sediments, morphology and sedimentary processes on ... · 42 carbonate systems during the...

SEDIMENTS, MORPHOLOGY AND SEDIMENTARY PROCESSESON CONTINENTAL SHELVES

Other publications of the International Association of Sedimentologists

SPECIAL PUBLICATIONS

43 Quaternary Carbonate and Evaporite SedimentaryFacies and Their Ancient AnaloguesA Tribute to Douglas James ShearmanEdited by Christopher G. St. C. Kendall andAbdulrahman S. Alsharhan2010, 494 pages, 263 illustrations

42 Carbonate Systems During the Olicocene-MioceneClimatic TransitionEdited by Maria Mutti, Werner E. Piller andChristian Betzler2010, 304 pages, 154 illustrations

41 Perspectives in Carbonate GeologyA Tribute to the Career of Robert Nathan GinsburgEdited by Peter K. Swart, Gregor P. Eberli andJudith A. McKenzie2009, 387 pages, 230 illustrations

40 Analogue and Numerical Modelling ofSedimentary SystemsFrom Understanding to PredictionEdited by P. de Boer, G. Postma, K. van der Zwan,P. Burgess and P. Kukla2008, 336 pages, 172 illustrations

39 Glacial Sedimentary Processes and ProductsEdited by M.J. Hambrey, P. Christoffersen,N.F. Glasser and B. Hubbard2007, 416 pages, 181 illustrations

38 Sedimentary Processes, Environments and BasinsA Tribute to Peter FriendEdited by G. Nichols, E. Williams and C. Paola2007, 648 pages, 329 illustrations

37 Continental Margin SedimentationFromSediment Transport to Sequence StratigraphyEdited by C.A. Nittrouer, J.A. Austin, M.E. Field,J.H. Kravitz, J.P.M. Syvitski and P.L. Wiberg2007, 549 pages, 178 illustrations

36 Braided RiversProcess, Deposits, Ecology and ManagementEdited by G.H. Sambrook Smith, J.L. Best, C.S.Bristow and G.E. Petts2006, 390 pages, 197 illustrations

35 Fluvial Sedimentology VIIEdited byM.D. Blum, S.B. Marriott and S.F. Leclair2005, 589 pages, 319 illustrations

34 Clay Mineral Cements in SandstonesEdited by R.H. Worden and S. Morad2003, 512 pages, 246 illustrations

33 Precambrian Sedimentary EnvironmentsA Modern Approach to Ancient DepositionalSystemsEdited by W. Altermann and P.L. Corcoran2002, 464 pages, 194 illustrations

32 Flood and Megaflood Processes and DepositsRecent and Ancient ExamplesEdited by I.P. Martini, V.R. Baker and G. Garzon2002, 320 pages, 281 illustrations

31 Particulate Gravity CurrentsEdited by W.D. McCaffrey, B.C. Kneller and J.Peakall2001, 320 pages, 222 illustrations

30 Volcaniclastic Sedimentation in LacustrineSettingsEdited by J.D.L. White and N.R. Riggs2001, 312 pages, 155 illustrations

29 Quartz Cementation in SandstonesEdited by R.H. Worden and S. Morad2000, 352 pages, 231 illustrations

28 Fluvial Sedimentology VIEdited by N.D. Smith and J. Rogers1999, 328 pages, 280 illustrations

27 Palaeoweathering, Palaeosurfaces and RelatedContinental DepositsEdited by M. Thiry and R. Simon Coincon1999, 408 pages, 238 illustrations

26 Carbonate Cementation in SandstonesEdited by S. Morad1998, 576 pages, 297 illustrations

25 Reefs and Carbonate Platforms in the Pacific andIndian OceansEdited by G.F. Camoin and P.J. Davies1998, 336 pages, 170 illustrations

24 Tidal Signatures in Modern and AncientSedimentsEdited by B.W. Flemming and A. Bartholoma1995, 368 pages, 259 illustrations

23 Carbonate Mud-moundsTheir Origin and EvolutionEdited by C.L.V. Monty, D.W.J. Bosence, P.H.Bridges and B.R. Pratt1995, 543 pages, 330 illustrations

REPRINT SERIES

4 Sandstone Diagenesis: Recent and AncientEdited by S.D. Burley and R.H. Worden2003, 648 pages, 223 illustrations

3 Deep-water Turbidite SystemsEdited by D.A.V. Stow1992, 479 pages, 278 illustrations

2 CalcretesEdited by V.P. Wright and M.E. Tucker1991, 360 pages, 190 illustrations

Special Publication Number 44 of the International

Association of Sedimentologists

Sediments, Morphology and Sedimentary

Processes on Continental Shelves

Advances in Technologies, Research, and Applications

Edited by

Michael Z. LiGeological Survey of Canada (Atlantic), Natural Resources Canada,

Bedford Institute of Oceanography, Dartmouth, Nova Scotia, B2Y 4A2 Canada

Christopher R. SherwoodU. S. Geological Survey, Woods Hole, MA 02543-1598 USA

Philip R. HillNatural Resources Canada, Sidney, BC, Canada V8L 4B2

This edition first published 2012 � 2012 by International Association of Sedimentologists

Blackwell Publishing was acquired by John Wiley & Sons in February 2007. Blackwell’s publishing program has been

merged with Wiley’s global Scientific, Technical and Medical business to form Wiley-Blackwell.

Registered office: John Wiley & Sons, Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK

Editorial offices: 9600 Garsington Road, Oxford, OX4 2DQ, UK

The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK

111 River Street, Hoboken, NJ 07030-5774, USA

For details of our global editorial offices, for customer services and for information about how to apply for permission to reuse

the copyright material in this book please see our website at www.wiley.com/wiley-blackwell.

The right of the author to be identified as the author of this work has been asserted in accordance with the

UK Copyright, Designs and Patents Act 1988.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any

form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the

UK Copyright, Designs and Patents Act 1988, without the prior permission of the publisher.

Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and

product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective

owners. The publisher is not associated with any product or vendor mentioned in this book. This publication is designed

to provide accurate and authoritative information in regard to the subject matter covered. It is sold on the understanding that

the publisher is not engaged in rendering professional services. If professional advice or other expert assistance is required,

the services of a competent professional should be sought.

Library of Congress Cataloging-in-Publication Data

Sediments, morphology, and sedimentary processes on continental shelves : advances in technologies, research, and

applications / edited by Michael Li, Christopher R. Sherwood, Philip R. Hill.

p. cm. – (Special publication / International Association of Sedimentologists ; no. 44)

Includes bibliographical references and index.

ISBN 978-1-4443-5082-1 (hardcover : alk. paper) 1. Sedimentation and deposition. 2. Sedimentology. 3. Sediment

transport. 4. Continental margins. I. Li, Michael. II. Sherwood, Christopher R. III. Hill, Philip R. IV. International

Association of Sedimentologists.

QE571.S423 2012

551.3’03–dc23

2011040562

A catalogue record for this book is available from the British Library.

Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in

electronic books.

Set in 10/12pt Melior by Thomson Digital, Noida, India

1 2012

http://www.wiley.com/wiley-blackwell

Contents

Preface vii

Part 1 Sediments andmorphology in shelfand coastal systems

Optimal use of multibeam technology in thestudy of shelf morphodynamics 3

John E. Hughes Clarke

Palaeogeographic reconstruction of HecateStrait British Columbia: changing sea levelsand sedimentary processes reshape aglaciated shelf 29

J. Vaughn Barrie and Kim W. Conway

Changes in submarine channel morphologyand slope sedimentation patterns from repeatmultibeam surveys in the Fraser River delta,western Canada 47

Philip R. Hill

Recent sedimentary processes in the Cap deCreus canyon head and adjacent continentalshelf, NE Spain: evidence from multibeambathymetry, sub-bottom profiles andcoring 71

A. Garcıa-Garcıa, T. Schoolmeester,D. Orange, A. Calafat, J. Fabres,E. Grossman, M. Field, T. Lorenson,M. Levey and M. Sansoucy

Geology metrics for predicting shorelinechange using seabed and sub-bottomobservations from the surf zone andnearshore 99

Jesse E. McNinch and Jennifer L. Miselis

Re-examination of sand ridges on the middleand outer New Jersey shelf based on combinedanalysis of multibeam bathymetry andbackscatter, seafloor grab samples andchirp seismic data 121

John A. Goff and Catherine S. Duncan

Sedimentary facies of shoreface-connectedsand ridges off the East Frisian barrier-islandcoast, southern North Sea: climatic controlsand preservation potential 143

Chang Soo Son, Burghard W. Flemmingand Tae Soo Chang

Part 2 Sediment transport processes,sedimentation and modelling

Recent advances in understandingcontinental shelf sediment transport 161

L.D. Wright

Recent advances in instrumentationused to study sediment transport 173

Jon J. Williams

Seabed disturbance and bedform distributionand mobility on the storm-dominated SableIsland Bank, Scotian Shelf 197

Michael Z. Li, Edward L. King andRobert H. Prescott

Temporal variability, migration rates andpreservation potential of subaqueous dunefields generated in the Agulhas Current onthe southeast African continental shelf 229

Burghard W. Flemming andAlexander Bartholoma

Measurement of bedload transport in acoastal sea using repeat swath bathymetrysurveys: assessing bedload formulae usingsand dune migration 249

Garret P.Duffy and JohnE.Hughes-Clarke

Analyzing bedforms mapped using multibeamsonar to determine regional bedload sedimenttransport patterns in the San Francisco Baycoastal system 273

Patrick L. Barnard, Li Erikson, DavidM.Rubin, PeteDartnell andRikkG. Kvitek

v

Sediment transport on continental shelves:storm bed formation and preservation inheterogeneous sediments 295

T. R. Keen, R. L. Slingerland, S. J. Bentley,Y. Furukawa,W. J. Teague and J. D. Dykes

Tidal influence on the transport ofsuspended matter in the southwesternYellow Sea at 6 ka 311

Katsuto Uehara and Yoshiki Saito

Origin, transport processes and distributionpattern of modern sediments in theYellow Sea 321

Xuefa Shi, Yanguang Liu, Zhihua Chen,Jianwei Wei, Sulan Ge, Kunshan Wang,Guoqing Wang, Shouye Yang, ShuqingQiao, Deling Cai, Zhenbo Cheng,Wenrui Bu and Hi-Ii Yi

Seasonal and spatial variation in suspendedsediment characteristics off the Changjiangestuary 351

Guoqing Wang, Xuefa Shi, Yanguang Liu,Xisheng Fang and Gang Yang

Factors controlling downward fluxes ofparticulate matter in glacier-contact andnon-glacier contact settings in a subpolarfjord (Billefjorden, Svalbard) 369

Witold Szczucinski and MarekZajaczkowski

Part 3 Application and management

On seabed disturbance, marine ecologicalsuccession and applications for environmentalmanagement: a physical sedimentologicalperspective 389

Peter T. Harris

Benthic habitat mapping from seabed acousticsurveys: do implicit assumptions hold? 405

Vladimir E. Kostylev

List of Reviewers 417Index 419

vi Contents

Preface

Sediments, morphology and sedimentaryprocesses on continental shelves: advances intechnologies, research, and applications

The application of multibeam and sediment trans-port measurement technologies and the adoptionof integrated techniques in the approach toresearch have greatly advanced our understandingof sediments, morphology and sedimentary pro-cesses on continental shelves in the last decade.This book focuses on the applications of newmultibeam mapping and sediment transport mea-surement technologies, the integrationofmorphol-ogy and processes and the utilization of shelfseabed property and process knowledge in coastaland ocean management.

The volume grew out of a technical session“Sediments and Sedimentary Processes on Conti-nental Shelves” organizedby the editors at the 17thInternational Sedimentological Congress held inFukuoka, Japan, from 27 August to 1 September,2006. Thirteen of the twenty contributions wereoriginally presented in that technical session. Theother sevenpaperswere solicited to cover the latestadvances in multibeam mapping technology,advances and applications of sediment transportmeasurement techniques, and the application ofsediment property and process knowledge in habi-tat mapping and ocean management. The articlesin this bookwere contributed by authors fromeightcountries and cover a range of topics. The aim ofthe book is to take stock of the impact that newadvances in technology, spatial analysis and mod-eling have brought to the understanding of shelfsedimentology. With the mix of primary researchand review papers, the boo will serve as a mile-stone for the world’s shelf sedimentology andocean management communities.

Thebook isdivided into three sections.Section I,“Sediments and Morphology in Shelf and CoastalSystems” opens with an article by Hughes-Clarkethat reviews the application of state-of-the-artmul-tibeamtechnology toshelf sedimentologyresearch.Resolution and accuracy in both bathymetry andbackscatter data are explored and examples illus-trate the optimal use of multibeam technology.The next three papers demonstrate the integrationof multibeam bathymetry mapping with seismicsurveying and coring. Barrie & Conway use thesetechniques to understand how changing sea levels

and sedimentary processes reshaped the glaciatedshelf in Hecate Strait, British Columbia. Hilldemonstrates how repeated multibeam surveyscan be used to interpret significant changes insubmarine channel morphology and slope sedi-mentation patterns in the Fraser River delta, wes-ternCanada.Garcia-Garciaetal. interpret sedimen-taryprocesses in theCapdeCreus canyonheadandadjacent continental shelf, NE Spain.

Two papers apply these integrated techniques tothe study of sand ridges on continental shelves. Goffand Duncan integrate multibeam bathymetry andbackscatter data in an analysis of sand ridgeresponse to the present-day hydrologic regime ofthe middle and outer New Jersey shelf. Son et al.investigate the sedimentary facies and preservationpotential of shoreface-connected sand ridges in thesouthern North Sea through analyses of multibeambathymetry, seabed samples and internal sedimen-tary structures derived from box-cores. The lastpaper in this section, by McNinch and Miselis,explores the links between shoreline changes, mor-phology and sediment distribution in the surf zoneoff the coast of North Carolina based on almost adecade of repeated field observations that includedinterferometric swath bathymetry, chirp sub-bottomprofiles, sediment vibracores and radar data.

Eleven papers are included in Section II, “Sedi-ment transport processes, sedimentation andmod-elling”. The first two contributions reviewadvances in instrumentation and understandingin the research of sediment transport processes oncontinental shelves. Wright reviews highlights ofshelf sediment transport research and identifiesnew directions, with an emphasis on gravity-driven transport,within negatively-buoyant hyper-pycnal layers, as an important mechanism ofacross-shelf sediment transport. Williams focuseson recent technological innovations in sedimenttransport instrumentation and examines, by exam-ple, how these novel technologies contribute tosimultaneous measurements of fluid flow, bedmorphology and sediment transport at unprece-dented spatial and temporal resolution in thelaboratory and in the field. He also predicts howdevelopments in instrumentation over the comingdecade may enable more accurate forecasting ofsediment transport processes.

The following papers present recent findingson seabed forcing, bedform distribution and

vii

migration, the implications of these to bedloadtransport calculations and regional sediment trans-port patterns and sedimentary strata formation andpreservation under a range of current and stormconditions. Regional bathymetric maps, sidescansonar and multibeam bathymetric surveys, sedi-ment samples and model predictions of seabeddisturbance are integrated by Li et al. to character-ize the distribution, metrics and mobility of bed-forms on the storm-dominated Sable Island Bank,Scotian Shelf. Flemming and Bartholom€a examinea serial sidescan sonar data set spanning nearly16 years to shed light on the temporal variability,migration rates and preservation potential of sub-aqueous dune fields beneath the Agulhas Currenton the southeast African continental shelf. Duffy &Hughes-Clarke use repeat multibeam bathymetrysurveys and velocity profile measurements toquantify long-term sediment transport and eval-uate the performance of several formulations forpredicting bedload transport rate over sand dunesin the Bay of Fundy. Barnard et al. present acomprehensive analysis of more than 3000 bed-forms measured from high-resolution multibeamsurvey data and demonstrate how this highlydetailed, quantitative information can be usedto both determine regional sediment transportpatterns and assist management of the sedimentresource in the San Francisco Bay coastal system.Keen et al. use observations and numerical modelpredictions to examine the source, transport pro-cesses and deposition of modern storm beds. Bycomparison with historical data, they evaluate therecurrence frequency of modern storm beds andthe preservation potential of such beds within astorm-dominated shelf sequence.

The last four articles in Section II discuss theorigin, characteristics, seasonal and spatial varia-tions, transport and sedimentation of suspendedparticulate matter. The understanding of cohesivesediment dynamics and deposition presents itsown set of challenges that can be addressed bynew technologies, integration of different spatialdata sets and numerical modelling. Shi et al. usevarious sediment characteristics, suspended par-ticulate matter concentration and trend analysis toestablish a depositional process model for theYellow Sea. Uehara & Saito use numerical model-ing to examine the tidal influence on the transportof suspended matter in the southwestern YellowSea in the mid-Holocene. Wang et al. demonstratethe use of in-situ particle size and concentration

measurements to understand seasonal changes insuspended particle dynamics off the Changjiangestuary. Szczuci�nski & Zajaczkowski integratehydrology and sediment trap data to investigateparticulate matter fluxes and controlling factors ina sub-polar fjord in the European Arctic.

Section III reviews the application of shelfsedimentology research to habitat mapping andocean management. Harris explores the inter-relationships between seabed disturbance, mar-ine ecological succession and ecosystem-basedmanagement, demonstrating the need for anunderstanding of sedimentology in the designof marine protected areas (MPAs). The volumeconcludes with a conceptual article by Kostylevthat examines the validity of the basic assump-tions in using acoustic seabed surveys to producebenthic habitat maps. He calls for a better under-standing of the processes linking benthic com-munities to seabed geology so that remotely-sensed geological information from, for example,multibeam surveys can be used more confidentlyin benthic habitat mapping.

This Special Publication volumewouldnot havebeen possible without the contribution and sup-port of a large group of people. First and foremost,the guest editors would like to acknowledge thesupport and cooperation from all the authors ofthe papers included in this book. We are gratefulto the Special Publication editors Drs Ian Jarvis,Stella Bignold and Thomas Stevens for theirsupport and advice during the editing process ofthis book. We acknowledge our respective institu-tions for allowing us to take on this endeavour inaddition to our normal organizational responsibili-ties. Special thanks go to Nina Parry for her veryable co-ordination of all correspondence duringthemanuscript reviewprocess. Finally, eachpaperin this bookwas evaluatedby at least two reviewersand their comments and critiques have greatlyimproved all the papers in this volume.We deeplyappreciate the effort of these reviewers who arelisted at the end of this book.

The completion of this Special Publicationvolume is made possible by the support of ourfamilies, particularly our wives Ping, Patricia andJennifer. We are grateful for their understandingand encouragement.

Michael Z. LiChristopher R. Sherwood

Philip R. Hill

viii Preface

Part 1

Sediments and morphology in shelf andcoastal systems

Optimal use of multibeam technology in the study of shelfmorphodynamics

JOHN E. HUGHES CLARKE

Department of Geodesy and Geomatics Engineering, University of New Brunswick, Canada (E-mail: [email protected])

ABSTRACT

Many of the recent advances in our understanding of sedimentary processes on thecontinental shelf have come about as a result of the use of multibeam sonar systems.These systems provide wide area coverage of seafloor variations in bathymetry andbackscatter at typical horizontal resolutions as small as � 2% of the water depth. Thenarrowest beamsystemsnowprovide backscatter data at resolutions approaching towedsidescan sonar while simultaneously providing co-registered, equivalent-resolutiontopography.

Evenmore valuable than the static viewof the seabed is an ability, through resurvey, tomonitor temporal variations in the seabed. By adding the time dimension, insights canbe provided into the sedimentary processes rather than just the resulting sedimentdistribution. To achieve this, however, requires particular attention to be placed on thelimitations of these survey systems,which affect repeatable accuracy. To assess the totalachievable accuracyoneneeds to account for all the integrated components of the surveysystem.

In this paper, the contributions of the various sources of systematic bathymetric andbackscatter errorwithin a typical shelfmultibeam survey are described. To optimize thebathymetric data, strategies for dealing with imperfections in tidal models and knowl-edge of the sound speed structure are described. In order to improve the backscatter data,strategies for predicting the combined effect of beam pattern residuals and the seabedangular response are detailed.

To illustrate a typical result, a pair of overlapping surveys employingwidely differingsource sensor resolution and accuracy is combined to try to predict the relativeimportance of active and relict shelf morphodynamic processes.

Keywords: Multibeam, multi-sensor integration, calibration, backscatter reduction.

INTRODUCTION

Routine application of multibeam sonar bathyme-try and backscatter has revolutionized our under-standing of continental shelf morphodynamics.The ability to view a near-continuous topographicsurface together with variations in seabed back-scatter strength provides an overview analogous toaerial photography, resulting in a vastly improvedability to interpret the seafloor sedimentary pro-cesses (Hughes Clarke et al., 1996).

One of the most immediate results of this newtechnology has been the recognition, for the firsttime, of the continuity and juxtaposition of longwavelength features such as drowned beach ridgesand reefs (e.g. Gardner et al., 2005), or morainecomplexes (e.g. Todd et al., 1999). But the real

challenge tomaximizing the usefulness of this datawill lie in the finer details revealed. The detail is inthe shorter wavelength morphology that lies closeto the limits of resolution of these systems.

After the first pass interpretation of the currentstate of the shelves, future researchwill be increas-ingly focused on monitoring their temporal evolu-tion. The first view provides a snapshot. Thatsnapshot allows inferences to bemade about likelysedimentary processes. However, proof of theactivity of those processes awaits repetitive sur-veying. Proof that the seabed has changed requiresconfidence in the absolute accuracy of both thebathymetric and backscatter output of the inte-grated sonar system.

Obvious change, such as new slide scars(Brucker et al., 2007), overprinted iceberg scours

� 2012 International Association of Sedimentologists and published for them by Blackwell Publishing Ltd. 3

Int. Assoc. Sedimentol. Spec. Publ. (2012) 44, 1–28

(Sonnichsen et al., 2005), freshly emplaced debrisflows (Kammerer et al., 1998) or significantly-migrated bedform positions (Duffy and HughesClarke, 2005) canbediscerned from imperfect data.However,moresubtle transitions, suchasaccretionof thin sand sheets, deflation of near shore sandbodies, deepening of pockmarks or migration ofripples requires a level of absolute accuracy thatlie at the limit of many of the integrated systems.

This paper explores the resolution and accuracycapabilities in both bathymetry and backscatterthat is realistically available from currently state-of-the-art multibeam sonar systems. Practicalexamples are provided, illustrating the advantagesand limitations of this sort of data for shelf mor-phodynamic research.

BATHYMETRY

Resolution

Thepower of amultibeamsystem lies in its ability toresolve sedimentary structures atwavelengths smallenough to infer the processes active. Many of thesediment transport mechanisms can be inferredfrom the short wavelength relief. Most notably, bed-forms, such as transverse dunes or ripples and long-itudinal ribbons provide a clear indication of activesediment transport. Similarly, erosional scour andpockmarks are indicative of modern or relict sedi-mentary processes. However, such features, whichhave spatial scales of decimetres to a few tens ofmetres, often lie at the limit of the spatial resolutionof the system. In the case of surface hull-mountedsonars, the resolution decays roughly linearly withdepth. However, the question needs to be asked:does the disappearance of a specific short wave-length morphology with depth indicate a changein sedimentary environment, ormerely adefocusingof the instrument over increasing range?

Sedimentologistswishing to conductmultibeamsurveys may not have the luxury of choice ofsystem due to logistical or financial constraints.When interpreting the available data, however, it isimportant to establish the achievable resolution ofthe utilized specific sonar system. To this end,there are a number of components that need tobe considered, including:

Beam width, spacing and detection algorithm

Sonar systems are routinely quoted with beamwidth dimensions. Such dimensions need to be

specified in two directions (Fig. 1A), along track(controlled by the transmit beamwidth) and acrosstrack (controlled by the receive beam width) asthey may differ (Miller et al., 1997).

In order to appreciate the potential of the beamfootprint, its solidangleneeds tobeprojected to theseabed over the range of depths and angles used. Itis readily apparent that the minimum, resolvabledimension is strongly linked to the size of thisfootprint (Fig. 2). Resolution needs to be describedseparately for along and across track.

For an amplitude detection (deMoustier, 1993),the resolvable dimension cannot be smaller thanthis footprint as the echo is integrated over thatdimension. Few sonars today, however, still useamplitude detection outside the near nadir or nearspecular region. Phase detection using a split aper-ture (deMoustier, 1993), in which the elevationanglewithin the beam footprint is definedbyphaserather than peak intensity, is almost universallyused. In this manner, discrimination across trackcan be achieved based on phase (Fig. 1C). For thelong, lower grazing angle echoes, phase (and thusfeature definition) can be discerned at a scale sig-nificantly finer than the beam footprint dimension(Hughes Clarke et al., 1998). Formost sonars this isachieved byhaving beamspacings across track thatare tighter than the beam footprint dimension. Themost commonexample of this is the “Equi-Distant”beam spacing (EDBS) mode (Fig. 1B) increasinglyoffered. For conventional phase detection, eachbeam still has only one depth solution (what istermed the “zero phase crossing, solution 0 inFig. 1C), but it is based on just the phase slopein the central part of the beam.

Figure 2 (left; EM1000 images) illustrates theresolution achieved using equi-angular beamspa-cing when the EDBS philosophy is not employed.As can be seen, the definition of the bouldersdegrades notably as one moves to the outer partof the swath. The compromise in EDBS is that, for afinite number of beams, the beam spacing in thenear nadir region has to be compromised to accom-modate the extra solutions at lower grazing angles(see beam spacing in Fig. 1B). For example for theEM1002, which has 111 beams over a 150� sector,in equi-angularmode (EABS) the near nadir beamsare spaced at 1.35�, whereas in EDBS they arespaced at 3.84� (resulting in lower nadir resolutionand wider than the 2� beam width, resulting incorrupted backscatter data).

Most recently, the limitation of EDBS has beenremoved through the use of “high definition” beam

4 John E. Hughes Clarke

forming (Kongsberg, 2005) in which, for phasedetection, multiple points on the phase slopeare used within a single beam footprint (Fig. 1C,solutions �1, þ 1 and þ 2). The physical beamspacing is actually equi-angular, but more depthsolutions than beams are generated by subdividingthe lower grazing angle beams. This gets aroundthe compromise in conventional EDBS as optimalbeam spacing for amplitude detection is retained.

However tight the beam spacing in the acrosstrack dimension, in the along track direction, thebeam dimension and its spacing will still limitresolution. Thus narrower transmit beam widthsare to be favoured. For a given transmit beamwidth and depth, the fore-aft dimension of thefootprint grows with obliquity. Thus for geolo-gical purposes, resolution will generally decayaway from the nadir region. Again it is importantthat this limitation be notedwhen interpreting the

distribution of features close to the limit of reso-lution such as ripples or boulders.

There is a wide variety of multibeam sonarsavailable, but the ones most commonly used onthe continental shelf are those in the �100kHzrange. The EM1000, operating at 95kHz with abeam width of 2.4� � 3.3�, first appeared in 1992and has been used extensively in continental shelfsurveys worldwide. Large tracks of the US conter-minous continental shelf have been covered withthis sonar (Gardner et al., 2005, Valentine, 2005,Butman et al., 2006). The RESON 8111 (100kHz,1.5� � 1.5�) appeared in � 1996 and has been usedcommercially for similar scale continental shelfmapping (Wilson et al., 2005, Intelmann et al.,2006). The EM1000 was superseded by the verysimilar but higher resolution EM1002 (2.0� � 2.0�)in 1998, but manywere still used until� 2005. TheEM1002 has been employed on a regional scale for

The “footprint” depends on beam width and depth

EQUI-DISTANT

EQUI-ANGULAR

+ High Definition Beam Forming

π

− π

Pha

se

Time

-1

+1 +2

0

Just picking beam centre

A

B

C

Fig. 1. (A) Representation of an oblique narrow beam footprint on a typical seabed terrain. (B) Variation of size and spacingof a series of multibeam profiles, comparing and contrasting the bottom detection solution (represented by stars alternatingblack and white from beam to beam) spacing offered by equi-angle, equi-distant and high density beam forming (multiplesolutions per beam). (C) The method of describing the across-track relief within a single beam footprint by looking at theevolution of differential phase over the across-track beam dimension. For this method, the centre of the beam corresponds tothe point at which the differential phase is zero. For conventional detection, this point is the only one located (by regressionthrough the phase slope), whereas for high definition, multiple points on the phase curve (corresponding to multiple angleswith respect to the beam centre) are identified.

Optimal use of multibeam technology in the study of shelf morphodynamics 5

continental shelf geological mapping by Canadianagencies (Pickrill & Todd, 2002, Conwayet al., 2004). The EM710 (Fig. 2, right hand sideimages) represents one example of the next genera-tion of sonar systems that are replacing the 1000/1002 series with beamwidths now as narrow as0.5� � 1.0�, and for the first time include yaw sta-bilization. The practical examples here compareand contrast the EM1000 and EM710 sonars.

Roll, pitch and yaw stabilization

In order to optimize the resolution, the soundingdensity along track shouldbe ashigh andas evenaspossible. Ping rate for single ping systems is con-trolled by the two way travel time (TWTT) to theoutermost beams. The wider the angular sector,the lower the ping rate. Thus for a given speed,resolution will decay with sector width resultingin a competition between lateral coverage and

resolution. This is starting to be solved with therecent use of multiple swaths per ping cycle sys-tem. This is now offered (but only delivered in July2008) by a number of manufacturers and promisesto improve this limitation.

Irrespective of the along track vessel movementbetween pings, the outermost beams may be dis-placed more or less depending on the vessel rota-tions and the form of stabilization (Fig. 3). Rollstabilization is essential if the full swath is to beused, but does not affect the along track density.Pitch stabilization is more important in deeperwater. But the biggest issue in continental shelfdepths is yaw. In a cross-sea, vessel heading is hardto maintain, and as the water depth becomes shal-lower the helmsman is forced to take strongercorrective action to maintain minimal survey lineoffset. The requirement for yaw stabilizationdepends on the inter-ping yaw shift and the trans-mit beam width. As narrow transmit beams are

1992 EM1000 -2.4°x 3.3°

50m

2006 EM710 -0.5°x1.0°

Backscatter (echo-strength)

Depths (sun-illuminated)

Fig. 2. Comparison of the backscatter and bathymetric imaging resolution of two generations of common multibeam sonarsystems. Wrecks and boulder fields in 25–40m of water. Note the increased definition in both bathymetry and backscatterimaging of the boulder targets. Note also the pronounced drop in resolution for the outermost beams in the case of the equi-angle beam spacing utilized by the EM1000. Beam widths are given for both scanners.


being used to increase resolution, the requirementfor yaw stabilization is increasing.

To achieve yaw stabilization requires the use ofmultiple sectors (Fig. 3). For a single sector system,the full swath is illuminated using a single broadtransmit beam that canonlyutilize a single steeringangle, which must be chosen as a compromisewhereby both sides of the swath are aligned as bestas possible. For the case of multiple sectors, asuccession of individual transmissions is gener-ated, closely spaced in time (separated in timeonlyby the length of each pulse). Each sector/transmis-sion addresses only a specific subset of the totalswath and can thus have a unique steering anglethat best aligns that subset of the swath. In thismanner the compromise inherent in single sectorsystems canbe avoided, allowing yawstabilizationthat requires, as a minimum, opposite-sense steer-ing angles for each side.Without yaw stabilization,

there will be zones of lower sounding density (onthe outside of shallow corners; Fig. 3) where thetarget resolution, and thus geological interpreta-tion, is compromised.

Accuracy

Achievable resolution is no guarantee of absolutesurvey accuracy at that level. Any survey consistsof a series of systematically offset corridors of data,normally called swaths. The combination ofmulti-ple swaths requires a common reference datum.Absolute accuracy limits will corrupt the data intwo ways: (1) when blending the overlap, the viewof the seabed in the region of overlap will bedefocused; and (2)when comparing the swathwithdata collected at other times, only scales of seabedchange larger than the combination of the achiev-able accuracies of both surveyswill be discernable.

Roll, Pitch and Yaw Compensation

Single Sector

Roll and Pitch Compensation

Multi (3) Sector

Pitched Only Yawed Only Resultant SoundingDensity

Spread out

Bunched up

Fig. 3. The strategy and result of active roll, pitch, and yaw stabilization. Note particularly the improvement in evensounding density achieved by using the multi-sector strategy. This strongly impacts on the ability to maintain resolution forall regions ensonified.


While manufacturers’ brochures tend to empha-size the sonar-relative range and angle accuracy,these usually input an uncorrelated random noisein the sounding data rather than a systematic bias.For operations where repeat surveys are requiredfor sedimentary change assessment, it will be thesystematic errors that are more important as theywill generate biases that can be confused with truesediment accretion or deflation. There are a num-ber of components in addition to the sonar rangeand angle measurement that contribute to theachievable degree of accuracy.

Positioning systems – horizontal

Positioning requirements are normally quite dif-ferent for horizontal and vertical. The Global Posi-tioning System is now used universally for thehorizontal component of marine surface surveys.The achievable accuracy depends on the type ofGPS chosen. Stand alone versions (non-differen-tial) will allow 10–15m accuracy, sufficient fordeep-sea operations (where the resolution is belowthis), but not for shelf investigations where someform of differential GPS will be required.

Differential corrections from a coastal (usuallyCoastguard) service will provide sub-2m horizon-tal accuracy, adequate for outer continental shelfsurveys. To obtain better accuracy than this wouldrequire an interpolated correction service, such asFugro OmniSTAR (Visser, 2007) or C&C CNav(Chance et al., 2003), often referred to as Glob-ally-correctedGPS (GcGPS). Such services providedecimetre level horizontally, meeting practicallyall the needs of shelf and inshore surveys.

To get to a centimetric level positioning requiresa local base station and “Kinematic GPS”(USACE, 2002). This is not practically needed forhorizontal positioning but, as outlined below doesprovide the necessary level of vertical positioningto account fully for tides and squat.

Assuming that the horizontal accuracy of thepositioning system meets the needs of the seabedchange detection requirement, one still has toensure proper integration of that position. Themost common issue is one of time delays betweensonar and positioning sensor clocks. Delays willresult in systematic, along survey-line displace-ment of the swaths of data. This will generateapparent migration of seabed features that couldbe confused with real change. Detection of suchoffsets is normally quite easy by comparing thedisplacement of linear targets such as bedrock

outcrop ridges or sand wave crests within a singlesurvey.

Angular measurements – accuracy and alignment

All sonar relative ranges and bearing need to beadjusted for array orientation at transmit andreceive operations. Generally the stated angularaccuracies (<0.05�) of the high-endGPS-integratedinertial motion sensors are more than adequate forthe accuracy levels needed for operations. How-ever, it is not the instrument accuracy that mostconcern us, but rather the integration of sensordata. Proper integration requires knowledge ofsonar to motion sensor alignment and timingcalibration.

Misalignment or mistiming of sensors relative toeach other can create both static biases (for exam-ple a roll bias) and dynamic residuals (so calledwobbles). For a full review of the sources ofdynamic motion residuals, the reader is referredto Hughes Clarke (2003). From the point of view ofthe sedimentologist, the effects have two end-member results. Firstly the static biases impedethe ability to measure change, and secondly thedynamicmotion residuals can be confusedwith, orobscure, real seabed terrain.

Water column sound speed structure

An integral component of an accurate depth mea-surement is the proper accounting of sound wavepropagation and refraction in the water column(Beaudoin et al., 2004). This depends on an ade-quate knowledge of the sound speed structure inthe ocean. Failure to account for this properly willresult in either a dynamic residual (HughesClarke, 2003) or a systematic, convex or concaveacross track bias (Fig. 4), the magnitude of whichdepends on the unmonitored changes in the watercolumn.Thewater column is changing continuallyin time and space and thus themagnitude and signof the errorwill reflect the timeand/ordistance thathas passed since the last sound speed measure-ment (Hughes Clarke et al., 2000). Figure 4 illus-trates a typical summer continental shelf oceano-graphic section, illustrating the rapid change inrefraction conditions from freshwater stratified, toa tidal mixed area to a thermally stratified oceanwithin distances of tens of kilometres.

In order to minimize the significance of soundspeed errors, a variety of strategies may be dev-eloped including continuously monitoring the


sound speed (Cartwright & Hughes Clarke, 2002),reducing the angular sector of the swath, andreviewing archived information about likely watermass variability and then designing the survey totake that into account.

The illustrated profile (Fig. 4) required updatedsound speed structure information at approxi-mately half hour intervals to maintain the full �65� swath within IHO order 1 specifications (Inter-national Hydrographic Organization, 1998). Sucha profiling frequency is not practical unless under-way profiling strategies are available. Prior knowl-edge of this oceanographic variabilitywould allowtheprudent user to breakup the survey into regionsof common watermass type.

In all cases, it should be appreciated that redu-cing the angular sector is the most reliable way of

minimizing these errors. This results in a low rateof coverage, but will improve the data density (asthe required maximum two-way travel time isreduced) and thus increase the resolution. A prac-tical example is presented (Fig. 5) showing inter-survey bathymetric surface differences for Squam-ish Delta in Howe Sound, British Columbia. Theupper delta is accreting �1myr�1 on average. Thedelta has been the subject of investigation withmultibeam since 2004 (Brucker et al., 2007). Main-taining sufficient sound speeddata to survey repre-sents a challenge due to the presence and varia-bility of a freshwater plume emanating from themouth of the Squamish River that is modulatedover a tidal cycle.

The first difference map (Fig. 5A) shows theapparent changes based on two regional surveys

44°N

66°W

Salinity

A

B

C

A B C

100m

200m

100m

200m

Temperature

A B C

30 33.5

10 16

Salinity (PSU)

Temperature °C

Saint John River Plume

Thermal Stratification Tidal Mixing

10km

Sound speed m/s

depth

m

AB

C

True seafloor

positive refraction bias

negative refraction bias

A

B

C

Tidal Fronts

New Brunswick

Nova

Scotia

Maine

Fig. 4. Variability in the sound speed field for a typical summer-time continental shelf. Example dogleg transect illustratedfrom A to B to C across the Bay of Fundy. Note transition from a fresh-water stratified environment, to a thermally stratifiedenvironment, punctuated by zones of increased mixing, separated by abrupt tidal fronts, spaced at times by only a fewkilometers along track. Such changes, which alter the location of the velocline, lower left, result in systematic across trackbiases in the resulting bathymetric data, lower right. Datawerederived fromacontinuously operating underwayprofiler at 12knots, with sample spacing of about 1 km.


that did not undertake extra sound speedmeasure-ments close to the river mouth. While it is imme-diately apparent that gross change has occurred inthe delta foreset channel and on the proximal lobeto theSW, there is a conspicuouspattern of stripingdeveloped over the difference map on the rest ofthe delta surface that does not obviously correlatewith likely depositional or erosional processes.These are a result of refraction residuals(Fig. 4D) in each survey. Note that the residualsare actuallyonly�0.5–1.5m in100–200mofwaterwhich is well within International HydrographicOrganization standards, which are typically � �1.5% of water depth (International HydrographicOrganization, 1998).

By contrast, the second difference map (Fig. 5B)was obtained using much more frequent soundspeed profiles collected in the local area

throughout the survey. As can be seen, the stripingis nearly absent (the two surveys were run ortho-gonally to each other and thus the contributionof each survey should be apparent). Only usingthese methods can one start to assess the scale ofthe over-bank sedimentation that contributes to thelong term growth of the delta front.

Tidal reduction, measurement and models

However good all the other integrated componentsof thedepthmeasurement are, ultimately thedepthmust be referenced to a stable vertical datum. Tidalreduction has always been a necessary step. Fortraditional coastal hydrographic surveys the stan-dard has been to install a local gauge.

This approach is valid for regions in the localarea that share the samephase andamplitudeof the

October 2005

- April 2004

November 2006

- March 2006

2004-2005 surveys :

• used sparse sound speed profiles

• predicted tides

(still within IHO Order 1 accuracy)

2006–surveys used:

• local tide gauge

• dense sound speed profiles

Bathymetry

Squamish Delta

500m

Depth in m

>= +3mdeposition

<= -3merosion

A

B

C

Fig. 5. Example of the effect of water column uncertainty on estimates of seafloor change. (A) difference map between twosurveys (EM1002) in 50–200m of water. (B) bathymetry of the Squamish Delta, both surveys using sparse sound speedprofiles, obtained several kilometres away. (C) differencemapbetween two surveys (EM3002). For both surveys, the greyscaleused is the same between �3m and þ 3m. Differences greater than this are thresholded to black or white.


tide. However, as onemoves along restricted coast-al areas or onto the open continental shelf, knowl-edge is required of the propagation of the tidalwave. This is often expressed in terms of a co-tidalchart, where regions are defined inwhich tides at areference station need to be scaled and delayed tobe valid in adjacent regions (Admiralty, 1969). Forcoastal areas, this knowledge is based on historiclocation of tide gauges at adjacent locations alongthe coast. However, as one moves out onto thecontinental shelf, adequate tidal data are lacking.For an open coastline the propagation of the tidalwave from the edge of the continental shelf waspreviously poorly known. However, recent model-ling, based on analysis of Topex Poseidon seasurface elevation data (e.g. Dupont et al., 2002) hasresulted in the development of hydrodynamic

models that predict the propagation of the waveacross outer-continental shelf regions. In thisman-ner, a dynamic tidal solution may be calculatedalong the track of the survey vessel, appropriate forthe location and time of the vessel at every point.

In Fig. 6, an example on one such model for theBay of Fundy is presented. The model is based onthe WebTide (Department of Fisheries andOceans, 2005) hydrodynamic model, which isavailable for the entire Canadian continentalshelf. The resolution is variable and uses a finiteelement triangulated network (Fig. 6A). The Bayof Fundy is a region in which the amplitude of thetide (Fig. 6B) more than doubles as one moves upthe bay and the phase is successively delayed(Fig. 6C) particularly at constricted regions(Greenberg, 1979).

(D)50 km

B C

A D

Finite Element Mesh

5.5

5.0

M2 Amplitude M2 Phase

1 contours (2.07 min s)~ 10 cm error at max flood

for 10m M2 tides

draught

EGM96 Ellipsoid

Geoid Separation-21

-22

-23

-20Separation (m) CCGS Matthew 2007

JD127 JD137

Fig. 6. (A) Resolution of the hydrodynamicmodel for the Bay of Fundy; (B) the resulting distribution of tidal amplitude; (C)the phase of the constituents (M2 illustrated here, but available for all ofM2, S2, N2, O1, K1); (D) in order to take advantage ofGPS heighting, the addition of an ellipsoid-geoid separation model is required. The profile represents the correctionnecessary to shift measurements from the ellipsoid to the geoid over a 20-day period of operation (the ship tracks areillustrated in D).


Of most concern are constricted regions wherethe tidal wave is impeded and the phase contours(Fig. 6C) are tightly spaced. In these regions, a localgauge can become invalid within a few kilometres.Themodel illustrated here has been adopted as theprime reference for reprocessing of multibeamsurveys from 1992 to 2007 in the region. Thisapproach was chosen over conventional tidegauges due to the problems associated with thenecessity of maintaining multiple gauges andaccurately defining the tide in the central bay.Disadvantages of tidal models are twofold:

1. Their accuracy offshore is hard to assess. It isgenerally only tested against point stations onthe coast;

2. They cannot predict non-tidal sea level signa-tures due to, for example, atmospheric pressurevariations, or wind-driven sea surface run up.

In the event of there being an unmodelled ampli-tude or phase error in the applied tidal profile, itwill not be immediately apparent if sequentialsurvey lines are just a few tens of minutes apart.This is because the magnitude and sign of theresidual error will change only with periods simi-lar to the tidal forcing. But if a pair of lines is runwith a time gap between them of several hours(more strictly with a significant change in tidalphase) then the sign and magnitude of the tidalresidual error is unlikely to be the same.

Figure 7 illustrates the analysis of a repeat surveyon the continental shelf in which the inter-surveydifferences are clearly dominated by the tidal sig-nature. Two types of residual are seen. Graduallychanging magnitude and sign of the differenceacross the survey progression indicate either aphase or amplitude error in the tide. Abrupt stepsin the sign of the difference along a shiptrackindicate that the survey has been broken for anunspecified period. Only the second type of errorwill show up in the short wavelength morphologyas an abrupt inter-line step. The first type of error,results in only a few centimetres difference in theerror between adjacent lines (even though both areactually wrong). The example in Fig. 7 is of twosurveys, one day apart and using an identical plat-form just using orthogonal survey line orienta-tions. As most of the error sources cancelled out,the inter-survey bias was minimal (1 cm), but thetidal errors are seen to be the dominant signatureeven though the tide gauge was only 10 km away.BothFig. 7 andFig. 5A illustrate that it is important

to know the survey line orientation when examin-ing surface difference maps. Any apparent linea-tion that is parallel to one of the two survey lineorientations should be treated with suspicion. If acertain sedimentary process that has a preferredgrain is suspected, the survey lines should beoriented so that any systematic biases would notbe confused with the natural process of interest.

As the magnitude and sign of the tidal error onlychange over periods of hours, a strategy of avoidinglong survey lines that only generate overlap afterseveral hours should be adopted, therebyminimiz-ing interline errors. The preferred sequence wouldbe to break up large areas into several regions withline lengths no more than about an hour. Note thatthe error is still present but is not manifested asabrupt line to line mismatches. This provides amuch clearer view of the geomorphology. Stepswill still be generated at survey region boundaries.Similarly survey strategies that involve “racetrack” strategies, where alternate lines are runwithfill in lines at other phases of the tides are to beavoided. This is often implemented for vessels thathave a large turning radius compared with the linespacing.

Note that this strategy of breaking up large areasinto smaller sub-regions is actually complemen-tary to the aim of minimizing water mass variabil-ity as the data collected are within a similar watermass, and sound speed sampling strategies can bedesigned around a shifting box survey region. Anincreasingly-used alternative to the tidal measure-ment and modelled interpolation is to use a GPS-derived ellipsoid height. Conventional differentialGPS heights are in the � 5m vertical accuracyrange and thus of no value. Kinematic GPS offersthe best solution but are limited by separation ofplatform and base station (generally to less than20 km, USACE, 2002).

An emerging approach involves the GcGPS ser-vices such as C-Nav or OmniSTAR which offer avertical accuracy of several decimetres. Withsmoothing, this provides an adequate result fortidal correction in continental shelf waters whereone is beyond the practical range of kinematic GPSand the tidal propagation models are uncertain.Themajor problemwith these services is reliability(Hughes Clarke et al., 2005). The vertical profileneeds to be filtered and edited to account for dis-continuities and interruptions.

Before any ellipsoid height model can be used,the separationbetween that surface and thedesiredreference vertical datum (usually either Chart


Datum or Mean Sea Level) needs to be established.For a small area (less than a few kilometres) a singleshift canoftenbeapplied,but typical geoid-ellipsoidsurface slopes are in the 3 to 10cmkm�1 rangeand thus for continental shelf areas, one needsto have a model of the geographic variation inthe separation. Figure 6D illustrates the EGM96(Lemoine et al., 1998) ellipsoid to geoid surfaceseparation used for the Bay of Fundy. The super-imposed ship tracks run several hundred kilometresup and down the bay and thus require a continu-ously varying separation to be applied to the data(profile inset in Fig. 6D). In this manner repeatsurveysmaybe conducted and referenced to a stabledatum (the ellipsoid) wherein one can start to esti-mate sedimentary change at a vertical scale of a fewdecimetres.

BACKSCATTER

Increasingly, spatial variations in the seabed back-scatter strength are being used as an additional toolto aid in interpretation of shelf sedimentary pro-cesses. In order to use this effectively, a properunderstanding of both the physical controls onseabed scattering and the effect of sonar radio-metric and geometric imaging is required.

Physical controls on seabed scattering

Seabed backscatter strength is driven by theseabed’s physical properties (Jackson et al.,1986) and thus is potentially a useful indicatorof sedimentary environment. A direct correlationbetween acoustic backscatter strength and a simple

mean difference: -1.0 cm

standard deviation: 11.5 cm

In these water depths (45-55 m), these

difference statistics (2σ of ~ 23 cm) are

comfortably inside IHO Order 1

(82 cm or 1.64 %).

Surface

Difference

greyscale used

is ± 30 cm

Maximum rate of

change of tide

Maximum rate of

change of tide

Break in

survey

Break in

survey

Change in vessel

Speed (squat)

Repeat multibeam survey - orthogonal direction

same vessel - 24 hours apart

Line

spacing

Line spacing

Fig. 7. Map of inter-survey differences for two EM1002 survey performed 24hours apart. The region is an extremely lowrelief sand seafloor in the English Channel. Two rock outcrops can be seen on either side of the map. They appear as adisturbance in the difference map due to slight timing and bottom tracking uncertainty. There was no real change in theseafloor between the surveys, yet the difference map illustrates that the inter-survey errors are not random. Rather they aredominated by imperfections in the tidal model. Such patterns need to be understood before any real sedimentary change canbe interpreted.


quantity such as grain size has been inferred (e.g.Borgeld et al., 1999) but in general remains elusivebecause spatial variations in backscatter mayreflect changes in one or all of the following:

. Impedance contrast of the seabed/seawater inter-face (controlled by the bulk density and soundspeed in the sediment);

. Interfacial roughness of that sediment waterboundary;

. Volume heterogeneity – changes in the patchi-ness and contrast in the very shallow subsurfaceimpedance;

. Changing grazing angle (Fig. 8).

Even at a fixed grazing angle, it can thus beambiguous as towhether a change viewed is result-ing from a change in impedance, roughness, orvolumeheterogeneity.Hamilton&Bachman(1982)

demonstrated that, for terrigenous sediments, theimpedance is strongly correlated with grain size. Itwould be convenient if this were the principalcontrol on backscatter strength but, for a givengrain size, the interface roughness is linked tootherfactors such as sorting or rippling or thepresence ofshell hash. For fine-grained sediment (where thereis significant penetration into the sediment), thevolume heterogeneity is controlled strongly bybioturbation and/or the presence or absence ofburied shell debris or glacial dropstones.

Distinguishing outcrop or cobbles from fine-grained unconsolidated sediments is not an issueas the backscatter strength contrast between graveland mud is unambiguous for all grazing angles(Fig. 8B). For the case of most temperate conti-nental shelves, however, the variations of interestoften range only frommuddy sands to sandymuds.Under these conditions, the simple grain-size

Grazing Angle (θ) 90 0

BS

(dB

)

-50

-10

Typical Sediment

Angular Response CurvesBottom Backscatter Strength

Angular Dependence

(built into swath imaging geometry)

(A) (B) (C)

θ

(C) BEYOND

CRITICAL ANGLE

head wave

θ

(A) VERTICAL

Ref

lect

ed a

nd

back

scat

tere

dVol. Scat.

θ

(B) OBLIQUE

A BC

A

B

C

Fig. 8. The effect of grazing angle on multibeam geometry and typical angular response curves. Cartoons illustrating thechanging role of the differing physical scattering processes for the three main parts of the angular response curve are shown(Vol. Scat.¼Volume Scattering, Critical¼ angle beyond which no sound is refracted into the seabed).


correlation can be obscured by other factors such assorting, rippling, bioclastic debris and bioturbation.

Grazing angle effects

Even for a given set of sediment physical proper-ties, the backscatter strength will vary with graz-ing angle (Fig. 8B). A typical swath will imagefrom vertical incidence (90� grazing) to grazingangles usually as low as 25� (Fig. 8A). Thus, ameasure of bottom backscatter strength will varyacross the swath, providing at first glance, a mis-leading picture of the sediment distribution. Forpractical mapping purposes, the geological inter-preter wishes to view an image that reflects regionalsediment variations without having to continu-ously be aware of the imaging geometry. To achievethat, a compensation algorithm needs to be est-ablished that effectively “flattens” the angularresponse curves (Fig. 8B). To do this, of course,requires a priori knowledge of the shape of thatcurve.

The curve shape however, is highly variablebetweendiffering sediment typeswith strong spec-ular peaks, of varying width and differing rates ofroll-off with low grazing angle (Fig. 8B). Thus thereis a need to locally adjust the compensation algo-rithm to reflect the local angular response (AR)curve. However, this is potentially a circular argu-ment, as the AR curve needs to be derived from theseafloor and thus one needs to assume that thesediment type is constant from side to side in asingle (or series of adjacent) swath. For continentalshelf depths (50–200m) this translates into anassumption of spatial sediment invariance overa distance of 200 to 800m. Without this assump-tion one risks interpreting across track sedimentchanges as unusualARcurveswhich could then becompensated incorrectly.

Sonar radiometric and geometric influence on thereceived scattering intensity

All of the above discussion assumes that one has acalibrated measure of the bottom backscatterstrength (BS). To achieve this requires a completeknowledge of the sonar system settings.

Source level and receiver gain settings

Depending on the sonar system, the source leveland receiver gains may or may not have alreadybeen compensated for. For Kongsberg systems, the

receiver gains are automatically set to adjust forsource level, spherical spreading, attenuationand pulse length and seabed backscatter variations(but assuming a locally flat seafloor) (Hammer-stad, 2000). The only compensation necessary forthese systems is slight adjustments for exact pulselength used, beam pattern residuals (see later sec-tion) and true seafloor slope (see later section). Incontrast, the RESON family of sonars maintain afixed receiver gain ramp but log all the radiometricparameters including source level, pulse lengthand fixed gain steps. Before data can be used forgeological interpretation, all the calculations needto be applied in post processing (e.g. Beaudoinet al., 2002).

The most fundamental measure is the sourcelevel of the sonar. Few multibeam systems areprecisely calibrated and thus an absolute levelcannot be relied upon. The usual proxy is that, forthe duration of a deployment, the source level is aconstant. Such an assumption will break down if asurvey consists of multiple deployments withchanging sonar hardware. Overlapping coveragebetween surveys performed with different hard-ware settings may be the only way to maintaina stable relative calibration (Hughes Clarke et al.,2008). Thus when trying to quantitatively assesswhether a change in backscatter imagery betweentwo surveys is real, the usermust attempt to grosslyshift the data to match in regions where it isbelieved that the seabed sedimentary environmentis unaffected. Even if the data in a certain region arefixed, one needs to account for the effect of chan-ging seasonal oceanography,which is expressed inthe seawater attenuation coefficient.

Seawater attenuation

The received intensity is a function of the attenua-tion taking place in the seawater. This attenuationis dependent on the frequency of choice and variessignificantly with temperature and salinity (Fran-cois & Garrison, 1982a, b). It is up to the user toapply the appropriate value. One of the main var-iations reflects the changes in freshwater influenceas onemoves within the coastal zone. For exampleat given salinities the received intensity (all at 10mdepth, 10 �C, 100 kHz):

. 33ppt salinity (typical coastal ocean):32dBkm�1;

. 27ppt salinity (typical distal river plume or fjordbasin): 27dBkm�1;


. 15ppt salinity (typical brackish estuary):16 dBkm�1.

Fortunately river plumes are normally restrictedto the upper few metres of the water column andthus the depth-averaged attenuation coefficient isless affected. But within fjord basins, separated bysills, the bulk change in the salinity from basin tobasin, if not accounted for, will alter the apparentbackscatterstrengthofthebasinfloor.Forexamplein100m water depth using a 60� beam (400m roundtrip) this corresponds to a 2dB change for a changefrom27to33ppt.Thus,unlesscompensated for (notstandard in most post-processing software), onecannot discern whether there is a change in thesediment type up the fjord, or merely a change inthe water mass.

Perhaps more misleading is the fact that such abias is depth dependent. For a single basin, theimage will appear consistent, but, with the wrongattenuation coefficient, the interpreter may infer adepth-correlated change in sediment type. Usingthe same example (33 v. 27 ppt), the same materialwill appear 4 dB weaker from the beach to 200mdepth and 8dB weaker at 400m depth. Comparedwith that, the BS variation between fine sand andmud is only 2–6dB. Many sedimentary environ-ments are depth-dependent as they depend onsurface wave activity or current strength and thusthe user is easily led into believing depth-relatedapparent sediment variations.

Another effect is the seasonality of the watertemperature (all at 10mdepth, 33 ppt S – 100 kHz):

. 5 �C – 27dBkm�1;

. 10 �C – 32dBkm�1;

. 15 �C – 35dBkm�1.

Thus if a regional survey starts in the spring(5 �C), but continues, or is compared with one inthe late summer (15 �C), a 100m depth solutionusing a 60� beam, (a 400m round trip)will exhibit a3.2 dB difference.

Pulse length changes

Except when operating in the shallowest rangeof depth, most sonar systems are operating atfull power the upper level of which is normallyrestricted by cavitation issues. As the water getsdeeper, the received signal strength will drop,resulting in a loss of signal to noise. To circum-vent this, one needs to increase the pulse length.

Doing so for the same source level increases theinstantaneously ensonified area resulting in botha stronger signal and, for narrow band signals, alower range resolution.

This has three effects on the geologicalinterpretation:

1. Unless compensated for, the seabed intensitywill appear to change. Even for those systemsthat do so, the compensation is never perfect;

2. If the interpreter is relying on the pixel specklecharacteristics to discern different sedimenttype, the speckle pattern will coarsen withlonger pulses;

3. If the interpreter is looking to resolve smallfeatures, the longer pulse will be defocused,making some short wavelength features suchas ripples or cobble fields disappear.

To compensate for effect 1, a measure of thepulse-length needs to be maintained with the dataand, based on analysis of the shift at changes, abulk and/or range and angle-dependent offsetneeds to be applied.

Thesecondeffectcanbemostdamagingtosomeoftheautomated texturalclassificationsoftwareonthemarket (e.g.Milvangetal., 1993;Prestonetal., 2001)both of which in part rely on the “Pace” features(Pace & Dyer, 1979; Pace & Gao, 1988). At this time,this approach cannot take into account pulse lengthchangesandthusautomatedclassification is limitedto regions where a single pulse length is used.

For the third effect, the loss in resolution isgenerally less of an issue because, at the point atwhich the pulse length needs to be shifted, the datawith the shorter pulse length are compromised inany case by the lower signal to noise levels.

Beam patterns – single and multiple sectors

Both the transmitter and each of the individualreceiver beam patterns have intensity/sensitivityvariations with elevation angle. The combinedeffect of these two beam patterns will generate var-iations in received intensity across the swath thatmight be confused with seabed sediment changes.

The simplest configuration is a single-headed,single sector, multibeam inwhich the entire swathis illuminated by one transmitter. In this case, thetransmit beam patterns are generally simple, vary-ing only slowly with angle (e.g. RESON 8111,Beaudoin et al., 2002). A notable exception is theoriginal EM1000 which used a barrel array for the


transmit and thus variations in the intensity fromthe staves within the barrel could produce com-plex transmit beam patterns. In both cases, thepattern is fixed with respect to the array. For theEM1000, because each receiver channel, which isroll-stabilized, uses a separate amplifier, any inter-amplifier differences will show up as an apparentvertically-referenced beam pattern residual.

An additional complication is found in multisector systems. Several systems: EM12, EM1002,EM300, EM120, EM710 use multiple sectors. Thisis done to provide advantages in multiple suppres-sion, improved pitch and yaw stabilization (Fig. 3),improved water column imaging (Hughes Clarke,2006) and to allow transmit focusing (Kongsberg,2005). Each sector has a unique center frequency toavoid interference between sectors.As a result eachsectormayhave slightly different calibrations.Alsothe seabed angular response can be subtly differentat the different frequencies and each frequency hasa slightly different attenuationcoefficient.All thesefactors conspire to make the sector boundariesshow up in the backscatter data (Llewellyn, 2005),potentially confusing geological interpretation.

In theworst case, both sonar-referenced and verti-cally-referenced beam pattern artefacts may beapparent in the data and require compensation.Figure 9 is an example of this, although using datafrom a malfunctioning sonar to better illustrate theeffect. The data are from an EM1002 that has threetransmit sectors whose beampatterns are fixedwithrespect to thesonar.Thesamesonar receiversare rollstabilized and thus vertically referenced. Thus onesees the effect of rolling transmit beam patternstruncated at the vertically-referenced sector bound-aries. An estimate of the beam pattern (describedbelow) has to be collected separately for each of thesectors. Once estimated, by combining this withknowledge of the vessel roll at transmit, the twosignatures may be predicted and removed fromthe data. As stated, this is an extreme example with> 10dB beam pattern nulls. However, such signa-tures at levels of only 2dB are still common andhamper interpretation of typical continental shelfseabed sediment signatures that are of similar mag-nitude (Iwanowskaetal., 2005).More typically,onlythe vertically referenced sector boundaries show up(Figs. 10 and 11).

Backscatter data manipulation strategies

Given all the imperfections outlined above, thereal-time backscatter output of the multibeam

sonar systems will contain artefacts that hamperthe ability to undertake regional sediment distri-bution analysis (e.g. Fig. 11A). Themost noticeableeffect is that of residual beam pattern and grazingangle distribution. Thus strategies need to bedeveloped to minimize these artefacts.

Estimating residual beam pattern and grazingangle variability

In order to remove the beam pattern and grazingangle effects, one ideally needs to know the trans-mit and receive beam pattern sensitivities (bysonar and/or vertically referenced angle, Fig. 9)as well as the local seabed angular response curve(by seafloor grazing angle, Fig. 8). As these are allunknowns, thismust beguessedbasedon the inten-sity variations by a combination of sonar-relative,vertically-referenced and seafloor-referenced angle.Unless one of the three signatures is dominant, itis practically impossible to separate them. Thus thevertically referenced angle is usually used, as,averaged over several 100 or 1000 pings, the sonar-relative angles will oscillate about zero and theseafloor slope will on average be level.

The operator is left with a choice of length scaleover which to average. The longer the averaging,the more likely that local across-track geologicalvariations will average out. An array of intensitiesby beam referenced angle is maintained (e.g.Fig. 9A) and the statistics of the average intensity,normally in 1� bins, is compiled. A reference levelrepresentative of the average signal strength is thenselected and intensity offsets (multipliers in linearintensity or additive offsets in logarithmic inten-sity) are calculated for each 1� bin. The data arethen adjusted so that all beams at a certain 1� binhave a fixed offset in intensity applied.

Coping with geographically varying angularresponse curve shapes

Given the strong beam pattern and grazing anglesignatures that will be present in multibeam back-scatter data, strategies need to be established tominimize these. Figure 10 illustrates strategies forachieving this.

For a single line, statistics can be gathered onintensity variations with incidence angle. Aver-aged over many geological terrains (as would becovered in a typical survey line), this representsthe best guess of the combined input of beampattern and grazing angular response.


The main limiting assumption is that the shapeof the AR curve for all sediment types, while ofdifferent mean level, is the same shape. However,the shape of the AR curve for different shelf sedi-ments is highly variable (Fig. 8B), some havingstrong and narrower specular peaks, others havediffering roll-off with low grazing angle, some-times including a critical angle cusp.

Thus it would be better to calculate this com-bined response separately for each sediment type.This requires, however, an a priori knowledge ofthe sediment distribution. One way to approachthis is to derive the statistics not regionally (i.e. thewhole line, or several lines) but locally (for a subsetof the line). However, the danger comes when onetries to define what locally is. If sediment typechanges over length scales of hundreds or

thousands of pings, one can select a similar lengthscale, but for such a small number of pings, onemay remove the valid assumption of geologicalrandomness. This is easiest to visualize by think-ing of sediment being different from one side of theswath to the other for the duration of the averagingperiod. An apparent lop-sided angular responseestimatewill result and the correctionwill attemptto flatten it.

Figure 10 illustrates this dilemma. Image (A)shows the EM1002 data with the Kongsberg flat-tening function applied. While the gross sedimentboundaries are visible, the sector boundary is over-printed and it is clear that the near-nadir back-scatter data is imperfectly flattened. Figure 10Bshows the result of estimating the response overthe entire line. The sector boundaries are now

Centre

98 kHz

Stbd.

93 kHz

Port.

93 kHz

Rolled–Starboard side up

Rolled–Port side up

Vessel Level

EM1002 with hardware damage

Cen

tre S

ecto

r

Po

rt S

ecto

r

Stb

d. S

ecto

r

AS ACQUIRED COMPENSATED

B

B

C

A

Fig. 9. Special case processing for multi-sector multibeam backscatter data. The example presented (data courtesy of theGeological Survey of Israel) has a hardware problem resulting in pronounced transmit beampattern residuals (shown byA, Band C). Statistics are compiled separately for each sector. The sector data are collected within specific vertically referencedsectors, but compiled by sonar-relative angle. While this is an extreme example for illustrative purposes, such patternsimpede geological interpretation and should be removed.


subdued (as they were present for the whole line)but it is apparent that the near nadir response isunder compensated at the north end of the line andovercompensated at the south end of the line.Figure 10C and D illustrate the effect of usingstatistics from just the northern or southern endof the line. In each case, the flattening algorithm issuperior in the region from which the statistics arederived but fails on the other sediment type.

The important factorhere is the contrast betweenthe flat near-nadir AR of gravels and the peakednear-nadir and steep low grazing angle drop offtypical of muds (Fig. 8B). In Fig. 10E a strategy ofcontinuously estimating the local incidence

angle–referenced response over a length scale of300 pings was employed. As can be seen, thisapproach best regionally suppresses the angularresponse. What is less apparent, however, is thatthismethod produces artefacts at sediment bound-aries where the sediment is not uniform from oneside of the swath to the other (“haloes” in Fig. 10E).

Figure 11 illustrates the approaches describedabove for a large continental shelf region (HecateStrait, Barrie, 2004, pers. comm.). Figure 11A isthe original data, while Fig. 11B shows the line-based strategy and Fig. 11C the rolling responsestrategy. A general improvement in the clarity ofthe likely sediment distribution is apparent. Two

As observed

with KM TVG

Average of

whole line

Average of

shallow gravel

Average of

deeper mud

Rolling 300 ping

local response

500 m

~10km “h

alo

es”

(A) (B) (C) (D) (E)North

South

Fig. 10. Empirical approaches to predicting and removing combined angular response and transmit-receive beam patternproducts. Data are a 10 km line collectedwith a� 65� sector, running from� 100mdepth gravels in the north to�200mdeepmuds in the south. (A) Beam trace data as delivered by Kongsberg Maritime (KM) using their predictive time varying gain(TVG) functions.Note the presence of pronounced beampattern residuals at the sector boundary transitions at� 50�. (B) Dataafter application of corrector for line-based (all 4600 pings) average intensity by vertically referenced angle. Note the removalof the sector boundary artifacts, but over compensation of the nadir response in the northern gravels andunder compensationof the nadir response in the southern muds. (C) Data after application of corrector based only on northern gravels (first 2000pings). Note good compensation for gravels, but under compensation for muds. (D) Data after application of corrector basedonly on the southern deep muds (last 2000 pings). Note the good compensation for the muds, but over compensation for thegravels. (E) Data after application of a rolling 300-ping-based local corrector. Note the excellent suppression of the nadirresponse throughout the whole line. One caveat is the “halo” effect on traversing abrupt oblique sediment boundaries.


final artefacts are still apparent, however. Thepulse length was increased for the lines in thesouthern end of the area. In Fig. 11D an empiricalcorrection was applied to account for this. Evenafter that, however, it is apparent that there is anabrupt small, but noticeable, gain offset half wayacross the image. This was due to replacement ofsonar hardware on board, mid way through thecruise. Thus, even after all these steps, it would behard to be confident in a regional change in seabedsediment type based on a resurvey.

While at first it may appear that the rollingresponse predictor (Fig. 10E and Fig. 11C) is theoptimal approach for interpretation of continentalshelf sediments, it is actually hiding valuableinformation from the interpreter. The grazing angleresponse of the sediment has been specifically

removed so that a particular sediment type willappear at the same grey level irrespective of enso-nification angle. However, there are many shelfsediment types that exhibit very similar AR curvesand that may be identical in the mid-grazing anglerange, but differ near nadir or at low-grazingangles. Figure 12 illustrates this concept. Image(A) is as collected. Image (B) is after rolling res-ponse correction. The second image is more pleas-ing to the eye, but between the two areas circled in(A), there is actually a change in sediment type.The two AR curves (Fig. 12C) are identical in themid range of grazing angles but one has a strongerspecular peak than the other. They are probablysimilar sediments with similar volume scatteringsignatures (that dominate the mid-range of grazingangles) but one has a smoother interface than the

(A) As observed (B) BP by line (C) Rolling BP (D) De-pulsed

EM1002 backscatter, 4 km x 10 km region – Greyscale range: -40 dB (Black) -10dB (White)

Fig. 11. Successive processing steps for EM1002 backscatter data. Datawere collected in� 150–250mofwater using a� 65�

sector. In the southern half of the survey, the depths were great enough to cause the sonar to jump from a 0.2ms pulse to a0.7ms pulse. (A) Original data with just Kongsberg 1st order TVG reduction. (B) Data empirically processed on a line-by-linebasis, reducing for the effects of average across track intensity by vertically referenced grazing angle (BP¼ beam pattern). (C)Data empirically processed with a rolling 500 ping local incidence angle function. (D) As in C but with an empiricallycalculated offset for the 0.7mspulse data (�2dB).Note that even after this it is apparent that, halfway across the survey, thereis a jump in backscatter (of only �2dB) due to in-field replacement of the transceiver electronics board.


sediments, morphology and sedimentary processes on ... · 42 carbonate systems during the...

Documents