Early Cretaceous was likewise a key stage in the evolution of the Dan structure, which developed a near circular outline during this period (Fig. 6c). As observed in analogue
models of domal uplift (Withjack & Scheiner 1982), radial faults develop over the central part of evolving structures to accommodate curvature in the overlying strata (Fig. 6c).
Upper Cretaceous
The Upper Cretaceous defined an overall change in the depositional pattern from the previous depocentre in the north, to N W - S E trending thickness contours, which characterize the Upper Cretaceous-Cenozoic untis (Fig. 6d). This marks a shift from the filling of rift-related bathymetry to more post-rift subsidence. Most significantly, while axial salt migration and withdrawal from the north enabled Jurassic-Early Cretaceous growth, including the transformation from salt wall to point source structure during the very latest Jurassic-Early Cretaceous, further growth of the Skjold diapir during the Late Cretaceous was facilitated by withdrawal from the south. This new trend in salt flow into the evolving structure resulted in a clearly defined N W - S E trending salt withdrawal cell and an
Fig. 9. NW-SE trending traverse through Dan. The insert within the associated interpretation indicates the location of the traverse. Notice that the Dan structure is composed of Zechstein salt at its original level, and intruded into at least two levels within Mid-Upper Triassic strata.
158
M. RANK-FRIEND & C. F. ELDERS
Fig. 10. 3D time structure map (ms TWT) of the Base Cretaceous. The insert is an amplitude image of the Skjold structure at Base Cretaceous level showing radial and concentric faults. associated narrow feeder channel into Skjold, with diapir growth becoming confined to a smaller central part of the structure (Figs 4 & 6d). This salt migration from the broad Jurassic-Early Cretaceous N W - S E trending salt ridge (swell) resulted also in a complete separation of the Kraka and U-1 pillow structures. Both elongated in outline, the U-1 structure obtained a slightly more N N W - S S E orientation than N W - S E trending Kraka. A similar N W - S E trending, elongated outline characterized Late Cretaceous growth associated with the Dan structure. Cenozoic
The Lower Cenozoic time structure maps display regional polygonal fault patterns throughout the entire study area (Figs 11 & 12). Typically these are associated with overpressured mud-rocks in the central North Sea basin
(Cartwright & Lonergan 1996), but are suppressed by a dense pattern of radial faults in the overburden of the Skjold diapir. Similar observations and interpretations of diapirassociated radial faults (which accommodate the processes that generate the polygonal faults) have been made in association with other Central Graben diapirs (Davison et al. 2000; Davis 2003). As both radial and polygonal faults are confined to strata-bound 'tiers', which suggests that they form episodically (Davis 2003), it would appear that diapir growth continued throughout the Cenozoic. This, combined with significantly reduced thicknesses over the central part of the structure and a lack of any associated nearby salt withdrawal areas, suggest clearly that Skjold evolved by passive diapirism throughout the Cenozoic (Figs 6e-h). In addition, a collapse graben formed over the Skjold diapiric head during the Early Cenozoic (Figs 4 & 6f), characteristic of late stage development of mature salt diapirs which
Fig. 11. Shaded relief Mid-Oligocene time structure map (ms TWT).
EVOLUTION OF CENTRAL GRABEN SALT STRUCTURES
159
Fig. 12. Shaded relief Mid-Miocene time structure map (ms TWT): Regional polygonal faulting is suppressed by radial faulting adjacent to the Skjold diapir which, like Dan and Kraka, is associated with an area of downwarping.
deform by passive growth (Davison et al. 2000; Alsop 1996). A deviation from the general passive growth pattern was induced by periods of inversion, most particularly associated with the Mid-Miocene compressional event (Badley et al. 1989), which caused a further amplification of the Skjold salt structure relief (Fig. 6h). Onlap of reflectors onto the Mid-Miocene unconformity occurs adjacent to Skjold (Fig. 4) and is also a typical phenomenon of other Central Graben salt structures at this stratigraphic level (Davison et al. 2000). Similarly, faulting in distinct zones above salt walls and diapirs in the southern North Sea has been related to mid-Paleocene and Miocene inversion events (Oudmayer & Jager 1993). Rejuvenation was also associated with the Dan and Kraka salt structures, where Cenozoic growth (apart from that associated with this Mid Miocene event) was negligible (Figs 5 & 6e-h). Thus Kraka and the U-1 structure, which had also ceased to grow by the end of the Cretaceous, remained deep-seated, non-piercing pillow structures.
Discussion The very different morphologies and growth histories associated with the three salt structures provide a good basis for discussing several aspects of salt structure evolution.
Early stage evolution: reactive growth Although difficult to conclude from the present-day Skjold diapir, the early, elongated salt ridge/swell formation associated with Kraka, and probably in the area of the Dan structure, does indeed show that salt migration into linear salt swells/ridges was prevalent during the Late Triassic period. This was contemporary with Late Triassic active faulting, and although it is impossible to show a direct relationship with salt structure formation, this, coupled with physical modelling (Vendeville & Jackson 1992a, b; Withjack & Caiioway 2000) suggests strongly that extensional faulting (thick- and thin-skinned) triggered initial salt swell/wall formation. However, fault reactivation during the Mid-Late Jurassic rift period appears to have played an essential role in further salt structure development. Thick-skinned extension along the Coffee Soil Fault, and to a lesser degree along basement faults within the study area itself
(Figs 4, 5 & 9), effectively caused overburden stretching and local thin-skinned faulting throughout the area, enabling renewed salt wall/swell growth. The initial Dan structure probably formed a similar, confined linear N N W - S S E trending structure during this period.
The role o f Triassic salt A series of seismic traverses, extracted from the 3D seismic data set at a high angle to the N N W - S S E trending fault system to the south of and through Dan structure, provide a good insight into its structural complexity (Figs 13 & 14). Conformably encased salt wedges thickening into and terminating against the fault, which detaches along an underlying steep-sided Zechstein salt core, can be observed at a minimum of two different stratigraphic levels within the Mid-Upper Triassic succession. However, whilst underlying Lower Triassic strata clearly characterized by parallel, undeformed sediments, are conformable stratigraphically with the underlying Zechstein unit, successions above the intra-Triassic salt are faulted as well as folded around the salt bodies (Fig. 14). The overall thicker unit associated with the presence of salt in the area is clearly visible on the vertical time thickness map of the Mid-intra-Upper Triassic unit (Fig. 13), where the present-day intra-Triassic salt distribution forms an elongated feature, situated mainly in the hangingwall of the NNW-SSE trending fault system. In the area of Dan, where intra-Triassic salt is interpreted to be present on either side of the fault, the thicker unit forms an almost circular outline similar to that associated with the Dan structure at Base Chalk level (Fig. 3). While the seismic stratigraphic configuration suggests isolated salt movement at different stratigraphic levels, with salt structure development within Triassic strata taking place independently from that at Zechstein level, a pivotal question is whether there is any generic relationship between intra-Triassic and underlying Zechstein salt. Although it is difficult in some places to establish a present-day connection between the stratigraphically separated salt, the Zechstein salt and intra-Triassic salt is juxtaposed clearly at other parts along the N N W - S S E trending fault. In the light of this, it is suggested that Zechstein salt was intruded vertically and along weak planes/horizons within the Mid-Upper Triassic successions to form an initial elongated N N W - S S E trending salt wall system. Whilst Late Triassic faulting may have triggered this initially, further growth and salt wall formation
160
M. RANK-FRIEND & C. F. ELDERS
Fig. 13. Vertical time thickness map of Base Mid-Triassic- intra- Upper Triassic with the location of transverses through the Dan structure.
was facilitated by renewed normal extension during the MidLate Jurassic rift phase.
Mid-stage evolution: salt wall to point source structure transition The significant change in salt structure morphology during the Cretaceous is linked closely to the transition from regional extension to post-rift thermal subsidence. The Late Jurassic depocentre was not yet filled in the Early Cretaceous and continued to influence deposition and maintain differential loading imposed upon the salt layer. Although salt migration paths were not cut off, a cessation in thick- and thin-skinned extensional faulting prevented further reactive salt wall growth, permitting other driving mechanisms to predominate. A domal, circular outline developed by the Dan structure during this period evolved at the intersection between thin-skinned N W SE and N N W - S S E faulting within Triassic and Jurassic strata, where a weaker and thinner overburden enabled confined structural growth, aided possibly by further salt intrusion. The reason why Dan (like Skjold) did not become diapiric is probably because the growth of Dan at this point was associated primarily with the allocthons within Triassic strata, and that these were not capable of piercing through the relatively thick, competent overburden. A salt wall to point source structure transformation, associated with Skjold, was probably facilitated by similar thin-skinned extension and accompanying lateral thickness variations and weaknesses in the overburden. Skjold pierced the overburden, and growth continued to keep pace with sedimentation into the early part of the Late Cretaceous. The availability of nearby mobile salt, which was actively withdrawn from a syncline to the south of the structure, facilitated this active growth (active phase sensu stricto, Vendeville & Jackson 1992a, b). However, once this source was exhausted, the structure was no longer capable of breaking through the overburden of competent chalk (Korstg~.rd et al. 1993; Warner 2001), causing a cessation of active growth and subsequent transition to downbuilding/passive growth. Up until this point, the crest of the structure remained close to the surface with dissolution probably influencing growth during this stage. The absence of an earlier point source transformation, subsequent to the Triassic rifting, is probably related to several aspects. Of these, a less evolved salt wall relief and a differential
imposed pressure are probably important factors. This may be the reason why the broad, low relief salt ridge system, of which Kraka and the U-1 structure were part, never became diapiric.
Late stage evolution: passive diapirism Growth associated with the mature Skjold diapir during the Cenozoic was similar in style to that of other Central Graben diapirs, which have been interpreted as having evolved by downbuilding/passive growth during this period (Davison et al. 2000). However, growth did not keep pace with rapid sedimentation and the crest did not remain close to the depositional surface, becoming buried progressively during the Cenozoic. Passive growth was punctuated during periods of inversion, most significantly during the Mid-Miocene, where induced compressional stresses were absorbed by the salt structure, which responded by vertical growth. The response similarly affected and rejuvenated the Dan and Kraka structures, which were otherwise dormant during the Cenozoic.
Conclusion The conclusions from this 3D seismic study can be summarized as follows: (1)
Whilst the location of the salt structures seems to have been determined by Late Triassic extensional faulting of both a thin- and thick-skinned nature, decoupled M i d Late Jurassic extensional faulting was essential in enabling further salt structure growth into linear salt wall/swell features. Thus, Skjold evolved as an elongated salt wall along an underlying N W - S E trending fault at Rotliegend level, while a broader salt swell formed in the area of the present Kraka and U-1 structures. Similarly to Skjold, the latter structure was most likely initiated due a combination of thick- and thin-skinned extensional faulting related to an underlying fault system. Lacking any apparent underlying faulting, a linear N N W - S S E trending salt structure formed in the area of and to the south of the present-day Dan structure. This linear feature is interpreted as consisting of intruded Zechstein salt. While this conclusion is similar to that of Sundsbr & Megson (1993),
EVOLUTION OF CENTRAL GRABEN SALT STRUCTURES
161
Fig. 14. The Dan structure: S W - N E trending seismic traverses through the Dan structure, with associated interpretations. For location of traverses see Figure 13.
162
(2)
(3)
(4)
(5)
M. RANK-FRIEND & C. F. ELDERS we do not see any evidence of strike-slip faulting within the study area, and conclude that the intrusion of Zechstein salt was facilitated by decoupled, extensional faulting, probably related to thick-skinned faulting along the graben bounding Coffee Soil Fault. The post-rift, thermal subsidence regime of the Cretaceous marked a significant change in the evolution of the salt structures. Underfilling of the Jurassic depocentre to the north of the study area continued to control Early Cretaceous deposition, promoting further differential loading, However, cessation of extensional faulting prevented further salt wall growth, forcing Skjold to become a point source structure. Dan, situated at a Triassic-Jurassic fault junction adjacent to the depocentre, established a similar sub-circular outline during the Early Cretaceous. Along-axis salt migration into Skjold during the Late Jurassic-Early Cretaceous was succeeded by Late Cretaceous salt withdrawal from the broad N W - S E trending salt ridge to the south of it. Kraka and the U- 1 structure, which formed part of this system up until this point, became individual pillow structures. Except for periods of compression, particularly during the Mid-Miocene, which triggered renewed growth and the development of unconformities with prominent onlap surfaces in the overburden of all study structures, growth had ceased for all but the Skjold diapir by the beginning of the Cenozoic. The mature Skjold diapir grew predominantly by downbuilding/passive growth during the Cenozoic. Although rapid sedimentation exceeded the rate with which the structure grew and the crest of the structure became deeply buried during the Cenozoic, passive growth is suggested by a combination of features. These include significantly thinner strata over the crest compared with correlative successions immediately adjacent to it, a crestal collapse graben, and radial faults in the overburden away from the diapir. The lack of crestal polygonal faulting, when otherwise dominating the area, and absence of clearly defined salt withdrawal cells also point towards passive growth. This deep-seated passive growth does not fit the passive growth stage of the Vendeville & Jackson model (1992a, b), which dictates a near/at surface growth. Instead it is possible that the overburden acted in a ductile manner, and growth was enabled due to a buoyancy/ loading effect.
The authors would like to thank S, Stewart, M. Lappin and K. Sorensen for constructive reviews. This work was carried out using software donated generously to Royal Hoiloway College, University of London, by Schlumberger-Geoquest. This paper has been published with permission from the Geological Survey of Denmark and Greenland (GEUS).
References ALSOP, G. I. 1996. Physical modelling of fold and fracture geometries associated with salt diapirism. In: ALSOP, G. I., BLUNDELL, D, & DAVISON, I. (eds) Salt Tectonics. Geological Society, London, Special Publications, 100, 227-242. ANDERSEN, C., OLSEN, J. C., MICHELSEN, O. & NYGAARD, E. 1982. Structural outline and development. In: MICHELSEN, O. (ed.) Geology of the Danish Central Graben. Geological Survey of Denmark, Series B, 8, 9-26. BADLEY, M. E., PRICE, J. D. & BACKSHALL, L. C, 1989. Inversion, reactivated faults and related structures: seismic examples from the southern North Sea. In: COOPER, M. A. & WILLIAMS,G. D. (eds) Inversion Tectonics. Geological Society, London, Special Publications, 44, 201-219.
BUCHANAN, P. G., BISHOP, D. J. & HOOD, D. N. 1996. Development of salt-related structures in the Central North Sea: results from section balancing. In: ALsOe, G. I., BLUNDELL,D. & DAVlSON,I. (eds) Salt Tectonics. Geological Society, London, Special Publications, 100, 111-128. CARTWRIGHT, J. A. 1987. Transverse structural zones in continental rifts--an example from the Danish sector of the North Sea. /tt: BROOKS,J. 8z GLENNIE. K. (eds) Petroleum Geology of North West Europe. Heydon, London, 441-452. 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, i 83-193. DAMTOFT, K.. NIELSEN. L. H., JOHANNESEN, P. N., THOMSEN, E. ANDERSEN, P. R. 1992. Hydrocarbon plays of the Danish Central Trough. In: SPENCER, A. M. (ed.) Generation Accumulation and Production of Europe's Hydrocarbons II. Springer-Verlag, 35-58. DAVIS, T. H. 2003. Tertian' faulting patterns and growth history of central graben salt diapirs. PhD Thesis, University of London. DAVISON, I., ALSOP, I., BIRCH, P., ELDERS, C., EVANS, N., NICHOLSEN, H., RORISON, P., WADE, D., WOODWARD, J. & YOUNG, M. 2000. Geometry and late structural evolution of the Central Graben salt diapirs, North Sea. Marine and Petroleum Geology, 17, 499-522. ERRAT, D. 1993. Relationship between basement faulting, salt withdrawal and Late Jurassic rifting, UK Central North Sea. hi: PARKER, J. R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 1211-1219. ERRAT, D., THOMAS,G. M. & WALL. G. R. T. 1999, The evolution of the Central North Sea Rift. b~: FLEET, A. J. & BOLDY, S. A. R. (eds) Petlvleum Geology of Northwest Europe: Proceedings of the 5th Conference. Geological Society, London, 63-82. FRANDSEN, N., VEJB~K, O. V., MOLLER, J. J. 8.~ MICHELSEN,O. 1987. A dynamic geological model of the Danish Central Trough during the Jurassic-Early Cretaceous. In: BROOKS,J. & GLENNIE,K. (eds) Pettvleum Geology of North West Europe. Heydon, London, 453 -468. FULLARTON, L. D. 2000. Structural anaIvsis of the jupiter and camelot areas, southern north sea. PhD Thesis, University of London. GOWERS, M. B. & S~BOE, A. 1985. On the structural evolution of the Central Trough in the Norwegian and Danish sectors of the North Sea. Marine and Petroleum Geology, 2, 298-318. GRAVERSEN, O. 1994. Interrelationship between basement structures and salt tectonics in the Salt Dome Province, Danish Central Graben, North Sea (abstract). In: ALSOP, A. (convenor) Salt Tectonics: Programme and Abstracts. Geological Society, London, 13. GREGERSEN, U. & RASMUSSEN,E. S. 2000. The subtle play-potential of Upper Jurassic-Lower Cretaceous block-faulted turbidites in the Danish Central Graben, North Sea. Marine and Petroleum Geology, 17, 691-708. JORGENSEN, L. N. 1992. Dan Field. hi: BEAUMONT,E. & FOSTER, N. (eds) Structural Traps VI. American Association of Petroleum Geologists Treatise of Petroleum Geology, Atlas of Oil and Gas Fields, 199-218. KORSTGARD, J. A., LERCHE, I., MOGENSEN, T. E. & THOMSEN, R. O. 1993. Salt and fault interactions in the northeastern Danish Central Graben: observations and inferences. Bulletin of Geological Socie o, of Denmark, 40, 197-255. MEGSON, J. B. 1992. The North Sea Chalk Play: examples from the Danish Central Graben. In: HARDMAN, R. F. P. (ed.) Exploration Britain: Geological hlsights for the Next Decade. Geological Society, London, Special Publications, 67, 247-282. OUTDMAYER,B. C. & JAGER DE, J. 1993. Fault reactivation and obliqueslip in the Southern North Sea. In: PARKER, J. R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 1281 - 1290. SUNDSBO, G. O. & MEGSON, J. B. 1993. Structural styles in the Danish Central Graben. In: PARKER, J. R. (ed.) Petroleum Geology of
EVOLUTION OF CENTRAL GRABEN SALT STRUCTURES
Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 1255-1267. VEJB~K, O. V. 1990. Rift zones in the continental crust of Europe: geophysical, geological and geochemical evidence; Oslo-Horn Graben. Tectonophysics, 178, 29-49. VEJB,~K, O. V. & BRITZE, P, 1994. Toppre-Zechstein (TI'W and depth ). Geological map of Denmark 1:750 000. Geological Survey of Denmark. Map Series, No. 45. VENDEVILLE, B. C. & JACKSON, M. P. A. 1992a. The rise of diapirs during thin-skinned extension. Marine and Peovleum Geology. 9, 331-353. VENDEVILLE, B. C. & JACKSON, M. P. A. 1992b. The fall of diapirs during thin-skinned extension. Marine and Petroleum Geology. 9, 354-389.
163
WARNER, M. J. C. 2001. Mesozoic structures of the central graben: implications for earlv diapir evolution. PhD Thesis, University of London. WARNER. M. J. C., ELDERS, C., DAVIS, T. & RANK, M. 2002. Salt tectonics above complex basement extensional fault systems: results from 3D seismic analysis of central Graben salt structures (abstract). American Association of Petroleum Geologists 2002 Annual Meeting Abstracts. 185. WITHJACK, M. O. & CALLOWAY, S. 2000. Active normal faulting beneath a salt layer: an experimental study of deformation patterns in the cover sequence. AAPG Bulletin, 85, 627-651. WrrmAfK. M. O. & SCHEINER, C. 1982. Fault patterns associated with domes--an experimental and analytical study, AAPG Bulletin, 66, 302-316.
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 BRUCE
D. TRUDGILL
1"3 & M A R K
G. ROWAN
2"3
lDepartment of Earth Science and Engineering, Imperial College, London SW7 2AZ, UK (Current address." Department of Geology and Geological Engineering, Colorado School of Mines, Golden, CO 80401, USA; (e-mail: [email protected]) 2Rowan Consulting Inc., 850 8th St., Boulder, CO 80302, USA 3Formerly at the Energy & Minerals Applied Research Center (EMARC), University of Colorado, Boulder, CO 80309, USA
Abstract: By integrating 3D and 2D seismic interpretation with structural restorations we have reconstructed the evolution of a complex, composite stepped counter-regional salt system in the West Delta/South Pass (WDSP) area of the northern Gulf of Mexico. Biostratigraphically calibrated well data allow the last I0 Ma of the evolution of the salt system to be divided into six stages: (1) sea-floor extrusion of isolated salt tongues fed from the Jurassic Louann salt through northward dipping feeders prior to 7.5 Ma; (2) amalgamation of the salt tongues to form a salt-tongue canopy between 7.5 and 6.4 Ma; (3) counterregional evacuation of the salt-tongue canopy as a result of enhanced sediment loading due to progradation of the shelf margin between 6.4 and 5.0Ma; (4) evacuation of salt into a series of salt walls linking salt domes between 5.00 and 2.55 Ma; (5) evacuation of the salt walls to form counter-regional fault welds between 1.95 and 0.5 Ma; and (6) final evacuation of most of the salt from deeper levels leaving a series of isolated salt domes connected by counter-regional fault welds. The counterregional evacuation of the WDSP salt systems illustrates the value and limitations of published 2D models for allochthonous salt, and the reconstructed evolution yields insights into the complex interactions between salt deformation and sedimentation. The results also suggest that the WDSP salt systems significantly affected sediment transport pathways, trap geometries and possibly late stage petroleum migration across evacuating salt welds.
It is now well established that the offshore northern Gulf of Mexico is characterized by a complex pattern of salt structures, many of which are allochthonous bodies, detached from their original Jurassic source layer (e.g. McGuinness & Hossack 1993; Diegel et al. 1995; Fletcher et al. 1995). This is most apparent at the Sigsbee Escarpment where shallow allochthonous salt has overridden deep basin strata for tens of kilometres. In contrast to this large tabular salt canopy, the eastern Louisiana shelf area comprises a series of isolated salt domes, linked by both basinward and landward (counter-regional) dipping faults (Fig. 1). In the West Delta and South Pass (WDSP) southern addition protraction areas, these isolated salt domes are particularly well defined (Fig. 2) and are associated with significant petroleum accumulations, e.g. the South Pass 89 field and the West Delta 133 field. The South Pass 89 field has been highly productive since its discovery in 1969. To date, 55 801 MBO, 46.15 BCF, and 2115 MBO condensate have been produced from Lower Pliocene/Upper Miocene reservoirs (Marathon production figures, pers. comm. 2000). The trap is a complex stratigraphic/structural trap along the steep to overturned eastern flank of the South Pass 89 salt dome (Dome B in Fig. 2). Adjacent salt domes are linked by a series of landward dipping, counter-regional faults (or fault welds), that show a large degree of expansion in their hangingwalls (Schuster 1995). These shallow salt bodies overlie a deeper, discontinuous system of allochthonous salt welds (evacuated salt bodies) and remnant salt (Schuster 1995). The main aims of this paper are to examine how the deeplevel salt systems evolved into the shallower salt dome geometries (in 3D and through time) and how this complex evolution may have influenced reservoir development, trap formation and petroleum migration from the underlying source rocks (see also Bland et al. 2000). Several different e n d - m e m b e r s for the evolution of allochthonous salt have been proposed and documented in the northern Gulf of Mexico, including stepped counter-regional
and roho systems (Schuster 1995; Diegel et al. 1995) and saltstock canopies (Rowan et al. 1994; Rowan 1995). The basic components of these systems are salt tongues and bulb-shaped salt stocks, respectively (Rowan 1997; Rowan et al. 1999). Ultimately, the only difference between these elements is the symmetry or asymmetry of salt extrusion: salt tongues are extruded basinward from basinward-leaning feeder stocks; whereas bulb-shaped salt stocks spread radially from vertical feeders. In the WDSP area the dominant salt style is that of steppedcounter regional salt systems. The 2D geometry and evolution of these salt systems as determined by Schuster (1995) has come to represent the type example of a stepped counter-regional salt system (Fig. 3) within the northern Gulf of Mexico basin. In this paper, we aim to build on Schuster's (1995) landmark publication by focusing on the three-dimensional geometry of the salt systems, and then use both 2D and 3D structural restorations to illustrate the changing salt geometry during its emplacement and subsequent evacuation. Finally, we use the restorations and isopach maps to illustrate the history of interaction between salt movement and sedimentary loading. The results of this study should be useful both to researchers who are investigating the processes of salt system evolution and to industry geoscientists who are exploring for petroleum around salt structures in stepped counter-regional salt systems worldwide.
Seismic and well database Two different seismic datasets were merged and interpreted (Fig. 4) for this study. The primary dataset used was a 1993 Western Geophysical, time-migrated 3D survey, with acquisition bin dimensions of 25 • 40m. We also interpreted a 1990 3 . 2 k m • grid of time-migrated 2D seismic data to extend our interpretations and tie some of the nearby wells.
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, 165-176. 0435-4052/04/$15 9 The Geological Society of London 2004.
166
B.D. TRUDGILL & M. G. ROWAN
Fig. 1. Tectonic map of the northern Gulf of Mexico (from Diegel et al. 1995, reproduced by permission of the AAPG) showing allochthonous salt (black), basinward-dipping faults (thin lines)+ counter-regional faults (thick lines), deepwater folds (dashed lines), and the location of the region shown in more detail in Figure 2 (boxed area). A selection of well and biostratigraphic data were available to calibrate our horizon interpretations (Fig. 4). Check-shot surveys from ten different wells were applied to 18 wells within the study area in order to overlay both foraminifera and nannofossil picks on the time seismic data. The biostratigraphy was then used in conjunction with the seismic data to identify seven maximum flooding surfaces (0.50, 1.35, 2.55+ 3.65, 4.30, 5.0 and 6.40Ma) that were correlated throughout the two seismic surveys where possible. Only one well penetrated the 6.40Ma surface, so the interpreted geometry of this deeper horizon is less well constrained than are those at shallower levels. Three further horizons (7.50, 9.10 and 9.90Ma), were tentatively correlated with older maximum flooding surfaces, in order to assign some age constraints to the early stages of salt development. It is emphasized that these ages are not based on nearby well correlations, but from regional seismic line correlations to published information and unpublished data from Andy Pulham (pers. comm. 1999).
the salt and adding this to the base salt time horizon to correct for velocity contrasts and approximate the depth geometry of the base salt surface) is a proxy for salt evolution in the area (Fig. 5). Assuming no subsequent deformation of the salt system, deeper areas identified by blue to violet colours represent regions where salt was extruded relatively early in the evolution of the system. In contrast, green to red colours represent shallower and therefore later stages of salt evolution. The geometry of the salt system is one of shallow salt domes linked by deeper salt welds that were interpreted by Schuster (1995) as representing an earlier, salt-tongue canopy (Fig. 3). The deeper welds generally lie to the north and west of the present day salt domes, indicating that much of the present-day salt was ultimately
Salt system geometry M a p geometry A map of the present day salt welds combined with a pseudodepth base salt (constructed by doubling the time-thickness of
Fig. 2. Map showing the distribution of salt systems and major fault families in southeast onshore and offshore Louisiana (from Schuster 1995, reproduced by permission of the AAPG). The boxed area covers the salt systems examined in this study.
Fig. 3. Idealized 2D model of a stepped counter-regional salt-tongue system (after Schuster 1995), developed from the evacuation of a salttongue system. Black areas are salt, and pairs of black dots indicate salt welds. Thin lines show the schematic form of stratigraphic horizons. The uppermost line in each figure represents the seafloor.
INTEGRATING 3D SEISMIC AND RESTORATIONS
Fig. 4. Map showing the distribution of seismic, well and biostratigraphic data used in the course of this study. A grid of 2D seismic lines that tie additional well data complements the area of 3D seismic data.
derived from deep feeders to the north of the study area. Locally, two deeper areas are identified within the WDSP salt systems. One of these is a N E - S W trending 'keel' structure that forms a distinct structural low across System A, the second is a deep level weld that links into system C (Fig. 5). A continuity time slice through the 3D seismic data volume at 2200 ms TWTT illustrates the map view geometry of the salt domes and associated faults at this time level (Fig. 6). Dark, rounded areas (low values of continuity) characterize the salt domes. A series of curvilinear faults connect the salt domes and extend northeastwards as a series of en echelon structures. To the northwest, a more linear, linked fault array is clearly developed (this is the rollover fault family in Fig. 7b). The development of these fault families is intimately linked to the evolution of the salt systems (Rowan et al. 1999), as will be shown by the restorations presented later in this paper. Profile geometry
Interpreted seismic profiles (Fig. 7) illustrate the geometry of the salt systems, fault families and stratigraphic horizons.
Fig. 5. Time structure map of the WDSP salt systems (A, B and C), derived from interpretation of both 3D and 2D seismic data donated by Western Geophysical. The map is a combination of welds and pseudo-depth base salt (constructed by doubling the time-thickness of salt and adding this to the base salt time horizon to approximate the depth geometry). Key shows the depth of the welds/pseudodepth base salt in seconds TWTF (yellow and red, shallow; pink and purple, deep). Thick yellow lines are seismic profiles illustrated in Figures 7a-c; thin blue lines are protraction area boundaries (SP, South Pass; WD, West Delta: MC, Mississippi Canyon).
167
The confidence level for the top salt interpretation is generally high, with the salt/sediment interface well resolved by the seismic data. The base salt interpretation is less well constrained, particularly beneath salt overhangs. Salt welds are difficult to resolve on individual lines, but confidence in their identification increased through rigorous mapping over a tight grid spacing across the whole dataset. To clarify the salt system geometries, the salt bodies on the seismic profiles in Figure 7 are shown in black, with the interpreted pseudo-depth base salt geometry. A typical dip profile oriented N W - S E and extracted from the Western 2D seismic survey is illustrated in Figure 7a. The main features of this profile are: (i) two shallow salt domes (Domes C and B), and (ii) a deeper counter-regional weld system that ramps up from the northwest, and connects Dome B with a deep part of salt system A. The counter-regional weld is interpreted based on horizon truncations and stratal geometries that are consistent across the 3D survey in this area. The weld system reaches a maximum depth of around 7.5s TWTT (approximately 12.5 km), where it may represent the Jurassic Louann salt (Schuster 1995). In the hangingwatl of the Dome A/B salt system, the shallower interpreted seismic horizons (6.4 to 2.55 Ma) thicken towards the weld. This is interpreted to represent the progressive, basinward evacuation of the salttongue system. Older seismic horizons show a more bowl shaped geometry, interpreted to represent initial vertical subsidence into the salt tongues (Rowan 1995: Rowan et al. 1999). The geometry of salt system A is well developed in Figure 7b. Dome A is the major feature on the profile, extending as a continuous body of salt from less than 1 s TWTT to over 6.5 s TWTT (approximately 9 k m of vertical relief). There is a distinct keel at the base of the salt body (see also Fig. 5), that represents a localized salt feeder (see below). The counterregional weld has a distinctly stepped profile, representing different levels of salt during the evolution of the system (Schuster 1995). On this profile, a family of rollover faults (Rowan et al. 1999) developed at the apex of the hangingwall rollover formed due to monoclinal bending of the hangingwall sediments during salt evacuation. A small salt body at 5 s TWTT represents the eastern part of the WD133 salt system that lies further to the northwest (Fig. 2). Salt dome B (the SP 89 Dome) occupies the central part of the salt system (Figs 5 & 6). The complex evolution of
168
B.D. TRUDGILL & M. G. ROWAN
Fig. 6. A continuity time slice through the Western Geophysical 3D seismic data volume at 2200 ms TWTI'. The dark, rounded areas (regions of low continuity) mark the geometry of the salt domes, with associated fault families marked by linear features in the repro~ cessed data.
this salt structure is only apparent through 3D visualization and restoration (see below). A single seismic profile (Fig. 7c) shows a broad salt dome that overhangs significantly on its southeastern flank. The deeper part of the system is linked to system A to the west. Interestingly, a small isolated salt body between systems B and C is linked to both by interpreted salt welds. This implies that salt dome B was sourced from more than one deep level salt system, an interpretation supported by some of the confusing drilling results encountered on the southeast flank of the salt dome (Ted Dohmen, pers. comm. 1998) where the main petroleum reservoirs are located. The deeper wells drilled in this area are generally deviated to the west, to track the overhanging flank of the salt dome. Initially these wells pass through strata that dip consistently to the southeast at about 65 ~. However, some distance below the 5.0 Ma horizon (Fig. 7c), the dipmeter data record very steep dips ( > 7 0 ~ towards the northwest. Biostratigraphic data for this northwest-dipping section indicate no apparent age increase with depth through a substantial thickness of section. One explanation for this change is that the wells are recording strata on either side of a steeply dipping salt weld. In Figure 7c, a weld is interpreted in this location that links Dome B to the small salt body to the southeast, and subsequently to Dome C. Alternatively, the geometry could be that of unconformity-bound halokinetic sequences that formed during passive rise of the diapir (Giles & Lawton 2002).
3D depth geometry In order to gain a better understanding of the complex 3D geometry of the linked salt systems in the WDSP area, we constructed a 3D model in several steps. First, gridded surfaces in two-way time were created for the top salt, base salt, welds, faults and stratal horizons. These were imported into Midland Valley's 3DMove and converted to depth using vertical ray paths, constant velocities for water and salt (1497m/s and 4572 m/s respectively) and an average depth-dependent velocity function for the surrounding sediments derived from nearby wells (z = 1 . 1 4 3 t - 322.1, where z is true vertical depth in metres and t is two-way travel time in ms). Figure 8a is a plan
view of the 3D model that has been greatly simplified from the original interpretation of the seismic data. Minor faults have been omitted and the salt and horizon surfaces have been edited and smoothed. The bright pink salt bodies show the present-day extent of salt in the sub-surface. Interpreted weld surfaces are shown in red (the deeper level weld in pale pink), and counterregional 'fault welds' (Rowan et al. 1999) in green. Two large basinward dipping roller faults (blue) that are part of a roho system to the east (Fig. 2) cross cut some of the counter-regional fault welds. A perspective view of the depth model from the north (Fig. 8b) shows the complex geometry of the composite salt system in the WDSP study area. The weld feeding into salt system A from the northwest has a stepped geometry that corresponds to that seen in the seismic profiles in Figures 7a and 7b. A series of westward dipping counter-regional faults bound the eastern flank of salt system A (Figs 8a & 8b). Salt system A is linked to salt system B by a deep level weld (Fig. 7a) and a counter-regional fault weld. Salt system B is also linked to salt system C by the steeply dipping weld described above (Fig. 7c) and a counter-regional fault weld. Small remnant pockets of salt exist across the fiat lying secondary (?) weld that feeds into salt system C (the central portion of dome C is outside the 3D data volume). This secondary weld level is linked to the deep weld that extends down to a depth of over 12 krn. The salt geometries have been further simplified in Figure 8c, which shows the geometry of only the salt bodies and welds, viewed looking towards the southwest. The multi-layered geometry of the composite salt system is particularly clear from this view. The true 3D geometry of the WDSP salt system is best viewed in a movie format (see Movie 1 on supplementary CD). The movie starts off viewing the simplified salt system from the east, swings round through the north and west and finally ends up viewing the system from almost overhead.
Interpretation The multi-layered geometry of the WDSP composite salt system is interpreted as a result of the progressive movement of salt, first from the Jurassic Louann source layer up through a series of feeders into salt tongues that linked in places and remain separate
INTEGRATING 3D SEISMIC AND RESTORATIONS
Fig. 7. Dip-oriented, time-migrated seismic profiles extracted from the Western 2D and 3D surveys (see Fig. 5 for locations). Salt bodies are shown in black, with the base salt being a pseudo-depth calculation (see Fig. 5). Welds are in black and highlighted with pairs of dots. Faults are coloured according to the classification of Rowan et al. (1999). Eight maximum flooding surfaces are illustrated and labelled: (a) Profile A crosses parts of all the salt systems present, illustrating the linked nature of the systems. (b) Profile B covers the whole of salt system A, showing the stepped nature of the counter-regional weld and the asymmetric shape of the salt diapir. (c) Profile C crosses salt system B and part of system C. Dome B is interpreted as linking to both system C and possibly system A. The interpretation of the weld dipping steeply to the southeast below the overhanging salt flank is discussed in detail in the text.
169
170
B.D. TRUDGILL & M. G. ROWAN
Fig. 8. Three-dimensional depth models of the WDSP salt systems: (a) Map view of the simplified 3D model. Present day salt is shown in pink, salt welds in red, counter-regional 'fault welds' in green, and two large roller faults from an adjacent roho system in blue. The yellow line marks the location of the restored cross section in Figure 9. (b) Perspective view of the simplified 3D model viewed from the north; (c) View towards the southwest with faults and fault welds removed to highlight the multi-layered and linked geometry of the salt systems. The depth model was constructed in 3DMove (see text for explanation). Tick marks on frame edges in (a) and (b) are spaced every l0 km, but the scale in each frame varies because of the perspective view.
in others, and then into the isolated shallow domes. The salt and stratal geometries of salt system A show a clear correspondence with the 2D evolutionary models of Schuster (1995), illustrated in Figure 3. The classic monocline thickening into the salt body above a stepped counter-regional weld testifies to the vertical subsidence of the overlying strata into an evacuating salt tongue. The more complex relationships described for salt systems B and C imply a somewhat different salt history however. To truly understand the complex interplay between sedimentation and salt evolution across the whole system it is necessary to restore the salt geometries back through time.
Restorations Methodology The 3D depth model was restored in both 2D and 3D to illustrate the three-dimensional emplacement and subsequent evolution of
the WDSP salt system. Each restoration was carried out following the methodology of Rowan (1993). For each stage, the top interval was stripped off and the underlying section decompacted and isostatically adjusted using appropriate porosity functions and densities. Porosity functions (Sclater & Christie 1980) were based on the estimated sand-shale percentages of the stratigraphic unit, derived from well logs. Isostatic adjustments to the decompacted section were calculated using a smoothed Airy isostatic model and standard densities (water, 1030 kg/m3; sedimentary grains, 2650kg/m3; salt, 2200kg/m3; mantle, 3300kg/m3). The suprasalt section was then restored to a sea-floor template determined by the isostatic calculations, and the subsalt section was restored to an adjusted deep baseline. Importantly, no area or volume conservation constraints are placed on the salt; its reconstructed shape at each stage is defined by the supra- and subsalt restorations. The assumptions and limitations of this restoration technique are discussed in more detail in Rowan (1993) and Bland et aL (2000).
INTEGRATING 3D SEISMIC AND RESTORATIONS 2D restoration
The section line located on Figure 8 is used to show the restoration results in two dimensions for salt system A. Figure 9 displays the present day depth geometry and sequential restorations to 0.50, 2.55, 3.65, 5.00, 6.40, 7.50 and 9.90 (?) Ma. Although the restorations were performed moving backwards through time, the restoration results are best illustrated and described moving forwards through time, showing the changing geometries as the salt system evolved (see also Movie 2 on supplementary CD). The restoration for 9.90 Ma is somewhat schematic as the age constraint for this horizon is poor and interpreting the position of the horizon beneath the salt is often difficult. The restoration shows a tongue at the surface, extending for approximately 12kin downdip. A northwestdipping feeder primarily fed this tongue from the Louann source layer. Based on the keel geometry of the base salt surface (Figs 5 & 7a) in salt system A, a second feeder is interpreted to have contributed salt to the tongue. The seafloor restoration at 9.90Ma produces water depths of between 1400 and 1800m along the section line at this time. Other studies in the northern Gulf of Mexico (e.g. Rowan et al. 2001) have compared water depths from restorations with paleoecologic bathymetry data, and shown that the water depths predicted are reasonable, even though eustatic sea-level changes were not incorporated into the restorations. By 7.50Ma the salt tongue had reached its
Fig. 9. Two-dimensional restoration of a cross section across the Dome A system (line orientation shown in Fig. 8), constructed using 2DMove and the methodology of Rowan (1993). Salt is in black, and pairs of dots indicate salt welds. The 9.90 Ma restoration is unconstrained by biostratigraphic data. No vertical exaggeration; see text for detailed discussion.
171
maximum down-dip surface extent of approximately 15 kin. Minibasins established on either side of the salt tongue were receiving thick deposits of sediment relative to the salt tongue, which was a bathymetric high presumably bypassed by the main sedimentary systems at this time. Water depths at 7.50.Ma had shallowed to around 1300 m across the profile. The salt tongue still had a significant surface extent at 6.40 Ma, but sediment loading along the northwest flank of the salt tongue was accommodated by further basinward extrusion of salt. The flanking minibasins were already filled with some 5kin of sediment at this time. Water depths by 6.40 Ma had shallowed significantly to 900-1000 m. This marks the start of a major progradation of the shelf margin across the WDSP area (Piggott & Pulham 1993; Andy Pulham, pers. comm. 1999) that occurred between 6.40 and 5.00Ma. By 5.00Ma water depths had shallowed dramatically to around 500 m and a major minibasin depocentre had developed across the area. The restoration shows that a maximum thickness of 4000 m of sediment accumulated in the northwestern minibasin between 6.40 and 5.00Ma, an average depositional rate of nearly 3 mm per annum. This rapid sedimentation rate and the subsequent differential loading of the salt tongue drove the salt basinward, constricting its surface extent, and coincides with start of the 'welding out' of the salt feeders to the system. A significant decrease in salt area is apparent by the 3.65 Ma restoration. This may be due to loss of salt by dissolution and/or salt moving out of the plane of
172
B.D. TRUDGILL & M. G. ROWAN
the restored section. The water depths continued to decrease gradually through 3.65 Ma and 2.55 Ma. A slight flaring of the near surface salt dome between 2.55 and 0.50 probably represents a decrease in the ratio of sedimentation rate to salt flow rate (Vendeville & Jackson 1991). By the present day, however, the salt body no longer has a surface expression on this 2D section, and has little influence on present-day depositional processes where water depths are less than 200 m across the whole of the section.
3D restoration Although the 2D restoration is instructive in terms of illustrating the salt evolution, it tells us very little about how the complex, composite salt system in WDSP evolved in three dimensions through time. A true 3D restoration allows the complex interaction between sediment loading and salt evacuation to be tracked through time. The three-dimensional evolution of the WDSP salt system is illustrated with a series of oblique views from the northeast of 3D restorations for each restored horizon surface (Fig. 10). The sequence of restorations in Figure 10, and in movie format as Movie 3 on the supplementary CD, shows the evolution of the main salt
systems and associated counter-regional fault welds. At 9.10Ma two distinct salt systems are identified: salt system A to the northwest, and the deep feeder (identified in Figs 8a, 8b and 8c) and salt tongue of salt system C. By 7.5 Ma the surface area of each salt tongue had expanded significantly. A minor salt body appeared at this stage that is unconnected to the two main salt tongues. This was the initial expression of salt system B. A re-examination of the seismic data in this area indicates that this salt tongue may have been sourced from a northerly-dipping feeder. Poor seismic data quality in this area prevented an earlier interpretation of this feature, illustrating the value of integrating restorations with structural interpretation of seismic data. By 6.40Ma salt tongues A and B had amalgamated, and may have been locally connected to salt tongue C, producing a major salt-tongue canopy across the area. The maximum extent of this canopy at around 6.40Ma was relatively short lived. By 5.00 Ma the source feeder for salt tongue A had started to weld out and the surface extent of the composite salt system had been significantly reduced. Again, this is attributed to the shelf progradation across the area producing significant loading of the salt that triggered the movement of salt into a series of linked salt walls. This process continued through 2.55Ma as the salt walls became
Fig. 10. Perspective views of the threedimensional restorations, constructed using 3DMove from 9.10Ma to the present day. Views are from the northeast. Volumes of salt at each stage are shown in pink. Welds that develop during evolution of the salt system are shown in red and late stage counter-regional 'fault welds' in green. Amalgamation of the original salt tongues to form a salt-tongue canopy was followed by evacuation and salt migration into a series of isolated salt domes. The restorations indicate that much of the evacuation of the weld between systems B and C took place in the last 500 000 years.
INTEGRATING 3D SEISMIC AND RESTORATIONS progressively narrower and the surface extent of the salt more restricted. By 1.35 Ma, the salt walls had evacuated completely to form a series of counter-regional fault welds, linking isolated salt domes at the seafloor. These features maintained their seafloor expression until 0.50 Ma when evacuation of salt from the major intermediate salt level in salt system C triggered the demise of the salt system as a significant influence on sedimentary processes across the area.
Salt system/minibasin interaction The 3D restorations of the WDSP salt system only show the evolution of salt bodies through time. Isopach maps provide another important part of the story for the different intervals (Fig. 11), as they reflect the changing accommodation and sediment fill through time. The accommodation is related to the seafloor topography which, in turn, is controlled by a combination of near-surface salt evolution and deep salt withdrawal (Rowan & Weimer 1998). The isopach maps can be draped over the map view restorations to illustrate the interaction of the salt system with the surrounding minibasins (Fig. 11). The maps show a variety of patterns. In the oldest intervals (9.10-7.50Ma) the isopachs are relatively consistent, with some small minibasins showing enhanced thickening adjacent to the growing salt tongues. Between 7.50 and 6.40 Ma significant thickening to the north of the salt tongue canopy indicates enhanced sediment loading in this region. The relative sediment thick south of the salt-tongue canopy may represent a fan lobe deposited as turbidite flows diverged after passing through the constriction to the northeast (e.g. Kneller & McCaffrey 1995). The depositional framework changed between 6.40 and 5.00Ma as shelf progradation across the area (Andy Pulham, pers. comm. 1999) led to counter-regional evacuation of the salt-tongue canopy. This created significant accommodation through salt withdrawal in the northern part of the area and thick minibasins developed during this time. South
Fig. il. Isopach maps of the interpreted stratigraphic intervals, derived from the 3D model. Yellowsand reds are thins, blues and purples are thicks. The isopachs are superimposedon the salt geometry for the end of each interval, and illustrate the relationship between evolving salt geometries and creation of accommodation for sediment infill.
173
of the evacuating salt systems however, the basinward progradation of the shelf margin resulted in this area becoming a sediment bypass zone, with relatively thin sequences developed during this timeframe. This depositional framework persisted until about 1.35Ma. Within the last 0.50Ma, the essentially inactive salt systems have only influenced sedimentation by creating minor depocentres through either differential compaction around the shallow salt domes, or ongoing minor subsidence into the salt. The structural restorations and isopach maps were combined to illustrate the evolution of the WDSP salt system and the surrounding minibasins (Fig. 12). In the following sections, we define six evolutionary stages and illustrate them in map view. We also relate the structural history to gradually shallowing water depths interpreted from the structural restorations.
Evolution
Isolated salt tongues (9. 90?- 7.50 Ma) The configuration at this stage is speculative, as there is no well control for these intervals. Sediment loading to the north of drove deep salt upward and basinward from the Jurassic Louann salt, leading to growth of isolated salt tongues at the seafloor. At this time the shelf margin lay somewhat to the north of the WDSP area (Andy Pulham, pers. comm. 1999).
Salt-tongue canopy stage (7.50-6.40Ma) During this interval, continued feeding of salt into the system led to growth and amalgamation of the salt tongues into a salttongue canopy. The salt reached its maximum aerial extent at the seafloor during this time.
174
B.D. TRUDGILL & M. G. ROWAN
INTEGRATING 3D SEISMIC AND RESTORATIONS
Start of salt tongue evacuation (6.40-5.00 Ma) As the shelf margin prograded basinward from the north across the WDSP the geometry of the salt systems changed rapidly. Sediment loading of the salt tongues from the north led to the start of counter-regional evacuation of the salt tongues (sensu Schuster 1995). The salt supply to the systems became limited as the deep feeders started to 'weld out'. Thick minibasins developed above the evacuating salt tongues.
Salt wall stage (5.00-2.55Ma) Sediment loading within the suprasalt minibasins continued to drive the evacuation of salt into salt walls as the original salttongue canopy was dissected. The isopach maps (Fig, 11) indicate that the thickest minibasin sedimentation gradually shifted from west to east (i.e. from system A to system C), possibly representing an eastward shift of the depositional systems at this time.
Salt dome stage (2.55-0.5Ma) A major westward shift of main depocentres away from the WDSP area lead to a decrease in sedimentation rate, allowing vertical, expanding growth of the salt domes at seafloor. To feed this growth, salt was finally evacuated from the salt walls into a series of counter-regional fault welds linking the salt domes (see also Handschy et al. 1998). Minor subsidence continued in the hangingwalls of these fault weld systems.
Burial stage (0.50 Ma to present day) As the main Mississippi depocentre moved back eastwards to its present day position just north of the WDSP area, the shallow salt domes were buried beneath a thick pile of sediment and salt activity diminished.
Discussion and conclusions The three-dimensional geometry and history of the composite WDSP salt system demonstrates both the utility and shortcomings of published 2D end-member models of allochthonous salt evolution. By integrating 3D seismic interpretation with structural restorations it is possible to obtain a more detailed understanding of the complex interplay between salt evacuation and sediment loading. The ground-breaking work of Schuster (1995) defined the evolution of a stepped counter-regional salt system through the evacuation of an individual salt tongue. In this study we have been able to reconstruct the 3D evolution of a complex, composite stepped counter-regional salt system over the last 10 Ma. During this time, growth of initially isolated salt tongues led to amalgamation of individual salt tongues to form a salt-tongue canopy. But this was not a continuous, constantlevel canopy, as sutures between different tongues developed only locally and at different times. Thus, any given seismic
175
profile may show: (1) a single, subhorizontal salt layer/weld; (2) a multi-level canopy: (3) isolated salt bodies/welds at similar levels; or (4) isolated salt or welds at different levels. Between 6.40 Ma and 5.00 Ma, a major progradation of the shelf through the WDSP dramatically altered the evolution of the salt system as the counter-regional evacuation of the salttongue canopy produced a complex series of salt walls linking isolated salt domes. New minibasins developed in response to this salt evacuation, and these minibasins subsequently became sites of enhanced sediment deposition and salt evacuation into the domes, which were linked by the counter-regional fault welds. Throughout this history, therefore, the primary control on both sediment deposition and fault development was the geometry of the evolving salt system. Within the region surrounding the WDSP study area the distribution of the different salt systems and fault families forms a complex structural pattern (Fig. 3). For example, directly to the east of the WDSP counter-regional system are a series of large, down to the basin roller faults (Rowan et al. 1999) that define a roho salt system involving a significant component of down-dip translation (Schuster 1995: Diegel et al. 1995). Such rapid lateral changes in salt style are probably related to the length of the original salt canopy in the downdip direction (Schuster 1995), and thus the amount of gravity head, i.e., a longer salt canopy results in basinward sliding and thus a roho salt system develops. In contrast, a shorter salt canopy has less of a gravity head and results in vertical collapse with no basinward translation and the development of a stepped counter-regional system. Regional analysis of this area by one of us reveals that this local switching of salt systems over relatively short distances does not appear to be a result of fewer feeders being present beneath the stepped counter-regional systems. We therefore interpret the variations as being primarily due to salt budget within the original Louann salt layer. Originally thin Louann salt rising through spaced feeders produces isolated tongues and small canopies and thus stepped counter-regional systems. Thicker Louann rising through same spacing of feeders gives larger canopies because of more salt volume, thus forming roho salt systems. On a larger scale, the boundaries between the two provinces (stepped counter-regional and roho) are commonly parallel to the underlying rift fabric ( N E - S W and N W - S E ) , as previously noted by Rowan (1997). The results of this study should prove beneficial to those exploring in similar areas or those investigating salt evolution and salt-sediment interaction. The detailed analysis of this composite salt system body and its surrounding minibasins illustrates many of the factors that combine to determine salt system geometry and evolution and its impact on sediment dispersal and deposition. Ultimately, it is hoped that the work presented here illustrates the importance of integrating 3D seismic interpretation in salt basins with sequential structural restorations. This will not only increase our understanding of salt tectonics in the northern Gulf of Mexico, but also improve our interpretations in analogous salt basins worldwide. We thank Western Geophysical for the 3D and 2D seismic data. We also acknowledge Burlington and Paleo-Data, Inc., for well and biostratigraphic data, and Landmark and Midland Valley for their software. The research was carried out at the Energy & Minerals Applied Research
Fig. 12. Schematic diagrams of the evolution of the WDSP salt system constructed from the 3D restorations and isochron maps. Black is salt at or close to the surface, dark grey is a salt weld: in (a), salt it is extruding up and as a series of isolated salt diapirs: in (b). the salt diapirs expand at the surface and amalgamate to form a salt-tongue canopy; in (c), the salt canopy starts to evacuate basinward as the shelf margin progrades across the WDSP area in (d), evacuation produces a series of linked salt walls (e). salt evacuates into the present day salt domes and "fault welds" develop: and in (f), the salt systems are buried beneath a thick sediment layer with only localized movement of salt in the system. The primary depocentres for each stage are shown in pale grey. See text for discussion. Thin lines are protraction area boundaries, and the black bar is 10km long.
176
B. D. TRUDGILL & M. G. ROWAN
Center (EMARC) at the University of Colorado, where we were funded by an industrial consortium consisting of Agip, Amerada-Hess, Amoco, Anadarko, BHP, BP, Burlington, Canadian Oxy, CNG, Conoco, Enterprise, Exxon, Marathon, Maxus, Mobil, Occidental, Phillips, Shell. Texaco, Union Pacific, Unocal, and Vastar. P. Weimer, A. Pulham and D. Knapp are thanked for their contributions within EMARC. We are particularly grateful to T. Dohmen for information regarding the confusing drilling results from the SP89 field that stimulated the interpretation of the complex salt weld geometries in this area.
References BLAND, S. C., GRIFFITHS, P. A. & DZUBER, R. 2000. Techniques and benefits of 3D salt restoration, GeoCanada 2000, Calgary. Alberta, Canada, May 29-June, 2000, Technical Session 06. DIEGEL, F. A., KARLO, J. F., SCHUSTER, O. C., SHOUP, R. C. & TAUVERS, P, R. 1995. Cenozoic structural evolution and tectonostratigraphic framework of the northern Gulf Coast continental margin. In: JACKSON, M. P. A., ROBERTS, D. G. & SNELSON. S. (eds) Salt Tectonics: a Global Perspective, American Association of Petroleum Geologists Memoir, 65, 109-151. FLETCHER, R. C., HUDEC, M. R. & WATSON, I. A. 1995. Salt glacier and composite salt-sediment models for the emplacement and early burial of allochthonous salt sheets, hz: JACKSON, M. P., ROBERTS, D. G. & SNELSON, S. (eds) Salt Tectonics: a Global Perspective, American Association of Petroleum Geologists Memoir, 65, 77-108. GILES, K. A. & LAWTON, T. F. 2002. Halokinetic sequence stratigraphy adjacent to the El Papalote diapir, northeastern Mexico. AAPG Bulletin, 86, 823-840. HANDSCHY, J., VAN DEN BEUKEL, J., GE, H. & DIEGEL. F, 1998. Salt dome geometries, Gulf of Mexico shelf. American Association of Petroleum Geologists Annual Meeting Program and Abstracts, 7, A269, KNELLER, B. & MCCAFFREY, B. 1995. Modelling the effects of salt induced topography on deposition from turbidity currents, h~: TRAVIS, C. J., HARRISON, H., HUDEC, M. R., VENDEVILLE,B. C. & PEEL, F. J. (eds) Salt, Sediment and Hydrocarbons. Gulf Coast Section SEPM Foundation 16th Annual Research Conference, 137-145. McGU1NNESS, D. B. & HOSSACK, J. R. 1993. The development of allochthonous salt sheets as controlled by the rates of extension, sedimentation, and salt supply. In: ARMENTROUT,J. M., BLOCH,R., OLSON, H. C. & PERKINS, B. F. (eds) Rares of Geological
Processes, Gulf Coast Section SEPM 14th Annual Research Foundation Conference, 127-139. PIGGOTT, N. & PULHAM, A. 1993. Sedimentation rate as the control on hydrocarbon sourcing, generation, and migration in the deepwater Gulf of Mexico. In: ARMENTROUT,J. M., BLOCH, R., OLSON, H. C. & PERKINS, B. F. (eds) Rates of Geological Processes, Gulf Coast Section SEPM 14th Annual Research Foundation Conference, 179-191. ROWAN, M. G. 1993. A systematic technique for the sequential restoration of salt structures. Tectonophysics, 228, 331-348. ROWAN, M. G. 1995. Structural styles and evolution of allochthonous salt, central Louisiana outer shelf and upper slope. In: JACKSON, M. P. A., ROBERTS, D. G. & SNELSON, S. (eds) Salt Tectonics: A Global Perspective, American Association of Petroleum Geologists Memoir, 65, 199-228. ROWAN, M. G. 1997. Allochthonous salt tectonics (abs). American Association of Petroleum Geologists Annual Convention Official Program, A 100. ROWAN, M. G. & WEIMER. P. 1998. Salt-sediment interaction, northern Green Canyon and Ewing Bank (offshore Louisiana), northern Gulf of Mexico. AAPG Bulletin, 82, 1055-1082. ROWAN, M. G., WEIMER, P. & FLEMINGS, P. B. 1994. Threedimensional geometry and evolution of a composite, multi-level salt system, western Eugene Island. offshore Louisiana. Transactions of the Gulf Coast Association of Geological Societies, 44, 641 - 648. ROWAN, M. G., JACKSON, M. P. A. & TRUDGILL, B. D. 1999. Saltrelated fault families and fault welds in the northern Gulf of Mexico. AAPG Bulletin, 83, 1454-1484. ROWAN, M. G,, RATLIFF, R. A., BARCELO-DUARTE. J. & TRUDGILL, B. D. 2001. Emplacement and evolution of the Mahogany salt body, central Louisiana outer shelf, northern Gulf of Mexico. AAPG Bulletin, 85, 947-969. SCHUSTER, D. C. 1995. Detormation of allochthonous salt and evolution of related salt/structural systems, eastern Louisiana Gulf Coast. In: JACKSON, M. P. A., ROBERTS, D. G. & SNELSON, S. (eds) Salt Tectonics: a Global Perspective, American Association of Petroleum Geologists Memoir, 65, 177-198. SCLATER, J. G. & CHmSTIE, P, A. F. 1980. Continental stretching: an explanation of the post-mid-Cretaceous subsidence of the central North Sea basin, Journal of Geophysical Research, 85, 3711-3739. VENDEVlLLE, B. C. & JACKSON. M. P. A. 1991, Deposition, extension, and the shape of downbuilding salt diapirs (abs). AAPG Bulletin, 75, 683.
Exploration 3D seismic over the Gjallar Ridge, Mid-Norway" visualization of structures on the Norwegian volcanic margin from Moho to seafloor S. M. CORFIELD
t, W . W H E E L E R
3'2, R . K A R P U Z
2, M . W I L S O N
4 & R. HELLAND
2
tDepartment of Earth Sciences, University of Manchester, Oxford Road, Manchester M13 9PL, UK (e-mail: stephen, corfield@ ntl. world.corn) 2Norsk Hydro Research Centre, PO Box 7190, N-5020 Bergen, Norway 3Centre for Integrated Petroleum Research, University of Bergen, Allegt. 41, N-5007 Bergen, Norway 4School of Earth Sciences, Leeds Universit3', Leeds LS2 9JT, UK
Abstract: We present an analysis of a unique 3D survey that allows us to relate the deep structure of the crystalline crust to the shallow structure of the overlying, potentially hydrocarbon-rich sedimentary basins. The survey is located over the Gjallar Ridge, Mid-Norway, and extends from a Moho-levei reflector at around 15 km depth to polygonal faulting and diapiric structures at or near the seabed. 3D visualization techniques using seismic workstations and the Cave immersive environment have been used to illustrate the geometries of these features. The deep reflector is correlated with the top of a deep, highdensity, high-velocity body that is interpreted to indicate the presence of magmatic underplating and is intimately related to localized uplift of the Gjallar Ridge. Abundant high-amplitude reflectors in the deep Cretaceous sections of the survey are interpreted as sills emplaced during the Palaeocene magmatic event and are therefore interpreted to be coeval with the magmatic underplate. In contrast, the shallow parts of the survey have numerous gas-charged mud diapirs and an extensive network of polygonal faults extending to the seabed. Study of such very deep or very shallow features is not standard industry practice. However, the intention here is to demonstrate that. by utilizing the full volume of 3D seismic data, it is not only of scientific interest but also results in a greater understanding of the tectonic history of a hydrocarbon prospect.
The interpretation of 3D seismic data for prospect generation generally focuses on the top 1 to 3 s TWT of the section. However, the majority of 3D seismic volumes have a record length of at least 6 s TWT. Consequently, large parts of the 3D volume are left uninterpreted. It is the intention of this paper to demonstrate that major insights into the evolution of a considerable thickness of the crust can be gained by using the full volume of the 3D seismic data. A possibly unique aspect of the data used in this study is the presence of a high-density, highvelocity body at depth which has been interpreted as a zone of magmatic underplating. The 3D geometry of this deep crustal body and associated igneous intrusions in the overlying sedimentary section has been examined with the extensive use of 3D visualization techniques, both on the seismic workstation and the CAVE immersive environment in the Norsk Hydro Research Centre in Bergen, Norway. An integrated model is presented that postulates a genetic link between the emplacement of igneous bodies of Palaeocene age in the Cretaceous section, and the subsequent formation of vents, diapirs and polygonal faults in the Tertiary overburden. The 3D data used in this study were acquired by Saga Petroleum (now Norsk Hydro) in 1996 as survey SG9604 with NW-trending inlines spaced at 6.25 m and a record length of 8 s TWT. Most of the 3D survey is located in Norwegian block 6704/12, with an extension to the SW into the SE part of block 6704/11 and to the NE into a very small part of block 6705/10 (Fig. 1). This part of the Ridge was tested in 1999 by well 6704/12-1 which resulted in shows of hydrocarbons but no commercial accumulations.
Deep basin and crustal structure The Gjallar Ridge is located on the western margin of the VOting Basin, Mid-Norway and consists of an arcuate high that parallels the continent-ocean boundary for about 250 km. It trends N - S in its southern part and N E - S W in its central and northern parts. There are three culminations along the Ridge, defined at Middle Cretaceous to Palaeocene stratigraphic levels and all three have a corresponding gravity high (Dor6 et al. 1999; Gernigon et al. 2003).
Multichannel reflection seismic lines have not been very successful at imaging the deeper structure of the thick sedimentary basins and crystalline basement along the Atlantic margin, primarily because sill intrusions severely attenuate near-vertical seismic energy. Hence the best constraint on the deep structure in the outer Voting margin, including the Gjallar Ridge, is derived primarily from seismic refraction and wide-angle reflection studies combined with gravity modelling. The wide-angle surveys include both expanding-spread (ESP) profiles (Eldholm & Mutter 1986) and ocean bottom seismometer (OBS) profiles (Mjelde et al. 2001). Two 2D OBS profiles and several 1D ESP experiments lie within or near the 3D survey discussed here. These constrain the deep velocity and reflectivity structure of the area, allowing otherwise ambiguous reflections in the conventional 2D and 3D surveys to be interpreted. The deep structure in the study area is also imaged by several moderate to long offset (11 km), long-record length (14 to 17s TWT) multichannel reflection seismic lines. The wide-angle seismic and gravity models indicate that the crystalline continental crust is locally extremely attenuated ( 4 - 5 km) in the Gjallar Ridge study area. Where the crust is thinnest, top crystalline basement lies at - 1 0 k m depth, and the base of the continental crust at - 1 4 k i n depth. This 'seismic crystalline crust' is characterized by P-wave velocities in the range 5.8 to 6.9 km/s and densities of about 2.85g/cm 3 derived from the gravity models. Beneath the continental crust lies a thick (6km) layer characterized by greater density (3.0 g/cm 3) and higher velocity (7.1-7.4 km/s). This layer is interpreted as a magmatic underplate emplaced during the Palaeocene magmatic event prior to continental break-up in the Eocene (Dor6 et al. 1999) and will be referred to in this study as the High Velocity Body (HVB). The velocities and densities are intermediate between granodioritic continental crust and mantle and are characteristic of basaltic igneous rocks. Below the underplate lies the true mantle with velocities over 8km/s and densities over 3.1 g/cm 3. More detailed discussion of the geometry and modelling of these layers can be found in Torn6 et al. (2003) and Mjelde et al. (2001 ).
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, 177-185. 0435-4052/04/$15 9 The Geological Society of London 2004.
178
S.M. CORFIELD E T AL.
Fig. 1. Location map of the Voting Basin. The red box outlines the area of the 3D survey. Abbreviations, main map: COB, Continent-Ocean Boundary; FB, Froan Basin; HB, Helgeland Basin; HG, Hel Graben; HHA, Helland-Hansen Arch; HT, Halten Terrace; MA, Modgunn Arch; ND, Naglfar Dome; NH, Nyk High; NR, Nordland Ridge; UH, Utgard High; VB, Voting Basin; VD, Vema Dome. Abbreviations, inset map: BS, Barents Sea; I, Iceland; N, Norway; NGS, Norwegian-Greenland Sea; NNS, Northern North Sea; R, Rockall Plateau; Sv, Svalbard. Abbreviations, cross-section: HVB, High Velocity Body; J, Jurassic and older sediments; P, Pre-Permian sediments beneath the Halten Terrace; SDR, seaward-dipping reflectors; TOC, transitional oceanic crust.
The magmatic underplate defined on the basis of the wideangle and gravity models is characterized by a laccolith-like geometry (Torn6 et al. 2003). In the study area the top of this layer corresponds to a prominent high-amplitude reflection in the 2D and 3D seismic surveys. Within the 3D seismic survey the top of the HVB appears as a strong continuous reflector (Fig. 2) at depths corresponding closely to those predicted by the ESP and OBS profiles. Gernigon et al. (2003) mapped the reflector (which they informally named the T reflector) over a large part of the outer V0ring basin and concluded that there is a good correlation between the free-air gravity anomalies and the T reflector i.e. structural highs defined on seismic data correspond with positive gravity anomalies and structural lows correspond with negative gravity anomalies. Gernigon et al. (2003), consider the HVB to be partly composed of high-pressure granulite or eclogitic material and they interpret a Caledonian age for its formation. This conclusion has major implications for the thermal history of the Ridge and it also minimizes the effects of Palaeocene magmatism. Here, we favour the interpretation that the HVB is (largely) a magmatic underplate, consistent with the seismic and gravity modelling (e.g. Mjelde et al. 2001; Torn6 et al. 2003). We note the difficulty of achieving enough eclogite or granulite in the crustal assemblage to match the observed velocity (7.2km/s), yet light enough to match the modelled density (3.0g/cm3). However, we are not against a model in which the HVB is a magmatic underplate, incorporating lenses of older crustal material such as the high-pressure facies mentioned by Gernigon et al. (2003).
The mechanism of the flow of sub-crustal melt was addressed in a discussion of the structural evolution of the neighbouring Hel graben and flanking Nyk High by Gernigon et al. (2003) who compared the observed structure with an analogue model using sand and silicone produced by Bonini et al. (2001). The experiment modelled the migration of subcrustal melt from the site of generation (the area of most rapid lithospheric thinning) to the areas beneath the flanking structural highs, where the Moho was shallower, causing further uplift of the highs. We regard this as a plausible model for the distribution of magmatic underplating beneath Mid-Norway in general and the Gjallar Ridge in particular. It addresses the question posed by Dor6 et al. (1999) of why the flanking highs should be underplated and uplifted and not the attenuated crust of the rift zone. Buoyant magma, generated beneath the loci of greatest Palaeocene lithospheric thinning, travelled horizontally along the base of the crust to beneath the thinnest crust. These loci of thin crust resulted from the series of tectonic events since the Devonian. Locally, the melts were ultimately focussed towards the pre-existing structural closure at depth beneath the Gjallar Ridge in a magmatic analogue of oil migration.
Uplift history of the Gjailar Ridge The apparently episodic uplift history of the Ridge presents an interesting problem. At first inspection, the presence of the HVB and its interpretation as a magmatic underplate appears to explain
3D SEISMIC VISUALIZATION OF THE GJALLAR RIDGE
Fig. 2. Seismic line from the 3D survey across the Cretaceous/Palaeocene culmination of the Gjallar Ridge above the HVB. Note the westerly increase in dip of the Cretaceous faults, the concentration of sills in the deep Cretaceous section to the SE of the Ridge and the location of the diapir above the crest of the highest fault block. Abbreviations: BUP, Base Upper Piiocene unconformity; O-CT, Opal-CT reflector; LC, Lower Cretaceous; P, Palaeocene; TE, Top Eocene; UC, Upper Cretaceous. the presence of the structural high at Upper Cretaceous and Palaeocene levels (Figs 2, 3 & 4) as a response to the emplacement of the locally thick, buoyant, underplate at depth. However, there is evidence that the Gjallar Ridge has been a structural high from as early as the Lower Cretaceous. Blystad et al. (1995) interpret the Gjallar Ridge as a structural high in the Lower
Fig. 3. View from the west of an intraCretaceous interpretation with a time slice at 7 s TWT. The High Velocity Body (HVB) is displayed as a concentric zone of high amplitudes at (a) that directly underlie the structural crest of this part of the Gjallar Ridge (b). High amplitude reflectors at (c) are interpreted as sills overlying the HVB.
179
Cretaceous, based on a general thinning of Lower Cretaceous strata towards the margins of the Voting basin. They specifically indicate post-Cenomanian onlap onto the Gjallar Ridge (Fig. 1). Similarly, Sanchez-Ferrer et al. (1999) show post late Aptian sediments onlapping onto the Ridge. In contrast, Gernigon et al. (2003) interpret onlap beginning in the Turonian and continuing to the Palaeocene. Clearly, dating of the deeper reflectors is problematical due to the sparse numbers of wells, and their location on structural highs, but there remains a consensus of opinion that the Upper Cretaceous onlaps the Ridge. PostCampanian onlap and thinning is attributable to uplift of the Gjallar Ridge in response to Upper Cretaceous to Palaeocene extensional event(s), while earlier onlap is not contemporaneous with evidence of extension anywhere along the conjugate Norwegian and East Greenland margins. The late Cretaceous extension on the Gjallar Ridge is characterized by easterly-tilted fault blocks, bounded by predominantly NE-trending, westerly dipping listric faults. The bounding faults become sub-horizontal at depth over the domed HVB reflector (Fig. 2), appearing to merge (Ren et al. 1998) into the dome-shaped intra-basement reflectors reported by Lundin & Dor6 (2002) and the top-HVB or T reflector reported here. Gernigon et al. (2003) postulated that middle Cretaceous mobile shales, characterized by a chaotic/transparent seismic facies, act as a detachment for the listric faults. They also identify an uplift phase ending in the midPalaeocene based on well 6704/12-1, which shows the Campanian-Maastrichtian syn-rift wedges to be unconformably overlain by Upper Palaeocene and/or Lower Eocene sediments. Palaeocene rifting along the Gjallar Ridge was contemporaneous with the extrusion of voluminous basalt flows with an age of 5 4 - 5 6 M a (Dor~ et al. 1999). It is the post-breakup (i.e. post-magmatic) uplift of the Gjallar ridge that is consistent with the realization of the buoyant effect of a consolidated sub-crustal magmatic underplate. Latest Eocene and Middle Miocene flexure of the Voting escarpment was recognized by Brekke et al. (1999), but attributed to 'intermittent episodes of compression due to changes in relative plate motions'. Here, we note stratigraphic evidence for a previously unrecognized later uplift of the Gjallar Ridge in the late Miocene to Early Pliocene by the presence of truncated Miocene reflectors beneath the base Upper Pliocene unconformity. The onset of this uplift is marked by a wedge of ?lower pliocene sediments onlapping a tilted reflector in the eastern part of Figure 2. Uplifts in many areas of the Voting basin occurred
! 80
S.M. CORFIELD ET AL.
Fig. 4. View from the south of an interpretation (a) of the top of the zone of high amplitudes that correlates with the top of the HVB. High-amplitude reflectors at (b) are sills intruded into the deep Cretaceous section. The culmination of the Ridge at top Palaeocene levels occurs at (c). at this time, for example the Helland-Hansen, Naglfar and Vema domes. These have in general been attributed to shortening induced by ridge-push compressional stress.
Sills 3D visualization of the zone above the deep reflector marking the top of the HVB on the workstation and in the CAVE immersive environment has revealed the presence of abundant high-amplitude reflectors and are interpreted as sills intruded during the Palaeocene magmatic event. Sills are a characteristic feature on seismic sections along the North Atlantic margin basinwards of the volcanic escarpments and are observed in the Veering, Mere and West of Shetlands basins and the Rockall Trough (Ritchie et al. 1999). The sills are most abundant
downdip of the culmination of the Gjallar Ridge in the faulted, Cretaceous synclinal depocentres to the SE and SW (Fig. 2). On 2D seismic sections they have the form of arcuate, upwardly diverging events 1 0 - 2 0 k m in length, 5 - 1 0 k m in width and with heights of 2 0 0 - 4 0 0 s TWT. When viewed in 3D in the CAVE immersive environment, the reflectors are resolved as sub-horizontal sheet-like bodies, parallel to the reflectors corresponding to bedding, and are observed to pass laterally into synclinal, inverted cones (Fig. 5). Sills also cut upwards through the bedding reflectors in a stepwise geometry (Fig. 5). Very high-amplitude fault plane reflections are interpreted as intrusions along faults (Fig. 6). Stacked saucer-like geometries are also observed, with the upper margin of a lower saucer merging with a saucer at a high level. The upper margins of the saucers and cones commonly display a ragged edge with hornlike protuberances[Q3] (Fig. 7).
Fig. 5. Sills in the Cretaceous section to the SE of the crestal area of the Gjallar Ridge. There is a range of sill geometries: the deepest sills (a) are laterally continuous. The sill at (b) displays a stepped profile as it climbs up-section, and the sill at (c) changes laterally from a sub-horizontal sheet to a deep synclinal geometry.
3D SEISMIC VISUALIZATION OF THE GJALLAR RIDGE
181
Fig. 6. Palaeo-sea-bed mounds or vents above the crest of a tilted fault block. The very high reflectivity of the fault plane at (a) is possibly due to the presence of an igneous intrusion. Overlying the fault tip at (b) are a number of mounds at the top Palaeocene level. Note the contrast between the reflective syn-rift wedges of the Upper Cretaceous at (c) and the low reflectivity of the Lower Cretaceous in the footwall block at (d). Abbreviations: BTU, Base Tertiary Unconformity; TP, Top Palaeocene. Clearly, there is a genetic relationship between the various geometries outlined above. Francis (1982) studied the Midland Valley Sill in Scotland and the Whin Sill in the NE England. Both examples are in coalfield areas and abundant borehole and mining records in addition to outcrop data enabled the production of structure contour maps and isopach maps of the sills. The 1600km 2 and 5000kin ~- extents of the Midland Valley and Whin Sills are larger than the individual sills imaged in the Gjallar dataset, which range in area from 150 to 500 kin". The Whin sill continues unmapped beneath the North Sea and thus its true extent, perhaps far greater but undocumented, may be analogous to the larger sills in the deeper parts of the Voting Basin. Francis noted the following characteristics, which are remarkably similar to our VOting observations:
hence they can be expected to be important factors in petroleum migration and accumulation along the North Atlantic margins. The major advantage of the analysis of 3D seismic data in relation to outcrop studies of intrusive bodies is the ability to view the geometrical relationships in 3D, aided by the very high acoustic impedance contrast between the igneous body and the surrounding sediment. However, a limitation of the seismic reflection method is the inability to satisfactorily image the relatively narrow ( < 5 0 m ) , sub-vertical ( > 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 (<0.25 m s - ~ ) and a continuously varying direction. At 234m depth the current
speed varied with a cycle period of a quarter of a tidal day and the azimuth flipped east-west with the same period. The boundary between surface and bottom currents occurs between approximately 500 and 600m depth (Bulat & Long 2001). The deeper current measurements shown in Figure 3 indicate higher speeds than those at shallow levels and also indicate a semidiurnal variation in speed (a period of half a tidal day). Significantly, while current speed varies greatly, cycling from 0.1 to 0 . S m s -~, the azimuth of current movement varies smoothly, and has little variation from a westerly trajectory. The timing of the initiation of flow of Northern Component Water (NCW), the equivalent of present-day NADW, through the Faroe-Shetland Channel is uncertain. Eldholm (1990) suggested that it is largely a Neogene phenomenon, whilst an Oligocene onset is preferred by Davies et al. (2001), from dating of contourite development. The reason that the Channel itself is and has been such an important conduit for NCW is that deep-water flow is impeded elsewhere between Greenland and continental Europe by the topographic highs of the Greenland-Iceland and Iceland-Faroe ridges. Together these ridges form a hydrographic barrier to south-going deep water (Fig. 1, Wright & Miller 1996; Jones et al. 2002, fig. 1). The Greenland-Faroe ridge system represents the anomalously thick oceanic crust formed at the mid-Atlantic spreading centre above the core of the Iceland mantle plume since continental break-up around 5 4 M a (Smallwood et al. 1999). Wright & Miller (1996) suggested, using isotopic measurements, that deep water formation in the northern North Atlantic has been pulsed since the Oligocene. This pulsing of deep water movement from More to Iceland Basins (Fig. 1) could be caused by changes in atmospheric temperature. Some isotopic studies suggest that deep ocean circulation became more vigorous during warm 'greenhouse' periods while NADW formation became suppressed during periods of severe global cooling (Raymo et al. 1992). A second possibility is that there was an episodic subsidence and uplift of the physical obstacle to deep-water exchange between the More and Iceland Basins. The oceanic crust of the Greenland-Faroe ridge system has subsided with age, as does most oceanic crust, but an additional factor controlling its elevation is the support from the underlying asthenosphere. Variation in the thermal activity of the Iceland mantle plume has been proved by variation in oceanic crustal thickness (Smallwood & White 1998)
DAVIES,R. J., CARTWRIGHT,J. A., STEWART,S. A., LAPPIN.M. & UNDERHILL.J. R. (eds) 2004.3D Seismic Technology: Application to the Exploration of Sedimentao, Basins. Geological Society, London, Memoirs, 29. 187-198. 0435-4052/04/$15 9 The Geological Society of London 2004.
188
J.R. SMALLWOOD bed topography consists of a series of scarps and hollows (Figs 4 & 5). Since the dominant deep-water current direction is to the southwest (Stoker et al. 1998) over the scarps at the northeast steep side of the hollows, the scarps are here termed the 'Judd Falls'. The sea bed
Fig. 1. Regional location map (a) bathymetry, contour interval 1000 m. RR, Reykjanes Ridge; KR, Kolbeinsey Ridge; ICE, Iceland; WVZ & EVZ, Western & Eastern volcanic zones; GIR, Greenland-Iceland Ridge; FIR, Faroe-Iceland Ridge; IB, Iceland Basin; MB, MOre Basin; WTR, Wyville-Thompson Ridge; FI, Faroe Islands; FSC, FaroeShetland Channel. The box east of the Faroes shows the area of Figure 2. (b) Free-air Gravity anomaly (Sandwell & Smith i 997). JM, Jan Mayen microcontinent; AR, Aegir Ridge. The white continuous lines are flowlines showing the position of oceanic crust formed at the Reykjanes Ridge since magnetic chron 24. The change in direction of the flowlines indicates a change in direction of plate movement at 40 Ma. The black line is a flowline showing oceanic crust formed over the head of the Iceland plume, incorporating the effect of ridge jumps, leaving more crust on the eastern side of Iceland than the west. Red lines show extinct ride axis positions on this flowline. The dashed white line shows the segmented oceanic crust, formed before the Iceland plume was in close proximity to the Reykjanes Ridge (c. 38 Ma; Jones 2003).
and igneous rare earth element geochemistry (White et al. 1995). This expression of variation in thermal activity of the mantle plume is shown by the series of 'V-shaped' ridges south of Iceland (Fig. lb) which record the passage of pulses of anomalously hot asthenospheric mantle southward along the Reykjanes Ridge. Changes in the temperature structure of the asthenospheric mantle are also thought to cause elevation changes of the Faroe-Shetland Channel area (e.g. Smallwood & Gill 2002). Such elevation changes to the Greenland-Faroe Ridge system, closer to the Iceland mantle plume, would have caused significant changes in the degree of physical impedance to deep-water exchange (White & Lovell 1997), and hence change the importance of the Faroe-Shetland Channel as a conduit for NCW.
Seismic mapping The Faroe-Shetland Channel narrows and shallows to the south (Figs 1 & 2). The morphology of the Channel is thus like a funnel, which causes the velocity of south-going water to increase southwards (Boldreel & Andersen 1995). Where the Channel is at its shallowest, at 60 ~ 20~N, 4~ (Fig. 2) the sea
The Judd Falls and the hollows, previously described as the 'Judd Deeps' (Stoker et al. 2003) have been mapped on highresolution 2D and 3D seismic reflection data (Fig. 4). The interpreted time horizons were converted to depth horizons using a water velocity of 1480 m s - l (Fig. 5). The longest scarp is 15km in length, with an irregular cuspate geometry on a 3 - 4 km length scale. The scarp has its greatest height near its centre, where the water depth jumps from 970 m on the updip side of the scarp to 1300m in the adjacent Deep (Deep 1 of Stoker et al. 2003). Along much of its length, the scarp is around 200 m in height, although this decreases towards each end. The scarp itself has an angle too steep to provide an acoustic return, even on the 3D migrated data, which suggests an angle greater than 45 ~. The main strike direction of the scarp is northwestsoutheast, although the scarp bends towards the southwest at each end, where it evolves into a more gentle slope, particularly at its southern end (Fig. 5a). The Deep adjacent to this long scarp is, like the other Deeps, asymmetrical in northeast-southwest cross-section, being bounded by the scarp to the northeast and a more gently dipping slope to the southwest. The second largest scarp and hollow (Deep 2 of Stoker et al. 2003) lies to the west of Deep 1. The scarp is semi-circular in plan view, 7 km in diameter, and its eastern half offsets the sea bed by about 180 m from a maximum water depth of 1240 m in the base of the Deep (Fig. 5). The scarp is scalloped into a series of seven small cuspate structures, each approximately 2-2.5 km across. Above the western half of the scarp the sea bed rises up a slope (with a gradient of approximately 10~ to a shallow point of 710 m water depth. This shallow point forms the eastern side of a further small Deep (Deep 3 on Fig. 5), bounded by a single semi-circular scarp 3 km wide. While both the major Deeps are flanked on their northeastern sides by single, continuous rises, the scarp slopes are terraced in some places, for example, the longest scarp near its northwestern end and the semicircular scarp around Deep 2 at its eastern side (Fig. 5). Two more minor scarps and deeps lie to the north and west of the major structures, both with scarp heights of approximately 100 m, and showing scalloped structuration similar to that of the larger scarp faces. Beneath the sea bed
Seismic data shot across the Judd Falls and Deeps show that the present-day seabed structures extend into the sub-surface. A prominent angular unconformity surface forms a previous base of the Judd Deeps (Fig. 6). Similar scarps to those at present-day sea bed can also be mapped in the sub-surface, particularly to the southeast of Deep 1 (Fig. 7; Bulat & Long 2001; Stoker et al. 2003). The major unconformity surface of the sea bed scarps forms a composite unconformity with the Top Paleogene Unconformity (TPU) surface (formerly termed the Latest Oligocene-Earliest Miocene Unconformity or LOEMU; Stoker 1999; STRATAGEM Partners 2002). Figure 5b shows the interpreted TPU surface, mapped on 3D seismic reflection timemigrated data and depth converted using a 'V0-K' technique, datummed from the sea bed using parameters derived from nearby wells (Smallwood 2002). The eastern end of the longest present-day scarp continues into a northeast-southwest scarp that is totally buried at the present-day (Figs 5, 7 & 8). The buried part of the scarp has a height of up to 325 m, and is
INVERSION AND EROSION, F A R O E - S H E T L A N D CHANNEL
189
Fig. 2. Bathymetry of the FaroeShetland Channel area. Line locations of Figures 6, 7, 11 & 12 shown in red. White circle marks the position of well 6004/16-1, red circle marks the position of well 204/16-1 (Current data in Fig. 3). Arrows indicate the movement of North Atlantic Deep Water (NADW) from Stoker et al. (1998).
Current Direction (de
Current Speed (kts) at depth 74m
"/Alrrl . . . . .
"'%
)th 74m
.~,
!
N W
]E 0
.
0
~
1
1
18:00
I
00:00
0?-0ct
I
I
I
Illl
06:00
I
I
I W
I
12:00
I
I
I1~11
18:00
I
I
i
I
I
I
00:00
08-0ct
I
I
I
I
I
06:00
i
i
1
-
N
18:00
17- 0 0
S E
.i 00:00
06:00
IZ:00
18:00
00:00
07-0ct Current Direction (de
N !
;,L_"
L ~
qL
06:00
08-0ct
~th
N
1Z:O0
Z34m
N W
,[~,
J
S
.~1.0
0 5~,
[
--
_
i
~
! ~-
t
-: dl i-i 06:00
. . . . . . . . . .
18:00
18:00 00:00 06:00 IZ:00 18 O0 00:00 06:00 IZ:00 07-0ct 08-0ct I I [ Current~ I Speedi ~[(kts) et de~th Ai6S0m
650m1 5
N
~
00:00
9
. . . . . . .
12:00
~
-
00:00
07-0ct
~ o ' ~- o l " k ~ .,4" o. ool, I ..... I ..... 18:00
00:00
07-0ct
06:00
1Z:00
' k ~ ..,4r I .....
06:00
I .....
12:00
18:00
00:00
08-0ct
N.
l/
I ,,L'7~,, 18:00
Current Speed
06:00
-rK .....
00:00
08-0ct
I ..... 06:00
I,,, 12:00
.~-
00:00
06:00
08-0ct
jr
9
.
t
1Z:00
>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,, ,'~<,, I ,, ,, ,I, ,, ', ~ , . . . . . I ,, , ',~ , , .... I ,, ,,"~,, , , 18:00
E
. . . . . . . . . . . . . . . .
18:00
07-0ct Current Direction (de
~il~l.
--
:00
i
I
I
I
I
00:00
I
07-0ct
I
I
I
t
06:00
l
I
I
I
I
1Z:00
1
I
1
I
1
18:00
I
I
J
I
I
00:00
I
08-0ct
I
I
I
I
06:00
l
I
I
I
[
1Z:00
I
1
N
Current Direction
Fig. 3. Current data measured using an acoustic Doppler current profiler for a 48-h period during the drilling of well 204/16-1 (Fig. 2). Speed and direction measurements from four depths are shown: 74 m, 234 m. 650 m and 890 m below sea level. The peak current measurement was 1.59 knots (2.95 km h - l or 0.82 m s-z) in a WSW direction at 650 m depth. The shallowest measurement of current (74 m) indicates an irregular, relatively low speed and continuously varying direction. At 234 m the current speed varies with a cycle period of a quarter of a tidal day and the azimuth flips east-west with the same period. The two de?epest measurements have a semidiurnal variation in speed (a period of half a tidal day) but show a smooth azimuth trend, with currents varying little] from westerly trajectories.
190
J.R. SMALLWOOD
Fig. 4. View of interpreted water bottom reflector around the 'Judd Falls' sea bed cliffs. The sea bed is structured with a series of cuspate scarps up to 300 m high and 15 km long. The main deepwater current direction is arrowed (NADW) and the shallow point marked X.
approximately 8 km long. The seismic mapping suggests that a further Deep southeast of Deep 2 has been completely buried beneath Miocene and younger sediments (Figs 5 & 6). Where the TPU has cut to its structurally deepest points, below the Judd Deeps, it is subcropped by Ypresian (lower Eocene) strata. Recent wells have found that the lithologies within the Ypresian are clastic sediments, deltaic sands, silts and muds. Seismic velocity analysis suggests that where the Judd Deeps are cut, the bulk of the lithology is generally finegrained.
The Westray complex A second, similar set of asymmetric hollows has been mapped in the TPU surface 50 km northwest of the Judd Falls (Fig. 9),
and is termed here the 'Westray Complex'. This complex consists of a set of elongate hollows now almost completely filled by Neogene-Quaternary sediments (Fig. 10). The largest hollow is 2.5km across (northwest-southeast) and 5 k m in length (northeast-southwest). The hollow has inward-dipping sides, in the form of a flute mark, with steep sides (approximately 50 ~) to every direction except the southwest, where the TPU surface has a dip of approximately 20 ~ Similarly shaped, smaller hollows are developed to both sides of the southwest flank of the largest hollow. All the hollows are almost completely filled with post-TPU sediments, and there is only a subtle outline of the structures on the sea floor. More precise absolute dating of the fill is again difficult, although the seismic reflector geometries within the fill of the hollows (Fig. 10) suggest a three-phase fill, and possibly an erosional phase again correlated with the INU.
Fig. 5. (a) Depth map of the sea bed at present day. Judd Falls scarps in white. Line locations of Figures 6 & 7 are shown in red. White circle marks the position of well 6004/16-1. (b) Depth map of TPU (Top Paleogene Unconformity) surface. Buried scarps shown in pink. Black dashed line shows the subcrop of the uppermost Mid-Eocene seismic reflector onto the TPU surface. The difference between the two maps reflects the parts of the Judd erosional complex that have been in-filled since the first erosional episode. Location shown in Figure 13.
INVERSION AND EROSION, FAROE- SHETLAND CHANNEL
191
Fig. 6. SW-NE seismic line across the Judd Falls Deep 1. Location shown in Figure 2. Subsequent deposition has occurred on the SW of the sea bed cliff. TPU is the Top Paleogene Unconformity reflector (formerly known as the Latest Oligocene-Earliest Miocene Unconformity reflector), INU is the Intra Neogene Unconformity reflector (following Stoker 2002), other seismic horizons named. Isochron and onlap geometries indicate the periods of formation of the Judd inversion anticline under the sea bed cliff. The depression of the Balder surface under the sea bed cliff is purely a time 'push-down' effect. Maroon arrow marks onlap onto small fold just below the Mid-Eocene reflector.
Discussion The seismic data and interpretation described in this study allow discussion of the origin of the observed structures+ have highlighted the presence of several erosional unconformities in the area and have allowed identification of several phases of compressional deformation.
Origin o f the structures I follow Bulat & Long (2001) and Stoker et al. (2003) in attributing the formation of the Judd Falls and Deeps and their buried extensions (together referred to here as the 'Judd Complex') to the action of vigorous deep-water erosion. Other explanations include the possibility that the Judd Falls are the seabed expression of faulting, are slump structures, or that the Deeps are volcanic calderas or fluid escape structures, perhaps subsequently modified by current action. The hypothesis that the Falls represent fault scarps can be rejected on the grounds that there are no offsets of sub-surface reflectors beneath the Falls or their buried extensions.
Fig. 7. NW-SE seismic line across Judd Deep 1. Location shown in Figure 2. A buried scarp, part of the Judd erosional complex, has no expression at present-day sea bed. Seismic horizons named as in Figure 6. Dashed line shows position of Figure 8.
The apparent offset of seismic reflectors beneath the Falls (e.g. the Balder reflector in Fig. 6) is entirely due to velocity 'push-down' on the time section caused by the relatively low seismic velocity of sea water in the Judd Deeps compared to sediments at the equivalent two-way time to the north (e.g. Smallwood 2002). When depth converted properly horizons are smooth and continuous across the position of the Falls. The identification of the Falls as slump features is unlikely, as the aspect ratio of the hollows both in depth to length and in length to width does not appear to be consistent with mass wasting features seen elsewhere, such as the AFEN slide (e.g. Long et al. 2004). Furthermore, the direction of the scarps (facing southwest) is not consistent with the slope of the basin through the Eocene when clastic input was from the south and east. The high quality of the seismic data also allows the hypothesis that the structures are 'pockmarks' of some kind to be eliminated. While there are some small vertical 'gas chimneys' above the 6004/16-1 oil and gas discovery (Fig. 6), there is excellent continuity of the reflectors beneath the Judd Deeps (e.g. Fig. 7), which precludes the disruption of the
192
J.R. SMALLWOOD
Fig. 8, Zoomed-in NW-SE seismic line across Judd Deep I. Position of data shown in Figure 7. Seismic horizons named as in Figure 6. Blue arrows mark truncated lower Eocene reflectors. The infill of the Deeps between the TPU and the INU consists of at least four seismically distinct units. The deepest unit of infiil of the TPU deeps (A) has a highly reflective character and is topped by an unconformity. The second package (B) is seismically chaotic (mass flow?) and is only present at the NW side of the deeps on this line. The third package (C), which is the most volumetrically significant, shows smooth, bidirectional onlapping reflectors and is topped by an erosional unconformity (arrowed). The lower part of the fourth package (D1) is largely transparent and fills the moat left by package C (mass flow?) and is topped by a veneer (D2) which lies disconformably above unit C to the northwest. The INU itself erosionally truncates package C (green arrow).
lithological layers by the passage of a significant volume of vertically migrating fluid or magma. There are a number of small igneous intrusive bodies beneath the area, but not only are these thought to have been intruded prior to formation of the Falls (Smallwood & Maresh 2002), but seismic mapping and potential field modelling suggest that this area was a major Paleocene depocentre and there is no underlying volcanic centre (Smallwood et al. 2001). Since the infill of the main Judd Deeps shows evidence of subsequent deep-water erosion and the underlying TPU is observed as a clear major regional angular unconformity (Fig. 7), an origin as deep-water erosional scours is most plausible and consistent with the hydrographic conditions in the area. As with the Judd Complex, the morphology of the Westray Complex suggests a deep-water erosive origin. The morphology of the complex is identical to that of outcrop-scale flute marks, although the scale of the structures mapped is unusually large. Flute marks are a type of sole marks, commonly observed at the bases of deep-water turbidite deposits. They form as turbulent
flow erodes a soft substrate. Eddy currents moving across one flute mark commonly evolve to form parasitic flutes on its down-current flanks, resulting in an enechelon assembly of erosional structures, similar to the combination of hollows observed in the Westray erosional complex (Fig. 9).
Unconformities In addition to the Judd Deeps themselves, the reflector geometries allow several additional unconformity surfaces to be identified in the area. The uppermost Lower Eocene seismic reflector is an erosional unconformity even where it is not coincident with the TPU surface, Minor erosional truncation of lower Eocene strata to the southeast of the Judd Falls (Fig. 8) appears to have been coincident with similar truncation further north (Fig. 11), which is overlain by Lutetian strata. A second unconformity identified in this study which truncates three distinct Lutetian packages correlates into an upper
Fig. 9. Views of (a) sea bed and (b) TPU surface above Westray inversion structure. Location shown in Figure 13.
INVERSION AND EROSION. FAROE-SHETLAND CHANNEL
193
Fig. 10. Orthogonal seismic lines across buried TPU structure. The main hollow is 2.5 km across in a NW-SE direction, and 5 km in length NE-SW. The NW side of the hollow has a dip angle of approximately50~and the SE slope 20~ Similar. smaller parasitic structures are present on the edges of the main depression to the SE side.
Lutetian reflector to the northwest (reflector I-Lutetian 3 on Fig. 11) and becomes coincident with the uppermost lower Eocene reflector to the southeast (Fig. 11). The ages of the intraLutetian reflectors are not well constrained, but the Lutetian section appears fully preserved on the northwestern side of the basin (Fig. 11), while it is largely cut out by this upper Lutetian unconformity across the basin to the southeast. Hence, as previously inferred by Stoker (1999), only the youngest middle Eocene section is present on the southeast side of the Judd Complex. The major unconformity in the area is the TPU itself, which forms the basal unconformity of the Judd Deeps and their buried predecessors. Successively older strata subcrop the TPU into the core of the underlying inversion structure (Fig. 6). Steps in the profile of the TPU as it cuts the Lutetian strata suggest variations in the competence of the Lutetian stratigraphy (Fig. 6). The 3D mapping shows that the erosion of the scarps of the Judd Complex has largely occurred where the middle Eocene was exhumed, and the uppermost Mid-Eocene seismic reflector now subcrops the TPU surface (Fig. 5b). Erosion is particularly strong where the basal Lutetian sequence has been exhumed. Cross-bedding within this sequence (just above the Judd Falls between yellow and pink reflectors on Fig. 6) suggests that it is partly sand-prone, while less seismically reflective sections are interpreted to be more clay-prone. Further unconformities are identified here within the Neogene-Quaternary sediments which have partially infilled much of the space within the hollows of the TPU surface within the Judd Deeps (Fig. 7). It is difficult to date the sedimentary infill of the Deeps accurately, as these deposits have not been directly sampled and the reflectors are partly isolated from regional seismic correlation. Following Stoker (2002) the most distinct erosional surface within the infill is interpreted to be the early Pliocene (Intra-Neogene) unconformity or INU (Fig. 8).
The infill of the Deeps between the TPU and the INU consists of at least four seismically distinct units (Fig. 8). The deepest unit of infill of the TPU deeps (here termed package A; Fig. 8) has a highly reflective character and is topped by an unconformity. The second package (B) is seismically chaotic and appears to be only present at the NW side of the deeps, possibly associated with slumping from the steep walls of the Deeps. The third package, which is the most volumetrically significant, includes at least two phases of deposition (C1 & C2) which show smooth, bidirectional onlapping reflectors topped by erosional unconformities. The lower part of the fourth package (D1) is largely transparent and fills the moat left by package C. This is topped (D2) by a veneer which lies disconformably above unit C to the northwest. The INU itself erosionally truncates the strata of package C. Overall, these packages of Judd Deeps infill are interpreted to represent sediment drift deposits and slumped units from the steep flanks of the structure, subsequently partly reworked by strong currents.
Compressional phases At the present-day there is a large anticlinal structure (Figs 11 & 12) beneath the Judd Complex, and the structures mapped and seismic reflector geometries show that there have been several episodes of compressional deformation in the area. The first compressional episode was at the end of the Ypresian. During the Ypresian stage, major deltaic systems prograded northward across the area. Topsets and foresets of these delta systems are imaged from 1.65 to 1.8 s TWT at the SW end of Figure 6. The 300 m high foresets indicate that there was a major marine transgression in early Eocene times, thought to follow removal
194
J.R. SMALLWOOD
Fig. 11. NW-SE seismic line to the northeast of the Judd Falls. Location shown in Figure 2. A major intraYpresian unconformity cuts out all but the youngest middle Eocene strata on the SE side of the basin. Seismic horizons named as in Figure 6. Blue arrow marks erosionally truncated lower Eocene reflectors. of dynamic support of the area by the Iceland mantle plume about 54 Ma (Smallwood & Gill 2002). The prograding systems show no evidence of any paleo-high in the area now occupied by the anticlinal structure. The middle Eocene section, however, thins dramatically onto the present-day high in the uppermost Lower Eocene seismic reflector (Fig. 12), and the deformation that folded this horizon can be subdivided into a phase prior to deposition of the Lutetian and later phases towards the end of the Lutetian. The lowest Lutetian section thins southward and onlaps onto a paleo-high in the uppermost lower Eocene unconformity (Figs 6, 7, 11 & 12). Middle Lutetian deposition, however, is virtually isochorous across the area (between Pink and Green reflectors on Figs 6 & 11). Renewed folding provided more relief on the growing structure onto which the younger Lutetian strata onlapped and thinned (Green to Cream reflectors on Figs 6 & 11). A major change in depositional geometries occurred at the start of the uppermost lower Eocene (Cream to Maroon reflectors on Figs 6 & 11), and this sequence thins to pinchout northwest across the basin. Although this sequence onlaps the underlying strata (Figs 6 & 7), since this wedge thickens slightly towards the present-day structural high in places (Fig. 11) it appears that there was no growth of the 'Judd anticline' at this time.
There is onlap onto a small fold in the latest Lutetian on the north of the main Judd anticline (an'owed on Fig. 6) suggesting a further phase of compressional inversion at latest Lutetian time. The overlying sequence, the upper Eocene, shows deposition of different styles on the northwest and southeast sides of the basin (Maroon to Brown reflectors on Figs 6, 7 & 11). Smooth bi-directional (climbing?) onlapping reflector geometries are observed on the northwest side of the basin while more chaotic, high-amplitude reflectors, possibly mass flow deposits, form a wedge down from the southeast (Fig. 11). Overall these packages strongly onlap and thin onto the Judd anticline (Fig. 6). There appears to have been a further period of deformation and growth of the Judd anticline in the Oligocene. The minor folds with a 10km wavelength on the north side of the Judd Complex were reactivated prior to the erosion of the TPU (Fig. 6) and the uppermost Eocene reflector is deformed to dip away from the inversion structure on both flanks of the basin (Brown reflector on Fig. 11). This phase of deformation may correlate with the Oligocene compressional event identified by Boldreel & Andersen (1993) on the W y v i l l e - T h o m p s o n Ridge. Boldreel & Andersen (1993) identified a further inversion phase of ' S W - N E ' oriented compression in the
Fig. 12. SW-NE seismic line across major east-west inversion anticline in the south of the Faroe-Shetland Channel area. Location shown in Figure 2. Seismic horizons named as in Figure 6. The INU can only be mapped confidently in the centre of the line which is covered by reprocessed 3D.
INVERSION AND EROSION, FAROE-SHETLAND CHANNEL
195
Fig. 13. (a) Map of Top Balder seismic reflector (near the lowermost Eocene). The main trough running NW-SE reflects the Cenozoic post-rift thermal subsidence of the basin. (b) Isochore between seabed and Top Balder surface. Anomalous thins in this interval in the centre of the basin reflect the development of inverted structures between the Eocene and the present day. Locations of Judd (J) and Westray (W) complexes boxed.
Faroe-Shetland Channel in the Middle or Late Miocene. The TPU is itself folded gently (Fig. 6), but it is difficult to establish whether this reflects tightening of pre-existing folds in the Miocene or original topography on the TPU.
The link between inversion structures and erosion
Both erosional complexes identified in this study are positioned over Cenozoic inversion structures. The Westray Complex lies above a discrete structural culmination at the north end of the Westray Ridge, associated with the North Westray volcanic centre (Naylor et al. 1999). Volcanic activity in the basin occurred in several phases, including extrusion of lava flows during the Paleocene (Naylor et al. 1999) and sill intrusion during the early Eocene (Smallwood & Maresh 2002). The North Westray structure appears to have been reactivated, during the late Paleogene, after cessation of igneous activity (Fig. 13). The larger Judd Complex lies above the east-west oriented compressionally inverted Judd anticline. This inversion structure is positioned where Paleogene basin fill has been buckled up against the southerly buttress of the 'Judd High' during north-south oriented compression (Smallwood 2002, fig. 1). The Judd and Westray inversion structures are shown by the shape of the Top Balder seismic reflector, which is a regionally extensive and clear near lowermost Eocene seismic marker (Fig. 6; Smallwood & Gill 2002). While the overall trend of the Top Balder surface reflects the post-rift thermal sag across the basin (Fig. 13), the east-west inversion anticline in the south of the basin and the inversion over the north Westray Ridge are apparent in both Top Balder structure map and relative 'thins' in the isochore between Top Balder and seabed reflectors (Fig. 13). As mapped, the TPU forms the base to both the Judd Falls hollows and the Westray hollows, suggesting that the erosion occurred during the Neogene. It is impossible to rule out an earlier phase of erosion, for which direct evidence has been removed by subsequent formation of the main structures at TPU times. Stoker (2002) interprets that the sea bed forms a composite erosion surface with the INU in an area of the Judd Deeps 'that has been eroding into the underlying Paleogene strata throughout the Neogene interval'. However, the data are consistent with the alternative interpretation that the formation of the Deeps and their buried extension occurred mainly in a single event at TPU time. Subsequently a significant proportion of the Deeps have been infilled (Fig. 8), although more erosion has occurred in other areas, evidenced by erosional truncation of the INU by the seabed (Stoker 1999, fig. 8).
The timing of the main phase of erosion of the Judd Deeps and their buried extensions is consistent with published dates for the opening of the 'Southern Gateway' for NCW movement through the Faroe-Shetland Channel (Davies et al. 2001: Fig. 14). As mentioned above, there is no direct evidence in the data presented here for Early Oligocene southward deep water movement (Davies et al. 2001), as the main phase of erosion of the Judd and Westray Complexes appears to be coincident with the more regionally significant TPU unconformity (Stoker et al. 2002), Boldreel & Andersen (1993) proposed a link between tectonic shifts in the north Atlantic and the compressional phases that they identified in the region, specifically the change in seafloor spreading from the east to the west of Jan Mayen around magnetic chron 12 (31 Ma; Cande & Kent 1995) with the compression responsible for movement on the WyvilleThompson Ridge. Jones et al. (2002) suggest that the observed plate tectonic shifts may be caused by time-dependent flow within the mantle plume, which has had a 5 - 6 Ma periodicity since at least 35 Ma (Fig. 14). Wright & Miller (1996) linked this time-dependent flow within the Iceland mantle plume to NCW exchange variations, as the spill over the Greenland-IcelandFaroe ridge system is very sensitive to elevation changes, which can be caused by changes in mantle temperature. Drawing these ideas together gives a possible causal mechanism underlying the formation of the dramatic erosional features mapped in this study. The timing of pulses in the mantle plume flow, ridge jumps, compressive deformation, erosive unconformities and periods of NCW flux through the Faroe-Shetland Channel are summarized in Figure 14. The times of initiation of V-shaped ridges south of Iceland (Jones et al. 2002; Fig. lb) indicate periods when Iceland mantle plume activity was high (green bars on Fig. 14). According to Wright & Miller (1996), this would concentrate the flow of NCW into the Faroe-Shetland Channel, as the crust around Iceland becomes elevated by support from the mantle and NCW flow through other spill-overs is impeded. The uplift above the plume head will also vary compressive stress in the lithospheric plates as gravitational 'head' at the spreading centre is maximized. Distinct changes in stress regime may occur if the plume variation triggers a 'ridge jump' or spreading axis relocation, such as that ongoing from Western to Eastern volcanic zones in Iceland at the present day (Saemundsson 1979; Boldreel & Andersen 1993; Smallwood et al. 1999; Smallwood & White 2002; Fig. 1). The unconformities identified in the fill of the Judd Deeps are tentatively correlated with the Miocene plume activity highs (Fig. 14).
196
J.R. SMALLWOOD
Fig. 14. Relative timing of Cenozoic events in the Faroe-Shetland Channel area. Stars show approximate timing of compressional phases, Mid-Atlantic spreading centre 'Ridge Jumps', erosional unconformities, NCW concentration in the Faroe-Shetland Channel and initiation of V-shaped ridges south of Iceland (high mantle plume activity), with timescale from Harland et al. (1990), magnetic reversal timescale from Cande & Kent (1995) and approximate stratigraphic position of coloured seismic markers in Figures 6, 7. 11 & 12. Question marks indicate where exact timing is unsure. This study has highlighted four periods of compressive deformation in the Judd area--the latest Ypresian (Figs 6, 7, 11 & 12), late Lutetian (Fig. 11), latest Lutetian (Fig. 6) and Pre-TPU (Fig. 6). Several unconformities in the Judd Deeps fill (Fig. 8) are of probable Miocene age (A-C) and are tentatively correlated with other significant events, although timing is not absolutely constrained. Overall there appear to be intervals of activity (pale green rectangles group the events), when compression in the basin and deep-water erosion correlates with plume activity, with a periodicity of 6-8 Ma, back to 35 Ma when the Iceland plume approached the Mid-Atlantic Ridge (Jones 2003). Two earlier phases of tectono-hydrographic activity also correlate with regional events. Additional relevant references shown in each column. The V-shaped ridges (Fig. 14) have only been identified back to 35 Ma, after the Iceland plume head is thought to have closely approached the Mid-Atlantic Ridge (Jones 2003). Prior to this time, the compressional pulses and unconformities identified in this study seem to correlate in time with a major plate movement direction change at 4 0 M a (Fig. lb; Smallwood & White 2002) and a westward ridge jump at 49 Ma which would also have altered the stress and hydrographic regimes (grey bars on Fig. 14). The later Lutetian unconformity (Fig. 11) also coincides with a change in thermal regime beneath the Mid-Atlantic Ridge which is marked by the change from a segmented to a continuous mid-ocean ridge axis south of Iceland (White et al. 1995; Jones 2003; Fig. lb). Time-dependent convection variations in the Iceland plume may thus be underlying both the compressional stresses which form inversion structures and also the concentration of bottom current water movement through the Faroe-Shetland Channel. At the present-day, the cross-sectional area of the Channel beneath the 500 m isobath decreases by a factor of ten from 63~ to 60 ~ 20~N (Fig. 2), continuity therefore requiring a ten-fold increase in water velocity from north to south. The Channel has probably had a similar geometry since late Eocene to Oligocene time. Where structural highs such as the Judd anticline were forming at the sea bed in the centre of the Channel, any deepwater flow through the palaeo-Faroe-Shetland Channel would have been further constricted and accelerated, initiating
formation of large-scale erosive eddy currents. Eddies or vortices tend to be set up in the lee side of obstructions to flow (e.g. Almeida e t al. 1992), and these appear to have had sufficient energy in this case to erode hollows proportional to the scale of the bottom-current water mass and the scale of the structural flow restriction.
Conclusions The Faroe-Shetland Channel is an important conduit for deepwater moving from the Norwegian-Greenland Sea into the North Atlantic. Present-day measurements of current velocity, made during the drilling of well 204/16-1, show that deep-water current velocity can peak at over 0.8 m s-~ (1.5 knots). Similar high-energy bottom currents, Northern Component Water, funnelled in and accelerated as the Channel narrows in the southern part of the basin, sculpted a major series of cuspate erosional scarps and hollows (Stoker et al. 2003). The cuspate erosional scarps are 3 - 1 5 krn in length, perpendicular to the predominant deep-water current direction, with slopes exceeding 45 ~ in places. The hollows, on the down-current side of the scarps, are over 200 m deep in places, and along with the scarps, continue into the sub-surface, as the hollows are partially filled in by ?Miocene and younger sediments. A second set of hollows, also interpreted as deep-water erosion scours, has been mapped
INVERSION AND EROSION, FAROE-SHETLAND CHANNEL over the north end of the Westray Ridge. These structures, the largest of which is 2.5 km wide and 5 km long, have the form of a gigantic series of flute marks, in the TPU surface. These hollows are almost totally filled in at the present-day. Both sets of hollows are interpreted to have a deep-water erosional origin. The basal erosion surface of the Judd Complex is correlated into the Top Paleogene Unconformity (TPU), which is a widespread regional u n c o n f o r m i t y surface. Several other erosional unconformities have been identified in addition to the previously documented Base Balder, Intra-Lutetian, Top Paleogene and Intra Neogene unconformities in the area, namely a T o p L o w e r E o c e n e u n c o n f o r m i t y and several m i n o r unconformities within the ?Miocene infill of the Judd Deeps. Both the Judd and the Westray erosional complexes are positioned directly above Tertiary inversion structures, and this study has identified several periods of compressional deformation in the Judd area, latest Ypresian (early Eocene), late Lutetian (middle Eocene) and latest Lutetian (middle Eocene) in addition to the previously documented Paleocene, Oligocene and Miocene phases. Compressional phases have previously been linked to plate tectonic shifts in the Atlantic, which in turn are related to mantle circulation variation (Boldreel 8,: Andersen 1993; Smallwood & White 2002; Jones et al. 2002). The interaction between the Iceland mantle plume and the lithosphere has also controlled the exchange of deep water, as elevation of the G r e e n l a n d - S c o t l a n d ridge system has varied (Wright & Miller 1996; White & Lovell 1997; Jones et al. 2002). When regional thermal support from the mantle plume was most active then the deep-water flow would have been concentrated in the deepest remaining conduit, the F a r o e - S h e t l a n d Channel, at the same time that compressional deformation was causing flow restriction within the Channel. Constriction of flow through the Channel would cause greater acceleration of bottom waters and be conducive to the development of turbulent eddy currents (Almeida et al. 1992) causing the major erosional scours that are observed.
The seismic data are shown by kind permission of Veritas DGC, PGS and BP. I thank M. Stoker, D. Long, J. Bulat, B. Austin, S. Jones, D. Prescott, J. Clark and Amerada Hess' Atlantic Margin team for their input. The manuscript was improved following helpful reviews by P. Knutz and an anonymous referee. The opinions and interpretations expressed herein are not necessarily those of Amerada Hess Ltd.
References ALMEIDA, G. P., DURAO, D. F. G. & HEITOR, M. V. 1992. Wake flows behind two dimensional model hills. Exp. Thermal and Fluid Science, 7, 87. ANDERSEN, M. S. & BOLDREEL,L. O. 1995. Effect of Eocene-Miocene Compression structures on bottom-water currents in the FaroeRockall area. In: SCRUTTON,R. A., STOKER, M. S., SH1MMIELD, G. B. & TUDHOPE, A. W. (eds) The Tectonics, Sedimentation and Palaeoceanography of the North Atlantic Region. Geological Society, London, Special Publications, 90, 141-143. BOLDREEL, L. O. & ANDERSEN,M. S. 1993. Late Paleocene to Miocene compression in the Faroe-Rockall area. In: PARKER, J. R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 1025-1034. BOLDREEL, L. O. & ANDERSEN, M. S. 1995. The relationship between the distribution of Tertiary sediments, tectonic processes and deepwater circulation around the Faroe Islands. In: SCRUTTON,R. A., STOKER, M. S., SHIMMIELD, G. B. & TUDHOPE, A. W. (eds) The Tectonics, Sedimentation and Palaeoceanography of the North Atlantic Region. Geological Society, London, Special Publications, 90, 145-158.
197
BULAT, J. & LONG, D. 2001. Images of the seabed in the FaroeShetland Channel from commercial 3D seismic data. Marine Geophysical Researches, 22, 345-367. CANDE, S. C. • KENT, D. V. 1995. Revised calibration of the geomagnetic polarity timescale for the Late Cretaceous and Cenozoic. Journal of Geophysical Research, 100, 6093-6095. CLIFF, P. D., TURNER, J. t~ OCEAN DRILLING PROGRAM LEG 152 SCIENTIFIC PARTY 1995. Dynamic support by the Icelandic plume and vertical tectonics of the northeast Atlantic continental margins. Journal of Geophysical Research, 100, 24473 -24486. DAVIES, R., CARTWRIGHT,J., PIKE, J. 8z LINE, C. 2001. Early Oligocene initiation of North Atlantic deep water formation. Nature, 410, 917-920. DEAN, K., MCLACHLAN. K. & CHAMBERS, A. 1999. Rifting and development of the Faroe-Shetland Basin. In: FLEET, A. J. & BOLDY, S. A. R. (eds) Petroleum Geology of Northwest Europe." Proceedings of the 5th Conference. Geological Society, London, 533-544. ELDHOLM, O. 1990. Paleogene North Atlantic Magmatic-Tectonic events: Environmental implications. Memoire della Societa Geological Italiana, 44, 13-28. HARLAND, W. B., ARMSTRONG,R. L., Cox, A. V., CRAIG, L. E., SMITH, A. G. & SMITH, D. G. 1990. A Geologic Time Scale. Cambridge University Press. JONES, S., WHITE, N. & MACLENNAN,J. 2002. V-shaped ridges around Iceland: Implications for spatial and temporal patterns of mantle convection. Geochemistr3', Geophysics, Geosystems, 3, 1059 doi: 10.1029/2002GC000361. JONES, S. 2003. Test of a ridge-plume interaction model using oceanic crustal structure around Iceland. Earth and Planetar?' Science Letters, 208, 205-218. 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., LAPPIN, M. & UNDERHILL, J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 53-61. MASSON. D. G. 2001. Sedimentary processes shaping the eastern slope of the Faroe-Shetland Channel. Continental Shelf-Research, 21, 825-857. NAYLOR, P. H., BELL, B. R., JOLLEY,D. W., DURNALL,P. & FREDSTED, R. 1999. Palaeogene magmatism in the Faroe-Shetland Basin: influences on uplift history and sedimentation. In: FLEET, A. J. & BOLDY, S. A. R. (eds) Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference. Geological Society, London, 545-558. RAYMO, M. E., HODELL,D. & JANSEN,E. 1992. Response of deep ocean circulation to the initiation of northern hemisphere glaciation (3-2 M.Y.). Paleoceanography, 7, 645-672. SAEMUNDSSON, K. 1979. Outline of the geology of Iceland. Jokull, 29, 7-28. SANDWELL, D. T. & SMITH, W. H. F. 1997. Marine gravity anomaly from Geosat and ERS 1 satellite altimetry. Journal of Geophysical Research, 102, 10039-10054. SMALLWOOD, J. R. 2002. Use of Vo-K depth conversion from shelf to deep-water: how deep is that brightspot? First Break, 20, 99-107. SMALLWOOD, J. R. & WHITE, R. S. 1998. Crustal accretion at the Reykjanes Ridge, 61-62~ Journal of Geophysical Research, 103, 5185-5201. SMALLWOOD, J. R. & MARESH, J. 2002. The properties, morphology and distribution of igneous sills: Modelling, borehole data and 3D seismic from the Faroe-Shetland area. In: JOLLEY, D. W. & BELL, B. (eds) The North Atlantic Igneous Province: Stratigraphy, Tectonic, Volcanic and Magmatic Processes. Geological Society, London, Special Publications, 197, 271 - 306. SMALLWOOD, J. R. 8r WHITE, R. S. 2002. Ridge-plume interaction in the North Atlantic. In: JOLLEY, D. W. & BELL, B. (eds) The North Atlantic Igneous Province: Stratigraphy. Tectonic, Volcanic and Magmatic Processes. Geological Society, London, Special Publications, 197, 15-37.
198
J.R. SMALLWOOD
SMALLWOOD, J. R. & GILL, C. E. 2002. The rise and fall of the FaroeShetland Channel: evidence from seismic mapping of the Balder Formation. Journal of the Geological Society of London, 159, 627- 631. SMALLWOOD,J. R., STAPLES,R. K., RICHARDSON,K. R., WHITE,R. S. & THE FIRE WORKING GROUP 1999. Crust formed above the Iceland mantle plume: from continental rift to oceanic spreading center. Journal of Geophysical Research. 104. 22885-22902. SMALLWOOD,J. R., TOWNS, M. J. & WHITE, R. S. 2001. The structure of the Faroe-Shetland Trough from integrated deep seismic and potential field modelling. Journal of the Geological Society of London, 158, 409-412. STOKER, M. S. 1999. Stratigraphic Nomenclature of the UK North West Margin: 3. Mid- to Late Cenozoic Stratigraphy. British Geological Survey, Edinburgh. STOKER, M. S. 2002. Late Neogene development of the UK Atlantic Margin. In: DORri, A, G., CARTWR1GHT, J. A., STOKER, M. S., TURNER, J. P. & WHITE, N. (eds) Exhumation of the North Atlantic
Margin: Timing, Mechanisms and Implications for Petroleum Exploration. Geological Society, London, Special Publications, 196, 313-329. STOKER, M. S., ACKHURST,M. C., HOWE, J. A. & STOW, D, A. V. 1998. Sediment drifts and contourites on the continental margin off NW Britain. Sedimentary Geology, 115, 33-51.
STOKER, M. S., NIELSEN, T., VAN WEERING, T. C. E. & KUIJPERS, A. 2002. Towards an understanding of the Neogene tectonostratigraphic framework of the NE Atlantic margin between Ireland and the Faroe Islands. Marine Geology, 188, 233-248. STOKER, M. S., LONG, D. & BULAT, J. 2003. A record of mid-Cenozoic strong deep-water erosion in the Faroe-Shetland Channel. In: MIENERT, J. & WEAVER, P. (eds) European Continental Margin
Sedimentar 3, Processes: An Atlas of Side-Scan Sonar and Seismic bnages. Springer, New York. STRATAGEM PARTNERS, 2002. The Neogene Stratigraphy of the Glaciated European Margbl from l_z)foten to Porcupine. Svitzer Ltd., Great Yarmouth. TURNER, J. D, & SCRUTTON, R. A. 1993. Subsidence patterns in western margin basins: evidence from the Faroe-Shetland Basin. In: PARKER, J. R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 975-983. WHITE, N. & LOVELL, B. 1997. Measuring the pulse of a plume with the sedimentary record. Nature, 387, 888-891. WHITE, R. S., BOWN, J. W. & SMALLWOOD,J. R. 1995. The temperature of the Iceland plume and origin of outward propagating V-shaped ridges. Journal of the Geological SocieO', London, 152, 1039-1045. WRIGHT, J. D. & MILLER, K. G. 1996. Control of North Atlantic deep water circulation by the Greenland-Scotland Ridge. Paleoceanography, 11, 157- ! 70.
3D seismic analysis of the geometry of igneous sills and sill junction relationships DORTHE
MOLLER
HANSEN
1 JOSEPH
A
CARTWRIGHT
l & DAVID
THOMAS
2
13DLab, School of Earth, Ocean and Planetary Sciences, Cardiff University, Main Building, Park Place, Cardiff CFIO 3YE, UK (e-mail: [email protected]) 2Shell E & P Ireland Limited, Corrib House, 52 Lower Leeson Street, Dublin 2, Republic of Ireland
Abstract: We use 3D seismic data in a novel way to describe the three-dimensional geometry of a number of igneous bodies intruded into the upper crust as well as to define and classify sill junction relationships. Igneous intrusions were emplaced into Upper Cretaceous and Palaeocene sediments of the Faroe-Shetland Basin during the Early Palaeogene and in many cases they adopt remarkable saucer- or trough-shaped geometries that are 2-8 km in diameter and have a vertical relief of several hundred metres. Individual intrusions are interlinked and form highly interconnected sill complexes. Three geometrically distinctive classes of sill junctions are defined and illustrated with examples from seismic data. Each class implies a specific evolutionary sequence of events and these are discussed for each of the classes of junction. The class of junction often changes along the line of junction with one class evolving in space to another. This has significant implications for spatial reconstruction of sill complexes based on two-dimensional outcrop and this is illustrated with reference to an example from a 3D seismic dataset.
3D seismic data have the potential to identify new features not previously recognized at outcrop because of scale and resolution limitations and most critically their lack of three-dimensionality. Interpretation of 3D seismic data has already led to many novel discoveries including polygonal fault systems (Cartwright 1994; Cartwright & Dewhurst 1998), large-scale density inversion structures (Davies et al. 1999), giant pockmarks (Cole et al. 2000), dyke-fed submarine volcanoes (Davies et al. 2002), and flow structures in a shallowly emplaced igneous sill (Trude 2004). Recent advances in our understanding of fault growth, sediment remobilization, and deltaic system development, as well as improved ability to perform detailed reservoir characterization are also closely linked with the increasing use of high-resolution 3D seismic datasets, both by industry and the academic community. Igneous sills have been studied at outcrop for more than a century and many field areas provide good insight into the geometry of sill complexes. Numerous models have been published to account for the geometry and emplacement of igneous intrusions at shallow level (duToit 1920; LoewinsonLessing 1936; Tweto 1951; Bradley 1965; Leaman 1975; Francis 1982; Lister & Kerr 1991; Kerr & Lister 1995; Chevallier & Woodford 1999) as well as the nature of associated hydrothermal systems (e.g. Einsele et al. 1980; Boulter 1996). Sills are generally 1-350 m thick and cover areas of many tens or even hundreds of kin 2. They are predominantly concordant with bedding, but often display concave upward cross-sectional geometries with transgressive, discordant limbs inclined at 5 - 2 0 ~ (e.g. duToit 1920; Francis 1982; Chevallier & Woodford 1999). Igneous sills intruded into sedimentary basin fills are strikingly well imaged on seismic data because of a significant difference in the acoustic properties of the host rocks and the intrusions. 3D seismic surveys therefore offer an outstanding opportunity to re-evaluate field-based models with the advantages that the dense sampling of 3D seismic data offer. In this paper we present a detailed 3D seismic analysis of a suite of sills in Tranche 67 of the Faroe-Shetland Basin. We describe for the first time the detailed three-dimensional geometry of sill junctions, propose a geometrical classification of sill junction relationships, and discuss the possible kinematics behind their development. This paper does not aim to discuss the physical processes involved in magma emplacement itself. The first part of the paper provides a brief introduction to the study area, some of the limitations of the reflection seismic
method in studies of igneous intrusions, and a detailed description of some of the typical sill geometries seen on the interpreted 3D seismic data. This is followed by a detailed description and classification of sill junction relationships based on real examples from the T67 survey and a comparison of these with published outcrop analogues. Finally, a number of kinematic models for the development of each of the defined junction classes are presented and evaluated. The conclusion summarizes the key observations presented in the paper and briefly comments on some of their implications.
Regional setting The Faroe-Shetland Basin is a major N E - S W trending basin situated on the NE Atlantic margin between the Faroes and the Shetland Isles (Fig. I). This relatively deep-water basin formed as a result of Mesozoic and Palaeocene extension (e.g. Bott 1975; Ridd 1981; Mudge & Rashid 1987; Dean etal. 1999). The Faroe-Shetland Basin underwent uplift and several phases of both intrusive and extrusive volcanism during the Early Palaeogene in response to the impact of the proto-Icelandic plume at the base of the lithosphere (e.g. Andersen 1988; White 1988; White & McKenzie 1989; Ritchie & Hitchen 1996; Naylor et al. 1999; Ritchie et al. 1999). As a result, the mud- and silt-dominated Upper Cretaceous and Palaeocene sections of the basin are heavily intruded by dolerite sills. The timing of sill emplacement is uncertain, with published radiometric dates ranging between 50 and 80 Ma (see review in Ritchie & Hitchen 1996). However, most intrusive magmatism is believed to have been concentrated at 5 5 - 5 3 M a , approximately synchronous with the onset of spreading in the North Atlantic (Ritchie & Hitchen 1996). Following Palaeocene and early Eocene magmatism, the Faroe-Shetland Basin underwent passive thermal subsidence during the middle and late Eocene. Subsequent regional Neogene uplift shaped the final basin configuration seen at present (Dord et al. 1999).
Seismic interpretation of sills A sill is traditionally defined as a tabular igneous intrusion with concordant surfaces of contact (Allaby & Allaby 1999). However, it is evident from seismic mapping that sill-like igneous intrusions adopt a wide range of geometries from fully
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, 199-208. 0435-4052/04/$15 9 The Geological Society of London 2004.
200
D. MOLLER HANSEN ET AL.
Fig. 1. Location map showing the main igneous complexes and structural elements of the Faroe-Shetland Basin (after Naylor et al. 1999). Note the location of the T67 3D seismic survey area.
concordant sheet-like bodies to highly complex discordant geometries. This section provides a description of the seismic characteristics of igneous intrusions, some of the problems encountered when imaging igneous sills on seismic data, and the terminology used in this paper to describe igneous bodies in the study area. Igneous bodies intruded at shallow crustal levels into sedimentary host rocks generally appear as extremely highamplitude reflections on seismic data due to a significant difference in acoustic properties between igneous material and sedimentary rocks (Fig. 2; Badley 1985). Further characteristics that can assist confident interpretation of high-amplitude reflections as igneous intrusions include reflection continuity, abrupt reflection terminations, and a discordant relationship with stratal reflections. A number of interpretational challenges and artefacts with special implications for interpretation of igneous intrusions on seismic data should be noted. There are only few examples of wells encountering sills within 3D seismic survey areas and well calibration of sills with 3D seismic datasets is therefore not often possible. This complicates the determination of the phase of the seismic wavelet, which again makes interpreting top and base of the sills difficult. Interpreting the top and base of a sill and estimating sill thickness is further complicated if the imaged sill is thinner than half the dominant wavelength of the seismic wavelet because in such a case the seismic responses from the two intrusive contacts cannot be separated. In addition the seismic response from a series of thin sills separated by thin sedimentary sequences may be indistinguishable from the response from a single thick sill (Badley 1965). A seismic wavelet loses frequency when it encounters a high impedance boundary, such as a sill-sediment contact, resulting in a decrease in both vertical and horizontal resolution
(Smallwood & Maresh 2002). As a result, the seismic response from geological features below sills is dimmed and they are poorly imaged, which complicates confident interpretation of deeper sills and thus the geometry of highly interconnected sill complexes. Confident interpretation of sill terminations on seismic data is particularly challenging. If a sill gradually thins towards its tip interference between the reflections from the top and base sill boundaries could give rise to a tapering reflection amplitude and it would be difficult to locate the true termination. Overmigration of the data also complicates detection of sill terminations, as resultant upward curving diffractions at sill tips or from discontinuities in the main body of the sill could easily be interpreted as true continuations of the sill. Careful interpretation of a dense grid of lines can, however, greatly increase the chance of eliminating this artefact, as diffractions generally are lower in amplitude and much less consistent between neighbouring lines, than genuine reflections. Finally, dykes, which play an important role in vertical magma transport, are difficult to recognize on seismic data, as the reflection seismic method is unsuited to image near-vertical features (Badley 1965). In the following some simple informal terms are introduced to assist in distinguishing between different levels of the threedimensional geometrical complexity of igneous bodies recognized on 3D seismic data (Fig. 3). Sills (Fig. 3a) are generally characterized by continuous, very high-amplitude reflections that exhibit both concordant and discordant relationships with stratal reflections. They often have a simple three-dimensional geometry and a relatively smooth periphery. The term compound sill (Fig. 3b) is here used to describe sills with a complex geometry that generally consist of a number of interconnected concordant and discordant segments. They often have complex, irregular peripheries. The terms above are distinguished from sill complex (Fig, 3c), which is used to refer to a highly interconnected network of sills and/or compound sills that developed through a succession of repeated intrusive events over discrete stratigraphic intervals.
The T67 study area Geological context
This paper is based on the T67 3D seismic dataset from the northeastern part of the Faroe-Shetland Basin, immediately southeast of Brendan's Dome (Fig. 1). The Brendan's Dome gravity and magnetic anomaly has been interpreted as an igneous intrusion and is believed to have fed igneous bodies, intruded into Upper Cretaceous mud-dominated strata in the area (Hodges et al. 1999). Magma transport from the magma chamber to the level of intrusion is likely to have been partly facilitated by migration along a series of deep-seated,
Fig. 2. Seismiccross-section through the T67 survey area. Igneous intrusions (highamplitude reflections) were emplaced into Upper Cretaceous and Palaeocene muddominated sedimentary rocks during the Early Palaeogene.
3D GEOMETRY OF IGNEOUS SILL JUNCTIONS
201
Fig. 3. Schematic cross-sections illustrating different levels of the threedimensional geometrical complexity of igneous bodies recognized on 3D seismic data: (a) Sill, (b) Compound sill, and (e) Sill complex. NW dipping tectonic faults formed during Early Cretaceous rifting. The intrusions are easily recognized on the seismic data as high-amplitude events (Fig. 2), and are primarily found in the southeastern part of the survey area.
The T67 3D seismic dataset The T67 3D seismic survey covers an area of approximately 500 km 2 and both inlines and crosslines lines are spaced at 25 m. Detailed interpretation of the seafloor reflection (positive acoustic boundary) suggests that the T67 data display an increase in acoustic impedance as a wide trough surrounded by narrower peaks. That is, the data are close to zero-phase migrated and displayed using negative standard polarity (SEG standards). This means that an increase in impedance, as seen at the top of a sill, is displayed as a wide trough surrounded by narrower peaks, whereas a decrease in acoustic impedance, as at the base of a sill, is marked by a wide peak surrounded by narrower troughs. At the depth where sills are intruded in the T67 survey area the dominant frequency is approximately 35 Hz and half the dominant wavelength, thus approximately 80m (assuming a sill velocity of 5.55km/s: Skogly pers. comm.). This means that the thickness of sills less than 80 m thick cannot be estimated from the seismic data.
Sill geometries On vertical sections, compound sills in the T67 survey area typically consist of a series of interconnected concave-upward shaped segments with relatively flat bases and limbs transgressing at angles of less than 20 ~ (true depth). When viewed in three dimensions, these concave shapes are often found to be remarkable circular saucer-shaped or elongated trough-shaped (Fig. 4). Interpretation of other 3D seismic datasets suggests that this characteristic saucer shape is a fundamental geometry adopted by sills throughout the Atlantic margin of the UK. A contoured map of an almost perfectly circular saucershaped sill is shown in Figure 5a. The planview geometry of the saucer measures 3 by 3.5 km and covers an area of approximately 7.5 km 2. The saucer has a concave relief of approximately 200 in (assuming a host rock velocity of 3000 m/s; this velocity estimate will be use to convert two-way-travel times throughout this paper), and is located at a present-day depth of 4 6 5 0 - 4 9 5 0 m . The base of the saucer (1.5 by 1.5km) is relatively flat and concordant with bedding. Towards its edge the saucer transgresses discordantly to bedding at an angle of approximately 10 ~ (true depth) in all directions. Another example of a circular saucer-shaped sill is shown in Figure 5b. The sill measures 4 by 4 km in planview and covers an area of
Fig. 4. Sills in the T67 survey area. (a) Seismic line showing a series of interlinked concave upward segments of a compound sill. (b) 3D display showing the three-dimensional saucer- and trough-shaped geometry of the interlinked segments.
202
D. MOLLER HANSEN E T A L .
Fig. 5. Saucer-shaped sills. (a) Contoured map and vertical seismic section showing a circular, saucer-shaped sill. The sill measures 3 by 3.5 km, has a vertical relief of approximately 200 m, and covers an area of 7.5 km 2. The base of the sill is relatively fiat and concordant with bedding, while the sill edges transgress gently at an angle of approximately 10~ Contour interval is 20 ms (b) Map and vertical seismic displays of a circular sill. The sill measures 4 by 4 km and covers an area of 13 km 2. It has a vertical relief of more than 300 m and the contour pattern shows that the sill comprises two adjacent low areas, each covering approximately half of the total area of the sill. Contour interval is 25 ms. close to 13 km 2. It has a vertical relief of more than 300 m and the contour pattern shows that the sill comprises two adjacent low areas, each covering approximately half of the total area of the sill. A N N W - S S E trending ridge separates the two low areas. The high-amplitude concave reflections described, and interpreted as sills, above, bear a striking resemblance to 'migration smiles' caused by over-migration of seismic data (Yilmaz 1987). However, sills of this geometry have been observed in numerous seismic surveys from along the NE Atlantic margin, which have all undergone their own individual migration procedures and are thus unlikely to be seismic artefacts. Furthermore, concave sills forming circular outcrop patterns have long been described in the literature (duToit 1920; Francis 1982; Chevallier & Woodford 1999) and have more recently been d o c u m e n t e d by satellite imagery. This strongly suggests
that the concave reflections interpreted as sills on the seismic data are genuine geological features and not a seismic artefact.
Sill junction relationships C o m p o u n d sills and sill complexes are c o m m o n l y developed in the F a r o e - S h e t l a n d Basin. Many examples of sill-to-sill junctions occur within these, simply because of the dense network of individual intrusions. This section provides a detailed description and classification of three-dimensional sill junction relationships observed in the T67 seismic survey and compares these with similar relationships previously described at outcrop. In the study area most sills have a consistent concave saucer or trough shape and the development of c o m p o u n d sills can in most cases be shown to result from the junction of two or more
3D GEOMETRY OF IGNEOUS SILL JUNCTIONS
203
Fig. 6. (a) Contour map showing the junction between sill 1 (blacks contours) and sill 2 (green contours) of compound sill 1 in the T67 survey area. The two sills are linked along a 5 km long NNESSW trending line of junction. (b) Along most of the line of junction, sill 1 abuts against sill 2 (shown in red on the contour display). (c) Towards the east there is an abrupt change in the geometry of the junction and the tips of sill 1 and sill 2 are linked (shown in purple on the contour display). In areas of uncertainty or lack of data the contours are left open and where two sills partly or fully overlie one another the contours of the underlying sill are dashed. The sills are colour coded with contours and seismic pick shown in the same coiour. The line of junction is drawn with a thicker line. All contour annotations are in ms TWT (this also applies to Fig. 7). separate saucer- or trough-shaped bodies. Some simple terms are introduced to assist in the description of the observed sill junctions. The junction between two sills defines an approximately linear trace on contour maps of the interlinked sills. This line is referred to as the line of junction, although it should be noted that strictly speaking it is not a line, but a narrow surface. On a given vertical section two interlinked sills will connect at a point with the two interlinking limbs defining an angle of junction. This point of junction is referred to as the junction point, although it is more correctly a short line than a point. The length of junction will vary from junction to junction, but will in almost all cases be defined by a line of junction and only in exceptional cases be defined by a single point of junction.
T67 case study Contoured maps and representative seismic sections showing examples of the interlinked sills comprising a c o m p o u n d sill in the T67 3D seismic surveys are shown in Figures 6 & 7. Junction geometries similar to those found within this c o m p o u n d sill have been seen between other sills in the T67 survey area and between interlinked sills in other 3D seismic surveys and these junction relationships are therefore regarded as more general geometrical features of sills. Sills are generally well imaged in the T67 dataset, however, as previously noted, where one sill is
overlain by another, the reflection amplitude of the lower is likely to be significantly reduced and therefore very difficult to interpret. Also, many of the interpreted sills extend beyond the boundary of the seismic survey and their overall geometry cannot thus be documented.
Compound sill 1. Figure 6a shows a contour map of two interlinked sills from the western part of the T67 survey area, The larger of the sills (sill l) has an elongated planform (9.5 by 4 km) and transgresses more than 600 m towards the NE. The sill covers an area of more than 30 km 2. The smaller of the sills (sill 2) is saucer-shaped with a diameter of approximately 4.5 km and covers an area of at least 15 km 2. Sill 2 is intruded at a presentday depth of 4800 to 5 4 0 0 m and has a vertical relief of approximately 600 m. The line of junction between the two sills is slightly irregular, but shows an overall N E - S W strike. The full length of the line of junction is approximately 5 km, and along most of this length sill 1 abuts against sill 2 (Fig. 6b). There is an abrupt change in the junction geometry towards the east, where the two sills are almost inseparable. Here, the junction between them is marked by a kink in the reflection (Fig. 6c). Within the same c o m p o u n d sills, sill 2 is linked to sill 3 (Figs 7a & 7b). Sill 3 cross-cuts sill 2 and the tip of sill 2 is as a result detached from the main sill body. The detached tip region of sill 2 is structurally elevated and is rotated slightly backwards
204
D. MOLLER HANSEN
ETAL.
Fig. 7. (a) Contour map showing the junction between sill 2 (green contours) and sill 3 (blue contours) of compound sill 1 in the T67 survey area. (b) Sill 2 is cross-cut by sill 3 and the tip of sill 2 is detached from the main sill body and is slightly structurally elevated and backwards rotated. (Fig. 7b). The contour pattern of the mapped part of sill 3 suggests that it is a saucer-shaped sill with a diameter of approximately 2.5 km and a vertical relief of less than 300 m. The line of junction is at least 1.5 km long and strikes N N E - S S W .
Classification As shown above, careful interpretation of a compound sill on 3D seismic data reveals a suite of sill junction relationships. Based on the geometrical analysis of junction relationships in the T67 survey area, and more widely in the Faroe-Shetland Basin, these junction geometries can be grouped into three distinct classes. The three main junction classes are defined in Figure 8. These definitions are based on seismically resolvable geometrical relationships.
Almost a century ago, duToit (1920) recognized that many sills intruded into Carboniferous-Permian sediments of the Karoo Basin, South Africa, during the Early Jurassic had a concave shape and formed circular outcrop patterns. Due to erosion, however, the rims of the outcropping sills are never fully preserved. Following this pioneering work, Bradley (1965) developed a model for sill development in which adjacent saucer-shaped sills, in a manner similar to that shown here as Class A junctions, connected to form undulating sheets.
9 Class A junction (Fig. 8a): two sills are linked at their tips. 9 Class B junction (Fig. 8b): one sill abuts against another. The junction may be at any angle and occur anywhere on either sill. 9 Class C junction (Fig. 8c): one sill cross-cuts another leaving the two segments of the earlier sill displaced by the continuous, later sill. Class B junctions are the most common and form the longest lines of junction. In no case has a Class A junction been observed along the full length of a line of junction but always appears to evolve into a Class B junction, with the latter extending for the main part of the line of junction. Only two good examples of Class C junctions have been interpreted in the survey area and in both cases is the class of junction developed along the full length of the line of junction.
Outcrop analogues Since the fully three-dimensional geometry of igneous systems cannot be visualized at outcrop, a direct comparison of junctions observed on 3D seismic data and outcrop analogues is not possible. However, junction relationships comparable to the ones defined above have previously been described and inferred from outcrops in classical field areas such as the Karoo of South Africa.
Fig. 8. Classification of sill junction relationships: (a) Class A junctions form where the tips of two sills are linked. (b) Class B junctions form where one sill abuts against another. (c) Class C junctions form where one sill cross cuts another.
3D GEOMETRY OF IGNEOUS SILL JUNCTIONS The Class B junction is the dominant type of junction in the T67 survey and two-dimensional sections through this type of junction have been recognized in many field descriptions of sills. Field sketches by duToit (1920) and Lombard (1952) show highly interconnected networks of sills in which sills abut against one another. Other field studies have shed light on the significance of this class of junction geometry using special techniques to establish the sequence of intrusion. Tweto ( 1951 ), for example, used mineral orientation as an indicator of flow direction and described examples of sill splitting and sill merging in the Pando area in Colorado. Where the splitting or merging involves one sill splitting into two or two sills merging to one the resultant geometries can be regarded as comparable to Class B junctions. Outcrop analogues to Class B junctions have also been described in detail from Jurassic sills near Hobart in Tasmania (Leaman 1975). Intrusion geometries varying from Y- to T-shaped are here developed near feeder sources of dolerite sills intruded into Permian and Triassic sedimentary rocks.
Discussion: the kinematic development of sill junction relationships The detailed seismic analysis presented has revealed the complexity and variability of sill junction relationships. As shown in the previous sections, three distinctly different sill junction relationships have been defined and classified. This section provides a number of models for the possible kinematic development of each of these three junction geometries and their development in space, as well as a number of interpretation pitfalls, related to the non-uniqueness of single vertical sections.
Class A junctions Three possible models for the kinematic evolution of Class A junctions are considered (Fig. 9). The first model explains Class A junctions as a result of intersection of two independently propagating sills that meet as a result of propagation towards each other (Fig. 9a). Whilst coincidental propagation might lead to an intersection, it is extremely unlikely that such a junction would take the form of a single line of junction, with no minor
205
splay sills beyond the line. It therefore seems implausible that class A junctions form as a result of coincidental intersection between two sills. The second model illustrated in Figure 9b shows a Class A junction forming where two interlinked sills propagate away from a feeder source that joins the sills at the line of junction. No feeders have been observed associated with Class A junctions. However, as previously mentioned, vertical dykes are impossible to image directly on reflection seismic data, which means that dykes associated with Class A junctions could be present but are undetectable with the seismic data. Field evidence from heavily intruded, well-exposed areas such as the British Tertiary Volcanic Province, does certainly imply that dykes are a fundamental and much more common component of sill complexes than the seismic data suggest (Harker 1904; Walker 1975; Bell 1984). The third model, which is considered most likely based on the interpreted data, shows how a Class A junction could form where a secondary sill spreads out from the edge of a primary sill (Fig. 9c). The initiation of the secondary sill could for instance be promoted where planes of weakness, such as bedding planes and unconformities, or local lithological barriers, re-direct the propagation path of the primary sill.
Class B junctions Class B junctions are interpreted to form in three distinct ways (Fig. 10). The first way in which a Class B junction is interpreted to form is illustrated in Figure 10a and involves obstruction of the propagation path of a sill by a previously intruded sill. This is very likely to happen in densely intruded basins, such as the Faroe-Shetland Basin, and probably accounts for many of the interpreted Class B junctions in the T67 seismic survey. The second way in which Class B junctions might form is by bi-directional lateral propagation away from a transgressive sill tip (Fig. 10b). The change from transgressive to lateral propagation could occur as a result of exploitation of a plane of weakness or due to the presence of a iithoiogical barrier. This kinematic explanation for Class B junctions implies that a sill intruded at one stratigraphic level can feed a sill intruded at a shallower stratigraphic level. This mode of sill-fed intrusions at successively shallower levels has very important implications
(~J Cross-section Step1
Step2
Fig. 9. Models for the kinematic development of Class A junctions. Class A junctions might form as a result of (a) intersection of two independently propagating sills that intersect as a result of propagation towards each other, (b) bi-directional propagation away from a feeder source (dyke) that joins the sills at the line of junction, or (c) where a secondary sill spreads out from the edge of a primary sill.
t
Planview I
206
D. MOLLER HANSEN
ETAL.
Fig. 10. Models for the kinematic development of Class B junctions. Class B junctions might form as a result of (a) obstruction of the propagation path of a sill by a previously intruded sill, (b) bi-directional lateral propagation away from a transgressive sill tip, or (c) sill splitting and sill branching. for the emplacement of igneous sills and the build-up of igneous sill complexes in general. The third way in which a Class B junction might form is through sill splitting and sill branching (Fig. 10c). Sill splitting or branching could potentially occur due to lithological or mechanical heterogeneity. This might be expected irrespective of the propagation direction i.e. upwards, downwards or laterally.
Class C junctions A single model is presented to account for the kinematic development of a Class C junction (Fig. 11 ). A Class C junction is considered most likely to form due to cross-cutting of an early sill by a later sill. The detached tip region is often structurally elevated to accommodate the added thickness provided by the thickness of the cross-cutting sill. Further, the tip may be rotated slightly backwards as a result of drag induced during the forcible intrusion of the cross-cutting sill. Both these geometrical characteristics are useful guides to interpretation of this type of junction and differentiation from Class A and B junctions.
Spatial changes along lines of junction It is commonly observed that there is a change from Class A to Class B junction along the line of junction between neighbouring sills (e.g. Fig. 6). There are several ways in which the change from a Class A junction to a Class B junction along a line of junction can be explained. Such a change in junction geometry could, for example, occur along sections of the line of junction where bi-directional magma propagation away from the feeder (transgressive sill tip) as illustrated in Figure 10b was locally obstructed and the magma, as a result, only propagated away from the feeder in one direction (Fig. 9c). Alternatively, sill branching and splitting (Fig. 10c) may only occur locally along a line of junction, in areas where preexisting heterogeneities such as faults are present. It is also possible that sill branching might even be induced by local stress perturbations in the region around the line of junction that are associated with sill propagation. It is kinematically feasible that a Class B junction can evolve in space to a Class C junction. Although this mode of junction development has not been observed in the study area, it might occur if local weaknesses along a line of junction allow an otherwise blocked sill (Class B junction; Fig. 10a) to penetrate and breach the blocking sill (Class C junction; Fig. 11).
Discussion: role of 3D seismic in revealing sill geometries and sill junction relationships
Fig. 11. Class C junctions most likely form where an earlier emplaced sill is cross-cut by a later emplaced sill.
The three-dimensional complexity of sill junctions evident for example from the 3D seismic analysis of the T67 survey presented herein has not been recognized or addressed by fieldbased studies of igneous sills. This lack of three-dimensional control of complex geometry is closely linked with the spatial limitations encountered at outcrop and clearly demonstrates the strength of 3D seismic data in studies of shallow-level igneous complexes. Indeed, it could be argued that the mapping of complex sill geometries as interpreted in this study is a type example of the potential of 3D seismic method for extending research in classical areas of field-based geology through better resolution of complex three-dimensional forms.
3D GEOMETRY OF IGNEOUS SILL JUNCTIONS
207
Fig. 12. Four schematic cross-sections through the junction between sill 1 and sill 2 of compound sill 1: (a) Class B junction relationship between sill 1 and sill 2. (b) Class A junction relationship between sill 1 and sill 2. (c) The junction relationship between the two sills just beyond the southern tip of the line of junction, where the two sills are no longer linked. Along this section the tip of sill 2 overlies and overlaps the tip of sill 1. and the two sills are spatially separated. (d) Section D is parallel to the line of junction and illustrates the relationship between sill I and sill 2 immediately west of the line of junction. Here, sill 2 is seen above sill i and the sills appear as two more or less parallel, gently curved sheets. The details that can be resolved in an outcrop-based reconstruction depend directly on the size, location, spacing, and orientation of the accessible outcrops. The sills interpreted in the T67 survey area cover many square kilometres and are tens of metres thick. Intrusions of this size would very rarely be well exposed at outcrop because of scale limitations. Even if large outcrops were available, identifying individual sills between outcrops would be difficult and related with a high degree of uncertainty, particularly if the outcrops are widely spaced. The implications of these limitations for reconstruction of igneous sill complexes can be illustrated with reference to the seismically well defined junction relationship between sill 1 and sill 2 of compound sill 1 of the T67 survey area (Fig. 6). Four schematic cross-sections through the compound sill are shown in Figure 12. If compound sill 1 was exposed in the field, patches of these sections might be the only outcropping evidence of the compound sill and the junction between sill 1 and sill 2. Semi-continuous sections of section A (Fig. 12a) would possibly show the Class B junction between sill 1 and sill 2. However, even if the full section is exposed it would provide no information about the length and geometry of the line of junction nor about the location of this relative to the overall geometry of the two interlinked sills. At an outcrop represented by section B (Fig. 12b), or part thereof, sill 1 and sill 2 could easily be interpreted as one continuous sill, and the significance of the Class A junction might be completely overlooked. Section A and section B both cross-cut the line of junction between sill 1 and sill 2, but at outcrop this would be very difficult to establish and the spatial change in junction geometry along the line of junction is unlikely to be resolved. Section C (Fig. 12c) and section D (Fig. 12d) both show sill 2 completely separated from the underlying sill 1 and offer no evidence of the nearby junction and possible kinematic relationship between the two sills. From this example it is obvious that a reconstruction of the junction relationship between sill 1 and sill 2, based solely on the four cross-sections described above, would be extremely aliased and would undoubtedly fail to reveal the detail evident from the 3D seismic analysis of compound sill 1. It is thus apparent that without 3D seismic imaging of sills, the subtleties of junction geometries and their kinematic significance would be difficult or even impossible to discuss from field data. The
description in this paper of these relationships thus represents an excellent illustration of the potency of 3D seismic data to bring new insights to old and well studied geological problems. The description of sills and their junction relationships presented in this study also highlights the complex interrelationships developed when a significant intrusive episode affects the shallow levels of a sedimentary basin. Although the sources of the magma cannot be precisely identified, the numerous and complex junctions between sills and compound sills point to an elaborate plumbing system at shallow crustal levels in which magma is distributed to near surface intrusions by branching and intersection within and between successively shallower compound sills.
Conclusions Three main classes of junction geometries exhibited by sills have been recognized from 3D seismic data from the northern Faroe-Shetland Basin. Class A junctions form where the tips of two sills are linked, most likely with one of the sills acting as a feeder to the other. Class B junctions are characterized by an abutment relationship between two sills. This type of junction is likely to form in densely intruded sequences where sills obstruct the propagation path of other sills. Class B junctions may also form as a result of bi-directional propagation away from a transgressive sill tip where planes of weakness or barriers to magma transport are encountered and therefore also represent sill-sill feeders. Class C junctions most likely form where an earlier sill is cross-cut by a later sill. This study demonstrates the complex three-dimensional spatial properties of sill junctions, which have not previously been recognized, due to the scale and two-dimensional limitations encountered at outcrop. The junction geometry between two sills often changes aIong the line of junction and a junction relationship seen on a vertical section (outcrop or 2D seismic line) is unlikely to be representative of the entire junction. Another important observation is that two apparently unrelated sills seen on a vertical section may be linked in space and even act as feeders for one another. The observations presented in this paper have considerable implications for the kinematic evolution of sill complexes, timing of intrusion, and hydrocarbon exploration in basins with shallow level intrusions.
208
D. MOLLER HANSEN ETAL.
We thank the Tranche 67 license group (Statoil (UK) Ltd, Enterprise Oil plc, and Mobil North Sea Ltd) for permission to publish images from the T67 3D seismic survey. Sponsorships (DMH) from the Danish Research Agency and Shell Expro UK are greatly appreciated. Schlumberger GeoQuest is acknowledged for providing seismic interpretation software and technical support. Sverre Planke and Brian Bell are thanked for their very constructive reviews. The views expressed in this paper are solely those of the authors and do not necessarily represent those of Enterprise Oil plc.
References ALLABY, A, & ALLABY, M. 1999. Dictional y of Earth Sciences. Oxford University Press. ANDERSEN, M. S. 1988. Late Cretaceous and early Tertiary extension and volcanism around the Faroe Islands. h~: MORTON, A. C. & PARSON, L. M. (eds) Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society, London, Special Publications, 39, 115-122. BADLEY, M. E. 1965. Practical Seismic Interpretation. International Human Resources Development Corporation, Boston. BELL, B. R. 1984, The geochemistry of Lower tertiary basic dykes in the Eastern Red Hill district, Isle of Skye, and their significance for the proposed magmatic evolution of the Skye Centre. Mineralogical Magazine, 48, 365-373. BOTT, M. H. P. 1975. Structure and evolution of the north Scottish shelf. the Faroe Block and the intervening region, hi: WOODLAND, A. W. (ed.) Petroleum and the Continental shelf of Northwest Europe. Wiley, New York, 105-115. BOULTER, C. A. 1996. Extensional tectonics and magmatism as drivers of convection leading to Iberian Pyrite Belt massive sulphide deposits? Journal of the Geological Society. London, 153, 181-184. BRADLEY, J. 1965. Intrusion of major dolerite sills. Transactions of the Royal Socie~' of New Zealand, 3, 27-54. CARTWRIGHT, J. A. 1994. Episodic basin-wide fluid expulsion from geopressured shale sequences in the North Sea Basin. Geology. 22, 447 -450. CARTWRIGHT,J. A. & DEWHURST,D. N. 1998. Layer-bound compaction faults in fine-grained sediments. Geological Society of America Bulletin, 110, 1242-1257. CHEVALL1ER, L. & WOODEORD, A. 1999. Morpho-tectonics and mechanism of emplacement of the dolerite ring and sills of the western Karoo, South Africa, South Aftqcan Journal of Geology, 102, 43-52. 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. DAVIES, R,, CARTWR1GHT,J. & RANA, J. 1999. Giant hummocks in deep-water marine sediments: Evidence for large-scale differential compaction and density inversion during early burial. Geology. 2% 907-910. 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. DEAN, K., MCLACHLAN, K. & CHAMBERS, A. 1999. Rifting and the development of the Faroe-Shetland Basin. ht: FLEET, A. J. & BOLDY, S. A. R. (eds) Petroleum Geology of Northwest Europe: Proceedings of the 5 th Conference. Geological Society, London, 533 -544. DORI~, A. G., LUNDIN,E. R., JENSEN, L. N., B1RKELAND,P. E. & F1CHLER, 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 5 th Conference. Geological Society, London, 41-61. DuTOlT, A. I. 1920. The Karoo dolerites. Transactions Geological Socie~ of South Africa, 33, 1-42. EINSELE, G., GIESKES, J. M. ETAL. 1980. Intrusion of basaltic sills into highly porous sediments, and resulting hydrothermal activity. Nature, 283, 441-445.
FRANCIS. E. H. 1982, Magma and sediment-I: Emplacement mechanism of late Carboniferous tholeiite sills in northern Britain. Journal of the Geological Society, London, 139, 1-20. HARKER, A, 1904. The Tertian" Igneous Rocks of Skve. Hedderwick, Glasgow. HODGES, S., LINE, C. & EVANS, B. (1999). The other millennium dome. Presented at the ! 999 SPE Offshore Europe Conference, Aberdeen, Scotland, 7 - 9 September 1999. Society of Exploration Engineers Inc. KERR, R. C. & LISTER, J. R. 1995. The lateral intrusion of silicic magmas into unconsolidated sediments: the Tennant Creek porphyry revisited. Australian Journal of Earth Sciences, 42, 223- 224. LEAMAN, D. E. 1975. Form, mechanism, and control of dolerite intrusion near Hobart, Tasmania. Journal of the Geological Society of Australia, 22, 175-186. LISTER, J. & KERR, R. 1991. Fluid-mechanical models of crack propagation and their application to magma transport in dykes. Journal of Geophysical Research, 96, 10049-10077. LOEWINSON-LESSING, F. 1936. A contributions to the mechanics of intrusions. XVI International Geological Congress Report, 333-352. LOMBARD, B. V. 1952. Karoo dolerites and lavas. 7)'ansactions Geological Society of South Africa, 55, 175-198. MUDGE. D. C. & RASHID. B. 1987. The geology of the Faroe Basin area. ht: BROOKS, J. & GLENNlE. K. W. (eds) Petroleum Geology of NW Europe. Heyden, London, 751-763. NAYLOR, P. H., BELL, B. R., JOLLEY, D. W., DURNALL,P. & FREDSTED, R. 1999. Palaeogene magmatism in the Faroe-Shetland Basin: influences on uplift history and sedimentation, hi: FLEET, A. J. & BOLDY. S. A. R. (eds) Petroleum Geology of Northwest Europe: Ptvceedings of the 5th Conference. Geological Society, London, 545-558. RIDD. M. F. 1981. Petroleum geology west of the Shetlands. Petroleum Geology of the Continental Shelf of North-West Europe. Institute of Petroleum, London, 414-425. RITCmE, J. D. & HITCHEN, K. 1996. Early Paleogene offshore igneous activity to the northwest of the UK and its relationship to the North Atlantic igneous province. In: KNox, R. B. O'B., CORFIELD, R. M. & DL'NAY. R. E. (eds) Correlation of the Early Paleogene in Northwest Europe. Geological Society, London, Special Publications, 101, 63-78. RITCHIE, J. D., GATLIFF, 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 5 th Conference. Geological Society, London, 573-584. SMALLWOOD,J. R. & MARESH, J. 2002. The properties, morphology and distribution of igneous sills: modelling, borehole data and 3D seismic from the Faroe-Shetland area. In: JOLLEY, D. W. & BELL, B. R. (eds) The North Atlantic Igneous Province: Stratigraphy, Tectonic, Volcanic and Magmatic Processes. Geological Society, London, Special Publications, 197, 271-306. TRUDE, K. J. 2004. Kinematic indicators for shallow level igneous intrusion from 3D seismic data: evidence of flow direction and feeder location, hi: 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, 209-217. TwE'ro, O. 1951. Form and structure of sills near Pando, Colorado. Geological Society of Anwrica Bulletin, 62, 507-532. WALKER, G. P. L. 1975. A new concept of the evolution of the British intrusive centres. Journal of the Geological Socie~.', London, 131, 121-141. WHITE, R. S. 1988. A hot-spot model for early Tertiary volcanism in the NE Atlantic. In: MORTON, A. C. & PARSON, L. M. (eds) Early Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society, London, Special Publications. 39, 3-13. WHITE. R. & MCKENZIE, D. 1989. Magmatism at rift zones: The generation of volcanic continental margins and flood basalts. Journal of Geophysical Research, 94, 7685-7729. YILMAZ, (), 1987. Seismic Data Processing, Investigations in Geophysics vol. 2, Society of Exploration Geophysicists, 468-473.
Kinematic indicators for shallow level igneous intrusions from 3D seismic data: evidence of flow direction and feeder location K. J. T R U D E
3DLab, School o f Earth, Ocean and Planetary Sciences, Cardiff Universi~.,, Main Building, Park Place, Cardiff CFIO 3YE, UK (e-mail: james @ ocean, c f ac. uk)
Abstract: This paper describes some of the results from a 3D seismic-based analysis of the mechanics of igneous sill emplacement in sedimentary basins. Detailed 3D interpretation of igneous intrusions flanking the Corona Ridge in the Faroe-Shetland Basin has led to the discovery of a sill (the Corona Sill) with a previously unrecognized morphology. Two potential feeder sources have been interpreted for the broadly rectangular intrusion, of which the surveyed portion measures approximately 15 by 4 kin. The Corona Sill has a linear NW margin and two lobate protuberances along the SE side. Arcuate ridges that radiate from a central point at the intersection of the lobes cover the imaged surface of the sill. The ridges have wavelengths of 220-350 m and amplitudes in the range 25-50 m. The ridged morphology on the surface of the Corona Sill has not previously been described from any seismic or outcropbased study of igneous sills. The ridges are interpreted to have formed as a direct result of the propagation mechanism, and are thought to have been influenced by the viscosity of the magma, host sediment and the depth of intrusion, which is likely to have been within 400 m of the sediment-water interface. It is suggested that during very shallow intrusion of viscous magma into soft, waterlogged sediments, magma is able to spread, creating a geometry similar to that expected for a high viscosity lava flow. Ridges were formed by compression of a more rigid outer layer of magma in the sill, retarded by the solidifying sill front. Forward movement of the surface layer is likely to be caused by viscous drag from within the sill body. The ridges on the sill top surface are a kinematic indicator for the flow direction of the magma, enabling identification of the feeder zone, which displays a clear link to an underlying sill. It is demonstrated that 3D seismic data has significant untapped potential for the study of magma transport and intrusive processes in the upper crust.
Spanning the considerable literature on the geometry of igneous sills, numerous previous workers have described the variety of ways that igneous sills deviate from a simple planar geometry (Du Toit 1920; Bradley 1964; Francis 1982). Field-based investigations have also revealed the spectrum of deformational styles at the periphery of sills, both within the sill itself and within the surrounding host sedimentary rocks (e.g, Tweto 1951: Pollard et al. 1975; Brooks 1995; Duffield et al. 1986). These reported deformational structures have generally been small scale, ranging from micro to mesoscopic in size. An important group of sill-related deformational features are commonly observed wherever sills intrude into soft sediment saturated with water. These include features such as peperites and fluidization structures, which have been documented by Schminke (1967), Kokelaar (1982), McPhie (1993) and Brooks (1995). These structures can be genetically related to phreatic eruption and explosion breccias due to heating and expansion of pore waters as observed by Grapes et al. (1972) in the Allen Hills region of Antarctica. The importance of high water contents in the host sediments for sill emplacement was emphasized by Einsele et al. (1980), who found large decreases in porosity in soft-sediment-sill contact zones in the Guamas Basin, Gulf of California, thus implying substantial migration of heated pore fluids out of the intruded sediments. Similar styles of interaction between hot magma and fluidsaturated sediments have also been recorded associated with lava flows. In some cases, subaerial lava flows of basic composition emplaced onto poorly consolidated sediments have been found to ingress into the underlying sediment. Schminke (1967) described basalt extruded onto a tuff. Where the accumulated ash fall was thicker and in an aqueous environment, the basalt had burrowed into it, producing spectacular breccias (peperites). The sediment-water interface can, under certain circumstances, be considered a transitional zone that may be
transgressed or invaded, both from below in the form of lava extrusion and from above in the form of lavas burrowing into sediment. Duffield et al. (1986) described a basalt sill emplaced at shallow depth in the Coso Range, California. It had been emplaced as a result of basaltic lava burrowing into soft sediment to a depth of 25 m, extending laterally over 1 kin, initially wedging beds apart and eventually bulldozing them into a 150 m long wedge. Even in areas with substantial exposure of sills and their host rocks such as the Karoo Province of Southern Africa (Du Toit 1920), field-based approaches to analysis of sill geometry and sill-related deformation are limited in their capacity to define both the sill and deformational geometry fully in three dimensions. This paper describes a novel form of sill-related deformation using 3D seismic data. 3D seismic data offer an ideal opportunity for the study of igneous sills emplaced into sedimentary basins and can produce new insight into processes such as magma transport and intrusion. This is because of the significant areal extent of the data (often > 1000kin 2) and its resolving power in three dimensions. Typical 3D surveys are acquired with line spacing of 2 5 m or less, and this in conjunction with other benefits associated with 3D seismic such as removal of side sweep reflections allows for accurate definition of the sill geometry with a spatial resolution of similar magnitude. Several previous publications have documented the general form and seismic expression of igneous sills from this area, including Gibb & Kanaris-Sotiriou (1988), Planke et al. (1999, 2000) and Davies et al. (2002), although these studies have not provided detailed images of sill surface morphology. The highresolution data used in this study have allowed the definition of a previously undocumented upper surface morphology of a large sill in the study area (referred to here as the Corona Sill). The upper surface of this sill is characterized by radiating concentric ridges that are well aligned with the lobate outer periphery.
DAVIES,R. J., CARTWRIGHT,J. A., STEWART,S. A., LAPPIN,M. & UNDERHILL,J. R. (eds) 2004.3D Seismic Technology: Application to the Exploration of Sedimentao, Basins. Geological Society, London, Memoirs, 29, 209-217. 0435-4052/04/$15 9 The Geological Society of London 2004.
210
K.J. TRUDE
On the basis of a detailed stratal and geometrical study it is considered that these novel features are related to the viscosity of the magma, depth of emplacement and the nature of the sediment into which the sill intruded.
Regional setting The three-dimensional seismic survey used in this study is located in a base-of-slope to basin floor position at a water depth of approximately 1500m in the Faroe-Shetland Channel. The Faroe-Shetland Trough is a confined deepwater basin that overlies a Mesozoic to Early Cenozoic rift basin along a NW trending axis between the Faroe Islands and Shetland Islands (Fig. la) (Mitchell et al. 1993). The deepest seismically distinguishable units are the Upper Cretaceous to Paleocene succession (Fig. 1). The Upper Cretaceous is a shale-dominated interval that varies greatly in thickness across the basin (Mudge & Rashid 1987) due to the protracted nature of Cretaceous rifting (Dean et al. 1999). The Paleocene succession consists largely of Upper Paleocene deltaic to coastal margin sediments such as claystones and sandstones with occasional coals and tufts towards the top of the sequence (Mudge & Rashid 1987). This succession is associated with Paleocene rifting and uplift, culminating in a compressional event during the Late Paleocene-Early Eocene, linked to the opening of the Atlantic Ocean (Dean et al. 1999). Renewed subsidence during the Eocene led to formation of a deep-water channel. Mixed contourite and debris-flow deposition was
initiated during the Oligocene and continued into the Pliocene (Davies et al. 2002). The Upper Cretaceous to Paleocene sequence is heavily intruded with doleritic sills, seen as generally discontinuous high amplitude concave reflectors in Figure lb. This set of igneous intrusions is a portion of the Faroe-Shetland Sill complex. Previously suggested dates for intrusion of the sill complex range from Late Cretaceous to Eocene age, e.g. Hitchen & Ritchie (1987). Naylor et al. ( 1 9 9 9 ) suggests a date of 55Ma from scrutiny of the published radiometric and compositional data of Hitchen & Ritchie (1987), though no quantitative assessment of accuracy is given. The geometry and size of sills found within the Faroe-Shetland Sill complex vary widely and are analogous with other great sill sediment complexes such as the Karoo in South Africa (Du Toit 1920; Bradley 1964: Chevallier & Woodford 1999). There are obvious differences also, for instance a terrestrial environment in the Karoo contrasted with shallow marine conditions existing off NW Europe at the time of emplacement.
3D seismic interpretation The three-dimensional seismic volume was acquired using six 4.5 km hydrophone streamers, creating 36-fold data that were migrated prior to being stacked, generating a seismic grid of 12.5 by 12.5 m (see Yilmaz 2001; Davies et al. 2002). The data volume was processed as zero phase. Therefore peaks and troughs coincide with interfaces between different impedances.
Fig. 1. (a) Location map of study area after Davies et al. (2002). (b) Seismic section through study area.
3D SEISMIC INTRUSIVE KINEMATIC INDICATORS Perhaps the most characteristic features of the igneous sills identified on the 3D seismic data are their discordant reflection geometry with respect to the encompassing sedimentary reflections and the extremely high-amplitude reflections from sill-sediment contacts (Figures lb, 2a, 2b). These characteristic features have been described in two-dimensional seismic study and petrophysical analysis by Skogly (1998) pers. comm. and Planke et al. (1999). The intrusive origin of the bodies causing such reflections is illustrated in Figure 2a, where the surface reflection of the feeder sill (Sill 1) below the ridged sill clearly cuts stratal reflections. The Corona Sill can likewise clearly be seen to cut up through stratigraphy towards the right-hand side of its profile (Fig. 2a). This cross-cutting character has been consistently observed for this one sill across many spatially correlated sections and this is taken as being diagnostic of an intrusive relationship and not as an expression of any type of seismic artefact e.g. over migration. Similarly, the ridges themselves can be distinguished from side effects and coherent
Fig. 2. (a) Seismic section through ridged sill. The upper sediment/sill interface is picked across a red horizon as the data is zero phase and red is interpreted as indicating a boundary of increased acoustic impedance. This reflector is identified as a sill because it discordantly cuts across stratigraphy away from the feeder zone. Ridge features of approximately 250-280 m wavelength can be seen towards the right hand half of the sill. The ridged sill is inferred as fed from Sill 1 below, as shown. Horizon C represents a close approximation to the sea floor at the time of intrusion. Note also that the surface of the feeder sill (Sill 1) is comparatively smooth and devoid of ridges compared to the ridged sill. For section line C - D see Figure 4b. (b) Seismic section through the ridged sill showing feeder area from Sill 1 below. Small ridge features are visible towards the left half of the ridged sill. The right half of the ridged sill does not show ridged features, as the section is parallel to axes. For section line E - F see Figure 4b. Seismic date courtesy of PGS Exploration (UK) Ltd. Interpretation software is GeoFrame courtesy of Schlumberger GeoQuest.
211
seismic noise such as reflected refractions and processing inaccuracy by the criterion of correlatability between differently orientated sections (Skogly 1998, pers. comm.). Sills were interpreted on the 3D seismic data by manual picking of closely spaced (12.5 m) vertical seismic profiles using an interpretation workstation and Schlumberger GeoFrame interpretational software. This involved the manual interpretation of every second 12.5 m-spaced line across the surface of the sill in perpendicular directions, thus creating a grid. To fill in the area between manually interpreted sections and reduce interpretation time the horizon was then processed through an auto-correlative based, seismic picking algorithm. Comparison of this technique with areas where every single seismic line was interpreted manually produced no observable difference. This interpretation procedure defined a sill with a ridged surface morphology not previously described in the field or on seismic data. The broadly rectangular intrusion (in plan view) has a linear NW margin and two lobate protrubances along its
212
K.J. TRUDE
SE side (Figures 3 & 4). The imaged portion is approximately 4 by 15 kin, with the elongate axis aligned in a N E - S W direction parallel to the Corona ridge. A N W - S W elongate trend (Fig. 3) is characteristic of a number of the sills in the survey area. The grossly planar Corona Sill with its ridged surface morphology contrasts with the surrounding sills. These exhibit a relatively smooth bowl-like nature (Fig. 3) with no ridged pattern and where the lowest part of a bowl is c o m m o n l y 0.5 s T W T below the rim (approximately 900 m, based on a seismic velocity of 3.6 k m s - l for the sediments contained within the
'bowl'). This difference in surface morphology cannot be attributed to loss of resolution due to increased depth or masking by other high-amplitude reflectors such as lavas, because sills found in the same survey at similar depths that are not shadowed by lavas do not display a ridged morphology. The most striking feature of the top surface of the sill is the presence of a series of ridges that track parallel to the perimeter of the lobate protrubances on the SE perimeter of the sill (Fig. 4) (i.e. in an arcuate nature). These concentric ridges can be traced back to a central focal point at the intersection of the two lobes
Fig. 3. Schlumberger, GeoViz, horizontal distance/vertical time image of interpreted top surface of the ridged sill and surrounding sills. Coiour (red, high; purple, low) does not indicate absolute depth but depth relative from the highest point of each sill. Sills 1-5 are all at a greater depth than the ridged sill. Sills 1-5 are interpreted as being connected to each other and the ridged sill. Red lines highlight the perimeter of sills. Note the different bowl shaped surface morphology and lack of ridges on the surface of sills 1-5 compared to the ridged sill. View is looking down on the sills towards N at approximately a 45 ~ angle.
Fig. 4. (a) Schlumberger GeoViz, composite horizontal distance/vertical time image of the top surface of the ridged sill. Image is lit from the SE. (b) Simplified interpretation of the ridge features with proposed magma flow directions and magma feeder zones. Location of Figure 2 section lines are shown.
3D SEISMIC INTRUSIVE KINEMATIC INDICATORS in the middle of the northern margin of the sill (marked Xt on Fig. 4b). To a lesser extent to the NE, ridges can be traced back to a zone on the northern margin of the sill (marked X2 on Fig. 4b). Within the southern lobe the ridges can be seen to have two distinct orientations (Fig. 4). The first is a western fingershaped region where the ridges tend to strike N W - S E . It is approximately 2 km across and runs from the junction of the two lobes to the SW tip of the sill. The region is bounded to the west by the margin of the sill and to the east by a second region where the ridges tend to strike N E - S W , running parallel to the perimeter of the lobe. The northern lobe of the sill also contains two distinct regions of ridge orientation (Fig. 4). The first runs broadly W - E from the intersection of the two lobes to the eastern margin of the sill. The region is approximately 2 km across and is bounded to the north by the linear termination of the ridge features and to the south by the sill edge, where the ridges run sub-parallel to it. Within this area the strike of the ridges changes by over 90 ~, indicating their arcuate nature. The second region in this lobe is the northern area, which does not contain pervasive ridge-like features, although it is evident that the extreme NE area of this lobe has 'feint' lineations that strike E - W . When viewed in detail (Fig. 2a) the ridge features across the interpreted upper surface of the Corona Sill can be seen to have an almost regular wavelength and amplitude. The mean wavelength is calculated as 237 m and the mean amplitude as 40m, using an assumed average interval velocity for the immediately overlying section of 3.6 km s - t for mudstones at equivalent burial depth from well 214/28-1 (approximately 9 0 k m to the south). No distinct trend in variation of either wavelength or amplitude across the sill body was observed. Horizontal resolution is equal to Fresnel zone size or the area of the reflector surface that returns energy to the hydrophone from the wave front within a half cycle i.e. one-quarter wavelength (Badley 1985). Brown (1999) states that 3D migration has the effect of reducing Fresnel zone size to a circle of diameter one-quarter wavelength for perfect migration. This equals 36 m in this instance given a frequency of 25 Hz+ velocity of 3.6 km s-1. Even if the Fresnel zone were twice the size the ridges on the sill top would still be resolvable given that they have a mean wavelength of 237 m. The limit of vertical depth differentiation between the crests and troughs on the upper surface of the ridged sill is related to the sampling interval, which is 4 ms in this survey (assuming no effect from Fresnel zone size and that a uniform stacking velocity has been applied across the area above the sill). Using a P wave velocity of 3.6 km s+- 1 for mudstones above the sill (at equivalent burial depth from well 214/28-1), this produces a vertical depth resolution of + 7 . 2 m . Therefore it can be concluded that the ridges are significant. The thickness of the sill is difficult to determine (Fig. 2a) as the seismic profile of the sill is characterized by a seismic peak (caused by an increase in acoustic impedance) at the top, followed by a trough, then another peak that is interpreted as 'incoherent noise' caused by problems (ringing) associated with imaging an object of high acoustic impedance. According to Badley (1985), top and base of objects can be resolved if their thickness is equal to or greater than half the wavelength. At a frequency of 25Hz and velocity of 6 . 0 k m s -~, half the wavelength is 120m. This represents the absolute maximum thickness for the ridged sill. Between half and quarter wavelength the two opposite polarity reflections from the top and base begin to overlap, until at quarter wavelength the two wavelets overlap and constructively interfere (Badley 1985). No distinct tuning effect is seen along the seismic profile of the ridged sill (Fig. 2). Therefore the sill thickness is probably between half and quarter wavelength and the base of the sill (for the maximum potential thickness assuming one sill is present) is
2!3
likely to be indicated by the trough associated with a decrease in acoustic impedance. This is in agreement with Skogly (1998, pers. comm.) on comparison of seismic data with a drilled sill. Assuming that this reflection is due to one sill and not a multilayer arrangement of several sills, a maximum sill thickness (between ridges) can be calculated yielding values ranging from 6 4 - 8 8 m with a mean of approximately 75 m using a P-wave velocity of 6 km s - t It is very difficult to determine the geometry of the contact between the sill and the underlying sediments for reasons previously mentioned. However, it is tentatively suggested that the lower surface of the sill may be relatively flat. Close inspection of Figure 2a reveals that below each ridge there is a small area of negative wavelets or troughs within the thicker peak 'noise' area. These lie directly below the crests of the ridges but appear to be in a straight line. They may be present due to the relative thickening of the sill below a ridge, allowing a separate negative wavelet to develop at the base, but only below the ridge. Conversely below the troughs on the sill surface no such features are seen. With the inclusion of the ridges the mean maximum thickness increases to 115 m. Using the imaged surface area of the ridged sill of 54 km 2 and an estimated thickness of 75 m, and making allowances for the extra thickness due to the ridges the volume of the sill is calculated as 5.1 km 3. This volume does not include the portion of the sill that is not imaged i.e. where the thickness is below vertical seismic resolution and the portion that extends beyond the limit of the seismic survey (Fig. 4). The ridged sill displays a close association with a highamplitude reflector below (Sill 1), which is suggested as a feeder sill (Figs 2a & 2b). In Figure 2a, Sill 1 cuts discordantly through stratigraphy and appears to meet the base of the ridged sill. Similarly in Figure 2b, Sill 1 transgresses upwards across stratal reflections to meet with the ridged sill forming another feeder zone. This interpretation is consistent with that of Davies et al. (2002) who suggest underlying sills cut up through stratigraphy to become feeders for submarine volcanoes. In both cases the lateral extent of the contact zone between the feeder and overlying sill is approximately 1-1.5 km. The extent of these feeder zones corresponds closely to the extent of the area that the ridges on the sill surface emanate from.
D e p t h o f intrusion
The depth of intrusion and composition of strata intruded may have important effects on the overall geometry and nature of sill-sediment contacts. The depth and consequent diagenetic state of the strata intruded can be partly constrained by knowing the age of intrusion. Here the nomenclature of Davies et al. (2002) is adopted for the identification of horizons in the study area. Horizons B and C have been confidently tied utilizing 2D and 3D data to wells 214/4-1 and 214/19-1 (Fig. la), (Davies et al. 2002). Horizon B (Fig. 2), immediately below the ridged sill is Selandian in age (58.5 Ma). Horizon C, lies close to the Eocene/Paleocene boundary and is 54.9 Ma (Davies et al. 2002). Some I5 kin to the SW, Horizon C has been locally uplifted by the intrusion of an underlying sill and the resultant topography at the sediment-water interface in-filled by onlapping sediments. The amount of topography created, the close proximity of the sill to the deformed horizon and the special correlation between the uplifted region and the shape of the underlying sill makes it unlikely that this structure is due to differential compaction. This process has resulted in the confinement by onlap of a package of sediment, consequently allowing the identification of the sea floor at the time of intrusion. Seismic correlation using the 3D survey shows clearly that the sill producing the local uplift is a constituent of the
214
K.J. TRUDE
complex of sills that underlie and connect to the Corona Sill. From this observation it is suggested that Horizon C represents a close approximation to the time of intrusion of the Corona Sill (to within 1 Ma). Taking Horizon C as the palaeo-sea bed at the time of intrusion, the compacted thickness of overburden ranges from 130-270m at the deepest level of the sill. Given that the section between Horizons B & C consists of mud/claystones with occasional intercalated sandstones, then using porosity curves from Einsele (1992), this thickness represents a decompacted thickness of 3 4 0 - 6 1 0 m using the present day burial to 4000 m, and assuming original average porosities of 0.6 and 0.65 respectively. As the majority of the ridged sill is found within 150m of Horizon C (Figures 2a & 2b) it can confidently be considered to have formed within 400m of the sediment-water interface (with compaction correction). Therefore, it would be expected that the sediment intruded (poorly consolidated muds and sands) would have contained a significant proportion of water (estimated at approximately 50%, based on Einsele 1992, p. 516, fig. 13.4).
Discussion The Corona Sill is unique in the dimensions of its ridged surface morphology, and the degree to which potential feeder systems can be imaged, both in comparison with previously described igneous sills in the field and on seismic datasets. The focus of this discussion is to offer an explanation for this unique property and to offer new insight into the plumbing of shallow intrusive systems.
Origin o f lobate geometry Tweto (1951) described lobate geometries exhibited by the peripheries of porphyry sills in the Pando area, Colorado. However lobate perimeters are not confined to intrusive bodies of magma, as such shapes are often adopted by basaltic lava flows extruded onto the Earth's surface or the sea floor. In such cases the rounded nature of the flow front is caused by the tendency of the lava to 'spread laterally under the influence of gravity' (Griffiths & Fink 1992). The development of lobes may be caused by slight variations in topography and hence division of the flow. In a similar way upon consideration of the propagation of a sill front moving through highly porous, poorly consolidated sediment at relatively shallow depth it may be expected that a similar geometry may be attained. The sill would spread laterally under the influence of the driving pressure of the magma entering the sill body. Perturbations in the advancing front would cause formation of lobes in a similar way to slight topography on the land surface. Pollard et al. (1975) employed similar arguments to attribute the initiation of finger formation at the periphery of sills to instability of the advancing interface between magma and host rock. The extremely linear western edge of the Corona Sill strikes parallel to the general structural trend of the area, which is N E SW. Indeed, the western margin of the sill can be seen on 3D seismic sections to run parallel to the Corona Ridge structure (Fig. 3), which has been interpreted as a basement fault block (Fig. lb) (Naylor et al. 1999; Dean et al. 1999). Therefore, the linear nature of the sill margin may well be associated with the presence of a block-bounding fault system. This compares with the observations of Tweto (1951), who found a number of sills with one straight edge marked by a dyke in the Pando area, Colorado. It may be considered that the linear margin of the sill on seismic may be due to an imaging problem such as masking by an overlying lava. This not the case, however, as though there is a lava close to the sill, its edge does not correlate with that of the
ridged sill. Whether a fault acted as a feeder for the Corona Sill is not certain, but the fact that the sill has a straight edge makes it more likely that a fault acted as a barrier along its western margin.
Origin o f ridges on the sill surface The most striking observation of the ridge features found on the upper surface of the Corona Sill is their orientation (Fig. 4). The alignment of the ridges with the sill margin strongly suggests that there is a genetic link between the propagation of the sill and the growth of the ridges. A lesser possibility is that the ridges were the result of some external process shaping the sill, for example either by local tectonics or by local adjustments to differential loading of the sill mass on the underlying strata. The contribution from such an external process (i.e. unrelated to the actual emplacement of the sill) seems unlikely, however, as 15 other sills were mapped surrounding the Corona Sill (some are shown on Fig. 3) and none exhibit similar ridge-like surface morphologies. Given the similarity in timing of intrusion, gross geometry, thickness and tectonic context of the Corona Sill and these surrounding sills, it is highly improbable that any such external process could have acted on the Corona Sill without affecting neighbouring sills, and thus we are forced to look at possible mechanisms related to the emplacement process itself. Small-scale linear features of various orientations have been described on the surface of sills, notably by Johnson & Pollard (1973) and Pollard et al. (1975). Johnson & Pollard (1973) described ridges that project between 1 and 5 m above the surface of a sill that have an arcuate geometry with portions parallel to the periphery of the intrusion. They suggested that most of these ridges record abortive attempts by the magma to send dykes into the overlying sandstone. The ridges described by Johnson & Pollard (1973) are however, not considered comparable to those described here from the Corona sill as they are much smaller in both lateral and vertical extent, less continuous, and result from intrusion into lithified rock at over 3kin depth i.e. under very different mechanical conditions compared to the Corona sill. Further possible explanations are connected to interactions between hot magma and water-saturated sediments during the emplacement of the intrusion. Several previous studies have described magma-sediment interactions such as peperites and the deformational features produced during this interaction such as fluidization structures and folds (Schminke 1967; Grapes et al. 1972; Kokelaar 1982; Hanson & Schweikert 1982; McPhie 1993, Brooks 1995). The features produced vary in character and magnitude depending on numerous factors such as: the sediment intruded, the state of consolidation of sediments, amount of water they contain, viscosity of the magma, magma volatile content and temperature, as well as the depth and hence confining pressure at intrusion. Confining pressure vs. temperature is one of the more important controls and governs the possibility of boiling, fluidization and steam explosions (cf. Kokelaar 1982). When confining pressure is relatively high, boiling and loss of volatiles tends to be suppressed, producing brecciation of the magma with little intermixing of magma and sediments (Hanson & Schweikert 1982). As pressure decreases relative to temperature, the degree of sediment-magma mixing and fluidization structures tends to increase. Brooks (1995) described mesoscopic scale folding in sediments above a shallow intrusion caused by fluidization. He described a basalt intruded into unconsolidated water saturated ash flow tufts where invasion of fractures on the top of the intrusion by fluidized ash and subsequent expansion of pore water resulted in the upward movement of detached basalt blocks into the overlying tuff. Where a block invaded the tuff, an elongate dome shaped forced fold was formed. The folds he
3D SEISMIC INTRUSIVE KINEMATIC INDICATORS described trend roughly parallel to bedding but are relatively small, being only metre-scale and discontinuous. Therefore they are discounted as a possible genetic analogue for the ridged sill. Grapes et al. (1972) described a curved conglomeritic dyke associated with the tongue shaped tip of a doleritic sill intruded into wet, unconsolidated sediment. They suggested the dyke formed as an explosive fissure growing from the extension space in front of the sill as it propagated forward. This curved dyke. being related to the front of the sill would have a similar arcuate geometry to the ridges seen on the top surface of the Corona ridged sill. If imaged on seismic it could produce an image similar to one flank of a ridge. This mechanism could not produce multiple dykes (or ridge like features) because the only means of preservation of the fissure is by venting to the surface causing the intrusion to cease. Some insight into a possible process for ridge development may be derived from consideration of the surface morphologies of lava flows. This might seem an unlikely avenue because traditionally the areas have been considered to be completely separate due to the presence of overburden above sills when they intrude and obvious lack of overburden above lavas during eruption. It may be considered that highly water saturated sediments might in some special cases provide such a low shear resistance to the intruding magma, that it could act like and exhibit some of the properties of surface flows. Brooks (1995) offers evidence that a very fine-grained water-rich ash at the top of an intrusion was fluidized by expanding steam and offered no resistance to the intruding magma. The geometry of ridges observed on the upper surface of the Corona Sill (Fig. 4) are strikingly similar to those observed on the upper surface of both basaltic pahoehoe lava flows and those seen on the upper surface of rhyodacitic coulees, notably the Chao Dacitic Coulee in northern Chile (Fig. 5) (Guest & Sanchez 1970; Fink 1980; De Silva etal. 1994; Francis 1996). The Chao Dacitic Coulee is the largest body of dacitic lava in the world according to De Silva et al. (1994), with a volume totalling 26 km 3. It is composed of three flow units. The largest, second phase has a volume of 22.5 km 3, covering 53 km 2, De Silva et al. (1994) describe huge 'ogives' or pressure ridges that pleat the surface. These ogives have very comparable
Fig. 5. Landsat Thematic Mapper image of the Chao dacitic coulee, northern Chile. From Francis (1996). The image is 15 km across. Massive pressure ridges or 'ogives' hundreds of metres apart and tens of metres high pleat the surface. The crater is indicated by the arrow from which arcuate ridges can be traced at the top right of the phase 2, and to a lesser extent phase 3 flows.
215
size and geometry to the undulations on the surface of the Corona Sill as is evident from a comparison of Figures 4 & 5. Guest & Sanchez (1970) stated that 'the top surface of the Chao Coulee is characterized by broad arcuate ridges; their shape reflects the movement of the flow and in part corresponds to the lobate front of the main flow'. Guest & Sanchez (1970) also reported an average wavelength of 200 m and height of 30 m for the ogives, where on the western margin, the ends of their bowed forms are aligned sub-parallel to the direction of bulk flow and the lava edge (Fig. 5), due to the more rapid advance of the central portion of the flow. This geometry compares very favourably with an average wavelength of 237 m and amplitude of 40 m as imaged on the Corona Sill (Fig. 4). Furthermore, along the eastern margin of the Corona Sill the ridges are also formed in an arc, sub-parallel to the sill edge in the same manner as the ogives on the western margin of the Chao terrestrial flow (Fig. 5). The ridges on the Chao Flow radiate from the eruption cone and are a kinematic indicator for the feeder zone and indicate the direction of lava flow. This may be the case for the ridged sill. From the similarities noted above, it seems there are grounds for considering the analogy between lava surface flow morphology and sill surface morphology a stage further.
Reconciliation of surface morphology of flows and sill It is clear that there is considerable similarity of form between the Corona ridged sill and those expected on the surface of an intermediate lava flow. It is suggested here, that although the Corona sill is intrusive, a high viscosity magma intruded into soft, water-logged sediment may act in a similar way to a lava extruded onto a free surface under conditions where the strength of the intruded sediments is very low or reduced to near zero. According to Fink & Fletcher (1978) and Fink (1980), regularly spaced surface folds in lava flows can be modelled as folds which form owing to the compression of a fluid whose viscosity decreases with depth. Surface folds can develop in lavas from basaltic to obsidian compositions. Fink (1980) concluded that although the ridges on acid flows are an order of magnitude larger, the morphological similarities between pahoehoe ropes and the ridges on rhyolitic obsidian flows suggest that the two structures form by a similar mechanism. He suggests the amplitude of folds is dependent on the temperature gradient, the contrast between interior and exterior viscosities and the ratio between the compressive stress due to flow and the gravitational stress due to the weight of the lava. Ramsay & Lisle (2000) suggested that the ropy folds seen on the surface of pahoehoe lavas are caused by movement of a glass skin downstream, dragged by the movement of lava in an underlying channel. Arcuate surface folds are formed normal to the local direction of magma flow. It is therefore reasonable to state that in pahoehoe lava, ropes develop as surface irregularities of low amplitude that are amplified at sites of compression, such as where forward movement of the surface layer is restricted. Fink (1980) suggested that brecciated rhyolite flows behave in a similar manner and that their flow fronts have an effective viscosity a few orders of magnitude higher than the bulk flow viscosity, hence retarding the advancing lava and causing compression as it backs up towards the vent. It is proposed here that a similar mechanism acted in the formation of the pressure ridges that cover the surface of the Corona Sill as that of the Chao Dacitic Coulee, whereby the drag of advancing internal magma against a more viscous outer layer and retardation of flow by the material at the nose of the sill caused subsequent compression and ridges to form. As stated above, the Corona Sill intruded at a depth of less than 400 m below the contemporaneous sea floor. Einsele
216
K.J. TRUDE
(Einsele et al. 1980; Einsele 1986) stated that basaltic sill injections preferentially occur in soft sediments some hundred metres below the sea floor. At depths shallower than several hundred metres most deep basinal sediments are soft and highly porous. As a result of sill intrusion however the surrounding sediment becomes indurated, consequently new sills will tend to intrude above older ones. The Corona Sill is found to be the highest sill stratigraphically in the surrounding complex (see Fig. 1b). This suggests that the sediments intruded by the Corona Sill would, at the time of its intrusion be unlikely to have suffered induration as would be expected in the underlying sediment surrounding the lower sills. The pressure due to the overlying 400 m of sediment at the time of intrusion can be calculated as 7.8MPa assuming a density of 2 . 0 M g m -3. The depth of seawater above the intrusion at the time of emplacement is less well constrained, but Kokelaar (1982) noted that boiling and steam explosions are suppressed during magma eruption in seawater or unconsolidated wet sediment at pressures above 31.2 MPa. In this case there would need to be about 2.4 km of water above the sill at the time of intrusion to prevent boiling occurring. The depth of water in the area at this time was less, 300-500 m, as inferred from clinoform geometry (A. Robinson 2002, pers. comm.). Therefore, it is likely that boiling and fluidization related features would have been able to develop. Kokelaar (1982) also stated that fluidization may be responsible for the mobilization or replacement of large amounts of sediment. Fluidization could have reduced the effective strength of the intruded sediments significantly, allowing the sill to invade or replace the surrounding sediment. Based on the interpretation that the top sill event is a primary reflection originating from the top of a single magmatic sill intrusion as well as drawing heavily on the strong similarity between the ridges observed on the Chao Dacitic Coulee, the following model is proposed to explain the ridges on the Corona Sill (Fig. 6). The sill initiates roughly horizontal intrusion at the feeder zones (marked X1 and X2 on Fig. 4b), fed by magma flow up the feeders from sill 1 (Figs 2a & 2b). These are essentially point or short linear intersections caused by intersection of a tip
(a)
(b)
( < 1.5 km wide) from the sill below, as previously discussed. As the magma radiates out, a surface layer of more viscous magma with numerous low-amplitude irregularities develops at the contact with the overlying sediment (Fig. 6a). As intrusion continues, viscous drag caused by forward movement of the sill interior forces the more viscous layer towards the solidifying nose of the sill, causing amplification of irregularities (Fig. 6b). Further forward movement of magma causes greater compression of folds (Fig. 6c) and draws out the arcuate nature of the ridges in the direction of flow (as seen on Fig. 4), thus indicating the direction of flow and the feeder zones. The base of the sill has been interpreted as relatively flat, and it is suggested that the weight of the overlying magma, coupled with sediment compaction below may prevent ridges forming. It is likely that magmatic pressure drove the lateral intrusion indicated by the areas where the sill has flowed upwards, and that high magma viscosity caused drag on the more solidified outer magma layer. The forward drag from the movement of the magma, coupled with the resistance caused by the solidifying frontal carapace, caused the formation of ridges. This statement can be reconciled with the observations of Tweto (1951), who reported wavy contacts as well as drag folds < 1 m high in the contact zones of sills and small thrust faults in the contact between porphyry sills and country rock in Colorado. The sills studied by Tweto (1951) were apparently intruded into competent rock, unlike the host sediments of the ridged sill. The creation of space in the sediment for the formation of ridges is consistent with observations made by Einsele et al. (1980) who asserted that total water loss in sediments caused from the uppermost sill in a complex intruded into soft-sediment is of the same order of magnitude as the thickness of the sill. This suggests that water removal from sediments creates space for the sill. Hence 3 0 - 4 0 m ridges should easily be accounted for by thermally induced differential compaction between and above ridges. Thermal fracturing and invasion of the sill top by fluidized sediments, recognized by Brooks (1995), could have created further space for ridge formation. It is very difficult to calculate meaningful flow or deformation rates for fold formation as the driving pressure of the magma or the viscosity of the magma is not known, although it is suggested that the magma must have been quite viscous because in terrestrial flows there does seem to be some correlation between fold size and magma viscosity (Fink 1980). If the size of the folds on the surface is due to high-viscosity lava, this either indicates that the magma was of a basic type but relatively cool or that it may have been more acidic in composition.
Conclusions
(c)
?
?
Magma flow direction
Higher viscosity surface layer Lower viscosity inner layer
0
500m
Fig. 6. Schematic cross section through the ridged sill at the time of intrusion. (a) The sill initiates intrusion. X marks spacing of four arbitrary surface irregularities. (b) Intrusion continues. Surface irregularities grow into folds. Y marks shortened distance between folds. (c) Distance between folds decreases to Z causing folds to grow further. '?' denotes uncertainty over geometry of the sill base. Vertical exaggeration is approximately • 5.
A roughly elliptical 'kidney-bean' shaped sill of at least 5 km 3 was emplaced at approximately 400 m below the sea floor in to poorly consolidated sediments. The surface of the sill is covered in arcuate ridges. Comparison with a terrestrial analogue indicates that the ridges are likely to have formed as a direct result of shallow intrusion of a viscous magma into soft sediment. During intrusion, compression of a more rigid outer layer retarded by the solidifying sill front caused amplification of irregularities between the sediment and the outer cooler layer of the sill. Forward movement of the surface layer is likely to be caused by viscous drag from within the sill. The ridges on the sill top surface are a kinematic indicator for the flow direction of the magma, enabling identification of the feeder zones, which display a clear link to an underlying sill. Though it is possible to observe sill networks in exhumed basin settings such as the Karoo of South Africa and the southern margin of Greenland, outcrops usually only allow observation of sills in 2D and rarely lend themselves to study of
3D SEISMIC INTRUSIVE KINEMATIC INDICATORS large-scale features, 3D seismic interpretation has significant potential to further our understanding of m a g m a transport and intrusive processes in the upper crust. Funding from Amerada Hess Ltd. is gratefully acknowledged. I am grateful to PGS Exploration (UK) Ltd for the data supplied, ExxonMobil International for help with data copying and to Schlumberger GeoQuest for use of IESX and GeoViz software. R. Davies and J. Smallwood are acknowledged for their comments and suggestions and A. Robinson for the original mapping work he undertook. I am also grateful to J. Cartwright for his guidance and editing this script. S. Planke, B. Bell and C. Line are thanked for their constructive reviews.
References BADLEY, M. E. 1985. Practical Seismic Interpretation. International Human Resources Development Corporation, BRADLEY, J, 1964. Intrusion of major dolerite sills. Transactions of the Royal Socie~ of New Zealand, 3, 27-54. BROOKS, E. R. 1995. Paleozoic fluidization, folding and peperite formation, northern Sierra Nevada, California. Canadian Journal of Earth Science, 32, 314-324, BROWN, A. R. 1999. Interpretation of Three-Dimensional Seismic Data. The American Association of Petroleum Geologists and the Society of Exploration Geophysicists, AAPG Memoirs 42, SEG Investigations in Geophysics, No. 9. CHEVALLIER, L. & WOODFORD, A. 1999. Morpho-tectonics and mechanism of emplacement of the dolerite rings and sills of the western Karoo, South Africa. South African Journal of Geology, 102, 43-54. 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. DEAN, K., MCLACHLAN, K. & CHAMBERS, A. 1999. Rifting and development of the Faroe-Shetland Basin. In: FLEET, A. J. & BOLDY, S. A. R. (eds) Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference, Geological Society, London, 1, 533-544. DE SILVA, S. L., SELF, S., FRANCIS, P. W., DRAKE, R. E. & RAMIREZ. C. R. 1994. Effusive silicic volcanism in the Central Andes: The Chao dacite and other lavas of the Altiplano-Puna Volcanic Complex. Journal of Geophysical Research, 99(B9), 17 805-17 825. DUFEIELD, W. A., BACON, C. R. & DELANEY, P. T. 1986. Deformation of poorly consolidated sediment during shallow emplacement of a basalt sill, Coso Range, California. Bulletin of Volcanology, 48, 97-107. DE TOlT, A. L. 1920. The Karoo Dolerites of South Africa: A study in hypabyssal injection. Transactions of the Geological Society of South Africa, 23, 1-42. EINSELE, G. 1986. Interaction between sediments and basalt injections in young Gulf of California-type spreading centres. Geologische Rundschau, 75, 197-208. EINSELE, G. 1992. Sedimentary Basins; Evolution, Facies, and Sediment Budget. Springer, Berlin. EINSELE, G., GIESKES, J. M. ET AL. 1980. Intrusion of basaltic sills into highly porous sediments, and resulting hydrothernaal activity. Nature, 283, 414-445. FINK, J. H. 1980. Surface folding and viscosity of rhyolite flows. Geology, 8, 250-254. F1NK, J. H. & FLETCHER,R. C. 1978. Ropy pahoehoe: Surface folding of a viscous fluid. Journal of Volcanology and Geothermal Research. 4, 151-170. FRANCIS, E. H. 1982. Emplacement mechanism of late Carboniferous tholiite sills in Northern Britain. Journal of the Geological Societ)', London, 139, 1-20. FRANCIS, P. 1996. Volcanoes: a Planetary Perspective. Oxford University Press.
217
GIBB, F. G. F. & KANARIS SOTIRIOU, R. 1988. The geochemistry and origin of the Faroe-Shetland sill complex, hi: MORTON, A. C. & PARSON, L. M. (eds) Earh' Tertiary Volcanism and the Opening of the NE Atlantic. Geological Society, London, Special Publications, 39, 241-252. GRAPES, R. H., REID, D. L. & MCPHERSON, J. 1972. Shallow dolerite intrusion and phreatic eruption in the Allen Hills region, Antarctica. New Zealand Journal of Geology and Geophysics, 17, 563-577. GRIFFITHS, R. W. & F1NK, J. H. 1992. Solidification and morphology of submarine lavas: A dependence on extrusion rate. Journal of Geophysical Research, 79, 19 729-19 737. GUEST. J. E. & SANCHEZ, R. J. 1970. A large Dacitic Lava Flow in Northern Chile. Bulletin of Volcanology, 33(3), 778-790. HANSON. R. E. & SCHWEIKERT,R. A. 1982. Chilling and brecciation of a Devonian rhyolite sill intruded into wet sediments, Northern Sierra Nevada, California. Journal of Geology, 90, 717-724. HITCHEN, K. & RITCHIE, J. D. 1987. Geological review of the West Shetland area. hi: BROOKS, J. & GLENN1E, K. (eds) Petroleum Geology of North West Europe. Graham & Trotman, London, 737-749. JOHNSON, A. M. & POLLARD,O. D. 1973. Mechanics of growth of some laccolithic intrusions in the Henry Mountains, Utah, I. Field observations, Gilberts model, physical properties and flow of the magma. Tectonophysics, 18, 261-309. KOKELAAR, B. P. 1982. Fluidization of wet sediments during the emplacement and cooling of various igneous bodies. Journal of the Geological SocieO', London, 139, 21-33. McPHIE, J. 1993. The Tennant Creek porphyry revisited: a synsedimentary sill with peperite margins, Early Proterozoic, North Territory. Australian Journal of Earth Sciences, 40, 545-558. MITCHELL, S. M., BEAMISH, G. W., WOOD, M. V., MALACEK, S. J., ARMENTROUT, J. D., DAMUTH, J. E. & OLSEN, H. C. 1993. Paleogene sequence stratigraphic framework of the Faroe Basin. In: PARKER, J. R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 1011-1023, MUDGE. D. C. & RASHID. B. 1987. The Geology of the Faroe Basin area. In: BROOKS, J. & GLENN1E, K. (eds) Petroleum Geology of North West Europe. Graham & Trotman, London, 751-763. NAYLOR, P. H., BELL, B. R., JOLLEY. D. W., DURNALL,P. & FREDSTED, R. 1999. Palaeogene magmatism in the Faroe-Shetland Basin: influences on uplift history and sedimentation. In: FLEET. A, J. & BOLDY, S. A. R. (eds) Petroleum Geology of Northwest Europe." Proceedings of the 5th conference. Geological Society, London, 1, 545-558. PLANKE, S., ALVESTAD, E. & ELDHOLM, O. 1999. Seismic characteristics of basaltic extrusive and intrusive rocks. The Leading Edge, March, 1999, 342-348. PLANKE, S., SYMONDS, P. A., ALVESTAD, E. & SKOGSEID, J. 2000. Seismic volcanostratigraphy of large-volume basaltic extrusive complexes on rifted margins. Journal of Geophysical researchSolid Earth, 105(B8), 19 335-19 35 !. POLLARD, O. D., MULLER,O. H. & DOCKSTADER,O. R. 1975. The form and growth of fingered sheet intrusions. Geological Society of America, Bulletin, 86, 351-363. RAMSAY, J. G, & LISLE. R. J. 2000. The techniques of modern structural geology. Applications of Continuum Mechanics in Structural Geology, Vol. 3, Academic, London. SCHM1NKE, H. W. 1967. Fused tuff and peperites in South-Central Washington. Bulletin of the Geological Society of" America, 78, 319-330. TWETO, O. 1951. Form and structure of sills near Pando, Colorado. Geological Society of America Bulletin, 62, 507-532. YILMAZ OZ 2001. Seismic data analysis processing, inversion and interpretation of seismic data. SocieO' of Exploration Geophysicists, 2027.
Visualization and interpretation of 3D seismic in the Carboniferous of the UK Southern North Sea JEREMY
J. L Y N C H
ConocoPhillips (UK) Ltd, Rubislaw House, Anderson Drive, Aberdeen AB15 6FZ, UK Present address: Perenco UK Ltd, 29 Duke of York Square, London, SW3 4LY (e-mail: jlynch @uk.perenco.com)
Abstract: Examples from the Carboniferous of the UK Southern North Sea are used to illustrate the application of visualization tools to increase the effectiveness of 3D seismic interpretation. Maintaining the seismic data to the fore throughout the interpretation workflow and decision-making process is vital for well planning in field development and trap validation in prospect evaluation. Scaling of the seismic into the "true-depth' domain and the automatic generation and visualization of layers using isochores derived from well data are the key elements in this interpretation workflow. Together these allow full integration of the seismic with various well data and interpreted elements. The main example is from Murdoch K, one of ConocoPhillips's most recent UK discoveries, part of a pre-Permian extensional graben inverted during the Early Tertiary tectonic phase. Other examples from the McAdam Field and an exploration prospect are used to highlight additional aspects of the method.
Advanced techniques for 3D seismic visualization, volume rendering and attribute analysis have, quite literally, transformed the way we look at seismic data. Different techniques, however, are suited to different areas and what works for one geological province or seismic dataset may not work for another. The Carboniferous of the UK Southern North Sea (SNS) is characterized by the following 'everyday' problems, which call for particular 'everyday' solutions: 9
Key reservoir and seal intervals that are often not adequately resolved in the seismic to allow direct interpretation. 9 Moderate structural complexity at reservoir level involving multiple episodes of faulting and erosion. Combination structural-stratigraphic traps are common, and to understand the geometry of truncated reservoir and sealing units is critical. 9 A complex velocity structure in the overburden. Seismic data, whether time or depth processed, must be converted or corrected to 'true depth', or at least what we believe to be 'true depth'. It is essential that the seismic be fully integrated and visualized alongside well data.
Unconformity, which separates the Carboniferous from the overlying Silverpit Formation of the Lower Permian Rotliegendes Group is well documented (e.g. Hollywood & Whorlow 1993; Quirk 1997: Quirk & Aitken 1997). The subcrop to this unconformity surface reveals the predominately west-east trending pre-Permian extensional fabric of the basin. Superimposed on this is a N - S to N W - S E trending compressional fabric related to the Early Permian uplift event itself. A simplified version of the subcrop is included in Figure 3. The Murdoch and Caister Fields produce from the Lower Westphalian B Murdoch sands. As expected, the 44/22a-10Z well encountered gas in stratigraphically younger Westphalian C/D sands. The stacked fluvial channel sands are ascribed to the Ketch and Lower Schooner formations based on lithostratigraphic, palynostratigraphic and chemostratigraphic correlation according to a scheme proposed by Collinson (1997) (Fig. 4). In order to evaluate the gas accumulation and design a development sidetrack, these sands must be accurately mapped away from this isolated well control. A detailed understanding of both stratigraphy and structure is critical in this regard.
MurdochK Interpretation workflow Murdoch K is one of five Carboniferous gas discoveries in Quad 44 to be developed over the course of 2002 to 2003 as part of the third development phase of the Caister-Murdoch System (CMSIII). These fields will be tied back to expanded and upgraded facilities on the Murdoch Field. Murdoch K itself straddles UK blocks 44/22a and 44/23a. It was discovered by well 44/22a-10Z, drilled early in 2001 directly between the Murdoch and Caister Fields (Fig. 1). These fields have been on production since 1993 and are described in Birrell & Courtier (1999) and Ritchie & Pratsides (1993) respectively. Murdoch, Caister and Murdoch K lie on the same N W - S E trending structural high at Top Carboniferous. Tertiary inversion of the 'Murdoch Fault' system on its northern boundary is responsible for most of this present day relief (Corfield et al. 1996). The seismic data, from a ConocoPhillips 1992 timemigrated 3D survey reveal a wedge of Westphalian sediments separated from the neighbouring fields by pre-Permian normal faults (Fig. 2). The seismic expression of the Base Permian
As is usually the case for the SNS Carboniferous, neither the individual Murdoch K reservoir sands nor the Base Permian top seal can be mapped directly in the seismic data without difficulty. Of the various Westphalian C/D analogues, only in the Hawksley discovery are the sands sufficiently thick and of sufficiently low impedance to produce a consistent mappable event (Fig. 5). The inherent resolution and impedance contrast limitations are compounded in this case by continuity loss beneath tightly folded rafts of Zechstein Z3 Plattendolomit and Hauptanhydrit. It is clear from the seismic and from extensive well penetration that these rafts often thicken dramatically in the troughs of these folds (Figs 5 & 6). This is presumably due to gravity-enhanced sliding of the relatively competent anhydrite induced by diapiric flow in the surrounding Stassfurt and Leine halites. The areas of poorest seismic data and velocity pull-up directly underlie these thickened rafts. Amplitudes extracted at the Top Rotliegendes event perhaps best illustrate this relationship (Fig. 6).
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, 219-225. 0435-4052/04/$15 9 The Geological Society of London 2004.
220
J.J. LYNCH digitized directly using various vertical and horizontal slices of the seismic data. Working in 3D forces the interpreter to create a structural model that is consistent with all mapped horizons-traditional 'fault polygon' maps are an optional final output for display purposes, rather than an interpreted input. Reverse faults such as the 'Murdoch Fault' do not present a problem (Fig. 6), as the software allows multiple z-valued surfaces. Regional well control is used to create isochores for the Silverpit and various Westphalian layers (in this example it is assumed that no faults were active during deposition), which are converted to time isochrons using velocities from the VSP. These are used to define the surfaces that cannot be picked directly on the seismic surfaces that are automatically truncated at faults and unconformities. The essential step is verification of the gridded 3D time model using the seismic data. The entire model and seismic data are then depth converted, iteratively if necessary, using a conventional layer-cake approach. The resulting depth model (Fig. 8) can be used for volumetrics, well planning and/or more detailed reservoir modelling. Critical displays such as geoseismic sections and subcrop maps can be generated quickly and easily.
Post-stack depth migration interpretation
Fig. 1. Map showing location of Murdoch K, McAdam and surrounding fields.
Time-migration interpretation Interpretation of these seismic data is therefore based on the only clear Carboniferous seismic marker, related to the 15~ Lower Westphalian B 'C3 coal' some 800 ~or so below the base of the Westphalian C/D. The workflow is summarized as a chart in Figure 7. Initially, Landmark's SeisWorks package is used for the generation of key horizons using conventional methods. Petrel software is then used for 3D fault interpretation, map gridding, depth conversion, well planning and visualization. Fault 'pillars' are
A fast-track post-stack depth migrated seismic cube is now available over Murdoch K as the preliminary output of an ongoing full pre-stack depth migration programme. The velocity model incorporates the Plattendolomit/Hauptanhydrit rafts and the result is a subtly improved image that permits interpretation of the thick uppermost Schooner sand over much of the Field. This allows a more detailed and focused interpretation over Murdoch K suitable for well planning purposes. The seismic are interpreted in the original depth domain, and the seismic, faults and horizons are all stretched to true depth using constant scaling factors in a layer-cake model. In all other respects the workflow is the same as above. The intra-Ketch surfaces, derived using the Top Schooner pick and constant isochores from the 44/22a-10Z well are consistent with the seismic data. Use of this 3D model has allowed a cost-effective development sidetrack to be planned (Fig. 9).
Fig. 2. Time-migrated seismic line along the axis of the Murdoch-Caister structural high. See Figure 1 or Figure 3 for location of section.
3D SEISMIC VISUALIZATION IN THE CARBONIFEROUS
221
Fig. 3. Base Permian subcrop map for the Quadrant 44 area compiled from 3D seismic and well data. Murdoch K is one of a series of broadly WSW-ENE trending pre-Permian extensional graben.
Structural interpretation Structurally, Murdoch K is distinct from the Murdoch Field, separated from it by a major curvilinear and listric prePermian extensional fault. Unusually, there is some evidence for thickening of the lower Westphalian B into this fault (Figs 2 & 5), indicating the fault was active somewhat earlier than others in the area. The Post-Permian evolution of the basin has previously been described by Van Hoorn (1987), Ritchie & Pratsides (1993) and Quirk & Aitken (1997). Of greatest importance here is that the fault has been partially reactivated and utilized by the 'Murdoch Fault', a major Early Tertiary reverse fault that defines the NE margin of the Cavendish, Murdoch, Murdoch K and Caister Fields. Detailed inspection of the 3D model reveals clear sand-onsand juxtaposition in the gas leg between Westphalian C/D sands in Murdoch K and Murdoch sands in the Murdoch Field. This, in conjunction with evidence from FMT and log data has caused the DTI to determine the 44/22a-'K' structure as part of the Murdoch Field, hence Murdoch K.
Fig. 4. Westphalian stratigraphic nomenclature for Quadrant 44 of the SNS, highlighting reservoir and seal units for the Murdoch K and McAdam Fields. exploration prospect A and neighbouring gas fields.
Other examples Following on from Murdoch K, these methods have been successfully applied to many other Carboniferous structures in the SNS of which two are described below.
McAdam Field Of the other CMSIII accumulations, 44/17c-McAdam, some 10kin north of Murdoch K (see Fig. 1), perhaps best illustrates
the use of depth-scaled seismic slices in well planning (Fig. 10). The development well, to be drilled immediately after the Murdoch K sidetrack, will aim to contact three distinct sand intervals: a lower Westphalian B sand-prone unit, the Murdoch interval and the uppermost Westphalian A. The well must be designed so that it encounters all three intervals above the gas-down-to recorded in the discovery well 44/17-1. The key uncertainty is the very low angle truncation of these intervals by the Base Permian Unconformity. Experimenting with isochores derived from well data and comparing with the
222
J.J. LYNCH
3D SEISMIC VISUALIZATION IN THE CARBONIFEROUS
223
Fig. 5. Examples of Carboniferous seismic response from the UK SNS. Note the generally poor imaging of the Base Permian Unconformity and Westphalian C/D sands except in the case of Hawksley and Murdoch K. Also note the generally isopachous nature of the Westphalian layers. In all cases the time range is 800 ms and the horizontal range approximately 4.5 km. Fig. 6. Top Rotliegendes TWT structure over Murdoch K highlighting the image disruption beneath thickened troughs in the Hauptanhydrit/ Plattendolomit rafts. Vertical exaggeration is approximately x 1.4.
depth-stretched uncertainty.
time-migrated
seismic
reduces
this
Exploration prospect A Similar techniques can be used in prospect evaluation, even if there is no direct or nearby well control. Prospect A (Fig. 11) invokes Westphalian C/D Ketch reservoir and combination Silverpit and Upper Ketch top seal, for which there are several analogues, most notably the Tyne Fields (O'Mara et al. 1998). Again, only the Top Rotliegendes and a (different) Westphalian B coal marker can be identified clearly in the seismic. The key trap-defining surfaces are
therefore defined using isochores for the Silverpit and intraWestphalian layers derived from regional well control, Multiple isochore contouring realizations are constrained by comparison with the depth-converted seismic data. The result is an appropriate range of predicted reserves and a more complete understanding of the trap risks.
Conclusions Software designed primarily for volumetrics and well planning, in this case Petrel, can also be used for the structural and stratigraphic interpretation of seismic data. To understand and
Fig. 7. Simplified time migration interpretation workflow diagram, with the input data and secondary output products at each stage. The bold arrows represent key QC steps using seismic and well data. Note that the 'reservoir zonation' and 'depth conversion' steps are interchangeable if there is no velocity contrast between reservoir layers.
Fig. 8. Murdoch K 3D depth model based on time-migrated data and interpretation scaled to depth. The C3 coal surface is shown, along with the intra-Westphalian C/D events derived from it and a depth slice at the 44/22a-10Z gas water contact. Vertical exaggeration is x 2.
224
J.J. LYNCH
Fig. 9. Murdoch K 3D depth model based on post-stack depth migrated data corrected to true depth. A depth slice at the 44/22a-10Z gas water contact and a vertical slice along the planned development sidetrack are shown. Vertical exaggeration is x 2.
visualize the layered reservoir system is the key to the Carboniferous play and a particular brand of seismic visualization must be used to effect that understanding. Scaling of the seismic into the 'true-depth' domain and the automatic generation and visualization of layers using isochores derived from well data are the key elements in this interpretation workflow. Note: Since submission of the original manuscript for this paper, both the Murdoch K and McAdam Fields have been
successfully brought development.
on
stream
as part
of the
CMSIII
I would like to thank ConocoPhillips (UK) Ltd and our partners Gaz de France Britain and Tullow Oil UK for permission to present these examples. A. Conway, S. Easton and R. Ings provided geological input to the 3D models and Figure 3. C. Gunn and J. Courtier performed much of the early seismic interpretation for these examples. I also wish to thank J. Farrell for continued help and support.
Fig, 10. McAdam Field 3D depth model based on time-migrated data scaled to depth. A depth slice at the 44/17-1 gas-down-to and a vertical slice along the planned development well are shown. Vertical exaggeration is • 4.
3D SEISMIC VISUALIZATION IN THE CARBONIFEROUS
225
Fig. 11. Use of isochore maps derived from (untruncated) well data for trap definition of exploration prospect A. Multiple realizations are rapidly generated and compared against the seismic data. Vertical exaggeration is • 3.
References
O'MARA, P. T., MERRYWEATHER,M. & COOPER, D. 1998. The Tyne Gas Fields. Proceedings of the PETEX Conference, Extended
BIRRELL, S. • COURTIER,J. 1999. Structural analysis of 3D seismic
Abstract C 10. QUIRK, D. G. 1997. Sequence stratigraphy of the Westphalian in the northern part of the Southern North Sea. In: ZIEGLER, K., TURNER, P. 8z DAINES, S. R. (eds) Petroleum Geology of the Southern North Sea: Future Potential. Geological Society, London, Special Publications, 55, 153-168. QUIRK, D. G. & AITKEN, J. F. 1997. The structure of the Westphalian in the northern part of the southern North Sea. In: ZIEGLER, K., TURNER, P. & DAINES, S. R. (eds) Petroleum Geology of the Southern North Sea: Future Potential. Geological Society, London, Special Publications, 55, 143-152. R.1TCHIE,J. S. ~z PRATSIDES,P. 1993. The Caister Fields, Block 44/23a, UK North Sea. In: PARKER, J. R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 759-769. VAN HOORN, B. 1987. Structural evolution, timing and tectonic style of the Sole Pit inversion. Tectonophysics, 137, 239-284.
data, using the correlation attribute: a case s t u d y ~ a r b o n i f e r o u s of the Southern North Sea (UK). In: FLEET, A. J. & BOLDY, S. A. R. (eds) Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference. Geological Society, London, 789-797. COLLINSON JONES CONSULTING,1997. Stratigraphy and Sedimentology of the Westphalian C/D Barren Red Beds in Quadrants 44 and 49, Southern North Sea. (proprietary report). CORFIELD, S. M., GAWTHORPE, R. L., GAGE, M., FRASER, A. J. & BESLEY, B. M. 1996. Inversion tectonics of the Variscan foreland of the British Isles. Journal of the Geological Socie~. ~,London, 153, 17-32. HOLLYWOOD, J. M. & WHORLOW, C. V. 1993. Structural development and hydrocarbon occurrence of the Carboniferous in the UK Southern North Sea basin. In: PARKER, J. R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 689-696.
Direct visualization and extraction of stratigraphic targets in complex structural settings HUW
JAMES,
ROB
BOND
& LUCY
EASTWOOD
Paradigm Geophysical, 820 Gessner, Houston, Texas, 77024, USA
Abstract: Advances in 3D visualization and volume detection have greatly improved our ability to find stratigraphic targets in areas of complex structure. In this paper we apply volume sculpting, seismic facies analysis and prediction of rock properties from seismic data to resolve a potential reservoir channel from a highly faulted setting in offshore Indonesia. These techniques use computational power to search for patterns in the data. These patterns may be presented to the interpreter or may be used to transform seismic and well data into estimates of rock properties. Such techniques should improve development efficiency in areas where complex structure has previously precluded the use of these techniques.
Volume interpretation methods for 3D seismic interpretation have been evolving over the last twenty years. Early attempts were modelled after traditional interpretation on paper. More modern volume visualization applications frequently use true 3D techniques borrowed from medical image analysis; one of these is sub-volume detection. This technique tries to track volume values that are contiguous and fall within a specified range. In the medical field sub-volume detection is used to extract segments of bone, vessels or tissue from 3D computer tomography or magnetic resonance imaging scans (Serra 1997). In seismic interpretation the same technique has been used to map amplitude anomalies, horizons and stratigraphic features. In the last few years techniques based on trace shape have been used to extract features from seismic volumes. This method closely matches the method that a human interpreter uses to identify horizons and other features in seismic volumes. Trace shape can also be used to classify regions of seismic volumes. The manual extraction of seismic facies from 3D volumes is extremely laborious. Neural networks can automatically classify trace intervals from single or multiple volumes. A simple classification based on trace shape can be performed by a computer and used to speed up the manual process. A more complex classification based on multiple volumes may find and classify patterns hidden the data that would be otherwise undetectable by a human interpreter. After a computer classification has been made an interpreter can discern whether there is any geological significance to the patterns found. Neural networks trained on both well and seismic data can be used to form estimated volumes of petrophysical parameters such as porosity. These estimated 3D volumes can be interpreted by selecting data based on value ranges of a single variable or by using regions from cross-plots of multiple variables in a similar manner that petrophysicists use to interpret 1D log curve data. When used collectively these techniques offer completely new workflows for the interpreter. The data set under consideration is from offshore Indonesia, in the zone of complex plate interactions related to the extrusion of Southeast Asia in response to the collision of India since the early tertiary. The general evolution consists of E o c e n e - O l i g o c e n e extension, with attendant graben formation followed by late tertiary inversion. A dominant fault runs the length of the 3D dataset from east to west and appears to have formed as a fight lateral tear fault with the formation of a small pull-apart basin at a fault bend. Geometric analysis of the pullapart suggests right lateral slip of roughly 1.5 km. Minor inversion has occurred in the late tertiary to further complicate the structure. 5000km of 2D seismic exploration has been conducted over the block since 1978. The first discovery well
was drilled in 1978 to a depth of 1600m reaching Oligocene sandstones in a faulted anticline. A deeper discovery at a depth of 1800m in Eocene sands in 1995 initiated a new exploration program that included the 10 0 0 0 k m 3D seismic survey used in this paper. The initial exploration targets were seismic amplitude anomalies that were both bright and flat (see Fig. 1). We began our interpretation with rapid scans through the seismic volume. These scans are simply animations back and forth in the inline, cross line and depth directions. Rapid animation accentuates changes in reflector and fault geometry. Rapid animation also enables the interpreter to comprehend features that fall on many successive slices; this is especially true for stratigraphic features that have been tilted subsequent to deposition. Most of the wells lie along the axis of a faulted anticline and are drilled to depths of 1600m penetrating numerous flat bright spots. There are two later wells, labelled A and B, that are drilled to different targets characterized by bright events terminating against normal faults. Investigating the seismic data close to well A yielded a possible channel. As soon as evidence of the channel was disclosed by scanning slices it was re-examined in section view and plan view. The width of the channel identified varies from 4 0 4 m to 360m in the region the downthrown side of the major fault. The interpreted depth of the channel feature is roughly 8m. The channel feature was picked using sub-volume detection and its reported area was sufficient to make it an economic target (see Figs 2a and 2b. However the area is strongly faulted so volume picks are terminated by faults. To interpret further needed more work.
Structural interpretation First a strong reflector that is representative of the structure at the depth of investigation was chosen, we call this a template horizon. Once chosen, seed points on the horizon were selected and regions grown using a shape based autopicker. This sequence was repeated until a fairly dense horizon pick was achieved. This pick was interpolated to fill in small gaps of less than four traces using a simple bi-linear operator. This surface was lightly smoothed and then used to flatten the volume. Precise correlation of the template reflector across faults is not critical since its purpose is simply to remove the structural complexity and further reveal detailed stratigraphy. The purpose of the light smoothing is to remove stratigraphic details in the template horizon so they will not be directly expressed on other horizons after flattening. Flattening disclosed additional channel segments including one on the upthrown side of the major fault.
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, 227-234. 0435-4052/04/$15 9 The Geological Society of London 2004.
228
H. JAMES ET AL.
Fig. 1. Sections from a seismic reflection volume from offshore Indonesia. Well D is the discovery well drilled into a faulted anticline. Numerous amplitude anomalies are shown. Well A is drilled into a deeper bright spot and has passed by a possible channel. This view is from the southeast.
If these segments belong to the same channel then they show evidence of about 1 . 5 k m of right lateral slip on the fault. The width of the additional segments is also between 300 and 4 0 0 m and their interpreted depth is also close to 8 m, supporting the correlation For this feature the structural interpretation revealed the stratigraphic detail, and the stratigraphy is revealing structural information so that each interpretation is benefiting from the other. Once the various segments of the original channel are disclosed, the whole feature was re-picked using sub-volume d e t e c t i o n based on a range of a m p l i t u d e values. All interpretation performed on the flattened volume can be
datumed to the original volume by un-flattening, this process restores both the seismic data and horizon interpretation to the original datum. The volume was sculpted to remove all data above the template horizon and all data below a similar deeper template. Then the opacity was set such that the channel amplitudes would be opaque and all other amplitudes would be transparent. This variable opacity display is rendered above a display of the deeper template horizon. The result is shown in Fig. 3. This image is rich with detail on the faults with hints of their timing and will clearly benefit future detailed interpretation. The striations in the channel data are artefacts of volume rendering sampled data. The thickness of the channel feature is
Fig. 2. (a) A time slab from 14601520 ms rendered with variable opacity so that only bright negative amplitudes are shown. The scale is given by the tick marks on the x-axis, these are at I km spacing. North is at the bottom of the slide. One large meander of the channel is evident, the width of the channel varies from 300-400 metres. (b) The same time slab as Fig. 2(a) with the channel picked using sub-volume detection. The channel is broken into three distinct segments by faults. (c) Shows the same interval as 2(a) and 2(b) after the volume has been flattened to the template horizon. Flattening has exposed channel segments of similar characteristics on both sides of the major fault.
DIRECT VISUALIZATION OF STRATIGRAPHIC TARGETS
229
Fig. 2. (continued).
almost constant at two samples but it is occasionally represented by either a single sample or by three samples. These variations cause the striations in the display. There is a possible crevasse splay to the fight of the main meander and indications of laminar deposits in the interior of the meander. The width of the channel is very consistent in this view using seismic amplitude data. All of this detail has been disclosed using six manual horizon seeds for input to an automatic horizon picker, followed by horizon interpolation and horizon sculpting. Well A clips the edge of this channel and the log curves for this well show evidence of a thin sand with some porosity at the depth of the channel feature. All of this analysis was accomplished within one hour from the initial scanning of the volume.
Trace classification The template horizon was also used as a reference to classify a trace interval automatically. This classification used a 60ms interval of seismic data from each trace of the survey starting at the depth of the template horizon. These trace interval samples were input to our proprietary classification program, Stratimagic. In this instance a hierarchical classification was made after the samples had been analysed and the dimensionality of the problem reduced using principal components analysis. The interpreter requested the program to sort the traces into 12 classes. This automatic comparison took a few minutes of computer time and the results are shown in Figure 4. This process has also identified the channel.
230
H. JAMES ET AL.
Fig. 3. A volume sculpted view of the data from the Northeast showing the channel and the major fault. Opacity has been set to show only large negative amplitudes. The offset nature of the two channels is apparent in this view. Additional details of the en echelon secondary faults along the main trend are also made evident.
The work so far has given us two views of the same channel, one based on sub-volume detection the other based on classification using trace shape. Sub-volume detection is based exclusively on connectivity and a value range for the samples, in this case we used seismic amplitude. The trace shape classification shown in Figure 4 is based on all the samples in the 60ms interval not just the value of a single sample. Trace shape in an interval may be dominated by any characteristic. For example an interval of 16 samples may be dominated by the width of a peak, the separation of two peaks or the height of a peak. The algorithm chooses automatically which characteristics are most important for the classification. In this example both techniques have shown the same meander, The trace shape algorithm used a sequence of samples so it will be affected by variations in lithology and fluid content of units above and below the channel. The data displayed indicates a potential sand rich target but seismic amplitude data is a poor indicator of lithology, porosity and fluid content since the amplitudes respond to changes in impedance not the impedance values themselves. We know that acoustic impedance and elastic moduli are better predictors of lithology and fluid content (Schuelke & Quirein 1998; Goodway et al. 1997) so the next step in the
interpretation sequence followed some substantial seismic data processing.
Petrophysical attributes A sub-set of the pre-stack data was pre-stack depth migrated and during the process the partially migrated data was input to Probe, our proprietary elastic inversion application, and used to derive estimates of P-wave and S-wave reflectivity from the amplitude variation with offset or angle. To obtain good estimates of reflectivity from AVO/AVA variations in areas of complex structure requires that the migration algorithms correctly preserve amplitude. Migration velocity data derived from this seismic data processing was used with the vertical velocity data from the wells to estimate the P-wave vertical velocity throughout the volume. This estimation was performed using geo-statistics with the migration velocity used as external drift. The P-wave reflectivity was integrated in depth to form a P-wave impedance volume. Missing low-frequency data was added using the well data and the P-wave velocity field. The S-wave reflectivity was similarly integrated to form the S-wave impedance volume with S-wave velocity estimated from
DIRECT VISUALIZATION OF STRATIGRAPHIC TARGETS
231
Fig. 4. Shows the results of a trace shape classification in Stratimagic. In this example a 60 ms interval below the template horizon has been sorted into twelve classes. These classes show the same channel as the amplitude displays.
P-wave velocity and empirical relationships between the two velocities. One of the by-products of pre-stack depth migration is an estimate of the migration velocity field. The accuracy of this field controls both the quality of the output image and the lateral accuracy of the migrated results. When we estimate the vertical average velocity from well data and the migration velocity field we can also estimate the accuracy of this result away from the wells. This accuracy controls the quality of our depth estimation. The vertical velocity was re-used to compute impedances and will be re-used later to compute elastic moduli. The quality of all these transformations largely depends on the quality of the initial velocity field. Therefore it is beneficial to all these transformations to use the best possible method to compute the migration velocity field. In this example the migration velocity field was picked manually and then updated using global tomographic procedures. Estimates of uncertainty would be a valuable extra to offer alongside each interpretation deliverable. These four volumes, P-wave reflectivity, P-wave impedance, S-wave reflectivity and S-wave impedance and porosity data from the wells were used to train a neural net, in our proprietary Vanguard application (Schuelke et al. 1997). Once trained at the wells this neural net was used to estimate a porosity volume for the survey. This volume is shown in two views below in Figures 5a and 5b. Data below 20% has been removed by making it transparent, the colours yellow, red and magenta represent porosities of 20%, 24% and 28% respectively. This technique has automatically detected parts of the channel using a porosity cut-off. Petrophysicists use porosity cut-offs to define porosity zones within a well. This is an example of the same
technique used to define a 3D zone within a 3D volume of estimated petrophysical data. The point needs to be stressed that in this example there is no overt detection of the channel. The only criterion used is the porosity cut-off. The fact that the body imaged is coincident with the channel is very encouraging. The log curves at a nearby depth on the other side of the fault indicate high porosity and low water saturation. We converted the P-wave and S-wave impedance volumes to volumes of elastic moduli. We define four physical parameters: kappa, the bulk modulus of elasticity; rho the bulk density; lambda and mu Lame's parameters, where mu is the shear modulus and lambda is bulk m o d u l u s - (2/3) * mu. These parameters are related as follows:
Vp = ~/[(h + 2* ~)/p] Vs = ~-~/01 h = K - (2/3)* p~. The driving force for lambda, mu, rho interpretation otherwise known as LMR, is that it nicely organizes hydrocarbons into low lambda*rho, low mu*rho regions (Goodway et al. 1997). The elastic moduli respond to changes in lithology, pore size and fluid content. These responses can be organized into Table 1. The boundaries of these classes are found by cross plotting "known data at the wells. The choice of which is the best indicator to use needs to be calibrated and tested for each local situation (Batzle & Han 2001). These boundaries are not precise and are not exclusive. When well data is available it is possible to
232
H. JAMES ET AL.
Fig. 5. Shows an estimated porosity volume with a cut-off of 20%. The picture shows a fault and well bore with water saturation log curve to the left and porosity curve to the right. There is no geometric selection of the channel, the whole volume has been thresholded so that values less than 20% are transparent. (a) shows well A from the northwest, (b) shows the same well from the southeast.
DIRECT VISUALIZATION OF STRATIGRAPHIC TARGETS Table 1. Distribution of lithologies in lambda*rho vs mu*rho cross-plot High mu*rho Cemented sands Carbonates Med mu*rho Gas sands Low mu*rho Coals Shales Low lambda*rho Med lambda*rho High lambda*rho
estimate the probability that data in each class is of the indicated type by counting the correct and incorrect predictions, For this survey we chose to cross plot kappa*rho vs mu*rho. Then we selected a region of low kappa*rho and low mu*rho to
Fig. 6. A cross plot of k*rho vs mu*rho (k is bulk modulus, mu is shear modulus and rho is density) at the left. The area selected is both low k*rho and low mu*rho indicating gas or coal. The points that lie in this region are marked in pink on volume display of P-wave impedance on the fight indicating the presence of hydrocarbons.
233
include all coals and gas sands that we had identified from the well data. The results are shown in Figures 6(a) and 6(b). The selection of this region the cross-plot indicating hydrocarbons has automatically selected part of the channel. The result of this interpretation is that geo-bodies in 3D can be detected automatically from petrophysical constraints and their shape can be interpreted directly by a geologist. These detections are quantitative so hydrocarbon volumes can be estimated immediately. Figures 6(a) and 6(b) show examples of elastic property detections using volume based cross plotting. No horizons or fault picks have been used in the interpretation.
234
H. JAMES ET AL.
Summary We have shown how attribute volumes may be calculated, m a n a g e d and interpreted using volume-based techniques. Modern depth imaging and amplitude inversion procedures produce products which help extract detailed stratigraphic information. If the attribute volumes produced are calibrated to borehole and laboratory data their value is further enhanced, since the 3D attributes can be used to infer rock and fluid properties. This value can be extracted quickly and economically by using modern volume-based interpretation methods. These workflows show how seismic and well data can be used together with information from the pre-stack domain to produce added value deliverables for the interpreter. The examples shown and the software used reflect the work of very many of our colleagues, too many to mention individually. We'd like to acknowledge Anat Canning, Joanne Wang for the AVO contributions with the Probe system. We'd like to thank Andy Peloso for the Seismic classification. Thanks are also due to Scot Krueger
and the anonymous reviewers who helped with the manuscript. The classification products Stratimagic and SeisFacies are based on patented technology from TotalFinaElf and ENI/AGIP, respectively. The seismic and well data was provided by courtesy of Clyde Petroleum.
References BATZLE, M. & HAN DE-HUA, M. 2001. Optimal hydrocarbon indicators. SEG Annual Meeting Abstract. GOODWAY, W., CHEN, T. & DOWNTON, J. 1997. lmproved AVOfluid detection and lithology discrimination using lame petrophysical parameters. CSEG Annual Meeting Abstracts. SCHUELKE, J. 8~ QUIREIN, J. 1998. Validation a technique for selecting seismic attributes. SEG Annual Meeting Abstracts. SCHUELKE, J., QUIREIN, J., SARG, F., ALTANY, D. & HUNT, P. 1997. Reservoir Architecnire and porosity distribution, pegasus field, West Texas. SEG Annual Meeting Abstracts. SERRA, L. 1997. Interactive vessel tracing in volume data. ACM Symposium on interactive 3D graphics.
Locating exploration and appraisal wells using predictive rock physics, seismic inversion and advanced body tracking: an example from North Africa G. PICKERING
I
E
KNIGHT
2 J. B L E T C H E R
1 R
BARKER
3 & M
KEMPER
4
I Troy-Ikoda Limited, The Coach House, 90 Alma Road, Windsor SL4 3ET, UK (e-mail: [email protected]) 2 Imperial College of Science, Technology and Medicine, Exhibition Road, London SW7 2AZ, UK 3 Burlington Resources (Energy Services) Inc. 1 Canada Square, Canao' Wharf, London El4 5AA, UK 4 Ikon Science Limited, 2 Castle Business Village, Station Raod, Hampton, Middlesex TW12 2BX, UK
Abstract: A case study is described that illustrates a complete reservoir property prediction workflow, from petrophysical analysis, through rock physics, impedance inversion, and on to interpretation and final drilling locations. This study is from onshore Algeria, where the hydrocarbons are found in several clastic reservoirs of varying ages and properties. The controls on the presence of both reservoir and hydrocarbon location are not straightforward, therefore reducing the risk of drilling locations has significant value. This study concentrated on the Triassic TAG-I formation, which forms the major reservoir in the study area. The prediction method used was an impedance based deterministic approach, using relationships based on rock physics, although the interpretation method includes fuzzy set classifications to take into account the "non-uniqueness" inherent in any seismic attribute. The petrophysical work ensured that the analysis of each well reconstructed 'virgin-zone" conditions. Rock physics models were then used to predict shear wave velocities. Acoustic impedance (AI), shear impedance (SI), and elastic impedance (EI) profiles were derived. Shear impedance was the best pure lithology indicator, with acoustic impedance showing a good relationship to porosity and elastic impedance most sensitive to fluid content. As the near angle seismic data were too noisy, gradient/intercept analysis was impossible, so shear reflectivity and consequently SI could not be derived. Although AI showed some lithological discrimination between sands and shales, it was not sufficient to be used as a single discriminator, so interpreted horizons were used to separate reservoir and non-reservoir intervals. The far-angle data were inverted to EI and this was analysed using a fuzzy logic approach. This method produces a classification volume and 3D body tracking was then used to find the best drilling targets. Generally, the analysis correctly predicted the results of the wells. both discoveries and non-discoveries. However, some of the discoveries were not predicted, which appears to be where the classification was not correctly calibrated. Work is now underway to improve the accuracy of the prediction process.
3D seismic data has revolutionized the way in which the subsurface is analysed. In addition to the improved structural definition that 3D provides, the advances in processing to true amplitude allow analysis of the data in terms of specific rock properties. The key strength of 3D seismic data in development studies, is the ability to predict rock properties away from the well control. A case study is described below which illustrates the complete workflow: from petrophysical analysis, through rock physics to predict optimal seismic attributes and on to impedance inversion, interpretation and final drilling locations. The case study is from onshore Algeria, North Africa, in the prolific Berkine Basin. The hydrocarbon system is not simple, with a variety of hydrocarbon phases found in clastic reservoirs of varying ages and properties. A generalized stratigraphic column is shown in Figure 1. The basin forms part of the North African Platform. During the Lower Palaeozoic, there was a clastic dominated depositional system on the north facing Gondwana passive continental margin (Boote etal. 1998). The F1 and F2 sandstones, which are the lower most productive intervals in this part of the basin, represent the most recent part of this succession. The top is marked by the Devonian/Carboniferous unconformity. The Carboniferous Upper RKF Formation and Visean formations consist mainly of claystones with thin fine-grained interbedded sands. The stratigraphy indicates a low energy environment. with water depth increasing through time. During the Hercynian Orogeny, there was progressive collision of Europe and Africa and the North African Palaeozoic platform deformed into a series of broad intra-cratonic sags and foreland basins. Uplift and erosion of the deformed platform resulted in the Hercynian Unconformity marking the top of the Visean shales. This unit can be seen to thin from over 700 m thick in the east to 100 m in the west due to uplift and erosion. Large N E - S W trending
normal faults formed during the formation of intra-cratonic sags and foreland basins related to the orogeny. As rifting waned, the North African platform subsided and was blanketed by an extensive sequence of Triassic fluvial sands. These sediments are characterized by significant vertical and lateral lhcies variations that can be related to the topography developed on the Hercynian Unconformity (Echikh 1998). Sediments were supplied from the major palaeo-highs such as the El Biod Arch to the west of the Ghadames basin (Fig. 2). Deposition took place on the flanks of these palaeo-highs within braided fluvial systems characterized by fining upwards channel sequences. The TAG-I sequence, which is the main reservoir locally, is around 60 m thick and consists of a stacked braided fluvial system. Continued rifting and extension relating to the opening of the Atlantic and Tethyan seaway (Guiraud 1998) led to the development of a series of en echelon normal faults and tilted fault blocks in the NW part of the Gbadames Basin and Berkine Basin. These fault sets can be demonstrated to control thickness and facies variations within the syn-tectonic Triassic sediments. Core from the TAG-I shows a fine to medium grained friable sandstone with porosities of 17-20% and permeabilities between 100 and 450mD. The basins were then buried beneath a passive margin of continental clastic deposits of the Carbonate shales and TAG-S sandstones. The Lower Jurassic sequence is dominated by the $3 unit which is a thick halite sequence formed by the desiccation of a closed basin. There are minor interbedded claystones. The Salifere, containing the S1 and $2, overlies the $3. It consists of halite, but interbedded claystone occurs in higher amounts, Dolomitic and anhydrite beds are also more common suggesting periodic marine influx into the closed basin. Within the basin, hydrocarbons have been found in sandstone within the TAG-I, RKF and Strunian formations. Within the study area only the TAG-I and Strunian F1 and F2
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, 235-248. 0435-4052/04/$15 9 The Geological Society of London 2004.
236
G. PICKER1NG ETAL. There are many possible approaches to making inter-well predictions using well data (geostatistics, neural networks etc.). The method used was an impedance based deterministic approach, using relationships grounded in rock physics, although the interpretation methods include fuzzy set classifications to take into account certain fundamental 'non-uniqueness'. The workflow used follows the schematic diagram shown in Figure 4.
Petrophysical analysis
Fig. 1. Generalized stratigraphic column for the Berkine Basin. sandstones contain significant quantities of hydrocarbons. Since little hydrocarbon bearing Strunian exists within the limited area of the seismic survey that was inverted, this study focuses on the TAG-I. The trapping mechanism is thought to be structural, mainly dipping fault blocks. However, the structural interpretation is not simple, as the contrast in AI between the top of the TAG-I formation and the Carbonate formation is often weak. Seal is provided by either the Jurassic salts or Triassic claystones. Source rocks are the oil-prone shales of the Silurian or Devonian. A large 3D seismic survey was recently acquired over the area, a subset of which forms the data for this study (Fig. 3). Despite difficult surface conditions this has been processed to produce a high quality data set. Even so there are two significant limitations to the data that have a direct effect on using the data for seismic reservoir characterization. One is the reduced frequency content at the reservoir level caused by attenuation of high frequencies by the evaporate/claystone sequence within the Lias Salifere formation. The other problem is the high degree of inner trace muting required to reduce the noise levels within the data. This has direct implications for the creation of useful seismic attributes and is discussed later.
The petrophysical work was carried out to ensure that the analysis of each well reconstructed 'virgin-zone' conditions. This is important as the seismic data represents these 'virgin" conditions and therefore a valid comparison requires that the log data be corrected. The first correction was for borehole washouts, where in the course of drilling the rock surrounding the borehole is removed by the abrasive effect of the passing drilling fluid and cuttings. Critical logs for seismic calibration are sonic and density logs. Of these the density log is often more affected by washouts, as it is a 'pad' tool that relies on contact with the borehole wall. Significant washouts will affect both tools. Correction method depends on the degree of the problem. If the density tool alone is affected, then this can be predicted in the washed out zones by using an appropriate relationship (e.g. Gardner et al. 1974) between the density and the sonic log. In more serious washouts, manual editing may be required. Unfortunately, one of the zones that suffers most from washouts is the Carbonate, due to the relatively soft shales present in this interval. As this lies just above the TAG-I, careful editing was carried out to minimize the effect on final well to seismic calibration. The second correction applied was for invasion of borehole fluids into the formation. As one aim of the study was the prediction of fluid fill, if logs are not reading the correct fluid response then this could jeopardise the final result. The invasion effect on the sonic and density tools depends particularly on the drilling fluid used. An invasion of a heavy water-based mud in an oil leg will affect the tool more than a similar invasion of a lighter oil-based mud. These corrections are made as part of the shear velocity prediction, which is described in the next section. The petrophysical interpretation of the TAG-I sequence is relatively straightforward. As shear velocity logs were not run in any of the wells used for the study, calculation of shear or elastic impedance requires that the shear velocity is estimated.
Shear velocity prediction and impedance calculation Acoustic impedance (AI) is calculated from standard sonic (Vp) and density (p) curves, which are acquired in almost all modem exploration and appraisal wells. Other forms of impedance, such as shear impedance (SI) and elastic impedance (El) (Connolly 1998, 1999) require shear velocity (V0. The definitions of these three impedance measures are given below: A I = p* Vp SI = o*V~
(1)
E1 = Vp ~1+tan z01, Vs ( - 8Ksin2 0 ) , pC 1-4Ksin 20)
where K = ( V s / V p ) 2 and 0 is the incidence angle of the appropriate seismic ray-path. AI and SI are straightforward products of the velocity and density. EI is a less intuitive quantity which was developed to assist in the inversion of AVO angle stacks. Angle stacks that approximate to near normal incidence (an average 0 value of close to zero) can be inverted to AI, and the EI equation can be
LOCATING WELLS USING 3D IN NORTH AFRICA
237
Fig. 2. Sedimentological framework of the Triassic deposits (from Echikh 1998). seen to reduce to the equation for AI in this case. However, stacks which are formed from traces that have non-zero values of 0 will include AVO effects, caused by the production of shear waves from mode conversion of the incident compressional wave. The derivation of an appropriate impedance value to use for these stacks is not trivial, and is covered in Connolly (1999). That it should include a shear wave velocity term is unsurprising, as the contrast in shear properties of the rocks will play a role in determining the amplitude of the modeconverted wave, hence its effect on the recorded compressional wave amplitude. There are various methods of estimating shear velocities. For consolidated sands and shales of the type found in the Triassic and deeper in this area, the Greenberg & Castagna (1992) empirical relationships would be expected to offer a good
% %
/ %
method. This was confirmed by testing the method on a nearby well, outside the area of this study, but with a similar lithological sequence. This well had been logged for shear velocity, and the agreement between predicted and measured shear velocity is good (see Fig. 5). The Greenberg-Castagna relationships, which relate V~ to Vp for both sands and shales, were developed for brine-filled rock. Gas and oil filled rock have different V~- Vp relationships to brine filled rock, therefore the velocity values taken from the sonic logs for the hydrocarbon filled zones must be transformed to their equivalent brine filled case. This is carried out using the Gassmann equation (Mavko et al. 1998). The Gassmann formula gives an expression for the compressional sonic mainly in these terms: 9 9 9 9 9
bulk bulk bulk bulk rock
modulus of the modulus of the modulus of the modulus of the shear modulus,
pore fluid, Kn mineral making up the rock, Kgr dry rock, /
Mavko et al. (1998, p. 172) gives a velocity form of Gassmann's relation as:
7
Vp2pb = Kp + Kmy + 4 / 3 ~
(2)
which is converted to:
2",
Vp2p = KflOp/dpR 2 -+- Kdry q- 4/31a
(3)
~7 or-
,oo/
Vp2pb
%
-
-
Kfl~b/~R 2 -~ Kd~ + 4/31x.
(4)
The right-hand side of this equation is independent of the fluid so assuming no grain fluid interaction we can write that: g p l 2 p l - Kfll~b/qbR 2 = Vp2292 - Kfl2~b/~bR 2
%% oO0
Fig. 3. Study area seismic basemap.
(5)
where Vp~ is the compressional sonic velocity for the rock when containing fluid 1 (m/s); Pl is the bulk density of the rock when containing fluid 1 (kg/m3); Kfl is the bulk modulus of fluid 1 (Pa); Vp2 is the compressional sonic velocity for the rock when containing fluid 2 (m/s); P2 is the bulk density of the rock when
238
G. PICKERING ETAL.
Petrophysics
I ~
Log correction, determination of lithology, porosity and saturations
I
RockPhysics
~""''~
Shear velocity prediction, alternative fluid scenarios
SeismicDataReview
Check seismic data is suitable for prediction, and determine what atlnbutes can be produced
Match synthetics from logs to seismic data to ensure a c c u r a t e inversion and develop velocity model
ii-
4,
4. I
BasicStatisticalAnalysis Determine basic relationships between seismic
I
I
AVOanalysis (AI) or far (El) Investigate if amplitude vades at offsets
4,
~
FuzzyLogicAnalysis Developreservoirparameterclassiflcationscheme
4,
,
~'I
I based on seismicattributes, and incorporate uncertainty I i
I
I
near
attributes and reservoir parameters (e.g.simple regression)
I
I
SeismicInversion
Invert seismic data to appropriate impedance(s)
4, 4, Applicationof FuzzyLogicAnalysis:
I
Apply classification to impedance volume(s) to build likelihood cubes I i
4, Volume Interpretation: ] Body tracking and identification
4, I Identify Drilling Locations {
containing fluid 2 (kg/m3); Kf2 is the bulk modulus of fluid 2 (Pa); dOis the porosity as a fraction; dORis the critical porosity as a fraction. This equation can be used to generate velocity curves for the water case, which can be used to predict the shear velocity using the Greenberg-Castagna (1992) relationship: Vs = Vsh * (0.76969 * Vp - 867.35) + ( 1 - V~h)*(0.80416* Vp - 855.88)
(6)
where Vsh is the shale fraction. With both density and velocity (shear and compressional) for the water case, the Gassmann equation can be used to calculate the velocities and density for the virgin case. Generally, shear impedance is best used as a lithological indicator, with AI combining lithology and fluid content, and El most sensitive to fluid content (Fig. 6). The angle used for the EI calculation in this example was 30 ~ which is the estimate of the average angle of incidence of the far-angle stack data used in
Fig. 4. Schematic of the reservoir prediction workflow.
this study. The shear velocity estimation allows calculation of SI and EI from logs at the well locations. However, these quantities will only be of use in the final predictive goal if they can be generated from the seismic data. Shear impedance can be generated from either true shear reflectivity data, or 'pseudoshear' reflectivity, which is generated by manipulation of the AVO incidence and gradient data (Spratt et al. 1993). As the near angle data were too noisy to be used, gradient/intercept analysis was impossible, so shear reflectivity could not be derived. AI showed some lithological discrimination between sands and shales, but not sufficiently to make it a useful predictive quantity. As EI is better at distinguishing porosity and fluid fill, and AI did not add sigfnificantly to this discrimination, it was determined that EI should be used as the predictive tool. The far-angle data were of good quality and could be inverted, however, for any given measure of reservoir quality, the EI value corresponding to a particular quality value is not unique. Therefore the relationship between reservoir quality and EI was analysed using a fuzzy logic approach.
Fig. 5. Plot showing predicted and measured shear velocity.
LOCATING WELLS USING 3D IN NORTH AFRICA
239
Fig. 6. Comparison of shear, acoustic and elastic impedance as reservoir property indicators.
Fuzzy logic analysis Fuzzy Logic is an extension of conventional Boolean logic (true or false) developed to handle the concept of 'partial truth' (Cuddy 2000). Well analysis can be used to derive lithofacies classes, based on such measurements as net-to-gross, porosity, and hydrocarbon saturation. The seismic inversion data samples can be assigned a fuzzy probability of association with these classes. In this manner, a given seismic impedance sample can then be classified as belonging to the most likely lithofacies case. The technique retains the possibility that a particular lithofacies type can give rise to any impedance value although some are more likely than others. Fuzzy analysis carries the inherent error term through the cabu~ation rather than ignoring or minimizing it. This is attractive because we no longer need to make a concrete decision between possible lithofacies classes which would generally be an over-simplification. Instead we can assign a probability to the quality of prediction giving us increased confidence in the level of certainty (or uncertainty) with which interpretations are made. The lithofacies classes, or membership functions, are derived from the well data and used to calibrate the inversion volume. The normal distribution was used as a first model for each membership function. The fuzzy possibility of measured EI value x belonging to lithofacies f is
F(xf ) = V'-ffe -(x-"' f- /2'rf2
(7)
where/x is the mean and or is the standard deviation. The objective of the analysis is to subdivide the impedance volume in terms of reservoir quality and pore fluid. Therefore three classes, or membership functions, have been used: 9 9
9
oil bearing reservoir water bearing reservoir non-reservoir.
A Vsh (shale volume) cut-off of 50% was used to distinguish reservoir from non-reservoir to derive the membership functions for the first two classes. EI values from the water bearing EI log which corresponded with a Vsh value greater than 50% were used to derive the non-reservoir class. The elastic impedance distributions for the oil bearing classes for each well are similar, with means typically between 71 and 72 EI units. There is an increase in EI values for every well for the water bearing case with means typically lying between 74 and 75 EI units. There is more variation in the spread of mean values for the non-reservoir class with averages between 73 and 78 EI units. The difference in membership
functions between wells was not considered sufficient to warrant the extra complexity of a laterally varying membership function definition. However, this is a simplification as some lateral variation would be expected in a fluvial-deltaic regime. Therefore all the data points for all the wells were analysed collectively to derive a global set of membership functions to be used for calibration of the inverted seismic data (shown in Fig. 7). The normal distribution appears to be a reasonable model although this can be a risky assumption. A QuantileQuantile (or Q - Q ) Plot can be used to check the deviations of the data from the normal distribution. A Q - Q plot is a graph of the observed normal distribution quantile value, and should reveal a straight line with a gradient of 1 if the assumed normal distribution is correct. Figure 7 shows Q - Q Plots for the oil, water and non-reservoir functions along with a line of gradient 1. The oil and water membership functions appear to fit the normal distribution reasonably well, However, a normal distribution assumption is clearly not applicable for the nonreservoir case. Fortunately, fuzzy logic does not require a normal distribution in order to work, and any type of distribution that describes each function can be used. However, it is important that some sort of statistical analysis be performed to confirm that the membership classes are significantly different and not simply derived by under-sampling the same population. The two-sample t-test, is used to compare the sample means and establish confidence intervals. However, this assumes that the parent distributions are all Gaussian. As has been shown by the Q - Q plots, this is not necessarily the case here. Nevertheless, the central limit theorem allows us to take groups of samples and compute their means so that the distribution of means will tend towards a normal distribution. The data were split by well and their means calculated (Table 1). The t-statistic measures the difference between the means of sets of samples, the aim being to prove that the means are significantly different. P(t) is the probability that the means differ solely due to chance, i.e. that the difference is just sampling 'error'. Therefore low P(t) implies a significant difference. The results show that there is a significant difference between the oil vs. water class and also the oil vs. non-reservoir class. However, the relatively high P(t) for the water and nonreservoir comparison provides much less support to the hypothesis that they are from separate groups. As the final membership functions were to be used to classify the inverted seismic data, the membership functions were recalculated after applying a 4 0 - 5 0 Hz high cut filter to the logs, in order to model the reduced resolution available from the seismic data. Figure 8 shows the functions derived for an example well. Filtering of the logs appears to have resulted in
240
G. PICKERING ET AL.
(b) 1.0 r o8~
i
//
(C) 1.0 I 0.8! I 0.6
0 o.s~ 0.4,
]
o.21
0.2~ 0.0
0.0
0.4,
0.2
,
0.4
0.6 0.8 Quantile
1.0
04 08 68 ObservedQuantJle
10
Observed
(d)| 1.0
0.0. . . . 0.0
0.2
0.4
Observed
0.6 0.8 QuanlJle
1.0
/
0.8. 0.6,
~0 . , / LU 0.2' 0.0
00
02
Fig. 7. (a) El distribution curves for each of the three classifications used in the fuzzy logic analysis for all the wells; Q - Q Plots for (b) oil, (c) water and (d) non-reservoir classes.
Table 1. Statistical t-test results for unfiltered E1 values Test 1
Mean Variance Observations
Test 2
Test 3 Oil
Non-Res
Oil
Water
Water
Non-Res
71.72
74.64
74.64
75.54
71.72
75.54
5.04
2.08
2.08
2.14
5.04
2.14
I0
10
I0
10
10
t Stat
3.460
1.382
4.504
P(T >- t) two-tail
0.0028
0.1840
0.0003
tighter distributions. However, some of the wells, as shown in Figure 8, exhibit a bimodal distribution. The cause of the bimodal distribution is a consequence of filtering the data as shown in Figure 8. This shows a simplified model of the TAG-I interval, encompassed by non-reservoir. The raw EI log shows an equal excursion from the non-reservoir trend for both oil bearing units. However, after applying a high cut filter, the two units show different EI values causing the two peaks in the membership functions. In general, thin bed effects may cause an under or over estimate of EI, for high EI or low EI zones respectively. All the data for all the wells were analysed to derive another set of membership functions using the filtered EI logs. The histograms and Q - Q plots are shown in Figure 9. From the grouped data it can be seen that the bimodality of the
10
distributions has been minimized, but the distributions are now increasingly non-normal. However, we will again apply the central limit theorem to use the t-test on the means. The results are shown in Table 2. The results show that there is a significant difference between the oil vs. water class but there is lower confidence in the two classes being separate when the filtered EI logs are used. The oil and non-reservoir classes are also significantly different although this statement is now made with less confidence than before. Confidence in a separation between the water and nonreservoir classes is dramatically lower. The m e m b e r s h i p functions chosen for the final analysis were those derived from the filtered EI logs, as these were expected to be more closely related to the inverted data. As a normal distribution
Fig. 8. (a) El distribution curves based on filtered logs from an example well, (b) an example El log demonstrating the cause of the bimodal distribution after filtering.
LOCATING WELLS USING 3D IN NORTH AFRICA
241
Fig. 9. (a) EI distribution curves for each of the three classifications used in the fuzzy logic analysis for all the wells based on the filtered logs; Q - Q Plots for (b) oil, (c) water and (d) non-reservoir classes derived from the filtered El logs.
Table 2. Statistical t-test results for filtered E1 logs Test 1
Mean Variance Observations
Test 2
Test 3
Oil
Water
Water
Non-Res
73.14
75.01
75.01
74.79
73.14
74.79
3.31
1.46
1.46
1.25
3.3 l
1.25
10
10
10
10
Oil
10
t Stat
2.704
0.422
2.442
P(T >-- t) two-tail
0.0145
0.6780
0.0251
Non-Res
l0
does not fit the data well, membership functions were generated by linear interpolation between bin centres. The curves were then normalised by setting the area under each curve equal to one. The curves have not been scaled by the relative occurrence of each class within the wells, as mentioned earlier in the fuzzy logic mathematics, because we are considering the EI scale of measurement for oil prediction. If no oil was encountered in any of the wells and the membership functions were scaled by the relative occurrence of each class then obviously no oil would be predicted throughout the volume. Different bin sizes have also been tested (Fig. 10). The final choice was a bin size of two EI units because this smoothed the distribution without removing the essential character.
Seismic interpretation and inversion At the target level the seismic data suffer from loss of amplitude and frequency content due to the Liassic evaporite layers preventing penetration of m u c h of the seismic energy. Factors c o n t r o l l i n g the quality o f the seismic data have been investigated by D r u m m o n d (2001), with the main problems being interbed multiples, refractions and statics resulting from some of the largest dunes in the world - up to 450 m high. Tests involving dynamite sources have showed significantly higher resolution than the vibroseis data used for this study, especially at reservoir levels. Five main horizons have been interpreted within the zone of interest (Fig. 11): (1) (2) (3) (4) (5)
Top Salifere Top $3 Top Carbonate Hercynian Unconformity Top F2 Sandstone.
Fig. 10. Normalized membership functions derived from the filtered El logs using (a) bin size of 1 El unit and (b) bin size of 2 EI units.
242
G. PICKERING ET AL.
~ t,~
E ~I.--
~. .o,.~
--
.
.:
"
_
~
~.
~t_-~t - ,,--.
2000.
,lliW
I
1500
I
I
I
I
f E ~-
2000
:_
P
...... o,
2500
|
I
I
i
I
I
I
I
_.
TopSalifere
Top~
2000
TopF2
I
~
2.5 km
~ -
Erosional Truncation
!
250s
Unfortunately, the top TAG-I reflection, which is the top surface of the productive reservoir, could not be distinguished from the overlying Carbonate because the increase in P-wave velocity in the reservoir was counter-acted by a decrease in density. Interpretation for an example inline shows some of the stratigraphic and structural relationships (Fig. 11). For this study one of the main purposes of the interpretation is to provide the structural template for the impedance model on which the inversion of the seismic data is based. The inversion process
. -...... :..
Fig. 11. An interpretation for an example line showing some of the stratigraphic and structuralrelationships.
converts the far offset seismic data into elastic impedance profiles, thus enabling an absolute measurement of elastic impedance throughout a 3D volume. For this procedure to be effective the input data must be properly processed through a sequence that has correctly preserved relative amplitudes (i.e., seismic amplitudes are proportional to impedance contrasts), removed multiples, and reduced noise. However, it should be noted that there is always likely to be some residual multiple and noise energy that will reduce the accuracy of the inverted volume.
LOCATING WELLS USING 3D IN NORTH AFRICA The inversion method used was a model-based inversion 9 This requires the generation of a geological model, which is then convolved with a seismic wavelet to produce a synthetic seismic trace. This trace is then compared to the seismic data and the model is iteratively updated to minimise the error between the synthetic and the real seismic data. To implement this approach both a low frequency model and a seismic wavelet are needed. The 'true' seismic wavelet is that wavelet that, when convolved with the reflectivity sequence of the earth sampled by the seismic trace, produces the final measured trace. A standard procedure is to estimate this wavelet by deconvolving the seismic data at a well location, where the well data can be used to derive the reflectivity sequence. However, there are important limitations to this technique: (a) The assumption that the seismic trace is formed only from the convolution of the reflectivity sequence with the wavelet is only an approximation. Noise, multiples and other effects will also be present 9(b) Logs are not recorded at the same lateral and vertical resolution as seismic data, which will mean that they do not represent the reflectivity sequence 'seen' by the seismic data. Therefore the wavelet is chosen to give the best inversion result (i.e. the minimum error between synthetic and real seismic data as detailed above), rather than the best match between a synthetic seismogram and the seismic trace at the well location. Examples of extracted wavelets and the final wavelet used for the inversion are shown in Figure 12. The final wavelet used was a zero phase wavelet extracted from the seismic data statistically. This wavelet has a broader bandwidth than the extracted ones and also does not contain the phase deviations seen in the extracted wavelet that are often caused by optimizing the synthetic match rather than being typical of the data as a whole. Note that this wavelet lacks frequencies in the 0 - 1 0 Hz range, as is typical of seismic data, and the geological model is
243
therefore required to provide the low-frequency component for an absolute estimation of impedance. The source of the model was a set of low frequency model traces at the wells that were interpolated throughout the volume using picked horizons to provide a stratigraphic constraint. Passing the elastic impedance curve through a sliding-gate mean filter generated the initial low frequency model traces at the wells. Figure 13 shows the elastic impedance log for well B-1 and the result of passing the El curve through three different averaging window lengths of 25, 50 and 75 m. The aim of the model was to retain the essential character of the background El in the logs, and the 50 m averaging window was chosen to best suit this purpose. The horizons interpreted in the earlier part of the study were used to guide the interpolation. Interpolation parameters were defined for each inter-horizon volume and are shown in Figure 14. Between the Hercynian Unconformity and the Top F2 Sandstone a top-lap relationship has been used to guide the interpolation. Although the seismic data shows on-lap of the RKF Sandstone onto the Top F2 Sandstone, the thickness variation within the RKF Sandstone is not significant and the well log correlation at the base of the Upper RKF appears conformable with a 20 m thick shale consistently overlying the F2 Sandstone. There is however significant thickness contrast in the Visean Shales due to the erosional truncation of the Hercynian Unconformity. A top-lap relationship attempts to model this so that where there is thinning of the sediments between the two surfaces it is correctly interpolated by thinning of the Visean Shale due to erosion. The unit between the top Carbonate and the Hercynian Unconformity has been interpreted as being a conformable sequence 9Again this is in contrast to a geological model of Triassic fluvial sands transgressing southwards, where a base-lap relationship may appear more reasonable. However, the complex reservoir architecture of a braided fluvial system means that any form of correlation of individual sand bodies between wells is difficult. A simple
(a)
~)pktm:le
f3~a~e (degrees)
9 0.5
, . . . . . . . . . . . .
-15
-50
~ 9
-75
. ,
0
/ i',.-/. . . . . . . Z5 50
U
5U
~-LeBe'mAmpktl,de"
Panplitude 1.0
r
IUU
--
Phase (e~vg: - 1IS.G0)
Amplitude
.
.
.
.
P~ase (e~rees)
.
0
" A / F
. . . . . .
- 75
~
%,
. . . . . . . . . . . . . . . . . . . . .
-50
-Z5
r
.
.
.
0
.
~
.
.
/ .
.
Z5
.
.
/ 0 .
.
.
.
50
75
5(3
Frequency
--Legend~l~tiale
100
F
(Hz) --
Phase {avg: -23.75)
,~lnpktude
~npkb~ue 1.0
Phase (degrees)
~.0-
!
(c) 1 [I..5
e.5-
o _..~ 4 ~ ,
,AL,~_ 0
oo
?5
.
nine (ms)
Fig. 12. Example wavelets (a) Example extracted wavelet, (b) Averaged extracted wavelet over all wells and (c) statistical zero phase wavelet.
~5
FrleqlJe,'~ y (Hz)
Tmle (m~)
0.5 .
.- 500
~
so
o
Time (ins)
r~
?5
50 Frequency
-- Lege~Al~h I,Ude
--
15
too
(Hz) Phase (avg: 0.B0)
i
244
G. PICKERING ET AL 9
Vertical depth from surface (m)
I
Time (ms}
Tops
*Density
(far
offs(
con"
*P-wave
J
El
El ZSm
2500 -~!
El 50m
J
-j,
.
i ;
El 75m
.
~17oo
i 84 _~.! I
Zf~O 1-
l.ias S~fere
Z700 -+
1750
-;
-,
i
"1800
:..! ~ -- '~, -.! -
-~.q. i ---~ i )
-
-1850
._
2900-
19 900
-_-;, -:
.
. . . .
,--,
...:--
,
3000-
~1950
-~ ~ . . t 84 3100 - - [ - ~ ..... I - Frias - " Carbonate -
'
9
:
:
:.~:
? 3200--1ks
TAG-I
Sandstone__ "
:
.
.
.
.
.
":-
-:
/
.
.
.
.
:
-"i
.....
'
;
i
'
!!:!..i_!
"
,.
. . . . . .
' i ; - !-"
,
_:2_2
"
:
!
- 2000
( . . . .
i - 2050
......
9 . . . . . . . . . :
'!i ~500-t~_T S~ds (upper z
i
,~ .
,
.
:
,,
;
9 ;--;r ' '~ . . . . .
.
~_
. . . . .
; .
! "
-2100
"'!" ' "
i
-Z150
34(10;i
. . . . .
9 ~!
35~t-trunian 3GO0-
~
,
~-
j
-
i
"
~ . . . .
i - 2200
- i -i- "
"
f
I
9
- ~wet
: z
.
"
,
.
.
.
.
:
.
.
.
. .::--: . . . . . . . . . .
ad~e~-:-.:~
..
-2250
i.
.
.
~--
-
.~i
3100-
i - - ' ---
- 2300 -
. . . .
0
~
.
.
O,02fi ktjh~a3
.
.
i
0.050
.
fl
.
.
.
~
.
.
50110 mt's
.
.
,
10000 0
.
.
.
.
i
.
.
.
i
.
:50 101) (m/s)'(kgh~3)
0
50
100
0
Imts)'(ko3m3)
9
i , . 9 .
: !
, i . . . . .
50 {mPs)" Ckgtm3 )
.
, i
,
1O 0
O
9
9
t
50
. . . .
l
100
(nes)'(wgtm3)
Fig. 13. Elastic impedance logs for an example well and the result of passing the El curve through three different averaging window lengths of 25, 50 and 75 m.
conformable interpolation of the low-frequency model which averages the sand and shale interbedded sequence should be sufficient to recover the background trend 9 Once the geological model was built, the model-based inversion was performed for the seismic trace at each well location. Figure 15 compares the inverted trace and target trace for two example wells and the mean absolute error between the two is also plotted. The mean error and the mean absolute error have been calculated for the whole trace and also over the range of the TAG-I interval. Deviation of the mean error from zero is an indication of a low frequency trend error. In order to assess the errors away from the well control, a blind-test was carried out by removing the well from the model built and comparing the final inversion result with the target. When the initial model trace is excluded, the mean error for the whole trace is --- 1 - 2 El units suggesting a minor low frequency trend error may be present away from the well control, although this error is only 3 - 6 % of the total range of El values. The mean absolute error is a measurement of the combined low and high frequency error. Values for the mean absolute error for the whole trace are typically 2 . 5 - 3 . 0 El units and are mostly due to high frequency errors derived from the seismic data. Errors increase slightly, to 2 . 5 - 3 . 5 El units, when the model trace at the well is removed 9Within the TAG-I interval the mean absolute error is generally smaller than for the whole trace
although some wells do show significant differences (e.g. well A in Fig. 15). This is likely to be a result of local noise or remnant multiple. When the model trace at the well is removed this has a limited effect on the difference between the inverted and target traces within the TAG-I interval, giving confidence that the inversion can predict impedance away from the well locations. Figure 16 shows the inverted result for an example inline. The Hercynian U n c o n f o r m i t y is clearer than in the original reflectivity data. The Top TAG-I still cannot be reliably distinguished from the Carbonate Shale, although where the reservoir properties are poor at the top of the TAG-I some separation can be made.
Application
of fuzzy
logic
The chosen membership functions shown in Figure 10 can be used to calculate the membership grade for three classes 9 For instance, an EI value of 73.0 has membership grades of 0.177 for oil, 0.087 for non-reservoir and 0.085 for water. The possibility of oil is given by: P,(oil) = F joil)(Fx(oil) + Fx(water) + Fx(non-resevoir)) (8) where Px(oil) is the possibility of oil and F~ (oil/water/nonreservoir) is the membership grade of oil/water/non-reservoir for a given E1 value, x. For an El value of 73.0 the equation
LOCATING WELLS USING 3D IN NORTH AFRICA
Top Salifere conformable
Top $3 conformable
Top Carbonate conformable
Hercynian Unconformity top lap
Top F2
occurred as one large sheet covering almost all of the area. This is clearly not representative of the known situation at the well locations. Firstly, well B-5 is dry at the TAG-I level. Secondly, if the seismic resolution (about 20 m) is considered, wells A-l, A-2 and B-3 would also be predicted as 'dry' due to the thickness of the oil column being below this resolution. In our comparison at the well locations, from which we derived a level of 47.5%, this resolution issue was not taken into account. Therefore, even on that basis, the level where oil-winners are accepted should be raised. From examination of Figure 17 it can be seen that at possibilities less than - 5 0 % there is still relatively large possibility of each of the other cases, therefore our confidence in predicting oil cases is corresponding low. However, if the threshold is raised to over 50%, then the possibility of oil rises sharply with a corresponding fall in the possibility of either of the other cases. Results for the level set at 50.0% and 52.5% are shown in Figure 18. An examination of Table 3 shows the following key results: (1) (2)
conformable
Fig. 14. Interpolation parameters for each inter-horizon volume.
would give (0.177/(0.177 + 0.087 + 0.085)) = 0.507 or 50.7%. Similarly, the possibilities of water and non-reservoir are estimated as 24.9% and 24.4% respectively. It is important that the word 'possibility' is used instead of probability, as the EI values could represent a rock type that we have not considered. Therefore it is true to say that there is an equal likelihood of this sample being in the oil class as one of the water or non-reservoir classes. It is also true to say that this sample is twice as likely to be oil bearing than water bearing, or oil bearing than non-reservoir, but it is incorrect to say that there is a 50% probability of there being oil. The elastic impedance volume for the TAG-I interval has then been processed through the fuzzy analysis using the interpreted Hercynian Unconformity as the base horizon and a top TAG-I surface derived from isochrons based on the Top Carbonate and Hercynian Unconformity. This TAG-I surface was derived by taking the average of surfaces predicted by mapping either the TAG-I thicknesses or the Carbonate thicknesses from the wells over the area. Oil class winners were those samples that had a higher possibility of oil than of water or non-reservoir. From Figure 17 this can be seen to be any sample with an EI value less than approximately 74.5, corresponding to a possibility of 35% or greater. Classification of the inverted traces at the well locations resulted in 74.3% of the samples falling in the oil class winning category whereas from the log data only - 30% of the samples would be expected to be oil bearing. The EI from the seismic inversion has overpredicted the amount of oil at the well locations. Therefore, some recalibration was required to increase the possibility 'hurdle' for oil class winners, until the volume represented by the oil class winning samples approximately equalled the expected oil volume over the area based on the well data. This was achieved at a level of 47.5%. Three-dimensional body tracking, using the Ikon Science software VolumeFinder, was implemented on the oil-class winners data volume. A confidence level of 47.5% was used to calculate the first scenario. The result was a large 'geobody' that
245
(3)
The results are different but have a similar "hit' rate of 60%. A more systematic error pattern can be seen at 52.5%, where: (a) The A wells are correctly predicted 'dry' at seismic resolution (b) The B wells are mixed (c) The C wells are incorrectly predicted to be dry. The prediction, particularly at 52.5%, is better at predicting 'dry' than it is at predicting oil bearing.
Result (2) is concerning, as the prediction at the well locations, which were all used in the inversion and classification process, should be good. Well B-6 is located close to a fault, and it is likely that this may be reducing the accuracy of the inversion at this point, and there is a body detected just to the east of the well location at both levels. This explanation cannot be applied to the C-wells where the inversion match was good and the wells are not located close to faults. However, at these wells the EI graph for oil-filled rock shows two peaks at El values of - 7 1 and - 7 8 units, which is not typical of the distribution across all the wells (Fig. 10). This then suggests that there is a spatial variation in the reservoir properties of the TAGI unit that cannot be neglected. Further work on developing a spatially variant classification, which may achieve a better prediction accuracy, is clearly necessary for wider application of the technique. Result (3) may, in part, be caused by the thin bed effects described earlier, where low EI values have been overestimated, suggesting water rather than oil-bearing rock. Whether being able to better predict dry wells is advantageous or not deserves some consideration. From an exploration and appraisal point of view, avoiding dry wells is clearly important. However, if opportunities are lost due to the surrender or neglect of acreage that is wrongly assumed to be unprospective then the savings could be outweighed by the 'opportunity cost'. This situation would have to be evaluated on a case-by-case basis. In development or production scenarios, there are different arguments, as the issue of acreage relinquishment may not exist. Successful development wells are clearly critical. The avoidance of dry wells, particularly where there are many possible locations, may be of more benefit than the 'loss" of not drilling a well in a good location due to prediction inaccuracy. However, in any situation, the better a technique is in predicting both good and bad locations, the more value it has to the project. With these results in mind, the original purpose of the study was to assist in the selection of future well locations. As the analysis is known to be doubtful towards the west, the lack of
246
G. PICKERING ET AL.
g :
c
:
,
[7
m
9
~
m
~
4~
V
w
c
u L
5
~ o
I~l~h 0
~o
~ w
g
0 i--
i:
z
_c --
I-
c
~
--
E
=~ ,a
0 "5
g
:
~ 8
;
9
g >ILl
~ ; ' . I :
8
8
8
o
8 O~h
i 84 ii
g
z,
=g 8
0 o
8
8
~
~
8
8
Depth
} g .
g ~.
j.
,
.,.
j
W 0
o
8
8
8
8 Depth
8
8
<
247
LOCATING WELLS USING 3D IN NORTH AFRICA
1500
1500
2000
2000
2500
2500 i
,,
65 Fig. 16. Example inverted inline.
-
i
J
70 75 80 Elastic Impedance Units
i
i
85
would be to the area east of B-l, and the area to the north of B-2 and the west of B-3. Although somewhat more higher risk due to its distance from the well control, would be the southern comer of the study area, where several geobodies have been detected, potentially suggesting a north-south region of better reservoir potential through the area. Clearly, other evidence needs to be considered alongside this study, for example the existence of valid structural closure. However, these results can at least be used to high-grade areas where further detailed study may be fruitful.
0.8 ,.4,..-' . m
ZO.6 or)
g_o.4 0.2 60
(~5
70 Elastic
75
80
85
Impedance
Fig. 17. Possibility of each class for a given EI value using normalised membership functions derived from the filtered El logs.
any geobodies in this area should not be considered as indicating no potential. There is more positive evidence for locations around the A and B areas, where the prediction accuracy is better. Good locations indicated by Figure 18(b)
Summary and conclusions By applying appropriate rock physics to properly conditioned well log data, it is possible to estimate density and velocity (compressional and shear) for the rock mass represented by seismic data. The shear velocity estimation allows the calculation of both shear and elastic impedance, of which the latter can be used to invert far angle stacks - in this study the best quality seismic data available. Given the inherent uncertainty with classifying lithology and fluid fill using impedance values, a fuzzy logic approach has been used. This
Fig. 18. Oil Class Winners with greater than (a) 50% and (b) 52.5% possibility tracked using 26-way Connectivity into Geobodies.
248
G. PICKERING ETAL.
Table 3. Summary table of prediction accuracy at well locations Well Oil bearing zone ~> 20 m Prediction at 50% Prediction at 52.5% A-I
~t
1,,"
A-2 B-1
~
1,I
iI
B-2
~
~
I~
B-3
1,I
~
lJ
B-4 B-5 B-6
1,~
c-1
J,~
C-2
~'
sequence also results in the low-amplitude, low-frequency imaging of the reservoir unit. This makes the interpretation and resolution of the TAG-I interval difficult and ambiguous in areas. Improved seismic data quality would inevitably improve the final results. Failure to correctly predict oil for wells C-1 and C-2 also suggests a more serious error in the calibration of the membership functions. These spatial variations in the membership functions are an area of ongoing investigation. The authors would like to thank Burlington Resources (Energy Services) Inc., Sonatrach and Talisman (Algeria) for their permission to publish this paper. Thanks are also due to those other members of the teams at Burlington and Troy-Ikoda that have assisted with this work: S. Cherry, T. Darling, I. Demaerschalk and R. Found. In addition, I would like to thank P. Connolly and M. Payne for their reviews, which have made a significant contribution to the final paper.
References ~,,
encapsulates both the uncertainty in prediction, and the nonuniqueness, where any given impedance value may represent different types of rock. In order to use this approach, lithofacies classes have been derived using the filtered elastic impedance logs. There is a measurable difference between the oil- and water-bearing classes, the oil and non-reservoir classes, and the water and non-reservoir classes. The latter, however, are not distinguishable on a statistical basis. Given the errors inherent in seismic inversion, the result when using the inverted seismic, will not be as good as indicated by the well analysis. However, it will still be useable for the prediction of pore fluids, albeit at a lower confidence than the well analysis would suggest. Seismic inversion of the far angle seismic migration was successfully performed and a 3D volume of elastic impedance has been generated. Error analysis provides some confidence in the inverted result away from the well control. After completion of the inversion, the fuzzy logic analysis was then applied to the inverted data. Fuzzy logic does an adequate job of predicting known oil cases and some confidence can be placed in its predictive power between well control. Interpretation of the reservoir interval is critical in order to exclude non TAG-I intervals from the fuzzy analysis. Confidence in the final prediction is therefore reduced in the north and south where there are no wells to constrain the interpretation. The study has identified several prospective areas with the main target for drilling located between wells B-1 and B-6, within a large rotated fault block bounded by a major N E - S W trending fault. The quality of the prediction is governed predominantly by the quality of the seismic data. The poor signal-to-noise ratio is evident from the poor correlation of the seismic to synthetic trace data and hence introduces errors into the inversion. The poor penetration of the seismic signal through the evaporite
BOOTE, D. R. D., CLARK-LOWES, D. D. & TRAUT, M. W. 1998. Palaeozoic petroleum systems of North Africa. In: MACGREGOR, D. S., MOODY, R. T. J. & CLARKE-LOWES,D. D. (eds) Petroleum Geology of Noah Africa. Geological Society, London, Special Publications, 132, 7-68. CONNOLLY,P. Calibration and inversion of non-zero offset seismic data, SEG 1998, Expanded Abstracts. CONNOLLY, P. 1999. Elastic Impedance. The Leading Edge, April, 438-452. CUDDY, S. J. 2000. Lithofacies and permeability prediction from electrical logs using fuzzy logic. SPE Reservoir Evaluation and Engineering, 3(4), August. DRUMMOND, J. 2001. Adapting to noisy 3-D data: Enhancing Algerian giant field development through strategic planning of 3-D seismic in Berkine Basin. The Leading Edge, July, 718-728. ECmKH, E. 1998. Geology and hydrocarbon occurrences in the Ghadames Basin, Algeria, Tunisia, Libya. In: MACGREGOR,D. S., MOODY, R. T. J. & CLARKE-LOWES,D. D. (eds) Petroleum Geology of Noah Africa. Geological Society, London, Special Publications, 132, 109-129. GARDNER, G. H. F., GARDNER, L. W. & GREGORY, A. R. 1974. Formation velocity and density - The diagnostic basics for stratigraphic traps. Geophysics, 39, 770-780. GREENBERG, M. L. & CASTAGNA, J. P. 1992. Shear-wave velocity estimation in porous rocks; Theoretical formulation, preliminary verification and applications. Geophysical Prospecting, 40, 195-209. GUIRAUD, R. 1998. Mesozoic rifting and basin inversion along the northern African Tethyan margin: an overview. In: MACGREGOR, D. S., MOODY, R. T. J. & CLARKE-LOWES,D. D. (eds) Petroleum Geology of Noah Africa. Geological Society, London, Special Publications, 132, 217-229. MAVKO, G., MUKERJI, T. DVORKIN, J. The Rock Physics Handbook. Cambridge University Press. SPRATT, R. S., GOlNS, N. R. & FITCH, T. J. 1993. Pseudo-shear - The analysis of AVO. In." CASTAGNA,J. P. & BACKUS,M. (eds) Offset Dependent Reflectivi~ - Theory and Practice of A VO Analysis. Society of Exploration Geophysicists, Tulsa, Oklahoma, Invest. Geophys., 8, 37-56.
Use of 3D visualization techniques to unravel complex fault patterns for production planning: Njord field, Halten Terrace, Norway CHRIS JAN
DART
1'7, I A N
C. RIVENAES
CLOKE
1, C E C I L I E
2'3
/kGE
HERDLEVA~R
OTTERLEI
l, E I V I N D
j'4, D O M I N I Q U E JOHNSEN
GILLARD
l & ANDERS
EKERN
5 1'6
JNorsk Hydro, Sandsliveien 90, 5020 Bergen, Norway 2ExxonMobil International Limited, St. Catherines House, 2 Kingsway, London, UK 3present address: ExxonMobil Exploration Company, 222 Benmar, Houston, TX 77067, USA 4present address: Statoil, Postboks 7200, 5020 Bergen, Norway 5ExxonMobil Upstream Research Company, 3120 Buffalo Speedway, Houston, TX 77098, USA 6present address: Schlumberger Information Technology Services Norge AS, Aslakveien 14A, 0753 Oslo, Norway 7present address: Ruhrgas Norge AS, Kongsg&rdbakken I, 4003 Stavanger, Norway
Abstract: In this paper we demonstrate the benefits of 3D seismic visualization techniques for fault interpretation where the
structural geology is complex and the seismic data quality is often poor. Production from the Njord field is strongly influenced by a complex pattern of segmented and linked extensional faults. The current interpretation of the well test data and production history is that the faults form barriers to fluid flow, reducing oil production, and limiting effective gas injection and pressure support. Drilling results demonstrate that geometrical uncertainties remain in the seismic interpretation. An additional problem is that currently available commercial reservoir modelling technology cannot easily handle a very complex fault pattern, making simulation grid construction difficult. Accurate well placement and production forecasting requires that fault geometries and properties are suitably represented in the reservoir simulation model. 3D visualization of depth-scaled volumes and depth-converted interpretations helped to decide how to best simplify the fault geometry for simulation, and compare automatically generated geological model components against seismic interpretations and data. The reservoir simulation model runs resulted in the identification of a number of well targets. 3D visualization techniques were then used to predict faults and structures that the proposed well trajectories may intersect.
This paper uses some of the latest 3D seismic interpretation, visualization, geological and simulation model building software to execute a typical workflow for production forecasting and well planning on an operative field. The workflow may be typical, but the field is not. The Njord field has one of the most complicated fault patterns on the entire Norwegian continental shelf and presents geometrical challenges that not only demonstrate the necessity of today's visualization techniques, but also define the improvements and additional functionality that will be required in the software of the future. This paper draws on work from the staff of the field's operating company, Norsk Hydro, and one of its partners, ExxonMobil. The work has been carried out both within the operative units and research/technology facilities of both companies. This paper uses movies as well as conventional figures. Most of the illustrations in the text are frames extracted from the movies. For clarity these are referred to as movies in the text, and not figures. The movies are numbered from 1-19 and are stored on the CD found in the back pocket of this volume. There is one movie illustration in the text for each movie on the CD. Because many of the illustrations are perspective views, the scale bars will be inaccurate, and not apply to the entire figure. The scale bars are meant to be illustrative only, and give an impression of the size of the particular structure/feature being discussed. The text in the paper is also organized around the movies, and the intention is that the reader plays the relevant movie while he/she reads the relevant section in the article. The Njord field (Koch & Heum 1995; Lilleng & Gundesr 1997) is about 6 km in diameter and located around 130 km NW of the operations base in Kristiansund (Fig. 1). It is operated by Norsk Hydro with ExxonMobil, Petoro, Paladin, ConocoPhillips, OER and Gaz de France as partners. The reservoir comprises a 120m thick Lower Jurassic Tilje Fm interbedded sandstones and shales that were deposited in a tidal/estuarine
setting (Dalland et al. 1988). The field has been involved in rifting phases at various times throughout the Triassic, Jurassic and early Cretaceous (Blystad et al. 1995). It is located in the hangingwall of the NW dipping master fault that defines the southeastern margin of the Halten Terrace. This fault has a complex geometry adjacent to the field and has been variously interpreted as either a ramp-flat-ramp (Osmundsen et al. 2002) or alternatively as part of a breached relay system. This geometrical complexity has resulted in the complicated and dense pattern of N E - S W and N - S trending faults observed at Njord (Fig. 2). Production on Njord began in 1997 and to date (early 2002) 14 production and injection wells have been drilled. Some faults with throws of < 25 m are sealing on a production timescale as a result of the high percentage of shale/clay content within the reservoir interval. This behaviour is especially observed in the Central Area of the field. In order to combat compartmentalization by sealing faults, the production wells have been drilled with U- and W-shaped geometries, designed to penetrate as many isolated fault compartments as possible, and to increase the chance of hitting the reservoir in locations where the reservoir has not been removed by displacement along faults. At the end of 2000 relatively poor new well performance resulted in a drilling stop and the establishment of a one-year Increased Oil Recovery (IOR) project to better define new well targets. The new drilling start-up is planned for 2002/2003. The IOR project was focused on the most complex part of the field, the Central Area. The existing seismic interpretation was edited to make it more consistent and easier to model, and a new geological and reservoir simulation model were constructed. Two new Central Area horizontal well locations were proposed at the end of 2001. 3D visualization was used as a through-going theme in all the work that was done (Bond 2001 also describes similar
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. 249-261, 0435-4052/04/$15 9 The Geological Society of London 2004.
250
C. DART E T A L . 6000,
7"00"
8"00 "
-
_
_
.
O~ .
10"00"
.
Heidrun Smerbukk
-
-
-
i~
t
' : ' r ~ ;
Trestakk
~ /1 *" J
Midgard a
. . . .
" . . . . . . .
Tyrihans Mikkel
TtoodtK~n
/
~a'/st/m~sur~
0
1 ~
common database structure. This is because reformatting problems take too much time to solve, and duplication of data creates confusion amongst a team of interpreters. Norsk Hydro uses GeoFrame ~' as a common database at asset project level. Interpretation data generated in Charisma | or GeoViz | (e.g. a fault stick, i.e. a set of coordinates that define a fault profile on a seismic line) are read from the database, modified, and then instantly returned to the database. Here they are stored, and can be fetched by the next geophysicist who is reworking the area, perhaps in the light of a new well or simulation model result. We have also added another workflow loop through Paradigm's Explorer :'~ software where depth conversion is carried out (Fig. 3). Here interpreted horizons and fault information, together with seismic cubes and seismic attribute cubes are scaled to depth using a velocity cube built from seismic velocities and well data. The use of depth converted seismic allows a direct link between the seismic data, and the model building/well planning process. Seismic, time maps and major faults are depth-scaled. Once the data are in depth then interpretation detail, such as the addition of minor faults and modifications to the horizon and major faults, are carried out. The final interpretation is used as the starting point for the RMS :"~ geological model building. Initial test RMS g' models highlighted fault/horizon inconsistency problems that needed to be edited in GeoViz ~, so an iterative process was required before the final RMS ~ input was produced.
i
Fig. 1. The location of the Njord field in relation to the Norwegian coastline.
working practices). Previous work on the field had shown that 2D modelling and mapping techniques involving vertical extrapolation down from a top reservoir map gave a very erroneous picture of the field because of the complex fault pattern (X-, Y- and inverted Y-fault intersections in crosssection) that often involved shallowly dipping ( < 4 5 ~) faults. Seismic interpretation was conducted using GeoViz ", Schlumberger's 3D interpretation and visualization software for Charisma '~. This was a new approach on the field. Previously interpretation had been done in Charisma :~ and then occasionally visualized in 3D in GeoViz ~ afterwards. Geological and simulation model building was done using Roxar's Reservoir Modelling System (RMS :*) 3D modelling and visualization package. RMS | functioned as a digital meeting place for geophysicists, geologists and reservoir engineers. ExxonMobil had produced a parallel fault interpretation using Schlumberger's IESX | GeoViz | and Framework3D :'< (Schlumberger's equivalent to RMS| and Paradigm's VoxelGeo ~ 3D visualization package. A procedure was established to import the ExxonMobil work into Norsk Hydro's Charisma GeoViz x project. This gave both operator and partner the possibility to directly compare and discuss alternative interpretations at the workstation, which resulted in a significant improvement in communication and mutual understanding. Another aspect of the paper is that movies are used as a presentation medium. 2D figures of 3D visualizations often contain too much overlapping information to be easily understood. Here we have used motion and progressive blending from one display to another, in an attempt to add perception and meaning to the presentation material.
Workflow It is our experience that a successful workflow is one that involves as few pieces of software as possible, and a central
D e p t h slice movies The first sequence of movies focus on visualization of the seismic and existing interpretation data in a horizontal plane. Movie 1 shows the Njord Central Area top reservoir map colour coded with an Edge Enhancement attribute (a function related to abrupt changes in the surface elevation); dark blue areas have high edge attribute values, and the major faults present in the existing interpretation are clearly visible as high edge anomalies. One can also see the existing U- and W-shaped production wells (three in total) and one vertical exploration well. All of these show geological markers (coloured diamonds) corresponding to top reservoir (pale blue) and the units just above. Note that as you follow the W-shaped production wells down from top right, they disappear beneath the top reservoir surface, only to reappear and disappear at their central apex, and finally reappear again as the top reservoir is penetrated for the final time at TD, where the well name is written. As the movie plays it zooms in on a horst structure which will be a focus of the rest of the paper. The next step is to examine the seismic data in conjunction with the top reservoir map. In Movie 2 the amplitude seismic data (red to black colour table) are rendered partially transparent and dip-steered discontinuity data (green to blue colour table) have been blended into the display. The discontinuity data are an attribute cube that has been generated by ExxonMobils Research Centre in Houston (ExxonMobil patented method similar to coherence; Marfurt et al. 1998). The method examines the similarity or dissimilarity of adjacent seismic traces and assigns a discontinuity value (see also Carter & Lines 2001; Chopra 2001: Worrel 2001, for similar methods). Geological faults cause offsets in geological strata leading to displacements in the character of adjacent seismic traces and therefore high discontinuity. Discontinuity cubes are therefore potentially useful for fault recognition and the interpretation of fault linkage in structurally complex areas. Dip-steering (see also Marfurt et al. 1999) of the discontinuity attribute is a refinement of the method that is intended to remove the effects that unfaulted dipping layers would have on the similarity of adjacent traces.
3D VISUALIZATION, NJORD FIELD
251
7130000
64~15'
Fig. 2. Structure map of the Njord field. Wells are shown by radial pattern of lines centered on the Njord platform. Colours indicate the subdivision of the Njord field into structural domains and into reservoir and non-reservoir regions. The central area fault blocks have the prefix CC.
7125000
410000
Only the most discontinuous portions (high values) of the discontinuity cube are displayed (Movie 2). As the movie pans down through the seismic cube, and over the top of the top reservoir map, note how the discontinuity data form 'walls' that correspond to the major faults. Also the intersection line of the top reservoir surface and the horizontal top surface of the seismic slab are shown in bright green. Producing a mental picture of the relationship between horizontal slices and mapped surfaces from 2D diagrams or a flat workstation screen requires experience and training. However, this relationship becomes immediately apparent using 3D visualization. Once the top reservoir map is removed (Movie 3) we can see the seismic data within the reservoir. By using the intersection lines between the top reservoir map and the top of the seismic slab (bright green), and a near bottom reservoir map and the top of the seismic slab (bright pink), we can quickly isolate and focus our attention on the portion of the seismic data that
I Charisma /Geoviz ] It GeoFrame I
~
Paradigm [ Explorer] Fig. 3. The software workflow used in this article.
415000
corresponds to the reservoir itself. The reservoir within the horst structure (introduced in Movie 1) lies in the central portion of the seismic cube that the slab pans through. Note that the horst is much wider at near base reservoir, than at top reservoir, level. Therefore the first impression provided by the top map gives an underestimation of the volumetric significance of the horst. A more correct impression is achieved by using 3D visualization. Ideally one would wish to use the discontinuity data in a horizontal plane to define fault traces. The major faults are clearly associated with linear discontinuity anomalies that resemble fault patterns seen in well-exposed horizontal planes through suitable outcrop analogues (e.g. Kilve Beach, Somerset, UK; Dart et al. 1995). The discontinuity data can be used to highlight minor faults within the fault blocks that are already interpreted. A useful experiment is to progressively blend in more and more of the less discontinuous data in an attempt to locate subtle faults. This is shown in Movie 4. The reader is encouraged to stop the animation when the most convincing fault pattern appears, but before the fault pattern is overwhelmed by the less discontinuous data. The choice of the amount of discontinuity blend for optimum fault trace recognition is obviously a subjective choice for the interpreter. The level of detail resolved is also the result of the operator length used in the discontinuity cube preparation. The operator length is the time interval of the sliding window over which the discontinuity attribute is calculated. A lower operator length will result in more structural resolution, but only at the expense of the enhancement of unwanted seismic noise. In the Njord case an operator length of 80 ms was found to be optimal.
252
C. DART ET AL.
Movie 1. Depth map of the top reservoir horizon in the Njord Central Area draped with an edge enhancement attribute. Dark blues correspond to high gradients and can be used as a fault indicator. Depths in the Central Area range from 2500- 3100 m.
Movie 2. A horizontal slab of blended seismic amplitude (red-black) and discontinuity (green-blue) data, shown against the top reservoir depth map.
Movie 3. As Movie 2, but with the top reservoir depth map removed.
3D VISUALIZATION. NJORD FIELD
253
Movie 4. As Movie 3, but with a variable blend of discontinuity data.
Movie 5 shows the final edited fault sticks bounding the horst that were used as input to the RMS | geological model. Faults are interpreted as sticks (bright blue, yellow, pink and orange) on vertical sections (both inlines and cross lines at 90 degrees to each other) through the seismic data (see also Movie 7). As Movie 5 is played note that care has been made to continue the fault interpretation right up to the intersection (branch) line between two faults. The sticks are regularly spaced and consistent from line to line. This provides a better geological modelling result. Note also that the linear discontinuity anomalies (blue to green colour table) and the interpreted faults are not in exactly the same place (a topic explained in Movie 6).
Inline movies Vertical sections through the data are the next step in expanding our understanding of the structure in three dimensions. In Movie 6 a vertical slab of blended amplitude (red to black colour table) and discontinuity data (blue to green colour table) are shown. The movie pans through the horst structure, and the fault sticks of the bounding faults flash on and off, as an inline where
Movie 5. As Movie 3, but together with interpreted fault sticks that bound a major horst.
a fault stick has been interpreted passes by. An existing production well is imaged at the same time. Notice that the discontinuity data have a vertical striped character that poorly represents the relatively shallow dipping major faults. An explanation is that discontinuities are best highlighted at the strongest reflectors. However, discontinuity is calculated over a vertical window (in this case 80ms), the length of which is dependent on the frequency content of the seismic data. This can result in a series of stripes hanging below the positions where the strongest reflectors are truncated by the dipping faults. The result is a series of vertical curtains, the positions of which need not necessarily correspond to the position of the fault on a particular horizontal slice (see also Movie 5). The discontinuity data are therefore useful if the 3D effect is understood and faults are mapped using both horizontal and vertical slices iteratively. Such an approach requires interpretation using a 3D visualization tool such as GeoViz ~ (or VoxelGeo ~:). In Movie 7 the process of adding detail to the interpretation by including a minor fault within the horst is illustrated. A disturbance within the seismic data that could be interpreted as a minor fault is found in the immediate footwall to the yellow horst bounding fault. As the movie plays, a fault slick
254
C. DART ET AL.
Movie 6. A vertical slab of blended seismic amplitude
(red-black) and discontinuity (green-blue) data shown together with interpreted fault sticks that bound a major horst. The movie pans through vertical sections across the entire study area.
(light blue) is placed on the displayed inline, and then the display is moved away from us by one inline. A new stick is drawn and the display is moved another inline away from us. Then the whole display is rotated to check that the two sticks are consistent with each other. The sticks are then lined up so that we can sight along the strike of the fault and look for a disturbance on the as yet uninterpreted inline. Once we feel confident, a third stick is added. Finally we rotate all three sticks to see how they look. The horst bounding faults are visualized at the same time, so that as we interpret we can ensure geometrical consistency with other thults, as well as with the seismic data. Notice also that the introduction of the new fault is not consistent with the horizon interpretation (bright green and pink). This will need to be modified bel-bre input to the geological modelling in RMS :R;
Sculpted cube movies Movie 8 is the first in a sequence of movies that illustrates the use of sculpted volumes. A sculpted volume is produced by cutting a cube of data by a surface and then removing all the data
situated on one side of the cut. For example the character of the seismic data at the top of the reservoir can be examined by cutting the seismic data cube at the top reservoir horizon and removing all the data above the top reservoir surface. In Movie 8 the existing near-base reservoir surface is shown colour coded with an edge attribute (as described in Movie 1). The upper surface of the seismic cube (a blend of amplitude and discontinuity data as described in Movie 2), which sits on top of the near-base reservoir surface, corresponds to the top reservoir surface. The seismic cube has been sculpted such that all of the seismic data that lies above the top reservoir surface have been removed. Effectively the upper surface of the cube shows the seismic data at the position of the top reservoir depth map. As the movie plays we zoom in on our horst structure. The major horst bounding faults, visible as offsets in the upper surface of the seismic cube, correlate quite well with discontinuity anomalies. The fault sticks (now bright green) that were interpreted in Movie 7 protrude above the upper surface of the sculpted cube. This is because they were continued above the top reservoir horizon when they were interpreted, It is important to check that the top reservoir marker (cyan) in the displayed production well coincides with the top of
Movie 7. As Movie 6, but also showing the interpretation
of a minor intra-horst fault.
3D VISUALIZATION. NJORD FIELD
Sculpted volume Production \
9
Movie 8. A sculpted volume (3D block diagram) of
blended seismic amplitude (red-black) and discontinuity (green-blue) data. The volume is shown together with a depth map of the near base reservoir in the Njord Central Area draped with a edge enhancement attribute.
E ] oo :]-"
the sculpted cube, indicating that the well and seismic data have been correctly calibrated during the depth scaling procedure. Geological modelling of fault planes can either use fault sticks or fault boundaries (the fault/horizon intersection lines) as input. Faults boundaries can also be referred to as fault polygons. In this case we chose to model the larger horst bounding faults using fault sticks as input, while the more minor internal faults are modelled using fault mid-lines (a type of fault boundary that lies midway between the upthrown and downtbrown fault boundary, i.e. midway between the footwall and hangingwall cut-off lines). As Movie 9 plays we first insert a fault boundary (white and then dark blue) that corresponds to the minor fault that was interpreted using fault sticks in Movie 7. We then blend in more of the discontinuity data in an attempt to find other minor faults. A possible fault boundary is then interpreted on a linear discontinuity anomaly about-half way across the horst (again white and then dark blue). The picture then changes to a display of the fault boundaries (now pale blue) superimposed on the top reservoir surface (again colour coded with the edge attribute as described in Movie 1). Further investigation of the second fault boundary on vertical inline slices revealed that a through-going fault could not be identified at this location and this proposed fault boundary was subsequently rejected.
Movie 9. A close up of the sculpted volume shown in Movie 8. Fault boundaries are interpreted on the upper surface (top reservoir map) of the sculpted volume.
~.
9
.
255
NW ,,,
'~,',,~'
w,o,
500 m
-
E~.;
'
'
Base reservoir edge attribute map
Vertical slices or slabs of seismic data can also be extracted from sculpted seismic cubes, In Movie 10 we use tbe same sculpted cube which is shown in Movie 8. We pan a slab of data towards us across the horst structure in order to focus in on the seismic data that lies within the reservoir. Good reflection continuity is observed just above the near base reservoir surface, and a trough (red) was tracked automatically using the automatic area tracking functionality (i.e. ASAP in Charisma:R:t within the GeoViz:" 3D visualization package. The resulting track is displayed as a brown horizon that fits the trough (red amplitude) well and highlights small offsets in the reflector that may correspond to minor faults. Movie 11 shows the automatically tracked intra-reservoir horizon surface that was generated in Movie 10. A detailed colour table is used to highlight small differences in elevation that may be associated minor faults. We are most interested in laterally continuous edges defining elevations changes that can be verified as fault offsets by panning through vertical inline slices. As the movie plays, the intra-reservoir horizon surface is rotated. Next, a set of inline interpreted (method described in Movie 7) fault sticks (bright green) is superimposed on the image. These fault sticks correspond to a laterally continuous elevation change edge, which is followed by more rotations and
256
C. DART E T A L .
Movie 10. A vertical slab of seismic data cut through the sculpted volume shown in Movie 8. The brown interpreted horizon just above the near base reservoir surface is automatically tracked. then a new automatically tracked intra-reservoir horizon that lies higher up in the reservoir is superimposed. A laterally continuous elevation change edge can also be observed on this new horizon, and can be seen to correspond to the interpreted fault sticks, confirming the interpretation. Movie 12 is the last in the sculpted volume sequence and shows a compilation of some of the data assembled for the horst so far. The two automatically tracked intra-reservoir horizons (Movies 10 & 11) are shown together with the modified horst bounding faults interpreted as sticks (Movie 5). The new intrahorst minor faults (Movies 7 & l 1) are shown as sticks (bright green), and the one to the left in the first frame also shows its corresponding fault boundary (pale blue) (Movie 9). Note that this boundary is interpreted at top reservoir level and therefore sits above the uppermost displayed intra-reservoir horizon. The proposed pale blue boundary (also Movie 9) seen in the centre of the first frame could not be related to a set of fault sticks and was excluded from the final model.
had used Schlumberger's Framework3D ~ geological modelling package to convert their fault sticks into gridded surfaces. A method was established to import these gridded surfaces to Norsk Hydro's GeoViz ~' project so that they could be visualized alongside Norsk Hydro's fault sticks. In this way the Norsk Hydro and ExxonMobil interpretations could be compared and contrasted with each other and, against the seismic and discontinuity data, in a 3D visualization environment. This helped to iron out disagreements and build confidence in each other's work. In Movie 13, two of ExxonMobil's movie 2 gridded fault surfaces (dark greens) are compared with Norsk Hydro's fault sticks (paler green and red). A depthslice is pivoted backwards and forwards about a horizontal axis and a horizontal slab of blended amplitude and discontinuity data accumulates around the faults and illustrates that the two interpretations are in broad agreement.
Integration of borehole and seismic data Partner data share Norsk Hydro's partner on the Njord field, ExxonMobil. conducted a parallel interpretation of the fault pattern. They
3D visualization also provides excellent opportunities to combine well data and seismic data. This is particularly important in areas with many intersecting faults, as it can often be difficult to understand which seismic mapped fault has
Movie 11. An automatically tracked surface displayed with a finely divided colour table to highlight subtle edges associated with minor faulting.
3D VISUALIZATION, NJORD FIELD
257
Movie 12. The automatically tracked surface shown in Movie 11 displayed together with interpreted fault sticks that bound a major horst.
been responsible for e.g. faulting out parts of the reservoir at the well location. In Movie 14 a vertical section (random line) has been extracted from the seismic data along a well path. Geological markers are shown along the well path (coloured diamonds). The near-base reservoir map (colour coded for depth; red, high; blue, low), and a selection of faults as sticks (pink, green and brown), are also shown. As the movie is played we zoom in on the horst. Within the horst itself (between the pink and green faults), in the first down-going reservoir well penetration, a complete stratigraphic column with all top zone markers is observed. In the following upgoing section (to the right), however, many of the markers (i.e. red. green, purple) corresponding to the central portion of the reservoir are missing. It can be debated whether it is the green or the brown fault that is responsible for the stratigraphic omission, or whether it is another fault that has not been interpreted, or a sub-seismic fault that needs to be introduced to explain the well results. In this case 3D visualization presents all the available data together, making it easier to weigh up the alternatives.
Movie 13. A horizontal slab (depth slice) of blended seismic amplitude (red-black) and discontinuity (greenblue) data shown together with comparable fault sticks interpreted by Norsk Hydro and gridded fault planes interpreted by ExxonMobil. Data are viewed obliquely from above.
Geological model building Movie 15 is the first in a sequence that deals with the final stages of the workfiow from geological and simulation model building to well planning. This work was conducted in Roxar's modelling package RMS x and uses GeoViz '~ interpreted data as input (Fig. 31. The fault modelling modules in RMS ~' are used to grid the fault sticks, interpreted for the major faults, into triangulated surfaces (e.g. purple surface), Care is taken to ensure that the surfaces are extended along-strike such that they pass through intersecting faults. It is geologically reasonable to extended the mapped portions of faults several hundred metres along-strike to allow for the low throw tip regions that lie below seismic resolution (e.g. Pickering et al. 1997). Where the faults meet, intersection (branch) lines are calculated and visualized (Movie 15). In the example the branch lines are the sub-vertical pale purple lines superimposed on the purple fault plane. In addition, the intersection lines between the horizons and the triangulated surfaces are calculated. Each horizon/fault intersection has an upthrown side and down-thrown side pair of lines
258
C. DART ET AL.
Movie 14. A W-shaped well with geological markers shown along a random line taken from the 3D amplitude cube. The depth map (rainbow colour table) is the near base reservoir horizon.
(footwall and hangingwall cut-off lines respectively: note that these lines are called fault lines by Roxar). These lines are used as the principal input for the representation of the fault offsets in the simulation grid (see Movie 17 later). In Movie 15 the near-base reservoir horizon is shown in a rainbow colour bar that corresponds to depth (red, high: blue, low), and top reservoir is shown as a partially transparent pale blue so that we can see the structures below top reservoir. As we rotate the model, a red fault appears which shows the triangles that are used to build the fault surface. Several more branch lines, and footwall and hangingwall cut-off lines, are shown as well as the intersection of these. Where the cut-off lines for different faults meet at branch lines, rigorous geometrical rules have to be obeyed. Once the seismically mapped major faults and horizons have been modelled, the intrareservoir geo-horizons defined by the geologist's well zonation can be added. Cut-off lines are then calculated for these. Movie 16 shows how import of seismic data to RMS ~ can be used to quality control the consistency of the calculated fault cut-off lines. This adds confidence to the results obtained. In the movie we pan across the horst block in RMS '~ and compare the triangulated fault surfaces (purple and redJ, and
their attendant cut-off lines (pale purple and bright red), to the original seismic data.
Simulation model building RMS :": has been used to construct the 3D simulation grid that is used for flow simulation in Eclipse ~'~. The mathematics behind the flow calculations requires that the simulation grid is as orthogonal as possible, i.e. that the grid cells are as cubic as possible. This geometry is at odds with the geometry of the fault blocks in reality, and therefore simulation girds are always a geometrical compromise. In Movie 17 we show a test simulation grid that is constructed between the top and near base reservoir surfaces. The locations of the fault offsets in the simulation grid appear to fit well with the locations of the calculated horizon/fault cut-off lines at the top reservoir horizon. Geometrical consistency between the geological model and the flow simulation model can be evaluated in detail by panning a section through the simulation model across elements selected from the geological model, for example the faults. In Movie 18 two faults are shown (the same faults in Movie 16). Note that
Movie 15. Geophysical input to the RMS geological model. Fault sticks are gridded to make fault planes. Branch lines are generated at fault intersections and cut-off lines are generated at horizon fault intersections. The top reservoir horizon is shown in translucent blue. and the near-base reservoir horizon is shown in a rainbow colour table.
3D VISUALIZATION. NJORD FIELD
Movie 16. Comparison of automatically generated horizon fault cut-off lines with the input seismic cube using RMS. Pale purple solid lines are footwall cut-off and branch lines; pale purple dashed lines are hangingwall cut-off lines. View is from the north.
Movie 17. Generation of a reservoir simulation grid using horizon fault cut-off lines to describe fault position and geometry.
Movie 18. Quality control of reservoir simulation grid against gridded fault planes.
259
260
C. DART E T A L .
there are discrepancies between the geological model fault surfaces and the offsets in the simulation grid, both in the location and dip of the faults. This approach can be used to compare the geological input to different methods of flow simulation grid construction, and to decide which methods should be employed.
Well planning Well planning on a structurally complex field such as Njord requires the integration of many different types of information. The well location needs to satisfy at least the following requirements: 9 it should penetrate the reservoir in an area where the most permeable zones are not omitted by faults: 9 because the faults are sealing it should penetrate as many different compartments as possible: 9 it should penetrate the reservoir above the oil/water contact: 9 it should penetrate the reservoir in areas that have not been pressure depleted by the existing production wells. In Movie 19 a selection of elements is shown to illustrate the principle. An oil-water contact is shown in dark blue and the near base reservoir surface is shown with a rainbow colour table that corresponds to depth (red, high; blue, low). The existing production wells are shown in blue and a proposed new well trajectory is shown in pink. The volumes of the reservoir that have been depleted by the existing production wells are taken from the flow simulation model and are shown as grey cubes. It is clear that the trajectory of the new proposed well will need to be modified if some of the depleted zones are to be avoided.
Summary and conclusion A summary of the strengths of methods illustrated shows the following: Use of the methods helps to solve complex structural problems for simulation model building and well planning. 3D visualization software can be used to interpret seismic data, and need not be restricted to the visualization of 2D, line by line interpretations made elsewhere.
9 The use of GeoViz ~ combined with GeoFrame | provides the user with a seamless read/write facility against a central project database. This is essential when one is dealing with a database of several hundred faults that are being simultaneously interpreted and reinterpreted by a team of several geophysicists. 9 3D visualization and data sharing methods have been devised that have resulted in a considerable improvement in the communication between a partner and the operator of the field. 9 In contrast to virtual reality (VR) visualization techniques that require expensive visionariums or caves, all the method shown can be achieved in the office using a standard work station. Sufficient dedicated time would not have been available to conduct this project solely in a cave. This is because of competition for cave time from other projects. 9 It is possible to conduct the GeoViz | parts of the work in the office using stereo vision. 9 The method presented integrates results from geophysics, geology and reservoir engineering in a complexly faulted reservoir. Quality control of different data types across the disciplines is implicit the work flow followed. In this way the different members of the asset team achieved a common understanding of how the data they generated was used, and the overall significance of adopting different strategies in their working practices. 9 As we hope this paper has shown, movie making greatly improves the possibilities that are available for effectively presenting 3D visualizations in the 2D format traditionally used in meeting rooms and conference venues. Although dialogue with the end user (the oil company) will be required, the development of improvements and additional functionality that could improve the method described ultimately rest with the software development companies. Undoubtedly we have yet to reach the goal of a single, integrated package that spans the workflow from interpretation, through geomodel building to reservoir simulation, for faulted fields. This goal is a very useful one, as it will minimize the large amounts of time that are currently spent when transferring and reformatting data, usually in sorting out the unexpected errors that occur. Finally, once this goal has been met. it will be easier to identify shortcuts and automated procedures that can realize the remaining potential
Movie 19. A new well (pink) is proposed in a 3D visualization environment against the backdrop of different types of geological objects. The near base reservoir depth map is shown in a rainbow colour table, and the oil-water contact is shown in blue. Existing producing wells (blue) are surrounded by depleted volumes (light gray). An example fault plane is shown by gridded network surface (red).
3D VISUALIZATION, NJORD FIELD for improved speed and accuracy in the model building and well planning process. It will also be easier to identify the fieldspecific additional functionality that will be required in addition to the 'standard' workflow. In the case o f Njord, this will be i m p r o v e d sub-seismic resolution fault modeling and fault property modeling. The authors thank the Njord field operator, Norsk Hydro, and the partners Petoro, ExxonMobil, Paladin, ConocoPhillips, OER and Gaz de France for permission to present the material in this paper. The opinions presented are the authors' own, and do not necessarily represent Norsk Hydro's or ExxonMobil's standpoints on software use or working practices. The following deserve special mention and without their efforts this work would not have been possible: T. Bridger, M. Augood and P. Hodgeson at Schlumberger Oil & Gas Information Solutions; A. Pegley and L. Eastwood at Paradigm Geophysical. GeoFrame, GeoViz, Charisma, IESX, Framework3D and Eclipse are registered trademarks of Schlumberger; Explorer and VoxelGeo and registered trademarks of Paradigm Geophysical; RMS is a registered trademark of ROXAR.
References BLYSTAD,P., BREKKE,H., F,'ERSETH,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. BOND, R. 2001. Volume-based visualisation and interpretation of 3D seismic data. First Break, 19, 4?2-478. CARTER, N. & LINES, L. 2001. Fault imaging using edge detection and coherency measures on Hibernia 3D seismic data. The Leading Edge, 20, 64-69.
261
CHOPRA, S. 2001. Integrating coherence cube imaging and seismic inversion, The Leading Edge, 20, 354-362. DALLAND, A., WORSLEY, D. & OFSTAD, K. 1988. A lithostratigraphic scheme for the Mesozoic and Cenozoic succession offshore mid- and northern Norway. Norwegian Petroleum Directorate Bulletin, 4. DART, C. J., MCCLAY, K. R. & HOLLINGS, P. N. 1995.3D analysis of inverted extensional fault systems, southern Bristol Channel basin, UK. In: BUCHANAN,J. G. & BUCHANAN,P. G. (eds) Basin Inversion. Geological Society, London, Special Publications, 88, 393-413. KocH, J. O. & HEUM, O. R. 1995. Exploration trends of the Halten Terrace. In: HANSLIEN, S. (eds.) Petroleum Erploration and Exploitation in Norway, Norsk Petroleumsforening Special Publications, 4, 235-251. LILLENG, T. & GUNDESO, R. 1997. The Njord field: a dynamic hydrocarbon trap. In: MOLLER-PEDERSEN,P. & KOESTLER, A. G. (eds) Hydrocarbon Seals." Importance for Exploration and Production, Norsk Petroleumsforening Special Publications, 7, 217-229, MARFURT, K. J., FARMER,S. L., KIRLIN, R. L. & BAHOR1CH,M. S. 1998. 3-D seismic attributes using a semblance based coherency algorithm. Geophysics, 63, 1150-1165. MARFURT, K. J., GERSZTENKORN,A., NISSEN, S. E., SUDHAKER, V. & CRAWFORD, K. D. 1999. Coherency calculations in the presence of structural dip. Geophysics, 64, 104-111. OSMUNDSEN, P. T., SOMMARUGA, A., SKILBREI, J. R. & OLESEN, O. 2002. Deep structure of the Mid-Norway rifted margin. Norwegian Journal of Geology, 82, 205-224. PICKERING, G., PEACOCK, D. C. P-, SANDERSON, D. J. & BULL, J. M. 1997. Modeling tip zones to predict the throw and length characteristics of faults. AAPG Bulletin, 81, 82-99. WORREL. A. 2001. Rapid and accurate 3D fault interpretation using opacity and optical stacking to reveal geological discontinuities. The Leading Edge, 20, 1381-1384.
Seismic characteristics of large-scale sandstone intrusions in the Paleogene of the South Viking Graben, UK and Norwegian North Sea MADS CLAUDIA
HUUSE
l, D A V I D E
G. G U A R G E N A JOSEPH
DURANTI 4
2 NORALF
PHILIPPE
A. C A R T W R I G H T
PRAT 5
STEINSLAND KRISTINE
1 & ANDREW
HURST
3,
HOLM 5 2
1 3D Lab, School o f Earth, Ocean and Planetary. Sciences, Cardiff University, Main Building, Park Place, Cardiff CFIO 3YE, U K (e-mail: m . h u u s e @ e a r t h . c f a c . u k ) 2 Department o f Geology and Petroleum Geology, Universi~' o f Aberdeen, U K 3 Statoil ASA, Forushagen, 4035 Stavanger, N o r w a y 4 Enterprise Oil Norge, now at A/S Norske Shell, P.O. Box 40, 4098 Tananger, N o r w a y 5 TotalFinaElf Petroleum Norge, Finnestadveien 44, Dusavika, 4029 Stavanger, N o r w a y
Abstract: Post-depositionalremobilization and injection of sand can significantlychange the geometry of deepwater clastic reservoirs. Features associated with these processes are particularly well developed in the lower Paleogene of the South Viking Grahen of the UK and Norwegian North Sea. Seismic scale sandstone intrusionscan be grouped in two classes. Class 1 comprises low-angle (20-40 degrees) tabular sandstone intrusions emanating from steep-sided in situ sand bodies within the Balder Formation. The intrusions may be 5-30+ m thick and crosscut 120-250+ m of compacted stratigraphic section. They terminate at an unconformity at the top of the Frigg interval where they may have extruded onto the palaeo-seafloor.Class 2 comprises conical sandstone intrusions that emanate some 50-300+ m upward from distinct apexes located 400-700+ m above the nearest depositional sand body. The conical intrusions may have been sourced from underlying sand bodies by clastic blow out pipes. Both types of intrusions seem to adopt their particular geometry independently of (but occasionally exploiting) polygonal faults within the encasing mudstones. Sandstone intrusions may be highly porous and permeable and are thus important both as reservoirs and as plumbing within thick mudstone sequences.
Over the last decade the increasing quantity and quality of borehole cores and 3D seismic data in the North Sea Basin has led to the recognition of a variety of features associated with remobilization and injection of deep-water sands. These features are particularly common within the Upper Paleocene and Eocene of the South Viking Graben and adjacent areas of the northern and central North Sea (Figs 1 & 2; Jenssen et al. 1993: Newman et al. 1993; Newton & Flanagan 1993; Dixon et al. 1995; Cole et aL 2000; Jennette et aL 2000; Templeton et aL 2002). This paper focuses on seismic scale features associated with sand remobilization and injection, such as steep-sided mounds and discordant seismic anomalies. The steep-sided mounds are particularly common in the Upper Paleocene and Lower Eocene of the South Viking Graben where sand rich deepwater fans were fed from the west into a deep narrow basin where the predominant background deposition consisted of hemi-pelagic, smectite-rich mud (Figs 1-3; Nielsen et al. 1986: Ziegler 1990; Dixon et al. 1995; Jennette et al. 2000; Thyberg et al. 2000). Steep-sided mounds in the North Sea Paleogene are often rimmed by crosscutting seismic reflections, which in some cases have been calibrated to thick sandstone units, interpreted as inclined sandstone dykes (Dixon et al. 1995; Lonergan & Cartwright 1999; MacLeod et al. 1999; Lonergan et al. 2000; Jones et al. 2003). Another class of clastic intrusions comprises V-shaped (in 2D) or conical (in 3D) amplitude anomalies apparently not attached to a source sand body. Similar structures have previously been interpreted as conical sandstone intrusions (Outer Moray Firth: Gras & Cartwright 2002; Molyneux et al. 2002; South Viking Graben: Leseth et al. 2003), injected tufts or seismic artefacts (North Viking Graben: L#seth et al. 2003). Sandstone intrusions seen in borehole cores and logs often have very high porosities and permeabilities, making them effective both as reservoirs, as 'plumbing' within and between adjacent reservoirs, and as potential thief sands; the effects of large-scale remobilization may also severely alter original depositional geometries (Dixon et al. 1995; Lonergan et al.
2000: Duranti et al. 2002a: Guargena et al. 2002). The effects of large-scale remobilization and injection on deep-water clastic reservoirs can thus be both positive and negative for exploration and production (cf. Lonergan et al. 2000: their fig. 1). It is thus vital to recognize remobilization and injection features from seismic and borehole data in order to efficiently appraise and develop remobilized reservoirs. This paper illustrates examples of steep-sided mounds and discordant amplitude anomalies seen on 3D seismic data from the Paleogene of the South Viking Graben. Similar discordant anomalies interpreted as sandstone intrusions have previously been reported from the Outer Moray Firth to the SW of our study area (MacLeod et al. 1999; Lonergan et al. 2000; Molyneux et al. 2002). The features discussed here have been tied to the available core and log data, but the main emphasis of this paper is on their seismic characterization, visualization and interpretation. The suite of features presented herein should facilitate the interpretation of similar features, both in the North Sea Paleogene and in other deep-water settings, such as the NW European Atlantic Margin and offshore West Africa.
Detection of sandstone intrusions Dimensions The dimensions of sandstone intrusions encountered in outcrop and in the subsurface range from mm-scale to a few tens of metres thickness, up to several hundred metres length along the direction of intrusion, and up to some kilometres width along strike (Fig. 4; Newsom 1903; Smyers & Peterson 1971; Parize 1988; Dixon et al. 1995; MacLeod et al. 1999; Thompson et al. 1999; Lonergan et al. 2000; Surlyk & Noe-Nygaard 2001; Duranti et al. 2002a, b). Because boreholes and outcrops are limited in extent and usually only yield 1D or 2D geometries, and because seismic data have limited resolution there is often an order of magnitude difference between sandstone intrusions
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. 263-277. 0435-4052/04l$15 9 The Geological Society of London 2004.
264
M. HUUSE ETAL. 4~ 60ON -
Topography
0~
-
4~
8~
12~
16~ -I- 60~
(m)
>1600 1400-1600
1200-1400 1000ol 200 800-1000 6OO-800 400-600 58~ 300-400 200-300 <200
58~
U. Cretaceous-Danian Chalk Group subcrop at base Quaternary 56ON
56~
Cenozoic, excluding (m) Danian, I s o p a c h
~J~. I
'~
~
"
54~ 1000-1500
5oo-2ooo
I
le
2000-2500 2500-3000 3000-3500
J
i: 52 ~ / 4~
> 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
<
2000
2500
-, ":
3000
=~Sl~etlal1~l.-;/~/ ~- ': . } ~ 6 - " i ; i ! - e ~ +, .... 9" ~. ' , t ~ ,
.....
~,
..
9.~.~:%,Y . . . . '~" "~+, . ~-' ~"
"
:" -
- ~
:
~
~
~
:
20oo
!
~
"
Ulsira H i g h "-"vT . . . . . > " ~ - ' , ~ i ~ : i '
2500
";
3000
3500
4000
,- ; ..~:~ .?+ . < . .....' -:~+ ;:.+~; :' 9
,. ,
' .~
.9. . . ~ , .:'+. ::
4500
9
'
.
9
"
'-
-
-
:
. "
-
-
.
"
,
t;,~--,
~ u. t. h. .-
.
~--.
Greben
:
.
' ,.":
.(,..;" .~:,/,~"
.T -.-:% i-~.~-,,.~ ~[~ '.:+.:~"~-..::~-,,...'d-;' L:Q "
-'
.:
"
-
,
..:
'
+
~ ' . . ,
'
:'
.
7
" -
..
,:
" '
~.
",'
-
~
- -
"
"
-
' ~
~-~ :" ,
"'"
" "
_: "
."
~ - , - g~' ~ -
"
"+
:.:;', + =
,-.
'~--.+
'.~<~"
. "-"
.?:
f
"' G
"
(
:/.~':."
v " " : : ". ~~. ; r +- : / .::
~
--
,~" :.:<,: .- , ,
.-
..:
,,5'.
~'~
i
". ~ "
~
", .'e~,25;"~
4oo0
9
9
9
~.
4500
.~
Fig. 2. Regional WSW-ENE seismic line showing the main structural and stratigraphic elements of the northern North Sea (--59 ~ 15~N). Remobilized fields are particularly common within the white oval, i.e. in the upper Paleocene and Eocene of the centralnorthern North Sea. Vertical scale is two-way travel time in milliseconds (ms TWT).
SANDSTONE INTRUSIONS, NORTH SEA AGE =,pO (Ma) ERA ~
EPOCH
LITHOSTRATIGRAPHY
STAGE
SOUTH V I K I N G
QUAT. P L E I S T O C E N E 1.8 5.2 23.5
.,i zu,I
+ O
N
O
: ". : .-: - . : :..
(J
..:.-..:: :-.....::'.:.".:-.
RUPELIAN LU
z
I,U .....
...... ;
:=.
:. . . . .
".~ ..........., , . :
O u.I
"
,_1
55
UTSBA
CHATTIAN
OLIGOCENE
Z U,I
HIGH
PLIOCENE
mm
34
UTSIRA
GRABEN
MIOCENE
O
265
~.~L~-.-:
~-'~-"
~--~
t '
~:
~:
:: ; T R ~
_P-v-~.
<
r
Fig. 3. Cenozoiclithostratigraphy of the northern North Sea. The timescale is nonlinear in order to emphasize details of the Paleogene stratigraphy.
65
Seismic characteristics Steep-sided mounds
The peculiar steep-sided mounds of the lower Paleogene in the South Viking Graben may have flank angles of the order of 10-40 degrees (Figs 5-10), which is much higher than common depositional angles of deepwater sand bodies. The mounds are usually defined as culminations on time-structure maps of seismic markers or as areas of anomalous thickness on timethickness maps and often contain tens of metres thick sandstones encased in smectite-rich, hemi-pelagic mudstones (Jenssen et al. 1993; Newman et al. 1993; Dixon et al. 1995; Cole et al. 2000: Templeton et al. 2002). The mounds discussed here are defined at the top Balder level and appear as areas of anomalous thickness on the Balder Formation time-thickness map (Fig. 5). The sandstone within the mounds may be highly reflective or inconspicuous on P-wave seismic data depending on the contrast in acoustic impedance between the sandstone and the encasing mudstones (Fig. 6). In this case one mound is seen as 'bright' and another as 'dim', which can be interpreted in terms of variations in pore fluid and cementation of the sandstone bodies. Borehole logs and cores show that the mound to the NE is sporadically cemented and partly oil saturated whereas the mound to the SW is uncemented and water saturated. Numerous steep-sided mounds have been intersected by exploration boreholes, proving the presence of several tens of metres thick sandstones within the mounds and often also an abundance of 'ratty' sandstones in the mudstones above the main mound (Newman et al. 1993; Newton & Flanagan 1993; Dixon et al. 1995; Cole et al. 2000). Both the upper part of the thick sandstones and the sandstones in the overlying mudstones are often hydrocarbon saturated, showing good vertical communication. Recently, core data have been published showing that the 'ratty' sandstones may be discordant to bedding and intrusive in nature, thus representing sandstone dykes and sills and injection breccias (Dixon et al. 1995: Lonergan et al. 2000; Duranti et al. 2002a; Templeton et al. 2002). Short cores from the intervals above the main depositional sandstone bodies in the two mounds discussed here (Figs 5 & 6) also showed numerous sandstone dykes, sills and breccias, suggesting that the mounds have been subject to postdepositional remobilization and injection.
EKOFISK
Chalk
^ ^ ^ Tuff
~
~
.? ..-. :-...~. Deep-water sands
Shale
In some cases, deposition of the thick sandstones appears to have been confined within an erosional scour (e.g. Newton & Flanagan 1993; Cole et al. 2000). However, in many cases it appears that basal erosion is negligible (Figs 6, 7, 9 & 10). The origin of such flat based sandstone mounds, lacking basal erosion or levees and encased in uniform thickness of mudstones is rather enigmatic. Models for their formation thus involve deposition from sandy debris flows (Jennette et al. 2000), postdepositional sand remobilization due to liquefaction (Brooke et al. 1995; Dixon et al. 1995), and confinement by loading of sand into underlying muds (Harald Brunstad, pers. comm. 2000). Regardless of the mode of sand emplacement, however, it is evident that differential compaction is a major control on the geometry of the deep-water sand bodies (cf. Cosgrove & Hillier 2000: Jennette et al. 2000). Crosscutting seismic anomalies
The uppermost Paleocene and Eocene of the South Viking Graben and adjacent areas of the northern North Sea contain numerous discordant events, many of which are conspicuous because of their high amplitudes, representing both 'hard' and 'soft' events. The interpretation of such events is highly ambiguous in the absence of well control, but several have now been calibrated to thick sandstone units seen in borehole cores, cuttings and logs, and are thus interpreted as inclined sandstone intrusions (Dixon et al. 1995; Lonergan & Cartwright 1999: MacLeod et al. 1999; Molyneux et al. 2002; Lcseth et al. 2003). Two classes of seismic anomalies may be defined based on their relation with the encasing shales and connection to inferred source sand bodies.
Class 1: Wing-like reflections emanating f r o m steep-sided mounds. One type of crosscutting event is seen as wing-like
reflections that emanate from the sides of steep-sided sandstonecored mounds (discussed above, Figs 6-12). The discordant events may be high or low amplitude, positive or negative reflections (cf. Figs 7, 9 & 10), depending on their porosity, cementation, thickness and pore fluid, as well as their continuity and geometrical complexity. Discordant reflections, inclined at 20-40 ~ to the horizontal, surround both of the mounds discussed above, but their seismic expression is completely different.
266
M. H U U S E ETAL.
~-
r~
8
"==~==
t
l.
z
~
~
.-~ ~ .~
o
~.~
o o~ ~ > 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,,+~. ~ ,~ .,~+,+ . . . . . . . . . . . . ,~~+<++' ;' 9 +++t-e+ ..-~+~. .+. .. .. . . . .. ~+' -+. +. +. + +..... . + ; , ~m ,-
.. +. ++~
. . . .
+
.
~
'
.
.
*
'
.
.
.
......
.
9
-"
~ %
' .
.
.
.
..
.,.. - - , , . .
.
.
.
'
.
.
~ .
.
.
~
~.~.-e,~',,,s
~"' ' " 9'
:
.
.
.
. ~' ~ " '
"+P.,++-~"~
. +
9
L-'.
-r~
l
-
.
.
.
.
.
.
.
.
.
.
.
.
.
. . . . .
Enlarged
-N' ~/
.
"
."-"
~+,++- +-- ~
~
~-;',:.+;-+': ;-~~,
~
".
""
;.:-+++ ......
-..,., - + . + , , +.~,~ +. -~ - - ,, -, . . . . . "--,.~ ..,-. . . .. . 9 , ~ . +, . , .. . . . . . . . . .
:.~+:.
-- .
.
.
, . . . .
+.... . .-.-tT+t,
.
.
.
. N
""
:::
.
~
"+iT.,
; :
i ,...
.
'
" " 9 2.."b".-,., _
~.+
.
.
" J
+ +9
-++-...+. .
- . 9
"
~ "
, +'.+,-:=+.7,+ ~ ,-
.
.
,'~.
~ . ~ " "E : " ~ " ' ~ ' - - ' ~+" ~ +'
"+"---,+--.."~:-
9
~
.
-
; -
"
.:'~---
.,,.,
.
" ~''~'~
'
:
r
9
-' ,.+,'+~ 1 ,
.-.,~ 9
- ~ . , . :
_,,+,._. . . . ~ r.'~, +-~ ~..~.,m
.---
";p
.
+~
-
" .+ "" ' ' ~
-
t '-. . . . . . .
+~.
~
.....
....
~
"
"
.
~
~
.
9 ....
"
.
~
~ ~
i.
........ .
'
.~_,_
9 % '-'~d~'+'~"
+";'+',"t"~'-. . . .
.
I
:,7-.,~..+
+ - ~-'-.- .-++:t~='+.,,:-+: + ~ .
.....
' . . . . ~'+ - " " . . . . . ";" " v , , . . . . . .: :.;,..,__.,,1. ~ " e ~ : t e t ~ ": : ~ . , , ,~2 ~ ' L, ,,+ ~ I ~2 " ' - ' .' " , ' ~ " ; ' r ' , f. -. ., ., .7 -~~ : ' d " + +"' ' ~ " ;"; ,' . ~ . ,"+" . . ' "
@ 160Ore/ms
"~-"'-"~+
-'-..,,~ .
"
.
~-:
-
",.-.~ . . . .
'
~
+.-:,.-.-'~"m~i.'.
"
.....
J :. 9 ,~ ~.
+"'~--~---'+~" ,."' " . . . . . "-''~""'+;~+':~';-:-:
~
0.0125~eC-~
_~,...._,~
7 "~'~t'-'rP'-
"
~ ......
"
-
-
...... ,. . . . .
-!-+++,
Fig. lO. Example of integration of seismic data from the Vgring Basin, Norwegian Sea. (A) Deep Tow Boomer sourced UhraHiRes 2D seismic profile across the intended 'Frontier' well location, (L) shows a relatively small fault. (F). The fault intersects the well trajectory L at depth approximately 280 m below sea bed, cutting s o f t - f i r m becoming stiff silty clays of the Pleistocene (Naust sequences). Fault F tips almost at the sea bed, which is deformed, where it is covered by a H o l o c e n e - R e c e n t veneer thickening into the hangingwall. Note T W T values are scaled to avoid conventional workstation restrictions when integrating highly sampled datasets. (B) HiRes 2D seismic along line N - N ' as in (A) above showing upper 0.5 km of Naust and Kai Plio-Pleistocene stratigraphy and a gas chimney (GC). Black box shows position of UltraHiRes line in (A). Regional mapping shows occasional faults that do cut up to the seabed. Others tip at various deeper levels whilst yet others are associated with development of gas chimneys allowing differential leakage of pore fluids such as GC. Here the deformed, mounded, chaotic reflection patterns within the upper Kai sequences display anomalous amplitudes, some almost flat, with significant masking of the seismic data beneath. (C) 3D seismic data suggest the faults link laterally as polygonal sets only crudely displayed across this particular Time Slice (TS) at 1.950 s TWT. Faulting initiating the development of gas chimneys is suspected to allow deeper ( 3 - 5 k s ) expulsion into assumed shallow reservoirs trapped within the dominantly clay overburden of the Naust sequences. The Time Slice shows the m a x i m u m areal extent of the well developed gas chimney (GC). (D) Reflection Intensity attribute display extracted from the exploration 3D seismic volume over the interval 0.050s above the TS displayed in (C) encompassing the N a u s t - K a i boundary (green pick in B). Anomalously higher values are reds and light blues are lowest Reflection Intensity values 9 This initial extraction formed the basis from which higher frequency 2D acquisition was targeted to then allow identification and analysis of drilling geohazards along the intended well trajectory. Red circle indicates the original intended drilling location that has high Health, Safety and Environmental risk.
288
B. AUSTIN
|
.
.'
,.....,..,.,(~
:~.~:~
"
".
'
k
. ~,
.
.
.
.
~
,,
,
I ".-
Z,O0--
II .r
~
,
"
.
!~.
~r 2..200 m " r~.4
.~'.~
_
- ,tl .-
/.,.,
. ~'~
,
~
I
.
.
i
1
~ ~"
Z300
o . .
,,
-
v
'
. . . . .,
,
.
'
: -
.......,
-
.
.... . , .
_
.
.
.
.
~ , ~
~,.,'. - . l,a.~ ,
.
~"
..
.
.
',..
, : ' J J M ~ ' ~
:.r.~u~,~7~
!
"/
0"
,,iL.~" ., :
.....
[.2klm
..,,r r,~,
Fig. 11. 3D vs HiRes2D seismic data comparison from the Voring Basin. The lines cross more or less orthogonally close to a proposed Frontier drilling location (L) where water depth is minimum of 1375+/- 15.25 m. The seismic profiles are considerably squeezed with vertical exaggeration being 4:1 for display purposes. (A) Initial exploration 3D seismic data (fast track-processed, near offsets only). (B) The HiRes 2D data is far less smeared and unambiguous. The faulting observed to cut virtually to seabed is barely discernable on the 3D seismic data and it would be unwise to identify and analyse geohazards using these data alone 9Note also the 0.025 s (18-19 m) time shift (Sh) at seabed (Sb). This is variable down the record length and is likely due to differences in stacking velocities selected covering the shallowest part of the section. The sub-parallel stronger reflection (OPAL) is interpreted to cut across the gently dipping and faulted seismic data. This event is also seen in Figure 10 at 2.375 s where it is more clearly cross cutting. It is considered to represent a fossilized Opal A to Opal C/T transformation as documented elsewhere along the NE Atlantic margin (Davies & Cartwright 2002). (GEM) Projects, (Atlanticon-GEM 2001). More emphasis can then be put upon design of smaller higher-resolution acquisition programmes. These are specifically targeted to first develop then flesh out a shallow geological and geotechnical model followed by detailed prognosis of geological conditions pertaining to any critical area of interest. Use of a 3D seismic data volume, once reliably 'calibrated', has allowed increased cost savings during field development since if undertaken correctly, less HiRes seismic data needs to be acquired 9The latter can often be targeted far more effectively to identify and investigate sea bed features, geotechnical considerations and constraints or to target optimum geotechnical borehole locations. Examples are given below of how calibrated integration of both 3D and HiRes 2D seismic data has been used in field developments in the Haltenbanken region of the Norwegian Sea, West of Shetland UK Quad 204, and in the Central North Sea.
Asgard field development, Haltenbanken Asgard is a subsea development on the Mid Norway shelf in 350 m water, a region notorious for encountering Quaternary overconsolidated tills, boulders and significant changes in geotechnical conditions. Regionally pervasive shallow gas sands exist in traps associated with unconformities and infilled iceberg scours. These present a risk to drilling in the region and caused the 'West Vanguard' shallow gas blowout in October
1985. The main foundation engineering objectives to be met were to ensure efficient emplacement of suction anchor/piles, sub sea templates and well cluster hardware. Conventional integration of exploration 3D seismic data and a well-planned HiRes 2D seismic grid allowed successful identification and prediction of foundation conditions. 2D seismic line spacing was 100m with additional 500 m spaced cross lines and location-specific tie lines across the field. Initially, standard gated attribute analysis of the 3D seismic data volume was used to map hazards and predict identified anomalous foundation conditions across the development area (Fig. 13). The shallowest Reflection Intensity (R.I.) attribute map provides confidence that most development well locations are sited away from the linear or concentrated patches of higher R.I. values shown in red and yellow (Fig. 13a). These high amplitudes are the seismic response to shallow gas accumulations in sands deposited during climatic amelioration which infill palaeo iceberg scour hollows 9 Renewed glacimarine deposition has later buried the features beneath clay and silty clay sequences that form a top seal. Given the low vertical resolution of the 3D seismic survey, a geotechnically plausible interpretation of the foundation conditions at any particular level can only be provided once the 3D seismic volume is adequately 'calibrated'. Calibration attempts to 'quantify' the likely geological/geotechnical conditions or changes contained within the broad 3D seismic wavelet and its anomalous neighbouring traces by detailed correlation to geotechnical borehole or other information.
INTEGRATED USE OF SHALLOW SECTION 3D SEISMIC
289
Fig. 12. (A) Irregularly topped, low mound shaped features interpreted as buried carbonate mounds, (M) occur coincident with the regional Late Oligocene-Early Miocene Unconformity (L) as resolved in profile by a quality HiRes 2D seismic line. The largest, most northeasterly mound, (M) measures 22 m thick x 1200 m broad whereas others are less broad. The line is located on Figure 1 near the base of the E. Faroes Slope where seabed (Sb) is greater than 1000 m and harder Leutetian sediments subcrop beneath the LOEMU. The Intra Neogene Unconformity, (blue horizon) and subsequent younger late Pleistocene Units 1CIII, ICII and IC1 create a 38 m thick overburden showing indications of differential compaction. Monoclinal Ypresian (Yp) to Leutetian strata are onlapped (O) by Priabonian and Oligocene above the brown base Oligocene horizon pick (O!). Continued movement into the Neogene is well illustrated by sharp truncation of the 'syncline', (stronger blue-red doublet reflections), beneath the largest mound M. (B) The full fold, 12.5 x 12.5 m binned 3D seismic line directly corresponding to A above illustrates significant differences caused by lower resolution and tuning phenomena. Same horizontal scale. Relative age of Units ICI, 1CII, LOEMU etc is shown in Table 2. Assumptions made during calibration depend upon suspected lithology, consolidation history, degree of diagenesis and other rock properties. Careful analysis of 3D vs. 2D seismic stacking velocities should be carefully accounted for as must the assumptions made in d e p t h - t i m e conversion whilst tieing from the geotechnical description logs and recovered samples into the HiRes 2D seismic. Variation in apparent sediment seismic velocity can be relatively wide from 1485-1700 or even 1900m/s or more. Such range is often significant given no electric well log data exists since it increases the fuzziness of the fundamental well-seismic ties. Calibration forms the critical remote link between geological ground truth (e.g. actually encountering an overconsolidated, rock-hard till or skirt buckling boulder bed) and the HiRes seismic with its promise of inexpensive lateral prediction once tied into the 3D seismic data volume. The Asgard HiRes 2D seismic data have been carefully 'groundtruthed' by tieing to geotechnical boreholes sited at three of the well locations shown on Figure 13a. Results show that gravel and boulder concentrations encountered in the Quaternary succession scatter higher frequency seismic energy giving rise to characteristic diffraction hyperbolae. This provides the chaotic internal seismic geometry observed on HiRes 2D seismic at the 2 0 - 3 0 m depths of the R.I. attribute extraction gate from the 3D seismic volume (Fig. 13b). The R.I. attribute maps display bright yellow/red anomalous amplitude patterns within an otherwise blue/green background exhibiting lower R.I. values (Fig. 13b). The anomalous values are interpreted to represent higher concentrations of cobbles and boulders within the otherwise fine grained succession since the less resolute 3D seismic reflection configuration is essentially
similar to the HiRes 2D seismic reflection pattern at the boreholes. Even deeper down, similarly anomalous conditions that would impair installation are noticeable but are restricted more to the north and extreme south of the development (Fig. 13C).
A development west of Shetland The Foinaven and Schiehallion oilfields lie partially within Block 204/20 on the West Shetland Slope of the UK where water depths increase towards the West from 2 0 0 - 5 0 0 m . Block 204/20 is highlighted in Figure 1 overlying the large debris flow features of the Rona Apron at the sea bed. Following a successful and intensive discovery and appraisal campaign in and around the block, occasional drilling problems caused interruption to otherwise smooth running field development drilling. Post-well analysis showed the problems were related to two separate intervals within the shallow section. These were seismically weakly defined and hence not wholly understood. In Quadrant 204 during the mid-1990s, predicting geotechnical conditions at exploration and development wells relied to a great extent upon analysis of the regional stratigraphy. Beneath sea bed a complex section of poorly dated Plio-Pleistocene glaci-marine sediment units was considered to exist. These units unconformably overlie gently dipping, faulted, presumed early Neogene sediments. For the purposes of drilling this makes the region appear quite similar to the better-understood and documented Northern North Sea (NNS) region where few drilling problems are expected. Given the previously low drilling activity levels West of Shetland compared to the NNS there was only a limited knowledge of essential unit lithology,
290
B. AUSTIN
Fig. 13. Reflection Intensity (R.I.) attribute maps of 0.032 s gates centred upon three important picked horizons from the sub-horizontally layered Pleistocene sequences encountered below seabed across the Asgard Field. White circles are drilling centres/piled facilities. Displayed area covers approx. 8 • 15 km. (A) Diffraction hyperbolae and chaotic internal seismic character, as observed on resolute 2D seismic data, have been calibrated by samples and logs from three geotechnicai soil borings to represent increased cobble and boulder concentrations in (over) consolidated glacial tills. These are correlated as the reds and yellows on the 3D attribute map indicating the dense spatial distribution of contained coarse material within the tills. Such conditions hinder ordinary emplacement of fixed piles at this level and could necessitate costly drilling and grouting if not predicted. (B) Fewer anomalous R.I. values, (An) are displayed on an extraction from the 3D seismic data at a slightly deeper level allowing prediction that the geotechnical soils profile becomes more benign with depth. (C) Enlarged view of Asgard where 3D seismic data has detected NNE/SSW trending lineaments, (within the Limit of Separability), interpreted as buried iceberg scours (Is). During climatic amelioration, these have become infilled with sands and later top-sealed under renewed glacial deposition. These conditions now act as conduits and traps for accumulations of shallow gas imaged as higher Reflection Intensity yellows and reds. thickness variation and geotechnical conditions within the PlioPleistocene and even less knowledge of the underlying geology. Perhaps unsurprisingly at well locations in and surrounding Block 204/20 geotechnical conditions encountered in the shallow section were found to vary quite considerably. This was the case not only when comparing subsurface conditions several kilometres apart but also with those similarly drilled locations separated by only a few hundred metres. Where noticed during passage of the bit, two particularly troublesome units, assumed to be sandy sediments became known informally as the 'Taylor Sands' above with the deeper 'Calvert Sands' below. Their seismic expression is shown on the 3D seismic and specifically targeted HiRes 2D seismic data, (Figs 14A and B). The presence and unpredictable variation of problem 'intervals' (geotechnical units) in terms of drilling behaviour is a concern for field development planning. Delays meeting drilling targets, (caused by stuck pipe and casing, fluid losses and caving or other loss of control), considerably increase costs. Furthermore they add to delays in appraisal decision making and planning. In turn, under worst case or 'ultra-fast-track' conditions, these affect even higher cost items such as facilities design and timing of order placements and procedure. Such geotechnical problems are only occasionally experienced in other parts of the NWECSS. They occur more frequently where
associated with deeper water environments with rapid rate of sedimentation and burial such as those of the Gulf of Mexico (Atberty et al. 1997). Here shallow section 'flowing sands', which are slightly overpressured and difficult to define seismically, are held responsible for unplanned and costly delays during intensive development e.g. at the Ursa field. Standard industry exploration and development practice, not just West of Shetland, means virtually no geological data younger than early Palaeocene is normally collected whilst drilling. Lithology and geotechnical conditions at the two problematic levels are therefore difficult to judge by inspection of either the extensive 3D seismic coverage or limited HiRes 2D seismic data shown as Figure 14. It is noticeable that a sequence of acoustically opaque and channel-like reflections occurs at the 'Taylor Sands' level as defined by drilling information. Little or no difference apparently exists, however, between the well locations (black vertical lines on Fig. 14), where drilling problems were experienced and where drilling continued without significant interruption, (Fig. 14C). The deeper 'Calvert Sands' appear to be associated with multi-layered and/or very low relief channel-form reflections exhibiting both convergent and divergent sub-parallel internal reflection geometry. This is an unexpected relationship since the channel is located within the presumed glacial outer shelf and is
INTEGRATED USE OF SHALLOW SECTION 3D SEISMIC unlike any seismic profile observed from the Central or Northern North Sea Plio-Pleistocene shallow section in the author's experience. S u c h reflection g e o m e t r i e s can, h o w e v e r , be associated with sediments deposited by N e o g e n e contourite current activity as observed in other parts of the F a r o e - S h e t l a n d Basin and b e y o n d (Stoker et al. 1998). Little prediction could be achieved from the interpretation of the available database defining the detail o f N e o g e n e structure and stratigraphy in the region. The stratigraphy is quite complicated c o m p a r e d to m a n y other regions o f the N W E C S S , i.e. akin to the Faroes sector further to the West (Table 2).
291
It became clear that to understand the seismic profile and prevent any re-occurrence of expensive drilling problems the stratigraphy, at least, needed to be investigated more thoroughly. This was partially achieved w h e n calibration to the nearest British Geological Survey (BGS) shallow b o r e h o l e - - 9 9 / 3 eventually b e c a m e available. BGS 99/3 was drilled in 1999 as part of a sophisticated industry funded campaign, (Fig. 1: Hitchen 1999). The use of an integrated approach (i.e. calibrating 3D seismic coverage with regional 2D seismic tied to a targeted reconnaissance HiRes 2D seismic data grid), allowed the complex seismic stratigraphic relationships shown in Figure 2
Fig. 14. (A) Uninterpreted, reconnaissance HiRes 2D seismic profiles shot to integrate well data (located close to black lines), over an unusual soils profile on the W. Shetlands Slope. (B) Exploration 3D seismic data that is directly comparable with the seismic shown in (A) above along line A - A ' located on Figure 1. Late Pleistocene hummocky debris flows, (DF) are well defined at sea bed overlying Unit 1 or undifferentiated Morrison Sequence 2 above the presumed INU surface (see Table 2). Highly condensed and truncated Units 2 and 3 Miocene and Oligocene reflections are seen where Oi is base Oligocene correlated from the BGS 99/3 shallow borehole located 40 km to SW (see Fig. 1). Eocene seismic sequences below show channelized and lensoid external geometries and subparallel to mounded internal reflectivity with complex onlapping relationships and small faults. Problems affecting drilling operations similar to those experienced under notorious shallow flowing sand conditions in the GOM have been experienced within these sequences. Two separate intervals appear to be responsible--the 'Taylor Sands' (T) and deeper 'Calvert Sands' (C). Top-hole drilling data depth estimates vary so widely, however that their location on the profiles is difficult to locate precisely where 0.100 s TWT is assumed to equate to 80 m. (C) Interpretation of the HiRes 2D seismic data from Figure 14A highlighting the differences in the shallow profile at the exploration and development well locations, (black lines). Here drilling problems, (Dp), were experienced at (T) and (C) or both whereas no difficulties occurred at one location, annotated zDP. Red vertical lines are acquisition line-crossing splices. Other annotation as per Fig. 14A. The interpreted stratigraphical relationships are based upon regional analysis and correlation to the shallow borehole BGS99/3, The "Calvert sands" interval between the base Oligocene, (O!) and the top Ypresian, (Yp) show internal reflection geometries on the HiRes 2D data indicating a complex erosional and depositional environment that is poorly understood. These require HiRes3D coverage to properly map in order to allow lateral prediction of likely subsurface conditions to aid drilling engineers and operations.
292
B. AUSTIN I I [ DpT+C
A
I I DpT
1
- - 6f,o
Slron~er amplitudes
[
I
I
DpC
A'
"
~
in
fill?
~,
'
DF
1/
Ii'
"~
,,),
; t i
~
+++. . .+++ :i
I
"
f,,.
.
++
.
,
.
.
.
~
.
.
.
e9s;,. ,
i~' -~~
I
1o
+
.
t
.-,+,L
~...~. ,
++'~
+'+
.
-v_p+.
,,,.,,,,.+ ,,.a+,,~
JLm~(cmern
O~iousdov~s in ? c l u m ~
.
" +"
!+ +++:
+...r
r
~"
t .~twsluiffuies)
"
oukr skdf e n ~ m m t ?~m setlm~ ~
, ~1 1
~. q mqlplmly l t s l l d ~ mf l ~ ~ mml.mmmiis ,
OIm~
~
i.~'~ .....
and Table 2 to be observed around Block 204/20. Regional analysis of the 3D seismic volumes followed. Initially, the seismic data was interpreted using common sequence stratigraphic and structural methods. This was followed by a specifically targeted but limited acquisition of HiRes 2D seismic data (Fig. 14A). Although the geological borehole 99/3 is located 40 km away, these data were successfully tied into the database through the local borehole-specific grid of UHiRes data. As much geological and geotechnical ground truth as possible was then correlated back to the problem areas from the borehole ties. Unfortunately given the variation in structure and sedimentology of the late Palaeogene and Neogene sequences over 40 km, the borehole geotechnical ground-truth could only be applied to the upper 25 m below sea bed. The problem drilling zones therefore still remain essentially poorly defined. The calibration exercise did, however, prove particularly beneficial for planning recent exploration wells located in the deeper, Faroese parts of the basin (Fig. 2). Much useful age-dated stratigraphic input into the regional interpretation was gained from the integration of borehole and seismic calibration. This confirmed the identification and presence of the Top Palaeogene Unconformity/LOEMU, (Table 2). Interpretation of the 3D seismic then allowed the LOEMU to be discriminated from a younger unconformity assumed to be the Intra Neogene Unconformity (INU), (Table 2; Stoker 1999), as well as from several erosive events (turquoise, brown and purple picks) observed beneath (Fig. 14C). Interestingly the stratigraphic correlation interpreted here shows how a reduced section of Neogene and upper Palaeogene strata exists less than 300m thick. This is much thinner than that interpreted 35km further west where Lutetian aged sandy basin floor deposits form the basinward development of the equivalent Palaeogene section, which is at least three times as thick (Fig. 2). Strong onlap within the post late-Ypresian
C
Fig. 14. (continued)
sequences and continued periods of tectonism into the late Neogene is further responsible (Fig. 12, Boldreel 1993 and Smallwood 2004). Using the combination of 3D and HighRes2D seismic data, much progress has been made in understanding the later Palaeogene and Neogene palaeogeographic environment and facies changes up the West Shetland Slope and Faroe-Shetland Channel. Correlation with the work of Davies et al. (2001) or Stratagem Partners (2002, 2003) for example indicates that the younger Taylor Sands could be of late Miocene age. There remains the possibility, however, they are in fact Pleistocene since the INU can be interpreted as being eroded out, by local Pleistocene incision in this area. The Calvert Sands can be tied seismically to lie beneath the base Oligocene (dark brown between 0.825 and 0.925s) and above the top Ypresian horizon--purple event (Fig. 14C). They appear to represent the outer shelf and failed upper slope during Lutetian times (Davies & Cartwright 2002). Although far from complete, the integrated use of various datasets has at least outlined and refined the stratigraphic context in which the drilling problems in Block 204/20 occurred. If development funding allowed, the next step could be to reprocess the 3D seismic data (as described in the Central North Sea example below). The occurrence and thickness of the problem interval could be mapped along with careful analysis of the attributes contained within the 3D seismic data. It is likely at this depth, however, that the conventional 3D seismic data would still be plagued by thin-bed reflection tuning effects. An example of potentially erroneous picking and local sequence boundary ambiguity created by this geophysical phenomenon is shown as Figure 12. Interpretation of 3D seismic data alone would suffer through lack of sufficient vertical resolution even assuming there was geotechnical control available from a betterplaced borehole. This leaves further HiRes 2D seismic (or even
INTEGRATED USE OF SHALLOW SECTION 3D SEISMIC better--HiRes 3D) acquisition as the remaining option should drilling problems persist. The case described above raises again the serious financial issue of precisely what is the appropriate budget programme and optimum acquisition to meet engineering requirements throughout field life? As recognized by Dutta (2002), many cases show it may often n o t be the route with the lowest bottom line when the decision to develop is taken. Operator cost sharing through joint consortia (as now occurs offshore the Faroes, Ireland, Norway and UK), allows the necessary holistic overview. This approach is recommended as being highly cost-effective for new deep and ultradeep developments over the lifetime of large producing fields, e.g. Angola, Caspian, NW Africa.
A Central North Sea development Several severe drilling problems with shallow gas, fluid losses, and stuck pipe necessitating re-drilling and other rig downtime were recognized by the Operator of a CNS development situated over a salt diapir structure shown in Figure 15. Production drilling plans required advanced design casing strings that required detailed prediction of any gas occurrence in the shallow section. A number of HiRes 2D seismic surveys had been acquired through time for each appraisal well. Sub-sea bed penetration is low due to p-wave dispersion through shallow gas and shallow seismic data could generally only be used to interpret the uppermost layers, channels, and anomalous sea bed
293
conditions. Since water depth is only 75 m, 3D seismic data imaging was typically poor in the shallow section. Budget was justified for short offset trace correction to be applied with reprocessing back to the original temporal sampling of 0.002 s to increase seismic bandwidth. The reprocessing resulted in a sixfold near-trace volume with tight 6.25 x 12.5 m bins covering a 6 x 6 km area across the entire development drilling area. By integrating the reprocessed 3D seismic data volume with the HiRes 2D seismic for the uppermost layers, shallow section problems could be investigated using detailed interpretative analysis qualified by back-prediction from the exploration and appraisal drilling history. Of prime importance was the recognition that an active gas chimney appeared to be linked to the shallow extensional fault pattern above the salt diapir. Wells were found to have encountered problems when passing through the interpreted fault planes. The fault zones were seen to be conduits for active, upward, gas migration (from well pressure data increases and mud log chromatograph readings). The location of three major faults, (green, blue and brown) mapped at the Mid-Miocene Unconformity level, together with some of the existing well paths and planned producers, is shown as Figure 16. Maximum fault heave is in the order of 7 5 - 9 0 m but reduces rapidly as the faults tip out in the Pliocene succession around 0.280 s TWT (Fig. 17). Significant quantities of gas occurring at three reflective levels in the shallow section were identified and mapped from the reprocessed 3D seismic volume. Structural trapping occurs in both footwall and hanging wall blocks bounded by the green,
!
It
!
Pfio-Pleistocene
Fig. 15. A recent full fold 3D seismic line over a Central North Sea development associated with a salt diapir, (SD) showing sub-regional setting. Near traces from a subset volume covering the shallow section, (yellow box), were reprocessed as described in the text. The dataset has allowed definition of gas migration pathways in the shallow section expulsing at the seabed (Sb). These are associated with a linked listric fault system, (colour coded F), extending above the diapir that can give rise to expensive problems whilst drilling wells. Several selected well trajectories are shown as black lines. Red boxes approximate the HiRes 2D site survey data, (SS) that were originally acquired to help solve the problems individually for each exploration and appraisal well surface location. These data are of reduced penetration due to effects of very shallow gas as shown on Figure 17.
r,
:
9
Miocene
. . . .
=
+
.
o=
/
Palaeogene
,#
-,
"'2 \
\ /
\
/
:
/
I0 tam
U. Cret.~
V.F-..
~ It 6 . 2 5
3.000
294
B. AUSTIN
Fig. 16. Structure map at the mid-Miocene unconformity identified in Figure 15. Map shows a salt induced, faulted domal structure in partial collapse where reds/yellows are high and greens/blues are deeper TWT values. Interpretation and mapping of the 3D seismic data using automated trace difference techniques shows a clear radial pattern of minor faulting with one concentric fault presumably due to diapir downbuilding and growth. Three normal listric fault planes are highlighted, (black zones) and have been correlated by hand. These dip approx. 60 degrees down towards the east and show an interesting pattern. In profile, (W) they can be viewed on the reprocessed 3D seismic data, shown as Figure 17 and at shallower mapped levels as Figure 18. The Blue Fault is generally planar throughout whilst the cuspate Green Fault becomes even more so as the compacting Neogene sediments slide off and around the flanks of the diapir. Although footwalls have similar aspect, the cuspate nature of the Brown Fault is, however, opposite to that of the Green Fault.
Fig. 17. Poor quality print of 3D seismic data along the deviated trajectory of the production well path, (W) located on Figure 16, from the 0.002 s sampled, sixfold, near-trace 3D seismic volume. The volume has been specifically reprocessed to capture the broader bandwidth and details of the gas prone shallow section above the salt diapir shown as Figure 15. During drilling both the Green and Blue listric fault zones were recognized by drillers whilst the Brown fault and carbonate cemented mid-Miocene unconformity were combined. The rather fuzzy depiction of the Blue and Brown faults here relates to poor control from the autopicking algorithm when attempting to provide depth maps of the fault planes themselves. The production well was steered safely and efficiently to absorb pressure kicks associated with the three fault zones at depth and avoid the worst accumulations of gas, notably those forming footwall traps at the Red (R) and Yellow (Y) levels.
INTEGRATED USE OF SHALLOW SECTION 3D SEISMIC
Fig. 18. Reflection Intensity (R.I.) attribute maps extracted from the shallower Red Level (A) and the deeper Yellow Level picked horizon, (B) as shown in profile on Figure 17. The maps show anomalously strong R.I. values in red and yellow defining gas accumulation within an overall domed structure with greater volumes in sands at the deeper Yellow Level, (B). The structurally controlled R.I. cut off is assumed to represent a gas-water contact (GWC). The tips of the Green, Blue and Brown Faults are identified (Gt, Bt). Maps (A) and (B) show fault footwall traps plus a doughnut shaped rollover in the hangingwall bounded by the Green Fault. The central core, however, within the blackdashed polygon (GC) shows low R.I. values. This area gives the appearance of being apparently non-gas filled but is interpreted as an artefact of the 3D seismic data also seen at the overlying Red Level. This is caused by disturbance of the seismic wave field and masking by a vertical gas chimney (GC) in generally unfaulted Pleistocene overburden through which the hydrocarbon system leaks naturally to the sea bed. The white area (Er) outlines a glacial channel eroding into the Red Level.
blue and brown faults. The Reflection Intensity attribute associated with the structural surface of the Yellow Level at around 0.630 s TWT (Fig. 17) shows the domed gas accumulation closure intersected by faults. It is interpreted that the structure is filled by gas migrating up fault zones and leaks naturally to the sea bed via the gas chimney (Fig. 18). Semblance and amplitude extractions from gated amplitude windows helped identify and focus upon deeper accumulations. This allowed quantitative distinction of risk levels between discontinuities created by fault activity and those caused by lithology contrasts at unconformities. After the integrated interpretation, 3D seismic visualization techniques aided definition of a drillers' earth model that enabled sophisticated well planning to avoid hazards and design trajectories to intersect the fault planes optimally (more acutely). Feedback from ongoing production drilling also allowed increasingly improved t i m e - d e p t h conversion to locate the fault plane zones (Figs 16 and 17). The carefully planned,
295
0
almala
~q[~O
1000
1500m
strongly deviated well bores were accurately geosteered efficiently and safely reached the intended target levels with far fewer unanticipated drilling problems and greatly reduced rig downtime.
Conclusions
In the glaciated margins of the NWECSS, 3D seismic data alone does not provide enough information from the seabed to around 500 m burial depth to satisfy all the requirements of engineering and drilling teams. This generally remains the case in underexplored deep and ultra-deep waters. Specialist reprocessing of the near offset traces can often be a reasonable approach to overcome some of the shortfalls. These data enable the identification and mapping of geological structure, many geotechnical changes and features such as shallow gas. Other more subtle features, e.g. the location of
296
B. AUSTIN
pressurized flowing sands, thin contourite sheets, or carbonate mounds, may still not be identified on these data. In these situations the safest option, to acquire HiRes 3D seismic data to target the problem zone(s) is also the most expensive. The expense can surely be justified in order to avert what Stewart & Holt (2004) term the costly 'train wreck'. Should this not be an option then reprocessing the 3D seismic data followed by careful calibration with HiRes 2D seismic could help to reduce risk. Once targeted HiRes 2D seismic data are acquired, careful interpretative calibration of the 3D seismic volume can take place. When tied back to useful geological control points the often complex shallow geology and sedimentary environments may be revealed. The results can be extremely powerful for interpretation and prediction of the geotechnical conditions given the c o m p r o m i s e in vertical resolution. This is the case, for example, with infield pipeline and cable routes, the upgrading of sea bed facilities or for enhanced recovery drilling programmes. Such planned, integrated use fully maximizes the value of the initial seismic investment throughout field life.
Thanks to P. Walker, G. Taylor, J. Trueman and T. Evans for many beneficial comments that improved the text. Also to J. Cartwright and D. Long for their initial reviews and to R. Davies for further editorial guidance. Grateful thanks to all friends and colleagues at the following organizations and companies who helped to acquire and process the datasets and also gain permission for their comparison and display in this paper: Agip, AHL, Anadarko, Baker Hughes/Western Geophysical, BP/Amoco, Exxon-Mobil, GECO, HAL, PGS, Phillips, Norske Hydro, Marathon, Shell, Statoil, Svitzer, VeritasDigicon, Atlanticon-GEM, BGS, RSG/PIP and SOC.
References ALBERTY, M. Er AL. 1997. Mechanisms of shallow-water flows and drilling practices for intervention. Offshore Technology Conference, Paper OTC 11971. ATLANTICON/GEM 2001. Faroese Geotechnical Work Group Website. http:/www.fo.com. AUSTIN, B. J. 2001. Geotechnical Work Group Summar3, Document. Atlanticon Report No: 00001437. BOLDREEL, L. O. & ANDERSON, M. S. 1993. Late Palaeocene to Miocene compression in the Faeroe-Rockall area. In: PARKER,J. R. (ed.) Petroleum Geology of Northwest Europe. The Geological Society, London, 1025-1034. BROWN, A. R. 1991. Interpretation of Three Dimensional Seismic Data. 3rd Edition, American Association of Petroleum Geologists, Memoir, 42. DAVIES, R. & CARTWR1GHT,J. 2002. A fossilized Opal A to Opal C/T transformation on the northeast Atlantic margin: support for a significantly elevated palaeogeothermal gradient during the Neogene? Basin Research, 14, 467-486.
DAVIES, R., CARTWRIGHT, J., PIKE, J. & LINE, C. 2001. Onset of Arctic-N. Atlantic deep water circulation during the early Oligocene. Nature, 410, 917-920. DUTTA, N. C. 2002. Deepwater geohazard prediction using prestack inversion of large offset P-wave data and rock model. The Leading Edge, 21, 193-198. HITCHEN, K. (compiler) 1999. Rockall Continental Margin Project Shallow Drilling Programme. Geological Report. British Geological Survey Technical Report WB/99/21C. KENYON, N. H., AKHMETZHANOV, A. M., WHEELER, A. J., VAN WEERING, T. C. E., DE HAAS, H. & IVANOV,M. K. 2003. Giant carbonate mounds in the southern Rockall Trough. Marine Geology, 195, 5-30. LAPPIN, M. ET AL. 2002. Integrated Use of 3D Seismic in Field Development, Engineering and Drilling--Extended Abstracts. 3D Seismic Conference held at the Geological Society London, November 14-16, 2001, http://www.geolsoc.org.uk/template. cfm?name=3d_seismic. 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 Sedimentary Basins. Geological Society, London, Memoirs, 29, 53-61. SALISBURY, R. S. K., DENLEY, M. R. & DOUGLAS,G. 1996. The Value of Integrating Existing 3D Seismic into Shallow Gas Studies, Offshore Technology Conference Paper No: 7990, Houston, May 1996. 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 Sedimenta~ Basins. Geological Society, London, Memoirs, 29, 187-198. STEWART, S. A. & HOLT, J. 2004. Improved drilling performance through integration of seismic, geologic 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 Sedimenta~ Basins. Geological Society, London, Memoirs, 29, 303-310. STOKER, M. S. 1999. Stratigraphic Nomenclature of the UK North West Margin: 3. Mid- to Late Cenozoic Stratigraphy. British Geological Survey, Edinburgh. STOKER, M. S., AKHURST,M. C., HOWE, J. A. & STOW, D. A. V. 1998. Sediment drifts and contourites on the continental margin off northwest Britain. Sedimentary. Geology, 115, 33-51. STRATAGEM Partners. 2002. The Neogene stratigraphy of the glaciated European margin from Lofoten to Porcupine. STOKER, M. S. (Compiler). A product of the EC-supported STRATAGEM project. STRATAGEM Partners. 2003. STOKER, M. S. (Compiler). Neogene evolution of the glaciated European margin. A product of the EC-supported STRATAGEM project, http//www.stratagemeurope.org. UK Offshore Operators Association. 1997. Guidelines for the conduct of mobile drilling rig site surveys, 1. Versionl.2.
4D/time-lapse seismic: examples from the Foinaven, Schiehallion and Loyal Fields, UKCS, West of Shetland G. BAGLEY,
I. S A X B Y ,
J. M C G A R R I T Y ,
C. PEARSE
& C. SLATER
BP Burnside Road, Farburn Industrial Estate, Dyce, Aberdeen AB21 7PB, UK (e-mail: [email protected])
Abstract: 4D or time-lapse seismic data has become a business of usual tool for reservoir management in high-cost environments for oil production such as the West of Shetlands. 4D seismic can be used to visualize fluid movement in a qualitative/semi-quantitative sense to visualize field behaviour, predict well performance and predict reservoir pressure. Whilst the data may be easily visualized using 3D visualization technology the challenges for the future include the need to obtain quantitative predictions through better quality seismic data.
The Foinaven, Schiehallion and Loyal Fields lie around 180 km west of the Shetland Islands within UKCS blocks 204/19, 20, 24 and 25 in water depths of between 350 and 550 m. The fields are produced through a sub-sea development scheme of horizontal producing wells and sub-vertical water injectors into Floating Production, Storage and Offloading Vessels (FPSO), from where oil is exported in shuttle tankers. This is a particularly high-cost environment where wells cost up to $30M dollars to drill, complete and tie-in; a Production Log can cost up to s (in addition to lost production) but a seismic survey covering all three fields can be acquired and processed for around $3M (Fig. 1). 4D or time lapse seismic technology has become a 'business as usual' tool for reservoir management on the Foinaven, Schiehallion and Loyal fields since the commercial validity of the method was proven by the Foinaven Active Reservoir Management (FARM) experiment in early 1999 (Bouska & O'Donovan 2000; Cooper et al. 1999a,b). Since that time a repeat survey of around 390 km 2 has been acquired over all three fields in 1999 and 2000, referenced back to a pre-production baseline survey acquired in 1993. It can be demonstrated that 4D seismic data can be used to visualize production characteristics of a field, in terms of water and gas movement, predict well performance and as a seismic PLT (Production Log). Currently all of these results are produced in a qualitative/semi-quantitative sense. The challenge for the industry is to calibrate the 4D seismic response to quantitative measurements of pressure and fluid saturation within the reservoir, which can be used to better manage fields and maximize production in the most environmentally friendly and sustainable way. This paper describes some examples of the results of three vintages of 4D seismic data over the fields, discusses methods of interpreting 4D seismic using 3D visualization techniques and raises challenges for the industry as we attempt to obtain quantitative data from 4D seismic.
November 1997, post-production surveys were recorded in November 1998, after 12 months of production. The results from these time-lapse surveys confirmed that production was impacting on the seismic response demonstrating the viability of 4D seismic technology and led to the acquisition of a 390 km 2 survey covering all three fields in 1999 with the deliberate intent of investigating the 4D seismic response. A further survey, with the same geometry, was acquired in 2000. It is the intent of BP and its co-venturers to acquire 4D on an annual basis, given the logistical limitations generated by an intense amount of marine activity in the area.
Qualitative explanation of the 4D response Imagine a homogeneous dipping reservoir section with a producing well up-dip and a water injector down-dip. As oil is extracted from the producing well the pressure within the reservoir around the well decreases. The oil in the fields is at bubble point, so any drop in pressure results in the evolution of gas from solution. At the water injector, the pressure around the well increases (dropping off away from the injector towards the producing well) whilst the water saturation increases. The rock and fluid properties are such that the elastic impedance (EI) (Connolly 1999) of the reservoir decreases progressively as water is replaced by oil, which in turn is replaced by gas. Similarly as the pore pressure increases, the EI decreases. On conventional reflectivity data a drop in EI is seen as a brightening in the top reservoir reflector, whereas an increase is seen as a dimming. The seismic reflection data will, therefore appear to brighten either due to an increase in reservoir G a s - O i l Ratio (GOR) or by an increase in pressure, assuming no phase change (Fig. 3).
Visualizing Field Behaviour West of Shetland seismic history 3D seismic data were first acquired over the BP/Shell acreage in Quad 204 following the discovery of the Foinaven Field in 1992 (Fig. 2). This dataset is a 2000 km 2 survey covering much of the acreage held by the partnership at that time, and was subsequently used to locate the successful Schiehallion and Loyal discovery wells. Following sanction of the Foinaven Field, permanent sea bed hydrophone cables were placed over a portion of the field and in 1995 seismic pre-production data were recorded into these cables and into a conventional surface-tow spread (the FARM experiment). Following first oil from the Foinaven field in
There are areas of the Foinaven field where the seismic amplitude response appears to brighten on the 1999 data compared to the baseline 1993 survey, and then appear to dim once more on the 2000 data (Fig. 4). Looking at the field production history (Fig. 5), it can be seen that the 1999 survey was acquired at a time when the field producing GOR was elevated, due a to a lack of effective voidage replacement through water injection. At the time of the 2000 survey, the GOR had returned to a value close to that of virgin reservoir conditions whilst the water cut for the field had started to increase. The dimming in the 2000 data can, therefore, either be explained in a reduction in GOR, or by an increase in 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 Sedimentary Basins. Geological Society, London, Memoirs, 29, 297-302. 0435-4052/04/$15 9 The Geological Society of London 2004.
298
G. BAGLEY E T AL.
Foinaven, Schiehallion & Loyal
Fig. 1. Location of the Foinaven, Loyal and Schiehallion fields.
Baseline 2000sqkm Exploration 3D
'99 Full Fields 4D ~400sqkm
'00 FlUllFields4D ~400sqkm
FARM'98 FARM'95 Schiehallion'96
1
i!o
9
i Iw
yal
Loyal
Schiehallion
Schiehallion
Foinaven
Foinaven
Discovery
First Oil
Fig. 2. West of Shetland seismic history.
Production Schematic
Pressure - Saturation change
High
3Sw ASg Low
Seismic Amplitude Change
Brighter -6 Peffect
§ P effect
effect
change
Dimmer
Fig. 3. Qualitative explanation of 4D response.
4D/TIME-LAPSE SEISMIC WEST OF SHETLAND
299
Fig. 4. Top reservoir seismic amplitude maps for one of the Foinaven producing reservoir intervals.
,
oooo
,
l
,ooooo r/ n~
...... ~ | 1
....
=11
llA lf ooo
oo
,200
60000
o
"o .~
6110 40000
~" ~
I--o"a'e Water Rate GOR
400
20000
200
0 ,~- ~" g " a, ~" " d o ~ ~ ~'~ ~ ~ ~
0 x" " a~'a~'d" ~ ~'~ ~ o ~
x" ~ " x " " a ~' ~ ~ ~ # o
Fig. 5. Foinaven Field performance and 3D acquisition.
saturation. One of the challenges is to be able to distinguish these two possible effects.
Predicting well performance Three vertical elastic impedance sections along a producing well bore within the Foinaven field show the impact of production on an individual well response (Fig. 6) The impedance drops (seismic brightens) in the 1999 data, interpreted to be due to the increasing GOR, whilst the dimmed data obtained in 2000 probably reflect the increasing water cut. Looking in more
detail, it can be seen that the 1999 data show significant brightening before the producing GOR of the well increases, reflecting the approach of critical gas saturation within the reservoir. The availability of real-time, fast turnaround 4D data would, therefore, have allowed the increasing GOR of the well to have been predicted and, possibly, managed.
Reservoir pressure prediction from seismic On the Schiehallion Field an attempt has been made to calibrate seismic attributes to reservoir pressure and compare this with
300
G. BAGLEY ET AL .
18000'
.
16000 ~
~A
14000" A
\
_~
~.~ 12000"
800
| II A_
700
Ji/\
61}0
10000" 400 8000, 30O ~ 6~0" 411110, 211110-
0
2OO
I1| / 1 19 9
100 9
/
. . .!1
-
0
Fig. 6. Predicting well performance.
dP -1500 500 -200
' 9 9 - '93 Calibrated Pressure Difference Map
0 +200 +1000 ~-3000
Simulator Pressure Difference
Map
Fig. 7. Pressure prediction from 4D seismic data compared to actual reservoir simulation predictions. a reservoir simulator prediction (Fig. 7). The yellow circles show areas of seismic brightening around water injectors. These correspond well to areas that the simulator predicts to have high reservoir pressure. The grey circles show areas around water injectors where no obvious brightening is seen. At these locations the increased injection pressure is either being offset
by an adjacent high rate producing well, or the pressure is being absorbed by the large down-dip aquifer below the oil-water contact. The red ovals show areas where the seismic amplitude is elevated, but the simulator predicts a drop in reservoir pressure. This is a consequence of our inability to fully deconvolve
4D/TIME-LAPSE SEISMIC WEST OF SHETLAND
301
Fig. 8. Improved seismic resolution will allow modelling of geological heterogeneity. pressure and saturation responses. The brightening is due to an increase in gas saturation due to a pressure drop caused by production.
Looking at 4D in 3D 'Traditional' interpretation of 4D seismic data involves the examination of vertical seismic sections, the preparation and manipulation of seismic attributes and perhaps the preparation of complex attribute volumes from time lapse data. As the cycle time between sequential time-lapse surveys decreases and the demands on the interpreter to incorporate more and more real-time well surveillance and production data into the
interpretation increase, we need to look to more sophisticated tools to integrate, visualize and interact with data. Many of the tools to do this already exist--the challenge is to change working practice and start to interpret 4D in 3D (Movies 1-7). More and more frequently reservoir simulator data is being conditioned by 3D seismic data, not just in terms of the structural configuration of the model but also in terms of reservoir characterization directly from seismic. We now have tools to co-visualize seismic and simulator data in the same environment (Movies 8-10). What remains a challenge, however, is the ability to update the simulator model with new data in a simple manner. In the future we can expect the history match of a reservoir
Fig. 9. Processing techniques that preserve the integrity amplitude vs offset will enable more quantitative use of 4D data to predict reservoir saturation and pressure.
G. BAGLEY ET AL.
302
Survey2
Survey I ~1~
I
~
i
J
9
I
Resolution. Fine scale reservoir models are now typically built with l m vertical resolution, in order to attempt to model the degree of heterogeneity present within producing reservoirs. The resolution of seismic data in the West of Shetlands is currently 1 5 - 2 0 m (Fig. 8). 9 AVO lntegri~'. The ability to interpret lithology/fluid and pressure/saturation from seismic data is likely to be highly dependent on the understanding of the A V O response of our reservoirs. The quality of AVO products such as intercept and gradient or near and far offset seismic data volumes is dependent on the quality of NMO correction and the multiple content of the data (Fig. 9). 9 Repeatabilio'. Changes in seismic response due to production are now being shown to exist in fields where it was previously thought to be unlikely. The repeatability of seismic surveys is key to improving the signal-to-noise ratio of seismic data to such a level that 4D or time-lapse seismic can be applied on any field as a matter of routine. Many differences between different vintages of seismic data can now be removed using 'parallel processing' techniques (Fig. 10). Methods for improving acquisition repeatability also exist. Cost m a n a g e m e n t in mature fields, however (where 4D is likely to have most benefit), is critical.
Conclusions Fig. 10. Seismic survey repeatability can be improved by techniques such as 'parallel processing'. simulator model to be constrained by quantitative pressure and saturation data from 4D seismic surveys. Tools are required that allow a simple, visual way to not only visualize but also interact with the datasets--to be able to reach out and touch the model in a 3D visualization environment.
4D or timelapse seismic technology is now being used routinely in a qualitative or semi-quantitative manner. Many examples exist within West of Shetland fields and in many other fields as well, where 4D seismic is making a material difference to reservoir m a n a g e m e n t and development. It is now possible to image fluid and pressure changes within the reservoir. The next step is to move to a fully quantitative, rapid turnaround of 4D interpretation.
Challenges for the future If 4D seismic data are to become a quantitative tool to be used routinely in reservoir management, there are a number of challenges that our experiences West of Shetland have identified as needing to be addressed.
The authors would like to thank the coventurers in the Foinaven, Loyal and Schiehallion fields (Marathon, Shell, Hess, OMV, Murphy and Statoil) and the management of BP for their permission to publish the paper. They would also like to thank the BP NW Europe 4D Implementation Team for their ongoing support. The views expressed in this paper are those of the authors and do not necessarily represent the views of BP or its coventurers.
P r e s s u r e a n d saturation p r e d i c t i o n Over recent years a number of techniques have developed, been published and are actively being marketed to enable the geophysicist to characterize lithology and fluid directly from seismic data. This is a major step forward in building a static reservoir model for exploration and development purposes. For reservoir m a n a g e m e n t using 4D data the next step is to extend these techniques, or develop new ones such that reservoir pressure and fluid saturation can be quantitatively extracted from seismic data.
S e i s m i c data quality The availability of high-quality, low-cost seismic data is fundamental to facilitating the practical use of 4D seismic data quantitatively. Specifically there are three areas of seismic data quality that are in need of improvement.
References CONNOLLY, P. 1999. Elastic Impedance: The Leading Edge, 438-452. BOUSKA, J. & O'DONOVAN, A. R. 2000. Exposing the 4D seismic timelapse signal imbedded in the Foinaven active reservoir management project. Paper OTC 12097 presented at 2000 Offshore Technology Conference, Houston. May 2000. COOPER, M., BOUSKA, J., THOROGOOD, E., O'DONOVAN, A., KRISTIANSEN, P. & CHRXSTJE, P. 1999a. Foinaven Active Reservoir Management: The time-lapse signal. 69th SEG Meeting, Houston USA, Expanded Abstracts, SRC3.6. COOPER, M., BOUSKA, J., WESTWATER, P., THOROGOOD, E., KRISTIANSEN, P. & CHRISTIE, P. 1999b. Foinaven Active Reservoir Management: Towed streamer and buried sea-bed detectors in deep water for 4D seismic. 69th SEG Meeting, Houston USA, Expanded Abstracts, SRC3.4.
Improved drilling performance through integration of seismic, geological and drilling data S.A. STEWART
& J. H O L T
B P Upstream Technology, Dyce, Aberdeen AB21 7PB, UK
Abstract: Unexpected incidents leading to lost time when the rig is on location cause unplanned cost to the hydrocarbon industry of over one billion dollars annually. Processing and interpretation of 3D seismic data usually focuses on reservoir levels. But from a drillers perspective, geological features of the overburden are often more significant than those at reservoir level, since over 90% of the well is typically spent drilling the overburden, coping with a wider variety of challenges than those associated with the reservoir itself. 3D seismic data defines overburden tectonostratigraphy, the framework of a geological model that can be used in well planning to reduce geological uncertainty, surprises and expense along the whole well track. Many technologies applied in reservoir modelling are equally valid in defining overburden features relevant to well planning. The overburden 3D volume can be populated with key parameters for well design, such as pore pressure and geomechanical attributes, though the complexity of the model will often be restricted by well cost and perception of drilling risk. The role of 3D seismic data in forming the tectonostratigraphic framework of multi-attribute, kilometre-scale Earth models, is illustrated here by a number of examples where model sophistication has been scaled to match project requirements. Overburden Earth models also provide a framework where several "academic" research themes, for instance 3D fault geometry, can be put into a commercial context. Construction of overburden models for well planning has also highlighted a number of future geological research areas that could have a significant impact on drilling performance. Some of these, such as hydraulic properties of fault systems, are highlighted here.
No petroleum industry operating company has a global monopoly of best in class drilling performance ('best in class' is defined here as fewest days per depth drilling performance in a given fairway or trend). Gaps between performance of a specific well and best in class arise from the sum of non productive time events (unplanned time where no hole is being made) and 'train wrecks' (major events where drilling has to be abandoned and restarted). Many of these incidents occur in the overburden section between the mud line and top reservoir (Fig. 1)--a zone often neglected in terms of geological input to well planning. The annual cost to the industry of these incidents in offshore wells alone is in the order of one and a half billion dollars (Table 1). This cost estimate discounts the effects of weather and time lost to equipment failure. A further element of the gap between performance of a given well and best in class can be attributed to operational inefficiency (Bond et al. 1996), which has not been factorized in the numbers shown in Table 1. 3D seismic data inherently provides the fundamental framework of a 3D model that spans the entire overburden. This represents a broader use of 3D seismic data than more usual reservoir-specific application, We term the kilometre-scale model of reservoir plus overburden an 'Earth model'. This space can be populated with mappable geological features (e.g. faults, channels), derived attributes (e.g. pore pressure) and drilling history data. In the context of this paper, it is used in conjunction with borehole stability analysis to determine the feasibility of a well track. Secondly, an Earth model is a key tool for refining the well concept by enabling the well track to be planned to take account of specific features in overburden geology (see also Cayeaux et al. 2001 ; Frantes et al. 2001 ). This approach represents a fusion of the shallow hazards mapping concept (e.g. Austin 2004) with the full range of subsurface tools employed at reservoir level, to tackle the geological problems along the length of the well. This paper begins with a short review of the possible elements of an Earth m o d e l - - a wide range of geological, geophysical, petrophysical and geomechanical attributes that can be generated in a given well project. Although the Earth model concept for well planning is globally applicable, the selection of attributes with which to populate a given model depends on the technical and commercial demands of the well or wells project. Devising
a workflow leading to an Earth model that makes best use of available tools and also balances available knowledge and commercial constraints is not necessarily straightforward but is an important decision to be taken by the wells team early in the planning phase. The second part of this paper focuses on the factors that influence selection of a fit for purpose set of attributes. Several Earth models tailored for different drilling challenges illustrate the application of this approach. The final part of the paper discusses some challenges met in the application of 3D Earth models that point to commercially valuable areas of future geoscientific research. We conclude by noting the contrast between application of 3D seismic data for reservoir definition alone, and use of 3D seismic data for planning well trajectories through holistic Earth models of overburden and reservoir.
Earth model components Possible components of an Earth model range from raw data supplemented, with increasing levels of sophistication, by various interpreted and calculated attributes. This part of the paper lists Earth model components arranged in three broad groups representing different levels of sophistication (Fig. 2). All of the Earth models described and illustrated here are designed for visualization using one of the many commercially available 3D visualization software packages (e.g. Sheffield et al. 2000).
Database The most basic type of Earth model consists of whatever digital data is to hand (usually 3D seismic data and some existing well tracks). In practise, very little data is actually in a 'raw' state. Any assumption that data in a given database are reliable and correlatable is invalid without comprehensive quality assurance procedures. 3D seismic data as used on the interpretation workstation is obviously the result of many acquisition and processing steps that are interpretive to various degrees. Although the seismic processing is usually assumed to be
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, 303-310. 0435-4052/04/$15 9 The Geological Society of London 2004.
304
S, A. STEWART & J. HOLT
$million (days x average day rate) 10
1000
"~
40
30
20
" =
performance
spread
~'-.. 2000
,e
overburden
"& 3 0 0 0 I= .,c
4000
5000
9
~.
.
.
-
~.
,
~
~ ~ ~
~-"~11~,
6000
Table 1.
Active rotary drilling rigs offshore Typical day rate Ex-WOW Non Productive time Mechanical failures and problems Total cost to offshore drilling
Fig. 1. Collation of days/depth data for 20 wells from a single trend (onshore South America) showing typical spread in drilling performance. The drilling time has been converted to cost based on an approximate rig day rate. Depths are plotted in metres MD. Wells on the lefthand side of the envelope represent the "best in class' and arrive at the reservoir at costs of approximately $15 million, whereas wells with average performance cost $20-25 million to get to top reservoir. A key observation is that the divergence in drilling cost accrues in the overburden section.
quality assurance after data loading or at project startup. Time for this fundamental work should be factored into project timelines. In addition to seismic and well data, it is possible to include drilling data in a digital database (e.g. Holt etal. 2000; Sanstrom & Hawkins 2000; Wright et al. 2001). Like seismic data, drilling data require some interpretative 'processing' before it can usefully be loaded into an Earth model. Relatively simple interpretation of well data could include, for e x a m p l e , assembling data relevant to constraining fracture gradient and m i n i m u m stress (e.g. White et al. 2002). Figure 3 shows a more involved, but still commonplace, example where m u d losses were recorded with increasing mud weight as a well was drilled. In this case post-well analysis showed that shallow sands in the open hole section, at approximately 7000feet MD, had low measured strength in offset wells, and were the likely problem zone. This example shows it can be misleading to use a drilling problem marker at the depth at which the problem was recorded in the daily drilling reports (i.e. the raw drilling data, c. 11 000 feet MD in this case). Instead, the marker should go several thousand feet shallower in the hole than the bit depth where the problem occurred. This marker should also record key information such as m u d weight, equivalent circulating density (ECD) and hole inclination to facilitate discussion when the model is interrogated. In addition to specific lost time incidents
550 $80 000 18% 50% $1.5 bn
The number of rigs operating offshore is a global total number (all companies) averaged over 2001. The day rate is an indicative approximation. The ex-waiting on weather lost time is an estimate from BP databases. The proportion of non-WOW lost time attributable to 'purely' mechanical failure is an estimate. quality assured, further problems can arise during loading, so survey datums, phase and scaling can contain errors. Nonetheless, well data are usually more likely than seismic to be corrupted by errors by the time the well header, deviation, curves and checkshots are digitally loaded. Deviation and c h e c k s h o t surveys are particularly error prone including systematic depth shifts and spurious data points (see also Williamson & Wilson 2000). The combination of possible processing and data loading errors for seismic and well data ensures that reasonable well ties to seismic (and to the spatial reference frame itself) are unlikely to be achieved without careful
Stress tensor from finite element model
~8 basin model
1
seismic velocity
Pore pressure from wireline logs
WeUboreStability I Geomechanics
..................... ____J~!U_]_L _! _J.
ii II Seismic Interpretation & structural restorations
Shallow Geohazards mapping i
Q~,~~!gital~~se
Wells, Drilling & Seismic data
Fig. 2. Geoscientific toolkit. Three categories of 'Earth model' are described in terms of the data and generated interpretations/results that can populate the Earth model volume. The ~tools' or work elements, are arranged into three broad tiers, with the top tier representing the most sophisticated resource intensive and longest lead time items. Building a workflow from a selection of these items (or others) is a key task that should feature early in a well project. An approach to this is discussed in the text.
305
IMPROVED DRILLING PERFORMANCE
losses
PPg 0
5
10
15
20 j
0 ,
~
1
(D
8000
~
qC ~....
, r'-.,
L
i
~summents
s
/
x
e
8000, a 10000
--
12000 14000
Pressure
~
~
measurements I ' '
or drilling problems, more subtle 'invisible lost time' drilling information can also be extracted and digitized, usually from a correlation of existing wells. For example, correlation between rate of penetration and drilling data such as torque can yield constraints on drilling parameters that are related to specific geological formations or features. The data types described so far, including 3D seismic, represent the basic building blocks of a 3D, digital database. An example visualization of a fairly simple digital database from a North Sea field is shown in Figure 4. Time restrictions and the generally poor quality of legacy databases are such that, in practise, very few wells are planned using Earth models with these fundamental data types fully loaded and quality assured.
Calculated point data At a slightly higher level of sophistication, there are a number of data types that can he derived from the seismic and well data. These include surfaces and faults interpreted on the 3D seismic data and petrophysical parameters derived from the seismic or well data, such as pore pressure (Fig. 5; see also Hillis 1995; Harrold et al. 1999; Japsen 1999). Mapping from related studies can also be imported, such as detailed site investigation mapping of the very shallow ( < 500 m) section, which is usually done on
Fig. 4. Example of database viewed in 3D (see also Sanstrom & Hawkins 2000). Vertical scale exaggerated • 3. Multicoloured surface is top of the reservoir, coloured planes are mapped faults, yellow lines are well tracks and boxes on the yellow lines are drilling incidents, note labels on each box. The boxes can be 'opened' for detailed information about the drilling event. The key utility of this approach to data management is the ability to make rapid spatial correlations between different data types, but there is a spin-off benefit of requiring the underlying database to be quality assured.
--._~sand fltacblre
'~176176
program
interpreted (event depth)
[ Fnactnreg mdient ]
'
-[ ~ ;
2000
Fig. 3. Processing drilling data prior to loading into an Earth model. Horizontal scale is pounds per gallon (ppg) and specific gravity (sg: g/cc). Drilling events are recorded at the depth of the bit, but in long open hole sections the event depth (in this example, where losses were occurring in fractured shallow sands) can be significantly different. The well data needs to be interpreted by a drilling engineer or operations geologist to translate bit depth data to appropriate event depths before display in 3D Earth models.
-
recorded (bit depth)
;
1
:
9 |
9
! 1.5
2
sg high-resolution 2D seismic data and is a regulatory requirement for most offshore wells.
Calculated geocellular data At the highest level of sophistication, the overburden can be broken down into a geocellular model, in the same way that reservoirs are modelled. The cells can then be populated with a range of attributes, such as pore pressure calculated from seismic velocities (e.g. Kan & Swan 2001) or 3D basin models (e.g. Giles et al. 1999; Fig. 6) and geomechanical parameters such as principal stresses (e.g. Wiprut & Zoback 2000; Gambolati et al. 2001). An Earth model populated with these attributes enables rapid comparisons of alternative well paths and designs in terms of pore pressure and fracture gradient in areas of complex overburden geology. Calculation of attributes for a geocellular overburden model can be time intensive, so may not be applicable to every new well that is drilled. On the other hand, database quality assurance, visualization and geomechanical studies are relatively cheap and can be applied to most wells. But given this range of attributes with which to populate an Earth model for well planning, a process is required to determine which level of
306
S.A. STEWART & J. HOLT
Fig. 5. Example of relatively simple Earth model (cf. Fig. 2). Horizontal scale applies to the area of wells at front of model. Vertical scale exaggerated x 3. This represents a framework for firstpass welltrack planning. Many of the elements (pore pressure from wireline, mapped overburden surfaces and so on) are also inputs to wellbore stability analysis.
sophistication is commercially appropriate in a given example. An approach to this is described in the next section of this paper.
W o r k f l o w selection Given that an Earth model and wellbore stability analysis will influence the selection of well concept in addition to assisting the detailed well design, it is essential that the appropriate subsurface workflow is defined as early as possible during well planning. Although it is not straightforward to define a generic process for selecting a fit for purpose subsurface workflow in any project, a framework for use in a well planning context is offered here. It is probably best used to steer a face-to-face discussion of the multidisciplinary well planning team, but it could be the basis of a formulaic approach to workflow choice.
Cost of well The cost of wells for hydrocarbon exploration and production spans four orders of magnitude from hundreds of thousands of dollars for an onshore well (e.g. onshore North America) to up to one hundred million dollars for difficult deep water, highpressure wells (e.g. Caspian sea and deep water Gulf of Mexico). Cost thresholds can be identified to give a first-order constraint on the type and amount of geological work to be carried out for a given well. The work programs for each band would correspond fairly directly to the level of sophistication in work options described in the previous section of this paper, with the option of doing minimum geological work at all for the cheapest wells. Although well cost is a primary parameter, the geological work options can be modified, prioritized and planned by considering additional parameters.
Fig. 6. Example of geocellular Earth model. Vertical exaggeration • 3. Illustrated attribute is the pressure difference between calculated pore pressure (from seismic velocities) and calculated fracture gradient (from overburden thickness). Attribute is shaded where this mudweight window is < 700 psi. In addition to planning specific well tracks, this model could also be used to influence selection of platform location.
IMPROVED DRILLING PERFORMANCE
Project context of well A project in this sense is the cycle of appraising unexplored acreage with a rank wildcat through appraising a discovery well, developing a field, to the very last well on an ultra-mature asset. Considering a well in this project context enables a view to be taken on whether, for example, a level of geological work justified by a single well might be modified in the light of this well being one of a series of ten. So a moderately sophisticated, field wide Earth model might in fact be worthwhile even though the well in question is cheap. Project context also helps prioritize operational aspects of the well, for example early appraisal wells on a major discovery are a key opportunity for data collection in the overburden as well as reservoir sections.
Well concept Existing well control, whether it consists of nearby exploration wells or the previous wells on a field, is a key database of relevant drilling experience. A history of relatively trouble free drilling provides a strong temptation to take a well design 'off the shelf', adding little new work in the planning stage. But the well concept should be compared with the previous wells in order to establish how relevant existing well plans actually are. Inclination and azimuth of a well concept have a major influence on wellbore stability. Wells typically become more complex as a field matures, usually by targeting smaller and longer-stepout locations. High inclination wells can require significantly different mud weights to drill the same formations and depths that may have been penetrated many times in the same field by vertical holes (e.g. Zhou et al. 1996; Willson et al. 19991). Location of a well with respect to the existing control also has a bearing on the requirement for a fresh look at overburden mapping, pore pressure and so on. Discussion of the well concept in terms of previous wells provides an opportunity to challenge assumptions, essentially questioning whether 'business as usual' really is appropriate for the next well.
Overburden geology Complexity. Overburden geology ranges from apparently simple endmembers to complex tectonostratigraphies with extreme pore pressures and exotic features such as mud volcanoes and gas hydrates. Cornplexity of overburden geology gives a direct indication of the sophistication and specific nature of geological work that might be of value in well planning, e.g. overburden mapping, or seismic reprocessing. But quantifying overburden complexity is subjective and drilling problems can and do occur in ostensibly benign overburdens.
307
Hesthammer et al. 2001 ). Each of these examples of uncertainty can be quantified, more accurately defined and sometimes reduced, but their representation in Earth models is not straightforward. This issue is mentioned again in the 'research challenges" section of this paper.
Setting The tectonic setting together with knowledge of the regional present-day stressfield, together with known local variations (e.g. Fuchs & Muller 2001; Grollimund et al. 2001), provides useful context if geomechanical work is identified as being of value to the well planning.
Database.
The state of the database needs to be defined before the workfiow can be specified. Access to the basic database elements discussed earlier in this paper (wells, seismic, drilling data) should be reviewed---of the data that exists, how much is available digitally? How much of this is quality controlled with appropriate levels of interpretation? Other relevant data include existing mapping and specific technical studies that might be relevant to well planning. Any gaps between known and available data represent obvious first steps in the workflow.
Defining the workflow Many of these influences on workflow have obvious or subtle interrelationships, but it is useful to consider the themes separately, for example as laid out here, in order to make sure all possible parameters have been examined. The interrelationships are usually project-specific, for example well cost will usually override geological complexity in determining workflow sophistication, but a multi-well program of wells that are individually inexpensive could justify a more thorough subsurface program. Consideration of all of these themes enables construction of a plan for subsurface work that is scaled to the requirements of the project, rather than ad hoc application of options. The process of defining the geological workflow is also an opportunity to ensure alignment with other work streams that may be relevant to the well, such as shallow gas surveying and mapping, or federal research and development projects. Overburden geoscience for well planning is potentially more complex than reservoir geoscience, not just because of the larger volume of Earth model involved, but the broader range of geological issues concerned. This contrast conflicts with the obvious requirement to devote the majority of the subsurface team resource to reservoir issues and highlights the importance of careful definition of overburden workflow.
Workflow application Uncertainty. Quantifying uncertainty is a significant challenge in any subsurface discipline. There are generally two 'families" of uncertainty. There are intrinsic uncertainties associated with the geological features themselves, for example fault seal, rock strength, stress magnitude, pore pressure, lateral continuity of stratigraphy and so on (e.g. Hesthammer & Fossen 2000). However, the tectonostratigraphic architecture, together with intrinsic uncertainties, is viewed through the 'lens' of seismic data--the resolution of the seismic introduces additional, extrinsic uncertainty. Seismic data quality is a function of parameters such as frequency, signal to noise, migration accuracy and depth conversion (assuming it is positioned and datumed correctly in the first place)--these introduce uncertainty in the location, geometry and even existence of subsurface features, irrespective of intrinsic uncertainties (e.g.
So far this paper has reviewed the components of a geoscientific workflow for well planning, and a framework for selecting the right pieces of work for a specific well. However, even a perfectly defined workflow is of little value unless instigated at the right time. Whatever framework for well planning and execution might be in place in a given company, the right time to begin the geological workflow is such that the results are delivered in time to influence the selection of well concept, i.e. before the detailed well design stage. The time requirement of specific work items, together with dependencies (e.g. wellbore stability should be done after pore pressure forecast) give an indication of the time window required for the geological input to well planning. So in order to allow time for resourcing issues to be
308
S.A. STEWART & J. HOLT
addressed and the work itself to be done and quality assured, workflow selection should ideally begin up to a year before the well is due to spud. The range of geological complexity that can be encountered in overburden mapping for well planning includes disciplines that are actively being researched, and others about which there is relatively little published information. These include potentially interesting and commercially applicable areas for future research--some of these are outlined in the final section of this paper.
Technical challenges H y d r a u l i c p r o p e r t i e s of f a u l t s Faults within reservoirs are usually considered to represent transmissibility barriers, with a degree of inhibition estimated from history matching production data, or from modelled fault rock properties (Knipe et al. 1998; Fisher et al. 2001). But abundant field evidence demonstrates that faults can also be fluid conduits (Eichhubl & Boles 2000; Sibson 2000: Gudmundsson 2001). This paradox appears to be resolved by the notion of fluid flow along faults during episodes of fault movement (Sibson 2000; Gudmundsson 2001) and this idea has been shown to be applicable to hydrocarbon systems (Losh et al. 1999; Finkbeiner et al. 2001). However, there has been very little published research on the effects of drilling through a fault. Two aspects of this problem are: (1) internal structure of the fault zone in terms of fracture orientation and effects on wellbore stability; (2) effect of the drilling fluid pressure on fractures in the fault zone in terms of mud losses and fault reactivation. (1)
Internal fault zone structure ('damage zone' geometry) has been modelled from outcrop, core and numerical standpoints (Knipe et al. 1998; Hesthammer et al. 2000), but these models do not currently take into account 3D variations in fault zone geometry. A typical, fairly simple, extensional fault system is shown in Figure 7. The fault planes in this figure are shaded according to a curvature attribute, but a number of other generic attributes are apparent to fully parameterize a fault system. These parameters could include displacement, displacement gradient, lithology, fault rock type (e.g. shale gouge
(2)
ratio), proximity to fault intersections, proximity to fault tips and fault population characteristics. This list is not intended to be exhaustive but to show that there is a significant gap between the present understanding of fault damage zone geometry and the level of understanding required to predict borehole stability within fault zones based on predrill parameters mapped on 3D seismic data. If drilling fluid pressure exceeds the minimum principal stress, then suitably oriented fractures (e.g. normal to $3) can open and losses will be experienced (e.g. Ito et al. 2001). Predicting the amount of losses would require knowledge of the fracture population distribution, including orientation and connectivity within the damage zone, as discussed above. Such a prediction would also require some assumptions about stressfield magnitude and orientation in the vicinity of the fault zone, both of which fluctuate with time around active faults (Kenner & Segall 1999), There is also the possibility that the magnitude of stress on the fault plane is such that a small increase in fluid pressure could cause reactivation of the fault (Finkbeiner et al. 2001).
The time dependent evolution of fault zone permeabilities in current seismic valving models (Sibson 2000; Gudmundsson 2001) do not offer any insight into the timescales over which fault permeability can vary before and after fault movement. These models do not rule out fluid-filled fractures and 'rubble zones' existing for some time after fault movement leading to significant mud losses on drilling. So there seems to be scope for much research in fault zone hydraulic properties as a function of mappable parameters. Even small steps forward, such as recognition that fluid flow focuses on fault intersections (e.g. Losh et al. 1999), represent useful rules of thumb for welltrack planning though faulted overburdens.
Depicting geological uncertainty in 3 D m o d e l s 3D Earth models as discussed in this paper are designed to show the geometry and spatial relationships of geological and manmade objects, but most visualization software is designed to render crisp images that do not take into account uncertainty on the position and form of the objects in the model (e.g. Fig. 8a). If the model shown in Figure 8a were redrawn to illustrate the degree of confidence in the depicted form and position of the various surfaces, it would look quite different, since the land surface is directly measured whereas all of the subsurface horizons and faults are in this case extrapolated from outcrop and constrained only by 2D structural restoration (Fig. 8b). There are a number of options for rendering uncertainty in 3D models (e.g. Davis & Keller 2000; Pang et al. 1997), but these have yet to be implemented in standard geoscience visualization software.
Unified g e o m e c h a n i c a l a p p r o a c h to o v e r b u r d e n and reservoir
Fig. 7. Typical extensional fault system, here seismically imaged in the Orinoco delta. Vertical scale exaggerated x 1.5. The 3D geometry is more complex than a 2D seismic slice might suggest. It is difficult to avoid drilling faults in this field, and a research challenge is identifying the best location to pass through a given fault. In other words, how can a fault system be parameterized in 3D in terms of hydraulic properties and wellbore stability?
Geomechanics and supporting studies have application to several different phases of well projects, from welltrack planning as discussed in this paper (e.g. Willson et al. 1999), to completion engineering, where for example perforating and sand production are influenced by geomechanical parameters (e.g. Wiprut & Zoback 2000). The disparate nature of customer disciplines can lead to separate geomechanical studies being conducted for different project elements. Use of an integrated geocellular model of the reservoir and overburden as a database of geomechanical data would lead to consistency and efficiency in geomechanical studies. The benefits to geomechanics are just
IMPROVED DRILLING PERFORMANCE
309
geological problems faced along the entire length of a welltrack and associated issues like uncertainty give rise to a number of potentially commercially valuable research areas, Some of these appear to be particularly complex issues, and others are essentially software development problems. A few have been described here to illustrate areas of leverage against the drilling performance challenge. This paper benefited from review by B. Austin and A. Welbon and input from numerous colleagues, particularly B. Barley, L. Beacom, S. Duppenbecker, T. Harrold and P. Mitchell.
References
Fig. 8. (a) Structural framework of bedding and fault surfaces constructed from outcrop control and poor seismic in an area with no well penetrations. Vertical scale exaggerated • 2.5. The land surface DEM is rendered transparent. The model looks precise but is a single, deterministic realization that does not convey any information about the degree of uncertainty in the location, form or existence of the structures. (b) There are several options for rendering uncertainty, here the same model is shown with transparency and blur linked to uncertainty. This is a more realistic representation of the degree of confidence in the subsurface structure and could be a more meaningful approach to sharing geological understanding across disciplines, for example to drilling engineers charged with well planning. a spin-off of the benefits of a 'full-field' geocellular m o d e l - several software platforms are capable of this scale of modelling, but practical application is probably limited by computer power at this time.
Summary A major proportion of capital expenditure for oil and gas companies arises from drilling wells. Poor drilling performance relative to best in class peers costs each supermajor in the order of hundreds of millions of dollars annually, and a large proportion of these costs arise from drilling plans failing to take account of predictable geological features and conditions in the overburden section. In this paper we set out a framework for subsurface work that can be undertaken as part of a well planning project. We also present a generic approach to the awkward task of choosing the appropriate pieces of subsurface work for a given well, bearing in mind the large range in well costs and geological challenges around the world. It is emphasized that there is only value in planning the geological work programme this carefully if it is done at the fight t i m e - - s u c h that the results can actually be reflected in the well design. This may require the workflow selection to have been made up to a year before planned well spud. The breadth of
AUSTIN, B. 2004. Integrated 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. & UNDERH1LL,J. R. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 279-296. BOND, D. F., SCOTT, P. W., PAGE, P. E. & WINDHAM, T. M. 1996. Applying technical limit methodology for step change in understanding performance, SPE, Paper number 35077. CAYEAUX, E., GENEVOIS, J.-M., CREPIN, S. & THIBEAU, S. 2001. Well planning qualit3' improved using co-operation between drilling and geosciences, SPE, Paper number 71331. DAVlS, T. J. & KELLER, C. P. 2000. Modelling and visualizing multiple spatial uncertainties. Computers and Geosciences, 23, 397-408. EICHHUBL,P. & BOLES,J. R. 2000. Rates of fluid flow in fault systems-evidence for episodic rapid fluid now in the miocene monterey formation, coastal California. American Journal of Science, 300, 571-600. FINKBEINER, T., ZOBACK, M., FLEMINGS,P. & STUMP, B. 2001. Stress, pore pressure, and dynamically constrained hydrocarbon columns in the South Eugene Island 330 field, northern Gulf of Mexico. AAPG Bulletin, 85, 1007-1031. FISHER, Q. J., HARRIS, S. D., MCALLISTER,E., KNIPE, R. J. & BOLTON, A. J. 2001. Hydrocarbon flow across faults by capillary leakage revisited. Marine and Petroleum Geology, 18, 251-257. FRANTES,T. J., DRENNEN,W. T. III, MAY, S. R. & UTSKOT,S. M. 2001. Impact of volume interpretation and visualization technologies on upstream business activities, Offshore Technology Conference, Houston, Paper Number 13294. FUCHS, K. & MULLER,B. 2001. World Stress Map of the Earth: a key to tectonic processes and technological applications. Naturwissenschaften, 88, 357-371. GAMBOLATI,G., FERRONATO,M., TEATINI.P., DEIDDA,R. & LECCA,G. 2001. Finite element analysis of land subsidence above depleted reservoirs with pore pressure gradient and total stress formulations. International Journal for Numerical and Analytical Methods in Geomechanics, 25, 307-327. GILES, M. R., INDRELID, S. L. ET AL. 1999. Charge and overpressure modelling in the North Sea: multi-dimensional modelling and uncertainty analysis. In: FLEET, A. J. & BOLDY, S. A. R. (eds) Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference. Geological Society, London, 1313-1324. GROLL1MUND,B., ZOBACK, M. D., WIPRUT, D. J. & ARNESEN,L. 2001. Stress orientation, pore pressure and least principal stress in the Norwegian sector of the North Sea. Petroleum Geoscience, 7, 173-180. GUDMUNDSSON, A. 2001. Fluid overpressure and flow in fault zones: field measurements and models. Tectonophysics, 336, 183-197. HARROLD, T. W. D., SWARBRXCK,R. E. & GOULTV, N. R. 1999. Pore pressure estimation from mudrock porosities in Tertiary basins, southeast Asia. AAPG Bulletin, 83, 1057-1067, HESTHAMMER, J. & FOSSEN, H. 2000. Uncertainties associated with fault sealing analysis. Petroleum Geoscience, 6, 37-45. HESTHAMMER, J., JOHANSEN, T. E. S. & WATTS, L. 2000. Spatial relationships within fault damage zones in sandstone. Marine and Petroleum Geology, 17, 873-893.
310
S.A. STEWART & J. HOLT
HESTHAMMER, J., LANDRO, M. & FOSSEN, H. 2001. Use and abuse of seismic data in reservoir characterisation. Marine and Petroleum Geology, 18, 635-655. HILLIS, R. R. 1995. Quantification of Tertiary exhumation in the United Kingdom southern North Sea using sonic velocity data. AAPG Bulletin, 79, 130-152. HOLT, J., WRIGHT, W. J., NICHOLSON, H., KUHN-DE-CHIZELLE. A. & RAMSHORN, C. 2000. Mungo Field: Improved communication through 3D visualization of drilling problems, SPE, Paper number 62523. ITO, T., ZOBACK, M. D. & PESKA, P. 2001. Utilization of mud weights in excess of the least principal stress to stabilize wellbores: Theory and practical examples. SPE Drilling & Completion, 16, 221-229. JAPSEN, P. 1999. Overpressured Cenozoic shale mapped from velocity anomalies relative to a baseline for marine shale, North Sea. Petroleum Geoscience, 5, 321-336. KAN, T. K. & SWAN, H. W. 2001. Geopressure prediction from automatically-derived seismic velocities. Geophysics, 66, 1937-1946. KENNER, S. & SEGALL, P. 1999. Time-dependence of the stress shadowing effect and its relation to the structure of the lower crust. Geology, 27, 119-122. KNIPE, R. J., JONES, G. & FISHER, Q. J. 1998. Faulting, fault sealing and fluid flow in hydrocarbon reservoirs: an introduction. In: JONES, G., FISHER, Q. J. & KNIPE, R. J. (eds) Faulting, Fault Sealing and Fluid Flow in tlydrocarbon Reservoirs. Geological Society. London, Special Publications, 147, vii-xxi. LOSH, S., EGLINTON, L., SCHOELL, M. & WOOD, J. 1999. Vertical and lateral fluid flow related to a large growth fault, south Eugene Island Block 330 field, offshore Louisiana. AAPG Bulletin, 83, 244-276.
PANG, A. T., WITTENBRINK,C. M. & LODHA, S. K. 1997. Approaches to uncertainty visualization. Visual Computer, 13, 370-390. SANSTROM, W. C. ~ HAWKINS, M. 2000. Perceiving drilling learning through visualization, SPE, Paper number 62759. SHEFFIELD, T. M., MEYER, D., LEES, J., PAYNE, B., HARVEY, E. L. & ZEITLIN, M. J. 2000. Geovolume visualization interpretation: A lexicon of basic techniques. The Leading Edge, May, 518-522. SIBSON, R. H. 2000. Tectonic controls on maximum sustainable overpressure: fluid redistribution from stress transitions. Journal of Geochemical Exploration, 69, 471-475. WHITE, A. J., TRAUGOTT, M. O. & SWARBR1CK,R. E. 2002. The use of leak-off tests as means of predicting minimum in situ stress. Petroleum Geoscience, 8, 189-193. WILLIAMSON, H. S. & WILSON, H. F. 2000. Directional drilling and earth curvature. SPE Drilling & Completion, 15, 37-43. WILLSON, S. M., LAST, N. C., ZOBACK, M. D. & Moos, D. 1999.
Drilling in South America: A wellbore stability approach for complex geological conditions, SPE, Paper number 53940. WIPRUT, D. ~ ZOBACK, M. 2000. Constraining the stress tensor in the Visund field: Norwegian North Sea: Application to wellbore stability and sand production. International Journal of Rock Mechanics and Mining Sciences, 37, 317- 336. WRIGHT, W., HOLT, J., REZMER-COOPER,I., MINTON, R. & RAMSHORN, C. 2001. Vizualisation: Game changer or gimmick?, SPE Paper number 67754. ZHOU, S. H., HILLIS, R. R. & SANDIFORD, M. 1996. On the mechanical stability of inclined wellbores. SPE Drilling & Completion, 11, 67-73.
4D seismic imaging of an injected CO2 plume at the Sleipner Field, central North Sea R.A.
CHADWICK
1, R . A R T S
2, O . E I K E N
3, G . A .
KIRBY
l, E . L I N D E B E R G
4 & P. ZWEIGEL
4
1British Geological Survey, Kingsley Dunham Centre, Keyworth, Nottingham, United Kingdom, NG12 5GG (e-mail: rach @bgs. ac. uk) 2Netherlands Institute of Applied Geoscience TNO--National Geological Survey, Kriekenpitplein 18, PO BoA" 80015, 3508 TA Utrecht, The Netherlands 3Statoil Research Centre, Rotvoll, N-7005 Trondheim, Norway 4Sintef Petroleum Research, N-7465 Trondheim, Norway
Abstract: CO2 produced at the Sleipner field is being injected into the Utsira Sand, a major saline aquifer. Time-lapse seismic data acquired in 1999, with 2.35 million tonnes of CO2 in the reservoir, image the CO2 plume as a number of bright subhorizontal reflections. These are interpreted as tuned responses from thin (< 8 m thick) layers of C02 trapped beneath intrareservoir shales. A prominent vertical 'chimney" of CO2 appears to be the principal feeder of these layers in the upper part of the reservoir. Amplitude-thickness scaling for each layer, followed by a layer summation, indicates that roughly 80% of the total injected CO2 is concentrated in the layers. The remainder is interpreted to occupy the feeder 'chimneys' and dispersed clouds between the layers. A prominent velocity pushdown is evident beneath the CO2 accumulations. Velocity estimation using the Gassmann relationships suggests that the observed pushdown cannot readily be explained by CO2 present only at high saturations in the thin layers: a minor proportion of low saturation CO2 is also required. This is consistent with the layer volume summation, but significant uncertainty remains.
CO2 separated from natural gas produced at the Sleipner field in the central North Sea (Norwegian block 15/9) is currently being injected into the Utsira Sand, a major saline aquifer some 26 0 0 0 k m 2 in area (Fig. 1). Injection started in 1996 and is planned to continue for about 20 years, at a rate of about one million tonnes per year. The Saline Aquifer CO2 Storage (SACS) project aims to monitor the injected CO2 by time-lapse seismic methods. Baseline 3D seismic data were acquired in 1994, prior to injection. A first repeat survey, covering some 26 km 2, was acquired in October 1999, with 2.35 million tonnes of CO2 in the reservoir, and a second repeat survey was acquired in September 2001 with 4.26 million tonnes of CO,, in situ. Current findings from the 1994 and 1999 surveys are described here, with vivid 4D seismic images of the CO, plume being used to illustrate the ongoing interpretive and modelling work.
Background to the injection operation The Utsira Sand forms part of the M i o - P l i o c e n e Utsira Formation (Gregersen et al. 1997; Chadwick et al. 2001). It is axially situated within the thick post-rift succession of the central North Sea, forming a basin-restricted lowstand deposit of considerable extent, over 400kin from north to south and typically 5 0 - 1 0 0 k m west to east. Sleipner lies towards the southern limit of the Utsira Sand (Fig. la), where the reservoir is some 800-1000 m deep and 2 0 0 - 3 0 0 m thick. Core measurements, petrographic analysis and well logs (Zweigel et al. 2001) show the sand to be clean and largely uncemented with porosities in the range 0.30-0.42, typically 0.37. Well logs from the Sleipner area however resolve thin beds of intrareservoir mudstone or shale, characterized by high ",/-ray readings (Fig. 2). The shales range in thickness from less than a metre to more than five metres, but with a well-defined modal peak at just over one metre (Zweigel et al. 2001). The Utsira Sand is overlain by the Nordland Formation (Isaksen & Tonstad 1989), which mostly comprises prograding deltaic wedges of Pliocene age. These generally coarsen upwards, from mudstones in the deeper, axial parts of the basin to silt and sand in the shallower and more marginal parts.
In the Sleipner area the lowest caprock unit, a 5 0 - 1 0 0 m thick silty mudstone, forms the immediate reservoir seal. CO2 was injected into the Utsira reservoir at a depth of 1012 m below sea level (bsl), beneath a gentle domal closure of some 12 m relief (Fig. 3a). The CO2 occupies an enveloping 3D volume which is here termed the 'CO2 plume'. The plume lies between the injection point and the top of the reservoir at about 800 m bsl, where estimated formation temperatures, based on a downhole measurement, are 36~ and 29~ respectively. At these conditions the CO, is in the form of a supercritical fluid with a roughly constant density of around 7 0 0 k g m -3, (the tendency for density to decrease with increasing temperature is counterbalanced by the increasing pressure, Span & Wagner 1996). The mass of 2.35MT injected by October 1999 would correspond therefore to a volume of about 3.3 x 106m 3 at reservoir conditions. Uncertainty in reservoir temperature and the effect of minor impurities such as methane, permit the possibility of lower densities, perhaps down to about 6 0 0 k g m -3, with a corresponding in situ C02 volume of 3.8 x 106m ~. Irrespective of the precise reservoir conditions, the principal driving force for the migration of CO2 up through the reservoir is buoyancy, due the density difference, Ap, between CO,_ and brine.
Reflectivity of the CO2 plume Introducing CO 2 into the Utsira reservoir has a dramatic effect on reflectivity. The 1994 pre-injection data (Fig. 3a), show moderate reflections from the top and base of the reservoir, with much weaker intra-reservoir events (the midUtsira reflection is a sea bed multiple of the prominent events near to the top of the reservoir). In contrast, the 1999 data show a clear image of the CO2 plume with strong reflections at a number of levels within the reservoir (Fig. 3b). These are interpreted as layers of CO2 accumulating or 'ponding' beneath the thin intra-reservoir shales. The CO2 related reflections do not show the gentle antiformal geometry of the Utsira stratigraphy as imaged on the 1994 data, but rather show a downward pointing V-profile, which becomes more pronounced down through the reservoir. This is interpreted
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, 311-320. 0435-4052/04/$15 9 The Geological Society of London 2004.
R.A. CHADWICK ETAL.
312
5o km
~
o
lOO
..... i ....................
~
i ......
i Shetlandl Isles
--
,:~ ~ / ".~,
-
.........
"
.,~
2oo
~'1~
B oo ...........
! .................... ii ................................. "i i
La...............
/
/
i
~
=,
..................
Fig. 1. (a) Limits and thickness of the Utsira Sand and location of the Sleipner injection point. (b) Cartoon of the Sleipner CO2 injection operation.
. . . . . . .
as an effect of velocity pushdown within the plume. A timeslice through the plume on difference data (1999 minus 1994) shows it to be markedly elliptical in plan, elongated N N E - S S W , with a major axis of about 1800m and a minor axis of some 6 0 0 m (Fig. 3c). The difference data also show complex structure within the plume, including vertical linear zones of amplitude reduction and relatively isolated volumes of CO2 (Fig. 3d). Reflections on the difference data beneath the injection point are interpreted as artefacts. These have two main causes: multiple energy (principally the sea bed multiple) from the overlying plume, and 'difference' signal generated by the effects of velocity pushdown rather than by changes in reflectivity. Up to twelve individual reflection horizons can be identified in the plume (Fig. 4). These were picked on wavelet troughs, signifying negative acoustic impedance contrasts, which correspond approximately to the top of each CO2 layer (see below). The probable presence of multiple energy and the likelihood that the plume reflections represent composite interference wavelets makes it difficult to produce an unequivocal horizon interpretation and other, more conservative interpretations with somewhat fewer horizons cannot be discounted. Some of the interpreted horizons are large features, comparable in plan area to that of the whole plume, others form much smaller outliers. The two small uppermost horizons are interpreted to lie fight at the top of the Utsira Sand, directly beneath the caprock (Fig. 4).
15/9-16
gr
sflu
injection well 15/9-A16
gr
rt
15/9-13
gr
sflu
Thin bed effects The twelve picked horizons have a total plan area of about 2.9 x 106 m 2. Taking an injected CO2 volume of 3.3 x 10 6 m 3, and a mean reservoir porosity of 0.37, if the CO-, were wholly distributed as reflective sub-horizontal layers, these layers would, on average, be only about 3 m thick. Because CO2 is also interpreted to be present as chimneys between the layers (see below), the actual average layer thickness would be less than 3 m. With layer thicknesses generally beneath the limit of seismic resolution (h/4, - 8 m for these data), the observed CO2 reflectivity is likely to be largely a consequence of thinlayer interference. With thin-layers, reflection amplitude is related directly to layer thickness, increasing from zero at zero layer thickness, to a maximum at the tuning thickness (Fig. 5). Thus, observed amplitudes on the picked horizons, which tend to increase systematically inwards, from zero at their outer edges to a maximum value near their centres (e.g. Fig. 6a), are consistent with a tuned response from thin layers of CO2 which thicken from zero at their outer edges to a maximum in the axial part of the plume, within the structural closure. The highest amplitudes moreover, are encountered in the central parts of the most areally extensive horizons. Dominantly thinlayer reflectivity is also consistent with the observed seismic waveforms, which comprise mostly interference doublets, rather than the near-symmetrical, near-zero phase processed
16/7-3
gr
rt
Fig. 2. Geophysical logs ('y-ray and sonic) in wells close to Sleipner. The Utsira Sand has much lower "v-ray (gr) signature than the caprock succession. 3,-ray peaks within the sand (main peaks arrowed) are interpreted as thin beds of shale. Note the injection well is strongly deviated and the drilled sequence will differ from that at the plume location.
4D SEISMIC IMAGING OF A CO, PLUME
313
-
W
W
Top Utsira Sand
\
900-
I 1000-"
/ a
11oo-,.
9 IP
Base Utsira Sand
ms
b
lkm
N
t Fig. 3. Time-lapse seismic images of the CO2 plume. (a) Crossline through the 1994 dataset prior to injection--IP denotes injection point. (b) Crossline as in (a) through the 1999 dataset, showing enhanced reflectivity and velocity pushdown. (c) Time-slice at 950 ms ( - 870 m) through the difference dataset (19991994). Blue denotes a negative acoustic impedance contrast. (d) Oblique line through the difference dataset (19991994) showing complex plume structure (arrow denotes position of time-slice). Enhanced amplitude display with re&yellow denoting a negative acoustic impedance contrast.
/
lkm
SSW
NNE
,,.. ~
"
900-
~, ~ ~ "
,o
I000-
w w
=~"
'
9
.~
~ " ,,o ,m
1100ms
9
.=
a
io-
b
input wavelet (the latter is well displayed at simple acoustic interfaces such as the sea bed). Assuming that the maximum amplitudes observed in the plume correspond to the tuning thickness of about 8m (in practice they may correspond to a somewhat lesser thickness), and making a simplifying linear interpolation, amplitude can be scaled directly to layer thickness for each horizon. For example, in the layer corresponding to Horizon X, maximum amplitudes are comparable with the highest amplitudes observed in the plume, so its maximum calculated thickness approaches 8 m (Fig. 6b).
C02 chimneys Intriguing detail is visible in parts of the plume (Fig. 7). Beneath the gentle closure at the top of the Utsira Sand, the main reflections show the characteristic V-profile velocity pushdown, which builds rapidly downwards. In the southern part of the plume, a vertical column of reduced horizon reflectivity corresponds precisely to a more localized pushdown, itself superimposed on the broader V-profile. The amount of this localized pushdown increases rapidly downwards from the reservoir top to reach a maximum of about 20ms at about
1kin
d
,,
1km
970 ms two-way time. It does not clearly change beneath this, but tends to smear somewhat, becoming rather diffuse at base Utsira level. The feature is interpreted as a vertical 'chimney' of moderate or high CO2 saturation, in the upper part of the plume. This causes a rapid build-up of pushdown within the chimney itself and a pushdown shadow below. Similar, though much less prominent seismic features seen elsewhere in the plume are interpreted as smaller CO2 chimneys. The relationship of the main CO2 chimney to the surrounding reflective layers is exemplified by Horizon Y (Fig. 7). The horizon dips in two-way time towards the chimney, due to the velocity pushdown in the axial parts of the plume. Horizon Y is the most extensive individual reflection within the plume and in plan view shows marked lateral amplitude variations (Fig. 8a). The chimney is visible as a 'hole' in the amplitude map where the horizon autotracker has not been able to pick the event (Fig. 7). It is surrounded by high amplitude reflections, particularly to the east, where a 'stream' of enhanced reflectivity is prominent in the east and north. A perspective view of the horizon (Fig. 8b), with reflection amplitudes draped over its two-way time topography, shows the prominent pushdown depression around the chimney, with linear ridgelike features to the north. The ridge crests correspond to markedly enhanced seismic amplitudes that are interpreted as due to small changes
R.A. CHADWICK ETAL.
314
the outer limit of the 95% confidence ellipse of the well position. It is tempting to suppose that the chimney location is linked directly to that of the injection point, however, because of the positional uncertainty, some form of pre-existing geological control cannot be ruled out. J
Verification aspects '
." ~
Top Utsira Sand
The 4D data provide two essentially independent means of quantitatively assessing the amount of CO2 in the subsurface.
Thin layer summation
"Horizon X
I
The capillary pressure, p~., between the formation brine and the injected CO2 will cause the CO2 saturation, Sco,, to vary with height, h, in each CO2 layer. The gradient can be computed by balancing the buoyancy, Ap,g,h, with the capillary pressure. In SI units:
Ap.g.h = Pc -----810.35( 1 - Sco :)-~
I/.,. (" Fig. 4. 3D view of the picked horizons within the 1999 plume superimposed on part of an inline. Horizons X and Y labelled. in thickness of the CO2 layer. Thus CO2 migrating laterally away from the chimney, beneath a thin shale, forms thicker 'ponds' beneath local topographic culminations. These give rise to higher reflection amplitudes as the CO2 layer approaches the tuning thickness (Fig. 9). Their ridgelike morphologies may be put down to primary sedimentary, channel-related structures within the Utsira Sand, or, perhaps more likely, to differential compaction within what remain largely unconsolidated strata (Zweigel et al. 2001). Whatever their underlying cause, it is likely that the amplitude variations are effectively mapping thickness changes in the CO2 layer down to less than one metre, which more-or-less corresponds to the noise threshold. It is notable that the main CO,, chimney is situated nearly, though not perfectly, above the injection point (Fig. 7), close to
[ ';i ~ - L " '~- L ~_ h_. ~_~. . . . [ , i: ~i . ! ........ ~t::liWatersaturated sand
L_
a:ii
The capillary pressure-saturation relationship was determined by centrifuge experiments on core material from the Utsira Sand (SACS unpublished data). The variation of Scot with h was thereby computed and also the average value of Scot for a range of layer thicknesses (Fig. 10). Using this information the layer thicknesses derived for each reflecting horizon (e.g. Fig. 6b) can be converted to net CO2 thickness (e.g. Fig. 6c). This was carried out at each grid point (CMP), by multiplying the layer thickness by the average CO2 saturation (Fig. 10), and by the reservoir porosity. Summation of these net thicknesses for each layer gives a first-order estimate of the total amount of CO2 imaged by the seismic data. For the interpretation presented here (Fig. 4), the total volume in thin layers is estimated at about 2.6 x 106m3; about 80% of the known injected volume. A number of factors, alone or in combination, will contribute to uncertainty in this figure. These include uncertainty in the horizon interpretation (including interference between adjacent tuning wavelets), errors in the simple amplitude to thickness conversion, the presence of dispersed (essentially unreflective) COz in between the reflective layers, dissolution of CO2 into the formation water and amplitude loss in the deeper plume due to various forms of signal attenuation.
Velocity pushdown The velocity pushdown of reflections beneath the CO2 plume (Fig. 11) provides an alternative means of estimating CO2 volume in situ. By interpreting the base Utsira Sand beneath the
C02 saturated , ........ ~_L
sand ~ ~
j
.....
I I mnea
response
layer
tuning thickness
resolved
(1)
Fig. 5. (a) Acoustic impedance model with a thin shale overlying a layer of CO2-saturated sand whose thickness increases from left to right. (b) Seismic response of the above model. Green pick marks seismic trough corresponding to the top of the CO2-saturated layer. In the Sleipner plume most reflectivity is a tuning response from CO,-saturated layers less than 8 m thick, where amplitude is controlled by layer thickness.
4D SEISMIC IMAGING OF A CO2 PLUME
315
0
20000
10000
0
%
8m
0
0
3m
4ms
N
Fig. 6. Plan views of Horizon X. (a) reflection amplitude (b) thickness of rock-CO2 layer (c) net thickness of CO2, assuming qb = 0.37 and saturationthickness function (d) velocity pushdown (ms) due to layer.
T 500m I
II
N
top Utsira Sand ,
900-Horizon Y
Iocalised pushdown
Fig. 7. Inline through the 1999 dataset. Note velocity pushdown at base of reservoir and also more localized pushdown interpreted as caused by a 'chimney' of CO2 in the upper part of the plume. Also note lateral amplitude variations on the individual reflections (e.g. Horizon Y with autopick). Blue denotes a negative acoustic impedance contrast. IP, approximate location of injection point (corrected for pushdown).
Horizon X
1000--
~,,
pushdown
/ base Utsira Sand
1100--" ms
IP~ ,,, 9 ~
4~ 84184 Ikm
R.A. CHADWICK ET AL.
316
W
N
t
E i
/; rr;
eh
q
mmlmm
I
chimney
d, b 500 m
Xline plume on both the 1994 and 1999 surveys it is possible to map the pushdown beneath much of the CO2 plume (Fig. llc). Significant uncertainty arises however because reflections on the 1999 data are locally degraded and the mapping shows some instability beneath the outer parts of the plume where pushdown values are small. An alternative approach is to map the pushdown automatically by cross-correlating a window of the sub-plume reflections on the 1994 and 1999 surveys, and thereby deriving a pushdown time-lag for each seismic trace (Fig. 11 d). Pushdown values derived in this way are more stable than the interpreted map beneath the outer parts of the bubble, but high pushdown values directly beneath the main CO_, chimney are not resolved, due to degradation of the crosscorrelogram by poor signal to noise ratios. Irrespective of the method of derivation, the pushdown anomaly is elliptical in plan, with time-lags in excess of 20 ms widely observed beneath the central parts of the plume and locally in excess of 40 ms. The total amount ofpushdown caused by the plume can be expressed as the individual time-lags at each CMP trace (or bin), summed over the entire anomaly. This is termed the Total Area Integrated Time Delay (TAITD). The pushdown mapped by interpretation of the Base Utsira Sand (Fig. 1 lc), has a TAITD of about 11 000m2s, whereas the pushdown from crosscorrelation (Fig. 1 l d) has a TAITD value of about 9200 m2s. Optimal mapping of the pushdown would probably incorporate both cross-correlation and local manual picking with a likely intermediate value of TAITD. The amount of pushdown can be related algebraically to the column of CO2 in the overlying strata (Fig. 12). stronger reflections at
UtskaSand
shale la~/er
1O
,
Fig. 8. Images of Horizon Y. (a) Plan view of reflection amplitude (highest amplitudes in yellow). (b) Perspective view from the SSE. Display shows reflection amplitude blue (low) to red (high), draped over two-way time topography. Note the prominent velocity pushdown depression around the chimney, and the high amplitudes corresponding to the ridge-crests farther north.
For each grid point (CMP bin):
AT.~x.~y = 2(Vsw - Vsc~
where AT is the time delay at each trace (T99 survey -- T94 survey); 8x is the x-dimension of bin (12.5 m for the SACS data); 8y is the y-dimension of bin (12.5 m for the SACS data); Vsw is the seismic velocity of water-saturated rock; Vsco, is the seismic velocity of rock saturated with CO_, (at saturation Sco,); Z is the thickness of rock saturated with CO2 (at saturation Sco,). Substituting reservoir porosity (r and CO, saturation (Sco3 and summing all the grid points over the whole pushdown anomaly:
EAT.~x.~y = 2(Vsw - Vsco2)'(Total injected volume of CO2) (Vsw'Vsco:)'r (3) In principle therefore, the TAITD, ~AT.Sx.Sy, can be related directly to the total volume of CO2 in the plume. In practice however there are significant uncertainties, particularly with respect to the expression below, here termed the 'Pushdown Factor': 2(Vsw -
Vsco~ )
( Vsw Vsco~)'r 1.0
if)8 0.5
// / /
/
point saturation
UtsiraSand
Fig. 9. Schematic representation of the vertical CO2 chimney acting as a feeder to a laterally migrating layer of CO2 trapped beneath a thin bed of shale. Slight undulations in the shale give rise to ponds of thicker CO2 and an enhanced 'tuning' response.
(2)
(Vsw'Vsco~_)
- - - - - - layer average saturation
0.0 height above layer base / layer thickness (m)
Fig. 10. Variation of average CO2 saturation with layer thickness.
4D SEISMIC IMAGING OF A CO2 PLUME
317
Top Utsira Sand
\
" 900 "
......
P
c
Y ~
0,,~" ,i,
i~I IwB"
-1000
-1100
/
Base Utsira Sand 1994
Base Utsira Sand 1999 lkm
b
lkm
TAITD ~11000m~s
Fig. 11. Velocity pushdown beneath the CO2 plume. (a) 1994 inline showing base Utsira Sand pick. (b) 1999 inline showing base Utsira Sand pick (1994 pick for reference). Note greater pushdown beneath chimney. (e) Map of two-way time pushdown based on manual interpretation of Base Utsira Sand (note high pushdown values SE of the injection point, beneath the main chimney). (d) Map of pushdown based on cross-correlation of a window of events beneath the plume (note lack of high pusbdowns associated with the chimney). IP denotes injection point. Black outline denotes outer edge of the plume reflectivity envelope.
The Pushdown Factor has units of s m -~ and expresses the amount of pushdown in seconds (or, more conveniently, milliseconds), per net metre thickness of CO2. To calculate the Pushdown Factor, seismic velocities in rock filled with CO2 at various saturations can be estimated using the Gassmann fluid substitution equations (Gassmann 1951). Velocities derived in this way show a decrease from the observed value of about 2050 m s- 1 in water-saturated sand, to about 1420 m s- 1 in wholly CO2 saturated sand (Fig. 13). Errors are related mostly to uncertainties in elastic parameters, principally the bulk moduli of the rock framework and of supercritical CO2. In addition, the Gassmann equations assume a homogeneous mix of fluids; a more patchy distribution would give a more linear behaviour of the velocity-saturation relationship. Pressure effects on the seismic velocities are expected to be negligible. No significant increase in pressure has been observed during the injection process so far, the CO2 flowing easily into the very high permeability reservoir. The pressure-temperature conditions of the reservoir around the
25
i
50 c-
o~"
=COEv 500 rn
"1
c1.
CO2 plume are such that the CO2 is expected to have remained in a supercritical state. Direct observation of velocity pushdown within the plume lends support to the Gassmann analysis. Around and within the CO2 chimney in the upper part of the plume (Fig. 7), a total pushdown of 22 ms develops over an estimated 60 m section of reservoir sand. This requires a seismic velocity of about 1 4 5 0 m s -~ within the chimney, broadly consistent with Gassmann-derived values for CO2 saturations in the range 0.3 to 1.0 (Fig. 13). This is in accord with both the moderate saturations for vertical conduits proposed by Johnson et al. (2001) and the higher saturations indicated by Lindeberg et al. (2001). Overall, sensitivity analysis suggests that velocity error does not comprise the main source of uncertainty in calculating the Pushdown Factor (see below). Most of the velocity decrease induced by CO2 takes place at low saturations, the velocity curve levelling out at values of Sco~ greater than about 0.3 (Fig. 13). This ambiguity, together with the fact that Sco. is an explicit term in the Pushdown Factor,
318
R.A. CHADWICK E T A L .
30 2200 I ---
[ I
velocity -- Pushdown Factor
0 o
2000
ffl
e-
-20 ~
1800
I
v~
g
8
-10 n (a)
/
(b)
J
E O
1400
~'-~" "-<--'F
'-<
ffl
! ~--/ t ,,,t--.4. / AT
1
\\ x
E
1200 !
Iz..~___ J 4,
""
i 0.5 Sc02
0.0
"0 1.0
Fig. 12. Schematic views of a vertical column of rock corresponding to a single cmp bin, underlain by a notional fiat reflector (dashed, shown in two-way time). (a) Rock column saturated with water. (b) Partial replacement of water by CO2 produces velocity pushdown AT.
Fig. 13. Variation of velocity and Pushdown Factor with CO2 saturation, according to Gassman equations.
renders the value of Sco, the main source of uncertainty in the pushdown calculation. Thus the Pushdown Factor varies from over 25 milliseconds per net metre of CO2 at very low saturations of CO2, to only 1 - 2 milliseconds per net metre of CO2 at high saturations. CO2 at low saturations is therefore a much more efficient pushdown agent than higher saturation CO2. This leads to inherent uncertainty; to calculate the pushdown from a known injected volume, it is necessary also to know the effective saturation of CO2 throughout the plume. Forward modelling can be used to address the problem. TAITDs have been calculated for a series of assumed plume saturation scenarios based on the total volume of the plume envelope (Fig. 14). The two saturation 'end-members' will be considered first. The minimum saturation case is represented by CO2 distributed homogeneously throughout the entire volume of the plume envelope (Fig. 14a). The CO2 has a uniformly low saturation (Sco, = 0 . 0 7 5 ) and generates a TAITD of 30 802 m2s. This-represents the theoretical maximum possible pushdown for the injected volume of CO2 and the observed plume geometry. The opposite end-member is the maximum saturation case, where CO2 is present only in a state of full saturation (Sco, = 1.0), such as in discrete fully saturated layers (Fig. 14b). The TAITD in this case is only 3801 m2s, which represents the minimum theoretical pushdown for the known injected volume of CO2. Neither of the end-member scenarios matches the observed TAITD values. The low saturation endmember generates a pushdown that is far too high, and
moreover, is not realistic in terms of the observed plume reflectivity. The full saturation end-member produces a pushdown that is much too low. Because the observed TAITD does lie between the endmember saturation limits, it can, therefore, be modelled by some intermediate saturation distribution. Bearing in mind the observed reflectivity, a reasonable saturation scenario is one where CO2 in the plume is partitioned into two separate components: a 'reflective' component of CO2 trapped in thin layers, each obeying the thickness-saturation function (Fig. 10), and an 'unreflective' component of diffuse, low saturation CO,, which occupies all or part of the volume in between the layers. Models based on this scenario took the component of CO2 in layers as the volume calculated by thin layer summation (see above). From this, the time-lag was calculated (Equation 2), using the layer thickness-saturation function, at each CMP, for each horizon (e.g. Fig. 6d). The TAITD for the component of CO2 in layers was then obtained by summing over the 12 horizons, giving a value of 3784 mZs (Fig. 14c). This is much lower than the observed TAITDs, but the model is incomplete, in that it contains only about 80% of the total injected amount of CO2. Additional pushdown will result from the remaining 20% of CO,,, which is assumed to form a diffuse, low-saturation component, in between the layers. The simplest two-component model (Fig. 14d) assumes that the remaining diffuse CO2 is homogeneously distributed
end-members
thin-layer models
II 1t ,~ Lt3',ers
uniform diffuse minimLJm saturation S_ = 0 0 7 5
100~ saturated layers maxiFfqUrn saturation S -- 1 O0
I
lrom :LJrqlrqg d.r:~L:htJd,.) and $sturatlor': fHFCh,}P :80 total irqiect('J .'oiL,r"- ,
th r1 la,,or~, plL,s dr1 f:}r'Ti (JlfILISP ( ~ )
th r'i la'y'~}rs ~)[~J~2 ~'r'/tre ",e g t~,t tiff use (-}C
I
a
b m
rXerp"
. . . .
c
d
3784ms
13697m s
| I -
e
m
- ,~ccrr
3801 ms
12847rn s
I Observed Total Area Integrated Time Delay ~9200 to ~11000 ms 30802 m's
Fig. 14. Computed total area integrated time delays for 3.3 x 106m3 of injected CO, in different plume saturation models. (a) CO2 distributed homogeneously throughout the plume volume. (b) CO2 present only in fully saturated form, e.g. in layers. (c) CO2 as given by thin layer summation and saturation function (i.e. - 80% of the injected volume). (d) CO2 as in (c), but with remaining CO2 dispersed uniformly between layers. (e) CO2 as in (c) but with remaining CO2 dispersed between layers and concentrated preferentially in axial part of plume. Observed TAITD as blue dashed lines (interp from interpreted Base Utsira Sand, xcorr from cross-correlation).
4D SEISMIC IMAGING OF A CO2 PLUME
319
of the plume, particularly in the NE and SW, farthest from the injection point, are characterized by low pushdown-amplitude ratios. These are interpreted as areas where CO2 is present only at high saturations in thin, reflective layers, which produce relatively small amounts of pushdown (cf. Fig. 14). In contrast, the inner parts of the plume show much higher ratios. These are interpreted as signifying the presence of diffuse, low saturation CO, between the layers, which produces additional pushdown but no additional reflectivity. A further effect, which would tend to reinforce the observed pattern, is a possible reduction in layer reflectivity where diffuse CO~- decreases the acoustic contrast of the high saturation layers. This is exemplified by the main CO2 chimney, which with its high pushdown, but subdued reflectivity (Figs 7 & 15b), is marked by a prominent localized area of high pushdownamplitude ratio (Fig. 15c). The interplay of different seismic effects is quite complex, but the underlying pattern is clear; elevated pushdowns in the central part of the plume do not correspond to similarly enhanced reflectivity, and thereby indicate the presence of diffuse, unreflective CO> This very much supports the preferred saturation model of Figure 14e and suggests that analysis of the relationships between velocity pushdown and plume reflectivity is a potentially powerful tool for mapping saturation distributions within the plume.
throughout the intra-layer volume. The additional pushdown due to this diffuse CO2 is 9913 m2s, amply demonstrating the very high pushdown efficiency of low saturation CO,. The resultant TAITD of 13 697m2s is however considerably higher than the observed range of 9200-11 000 m2s. A refinement of the model can be effected by the intuitively reasonable step of preferentially concentrating the diffuse CO2 in the central, axial parts of the plume. A simple concentric saturation distribution, increasing linearly from Sco, = 0.0 at the plume edge, to Sco~ - 0.06 at the plume centre (Fig. 14e) has the same total injected volume but with a TAITD of 12 847 m2s, significantly closer to the observed range. Further increasing the heterogeneity of the diffuse CO2 component, by concentrating it into localized volumes of higher saturation, has the effect of further decreasing the overall pushdown. Thus, the likely presence of chimneys of CO2 would effect an additional reduction in the calculated pushdown, probably to within the observed range. Alternatively, an observed pushdown lower than the calculated value may simply signify that rather more CO2 is trapped in the thin layers, at high saturations, than is indicated by the simple amplitude-thickness transformation. The effects of dissolution should not be discounted either, because dissolved CO_~ would effectively become seismically invisible, rendering observed pushdowns smaller than predicted. Johnson et al. (2001) indicate however that, in the first three years of injection, even with lateral dispersal of CO-, by trapping beneath shales, amounts of CO2 dissolving in the formation waters are likely to be small (<5%).
Conclusions and discussion The time-lapse seismic data clearly image CO, within the reservoir, both as sub-horizontal high amplitude reflections and also as a pronounced velocity pushdown. It is likely that much of the CO2 is present as thin layers, trapped beneath thin beds of low permeability shale. Though the CO2 layers themselves are mostly beneath the limit of seismic resolution, amplitude changes appear able to resolve thickness changes down to one metre or less. The data also resolve a prominent vertical feature, interpreted as a chimney of CO2. Reflection amplitudes and the build-up of velocity pushdown around the chimney indicate that the higher CO2 layers in the plume tend to thicken towards it, consistent with it forming the primary feeder of CO2 in the upper part of the plume. A number of similar though much less prominent features seen elsewhere in the plume may correspond to smaller chimneys. This is supported by some reservoir flow simulations (e.g. Lindeberg et al. 2001 ), which require a number of chimneys to feed the observed CO2 layers. The verticality of the main chimney may also offer intriguing insights into the way that CO_, migrates through the reservoir. Given that the probability of fortuitous vertical alignment of stratigraphical holes in the shale layers is rather small, the chimney seems to require that the column of CO,, is able to find its way rather easily through the thin shale beds. How this occurs is unclear.
P u s h d o w n - a m p l i t u d e relationships In the above, reflection amplitudes and velocity pushdown give estimates of in situ CO2 volume that are essentially independent. It is also fruitful to examine these two seismic parameters together, as their inter-relationships provide useful additional insights. Velocity pushdown increases strongly towards the centre of the plume with a pronounced area of elevated values ( - 40 ms or greater) around and east of the injection point (Fig. 15a). Total plume amplitudes show a different pattern, however, (Fig. 15b), particularly across the central part of the plume where they are more evenly distributed, without notably increased values east of the injection point. This different behaviour can be quantified as variation in the pushdownamplitude ratio (Fig. 15c). The pushdown-amplitude ratio is analogous to the Pushdown Factor in that it measures pushdown per unit total reflection amplitude (the latter being related to total CO2 layer thickness). The observed variation of pushdown-amplitude ratio (Fig. 15c) can therefore be interpreted as providing qualitative insights into saturation distribution. The outer parts
20
t40
Fig. 15. (a) Two-way time pushdown beneath the plume (IP denotes injection point). Mapping incorporates both crosscorrelation and manual interpretation. (b) Total absolute reflection amplitude of the plume from seismic difference data (analysis window 850-1070 ms). (c) Pushdown-amplitude ratio i.e. [grid(a)/grid (b)]. Black outline denotes outer edge of the plume reflectivity envelope.
! 30 20 10
t10
ml-0
lm-0 ._o
vE O
-:_o o
e-,
320
R.A. CHADWICK ET AL.
Sample data are limited, but intrinsic permeability seems unlikely, the buoyancy pressure of the CO2 column being probably too small to overcome the capillary entrance pressure (Lindeberg 1996). The 1994 data show little clear evidence of pre-injection faulting within the upper part of the reservoir, but small faults, close to the limit of seismic resolution, with displacements sufficient to provide pathways through the thin shale beds may be present, perhaps as a consequence of differential compaction. Another possibility is that the CO-, dehydrates the shales and thereby induces shrinkage cracks. Alternatively, in these weak, unconsolidated sediments, it may be that the buoyancy force of the CO2 column in the chimney is able to displace the thin shale layers by purely mechanical means. More circumstantial evidence for minor faulting in the reservoir is the likelihood of diffuse CO2 in between the main layers, which suggests that the thin shale horizons possess some degree of permeability. An important aspect of the time-lapse seismic imaging is its ability to quantify the amount of injected CO: and any changes that subsequently occur due to leakage or dissolution. The studies carried out so far suggest that the observed reflectivity and velocity pushdown are broadly consistent with the known injected volume of CO,. However considerable uncertainty remains in a n u m b e r of areas, in particular, the likely presence of low saturation 'diffuse' CO-, in between the more concentrated layers. Because this diffuse CO~ is both unreflective and makes a disproportionately large contribution to the total p u s h d o w n , it introduces a strong e l e m e n t of nonuniqueness to the saturation models. The presence of lowsaturation CO2 within the plume is consistent with reservoir flow models carried out in the SACS project using the SIMED simulator (van der Meer et al. 2000) and with the r e a c t i o n transport models of Johnson et al. (2001), which incorporate semi-permeable (microfractured) intra-reservoir shales. Work is continuing on the time-lapse datasets and, with acquisition of the second repeat survey in September 2001, it is anticipated that understanding of the CO2 migration and dispersal will improve still further. We thank the SACS consortium for permission to publish this work. Permission to publish is also given by the Executive Director, British Geological Survey (NERC). SACS is funded by the EU Thermie Programme, by industry partners Statoil. BP, Exxon, Norsk Hydro, TotalFinaElf and Vattenfall, and by national governments. R&D partners are BGS (British Geological Survey). BRGM (Bureau de Recherches Geologiques et Minieres), GEUS (Geological Survey of Denmark), IFP (Institute Francais du Petrole), TNO-NITG (Netherlands Institute of Applied Geoscience--National Geological Survey) and SINTEF Petroleum Research.
References
CHADWICK, R. A., HOLLOWAY, S., KIRBY, G. A., GREGERSEN, U. &; JOHANNESSEN, P. N. 2001. The Utsira Sand, Central North Sea--an assessment of its potential for regional CO, disposal. In: WILLIAMS, D. J., DURIE, R. A., MCMULLAN, P., PAULSON, C. A, J. & SMITH, A. Y. (eds) Greenhouse Gas Control Technologies. CSIRO Publishing, Collingwood, Australia, 349-354. GASSMANN, F. 1951. Ober die Elastizit~it por/Sser Medien. Vierteljahresschr, der Naturf. Gesellschafl in Ztirich, 96, 1-23. GREGERSEN, U., MtCHELSEN, O. & SORENSEN,J. C. 1997. Stratigraphy and facies distribution of the Utsira Formation and the Pliocene sequences in the northern North Sea. Marine and Petroleum Geology, 14, 893-914. ISAKSEN, D. & TONSTAD, K. 1989. A revised Cretaceous and Tertiary lithostratigraphic nomenclature for the Norwegian North Sea. Norwegian Petroleum Directoo', Bulletin. 5. JOHNSON, J. W., NITAO, J. J., STEEFEL, C. I. & KNAUSS, K. G. 2001. Reactive transport modeling of geologic CO2 sequestration in saline aquifers: the influence of intra-aquifer shales and the relative effectiveness of structural, solubility, and mineral trapping during prograde and retrograde sequestration, Proceedings of the First National Conference on Carbon Sequestration, Washington, DC, May 14-17, 2001, UCRL-JC-146932. LINDEBERG, E. G. B. 1996. Escape of CO2 from Aquifers. Energy Conversion Management, 38(Supplement), s229-s234. LINDEBERG, E., ZWEIGEL, P., BERGMO, P., GHADERI, A. & LOTHE, A. 2001. Prediction of CO, distribution pattern by geology and reservoir simulation and verified by time lapse seismic. In: WILLIAMS, D. J., DURIE, R. A., MCMULLAN,P., PAULSON,C. A. J. & SMITH, A. Y. (eds) Greenhouse Gas Control Technologies. CSIRO Publishing, Collingwood, Australia, 372-377. SPAN. R. & WAGNER, W. 1996. A new equation of state for carbon dioxide covering the fluid region from the triple-point to 1100 K at pressures up to 800MPa. Journal of Physical and Chemical Reference Data, 25/6. VAN DER MEER, L. G. H.. ARTS, R. J. & PETERSON,L. 2000. Prediction of migration of CO,_ injected into a saline aquifer: Reservoir history matching to a 4D seismic image with a compositional Gas/Water model. In: WILLIAMS, D. J., DVRIE, R. A., MCMULLAN, P., PAULSON, C. A. J. & SMITH, A. Y. (eds) Greenhouse Gas Control Technologies. CSIRO Publishing, Collingwood, Australia, 378-384. ZWEIGEL, P., ARTS, R., BIDSTRUP, T., CHADWICK, A., EIKEN, O., GREGERSEN, U., HAMBORG, M.. JOHANESSEN, P., KIRBY, G., KRISTENSEN, L. & LINDEBERG, E. 2001. Results and experiences from the first Industrial-scale underground CO: sequestration case (Sleipner Field, North Sea). American Association of Petroleum Geologists, Annual Meeting, June 2001, Denver, abstract volume (CD).
Towards an automated strategy for modelling extensional basins and margins in four dimensions NICKY
WHITE,
JOHN
HAINES,
STEPHEN
JONES
& DETLEF
HANNE
Bullard Laboratories, Madinglev Rise, Madingley Road, Cambridge CB3 0EZ, U K
Abstract: There is a need for quantitative models which predict the structural and thermal evolution of sedimentary basins and margins in three dimensions. Although many different, two-dimensional algorithms exist, most of them are forward models which assume that rifting is instantaneous. We outline a three-dimensional optimization strategy which calculates spatial and temporal variations in strain rate. This approach is a generalization of an existing two-dimensional inversion algorithm which already tackles three issues of interest to the hydrocarbon industry. First, the residual misfit between observed and predicted basin geometries allows competing structural and stratigraphic interpretations to be objectively tested. Secondly, the animated evolution of basin and margins can be produced using the strain rate tensor. Thirdly, spatial and temporal variations of strain rate control basal heatflow, which in turn constrains the temperature and maturation histories of the sedimentary pile. Here, we present a small selection of two-dimensional results and show how our three-dimensional formulation is a logical extension of earlier work. A three-dimensional algorithm is under development.
We outline a strategy for analysing the three-dimensional structural and thermal evolution of extensional sedimentary basins and continental margins. The cornerstone of our approach will be the use of inverse theory to extract strain rate histories from normal faulting, subsidence and crustal thinning observations, although for now we do not model the normal faulting process. These three sets of observations are subject to differing degrees of uncertainty and it is important that such uncertainties are incorporated into a modelling strategy. An important function of inverse modelling is to investigate how observational errors are mapped into the solution space. Our primary aim is to develop a set of 2D and 3D inverse algorithms which determine the strain rate histories of continental margins as a function of time and space. These histories will define both structural and thermal evolution and thus help to refine maturation and fluid flow models of margins. Inverse algorithms can be used to carry out a systematic analysis of conjugate margins systems located worldwide. We anticipate that our algorithms will be useful de-risking tools when hydrocarbon exploration is directed toward deeper, more highly extended, margins. Our strategy has developed out of two independent approaches pioneered by two of us over the last five years. The first approach is concerned with a general quantitative description of the way in which continents actively deform. Haines & Holt (1993) showed that strain rate data based upon earthquake focal mechanisms and GPS measurements can be inverted to obtain the complete horizontal motions within zones of active distributed deformation. This approach is also used to analyse Quaternary fault slip rate data (Holt et al. 2000). Application of these inverse algorithms is generating important insights into active shortening (e.g. India, Tibet and Southeast Asia: Holt et al. 2000), active oblique deformation (e.g. New Zealand, California: Beavan & Haines 2001; Flesch et al. 2000) and active extension (e.g. Aegean Region, Basin & Range Province: Holt et al. 2000). A growing body of accurate GPS measurements will make the determination of a global strain rate model tractable (Kreemer et al. 2000; http://archive.unavco. ucar.edu). The second approach has grown out of a long-standing interest at Cambridge University in the development of sedimentary basins (McKenzie 1978). White (1994) has shown how one-dimensional subsidence data derived from well-log information can be successfully inverted by permitting strain rate to vary as a function of time. This algorithm has been
applied to - 2 0 0 0 stratigraphic sections derived from field logging, industry wells and to seismic reflection data by Newman & White (1999). The resultant strain rate distributions have been tested using independent information about the duration and magnitude of stretching periods. Strain rate distributions can then be used to constrain the rheological properties of stretching lithosphere. Later, Bellingham (1999) and Bellingham & White (2000) developed a two-dimensional inverse model which extracts the spatial and temporal variation of strain rate from individual basin transects. These 1D and 2D algorithms make no assumptions about the number, duration, intensity and distribution of rifting episodes within a basin. Instead, strain rate is allowed to vary smoothly through time and space until the summed misfit between observed and predicted stratigraphy is minimised. Potentially important twodimensional effects such as lateral heat flow and varying elastic thickness are incorporated. Bellingham & White (2002) have applied their 2D inverse model to a set of well-known extensional sedimentary basins. We now wish to exploit these recent and significant advances to tackle the challenging problem of 3D finite deformation as a function of geological time. In essence, we will combine the planform analysis of Haines & Holt (1993) with the cross-sectional analysis of Bellingham & White (2000). In its general form, 3D inverse modelling will be a significant advance which will extract considerable added value from 3D seismic reflection volumes. We note that Gemmer & Nielsen (2000, 2001) have developed a 3D inverse model for the thermal evolution of extensional sedimentary basins but this model does not include finite deformation.
Nature of the problem In the physical and medical sciences, inverse modelling plays a crucial role in extracting information from data. A familiar example is medical tomography whereby detailed images of a human body are generated by inversion of scanning information. None of us would be happy if the medical profession used forward modelling to generate images of body structure since this approach is both slow and unreliable. Forward-modelled calculations of the thermal and kinematic evolution of passive margins should be treated with similar scepticism unless the model space is systematically explored. Despite the explosive growth of three-dimensional seismic reflection coverage, no serious attempt has been made to pose and solve the inverse
DAVIES,R. J., CARTWRIGHT,J. A., STEWART,S. A.. LAPPIN.M. & UNDERH[LL,J. R. (eds) 2004.3D Seismic Technology: Application to the Erploration of Sedimentary Basins. Geological Society, London, Memoirs. 29. 321-331. 0435-4052/04/$15 9 The Geological Society of London 2004.
N. WHITE ET AL.
322
problem of 3D basin/margin evolution. Instead, a variety of different 2D forward models have been built, many of which assume instantaneous rifting. This restrictive assumption requires strain rate to be infinite which is physical unrealistic. Furthermore, it is difficult to assess the quality of solutions and to investigate how observational error affects the results obtained. Here, we are interested in solving the inverse problem so that we can answer the following questions: does a solution exist? Is there a unique solution or are there many possible solutions? If a solution is obtained, how well resolved is it? Are solutions affected by trade-off between different parameters? How do stratigraphic and crustal thickness uncertainties map formally into solution space? To solve for 3D finite deformation, we must be able to calculate the detailed geometry of a sedimentary basin from any given spatial and temporal distribution of strain rate. This calculation is defined as the forward problem. Strain rate patterns determine the horizontal and vertical components of the deformation velocity field which in turn determine the evolution of crustal and lithospheric stretching, ~(x,y, z, t). The inverse problem is solved by determining the spatial and temporal variation of strain rate required to fit 3D subsidence and crustal thinning observations according to appropriate misfit and smoothing criteria (Bellingham & White 2000). Inversion is only computationally tractable if the forward problem can be calculated rapidly since the 3D inverse problem typically has - 1 0 6 dimensions. Why is this problem important to industry? Successful exploration for hydrocarbons relies upon accurate risk analysis. In an extensional system, stratigraphic interpretation, structural evolution and thermal maturation are important sources of risk. Our inverse strategy seeks to quantify these three sources of risk which at deep, highly extended, passive margins are especially significant.
strain rate tensor to vary smoothly through time and three spatial dimensions. The importance of strain rate distributions cannot be underestimated: these patterns directly determine the manner by which basins and margins grow. Strain rate also controls the variation of basal heat flux across the margins and decompression melting of the asthenosphere. In three dimensions, we shall be able to analyse margin development without a priori knowledge of extension direction. Our results will enable us to make animations of the combined thermal, structural and stratigraphic evolution.
Continuum deformation Finite deformation, velocity fields and strain rates are a central concern for 3D basin modelling. A continuum approach will be used to model the average or smooth deformation of the lithosphere as a function of time and space. For now, the normal faulting process shall not be included. Accordingly, we must solve DF Dt
-- L F
where F is the deformation gradient tensor and L is the velocity gradient tensor whose elements are L,j --
Vi
axj
The fundamental quantity which governs the thermal and structural evolution of these margins is the variation of strain rate through time and space (Fig. 1). We wish to invert threedimensional subsidence, well-log information, normal faulting, crustal thinning and free-air gravity datasets by allowing the
z
t
Fig. 1. Cartoon illustrating principles of 3D basin modelling. Thick black lines, deforming lithosphere; thin black lines, xyz axes; u. v and w, particle velocities; yellow panel, xz, cross-section of lithosphere where solid dots illustrate particles which have moved according to velocity field defined by u and w; pink panel, yz cross-section of lithosphere where stationary solid dots indicate that w = v = 0 in this plane: white horizontal panel, surface of lithosphere where solid dots indicate that displacement is governed by u alone.
(2)
where vi are the velocities in the xj directions (Malvern 1969). The expression D / D t in Equation (1) is the substantive or Lagrangian derivative which means that the time derivative is applied to a vector joining one pair of particles. At any time t, the deformation of a short line, p(t), within the continuum is given by
p(t) = F(t)p(0) A 3D formulation
(1)
(3)
where p(0) is the vector joining the two particles at t = 0. Equation (1) is valid for any temporally and spatially varying velocity field. In Eulerian form, this equation is written as Dt
+ v.V F = L F
(4)
where v.V are advective terms which take account of the transport of elements within the continuum. Simple analytical solutions to these equations exist if these advective terms are zero. Since v cannot be zero, this condition means that the spatial gradients of F would have to be zero everywhere. Thus L is constant and it follows that all velocity gradients must be constant. Of course, this simplifying assumption cannot apply to extensional sedimentary basins and margins where velocity gradients necessarily vary as a function of time and space. Equation (4) describes the three-dimensional finite deformation of the lithosphere. For any velocity field, we must also solve the evolving thermal structure of the lithosphere, T(x,y.z.t). T is calculated by solving the three-dimensional heatflow equation with appropriate advective terms,
~5+v.V r = ~ K0~.
(5~
We ignore crustal heat production and sediment blanketing which have a secondary effect on the thermal evolution of slowly extending basins (McKenzie 1981). A future implementation could easily include these terms. Equations (4) and (5) are cornerstones of 2D and 3D inversion algorithms but neither can be solved analytically for arbitrary velocity fields. Standard finite-element and/or finite-difference models are used (see e.g. Beavan & Haines 2001; Bellingham & White 2000).
MODELLING BASINS IN FOUR DIMENSIONS
Tackling brittle deformation The second stream of modelling will be concerned with deformation of the brittle lid (i.e. the upper crust, which deforms by faulting). Our approach will differ from that of Bellingham & White (2000) because faulting of the brittle layer at the top of the lithosphere will be explicitly taken into account. The Haines & Holt (1993) approach was based upon thin sheet dynamics which can be justified at very large spatial scales. The depth dimension will be added for the finer spatial scales involved in margin formation and evolution. The key factor is the existence of high-resolution data for basins that make the move to a full three-dimensional treatment justified. Apart from the direct inclusion of faulting, the physical model is similar to that of Bellingham & White (2000). Some differences in implementation will be necessary because of the greater generality of the overall description. Faulting is undoubtedly a key factor in the evolution of extensional sedimentary basins. In itself, faulting is a hugely complex process, and a complete treatment is beyond the scope of our formulation. Fortunately, however, almost all of the complexities such as the earthquake cycle and visco-elastic coupling between the brittle-elastic layer and the region below, occur on timescales (102-105 years) which are too short to have a major, long-term, bearing upon basin evolution over millions of years. Instead, on the time scale of basin evolution, the brittleelastic layer can be considered to have the following ingredients: (i) the faults where concentrated deformation occurs; (ii) essentially rigid blocks, which suffer from no significant long-term strain (this assertion largely holds for active fault populations but breaks down to some extent for ancient fault populations); and (iii) long-term elastic strength sufficient to support the topography of the blocks between the faults. The concentration of deformation on faults and the asymmetric geometry of extensional faulting are the major differences from the existing 2D inverse model, which assumes vertically uniform and horizontally continuous strain at all depths in the lithosphere. An important unknown is whether faulting, rather than the elastic strength of the brittle-elastic layer, accommodates the bending required to achieve gravitational stability when the lithosphere is non-uniformly stretched. Both possibilities will be built into the modelling, and the implications for the thickness of the elastic layer will be evaluated. Other differences result from the asymmetric orientation of faulting. The underlying flow of lower crustal and lithospheric mantle material together with advection of hot asthenospheric material must also be asymmetric with respect to the vertical, though the extent of the asymmetry may decrease with depth. To help in evaluating the pattern of fluid flow, we have the constraint imposed by the force balance equations. These equations require that depth-dependent horizontal gradients in lithostatic pressure are balanced by viscous stresses associated with the flow, particularly during the extensional phase. (To use these constraints in the inversions, estimates are needed only for relative values of apparent viscosity and not for full rheologies). Quantitative comparisons will be made between all aspects of these complicated phenomenologies and the simpler assumptions of the Bellingham & White (2000) model leading to a final inversion methodology that is a practical compromise between analytical rigour and computational efficiency.
2D approximations A generalized 3D inversion algorithm does not yet exist because of computational limitations. For a low-density grid with
323
20 points in the x, y, = and t directions, the inverse problem has 160 000 dimensions. Since the forward problem can take several seconds of CPU time to solve, a generalized 3D problem is substantial but tractable (cf. the tomographic problem of global seismology). Here, we briefly describe two simplified and complementary implementations which tackle different aspects of the simpler and faster 2D problem. The first was developed by Haines & Holt (1993) and is concerned with retrieving velocity fields from earthquake and GPS data. The second was developed by Bellingham & White (2000) and obtains the spatial and temporal distribution of strain rate from subsidence and crustal thinning observations. Elements of both approaches will form the basis of a generalized 3D approach.
Active deformation in plan form Haines & Holt (1993) described a general method for obtaining relative horizontal motions on the surface of a sphere from strain rate data. Strain rate data are extracted from earthquake moment tensor analysis, from Global Positioning System (GPS) and Very Long Baseline Interferometry (VLBI) data, and from Quaternary rates of deformation on faults. The approach stems from earlier work by Haines (1982) who showed that if the spatial distribution of strain rates is everywhere defined, then the full velocity gradient tensor is uniquely defined. In other words, if a generally variable strain field in which the rates of horizontal shear strain are everywhere known then the complete horizontal velocity field and inferred rotation can be recovered if one line is known to remain unstrained and unrotated. By using a leastsquares inversion, measured strain rates are used to recover the active deformation field. This approach been used to calculate velocity fields of actively deforming regions on Earth (see Holt et al. 2000 for review). One of the best-known examples of active extension is the Aegean Sea and surrounding areas. Jackson et al. (1992) used earthquakes with M~ > 6.0 within the period 1911-1992 to determine the spatial distribution of seismic strain rates throughout the Aegean. They showed that a smooth velocity field matches the distribution of seismic strain rates and explains previously inferred rotations and translations. Their results have been compared with Satellite Laser Ranging (SLR) velocity measurements which corroborate the style of deformation despite the fact that the earthquake strains only constitute - 5 0 % of the total strain inferred from SLR data. The smooth velocity field is shown in Figure 2. Average strain rates can be inferred using a seismogenic thickness of 10kin and are - 10- ~5s- ~ within the faster straining regions. The velocity field represents a snapshot of active deformation and clearly shows the two-dimensional nature of the planformal deformation. The most important feature is that approximately westdirected vector azimuths in northwestern Turkey are transferred into southwestern motions within the central Aegean. There is also a marked change in velocity from mainland Greece toward the Gulf of Corinth and the Peloponnese. This planform method is not concerned with the temporal variation of velocity fields. The obvious reason for this simplification is the nature of the observations. GPS and earthquake moment tensor data are only available tbr a limited time period of t0~ years. In the Aegean, independent geological constraints suggest that extension has lasted - 5 X 106 years. Although estimates of slip upon Quaternary faults can lengthen the sampling period, it is unlikely that strain rate data can be obtained over timescales which are representative of typical rift durations. In any case, most extensional systems are no longer active.
N. WHITE ET AL.
324 18" 44*
20 ~
22 ~
24"
~j
26 ~ 28 ~ ~ ~ ~',!~:: " : . . . . . .
30"
32" t44"
the two-dimensional heatflow equation with appropriate advective terms. Thirdly, loads are imposed on the lithosphere using regional as opposed to local isostasy. Our philosophy is to keep the two-dimensional model as simple as possible whilst incorporating the essential elements; complexity can be added if required by observation. The most important simplifications are that the horizontal velocity is constant as a function of depth and that short-wavelength normal faulting is not included. Both of these assumptions are controversial but we intend to relax them in a future implementation. We also do not include the effects of sediment blanketing. Our starting point is a lithospheric template whose properties are defined by the parameter values listed in Table 1. As in most models, we assume that density is a linear function of the thermal expansion coefficient and that temperature varies linearly with depth. The algorithm is divided into four parts. First, the variation of strain rate through space and time is defined and used to calculate the velocity field. Secondly, this velocity field determines the evolving thermal structure. Thirdly, the temperature structure constrains the changing density structure which d e f n e s the loading history. Finally, subsidence history is calculated by imposing loads through the flexural equation.
~rf~ 40"
,: f . , /
36 ~
~
/ 20mm/yr
:N~
~/
~
v.. .,
36"
'
/
:34"
km ' 32" ~ 18"
20"
,
0
~ 24"
22 ~
'
soo
"
26 ~
28 ~
~32" 32"
30"
Fig. 2. Velocity field of the Aegean region relative to Eurasia obtained from the distribution of seismic strain rates from earthquakes of M~ --> 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