New Techniques in Sediment Core Analysis (Geological Society Special Publication No. 267)

New Techniques in Sediment Core Analysis

Geological Society Special Publications

Books Editorial Committee Chief Editor BOB PANKHURST(UK)

Society Book Editors JOHN GREGORY (UK) JIM GRIFFITHS(UK) JOHN HOWE (UK) PHIL LEAT (UK) NICK ROBINS (UK) JONATHAN TURNER (UK)

Society Book Advisors MIKE BROWN (USA) ERIC BUFFETAUT(FRANCE) RETO GIERI~(GERMANY) JON GLUYAS (UK) DOUG STEAD (CANADA) RANDELL STEPHENSON(NETHERLANDS)

Geological Society books refereeing 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 Book Editors 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. More information about submitting a proposal and producing a book for the Society can be found on its 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: ROTHWELL, R.G. 2006. New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267. CHEN, Q. RACK, F.R. & BALCOM, B.J. 2006. Quantitative magnetic resonance imaging methods for core analysis. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 193-207.

GEOLOGICAL SOCIETY SPECIAL PUBLICATION NO. 267

New Techniques in Sediment Core Analysis

EDITED BY R. G. R O T H W E L L National Oceanography Centre, UK

2006 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 over 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 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 WIJ 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 (Orders: Tel. +44 (0)1225 445046, Fax +44 (0)1225 442836) Online bookshop: www.geolsoc.org.uk/bookshop

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. The Geological Society of London 2006. 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 W1P 9HE. Users registered with the Copyright Clearance Center, 27 Congress Street, Salem, MA 0 ! 970, USA: the item-fee code for this publication is 0305-8719/06/$15.00.

British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. ISBN-10:1-86239-210-2 ISBN-13:978-1-86239-210-6 Typeset by 4word Ltd, Bristol Printed by MPG Books Ltd, Bodmin, UK

Distributors North America For trade and institutional orders: The Geological Society, c/o AIDC, 82 Winter Sport Lane, Williston, VT 05495, USA Orders: Tel +1 800-972-9892 Fax +1 802-864-7626 Email [email protected]

For individual and corporate orders: AAPG Bookstore, PO Box 979, Tulsa, OK 74101-0979, USA Orders: Tel +1 918-584-2555 Fax +1 918-560-2652 Email [email protected] Website http://bookstore.aapg.org India

Affiliated East-West Press Private Ltd, Marketing Division, G-1/16 Ansari Road, Darya Ganj, New Delhi 110 002, lndia Orders: Tel. +91 11 2327-9113/2326-4180 Fax +91 11 2326-0538 E-mail [email protected]

Contents Acknowledgements ROTHWELL,R. G. & RACK,F. R. New techniques in sediment core analysis: an introduction HASCHKE, M. The Eagle III BKA system, a novel sediment core X-ray fluorescence analyser with very high spatial resolution RICHTER, T. O., VAN DER GAAST, S., KOSTER, R., VAARS,A., GIELES, R., DE STIGTER, H. C., DE HAAS, H. & VAN WEERING, T. C. E. The Avaatech XRF Core Scanner: technical description and applications to NE Atlantic sediments CROUDACE, I. W., RINDBY, A. & ROTHWELL, R. G. ITRAX: description and evaluation of a new multi-function X-ray core scanner THOMSON, J., CROUDACE, I. W. & ROTHWELL, R. G. A geochemical application of the ITRAX scanner to a sediment core containing eastern Mediterranean sapropel units ROTHWELL, R. G., HOOGAKKER,B., THOMSON,J., CROUDACE,I. W. & FRENZ, M. Turbidite emplacement on the southern Balearic Abyssal Plain (western Mediterranean Sea) during Marine Isotope Stages 1-3: an application of ITRAX XRF scanning of sediment cores to lithostratigraphic analysis ROGERSON, M., WEAVER,P. P. E., ROHLING, E. J., LOURENS,L. J., MURRAY,J. W. & HAYES, A. Colour logging as a tool in high-resolution palaeoceanography NEDERBRAGT,A. J., DUNBAR,R. B., OSBORN,A. T., PALMER,A., THUROW,J. W. & WAGNER, T. Sediment colour analysis from digital images and correlation with sediment composition JARRARD, R. D. & VANDEN BERG, M. D. Sediment mineralogy based on visible and nearinfrared reflectance spectroscopy RIBES, A. C., RACK, F. R., TSINTZOURAS,G., DAMASKINOS,S. & DIXONA. E. Applications of confocal macroscope-microscope luminescence imaging to sediment cores SCHULTHEISS, P. J., FRANCIS, T. J. G., HOLLAND, M., ROBERTS, J. A., AMANN, H., THJUNJOTO, PARKES, R. J., MARTIN, D., ROTHFUSS, M., TYUNDER, F. & JACKSON, P. D. Pressure coring, logging and subsampling with the HYACINTH system FREIFELO, B. M., KNEAESEY,T. J. & RACK, F. R. On-site geological core analysis using a portable X-ray computed tomographic system KLEINBERG, R. L. Nuclear magnetic resonance pore-scale investigation of permafrost and gas hydrate sediments CHEN, Q., RACK, F. R. & BALCOM,B. J. Quantitative magnetic resonance imaging methods for core analysis JACKSON,P. D., LOVELL,M. A., ROBERTS,J. A., SCHULTHEISS,P. J., GUyN, D., FLINT, R. C., WOOD, A., HOLMES, R. & FREDER1CHS,T. Rapid non-contacting resistivity logging of core GOLDBERG, D., MYERS, G., ITURRINO, G., GRIGAR, K., PETTIGREW, T. & MROZEWSKI, S. Logging-while-coring - new technology from the simultaneous recovery of downhole cores and geophysical measurements JENKINS, C., FLOCKS, J. & KULP, M. Integration of the stratigraphic aspects of very large sea-floor databases using information processing MOORE, C. J. • HABERMANN,R. E. Core data stewardship: a long-term perspective MITHAL, R. & BECKER,D. G. The Janus database: providing worldwide access to ODP and IODP data

vi 1

31 39

51 65 79

99 113 129 141 151

165 179 193 209 219

229 241 253

Acknowledgements The editor thanks the following colleagues and friends who kindly helped with reviewing the papers submitted for this volume: W. Balsalm A. Best B. Blfimich S. Carey A. Cattaneo C. Clauser S.E. Clavert T. Courp I.W. Croudace

Keh-Jim Dunn K. Emeis P. Francus L. Giosan D. Goldberg C. Graham C. Grattoni I. Hall

J. Harff J. Helmke J. Howe R. Jarrard D. Long M. Lovell M. Lyle D. Maloney

C. Moore A. Nederbragt L. Poppe B. Schnetger R. Taylor J. Thomson M. Weber W. Winters

The following companies are thanked for their contributions towards colour printing costs.

RIVILYTRI:L

k/

X-ray

~

O

Q

u

UNIVERSAL SYSTEMS

0Analytical Systems

New techniques in sediment core analysis: an introduction R. G U Y

ROTHWELL 1 & FRANK

R. R A C K 2

1National Oceanography Centre, Empress Dock, Southampton S014 3ZH, UK (e-mail." [email protected]) 2Joint Oceanographic Institutions, 1201 New York Avenue, NW, Suite 400, Washington, DC 20005, USA Abstract: Marine sediment cores are the fundamental data source for information on seabed

character, depositional history and environmental change. They provide raw data for a wide range of research including studies of global climate change, palaeoceanography, slope stability, oil exploration, pollution assessment and control, and sea-floor surveys for laying cables, pipelines and siting of sea-floor structures. During the last three decades, a varied suite of new technologies have been developed to analyse cores, often non-destructively, to produce high-quality, closely spaced, co-located downcore measurements, characterizing sediment physical properties, geochemistry and composition in unprecedented detail. Distributions of a variety of palaeoenvironmentally significant proxies can now be logged at decadal and, in some cases, even annual or subannual scales, allowing detailed insights into the history of climate and associated environmental change. These advances have had a profound effect on many aspects of the Earth Sciences, particularly palaeoceanography. In this paper, we review recent advances in analytical and logging technology, and their application to the analysis of sediment cores. Developments in providing access to core data and associated datasets, and data-mining technology, in order to integrate and interpret new and legacy datasets within the wider context of sea-floor studies, are also discussed. Despite the great advances in this field, however, challenges remain, particularly in the development of standard measurement and calibration methodologies and in the development of data analysis methods. New data visualization tools and techniques need to be developed to optimize the interpretation process and maximize scientific value. Amplified collaboration environments and tools are needed in order to capitalize on our analysis and interpretation capability of large, multi-parameter datasets. Sophisticated, yet simple to use, searchable Internet databases, with universal access and secure long-term funding, and data products resulting in user-defined data-mining query and display, so far pioneered in the USA and Australia, provide robust models for efficient and effective core data stewardship.

Sea-floor sediment cores are the fundamental data source for information on seabed character, depositional history and environmental change. Research into global climate change, slope stability, oil exploration, pollution assessment and control, surveying for laying telecommunications cables and offshore pipelines all rely on data obtained from marine sediment cores and samples (Table 1). Important oceanographic and earth science disciplines such as palaeoceanography rely on core material to determine past climate changes and changes in ocean circulation. Models of past climate changes can only be validated by examining the past record preserved in marine sediment and ice cores, and such records allow us to understand the past and predict the future world. References to marine sediments occur in Ancient Greek and Roman texts, but it was not until 1773 that the first recorded sediment was recovered from the deep sea. In that year 'fine soft blue clay' was sampled with the first

recorded deep-sea sounding made by Captain John Phipps on HMS Racehorse in 1250m water depth on the southern margin of the Voring Plateau north of Norway. Forty-five years later in 1818, Sir John Ross recovered 2.7 kg (6 lbs) of greenish mud from the floor of Baffin Bay, offshore Canada, using a deep-sea grab. This recovery from 1920m water depth represents one of the first recorded successful substantial deep-sea sediment recoveries. In 1851 the first functioning submarine telegraph cable was laid across the Straits of Dover and the advent of submarine cable laying as a new means of intercontinental communication led to a rapid growth in the collection of deep-sea soundings and samples. However, it was not until the Challenger expedition of 1872-1876 that enough deep-sea samples were recovered to produce the first global sea-floor sediment map. Victorian intellectual curiosity led Sir Charles Wyville Thomson, Professor of Natural History at Edinburgh University, and his Canadian-born

From: ROTH'WELL,R.G. 2006. New Techniquesin Sediment Core Analysis. Geological Society, London, Special Publications, 267, 1-29. 0305-8719/06/$15.00 9 The Geological Society of London 2006.

2

R.G. ROTHWELL & F.R. RACK

Table 1. End users of core and sea-floor sample data Scientific research

Industry-related research

Training

Research into environmental change

Sea-floormapping and surveys (ground truth)

Undergraduate, postgraduate and professional training

Palaeoceanographic research

Hydrocarbon exploration

Studies of slope stability

National resource assessment

Geochemical studies

Pollution control and assessment

Geochronological studies

Environmental protection and monitoring

Studies of sedimentary processes and dynamics

Surveying for laying submarine cables, pipelines and siting sea-floor structures

Benthic surveys

Studies of acoustic response and defence applications

Biological and productivity studies

student John Murray to conceive an oceanographic expedition of global extent. They successfully persuaded the Royal Society to back such a voyage and the British Navy provided HMS Challenger, a three-masted square-rigged wooden ship of 2300 tons displacement and some 226 feet long overall, for the purpose. The Challenger expedition was the first large-scale expedition devoted to oceanography. During its 4-year voyage, the ship recovered a large number of sea-floor samples from 362 observing stations, spaced at uniform intervals, along the cruise track, which spanned 128000km. The initial analysis of these samples was made by John Murray who edited the Challenger reports following Wyville Thomson's death in 1882. In 1891, with his co-worker A.F. Renard, he published the milestone Challenger report on 'Deep-sea Deposits', the first comprehensive volume on sediments of the deep-sea floor (Murray & Renard 1891). This volume introduced many descriptive terms still used today such as 'Globigerina ooze' and 'red clay', and provided a firm basis for further deep-ocean sediment studies. A fundamental step forward in the recovery and investigation of deep-sea sediments was the invention of the gravity corer by German researchers. This allowed recovery of longer continuous sections of sediment, although these were generally restricted to 1-2m in length. Gravity corers were used to collect several 2 mlong cores on the German South Polar expedition (1901-1903) and these were described by E. Philippi in 1910. These cores provided the first evidence that some deep-water sediments were stratified. From 1925 to 1938, Germany ran a number of oceanographic expeditions

using the ship Meteor, which recovered several 1 m-long cores from the South Atlantic and Indian oceans. Wolfgang Schott used these cores to demonstrate changes in foraminifer species with depth, initiating the new field of palaeoceanography. The Swedish Deep Sea Albatross expedition of 1947-1949 saw another fundamental advance in the recovery of deepsea sediments by deploying a new kind of coring device developed by B6rje KuUenberg, a marine geologist working in Gothenburg, Sweden. This was the piston corer, an innovative modification of the traditional gravity corer, in which the coring tube fell past a stationary piston at the end of the wire. This mechanism expelled water from the falling tube above the piston admitting sediment from below. This allowed retrieval of much longer (typically 10 m long or more) and much less disturbed sediment cores. Acquisition of long piston cores made possible the study of Pleistocene ocean history and initiated the era of modern deep-sea sampling. Although other types of corer have been developed, piston coring still forms one of the main methods for sampling the deep-sea sedimentary record. Sizing up of this design has led to the development of giant piston corers, now capable of obtaining sediment cores of up to 60m in length. In 1990, the French vessel Marion Dufresne obtained a 54m-long piston core covering 4 M a of sedimentation from the Indian Ocean - one of the longest piston cores so far recovered. However, it is only within recent years that we have understood the mechanics of coring process, especially in the taking of long piston cores. Thouveny et al. (2000) demonstrated through magnetic susceptibility measurements of sediment fabric and

NEW TECHNIQUES IN CORE ANALYSIS sedimentation rate studies of long piston cores from the Portuguese margin that the sequence recovered in some, if not all, long piston cores can appear up to 1.5-2 times longer when compared to the same sequence recovered by conventional piston cores on the same site. This is due to syringing or oversampling of the sediments in the upper portion of the core resulting from cable rebound that causes upward piston acceleration and a microfabric rotation to the vertical during the coring process. Skinner & McCave (2003) studied the coring mechanism from a soil mechanics perspective and confirmed that the upper 5-25 m of long piston cores are affected by such oversampling. They suggested that to get the least deformed record for stratigraphic analysis, a large diameter (De of c. 20-30cm) square-barrel gravity corer for the top 10-12m of the sediment section should be combined with a cylindrical piston corer for below - 10 m sub-bottom. The capability to routinely acquire long, undisturbed marine sediment sections has provided a major impetus to palaeoceanographic research and to the development of associated international core collection programmes. One of the best known of these is the IMAGES (International Marine Past Global Changes Study) programme, a global project, running since 1995, which organizes cruises to collect long piston cores specifically to gain greater understanding of the mechanisms and consequences of climatic change. Eleven international IMAGES coring cruises were completed between 1995 and 2003, collecting in total 11.5 km of core. As part of the programme, in 1996 IMAGES created a technical standing committee (called 'New Techologies in Sediment Imaging') to research new data acquisition tools and relationships with proxies. The advent of the Deep Sea Drilling Project (DSDP) in 1968 took deep-sea sampling technology further and heralded a new era in the exploration of the deep-ocean sedimentary record. Using the dynamically position drillship Glomar Challenger, DSDP set out to recover long geological time duration continuous or semi-continuous sediment records from the world ocean. Prior to DSDP, the global inventory of cores containing pre-Quaternary sediments was less than 100. DSDP and its successor, the Ocean Drilling Program (ODP) which began drilling operations in 1985 using a new drillship with improved capabilities, JOIDES Resolution, have led to major advances in our understanding of Earth's history, the processes of plate tectonics, and the Earth's crustal structure and composition. The Ocean Drilling

3

Program collected a little over 222km of core from January 1985 to September 2003, which are stored at a number of repositories in the USA and at Bremen, Germany. Recently, international deep-sea drilling has embarked on a new phase with the advent of the Integrated Ocean Drilling Program (IODP, to run from 2003 to 2013), which builds upon the success of ODP and DSDP but expands the research of these programmes by using several drilling vessels, including riser, riserless and mission-specific platforms to achieve specific research goals. Specific research areas within IODP are the deep biosphere, environmental change, and solid earth studies and geodynamics (IODP 2001). Within IODP, the USA will continue to deploy a riserless vessel for 2-month drilling legs around the world. In addition, a riser vessel, Chikyu, supplied and operated by Japan, will provide a platform for long-term expeditions at specific locations. Mission-specific platforms for drilling in ice-covered and shallow-water regions will be operated by the European Consortium for Ocean Research Drilling (ECORD) as part of IODP. The first ever drill sites drilled through the ice-covered Arctic Ocean, implemented as part of the IODP programme, found evidence of a warm, ice-free Arctic during the Palaeocene-Eocene Thermal Maximum some 55 Ma ago, a precursor of conditions that may return in a greenhouse world of 2100 (Kerr 2004), a graphic example of how studies of cored marine sediments from the past can be used to predict conditions in a future world. The great growth in core collection over the last five decades and the ever-increasing need to extract more high-resolution information from cores has led to the development of a range of often non-invasive, non-destructive, core-logging and imaging techniques (Table 2). These techniques provide researchers with automated instruments that can continuously and rapidly characterize sediments downcore in terms of their density, physical, magnetic, optical, geochemical and acoustic properties, providing far more measurements than ever could be reasonably obtained using traditional invasive analytical methods. Such logging instruments when used at sea provide immediate characterization of sediment properties, allowing costeffective use of expensive sea time and almost immediate guides for stratigraphic correlation and further sampling. Although non-destructive core-logging techniques can, in some instances, be less precise than conventional laboratory analyses (for instance, X-ray fluorescence (XRF) data derived from XRF core scanners compared to destructive laboratory XRF analysis), this

4

R.G. ROTHWELL & F.R. RACK

0

P

E

~-~

~ ~ ~ ~

~ ._~ ,.~, ~,~

0

._

~ ~:~2

.~

, o=

=

~,

~8 ~ - ~

~9

~.__. ~ . ~ [

9

g

o

9

9

9

0

..-=

9-

g

"~ ~ "~

&

.o

-E

r~

"'-

s

E

~

0

r~ g

g ..h

;~

"*~Ua -

N] 0

O9

e4

on

~o

-" .=~o~

9

9

NEW TECHNIQUES IN CORE ANALYSIS

r-,l

r.~

~ _ ~

0

N zz

~

0

0

. ....;

~

O x , -o~x ~,v-

",.-./

c~

~m

"~

.~ ~ ' ~ o

~" - "

~.~

~

~

.__.=~ ,~o

9

9

9

e

9

9

9

==

~_

9

9

9

~

9

9

i .~8~

r~

r-L~

81

~~~

~

~=.~

_~

~ ~

~o~

.~

~v}o~

-~

~.. ~

~-= ~

8r~ 8

D~

.,--~ ~=~0~

0 o

o1}

O~ o

8

{/} c~

z~

>

~

.~

~8

r~

,._,,

.~

~ o _ " ~

~ :~

~

~

~

~

-~'-

. . . . . . . . . . . .

o

2.~

o

~

~.

~ o~

~ ~

-~ "~'~

2~;,~

o:~':

6

R.G. R O T H W E L L

o

& F.R. R A C K

._.,

t'--I

~,

.~

,~ o

b

~3

=

~, ~ ._ ~ , ~

~

o

~

~

N.~ ~.~.-~ ~ ~

~-= ~

..,

.~ ~ ~

.~

O ",'-'

~,-

~ 9,~

N

~

.~.,-

~"-' ~

~, , . ~ - .

~

~

~

~.~ 9 ~ ~ ~-~

~ ~ o ~ o ~

'~= ~ . = o~ o ~ _ ~~ ,X~ ~ .~ -~

,=o~

~F~ ....

,-,

~

i.= aS I

9

0

9

9

I

9

9

9

9

9

9

9

9

I

9

9

0

~

~

9~

o

.~

~._

o= o

o

9

, •

~ ~-~o=

~'~

.-~ i~b~o

i

•

;g&

r-,

o

.B o o

"~

• ,<

.~,

~

O ..-~ . ~

"-7'.

e~

3b

o 0

E~

~ o

e-,

~'~ , •

~Oo ~.~

;>',

8

" ~

O

.~ .~ .o .~ o

~N

~9

~'~'~

-~ .r-, .--

o

.-~

._-

~

=~

= o ._

~

o

9

9

NEW TECHNIQUES IN CORE ANALYSIS disadvantage is offset by statistical advantages derived from gaining far larger numbers of point data measurements and constantly improving processing techniques that improve the signal to noise ratio. Other advantages are rapidity of acquisition of datasets and relative lack of operator effort compared to laboratory analysis. These automated or semi-automated logging techniques have evolved rapidly and have been reviewed in a series of reports and papers. In July 1997, the French Research Institute for the Exploitation of the Sea (IFREMER), Brest, organized a Core Logging Workshop, under the auspices of the EU-funded CORSAIRES European Concerted Action (1996-1997), which reviewed developments in a wide range of logging methods (Auffret 1997). Shortly afterwards, the IMAGES standing committee on 'New Techologies in Sediment Imaging' produced an interim report surveying emerging trends in non-destructive measurements for the geosciences (Rack 1998). This was followed by a benchmark paper by Ortiz & Rack (1999) on non-invasive sediment monitoring methods reviewing current and future tools for high-resolution climate studies. Since then, development of established and new core analysis technologies and identification of detailed proxies for physical, geochemical and environmental processes have continued apace. In 2001, under the sponsorship of the Japan Agency for Marine Earth Science and Technology (JAMSTEC) and their programme Ocean Drilling in the 21st Century (OD21), a community workshop on 'Advances in Coring, Drilling and Non-Invasive Measurement Technologies for Palaeoceanographic Investigations: A Community Discussion on the State of the Art' was held at ICPVII (7th International Conference on Paleoceanography) in Sapporo, Japan. This workshop, supported by JOI/ODP and the IMAGES programme, was convened specifically to discuss calibration issues, data handling and how non-invasive measurements be best used to develop sedimentological and palaeoceanographic proxies. In September 2003, the community met again at Southampton Oceanography Centre, UK, to review current developments in core logging and imaging. In this paper we review current developments in established and emerging technologies for sediment core research, particularly those in recent years.

The development of core-logging systems During the last three decades, the development of non-destructive core-logging techniques has

7

revolutionized our capability to analyse sediment properties and gain insights of environmental, geochemical and physical processes through characterization of detailed proxies. Previously researchers had to extrapolate using low-resolution data of variable quality, but the development of high-resolution core-logging systems means that scientists today can rapidly acquire a wealth of high-quality, co-located, closely spaced measurements that provide details of downcore variability at centimetre, millimetre and, even, micrometre scales. Such technology makes it possible to discriminate environmental and climatic variability at sub-Milankovitch, millennial, centennial, decadal and, even, subdecadal timescales. At the same time, our ability to visualize sediment in two or three dimensions has increased markedly with the development of new generation digital cameras, high-resolution digital X-radiography, three-dimensional Xradiography (CT scanning), magnetic resonance imaging, confocal laser macro/microscopes and wavelength-specific visualization. Another major advance is the capability to recover cores and preserve cores at in situ pressures, allowing study of subsurface biota and clathrates under real subsurface conditions, providing unprecedented insights into the deep biosphere and methane diagenesis and clathrate stability. Early (post World War II) studies of physical properties of marine sediments were aimed at addressing engineering requirements for designing sea-floor structures and platforms or to support military research or geophysical mapping. Consequently, such studies tended to concentrate on parameters such as index properties and sediment shear strength to understand sediment consolidation and its response to loading, or determining the acoustic properties of sediments and their interaction with sound waves. Bulk sediment properties were routinely studied during the early DSDP legs, but the resulting data were often compromised due to poor sample recovery, drilling-induced core disturbance and poor stratigraphic resolution. However, as early as 1979, Mayer demonstrated in cores from the equatorial Pacific that changes in key physical properties, specifically saturated bulk density and porosity, related to changes in carbonate content and hence to climate variability (Mayer 1979). During the 1980s DSDP/ODP drilling technology improved significantly. The development of the Hydraulic Piston Corer (HPC) and the Advanced Piston Corer (APC) as standard ODP drilling technology (Storms et al. 1983; Deep Sea Drilling Project 1984) improved core recovery and reduced core disturbance considerably, making continuous logging

8

R.G. ROTHWELL & F.R. RACK

a realistic option for ODP cores. The first stratigraphic application of magnetic susceptibility measurements, to determine changes in magnetic mineral concentrations and interpretation of variability in palaeoenvironmental terms, was made by Bloemendal (1983), soon after the introduction of the HPC, on cores from DSDP Site 514 in the SE Atlantic. Since then there has been rapid expansion in this type of investigation, with progress from discrete low field magnetic susceptibility measurements made on samples removed from cores and stored for measurement in small plastic cubes, to continuous palaeomagnetic profiling made directly from the surface of split sediment cores or using long U-channel samples. The possible near real-time availability of detailed continuous physical property logs has allowed detailed stratigraphic correlation of sequentially recovered cores whilst still at sea. On ODP Leg 138, non-invasive sediment logging was used to monitor core recovery and the contained stratigraphy from multiple drill holes to ensure full recovery of the target sequence (Hagelberg et al. 1992, 1995). Intercomparison of G R A P E (Gamma Ray Attenuation Porosity Evaluator), diffuse reflectance and P-wave measurements, obtained through automated logging, allowed all cores in each drill hole to be placed on a new depth scale in their correct stratigraphic position, producing a composite section that allowed drillers to recover potentially missing sequences. This strategy repeated at other ODP sites has produced high-quality data allowing detailed core log integration and the development of an orbitally tuned timescale for much of Tertiary time. Present-day researchers have available a varied suite of technologically advanced, often nondestructive, core-logging and imaging techniques to extract maximum environmental information from marine sediment records (Table 2). Several of new core imaging technologies that have emerged in the last decade (e.g. X-ray computed tomography, magnetic resonance imaging, confocal microscopy) were originally developed for, and have long-standing, medical applications.

Historical development of core-logging systems The development of core-logging can be traced back to the 1940s with the first use of wireline logging in boreholes. Pointecorvo (1941) first used gamma-ray attenuation to estimate sediment density in logging oil wells. Similar methods were applied to soils by Bernhard & Chasek (1955). In

the 1950s and 1960s, the Schlumberger Well Surveying Corporation (now Schlumberger Ltd) pioneered wireline logging of boreholes with the development of their Formation Density Logger, a gamma-ray-based wireline density logging tool. One of the first core-logging systems that logged actual core on an automated track was the GRAPE system developed by Evans (1965). This instrument compared the attenuation of gamma rays through a core section with that through an aluminium standard, to make incremental measurements of bulk density downcore. The absorption and scattering of gamma rays by elements in the sediment is determined by the mass absorption coefficient, which is a function of an element's atomic number to mass ratio (Z/A). Most elements found in deep-sea sediments have a Z/A of c. 0.5, but hydrogen found within sea water has a Z/A of c. 1, allowing porosity to be evaluated. The GRAPE logger was used from DSDP Leg 1 and continues in use today. An automated track for making automated incremental core physical property measurements was permanently installed on the ODP drillship JOIDES Resolution. Over time, other sensors were added to this track. Schultheiss & McPhail (1989) added a means of characterizing compressional soundwave velocity (P-wave) on ODP Leg 108. This provided data useful for seismic and stratigraphic correlation, as well as for constructing synthetic seismograms and useful data for assessing quality of uncut cores. On ODP Leg 115, Robinson (1993) used a loop sensor to measure magnetic susceptibility, a very useful indicator of terrigenous material with applications in correlation, provenance and climatic studies. These new sensors and applications resulted in a multi-sensor logging track that was the precursor to modern systems.

Modern core-logging systems The development of automated core-logging systems over the last four decades has been rapid and such instruments are now in routine use. Most types now make measurements on whole or split sediment cores of specific physical or chemical parameters and the resolution of measurement has moved continually higher, producing rapidly acquired high-quality data. Common types in use include multi-sensor core loggers, acoustic logging systems and elemental scanners.

Multi-sensor core loggers Automated tracks, as used on DSDP/ODP, provided ready and convenient platforms for a

NEW TECHNIQUES IN CORE ANALYSIS variety of non-destructive sensors and logging tools for physical property measurement leading to the commercial development of multi-sensor core loggers. Modern multi-sensor core loggers (Weaver & Schultheiss 1990; Schultheiss & Weaver 1992; Gunn & Best 1998) produce routine, high-quality incremental measurements of gamma-ray attenuation bulk density, P-wave velocity, natural gamma and magnetic susceptibility, and are capable of resolving subtle changes in sediment properties of geological significance. The GEOTEK multi-sensor core logger (MSCL) is now the industry standard with 84 instruments (Fig. 1), operating worldwide by June 2005. The range of parameters that can be measured using the standard logger configuration includes P-wave velocity, wet bulk density (via gamma attenuation method), magnetic susceptibility, electrical resistivity, colour imaging and gamma spectroscopy. Typically, the MSCL can log material at rates of 12 m h -I and at sampling intervals of down to 1 mm. It can analyse either whole or split cores, which for magnetic susceptibility measurements has advantages both ways as point-sensor magnetic susceptibility measurements provide high spatial resolution, while the whole-core magnetic susceptibility loop sensor has a higher signal to noise ratio and so is better for measuring sediments with low magnetic susceptibility (Gunn & Best 1998; Ortiz & Rack 1999). A major improvement in recent years is the replacement of the P-wave transducer mechanism, which previously rose and fell to couple with the sediment after each incremental core movement with a roller mechanism (acoustic rolling contact ARC - transducers). This provides better sediment contact, and improves the consistency of the acoustic coupling between the core liner and the transducer during the logging process. This arrangement improves data quality and consistency, and removes the necessity and inconvenience of using coupling fluids (previously sprayed water) which eliminates data dropouts caused by operator error. Additional sensors currently in development for multisensor loggers include radar scattering (for determining water content), ultraviolet, visible and infrared spectroscopy (for mineralogy), permeameters (for measuring permeability), highfrequency acoustic imaging (for porosity and grain size) and X-ray imaging (for determining sediment structure). A configuration that can log cores vertically (called MSCL-V) has been developed by GEOTEK and is designed for cases where the properties of the sedimentwater interface need to be studied, or for installation in laboratories or ships where available

9

space is limited (Fig. 1). In this system a sensor platform moves, under motor power, either in manual or computer control along the complete vertical length of the core section. The normal logging direction is from top to bottom. The sensor platform normally contains the systems required for measuring P-wave velocity, gamma density and magnetic susceptibility. Another recent innovation is an X Y Z logging system on which several split-core sections can be laid out on a frame and magnetic susceptibility, using a Bartington point sensor, natural gamma and spectrophotometer spectral measurements can be made incrementally and automatically over successive core sections (Fig. 1). This system can be set up and left to run overnight enabling greater core throughput and data collection with time. The collection of large amounts of core in many present-day coring programmes and the necessity to analyse these quickly, often with limited manpower, means that core loggers that can automatically analyse several core sections and run unattended overnight offer great advantages in cost-effective and time-efficient data collection. Such instruments are a major innovation in data acquisition. The data collected by multi-sensor core loggers can be related to sediment character in a number of ways. Velocity~tensity/porosity measurements can be used to infer lithology in many environments as siliceous, calcareous and terrigenous sediments group into specific fields when P-wave velocity is plotted against bulk density (Hamilton 1976; Ortiz & Rack 1999, fig. 5). Weber et al. (1997) and Weber (1998) used acoustic impedance (the product of bulk density and P-wave velocity) as a grain-size estimator in carbonate-free sediments. P-wave data also allow identification of grain-size variation through acoustic properties, being particularly useful in identifying sand and silt interbeds. However, core-logging measurements of physical properties differ considerably from in situ values due to pressure reduction, porosity rebound, and changes in sediment rigidity due to release of overburden pressure and temperature changes during core recovery. Indeed, density variations may relate to changes in grain size and packing rather than changes in lithology, as demonstrated by Rack et al. (1996) and Rack (1997) from ODP sites 909 (Fram Strait, Arctic Ocean) and 980 (Feni Drift, Rockall, NE Atlantic) where peaks in density correlated with increased ice volume, through increased delivery of poorly sorted ice-rafted debris to the seabed. One property now routinely measured using multi-sensor loggers is magnetic susceptibility, which relates to sediment composition. Sediment

10

R.G. R O T H W E L L & F.R. R A C K

(A) MSCL- Standard configuration

(B) Vertical MSCL

(C) XYZ-MSCL

Fig. 1. The Geotek multi-sensor core logger, shown in its current three configurations. The standard configuration (A) can measure sediment bulk density through gamma-ray attenuation, P-wave velocity, magnetic susceptibility and electrical resistivity. The vertical configuration (B) is used where there are space constraints or a need to study properties of the sediment-water interface. The X Y Z multi-sensor core logger (C) is a repository tool that allows automated measurement of a number of sections on unattended runs.

NEW TECHNIQUES IN CORE ANALYSIS

11

Fig. 2. Multi-sensor core logger records (natural gamma, grey-scale reflectance and magnetic susceptibility) of the last 150 ka (MIS 1-6) of archived half-core from DSDP Site 594. The site is located at latitude 45 ~ SW Pacific Ocean, east of the South Island of New Zealand. The site lies just south of the Subtropical Front, at a water depth of 1000 m beneath north-flowing Antarctic Intermediate Water. During glacial periods, a mountain ice cap developed on the Southern Alps and melt waters delivered abundant terrigenous mud to Site 594. As first described by Nelson et al. (1985), alternations in the core between biopelagic carbonate and terrigenous mud therefore record a striking climate cyclicity. Geophysical scanning, especially the natural gamma record (cf. Carter & Gammon 2004), confirms the macroscopic lithological pattern but provides a far higher resolution climate record, at centennial scale for the datasets shown and at decadal scale for the colour imagery (not shown). The core scans were performed at the West Coast Core Repository at Scripps Institute of Oceanography, USA, using a GEOTEK M S C L - X Y Z system, which, after loading, is capable of unsupervised collection of data from a full 10m-long DSDP or ODP core.

containing abundant ferro- or paramagnetic minerals (e.g. magnetite, pyrrhotite, hematite, olivine, biotite, pyrite, and iron-oxide-stained rock and mineral fragments) show high magnetic susceptibility, whilst biogenic material, such as calcite and silica, and quartz and feldspar show low or even negative magnetic susceptibility values (Robinson 1993). Relative changes in magnetic susceptibility and absolute values can thus be important parameters relating to sediment provenance, palaeoclimate, bottom-water flow conditions and regional stratigraphy (Fig. 2). Chi & Mienert (1996) and other authors have demonstrated the value of G R A P E density and magnetic susceptibility in identifying Heinrich Events in marine sediments, and the importance of these parameters in reconstructing climatic changes and related events in the Late Pleistocene.

Most stratigraphic applications of MSCL logs rely on the relative values of logged parameters for detecting geological events. However, absolute MSCL parameters are also of great potential value. Measurement of sediment bulk density and P-wave velocity downcore using multi-sensor core loggers allows acoustic impedance to be calculated from the product of velocity and density. These data are important for quantifying sound propagation through the sea floor and reverberation modelling (Best & Gunn 1999). However, there are difficulties in using absolute parameter values measured using MSCL logging. Gerland & Villinger (1995) and Weber et al. (1997) show that gamma density MSCL measurements must be calibrated to give meaningful results as they are susceptible to significant instrument errors. Best & Gunn (1999) proposed using a short length

12

R.G. ROTHWELL & F.R. RACK

(c. 24cm) stepped aluminium core placed in a water-filled sealed section of standard piston core liner as a standard calibration tool for the MSCL to simply quantify systematic errors during MSCL logging and intercalibrating MSCLs at different laboratories. Aluminium is chosen as its density (2.70 g cm -3) is not too dissimilar to quartz ( 2 . 6 5 g c m -j ) and calcite (2.71 gcm-3), both common minerals in marine sediments. The steps in the block simulate different porosity sediments.

Digital imaging Sediment cores can be difficult to photograph, particularly if the core surface is wet and/or uneven. After splitting cores can oxidize and change colour, losing important information on their original character. Even if cores are photographed immediately after splitting using traditional film-based cameras, the emulsion dyes may change within a few years of exposure, compromising accurate recording for archival purposes (Merrill & Beck 1995). However, digital images preserve accurate colour data indefinitely, allowing quantitative study of what is often an ephemeral physical property long after the cores were collected. The wide availability of affordable digital cameras and application of digital imaging to traditional film-based techniques (visual spectrum photography, X-radiography) means that images can now be rapidly acquired, stored and distributed electronically, and analysed using universal data-processing/ data-mining tools. The visible spectrum of light (400-700 nm) is used for visual core description where sedimentary structures, bioturbation, layer thickness, textures and fractures are recorded. It is also used for colorimetry, where colour values are quantitatively recorded (e.g. colour space, x, y, Y; L*a*b*, Munsell notation) by spectrophotometers that are either hand-held or integrated into an automated logging tool. Traditional film-based systems (particularly transparencies) have high sharpness and can record and reproduce a wide range of colours. The relative sensitivities (speeds) of the photosensitive layers are balanced, such that properly colour-balanced images result when scenes illuminated by the reference scene illuminant are photographed. With digital imaging, the image is captured using a digital camera, the signal processed (typically on a computer workstation), and the image formed and displayed on a video monitor. The spectral power distribution of a colour stimulus is generally a product of the spectral power distribution of a light source

and the spectral reflectance of an object (as a function of wavelength). Several digital imaging systems are now in use. The Geotek GEOSCAN III calibrated colour core imaging system collects digital images using a line-scan camera linked to the MSCL core conveyor stepper motor to generate synchronous output of image data. These resulting image data can then be integrated with other MSCL data, making it possible to carry out a wide range of statistical and multivariate analyses. Individual interference filters in front of each of the three CCD (chargecoupled device) line arrays within the camera body, coupled with appropriate calibration protocols, ensures that colour from each CCD array is calibrated and non-overlapping. Highresolution images are created that can be analysed in terms of three colour components, irrespective of the time, place or the MSCL system used. In this way, image data from institutions around the world can be compared and combined without the need for reference to Munsell colour charts and ensures laboratory intercalibration. During ODP Leg 207, the PalaeoceneEocene and Cretaceous-Tertiary boundary were drilled on the Demerara Rise, western Altantic (Erbacher et al. 2004). Both these important boundaries have dramatic lithological expression at the sites drilled (Fig. 3). The Palaeocene-Eocene (P-E) boundary occurs at the base of an approximately 30 cm-thick greenish clay bed that is an unique lithology at each site where it is present. The clay may record the shoaling of the calcium carbonate compensation depth following the P-E boundary event and is thought to coincide with the benthic foraminifera extinction event and a carbon isotopic excursion

Fig. 3. (A) Close-up photographs of the PalaeoceneEocene (P-E) boundary drilled at five drill sites on ODP Leg 207. The boundary occurs at the base of an approximately 30 cm-thick greenish clay bed that is a unique lithology at each drill site. The clay layer is thought to record the shoaling of the CCD following the P-E transition and is thought to coincide with a benthic foraminifera extinction event and a carbon isotope excursion. (B) Core photographs of the Cretaceous-Tertiary (K-T) boundary as drilled at three drill sites on ODP Leg 207. The boundary is marked by an ejecta layer, just under 2 cm thick, composed of normally graded green spherules, interpreted as ejecta from the K-T Yucatan bolide impact. Both the P-E and the K-T boundaries can be identified and correlated at the drill sites using core images due to their consistent lithological character. (C) Map showing the locations of the Leg 207 drill sites on the Demerara Rise, offshore of Suriname. Water depths are given in metres. From Shipboard Scientific Party (2004 in Erbacher et al. 2004).

NEW TECHNIQUES IN CORE ANALYSIS

~E .r-

.v.tt~

~E .r-

CO

t~E t'M v - CO t~

O

toe

t'kl r-. 9 to

O

tt'Mo e

to

.,1..,.

130

t~E

cO

t~E

.tO')

13

14

R.G. ROTHWELL & F.R. RACK

Fig. 4. The Cretaceous-Tertiary (Late Maastrichtian-Danian) boundary at Site 1260 drilled on ODP Leg 207 with interpretation. Cretaceous and Palaeogene planktonic foraminiferal zones are also shown for drill sites 1257-1261, across the K-T boundary. From Shipboard Scientific Party (2004 in Erbacher et al. 2004).

(Erbacher et al. 2004). The Cretaceous-Tertiary boundary (K-T) also shows a remarkably consistent character at the three sites where it was drilled (Fig. 3). The boundary is marked by an ejecta layer, just under 2 cm thick, composed of normally graded green spherules, interpreted as ejecta material from the K - T Yucatan bolide impact (Fig. 4). Both the P-E and K - T boundaries can be correlated at the ODP Leg 207 drill sites using core images due to their consistent lithological character. Colour has been shown to be a highly diagnostic property of marine sediments. Lightness is a robust indicator of certain sediment components such as carbonate, iron-bearing minerals (e.g. goethite, hematite, pyrite), free iron and clay. Colour can also give information on the oxidation state of iron (e.g. Mix et al. 1995; Giosan et al. 2002). Rogerson et al. (2006) demonstrate the potential of colour logging in developing an initial chronostratigraphic model for a high-resolution record from the late Quaternary using a core from the western Gulf of Cadiz (SW Spain). Nederbragt et al. (2006) discuss methods for extracting calibrated colour values from digital images of sediment core surfaces, which can be correlated with geochemical composition. In the carbonatepoor laminated sediments studied, total organic

carbon had the dominant effect on colour (Nederbragt et al. 2006). A recent development in imaging sediment cores is the application of infrared (IR) thermal imaging. On ODP Leg 201, a Thermacam thermal imaging IR camera was used to image methane hydrate on immediately recovered core from the Peru margin (D'Hondt et al. 2003; Ford et al. 2003). This was the first time IR cameras were used to identify gas hydrate prior to its dissociation, which usually occurs rapidly due to pressure release and temperature increase. Hydrate dissociation is an endothermic process, creating a negative temperature anomaly (cold anomaly, on average 4 ~ cooler) that can be rapidly imaged using an IR camera (Ford et al. 2003). Hydrate volume could be estimated from the processed images. In addition, voids in the sediment could be detected as warm anomalies. Another development of IR imaging also explored on ODP Leg 201 was its possible use in lithological characterization of ambient-temperature cores, due to slight variation in thermal emission properties reflecting differences in sediment composition or water content. Dissociation of the hydrate resulting in the creation of gas expansion voids caused increased variability in resistivity, P-wave velocity and natural gammaray emission logs. Although the data produced

NEW TECHNIQUES IN CORE ANALYSIS

Fig. 5. The relationship between a thermal anomaly in a core containing clathrate, from the Cascadia continental margin, offshore Oregon, drilled during ODP Leg 204, and chloride concentration anomalies. From Shipboard Scientific Party (2003 in Tr~hu et al. 2003).

are essentially qualitative, careful monitoring of core-handling times and ambient temperature, integration of the IR camera with a logging track, standardization of image analysis software and quantitative treatment of emissivity differences between different sediment types would make a more quantitative method (Ford et al. 2003). During ODP Leg 204, extensive use was made of IR cameras immediately after core retrieval to determine the distribution and texture of gas hydrates recovered from the Cascadia continental margin, offshore Oregon (Tr~hu et al. 2003). Thermal anomalies recorded by the IR imaging camera provided a robust record of gas hydrate distribution that could be calibrated using estimates of in situ gas hydrate concentration determined from pressure core samples and chloride concentration anomalies (Fig. 5). Infrared imaging of hydrate-bearing core seems a valuable potential tool for gas hydrate identification and quantification.

Non-imaging optical systems A great deal of useful scientific information can also be gathered from split sediment cores using non-imaging optical systems. These include grey reflectance measurements and diffuse spectral reflectance analysis. Grey reflectance analysis reduces information from a black and white core photograph to a single downcore profile with relative intensities ranging from 0 (pure white)

15

to 255 (black). Grey reflectance changes in deep-sea sediments are commonly related to variations in the calcium carbonate/detrital ratio, which in Plio-Pleistocene pelagic sequences are often broadly related to productivity variations during glacial-interglacial climatic cycles. Such measurements have been calibrated by oxygen isotope analysis and micropalaeontological studies, providing a useful, rapidly acquired, chronological tool (Broecker et al. 1990; Bond et al. 1992) that can be acquired using relatively simple instrumentation. Greyscale (lightness) has been successfully used for developing preliminary stratigraphies on long sediment cores, particularly those collected during the Ocean Drilling Program (e.g. Ortiz et al. 1999a, b; Grutzner et al. 2002) (Fig. 6). Sediment colour is a major indicator of composition and detailed quantitative colour data, showing character and variability beyond that assessable with the naked eye, can be collected by modern spectroscopy instruments. These include commercially available spectrophotometers, such as those produced by Minolta (the CM-2002 and CM-2022 models are the most commonly used in core analysis). These compact hand-held instruments combine measurement, data-processing and display functions in a single unit, and are ideally suited for laboratory and shipboard use. The data can be displayed in a variety of ways, graphically as spectral reflectance or colour difference, numerical absolute and/or difference values for standard defined colour spaces, or as Munsell notation. When attached to jigs or tracks to allow incremental measurement downcore, colour variation related to climate variability on scales as small as l cm have been documented (e.g. Chapman & Shackleton 1998). The Geotek XYZ-MSCL, where a spectrometer is a standard sensor, allows automated collection of such data (Fig. 1). Reflectance spectroscopy has been widely used in mineralogical and geochemical studies (e.g. Hunt 1977; Gaffey 1986) and has been applied to deep-sea sediments by Chester & Elderfield (1966), Barranco et al. (1989) and others. Oregon State University (OSU) have developed an automated reflectance spectroscopy logger to measure diffuse reflectance spectra in the ultraviolet, visible and near-infrared bands on split sediment cores (Mix et al. 1992, 1995). Measurement of wet sediments and splitcore surface roughness will affect reflectance, but water effects can be minimized by careful selection of the wavelengths studied and scraping with glass slides will mitigate roughness effects (Mix et al. 1992). Use of the OSU reflectance

16

R.G. R O T H W E L L & F.R. R A C K

NEW TECHNIQUES IN CORE ANALYSIS spectroscopy logger on ODP Leg 138 proved the technique as a tool for rapid, non-intrusive characterization of major lithologies and sediment oxidation state (Mix et al. 1992, 1995). A second-generation logger, examining a wider frequency band and with an improved signalto-noise ratio, was used on some subsequent ODP legs (e.g. Shipboard Scientific Party 1997; Ortiz et al. 1999a, b). Visible light spectroscopy, through providing quantitative measurement of optical lightness, has been used for determining carbonate content of marine sediments (Balsam et al. 1999), and spectral data have been used to estimate concentrations of opal, organic carbon, chlorite and hematite (Balsam & Deaton 1991, 1996; Deaton & Balsam 1991). Sediment mineralogy based on visible and nearIR reflectance spectroscopy is discussed by Jarrard & Vanden Berg (2006). X-ray fluorescence core scanners

A major innovation in the last decade has been the development of X-ray fluorescence (XRF) core scanners allowing rapid collection of highresolution continuous downcore records of element distributions. In XRF analysis, sediment is excited by incident X-radiation that causes electrons to be ejected from inner atomic shells. The resulting vacancies are subsequently filled by electrons falling back from the outer shells, and the surplus energy (i.e. the energy difference between the vacant inner and the outer shells) is emitted as a pulse of X-radiation (Jenkins & De Vries 1970). Atoms of specific elements will emit characteristic energy and wavelength spectra allowing recognition and estimation of element abundance. Elemental variations determined by XRF core scanners can be used for inferring environmental changes, diagenetic processes and pollutant inputs, and can assist in sediment correlation and process studies. A number of XRF core scanners have now been developed. The Netherlands Institute of Sea Research (NIOZ), Texel, The Netherlands, developed the Fig. 6. Lithostratigraphic composite for ODP Site 1262 (Walvis Ridge, SE Atlantic) showing how the variation in properties measured by non-destructive core-logging can be used to distinguish lithostratigraphic units, in this case, magnetic susceptibility (MS), natural gamma radiation (NGR), carbonate content and lightness (L*). Note how lightness (together with magnetic susceptibility and natural gamma) clearly distinguishes the boundary between lithostratigraphic units IIIA and IIIB and the Cretaceous-Tertiary boundary. From Shipboard Scientific Party (2004 in Zachos et al. 2004).

17

first scanner of this type in 1988 (Jansen et al. 1998). This core scanner made measurements on split sediment cores and was containerized to allow sea-going use. A modified version of this XRF core scanner was installed at the University of Bremen in 1997, and a second-generation core scanner, the AVATECH XRF core scanner, has been operational at both NIOZ and Bremen since 2002 (Richter et al. 2006). R6ntgenanalytik Messtechnik GmbH, of Taunusstein, Germany, have also produced a X R F core scanner, known as the Eagle I,tProbe (see Haschke 2006). In this instrument the incident X-rays are focused by a polycapillary lens to irradiate a very small area of the core allowing very high-resolution measurement with typical spot sizes in the range 30-50 p~m. Another instrument is the ITRAX core scanner, manufactured by Cox Analytical Systems of Gothenberg, Sweden (see Croudace et al. 2006), which also uses capillary optics in the form of a flat glass waveguide to focus the incident X-radiation in order to achieve a very small step size (down to 100~tm). In addition to element profiling, the ITRAX core scanner can also acquire very highresolution continuous digital X-radiographic images through the centre of the split cores. Conventional laboratory XRF analysis requires samples to be homogeneous, dry and have a flat smooth surface. These requirements cannot be met when analysing split sediment cores. Cut core surfaces, even when very carefully cut, will have topography and variation in grain size will cause surface roughness. Sediment compositional variability, water content, and textural and porosity changes all mean that XRF logging using core scanners will be semi-quantitative at best. Light element (Z<Si) measurement is often compromised by attenuation of excited Xrays in the air gap between the X-ray detector and the core surface. However, XRF scanning of marine sediment cores has allowed discrimination of even decadal, annual and subannual environmental changes. For example, Haug et al. (2003) used the Eagle laProbe (see Haschke 2006) to measure Ti in sediments from the Cariaco Basin, offshore Venezuela, at subannual resolution. Titanium was used as proxy for terrigenous sediment delivery to the basin from surrounding watersheds, hence providing an index of hinterland rainfall, with lower Ti reflecting lower precipitation resulting in lower river runoff. The resulting data identified three periods of drought, each lasting for a decade or less, coincident in time with the three phases of Maya city abandonment around 810, 860 and 910AD previously identified on archaeological evidence. Whatever social stresses were responsible for the

18

R.G. ROTHWELL & F.R. RACK

collapse of Mayan civilization, repeated prolonged regional droughts seem implicated as a probable underlying cause. Within an oceanic setting, much research using XRF core scanning has concentrated on palaeoceanographic studies (e.g. Jahn et al. 2003; Kuhlmann et al. 2004; Lamy et al. 2004; and others), although the technique also has value in lithostratigraphic analysis and provenance studies of allochthonous beds (Rothwell et al. 2006) and studies of diagenetic processes (Thomson et al. 2006). In the North Atlantic, the Ca/Fe ratio offers a good proxy for discriminating glacialinterglacial cycles as pelagic sediments deposited during interglacials have higher carbonate contents than those deposited during glacials (Balsam & McCoy 1987). Hence, downcore major element XRF records can rapidly provide preliminary stratigraphies by showing the relative abundance of biogenic carbonate and clay, respectively (Richter et al. 2006). Interelement relationships can also be useful for discriminating multiple sources for some elements if present. For example, the Sr/Ca ratio can show the relative contribution of aragonite to particular sediments, as aragonite generally has high Sr and calcite low Sr (Thomson et al. 2004; Richter et al. 2006). The Si/A1 ratios can show the relative contributions of terrigenous-derived Si compared to biogenic opal. When A1 data are poor, such as where it is at the limit of detection, then Ti or Rb can be used as suitable detrital divisors (Rothwell et al. 2006).

X-ray computed tomography]digital X-ray imaging The opacity of sediments hides their internal structure, processes that have occurred or may be occurring within them, and their response to stresses and other natural processes. This limits our understanding of important natural processes, such as sediment deposition, post-formational geochemical and physical alteration, bioturbation, dewatering and consolidation, erosion, in-sediment gas-bubble formation and migration, stress and failure, and many other phenomena. X-ray computer tomography (3D X-radiography or CT scanning) provides multidimensional real-time imaging of the structure of solid objects and can produce digital images (Fig. 7). Outside of its well-known medical uses, it is increasingly being used as a research tool for dendrochronology, archaeology and oceanography. Since the dataset is three dimensional, images can be visualized for any plane. Resolutions down to 1001am are now readily

available and promise to provide unparalleled insights into core structure, physical properties and, hence, sedimentary processes. Commercially available models specifically made for scanning rock or sediment cores have become increasingly affordable in recent years. CT scanners have been used by the oil industry since 1980, and CT scanning is now standard procedure in North America in the analysis of borehole samples, and rock and sediment cores, with a number of institutions having CT scanners specifically for geological research. A substantial body of literature has now been published resulting from CT scan analyses of sediments and rocks. These cover diverse applications including fabric studies, granulometry, porosity and permeability studies of source and reservoir rocks; structure and density of ice and cohesive muds; mineral assemblages in rocks, sediments and placer deposits; hydrocarbon contamination in soils; fracture and fissure patterns in rocks and consolidated sediments; composition of deepsea polymetallic nodules; characterization of slump and mass failure deposits; and pollutant migration in cores and characterization of methane hydrate in marine sediments. Until recently, most CT work on sediment cores has been performed using medical-type CT scanners, which were originally designed for imaging the human body and not optimized for core characterization. Such CT scanners are expensive, are generally non-portable (although tractor-trailer mounted units exist), and have high maintenance and installation costs (having large footprints and requiring lead-lined rooms). Recently, however, smaller specifically petrophysical CT scanners have been developed, some of which are self-shielded; negating the necessity of specially constructed lead-lined rooms. Freifeld et al. (2006) have designed and built a portable CT imaging system specifically for imaging whole-round cores at the drilling site. In this instrument core is loaded vertically and rotated about its vertical axis, and the X-ray source and detector move within a shielded horizontal gantry over the core. Careful design of the radiological shielding minimizes the size and weight of the instrument, resulting in a much smaller footprint than conventional systems and complete portability. Power consumption is low. The recent availability of smaller self-shielded CT scanners, specifically designed for imaging core, with their lower capital and maintenance costs compared to conventional medical-type scanners, makes this technology more affordable than ever before. As a result such systems are likely to come into routine use to provide extremely fine-scale density analysis.

NEW TECHNIQUES IN CORE ANALYSIS

19

Fig. 7. X-ray CT images of a pressure-sealed gassy sediment core taken from Dibden Bay, Southampton Water, UK. The top three pictures image, in three dimensions, gas bubbles and worm tubes by bandpass filtering grey levels. The bottom three images are 2D image slices through the core and show crack-like gas bubbles in mud, subspherical bubbles in sand and gas trapped in gastropod shells. Gas, probably methane, appears as black, calcium carbonate as white, and other minerals as intermediate grey levels. These images show the wealth of detail that can be revealed by both 2- and 3D X-radiography. (Reproduced courtesy of Dr A. Best, National Oceanography Centre, Southampton, partially adapted from Best et al. 2004.)

Nuclear magnetic resonance/magnetic resonance imaging (NMR/MRI) Nuclear magnetic resonance (NMR), also known as magnetic resonance imaging (MRI), is another investigative technique borrowed from medical imaging technology. N M R is increasingly being used to investigate physical properties, particularly porosity and pore-size distribution, of cored sediments. Within the hydrocarbon industry, N M R is now widely used to characterize porosity of source rocks and, hence, evaluate the productive potential of hydrocarbon reservoirs (Kleinberg 1996, 2006). Magnetic resonance imaging most frequently relies on the relaxation properties of excited hydrogen nuclei in water or inter-pore fluids. When the object to be imaged is placed in a powerful, uniform magnetic field, the spins of the atomic nuclei with non-zero spin numbers within the material all align in one of two opposite directions: parallel to the magnetic field or

anti-parallel. MRI systems generate images of nuclear spin density or magnetic resonance relaxation times (i.e. the return of spinning nuclei to a lower energy state). The SPRITE (Single Point Ramped Imaging with T1 Enhancement) MRI technique (Balcom et al. 1996; Chert et al. 2006) can produce images related to porosity, pore-size distribution, internal sediment structures and heterogeneity. Such data are essential for evaluating fluid flow characteristics in reservoir rocks. Kleinberg (2006) demonstrates how the technique can be used to quantify frozen and unfrozen phases in interparticle spaces and to understand the growth habit of ice and methane hydrate (clathrate) within sediment.

Confocal macro/microscopy Confocal microscopy is an imaging technique that uses a laser as the light source that is tightly focused through a slit or pin-hole aperture, limiting the depth of field to a single plane. By varying

20

R.G. ROTHWELL & F.R. RACK

the focal point, multiple images of different surface planes can be combined to produce a sharply focused, three-dimensional (3D) computer image of the object. Filters can be used to modify both the incident and reflected laser light, and these can be used to discriminate compositional data through reflectance or fluorescence spectroscopy. Although this technology was developed for medical application, a system specifically adapted for imaging rock and sediment has been produced (the confocal scanning laser Macroscope/Microscope, abbreviated cslM/m; Dixon et al. 1995). This instrument offers a range of fields of view and magnifications and can image samples ranging in size from 25 x 25 gm up to 7.5 x 7.5 cm in under 10 s using reflected light or photoluminescence as contrast mechanisms (Ribes et al. 2006). With this instrument, the sample can be rapidly surveyed in macroscope mode providing a photoluminesence overview image and areas of interest identified which can then be viewed in microscope mode to provide ultra-high-resolution images, resolving submillimetric features within the sediment. This instrument offers great potential for rapidly imaging core sections in low-resolution macroscope mode (similar to taking traditional core photographs) with the option of using the high-resolution microscope for sediment microfossil and microfabric examination, and microscale mineralogical analysis (Ribes et al. 2006). Indeed, an imaging survey of ODP cores has shown that the instrument can be used in fluorescence response mode to discriminate between different sediment types and to describe the distribution, and discriminate the relative percentage, of microfossils in sediments (Ribes et al. 2006).

Chemical fossils (biomarkers) and molecular stratigraphy During the last two decades discovery of biologically derived chemical markers (biomarkers) within marine sediment cores has led to remarkable advances in understanding past environmental changes. Biomarkers are organic compounds of known biological origin (primarily lipids, alkenones, and, for higher plants, n-alkanes, n-alkanols, n-alkanoic acids and wax esters) commonly preserved in marine sediments that can be used to develop proxy records of environmental change. Both marine and terrestrial plants biosynthesize molecules that are in some cases highly source-specific. On death of the source organism, some of these molecules, which are resistant to degradation, are transferred to sediment sinks and preserved as chemical fossils.

Currently, the most well-defined and well-studied lipid biomarkers are a series of long-chain unsaturated ethyl and methyl ketones, known as alkenones (see Herbert 2004 for a review). These are principally produced in the modern ocean by two species of widely distributed coccolithophorid algae, Emiliana huxleyi (which can make up 6080% of coccolithophore assemblages) and Gephyrocapsa oceanica. The function of alkenones in the source cell is poorly understood but may regulate membrane fluidity at different temperatures; however, they are clearly a physiological response to growth temperature. The overall degree of unsaturation in alkenones synthesized by these coccolithophores varies inversely with water temperature (Brassell et al. 1986). Alkenones are resistant to degradation within sediments and can survive for long temporal spans (several tens of Ma). Further, alkenones survive during long-term core storage (Sikes et al. 1991). Alkenones are usually measured by gas chromatography coupled with a suitable detector, typically a flame ionization detector (GC-FID) or mass spectrometer (GC-MS). Variation in the proportion of di-, tri- and tetraunsaturated alkenones in sediments is quantified as an alkek' none unsaturation index, termed u3k7 or U37 (Brassell et al. 1986). Culture experiments have shown that there is a linear relation between uk7 and growth temperature; and analysis of core tops and comparison with sea-surface temperatures (SST) in many parts of the world ocean has resulted in a universal and relatively robust calibration at least for typical open ocean regimes (Mfiller et al. 1998). However, analysis of alkenones may give temperature values for the water depth at which the source organisms lived rather than true SST. Even so, alkenone thermometry has become an important and widely used tool by palaeoceanographers and SST records have now been constructed for temporal spans from years to hundreds of kiloyears (ka). Recent work by Cacho et al. (1999, 2002) demonstrates how alkenone records can provide astonishingly detailed records of the pervasive nature of millennial events within the late Quaternary regional climate of Europe and the North Atlantic. Western Mediterranean Sea alkenone records show all the millennial-scale features seen in Greenland ice cores over the last 50 ka as conspicuous SST changes, with some changing as quickly as 6~ per century (Cacho et al. 2002). Biomarkers from higher plants and terrestrial sources also occur in marine sediments, particularly in areas with high fluvial input, or areas of open ocean where aeolian terrestrial inputs can be greater than pelagic ones (Pagani et al. 2000).

NEW TECHNIQUES IN CORE ANALYSIS These biomarkers can give insights to hinterland vegetation from which palaeoclimatic changes can be inferred (Pancost & Boot 2004). Indeed, the often large hinterland areas that contribute sediments to offshore basins, the high sampling resolution possible and the long temporal records often available means that marine sediments can provide a better assessment of continental climate change than comparable terrestrial sites, such as peats and lake sediments (Pancost & Boot 2004). Measurement of carbon isotopic composition of sedimentary organic carbon can also distinguish that derived from different, yet isotopically definable, sources - marine plankton and vascular plants (Degens 1969).

Recovery and analysis of core at in situ pressure In the last two decades two particular aspects of the subsea-floor sedimentary environment have been recognized as being of global importance. The first is the widespread occurrence of clathrates (gas hydrates) in marine sediments in some areas of the world. Gas hydrates occur both in Arctic regions affected by permafrost and in marine subsurface sediments in the polar regions (shallow water) and in continental slope sediments (deep water), where pressure and temperature conditions combine to make them stable (Kvenvolden 1993, 1998). Gas hydrate is a crystalline solid with a simple structure, consisting of gas molecules (usually methane) each of which is surrounded by a 'cage' of water molecules (Sloan 1990). In appearance, gas hydrate looks very much like water ice (Fig. 5). Methane hydrate is stable in ocean-floor sediments at water depths greater than 300m and, where it occurs, it is known to cement loose sediments in a surface layer of up to several hundred metres thick. Gas hydrates within sediments store immense quantities of concentrated methane, with major implications for potential future energy resources and climate. Release of methane from the clathrate reservoir, perhaps due to sea-level change-related sediment unloading, may contribute to significant global warming and, hence, may control longterm climate change (Kvenvolden 1988; Paull et al. 1991; Harvey & Huang 1995; Kennett et al. 2003; and others). Within subsea-floor sediments, gas hydrate occurs within interparticle pores and cements sediment grains. It therefore can have a significant effect on sediment strength and its formation and breakdown may influence the occurrence and location of submarine landslides (e.g. Kayen & Lee 1993). Such landslides may release methane into the atmosphere, affecting

21

global climate. When drilled, pressure release results in rapid dissociation of gas hydrate resulting in a frothy sediment of mousse-like appearance. As a result the study of gas hydrates has been hampered by lack of a way of recovering and preserving hydrate-bearing core at in situ pressure. The second feature of subsea-floor sedimentary environments to be recognized, of critical importance in the past two decades, is the deep biosphere (Parkes et al. 1994, 2000; Schippers et al. 2005). Living and diverse bacteria adapted to high-pressure, often high-temperature, regimes exist within sediments hundreds of metres below the sea floor, to depths of at least 850m below the seabed. The extent and importance of this remarkable microbial habitat, which may account for perhaps 10% of total global biomass, has only been recently recognized. Indeed, Whitman et al. (1998) estimated that deep marine sediments may contain over 60% of global bacterial biomass, making the subsea floor the largest bacterial habitat on Earth. The deep biosphere may play a major role in the global cycling of elements and form a significant reservoir of organic carbon. These obligate barophilic bacteria may not survive depressurization during the coring process. Yet, their diversity and remarkable adaptation to a highly extreme environment makes deep biosphere microbes of great biotechnological potential if means to recover and culture them are developed. Recently, the European-funded HYACE (HYdrate Autoclave Coring Equipment) project, and its successor HYACINTH (Deployment of HYACe tools In New Tests on Hydrates), have developed a pressure corer designed to recover cores in liners at in situ pressure and transfer them to pressure vessels (Schultheiss et al. 2006). As a result, subsamples can now be taken from cores and transferred to chambers, all at in situ pressures, where intrusive measurements and experiments can be made. The HYACINTH system has already been used on ODP Leg 204, allowing core containing gas hydrate to be logged at in situ pressure in the laboratory (Tr6hu et al. 2003; Schultheiss et al. 2006). Further, during the HYACINTH project, new subsampling equipment was developed to allow further biological, chemical, geophysical and geotechnical investigations at in situ pressures in the future.

Developing global accessibility to sea-floor samples and core data The huge numbers of sediment cores now acquired globally and the immense corpus of

22

R.G. ROTHWELL & F.R. RACK

data, much of it high resolution, collected from these cores necessitates a means of making cores and the data from them accessible to the wider scientific community. After cores have been analysed for the purpose for which they were taken, cores and sea-floor samples are normally stored under controlled conditions to ensure optimum preservation for further use. As new measurement techniques and instrumentation become available, and new concepts evolve, existing cores can be re-sampled to add to the knowledge base. The financial and scientific investment in collecting and storing cores is considerable, and stored cores provide a legacy of continuing scientific usefulness and importance. Secondary usage of cores and core-derived data, after the primary data collection requirement has been served, enables cost-effective use and maximizes the scientific return on the cost of core collection. The problem of providing accessibility to large sediment sample archives has been long recognized in the USA. As early as 1977, several oceanographic research institutions got together in a collaborative effort to help researchers locate marine sediment and rock material for further analysis. Today, 20 oceanographic institutions (mainly US-based, but including repositories in the UK, Canada and Germany) make up this effort, submitting core data to the 'Index of Marine and Lacustrine Geological Samples' at the US National Geophysical Data Centre (NGDC)/World Data Centre A for Marine Geology and Geophysics, at Boulder, CO (see Moore & Habermann 2006). This database currently holds information on nearly 157 000 sea-floor cores, grabs, dredges and drill samples, and can be searched through the Internet. The database is searchable by any parameter or combination of parameters. Inventories, data listings, data in digital form and plots by station/ lithology/texture are available. Samples are normally available for further scientific study, on request, from the participating institution. The importance of the 'Index of Marine and Lacustrine Geological Samples' and the service its provides to the international marine science community was recognized by a resolution of the International Oceanographic Commission (IOC) Committee on International Data & Information Exchange (IODE) passed by the IOC in 1994, which states: The IOC Committee on International Oceanographic Data and Information Exchange, recognizing the importance of analyses deriving from ocean sediment cores to studies of past climates and to palaeoceanography, being concerned with the diminishing amount of sample material and with the difficulty in locating material available for analysis ....

Noting the need to identify, catalogue and curate all such remaining material so these materials can be fully utilized for analyses beyond those for which the samples were collected originally .... Encourages Member States to locate and catalog marine sediment cores available for sampling and analysis and contribute information (meta-data) about these cores to the Index to Marine Geological Samples database maintained by W D C - A - M G G ; .... Urges Member States to establish procedures to provide access to these cores for sampling. (International Oceanographic Commission (IOC) Committee on International Data & Information Exchange (IODE) Resolution IODE-XIV.2, 1994: published on the Internet at http://www.ngdc.noaa. gov/mgg/curator/ioc_resolution.HTML.)

NGDC is developing the Index to create common cross-reference to additional data held in its archives. These include a full suite of DSDP core data, around 70 000 pages of scanned cruise reports, more than 50 000 core photographs, 7500 core X-radiographs, 20 000 sea-floor photographs, 14000 pages of paper documents, and 5000 pages of core logs and text descriptions (Moore & Habermann 2006). Thus, the Index is developing into a valuable comprehensive data resource for the global marine and earth science community. Planned importation of the Index into centre-wide geospatial database systems will allow implementation of standard quality assessment tools, as well as their integration into desktop and Internet mapping applications. This standards-based interoperable agenda promotes global use of the archived data and ensures its continuing value to the worldwide scientific community (Moore & Habermann 2006). Representatives from institutions that submit metadata and data to the Index meet about every 2 years to hear facility presentations, discuss common issues of interest, such as core-based research projects and facility information systems, and hold round-table discussions on information needs and strategies for co-operation. The 'Index' therefore also provides a valuable forum for idea exchange and discussions on core curation and analysis (Mix et al. 2003). The long-term funding and commitment to develop the Index and its robust data service provision, with its inherent networking amongst curators and associated community, provides an excellent model for data infrastructure for Europe and other countries. Within the ODP, and successor IODP projects, an online relational data management system called JANUS has been implemented. The database includes palaeontological, lithostratigraphic, chemical, physical, sedimentological and geophysical data for sediments and rocks collected during these programmes. The central JANUS database consists of more than 450 oracle tables divided into 25 subject areas that can be accessed,

NEW TECHNIQUES IN CORE ANALYSIS queried and visualized through a number of different tools (Mithal & Becker 2006). Data can be extracted using commercial applications such as SQL*Plus and Microsoft Access. Core metadata and core-derived data can be stored, as in the Index or JANUS, and made accessible through the Internet worldwide. However, an important development is the ability to mine existing datasets in order to effectively integrate diverse core-related and other sea-floor data collected over decades in order to provide ocean-bottom information at many spatial scales. A major innovation in this respect is the development of dbSEABED - an information-processing system for marine substrates originally developed at the Ocean Science Institute, University of Sydney, and used to provide thematic sea-floor maps of the Australian Maritime Region based on data collected from over 275000 attributed sample sites. DbSEABED is a unique, versatile and detailed information-processing system. It works through data mining, with modules devoted to extraction of diverse geological attributes, linguistic parsing of geological descriptions (e.g. core logs, photographs, descriptions, etc.) and on calculation of some parameters (see Jenkins et al. 2006). In order to preserve spatial resolution of original sampling campaigns, it deals in point data, although polygon and poly-line data types are also used. The structure is such that new attributes and new datasets are readily added. DbSEABED is now operated by a number of co-operating institutions in the USA and Australia, who import, integrate, process and display regional datasets under a common format and software set-up. They share data, code and innovations. The dbSEABED database structure currently holds in excess of 1 x 106 attributed sea-floor describing sites worldwide with over 400 000 of these within US waters. Processing of the data from US sites to produce multi-parameter thematic maps forms the usSEABED project run by the United States Geological Survey (USGS) (Williams et al. 2003). Data mining, through software such as dbSEABED, maximizes the benefits of legacy data and provides a model for efficient and effective exploitation of core-based and related datasets.

Data visualization and amplified collaboration environments The rapid development of core analysis and data-mining techniques, and the high degree of collaboration between scientific and technological specialists in present-day core-based programmes and research projects, requires development of

23

new data visualization methods and amplified collaboration environments. A recent innovation in data visualization is the Geowall project (www.geowall.org). This has developed 3D scientific visualization tools to aid (principally stereoscopic projection) in the understanding of spatial relationships. Although stereoscopic projection is not new for 3D visualization, the Geowall consortium has developed for the first time a low-cost, portable system that broadens the use of virtual reality visualization and associated tools in Earth Science research and education. A basic system consists of a desktop computer (running MacOS, Linux or Windows) with a fast graphics card, two projectors and a screen. The projected image is polarized using filters within the projectors. If the image is perceived through 3D polarized glasses, then a near-true 3D image is observed (passive stereo). GeoWall visualization systems are now widely used in educational and research institutions. A related development to GeoWall is CoreWall (www.corewall.org), an integrated visualization tool for studying sediment cores developed through collaboration by the University of Illinois at Chicago's Electronic Visualization Laboratory, the US National Lacustrine Core Repository (LacCore) at the University of Minnesota and the Integrated Ocean Drilling Program (IODP). CoreWall uses high-resolution tiled LCD displays to show images of core sections along with discrete data streams of measured parameters downcore and nested images, allowing rapid multi-disciplinary interpretation (Fig. 8). CoreWall users can link to databases of core images and sensor logs, and

Fig. 8. The Corewall core data visualization system. Corewall uses high-resolution tiled LCD displays to show images of core sections along with discrete data streams of measured parameters downcore and nested images, allowing rapid multi-disciplinary interpretation. Corewall users can also link and download data from remote databases, allowing access to dispersed datasets for interpretation. Sophisticated data visualization tools like Corewall will lead to the development of amplified collaboration environments for core analysis. (Reproduced courtesy of Arun Roa and the Corewall Consortium).

24

R.G. ROTHWELL & F.R. RACK

fetch data from remote CoreWall repositories. Links to the ODP JANUS database at Texas A&M University are anticipated in the future. A prototype desktop application for CoreWall has also been developed that uses a single computer to drive four LCD panels and a single screen GeoWall display. Using this system, high-resolution core images and sensor data can be examined together with stereoscopic visualizations (Rao et al. 2004, 2005).

Conclusions Our capability to extract multi-parameter highresolution data, including important and varied palaeoenvironmental proxies, has increased dramatically during the last two decades and the impact on marine geology, particularly palaeoceanography, has been profound. A variety of non-destructive logging technologies now allow rapid acquisition of high-resolution physical property, geochemical and compositional data, providing considerable savings in time, cost and effort normally required to collect such data by more traditional laboratory means. Variation in palaeoenvironmental proxies can now be examined at decadal and, in some cases, annual and, even subannual scales, allowing unprecedented insights into the history of climate and associated environmental change. X-ray fluorescence core scanning, 3D Xradiography, visual spectrum, infrared and ultraviolet spectral reflectance, confocal macro/ microscopy, magnetic resonance imaging and developments in molecular stratigraphy all promise to provide powerful tools for 21st century geoscientists to gain greater understanding of the Earth and its history in unprecedented detail. However, challenges remain, both in the development of standard measurement and calibration methodologies and in the development of data analysis methods. New data visualization tools and techniques need to be developed to optimize the interpretation process and maximize scientific value. In order to provide an adequate interpretation infrastructure necessary for the analysis and collation of multi-parameter datasets new amplified collaboration environments and tools need to be developed. In addition, data archiving and data-mining software needs to be developed and maintained at 'stateof-the-art' level in order to integrate and interpret new and legacy datasets within the wider context of sea-floor studies. Equally important is the need to make core data and associated datasets and products accessible to, and easily searchable for, the worldwide scientific commu-

nity. Sophisticated, yet simple to use, searchable Internet databases, with universal access and secure long-term funding, and data products resulting from user-defined data-mining query and display, all so far pioneered in the USA and Australia, provide robust models for efficient and effective core data stewardship. GEOTEK of Daventry (UK) are thanked for allowing reproduction of Figure 1. We are grateful to Dr R. Carter for reproduction of Figure 2. The Ocean Drilling Program, College Station, TX, is thanked for permission to reproduce Figures 3, 4, 5 and 6. Dr A. Best is thanked for supplying Figure 7. We are grateful to A. Rao and the Corewall Consortium for permission to reproduce Figure 8. This paper was improved by comments from Dr J. Howe.

References AUFFRET, G.A. (ed.). 1997. CORSAIRES-IFREMER, Core Logging Workshop, 24-26 July 1997, Plouzan6, France, Abstracts Volume (unpublished). BALCOM,B.J., MACGREGOR,R.P., BEYEA,S.D., GREEN, D.P., ARMSTRONG,R.L. & BREMNER,T.W. 1996. Single point ramped imaging with Tl enhancement (SPRITE). Journal of Magnetic Resonance, Series A, 123, 131-134. BALSAM,W.L. & DEATON,B.C. 1991. Sediment dispersal in the Atlantic Ocean: Evaluation by visible light spectra. Reviews in Aquatic Science, 4, 411-417. BALSAM,W.L. & DEATON,B.C. 1996. Determining the composition of late Quaternary marine sediments from NUV, VIS, and NIR diffuse reflectance spectra. Marine Geology, 134, 31-55. BALSAM, W.L. & McCoY, F.W. 1987. Atlantic sediments: glacial/interglacial comparisons. Paleoceanography, 2, 531-542. BALSAM, W.L., DEATON, B.C. & DAMUTH, J.E. 1999. Evaluating optical lightness as a proxy for carbonate content in marine sediment cores: Implications for marine sedimentation. Marine Geology, 161, 141-153. BARRANCO, F.T., JR, BALSAM,W.L. t~ DEATON, B.C. 1989. Quantitative reassessment of brick red lutites: evidence from reflectance spectrophotometry. Marine Geology, 89, 299-314. BERNHARD, R.K. & CrtASEK, M. 1955. Soil density determination by direct transmission gamma rays. Proceedings of the American Society for Testing Materials, 55, 1199-1223. BEST, A.I. & GUNN, D.E. 1999. Calibration of marine sediment core loggers for quantitative acoustic impedance studies. Marine Geology, 160, 137-146. BEST, A.I., TUEFIN, M.D.J., Dlx, J.K. & BULL, J.M. 2004. Tidal height and frequency dependence of acoustic velocity and attenuation in shallow gassy marine sediments. Journal of Geophysical Research, 109, B0810l, doi: 10.1029/2003JB002748. BLOEMENDAL, J. 1983. Palaeoenvironmental implications of the magnetic characteristics of sediments from Deep Sea Drilling Project Site 514, Southeast

NEW TECHNIQUES IN CORE ANALYSIS Argentine Basin. In: LUDWIG, W.J., KRASHENINNIKOV, V.A. ET AL. (eds) Initial Reports of the Deep Sea Drilling Project, 71. US Government Printing Office, Washington, DC, 1097-1108. BOND, G., BROECKER, W., LOTTI, R. & MCMANUS, J. 1992. Abrupt colour changes in isotope stage 5 in North Atlantic deep-sea cores: implications for rapid change of climate-driven events. In: KUKLA, J. & WENT, E. (eds) Start of a Glacial. NATO ASI Series. Springer, Berlin, 185-205, BRASSELL, S.C., EGLINTON, G., MARLOW, I.T., PFLAUMANN, U. & SARNTHEIN, M. 1986. Molecular stratigraphy: a new tool for climatic assessment. Nature, 320, 129-133. BROECKER, W., BOND, G., KLAS, M., CLARK, E. 8,~ MCMANUS, J. 1990. Origin of the northern Atlantic's Heinrich events. Climate Dynamics, 6, 265-273. CACHO, I., GRIMALT, J.O. & CANALS, M. 2002. Response of the western Mediterranean Sea to rapid climate variability during the last 50,000 years: A molecular biomarker approach. Journal of Marine Systems, 33/34, 253-272. CACHO, I., PELEJERO, C., GRIMALT, J.O., CALAFAT, A.M. & CANALS,M. 1999. C 37 alkenone measurements of sea surface temperature in the Gulf of Lions (NW Mediterranean). Organic Geochemistry, 30, 557-566. CARTER,R.M. & GAMMON,P. 2004. New Zealand maritime glaciation: millennial-scale southern climate change since 3.9Ma. Science, 304, 1659-1662. CHAPMAN,M.R. & SHACKLETON,N.J. 1998. What level of resolution is attainable in a deep-sea core? Results of a spectrophotometer study. Palaeoeeanography, 13, 311-315. CHEN, Q., RACK, F.R. & BALCOM,B.J. 2006. Quantitative magnetic resonance imaging methods for core analysis. In: ROTHWELL,R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 193-207. CHEN, Q. & SONG, Y.-Q. 2002. What is the shape of pores in natural rocks? Journal of Chemical Physics, 116, 8247-8250. CHESTER, R. & ELDERF1ELD, H. 1966. The infra-red determination of total carbonate in marine carbonate sediments. Chemical Geology, 1,277-290. CHI, J. & MIENERT,J. 1996. Linking physical property records of Quaternary sediments to Heinrich events. Marine Geology, 131, 57-73. CROUDACE,I.W., RINDBY,A. & ROTHWELL,R.G. 2006. ITRAX: description and evaluation of a new multifunction X-ray core scanner. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 51-63. D'HONDT, S.L., JORGENSEN, B.B. ET AL. (eds). 2003. Proceedings of the Ocean Drilling Program, Initial Reports, 201. Ocean Drilling Program, College Station, TX. DEATON, B.C. & BALSAM,W.L. 1991. Visible spectroscopy - a rapid method for determining hematite and goethite concentrations in geological materials. Journal of Sedimentary Petrology, 61,628-632. DEGENS, E.T. 1969. Biogeochemistry of stable carbon isotopes. In: EGLINTON, G. & MURPHY, E.T.J.

25

(eds) Organic Geochemistry, Springer, Berlin, 304329. DEEP SEA DRILLING PROJECT. 1984. Design and Operation of an Advanced Hydraulic Piston Corer. IPOD/DSDP Development Engineering Technical Report, 21. DIXON, A.E., DAMASKINOS,S., RIBES, A. & BEESLEY, K.M. 1995. A new confocal scanning beam laser MACROscope using a telecentric, f-theta laser scan lens. Journal of Microscopy, 178, 261-266. EVANS, H.B. 1965. GRAPE - A device for continuous determination of density and porosity. In: Proceedings of the 6th Annual SPWLA Logging Symposium, Society of Professional Well Log Analysts, Dallas, TX, Volume 2, B1-B25. ERBACHER, J., MOSHER, D.C. ET AL. (eds). 2004. Proceedings of the Ocean Drilling Program, Initial Reports, 207. Ocean Drilling Program, College Station, TX. FORD, K.H., NAEHR, T.H., SKILBECK, C.G. AND THE LEG 201 SCIENTIFICPARTY. 2003. The use of infrared thermal imaging to identify gas hydrate in sediment cores. In: D'HONDT, S.L., JORGENSEN, B.B., MILLER, D.J. ET AL. (eds) Proceedings of the Ocean Drilling Program, Initial Reports, 201. Ocean Drilling Program, College Station, TX [CD-ROM]. Available from Ocean Drilling Program, Texas A&M University, College Station, TX 77845-9547, USA. FREIFELD, B.M., KNEAFSEY,T.J. & RACK, F.R. 2006. On-site geological core analysis using a portable X-ray computed tomographic system. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 165-178. GAFEEY, S.J. 1986. Spectral reflectance of carbonate minerals in the visible and near infra-red (0.35 to 2.55 microns): calcite, aragonite and dolomite. American Mineralogist, 71, 15 I-162. GERLAND, S. • VILLINGER,H. 1995. Non-destructive density determination on marine sediment cores from gamma ray attenuation measurements. GeoMarine Letters, 15, 111-118. GIOSAN, L., FLOOD, R.D. & ALLER, R.C. 2002. Palaeoceanographic significance of sediment color on Western North Atlantic drifts: I, Origin of colour. Marine Geology, 189, 25-41. GRUTZNER, J., GIOSAN, L. E T A L. 2002. Astronomical age models for Pleistocene drift sediments from the western North Atlantic (ODP Sites 10551063). Marine Geology, 189, 5-23. GUNN, D.E. & BEST, A.I. 1998. A new automated nondestructive system for high resolution multi-sensor core logging of open sediment cores. Geo-Marine Letters, 18, 70-77. HAGELBERG, T.K., PISIAS, N.G., SHACKLETON, N.J., Mix, A.C. &: HARRIS, S. 1995. Refinement of a high-resolution, continuous sedimentary section for studying equatorial Pacific Ocean palaeoceanography, Leg 138. In: PISIAS, N.G., MAYER, L.A., JANECEK, T.J., PALMER-JULSON,A. & VAN ANDEL, T.H. (eds) Proceedings of the Ocean Drilling Program, Scientific Results, 138. Ocean Drilling Program, College Station, TX, 31-46.

26

R.G. ROTHWELL & F.R. RACK

HAGELBERG, T.K., SHACKLETON, N.J., PISIAS, N. & SHIPBOARD SCIENTIFIC PARTY. 1992. Development of composite depth sections for Sites 844 through 854. In: MAYER, L., PISIAS, N., JANECEK,T. ET AL. (eds) Proceedings of the Ocean Drilling Program, Initial Reports, 138. Ocean Drilling Program, College Station, TX, 79-85. HAINSWORTH,J.M. & AYLMORE,L.A.G. 1983. The use of computer assisted tomography to determine the spatial distribution of soil water content. Australian Journal of Soil Research, 21,435M43. HAMILTON,E.L. 1976. Variations of density and porosity with depth in deep-sea sediments. Journal of Sedimentary Petrology, 46, 280-300. HARVEY, L.D. & HUANG, Z. 1995. Evaluation of the potential impact of methane clathrate destabilization on future global warming. Journal of Geophysical Research, 100, D2, 2905-2926. HASCHKE, M. 2006. The Eagle III BKA system, a novel sediment core X-ray fluorescence analyser with very high spatial resolution. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 31-37. HAUG, G.H., GONTHER, D., PETERSON, L.C., SIGMAN, D.M., HUGHEN, K.A. & AESCHLIMANN, B. 2003. Climate and the collapse of Mayan civilisation. Science, 299, 1731-1735. HELMKE, J.P., SCHULZ, M. t~; BAUCH,H.A. 2002. Sediment color record from the Northeast Atlantic reveals patterns of millennial-scale climate variability during the past 500,000 years. Quaternary Research, 57, 49-57. HERBERT, T.D. 2004. Alkenone palaeotemperature determinations. In: HOLLAND, E.D., TUREKIAN, K.K. & ELDERFIELD, H. (eds) Treatise on Geochemistry, Volume 6, The Oceans and Marine Geochemistry, Elsevier-Pergamon, Oxford, 391432. HUNT, G.R. 1977. Spectral signatures of particulate minerals in the visible and near infrared. Geophysics, 42, 501-513. IODP. 2001. Earth, Oceans and Life." Scientific Investigation of the Earth System Using Multiple Drilling Platforms and New Technologies. Integrated Ocean Drilling Program Initial Science Plan, 2003-2013. JAHN, B., DONNER, B., MULLER, P.J., ROHL, U., SCHNEIDER,R. & WEFER,G. 2003. Pleistocene variations in dust input and marine productivity in the northern Benguela Current: evidence of evolution of global glacial-interglacial cycles. Palaeogeography, Palaeoclimatology, Palaeoecology, 193, 515533. JANSEN, J.H.F., VAN DER GAAST, S.J., KOSTER, B. & VAARS,A. 1998. CORTEX, a shipboard XRF-scanner for element analyses in split sediment cores. Marine Geology, 151, 143-153. JARRARO, R.D. & VANDENBERG, M.D. 2006. Sediment mineralogy based on visible and near-infrared reflectance spectroscopy. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 129-140.

JENKINS, C., FLOCKS, J. & KULP, M. 2006. Integration of the stratigraphic aspects of very large sea-floor databases using information processing. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 229-240. JENKINS, R. & DE VRIES, J.L. 1970. Practical X-ray Spectrometry. Macmillan, London. KAYEN, R.E. & LEE, H.J. 1993. Slope stability in regions of sea-floor gas hydrate: Beaufort Sea continental slope. In: Submarine Landslides: Selected Studies in the US Exclusive Economic Zone. US Geological Survey Bulletin, 97-103. KAYEN, R.E., EDWARDS, B.D. & LEE, H.J. 1999. Nondestructive laboratory measurement of geotechnical and geoacoustic properties through intact core liner. In: MARR, W.A. AND FAIRHURST, C.E. (eds) Nondestructive and Automated Testing for Soil and Rock Properties, A S T M STP 1350. American Society for Testing and Materials, West Conshohocken, PA. KENNETT, J.P., CANNARIATO, K.G., HENDY, I.L. & BEHL, R.J. 2003. In: Methane Hydrates in Quaternary Climate Change: The Clathrate Gun Hypothesis. American Geophysical Union, Washington, DC, Special Publications, 54. KERR, R.A. 2004. Signs of a warm ice-free Arctic. Science, 305, 1693. KLEINBERG,R.L. 1996. Well logging. In: Encyclopedia of Nuclear Magnetic Resonance, 8. Wiley, Chichester, 49604969. KLEINBERG, R.L. 1999. Nuclear magnetic resonance. In: WONG, P.-Z. (ed.) Experimental Methods in the Physical Sciences, Volume 35, Methods in the Physics of Porous Media. Academic Press, San Diego, CA, chap. 9. KLEINBERG, R.L. 2006. Nuclear magnetic resonance pore-scale investigation of permafrost and gas hydrate sediments. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 179-192. KVENVOLDEN,K.A. 1988. Methane hydrates and global climate. Global Biogeochemical Cycles, 2, 221-229. KVENVOLDEN, K.A. 1993. Worldwide distribution of subaquatic gas hydrates. Geo-Marine Letters, 13, 32-40. KVENVOLDEN, K.A. 1998. A primer on the geological occurrence of gas hydrate. In: HENRIET, J.-P. & MIENERT, J. (eds) Gas Hydrates." Re&vance to Worm Margin Stability and Climate Change. Geological Society, London, Special Publications, 137, 9-30. KtJHLr,IANN, H., FREUDEYrHAL, T., HELMr~, P. & MEGGERS, H. 2004. Reconstruction of paleoceanography off NW Africa for the last 40,000 years: influence of local and regional factors on sediment accumulation. Marine Geology, 207, 209-234. LAMY, F., KAISER, J., NINNEMANN, U., HEBBELN, D., ARZ, H.W. & STONER,J. 2004. Antarctic timing of surface water changes off Chile and Patagonian ice-sheet response. Science, 304, 1959-1962. MAYER, L.A. 1979. Deep-sea carbonates: Acoustic, physical and stratigraphic properties. Journal of Sedimentary Petrology, 49, 819-836.

NEW TECHNIQUES IN CORE ANALYSIS MERRILL, R.B. & BECK, J.W. 1995. The ODP color digital imaging system: Color logs of Quaternary sediments from the Santa Barbara Basin, Site 893. In: KENNETT, J.P., BALDAUF, J.G. & LYLE M. (eds) Proceedings of the Ocean Drilling Program, Scientific Results, 146, (Part 2), 45-60. MITHAL, R. & BECKER,D.G. 2006. The JANUS database: providing worldwide access to ODP and IODP data. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 253-259. MIX, A., CONRAD,B. E T A L . 2003. Curators of sea-floor and lakebed samples celebrate 25 years of service. COS, Transactions of the American Geophysical Union, 84(20), 191-192. MIX, A., HARRIS, S. & JANACEK,T.R. 1995. Estimating lithology from nonintrusive reflectance spectra: Leg 138. Proceedings of the Ocean Drilling Program, Scientific Results, 138, 413-427. MIX, A., RUGH, W., PISIAS,N.G., VEIRS, S. & LEG 138 SEDIMENTOLOGISTS AND SCIENTIFIC PARTY. 1992. Color reflectance spectroscopy: A tool for rapid characterisation of deep-sea sediments. Proceedings of the Ocean Drilling Program, Initial Reports, 138, Part 1, 67-77. MOORE, C.J. & HABERMANN, R.E. 2006. Core data stewardship: a long-term perspective. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 241-251. MULLER, P.J., KIRST,G., RUHLAND,G., VON STORCH,1. & ROSELL-MELI~, A. 1998. Calibration of the t alkenone palaeotemperature index uk7 based on core tops from the eastern South Atlantic and global ocean (60~176 Geochimica et Cosmochimica Acta, 62, 1757-1772. MURRAY, J. & RENARD,A.F. 1891. Report on Deep-sea Deposits Based on the Specimens Collected During the Voyage of H M S Challenger in the Years 1872 to 1876. Challenger Report, HMSO, London. NEDERBRAGT, A.J. & THUROW, J. 2004. Digital sediment colour analysis as a method to obtain high resolution climate proxy records. In: FRANCUS, P. (ed.) Image Analysis, Sediments, Paleoenvironments. Developments in Paleoenvironmental Research, Kluwer, Dordrecht, 7, 105-124. NEDERBRAGT, A.J., DUNBAR, R.B., OSBORN, A.T., PALMER, A., THUROW, J.W. & WAGNER, T. 2006. Sediment colour analysis from digital images and correlation with sediment composition. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 113-128. NELSON, C.S., HENDY, C.H., JARRETT, G.R. & CUTHBERTSON, A.M. 1985. Near-synchroneity of New Zealand alpine glaciations and Northern Hemisphere continental glaciations during the past 750 kyr. Nature, 318, 361-363. ORTIZ, J., MIX, A., HARRIS, S. & O'CONNELL, S. 1999a. Diffuse spectral reflectance as a proxy for percent carbonate content in North Atlantic sediments. Palaeoceanography, 14, 171-186. ORTIZ, J.D., O'CONNELL, S. • MIX, A. 1999b. Data report: spectral reflectance observations from

27

recovered sediments. Proceedings of the Ocean Drilling Program, Scientific Results, 162, 259-264. ORTIZ, J.D. & RACK, F.R. 1999. Non-invasive sediment monitoring methods. In: ABRANTES, F. & MIx, A. (eds) Reconstructing Ocean History: A Window into the Future. Kluwer/Plenum, New York, 343-380. OSMENT, P.A., PACKER, K.J. s aL. 1990. NMR imaging of fluids in porous solids. Philosophical Transactions of the Royal Society of London, Series A, 333, 441-452. PAGANI, M., FREEMAN,K.H. & ARTHUR, M.A. 2000. Isotope analyses of molecular and total organic carbon from Miocene sediments. Geochimica et Cosmochimica Acta, 64, 37-49. PANCOST, R.D. & BOOT, C.S. 2004. The palaeoclimatic utility of terrestrial biomarkers in marine sediments. Marine Chemistry, 92, 239-261. PARKES, R.J., CRAGG, B.A. ET AL. 1994. Deep bacterial biosphere in Pacific Ocean sediments. Nature, 371, 410--413. PARKES, R.J., CRAGG, B.A. & WELLSBURY, P. 2000. Recent studies on bacterial populations and processes in subsea-floor sediments: a review. Hydrogeology, 8, 11-28. PAULL, C.K., USSLER,W. & DILLON, W.P. 1991. Is the extent of glaciation limited by marine gas-hydrates? Geophysical Research Letters, 18, 432-434. PETROVIC, A.M., SIEBERT,J.E. & RIEKE,P.E. 1982. Soil bulk density in three dimensions by computed tomographic scanning. Soil Science Society of America Journal, 46, 445-450. POINTECORVO, B. 1941. Neutron well-logging. Oil and Gas Journal, 40, 32-33. RACK, F.R. 1997. Geotechnical stratigraphy of the Nordic Seas: Implications for palaeoceanography. In: Development of Palaeooceanography as a New Field of Science. Meeting Commemorating the 50th Anniversary of the Swedish Deep Sea Expedition, 18-21 August. The Royal Swedish Academy of Sciences, Stockholm, Sweden, 87. RACK, F.R. 1998. Tomorrow's Technology Today. Interim report of the IMAGES Standing Committee on New Technologies in Sediment Imaging. (Available http://www.mpip-mainz.mpg.de/,,~bluemler/ mouse/info/T3_report.html, cited 17 July 2006.) RACK, F.R., BLOEMENDAL,J. ET AL. 1996. Development of physical properties relationships, interhole composite depth profiles, and sedimentological ground truthing of multi-sensor core measurements: A synthesis of results. In: THIEDE, J., MYHRE, A., FIRTH, J. E T A L . (eds) Proceedings of the Ocean Drilling Program, Scientific Results, 151, Ocean Drilling Program, College Station, TX, 595-626. RAO, A., KAMP, W. E T AL. 2004. CoreWall: A visualisation environment for the analysis of lake and ocean cores. Geological Society of America, Annual meeting and Exposition, Denver, CO, USA, 7-10 November 2004, Abstracts volume. (Available http://www.evl.uic.edu/cavern/corewall/pubs/arao_ GSA2004.pdf, cited 1 August 2005.) RAO, A., RACK, F. ET AL. 2005. CoreWall: A scalable interactive tool for visual core description, data

28

R.G. ROTHWELL & F.R. RACK

visualization, and stratigraphic correlation. Los, Transactions of the American Geophysical Union, 86(52). Fall Meeting Supplement, San Francisco, CA, USA. (Available http: //www.evl.uic.edu/ files/pdf/CoreWall_Abstract_AGU.pdf, cited 24 March 2006.) RmES, A.C., RACK, F.R., TSINTZOURAS,G., DAMASKINOS,S. & DIXON,A.E. 2006. Applications ofconfocal macroscope-microscope luminescence imaging to sediment cores. In: ROTHWELL,R.G. (ed.) New Techniques in Sediment Core Analysis'. Geological Society, London, Special Publications, 267, 141-150. RICHTER, T.O., VAN DER GAAST, S. ET AL. 2006. The Avaatech XRF Core Scanner: technical description and applications to NE Atlantic sediments. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 39-50. ROBINSON, S.G. 1993. Lithostratigraphic applications for magnetic susceptibility logging of deep-sea sediment cores: examples from ODP Leg 115. In: HAILWOOD, E.A. & K1DD, R.B. (eds) High Resolution Stratigraphy, Geological Society, London, Special Publication, 70, 65-98. ROGERSON, M., WEAVER, P.P.E., ROHLING, E.J., LOURENS, L.J., MURRAY, J.W. & HAVES, A. 2006. Colour logging as a tool in high-resolution palaeoceanography. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 99-112. ROTHWELL, R.G., HOOGAKKER,B., THOMSON,J., CROUDACE, I.W. & FRENZ, M. 2006. Turbidite emplacement on the southern Balearic Abyssal Plain (western Mediterranean Sea) during Marine Isotope Stages 1-3: an application of ITRAX XRF scanning of sediment cores to lithostratigraphic analysis, hT: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 79-98. SCHIPPERS,A., NERETIN,L.N., KALLMEYER,J., FERDELMAN, T.G., CRAGG, B.A., PARKES, R.J. & JORGENSEN, B,B. 2005. Prokaryotic cells of the deep sub-sea-floor biosphere identified as living bacteria. Nature, 433, 861-864. SCHULTHEISS, P.J. & MCPHAIL, S.D. 1989. An automated P-Wave Logger for recording fine-scale compressional wave velocity structures in sediments. In: RUDDIMAN, W., SARNTHEIN, M. ET AL. Proceedings of the Ocean Drilling Program, Scientific Results, 108. Ocean Drilling Program, College Station, TX, 4 0 7 4 13. SCHULTHEISS, P.J. & WEAVER, P.P.E. 1992. Multisensor core logging for science and industry. In: Proceedings of Ocean '92, Mastering the Oceans Through Technology, 26-29 October 1992, Newport, Rhode Island, Volume 2, The Institute of Electrical and Electronics Engineers Inc., New York, USA, 608-613. SCHULTHEISS, P.J., FRANCIS, T.J.G., ET AL. 2006. Pressure coring, logging and subsampling with the HYACINTH system. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 151-163.

SHIPBOARDSCIENTIFICPARTY. 1997. Explanatory notes. Proceedings of the Ocean Drilling Project, Initial Reports, 167, 15-39. SIKES, E.L., FARRINGTON, J.W. & KEIGWIN, L.D. 1991. The use of the alkenone unsaturation ratio U~'7 to determine past sea surface temperatures: core-top SST calibrations and methodology considerations. Earth and Planetary Science Letters, 104, 36~7. SKINNER, L.C. & MCCAVE, I.N. 2003. Analysis and modelling of gravity and piston coring based on soil mechanics. Marine Geology, 199, 181-204. SLOAN, E.D. 1990. Clathrate: Hydrates of Natural Gases. Marcel Dekker, New York. STORMS, M.A., NUGENT, W. & CAMERON, D. 1983. Hydraulic piston coring - A new era in ocean research. In: Design and Operation of the Hydraulic Piston Corer. IPOD/DSDP Development and Engineering Technical Report, 12, 1-24. THOMSON, J., CROUDACE, I.W. • ROTHWELL, R.G. 2006. A geochemical application of the ITRAX scanner to a sediment core containing eastern Mediterranean sapropel units. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 65-77. THOMSON,J., CRUDELI,D., DE LANGE,G., SLOMP, C.P., ERBA, E., CORSELLI, C & CALVERT, S.E. 2004 Florisphaera profunda and the origin and diagenesis of carbonate phases in eastern Mediterranean sapropel units. Paleoceanography, 19, PA3003 doi: 10.1029/2003PA000976. THOUVENY, N., MORENO, E., DELANGHE,D., CANDON, L., LANCELOT, Y. & SHACKLETON, N.J. 2000. Rock magnetic detection of distal ice-rafted debris: clues for the identification of Heinrich layers on the Portuguese margin. Earth and Planetary Science Letters, 180, 61-75. TREHU, A.M., BOHRMAN,G. ET AL. (eds). 2003. Leg 204 Summary. Proceedings of the Ocean Drilling Program, Initial Reports, 204. Ocean Drilling Program, College Station, TX. VINEGAR, H.J. 1986. X-ray CT & N M R imaging of rocks. Journal of Petroleum Technology, 38, 257259. WEAVER, P.P.E. & SCHULTHEISS, P.J. 1990. Current methods for obtaining, logging and splitting marine sediment cores. In: HAILWOOD, E.A. & KIDD, R.B. (eds) Marine Geological Surveying and Sampling. Kluwer, Dordrecht, 85-101. WEBER, M.E. 1998. Estimation of biogenic carbonate and opal by continuous non-destructive measurements in deep-sea sediments: application to the eastern Equatorial Pacific. Deep-Sea Research I, 45, 1955-1975. WEBER, M.E., NIESSEN,F., KUHN, G. & WIEDICKE,M. 1997. Calibration and application of marine sedimentary physical properties using a multi-sensor core logger. Marine Geology, 136, 151-172. WELLINGTON, S.L. & VINEGAR, H.J. 1987. X-ray computerised tomography. Journal of Petroleum Technology, 39, 885 898. WHITMAN, W.B., COLEMAN,D.C. & WmBE, W.J. 1998. Prokaryotes: The unseen majority. Proceedings of

NEW TECHNIQUES IN CORE ANALYSIS the National Academy of Sciences of the United States, 95, 6578-6583. WILLIAMS, S.J., JENKINS, C. ET AL. 2003. New digital maps of U.S. continental margins: insights to seafloor sedimentary character, aggregate resources and processes. In: Proceedings of the International Conference on Coastal Sediments 2003, Corpus Christi, Texas, World Scientific Publishing Corporation and East Meets West Productions, Corpus Christi, Texas, USA.

29

ZACHOS, J.C., KROON, D. ET AL. 2004. Proceedings of the Ocean Drilling Program, Initial Reports, 208. Ocean Drilling Program, College Station, TX. ZOLITSCHKA, B., MINGRAM, J., VAN DER GAAST, S., JANSEN,J.H.F. & NAUMANN,R. 2001. Sediment logging techniques. In: LAST, W.M. & SMOL,J.P. (eds) Tracking Environmental Change Using Lake Sediments. Basin Analysis, Coring and Chronological Techniques, Developments in Paleoliminological Research. Kluwer, Dordrecht, 1-17.

The Eagle III BKA system, a novel sediment core X-ray fluorescence analyser with very high spatial resolution M. H A S C H K E R6ntgenanalytik Messtechnik GmbH, Georg-Ohm-Strasse 6, 65232 Taunusstein, Germany Present address." IFG - Institute for Scientific Instruments, Rudower Chaussee 29, D-12489 Berlin, Germany (e-mail: haschke@ ifg-adlershof de) Abstract: Sediment cores contain valuable information on geological history and palaeoclimate, and it is necessary to study cores at high spatial resolution to recover useful palaeoenvironmental data. This paper describes an innovative micro-X-ray fluorescence spectrometer that can measure spot sizes in the range of 50 ~tm, with incremental step sizes down to 10~tm. The analysis is non-destructive so the sediment core is preserved and available for other investigations. This paper describes the new instrument, the Eagle III BKA system, which employs some novel features such as a polycapillary lens, moveable sample stage and X-ray head, a Varispot focusing system and an environmental sample cabinet. Some examples of recent applications are described.

Sediment layers contain information on the geological and climate history of the Earth. Elemental and mineralogical analysis of such layers in boreholes or sampled from marine or lacustrine environments can reveal important palaeoenvironmental information. Core samples taken from deep lakes or the ocean floor are important data sources as, potentially, they are less disturbed by anthropogenic influences. The Ocean Drilling Program (ODP), for example, has produced many thousands of metres of core samples that have provided a wealth of important palaeoenvironmental information. In core analysis there are many different parameters of interest including, for example, colour, magnetic susceptibility, density and elemental composition. Elemental composition can be determined by cutting the core into thin slices, dissolving the slices into solution and then using conventional analytical methods such as atomic absorption spectroscopy (AAS) or, preferably, inductively coupled plasma emission spectroscopy (ICP-AES) for analysis. However, analysis by these methods means that some sample is often destroyed and, hence, no longer available for further analysis. In principle much smaller samples can be taken for conventional analysis (c. 0.5g: Croudace et al. 2006) but spatial resolution by these destructive methods is often not better than 5 mm. Another technique for element analysis is Xray fluorescence (XRF) (Hahn-Weinheimer 1984; Erhardt 1989), and conventional 'bulk' Xray spectrometers use an analysed area of a few centimetres diameter, which is inappropriate for the high-resolution necessary for many modern palaeoenvironmental studies. Micro-beam XRF

scanners, with their inherent small beam sizes, allow non-destructive analysis at high spatial resolution and good sensitivity for several elements (Janssens et al. 2000; Haschke et al. 2002). The high spatial resolution of these scanners readily provides the temporal resolution necessary to analyse thin sediment layers. In this paper, an innovative X R F scanner is described that can be used for this type of core analysis and some initial applications are described.

Initial instrument design requirements Sediment cores typically have a high moisture content and should not be allowed to dry out otherwise useful textural relationships may be destroyed or damaged when drying as shrinkage and cracking can occur. This damage results in potential loss of scientific information and, therefore, any water evaporation, even that occurring during analysis, needs to be minimized. Cores are normally kept at 4 ~ in refrigerated stores for optimum preservation, and during any measurement some environmental control is ideally required. This can be achieved by cooling the sample and/or maintaining a high humidity within the measurement chamber. It is preferable, if possible, to measure parameters along the complete core length to avoid further damage to the core caused by cutting into shorter sections. A spatial resolution down to 20 Mm is required to provide the detailed information necessary for many modern palaeoenvironmental studies. This means that both the spot size of the excitation beam and the step-size increment of the sample stage should be in this

From: ROTHWELL,R.G. 2006. New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 31-37. 0305-8719/06/$15.00 9 The Geological Society of London 2006.

32

M. HASCHKE

range for optimal analysis. However, such highresolution measurements can have their own problems when the size of the constituent minerals are of the same order as the X-ray beam diameter. In micro-XRF analysis all important elements need to be excited and detected efficiently. Closely positioned low-power X-ray tubes are convenient since their power and cooling requirements are modest. An energy dispersive X-ray detector is an optimal choice as it offers simultaneous measurement of all elements present above detectable limits. Design of the Eagle III BKA instrument The Eagle III BKA micro-XRF core scanner was developed from an instrument that was already on the market, called the Eagle ~Probe. In the Eagle BKA instrument the same well-proven measuring unit is used but with a new sample chamber. A photograph of the Eagle BKA instrument is shown in Figure l.

Sample handling The requirement to analyse the core over its total length can be achieved either by moving the measuring unit (i.e. the X-ray tube and detection system) above a fixed sample or by moving the sample below a fixed measuring unit. In the second case the sample chamber needs to be at least twice the length of the sample itself, in our case 3000 mm. The Eagle III BKA system presented here uses a combination of these options. The sample chamber has a length of 2000 ram. The sample is positioned on a stage that can be moved by 300 mm (core axis) x 150 mm (perpendicular) x 100 mm (height), with a minimum step size of 10~tm. In this way only part of the complete sample can be measured in one run;

Fig. 1. The Eagle III BKA sediment analyser.

however, if the measuring head is moved by 300mm into another measuring position, the same movement is possible again. The design of the instrument, with its combined movable sample holder and measuring head, is shown schematically in Figure 2. A step size of 50 ~tm requires 6000 measurement steps for the maximum stage movement. For a measurement time of 60s per point, this gives a total measuring time of more than 100h. This means that a change to the measuring head position should not to be performed too often. The maximum number of steps that can be handled in the operating program is 15000. For every measured position the complete spectrum is saved.

Sample environment Reduction of the partial pressure of the water in the sample to avoid drying of the sample during measurement can be carried out in different ways - by cooling the sample itself or by acclimatizing the sample chamber as necessary. Cooling of the sample requires significant energy and may reduce the temperature markedly. This can lead to condensation of water onto the sample but also onto parts of the instrument. Water drops on the sample will cause absorption/scattering phenomena that will adversely affect the data collected. Possible condensation onto the equipment surfaces and components requires careful selection of the fabrication material to avoid corrosion. By controlling the sample chamber humidity and temperature it is possible to reduce the partial pressure, but at temperatures higher than the dew point condensation can be avoided. In the instrument described, the sample chamber can be set at a temperature of between 4 and 40 ~ and the relative humidity can be controlled between 30 and 96%. In this

THE EAGLE III BKA SYSTEM

33

Fig. 2. Schematic view of the Eagle III BKA sediment analyser showing configuration and design. way evaporation of water from the sample is reduced, so there is little danger that the sample will dry out during analysis. Further, it is possible to protect the areas of sample that have just been analysed or are awaiting measurement by covering with a thin plastic film to prevent water loss.

Analytical conditions Excitation of the sample is induced using a lowpower air-cooled Rh X-ray tube, which gives the best compromise between a high excitation intensity and a low sensitivity to other elements by the tube spectrum. The only element that is slightly influenced by the tube spectrum is chlorine. The excitation parameters of the tube are controllable up to 40 kV and 1 mA. The radiation generated by the tube is concentrated by a capillary optic, a polycapillary lens, to irradiate a very small area of the sample. The configuration of the polycapillary lens is designed to capture a large solid angle of the radiation from the tube, and by multiple total reflection along the component capillaries of the lens this radiation is propagated through the lens and concentrated into a small area on the sample. Typical spot sizes are in the range of 30-50 ~tm. To obtain larger spot sizes, the distance between lens and sample can be increased by moving the complete tube with the polycapillary lens along the lens axis. In this situation the sample is not necessarily now at the 'focal point' of the lens; the spot size is enlarged but the number of excitation photons is still the same, i.e. the total emergent flux from the tube will not be changed but the quanta per unit area will change. In this way

the spot size can be varied between 50 and c. 400 gm. This is the principle of the 'VariSpot' configuration, illustrated in Figure 3. The sample should be at the same distance, d, to the detector because this distance influences the measured intensity by lid 2. Therefore, changes in the 'sample~letector' distance should be avoided. For this reason an optical autofocus system is provided which guarantees that the measuring point is at the correct 'sample-detector' distance. The fluorescence radiation of the sample is detected by an energy-dispersive detector (LN2 cooled Si(Li), energy resolution better than 140eV, sensitive area 30mm2). This type of detector gives a sufficient energy resolution and sensitivity for trace-element analysis. The main advantage of energy dispersive detection is its simultaneous measurement of all elements between light elements (AI, Si) and U. The sample and its holder are placed into the measuring chamber. The measuring points can be set with the help of a video-microscope. This shows the sample and the measuring point at several magnifications, the video images of which can be saved as bitmap files. The measuring points can be saved - each point can be defined as a line or matrix for the determination of linear or area element distributions. In the case of a line profile, the start and end points together with the number of equidistant points must be defined. The measurement commences automatically after selecting the measurement conditions and the acquired spectra are automatically saved after each incremental acquisition. This means that spectra can be re-evaluated later, thus

34

M. HASCHKE

Fig. 3. Schematic diagram showing the principle of the VariSpot X-ray focusing system.

allowing for identification of additional elements, new calibration, etc. Quantification of the X R F data is achieved using a standardless Fundamental Parameter Model (Sherman 1955; Elam 2002). This model calculates X-ray fluorescence intensities for an artificial sample and compares this with the measured intensities. By an iterative process the concentrations in the artificial sample are changed until predicted and measured intensities are identical. This calculation can be made using atomic parameters such as mass attenuation coefficients, transition probabilities, fluorescence yields, etc. No measurement of standards is

necessary. With this model the accuracy of concentrations is in the range of 1-3%. This can be improved by incorporating several reference materials. Then the same model allows an improvement by a factor of 2-3. The limits of detection depend on the element of interest. The excitation with X-rays and energy dispersive detection of fluorescence radiation provides detection limits in the range of 20-50 ppm for transition metals (Ti-Zn) and approximately 100 ppm for heavy elements. Technical specifications for the Eagle gProbe and the Eagle BKA instruments are given in Table 1.

Table 1. Specification of the Eagle pProbe and BKA instruments Parameter

Specification Eagle gProbe

Specification Eagle BKA

Excitation Tube parameters Spot size

Rh-fine focus tube 40 kV, 1000gA 30 gm

Rh-fine focus tube 40 kV, 1000gA 50-400 gm (Varispot)

Detection Detector parameters

Si(Li)-detector 30mm2, <140eV (MnKa)

Si(Li)-detector 30mm2, <140eV (MnKa)

Sample positioning Stage parameters

X-Y-Z stage Max. travel: 100 x 100 x 100mm Step size: 5 lam

X-Y-Z stage Max travel: 300 x 150 x 100mm Step size: 10pm

Sample view

Optical microscope with fixed magnification and autofocus function

Optical microscope with fixed magnification and autofocus function

q5330 x 350 mm air, vacuum not available not available

2000 x 600 x 600 mm air, radiation path can be He-flushed 4 ~ to room temperature, • 1 ~ 30-96% relative humidity

Sample chamber Size Measurement medium Temperature control Humidity control

THE EAGLE III BKA SYSTEM

Examples of applications Early measurements on sediments were performed with the Eagle gProbe (Haschke et al. 2002) and are published by Haug et al. (2003). The sample chamber of the Eagle ~tProbe is relatively small (see Table 1) and therefore only short cores can be examined. The analysed core was taken from the Cariaco Basin, offshore Venezuela, and the data have been interpreted as suggesting climate changes (regional drought) that may have contributed to the collapse of the Mayan civilization. Earlier studies on cores from a similar location with lower spatial resolution are described by Haug et al. (2001). The spatial resolution measured in these earlier studies was approximately 2ram. This corresponds to a temporal resolution of about 5 years. With the higher spatial resolution achievable using the Eagle gProbe the temporal resolution achieved

35

was approximately 2 months. This allows annual or subannual changes in sedimentation to be examined. The differences between these two sets of measurements are shown in Figure 4. Bulk titanium content was used in this study as a proxy for terrigenous sediment delivery to the Cariaco Basin from surrounding watersheds and provides a proxy for the strength of regional hydrological conditions. The upper profile shows the Ti content with the more limited resolution over the last 2000 years. In comparison, the lower profile, measured at higher spatial resolution, shows the Ti content variation for a period of about 150 years. The fluctuations of the Ti content in these measurements are seen in much more detail. Lower Ti reflects lower precipitation resulting in lower river runoff, and the high resolution achieved allows discrete multiyear-duration drought events (Ti minima) to be identified. Three periods of drought, each lasting

Fig. 4. Ti content in two cores from the Cariaco Basin, offshore Venezuela, measured with different spatial resolution. Ti is used here as a proxy for terrigenous sediment delivery to the basin from surrounding watersheds and reflects changes in rainfall. Such data have been crucial in determining the probable role of regional drought in the collapse of Mayan civilisation (adapted from Haug et al. 2003, figs 2 & 3).

36

M. HASCHKE

Fig. 5. Optical image of part of a sediment core from the Meerfelder Maar lake in the Eifel Mountains, western Germany. Measurement transects are shown by red lines (see Figs 6 & 7 and the text).

Fig. 6. Distribution of Fe and Si for the complete length of the sediment core from the Meerfelder Maar.

for a decade or less, were identified that were coincident with the three phases of Maya city abandonment around 810, 860 and 910AD previously postulated on archaeological evidence. Another study using the Eagle BKA core scanner was made on lake sediment from the Meerfelder Maar (Dulski pers. comm.), a lake located in the Eifel Mountains in the west of Germany. The measured part of the sediment core had a length of approximately 100 mm and is illustrated in Figure 5.

The conditions for Eagle BKA measurements were: Excitation conditions: Measuring time: Spot size: Step size:

40kV, 170 ~tA 60 s 50 pm 30 ~m

Measurements were made along two parallel transects 1.6mm apart (see Fig. 5). Figure 6 shows the distribution of Fe and Si over the interval with a length of 100mm. This measurement

Fig. 7. Distribution of several elements along 6 mm of the measurement line in the Meerfelder Maar core (Fig. 5).

THE EAGLE III BKA SYSTEM took approximately 75 h. All spectra of this run were saved and thus remain available for later evaluation. The element distribution in Figure 6 is shown at large scale and, hence, fine details are lost. It is, however, possible to zoom into small intervals of core to examine element distribution in greater detail. Element distribution along the two parallel transects, displayed as red lines in Figure 5, for a region covering only 6 mm is shown in Figure 7. A good agreement between the distribution of Fe and the optical image is observed. The element distribution along both lines is very similar, but there is a slight shift due to the structures not being perpendicular to the axis of the sediment core.

Conclusions A new micro-XRF instrument, that incorporates several novel features, is presented that can be used for the analysis of split sediment cores. The innovative character of this instrument is that it provides the capability to analyse sediment at very high spatial resolution, down to a 3 0 g m spot size and a 10gm step size. This allows identification of fine microstructure and associated features in sediment layers. The measurements can be collected via an automated 'auto-run' procedure and the results automatically prepared for visual presentation. The author is grateful to Drs I. Croudace and G. Rothwell of the National Oceanography Centre, Southampton, UK, for their reviews and assistance in improving the text and figures.

37

References CROUDACE,I.W., RINDBY,A. & ROTH'vVELL,R.G. 2006. ITRAX: description and evaluation of a new multifunction X-ray core scanner. In: ROTHWELL,R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 5 i-63. DULSKI, P. GF7 Pulsdam, personal information. ELAM, W.T., RAVEL, B.D. & SIEBER, J.R. 2002. New atomic database for X-ray spectroscopic calculations. Radiation Physics and Chemistry, 63, 121128. ERHARDT, H. 1989. Rfntgenfluoreszenzanalyse, Anwendungen in Betriebslaboratorien. Dt. Verlag fiJr Grundstoffindustrie, Leipzig. HAHN-WEINHEIMER,P. 1984. Grundlagen undpraktische Anwendung der Rfntgenfluoreszenzanalyse (RFA ). Vieweg, Braunschweig. HASCHKE, M., SCHOLZ, W., THE1S, U., NICOLOSI, J., SCRUGGS, B. & HERZCEG, L. 2002. Description of a new Micro-XRay spectrometer. Journal de Physique IV France, 12, 6-83, doi:10.1051/jp4:20020216. HAUG, G.H., GONTHER, D., PETERSON,L.C., SIGMAN, D.M., HUGr~N, K.A. & AESCnLIMANN, B. 2003. Climate and the collapse of Mayan civilization. Science, 299, 1731-1735. HAU6, G.H., HUGHEN, K.A., SIGMAN, D.M., PETERSON, L.C. & R6HL, U. 2001. Southward migration of the intertropical convergence zone through the Holocene. Science, 293, 1304-1308. JANSSENS,K.H.A., ADAMS,F.C.V. & RINDBY,A. (eds). 2000. Microscopic X-ray Fluorescence Analysis. Wiley, Chichester. SHERMAN,J. 1955. The theoretical derivation of fluorescent X-ray intensities from mixtures. Spectrochimica Acta, 7, 283-306.

The Avaatech XRF Core Scanner: technical description and applications to NE Atlantic sediments T H O M A S O. R I C H T E R 1, S J E R R Y V A N D E R G A A S T l, B O B K O S T E R 2, A A D V A A R S 3, R I N E K E G I E L E S l, H E N K O C. D E S T I G T E R 1, H E N K D E H A A S a & T J E E R D C. E. V A N W E E R I N G 1

1Department of Marine Chemistry and Geology, Royal Netherlands Institute for Sea Research (NIOZ), P.O. Box 59, 1790 AB Den Burg, The Netherlands (e-mail: [email protected]) 2Department of Marine Technology, Royal Netherlands Institute for Sea Research (NIOZ), P.O. Box 59, 1790 AB Den Burg, The Netherlands 3Avaatech Analytical X-Ray Technology, Wagenmakerstraat 11, 1791 EJ Den Burg, The Netherlands Abstract: X-ray fluorescence (XRF) core scanning provides rapid high-resolution (down to

1 mm) records of chemical composition on split sediment cores. The measurements are non-destructive and require very limited sample preparation. The new Avaatech XRF Core Scanner, operational since 2002, covers the atomic mass range from A1 to U. Instrument parameters, especially tube voltage, can be adjusted to provide optimum settings for selected elements or sets thereof. Owing to the nature of the surface of split sediment cores, particularly effects resulting from sample inhomogeneity and surface roughness, results are semiquantitative, yet provide reliable records of the relative variability in elemental composition downcore. Selected case studies from diverse sedimentary settings in the NE Atlantic Ocean illustrate a range of applications of XRF logging data. These include preliminary stratigraphic interpretations (glacial-interglacial cycles), provenance studies of the terrigenous sediment fraction, lithological characterization, early diagenetic processes and distinction between carbonate phases (aragonite v. calcite).

Various logging methods, mostly dealing with physical sediment properties, are presently widely used to produce high-resolution or quasi-continuous records for rapid, non-destructive characterization of marine and lacustrine sediment sequences (e.g. Weaver & Schultheiss 1990; Zolitschka et al. 2001). These measurements, frequently performed shortly after core retrieval during research cruises or field campaigns, can already provide preliminary stratigraphic information. More generally, they represent a first means of assessing the quality of sedimentary records and provide constraints on sampling strategies for subsequent analysis on discrete samples, commonly at lower resolution and/or over selected core intervals. X-ray fluorescence (XRF) logging was first developed at The Netherlands Institute for Sea Research (NIOZ) in 1988 (Jansen et al. 1998). An updated version of this C O R T E X scanner was delivered to the University of Bremen in 1997, and the second-generation Avaatech X R F Core Scanner (ACS) has been operational since 2002 at N I O Z and University of Bremen. X R F core scanning essentially tracks downcore

changes in chemical composition (major, minor and trace elements). By contrast, most other logging data (e.g. magnetic susceptibility, gammaray attenuation, seismic velocity, sediment colour) usually depend on a combination of sediment parameters, such as grain-size changes combined with changes in the ratio of terrigenous v. biogenic components (e.g. Bassinot 1993; Weber et al. 1997). Applications of X R F core scanning include, among others, initial correlation between cores, preliminary stratigraphic interpretations including astronomical tuning of sedimentary sequences (e.g. Norris & R6hl 1999; Pfilike et al. 2001), investigating terrigenous input patterns and provenance of terrigenous material (e.g. Haug et al. 2001; Lamy et al. 2004), tracing early diagenetic processes (Funk et al. 2004) and recognition of sedimentological events by specific lithologies (e.g. turbidites, sapropels, ash layers). In this contribution, we first provide a technical description of the X R F logging system, including recent developments and improvements of the new Avaatech X R F Core Scanner (http://www.avaatech.com). Then, we discuss

From: ROTHWELL,R.G. 2006. New Techniquesin Sediment Core Analysis. Geological Society, London, Special Publications, 267, 39-50. 0305-8719/06/$15.00 9 The Geological Society of London 2006.

40

T.O. RICHTER ET AL.

applications of X R F core scanning based on case studies from the NE Atlantic Ocean.

X-ray fluorescence (XRF) logging The principle of X-ray fluorescence is discussed by Jenkins & De Vries (1970). Briefly, under the influence of incoming X-ray radiation an electron is ejected from an inner shell of an atom. The resulting vacancy is subsequently filled by an electron falling back from an outer shell, and the energy difference between both shells is emitted as electromagnetic radiation. The wavelength(s) of emitted radiation are characteristic for each element, and the amplitudes of peaks in the X R F spectrum are proportional to the concentration of corresponding elements in the analysed sample.

Set-up o f X R F logging system and comparison o f C O R T E X (1988, 1997) and A vaatech (2002) scanners The CORTEX and Avaatech scanners perform continuous downcore XRF analysis on the surface of split sediment cores. Technical details for both scanners are presented in Table 1. Both instruments are installed in sea-going containers and can thus be used onboard research vessels. The basic instrument set-up is virtually identical for both scanners (Figs 1 & 2). A prism is lowered onto the sediment surface at each measurement position, and incoming radiation generated in the X-ray tube enters at an angle of 45 ~. The length of the X-ray beam (in downcore direction) is determined by a horizontally mounted variable slit system below the Xray tube, the width of the beam is fixed at 12 or 16mm, respectively (Table 1). The detector for outgoing radiation is likewise installed at an angle of 45 ~. The entire measurement system is flushed with helium to prevent partial or

complete absorption of emitted soft radiation from the sample by air. With a net counting time of 30 s as commonly applied, each individual multi-element analysis takes approximately 1 min. For exploratory analysis of major elements only, reliable count statistics can be achieved with lower counting times. For example, 1 m of sediment can be logged for Ca and Fe to obtain a preliminary stratigraphy of marine sediments (cf. below) in 15-20 min (2 cm resolution, counting time 10 s). In the CORTEX scanner, split core halves are transported along the measurement system. In the Avaatech instrument the measurement system instead moves along the half core, thus reducing the overall instrument size. The Avaatech XRF Core Scanner allows for a larger atomic weight range of elements to be analysed (Table 1). This was made possible because measurement conditions, especially tube voltage, can be varied to obtain optimal settings for specific elements or sets thereof (Table 2). Filters can be placed in the incident beam to attenuate part of the hard radiation from the X-ray tube spectrum, thus improving sensitivity for a range of elements of interest. Detection limits of elements in the atomic mass range K-Sr are also significantly lower for the Avaatech scanner compared to the earlier CORTEX system (Table 3).

Samples and sample preparation For X-ray fluorescence analysis with conventional laboratory instruments, samples need to be homogenous, dry, and have a flat and smooth surface. Split sediment cores do not meet these requirements. The water content generally ranges from 30 to 70% in the uppermost metres of marine sediments, with a rapid decrease immediately below the sediment surface. Effects of sample inhomogeneity and surface roughness are particularly pronounced for sediments containing abundant

Table 1. Technicaldetails of CORTEX and Avaatech XRF scanners

X-ray tube Target X-ray tube Film of measuring cell Film on sample Irradiated sample dimensions Detector Window of detector Elements analysed (atomic weight range) Processing software of measured spectra Accuracy along the X-axis of the scanner

CORTEX (1988, 1997)

Avaatech (2002)

KEVEX Mo Mylar 4 lam Polypropylene 100 lam L = 10mm, W = 12mm Energy dispersive, KEVEX SuperDry Si(Li), Peltier cooled Be 10 Bm K-Sr KEVEX not determined

Oxford Rh Ultralene 4 I.tm Ultralene 4 I.tm L = 0.5-10ram, W = 16mm Energy dispersive, AMPTEK Si-P1N,Peltier cooled Be 13 lam A1-U CANBERRA +0.035 mm

AVAATECH XRF CORE SCANNER

(a)

41

(b)

(c) Fig. 1. Views of the Avaatech XRF Core Scanner. (a) General overview of the instrument. (b) Measurements on a split sediment core. Arrow on core surface depicts the (stepwise) movement of the measurement system along the core surface. (c) Detailed view of the X-ray tube (A), filters that can be placed in the incoming X-ray beam (B), prism lowered onto the sediment surface covered with Ultralene film (C) and XRF detector (D). medium-coarse sand-sized particles such as shell fragments in coastal environments or foraminifera and ice-rafted grains in deep-sea settings, but are less significant in most fine-grained marine sediments.

Sample preparation includes careful flattening of the sediment surface to remove irregularities from core slicing. The sediment surface is subsequently covered with thin (4 lam) Ultralene film, further diminishing surface roughness and

Fig. 2. Simplified diagram showing principle of XRF logging on split sediment cores and response depth (not to scale) of selected elements to incoming X-ray radiation (see text).

42

T.O. RICHTER ET AL.

Table 2. Instrumental settings of the Avaatech XRF Core Scanner for specific sets of elements Tube voltage (kV)

Filter

Elements analysed

5 l0

none none

30 50

Pd thick Cu

(Mg), Al*, Si* A1, Si, P, S, C1, K, Ca, Ti, Mn, Fe, Cu, Zn Br, Rb, Sr, Zr Ba, Pb, U

32.2keV) is 2-4ram. For lighter elements, the detection depth is progressively smaller, for example 1-2mm for Sr (14.2keV), 1 mm for Fe (6.4 keV), 0.5 mm for Ca (3.7 keV) and 0.05mm for AI (1.5keV). Consequently, the detection limit for various elements also, to a first approximation, decreases with increasing atomic weight (Table 3). For heavy elements, such as Zr and Ba, the detection limit again increases because the XRF detector is slightly less efficient at the high wavelengths of their emitted radiation.

* At low bulk concentrations or for higher accuracy. preventing contamination of the prism unit during core logging. The poor quality of samples used for XRF logging, compared to fused beads or powder pellets of conventional laboratory X-ray fluorescence, necessarily means that XRF logging data are only semi-quantitative. However, the very limited sample preparation and rapid analysis allow for high sample throughput, and XRF logging records faithfully trace relative downcore variations in the elemental composition of sediments. Response depth to incoming X - r a y radiation and element detection limits

The response depth of elements to incoming Xray radiation is dependent on the wavelength of emitted radiation, itself related to atomic weight, and on the chemical composition of the matrix (Jenkins & De Vries 1970). For average marine sediments with significant amounts of Si, A1 and/or Ca and some Fe, the maximum detection depth for Ba (emitted wavelength

Spatial resolution

The irradiated sample length (ISL) of the Avaatech XRF Core Scanner can be varied between 0.5 and 10mm. Generally, a maximum resolution of 1 mm is recommended in order to maintain sufficient analytical volume (sample volume from which emitted X-rays are detected) for reliable count statistics. Further, the effects" of sample inhomogeneity and surface roughness become increasingly important at higher spatial resolution, resulting in more strongly fluctuating X-ray intensities between adjacent samples and a lower signal-to-noise ratio. In fine-grained clayey sediments, these effects are less pronounced to negligible, and an ISL and spatial resolution of 0.5 mm may be applied. For heavy elements (e.g. Sr, Ba) interaction of their emitted radiation with (lighter) elements present in the sample matrix will result in the X R F signal effectively being derived from a larger area in downcore (and crosscore) direction, hence decreasing the true spatial resolution. Thus, a step size and ISL of 3 and 2 mm for Ba and Sr, respectively, is generally sufficient, as

Table 3. Detection limits of the CORTEX and A vaatech XRF core scanners (selected elements) Detection limit (ppm) Element

Atomic weight

Ks line (keV)

Mg A1 Si P S K Ca Ti Mn Fe Sr Zr Ba Pb

24 27 28 31 32 39 40 48 55 56 88 91 137 207

1.25 1.49 1.74 2.01 2.31 3.31 3.69 4.51 5.89 6.40 14.14 15.74 32.19 10.84 (L~)

n.a., not applicable

CORTEX

Avaatech

n.a.

20000 2000 1000 500 500 400 200 500 100 45 5 20 40 10

n.a. B.a.

n.a. n.a. 2000 800 1200 1400 210 260 n.a. n.a. n.a.

AVAATECH XRF CORE SCANNER

43

Table 4. Locations and settings of sediments cores, and origin of XRF records Core

Latitude Longitude Water depth (m) Setting

Elements shown

MD992282 ENAM9606 ENAM9321 STRAT01-06 M2001-05 ENAM9706

59~ 55~ 62~ 62~ 53~ 53~

Ca, Fe CORTEX Ca CORTEX K, Ti CORTEX Ca, Si, A1, Fe, Mn, S Avaatech Ca, Fe, Sr Avaatech Ca, Sr Avaatech

10~ 13~ 4~ l~ 14~ 15~

1360 2543 1020 1576 840 2131

measurements at higher resolution will not provide additional information on the downcore variability of these elements.

Applications to NE Atlantic sediments Examples from NE Atlantic sediments, covering various timescales and diverse sedimentary settings, serve to illustrate a wide range of applications of XRF core scanning (Table 4, Fig. 3). We focus on preliminary conclusions that can be derived from XRF logging data alone or in combination with other non-destructive shipboard datasets (e.g. magnetic susceptibility records, lithological description). As the XRF records shown below were obtained partly with the CORTEX (1988) instrument and partly with the Avaatech Core Scanner, absolute element count rates are not directly comparable between all records.

Seamount Sediment drift Sediment drift Mud volcano Carbonate mound Continental margin

Instrument

glacial-interglacial cycles, with higher carbonate concentrations during interglacial periods (Kennett 1982; Balsam & McCoy 1987). Hence, preliminary stratigraphic interpretations can be based on downcore X R F records of Ca and Fe, tracing fluctuations in the relative abundance of biogenic carbonate and terrigenous material, respectively. Core MD992282 from the southern flank of Rosemary Bank seamount contains 34.35m of alternating light-coloured silty sand and darkercoloured silty clay. The X R F record (Fig. 4) displays recurrent antiphase oscillations of Ca and Fe, with higher Ca count rates corresponding to coarser-grained light-coloured sediments.

Ca (carbonate) records: Milankovitch and sub-Milankovitch variability It is well established that calcium carbonate records in the Atlantic Ocean can be related to

Fig. 3. Simplified bathymetry of the NE Atlantic Ocean showing locations of sediment cores used in this study.

Fig. 4. Ca and Fe count rates for core MD992282 (Rosemary Bank) showing glacial-interglacial variability. Shaded rectangles in upper part of figure indicate interglacial intervals (light-coloured sediments, high Ca, low Fe). Data gaps are due to (partial) voids in sediment sections.

44

T.O. RICHTER E T AL.

Fig. 5. Left panel: tentative correlation of MD992282 Ca record with smoothed and spliced benthic oxygen isotope record from Feni Drift ODP sites 980/981 (Raymo et al. 2004 and references therein). Stippled lines indicate tie points between both records. Numbers next to oxygen isotope records represent selected interglacial isotope stages. Right panel: resulting age model. The thick black line represents the general age-depth relation; stippled horizontal lines depict upcore increases in mean long-term sedimentation rate. The grey line indicates the agedepth relation for individual glacial-interglacial cycles. Numbers are sedimentation rates (cm ka -1) between peak interglacial oxygen isotope events (grey diamonds). The shaded rectangle corresponds to an increase in ice volume-related mean 6J8 0 (transition midpoint: 922 4- 12 ka, duration 40 • 9 ka, after Mudelsee & Schulz 1997). Downcore, an obvious shift towards higher-frequency variability in the depth domain occurs below approximately 18 m core depth. Figure 5 illustrates the tentative correlation between the record of Ca count rates and a record of benthic oxygen isotopes (Raymo et al. 2004 and references therein) from the nearby Feni Drift (ODP sites 980 and 981) tracing global ice-volume fluctuations over the last 1.4Ma. According to this stratigraphic scheme, the shift in the character of the X R F record around 18 m core depth corresponds to the Mid-Pleistocene transition from predominantly 41 ka cyclicities of global ice volume in the lower Pleistocene to larger mean global ice volume and pronounced 100 ka cycles in the upper Pleistocene. The exceptional character of Marine Isotope Stage 11 (highest Ca count rates of entire X R F record) is consistent with growing evidence for the unusual nature of this interglacial period (e.g. Howard 1997).

Fig. 6. Ca count rates and CaCO3 concentrations from Feni Drift core ENAM9606.

Average sedimentation rates for successive glacial-interglacial cycles show a two-step increase upcore. The first increase closely coincides with an increase in ice-volume-related mean 6180 between 940 and 900 ka BP (Mudelsee & Schulz 1997), suggesting a coeval change in depositional conditions possibly related to changing average current regimes. The second increase towards distinctly higher sedimentation rates during the last glacial cycle could represent an artefact of core stretching, as commonly observed in the uppermost metres of giant piston cores (Sz6r6m6ta et al. 2004). A high-resolution record from Feni Drift, sediment core ENAM9606, provides evidence for carbonate variability on sub-Milankovitch timescales (Fig. 6). Following a rapid increase in Ca count rates during the later part of Termination I (12-10ka), a further slight increase spanning nearly the entire Holocene is punctuated by pronounced cyclic variability. The X R F record shows generally good agreement with a lower-resolution carbonate record based on discrete samples. The apparently increasing difference between both records in the late Holocene is ascribed to the higher water content of nearsurface sediments. Holocene carbonate variability may be related to variable dilution by bottom-current-transported terrigenous material

AVAATECH XRF CORE SCANNER

45

Fig. 7. XRF and magnetic susceptibility records for Faeroe Drift core ENAM9321. Stratigraphy after (Rasmussen et al. 1996, 1998). Shaded rectangles correspond to Heinrich events 1-5; numbers represent selected ice-core interstadials; B/A, Bc~lling-Allerod;YD, Younger Dryas. (e.g. Chapman & Shackleton 2000) and/or to fluctuating carbonate productivity. The nature of this centennial- to millennial-scale variability and possible forcing mechanisms will be fully discussed elsewhere. P r o v e n a n c e o f terrigenous m a t e r i a l a n d Dansgaard-Oeschger

cycles

Millennial-scale climate variability during the last glacial cycle was initially deduced from Greenland ice-core records (Dansgaard et al. 1993). Correlative variations were first described for marine sediments from the Faeroe Margin (Rasmussen et al. 1996), and subsequently in a growing number of marine and terrestrial palaeoclimate records worldwide (Voelker &

Workshop Participants 2002). These Dansgaard-Oeschger cycles are characterized by recurrent shifts in North Atlantic deep-circulation patterns. Magnetic susceptibility records indicate enhanced deposition of magnetic minerals by bottom currents during interstadial (warm) events, and reduced bottom-current activity during stadial (cold) periods and Heinrich events (Rasmussen et al. 1996; Kissel et al. 1999). Following some earlier evidence based on clay mineral assemblages and quantitative chemical analysis (e.g. Blamart et al. 1999; Rasmussen et al. 1998), we show here (Fig. 7) that the Dansgaard-Oeschger cycles in Faeroe Margin sediments are also accompanied by repetitive shifts in the composition of the terrigenous fraction.

46

T.O. RICHTER ET AL.

Potassium and titanium can be dominantly linked to acidic (average continental crust) and basaltic sources, respectively. The XRF intensity ratio of these elements (K/Ti), tracing the relative importance of both terrigenous sources, shows a systematic relationship with magnetic susceptibility which was itself previously correlated to the Greenland ice-core record. This implies enhanced deposition of bottom-current-transported basaltic material derived from the North Atlantic volcanic province during interstadial periods (low K/Ti), and an increased contribution of continentally derived material, possibly through ice rafting from Fennoscandia, during stadial periods and, especially, Heinrich events.

M u d volcano sediments at the FaeroeShetland Margin." lithological characterization and early diagenetic processes Core STRAT01-06 was taken in the northern Faeroe-Shetland Channel on top of a mud volcano, which forms part of a group of relatively small structures rising several tens of metres above the surrounding seabed (Long et al. 2003). It contains two lithological units separated by a sharp boundary at 260cm core depth. The lower unit, characterized by numerous ram- to cm-scale clasts embedded in a fine-grained matrix, is interpreted as being related to mud volcano activity. The overlying upper unit is similar to typical hemipelagic sediments in the wider study area. These two units are clearly distinguished on the XRF logging and magnetic susceptibility records (Fig. 8). The upper unit shows cyclic fluctuations of magnetic susceptibility consistent with the regional pattern during the last glacial period (compare with Fig. 7), the record of Ca count rates depicts a thin (c. 3cm) Holocene section at the core top. The boundary with the underlying mud volcano unit shows an approximately 10-fold decrease in both magnetic susceptibility and Ca count rates, with consistently low values for both parameters in underlying sediments. The mud volcano deposits can be divided into two subunits based on a second magnetic susceptibility decrease at approximately 540cm core depth. The lower mud volcano unit contains occasional mm-sized light-coloured limestone clasts, which cause sharp short-lived maxima of Ca count rates in the XRF record. While Si count rates remain comparatively constant across the lithological boundary, the relationship between Si and A1 indicates different sources of silicon for the

Fig. 8. Magnetic susceptibility and XRF records from core STRAT01-06 (Faeroe-Shetland margin) showing the sharp lithological boundary between mud volcano deposits and overlying hemipelagic sediments. mud volcano deposits and hemipelagic cover sediments (Fig. 8). Fairly constant Si/Al-ratios in the hemipelagic unit imply that Si is mainly derived from terrigenous aluminosilicates, whereas higher and more variable ratios in mud volcano deposits suggest a significant contribution from biogenic opal. Count rates of Fe, Mn and S trace early diagenetic processes within the mud volcano unit (Fig. 9). Sulphur count rates show two abrupt downcore increases at 260 and 330cm core depth. The first increase coincides with the main lithological boundary, whereas the second increase seems to mark the limit between oxidizing and reducing conditions within the mud volcano unit. The oxidized upper part shows three sharp maxima in Fe count rates and a pronounced double Mn peak, all of which are interpreted as

AVAATECH XRF CORE SCANNER

47

palaeo-oxidation fronts. The diagenetic origin of the Fe spikes is confirmed by sharp increases in the Fe/Ti ratio. While Fe and Ti are closely related to each other in the terrigenous fraction, Fe is (partly) prone to diagenetic remobilization in pore waters, whereas Ti is inert to diagenetic processes. The reduced lower part of the mud volcano unit displays numerous short-lived maxima of S count rates, consistently accompanied by concomitant Fe maxima, implying precipitation of pyrite or other Fe-sulphide phases within reducing microenvironments. The conclusions discussed above are also illustrated by element cross-plots (Fig. 10). The plot of Siv. AI demonstrates that data from the two lithological units plot in different fields, consistent with different sources of silicon (cf. above). In the plot of Fe v. S (only data from lower unit shown) there is little relationship between both elements for most data points where Fe is presumably of terrigenous origin. On the other hand, diagenetic Fe enrichments fall outside the main cluster of data points, and Fe-oxide and Fe-sulphide phases can be easily distinguished based on S count rates.

Cold-water carbonate mounds: detection o f hardgrounds and coralline aragonite

Fig. 9. XRF records from core STRAT01-06 to illustrate early diagenetic processes within mud volcano deposits. Shading indicates oxidized top of mud volcano unit (light grey pattern) and underlying reducing environment (dark grey pattern).

Cold-water carbonate mounds with deep-water corals (mainly Lophelia pertusa and Madrepora oculata) are widespread along both margins of Rockall Trough in the NE Atlantic Ocean, rising up to about 350 m above the surrounding sea floor (e.g. de Mol et al. 2002, Kenyon et al. 2003, van Weering et al. 2003). Here we present results from piston core M2001-05, which was taken on the lower flank of a mound within a mound cluster on SE Rockall Trough margin.

Fig. 10. Diagnostic element cross-plots from core STRAT01-06 showing lithological characterization (left panel) and different sources of iron in the mud volcano deposits (right panel).

48

T. O. RICHTER E T AL. This core consists dominantly of cemented foraminiferal sands with intercalated hardgrounds. Numerous core intervals contain fragments of cold-water corals, presumably reflecting temporal variability of coral growth and/or variable preservation of coralline aragonite. Magnetic susceptibility and X R F records are shown in Figure 11. Calcium count rates are high throughout the sediment record (excluding the top 20cm), consistent with the dominant lithology. The scatter in the Ca record can be largely ascribed to increased surface roughness and scattering effects from large particles in sandy sediments. Hardgrounds display sharp maxima of magnetic susceptibility and Fe count rates. They are also characterized by high count rates of numerous other terrigenous elements (not shown). Compared to the Ca record, Sr count rates are highly variable; sediment intervals with coral fragments consistently have higher Sr count rates than over- and underlying intervals. Hence, Sr count rates in this setting serve to distinguish between foraminiferal calcite and coralline aragonite, the latter carbonate phase containing significantly higher percentages of Sr. This is further illustrated by the XRF records from a nearby pelagic reference core (Fig. 12), where Ca and Sr show largely consistent patterns of variability over several glacial-interglacial cycles. Maximum Sr count rates in the carbonate mound record are at least twice as high as those in the pelagic reference core.

Fig. 11. Magnetic susceptibility and XRF records from cold-water carbonate mound core M2001-05 (SE Rockall Margin). Shaded rectangles in the upper panel indicate hardgrounds with enhanced terrigenous input. The line with symbols in the lower right panel represents a five-point running mean of Sr count rates; the thin grey line corresponds to raw data. Shaded rectangles indicate core intervals containing fragments of cold-water corals.

Concluding remarks X-ray fluorescence core scanning provides high-resolution (quasi-continuous) palaeoenvironmental information in a variety of sedimentary settings. While the results are inherently semiquantitative due to the nature of the surface of split-sediment cores (particularly due to effects of sample inhomogeneity and surface roughness),

Fig. 12. XRF records (Ca and Sr) from Porcupine Margin core ENAM9706 (pelagic reference site for carbonate mound core M2001-05).

AVAATECH XRF CORE SCANNER X R F logging records faithfully trace relative downcore variability in the elemental composition of sediments. As the instrument is installed in a sea-going container, results can potentially be obtained onboard research vessels within hours after core retrieval and can be used to adapt ongoing sampling programmes. The measurements are non-destructive and can provide constraints on sampling strategies for subsequent analysis on discrete samples. X R F logging results can be interpreted in combination with other non-destructive (shipboard) data such as magnetic susceptibility records and lithological description. Cross-elemental plots and element intensity ratios can be applied to enhance the signal-to-noise ratio, because - to a first approximation - variable surface roughness effects (for example, related to downcore grainsize variability) will affect X R F count rates for all elements in a similar way. Inter-element relationships can also be helpful to distinguish between multiple sources for some elements and provide clues on sediment mineralogy. For example, combined distribution patterns of Si and AI trace the relative contributions of aluminosilicates v. biogenic opal to total sedimentary silicon. The coupled use of Fe, Ti and S can differentiate between a dominantly terrigenous supply o f iron and distinct early diagenetic iron enrichments in oxidizing and reducing environments. Sr and Ca distribution patterns taken together define the relative importance of aragonite (high Sr) v. calcite (low Sr). We thank K.-C. Emeis, one anonymous reviewer and the editor, G. Rothwell, for evaluating the manuscript.

References BALSAM,W.L. & McCoY, F.W., JR. 1987. Atlantic sediments: Glacial/interglacial comparisons. Paleoceanography, 2, 531-542. BASSINOT,F. 1993. Sonostratigraphy of tropical Indian Ocean giant piston cores: toward a rapid and highresolution tool for tracking dissolution cycles in Pleistocene carbonate sediments. Earth and Planetary Science Letters, 120, 327-344. BLAMART, D., BALBON, E., K1SSEL, C., TURPIN, L., ROBIN, E., LABEYRIE,L. & DECONINCK,J.-F. 1999. Deep water circulation variability in the southern Norwegian Sea during the last glacial period. Terra Abstracts, 4, 164. CHAPMAN,M.R. & SHACKEETON,N.J. 2000. Evidence of 550-year and 1000-year cyclicities in North Atlantic circulation patterns during the Holocene. The Holocene, 10, 287-291. DANSGAARD, W., JOHNSEN, S.J. ET AL. 1993. Evidence for general instability of past climate from a 250kyr ice record. Nature, 364, 218-220.

49

DE MOL, B., VAN RENSBERGEN,P. E~" A/.. 2002. Large deep-water coral banks in the Porcupine Basin, southwest Ireland. Marine Geology, 188, 193-231. FUNK, J.A., VON DOBENECK,T. & REITZ, A. 2004. Integrated rock magnetic and geochemical quantification of redoxomorphic iron mineral diagenesis in Late Quaternary sediments from the equatorial Atlantic. In: WEFER,G., MULITZA, S. & RATMEYER, V. (eds) The South A tlantic in the Late Quaternary: Reconstruction of Material Budgets and Current Systems. Springer, Berlin, 237560. HAUG, G.H., HUGHEN, K.A., SIGMAN, D,M., PETERSON, L.C. & ROHL, U. 2001. Southward migration of the Intertropical Convergence Zone through the Holocene. Science, 293, 1304-1308. HOWARD, W.R. 1997. A warm future in the past. Nature, 388, 418~419. JANSEN, J.H.F., VAN DER GAAST, S.J., KOSTER, B. 84 VAARS,A. 1998. CORTEX, a shipboard XRF-scanner for element analyses in split sediment cores. Marine Geology, 151, 143-153. JENKINS, R. & DE VRIES, J.L. 1970. Practical X-ray Spectrometry. Macmillan, London. KENNETT, J.P. 1982. Marine Geology. Prentice-Hall, Eaglewood Cliffs, NJ. KENYON,N.H., AKHMETZHANOV,A.M., WHEELER,J.W., VANWEERING,T.C.E., DE HAAS, H. & IVANOV,M.K. 2003. Giant carbonate mud mounds in the southern Rockall Trough. Marine Geology, 195, 5-30. KISSEE, C., LAJ, C., LABEYRIE, L., DOKKEN, T., VOELKER, A. & BLAMART,D. 1999. Rapid climatic variations during marine isotopic stage 3: magnetic analysis of sediments from Nordic seas and North Atlantic. Earth and Planetary Science Letters, 171, 489-502. LAMY, F., KAISER, J., NINNEMANN,U., HEBBELN, D., ARZ, H.W. & STONER,J. 2004. Antarctic timing of surface water changes off Chile and Patagonian ice-sheet response. Science, 304, 1959-1962. LONG, D., HOULT~R. ET AL. 2003. Mud mound/?diapiric features in the Faroe-Shetland Channel. Geophysical Research Abstracts, 5, 11201. MUDELSEE, M. & SCHULZ, M. 1997. The Mid-Pleistocene climate transition: onset of 100 ka cycle lags ice volume build-up by 280 ka. Earth and Planetary Science Letters, 151, 117-123. NORRIS, R.D. & R6HL, U. 1999. Carbon cycling and chronology of climate warming during the Palaeocene/Eocene transition. Nature, 401, 775 778. P,~,LIKE,H., SHACKLETON,N.J. & ROHL U. 2001. Astronomical forcing on late Eocene marine sediments. Earth and Planetary Science Letters, 193, 589-602. RASMUSSEN,T.L., THOMSEN,E. & VANWEERING,T.C.E. 1998. Cyclic sedimentation on the Faeroe drift 5310 ka BP related to climatic variations. In: STOKER, M.S., EVANS,D. & CRAMP,A. (eds) Geological Processes on Continental Margins." Sedimentation, Mass-wasting and Stability. Geological Society, London, Special Publications, 129, 255-267. RASMUSSEN,T.L., THOMSEN, E., VAN WEERING,T.C.E. & LABEYRIE, L. 1996. Rapid changes in surface and deep water conditions at the Faeroe Margin during the last 58,000 years. Paleoceanography, 11,757-771.

50

T. O. RICHTER ET AL.

RAVMO, M.E., OaPo, D.W. ET AL. 2004. Stability of North Atlantic water masses in face of pronounced climate variability during the Pleistocene, Paleoceanography, 19, PA2008, doi: 10.1029/ 2003PA000921. SZI~REMI~TA,N., BASSINOT,F., BALUT,Y., LABEYRIE,L. & PAGEL, M. 2004. Oversampling of sedimentary series collected by giant piston corer: Evidence and corrections based on 3.5-kHz chirp profiles. Paleoceanography, 19, PAl 005, doi: l 0.1029/2002PA000795. VAN WEERING, T.C.E., DE HAAS, H., DE STIGTER,H.C., LYKKE-ANDERSEN, H. & KOUVAEV, I. 2003. Structure and development of giant carbonate mounds at the SW and SE Rockall Trough margins, NE Atlantic Ocean. Marine Geology', 198, 67-81. VOELKER, A.H.L. & WORKSHOP PARTICIPANTS. 2002. Global distribution of centennial-scale records for

Marine Isotope Stage (MIS) 3: a database. Quaternao' Science Reviews, 21, 1185-1212. WEAVER, P.P.E. & SCHULa~tEISS, P.J. 1990. Current methods for obtaining, logging and splitting marine sediment cores. Marine Geophysical Researches, 12, 85-100. WEBER, M.E., NIESSEN, F., KUHN, G. & WIEDICKE, M. 1997. Calibration and application of marine sedimentary physical properties using a multi-sensor core logger. Marine Geology, 136, 151-172. ZOLITSCHKA, B., M1NGRAM, J., VAN DER GAAST, S., JANSEN, J.H.F. & NAUMANN, R. 2001. Sediment logging techniques. In: LAST, W.M. & SMOL, J.P. (eds) Tracking Environmental Change Using Lake Sediments. Volume 1: Basin Analysis, Coring, and Chronological Techniques. Kluwer, Dordrecht, 137-153.

ITRAX: description and evaluation of a new multi-function X-ray core scanner I A N W. C R O U D A C E

1, A N D E R S

R I N D B Y 2 & R. G U Y

ROTHWELL 1

1National Oceanography Centre, Empress Dock, Southampton S014 3ZH, UK 2Cox Analytical Systems, Ostergardsgatan 7, SE-431 53 Molndal, Sweden Abstract: A new automated multi-function core scanning instrument, named ITRAX, has

been developed that records optical, radiographic and elemental variations from sediment half cores up to 1.8 m long at a resolution as fine as 200 pm. An intense micro-X-ray beam focused through a flat capillary waveguide is used to irradiate samples to enable both Xradiography and X-ray fluorescence (XRF) analysis. Data are acquired incrementally by advancing a split core, via a programmable stepped motor drive, through the flat, rectangular-section X-ray beam. Traditional XRF determination of element composition in sediments provides high-quality data, but it takes a considerable time and normally consumes gram quantities of material that is often only available in limited quantities. The ITRAX core scanner non-destructively collects optical and X-radiographic images, and provides highresolution elemental profiles that are invaluable for guiding sample selection for further (destructive) detailed sampling. This paper presents a description of the construction, characteristics and capabilities of the ITRAX system. High-resolution ITRAX data obtained from sediment cores are also presented and compared with results from traditional wavelength-dispersive XRF analysis at lower resolution. Finally, some recent technical developments linked to the second-generation ITRAX are presented.

Elemental and other sediment property variations along core profiles can be used to infer environmental, sedimentological and diagenetic changes, pollution inputs and to assist in correlation studies. Traditional methods of acquiring solidphase geochemical data from sediment cores are time-consuming and involve incremental sampling to obtain gram quantities of material and further processing before analysis by techniques such as X-ray fluorescence (XRF) analysis. Conventional X R F analysis requires dried and ground sediment and two sample preparation methods are commonly used (Croudace & Williams-Thorpe 1988; Croudace & Gilligan 1990; Jenkins 1999). The fusion method produces a solid solution in the form of a glass bead by dissolving about 0.5 g of sample in a lithium borate flux for determination of major and the more abundant trace elements. The pelletization method utilizes at least 3 g of ground sample, pressed into a briquette at 15-20 tonnes, and allows major and trace elements to be determined. Notably, material from these X R F pellets may be re-used for other analyses or tests. In conventional X R F analysis grinding and pelletization of samples reduces mineral and particle-size effects and density variations while the fusion variant circumvents particle-size and mineralogical problems and reduces inter-element effects by dissolving the sample and providing a consistent matrix. The entire preparation and

analytical process for a 1 m core subsampled at 1 cm increments would take up to 2 weeks. Non-destructive scanners that incorporate X R F analysis provide useful, high-resolution geochemical records from terrestrial and marine sediment and drilled rock cores (e.g. Jansen et al. 1998; Rothwell et al. 2006; Thomson et al. 2006). Particle size, mineralogy and density effects will exist in cores, and direct analysis will lead to some degradation of data quality compared to the wellcontrolled approaches used in the conventional X R F analysis procedures above. Some of the new scanners do, however, allow relatively rapid and continuous analysis of split sediment cores to provide records of downcore geochemical and textural variations. At present these instruments generally serve the valuable role of providing initial detailed characterization of sediment cores. Sections of interest can then be analysed at higher accuracy (although generally at poorer spatial resolution) using well-established, variably destructive methods such as X R F analysis or inductively coupled plasma optical emission spectroscopy.

The ITRAX core scanner

The I T R A X core scanner (Fig. 1) is unique among the current generation of core scanners

From: ROTHWELL,R.G. 2006. New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 51-63.0305-8719/06/$15.00

9 The Geological Society of London 2006.

52

I.W. CROUDACE ET AL.

Fig. 1. Front view of the ITRAX core scanner with open hoods. Split core sections are moved incrementally from the left extension to the right extension during analysis.

in being designed to gather optical and microradiographic images and micro-X-ray fluorescence spectrometry (gXRF) elemental profiles for the same sediment core section. It can operate on split cores of sediment or rock with a maximum length of t 800 mm and a diameter ranging from a few cm up to 12 cm. The instrument consists of a central measuring tower incorporating an X-ray focusing unit and a range of sensors. These sensors include an optical-line camera, a laser topographic scanner, an X-ray line camera for measuring the transmitted X-rays and a high count-rate X R F detection system. Further information on the technology incorporated into the I T R A X scanner is available from the literature (Rindby et al. 1997; Vincze et aL 1998; Janssens et al. 2000; Streli et al. 2004; Tsuji et al. 2004). On either side of the tower are two fixed projections that are part of the motorized splitcore transport bed. 'Flat-beam' scanning technology (Bergstr6m et al. 2001) is used to provide highly resolved, two-dimensional radiographic images and profiles of X R F spectral data over the selected sample length. The system is fitted with numerous safety interlocks to ensure mechanical and radiological security. S c a n n i n g p r o c e d u r e a n d control s o f t w a r e The I T R A X scanner is controlled from a central computer that runs on a Windows XP platform. Users control the system through a graphical user interface called the Core Scanner Navigator where standard operation procedures can be

implemented and monitored (Table 1). A second program, Q-Spec, interacts with this interface and is activated automatically during certain operations. Q-Spec is primarily concerned with the display and on-the-fly analysis of X R F spectral data, but it can also be used to refine the X-ray spectral analysis after core scanning on the instrument is complete. The standard ITRAX procedure starts by loading a split sediment core onto a horizontal cradle with the core-top positioned to the right. Scanning is initiated from the software and follows a logical and guided sequence involving inputs: (i) to define the core length to be scanned; (ii) setting the excitation voltage and current to the X-ray tube; and (iii) initiating a surface topographic scan of the core. This topographic scan is made in relation to a horizontal reference plane and is used to ensure that any subsequent positioning of the X R F detector does not lead to a collision of the detector with the sediment surface and, importantly, that the detector-sample distance is monitored and remains constant. The scanning occurs by a regular left to right incremental movement of the core perpendicular to the long axis of the rectangular beam (Fig. 2). This initial scan takes approximately 5 rain after which the core is automatically returned to its start position. The next procedure (iv) involves defining the likely elements in the sample from a periodic table as well as adjusting and refining the peak-fitting parameters using one or more representative parts of the core sample. This optimization takes place with the operator ensuring

THE ITRAX CORE SCANNER

53

Table 1. Summary of typical ITRAX operatorprocedures" Operator task or input

Result

Load split sediment core on to horizontal sample cradle

Sample ready for scanning

Define kV and mA setting for 3 kW Mo X-ray tube

Excitation condition set

Define core dimension to be scanned

Dimensional limits set for scanning

Initiate surface scanning/photographic procedure (approximately 5 min)

Surface topography profile determined and digital photographic image of core captured

Enter the scan increment size and the dwell time for the radiographic scan

Radiographic parameters set ready for the radiographic scan

Establish XRF parameters and select the elements likely to be present

Elements selected and spectral-fitting parameters refined

Set reference response for the radiographic camera by automatically removing the core from the X-ray beam

Calibration of the X-ray line camera diode array

Enter dialogue menu where data file storage locations are named and the XRF count time defined

Automated process commences with the acquisition of an incremental (digital) radiographic scan. The core is then returned to zero ready for next stage

Instrument commences XRF analysis

Incremental XRF scans acquired and stored

that the best peak-fitting functions are chosen. The next step (v) is to record a reference response for the X-radiographic detector by automatically driving the core out of the path of the X-ray beam. This step calibrates and normalizes the response of the X-ray line camera diode array and takes approximately 1 min after which the core is returned to its start position. The final step in the Navigator panel is (vi) to enter the Batch Analysis Mode where the user defines the

core name, reviews instrument count times, dwell times and scan limits before starting the scan process. The first operation of the ensuing automated process is to acquire and construct the X-radiographic image, after which the core again returns to the start position and then begins the incremental acquisition of XRF elemental profiles. The XRF spectral data are all stored and can, if necessary, be re-processed and refined by the user after scanning is complete. Such re-analysis may be required if the user has neglected to include an element for output or if some further refinement to the background or peak-fitting process is deemed necessary. A straightforward facility exists for efficiently reprocessing a batch of spectra.

X R F spectral analysis software

Fig. 2. Schematic of the ITRAX system showing the optical-line camera (A), laser triangulation system (B), motorized XRF Si-drift chamber detector (C), 3 kW X-ray tube (D), flat-beam X-ray waveguide (E) and the X-ray line camera and slit system for the radiographic line camera (F). The horizontal arrow shows the incremental motion direction of a core and the vertical arrows the movement directions of the XRF detector.

The Q-Spec spectral analysis software employs standard fitting procedures to extract the individual elemental peak areas from the spectrum. Each peak is described by a Gaussian function with an exponential tailing on the low-energy side with a step-like background function. The general spectral background is described with a multi-parameter function including polynomial as well as exponential terms. As previously mentioned, the operator selects from a periodic table the elements to be extracted from the Xray spectra. Any incorrect or unnecessary elemental choices or incorrect fitting parameters can be adjusted later through a batch-controlled post-processing of the spectra.

54

I.W. CROUDACE ET AL.

Instrumental components Optical camera system. An optical-line camera system is used to generate good-quality RGB digital images of the sediment sample surface before the X-ray scan. The line camera incorporates a light-sensitive 2048 pixel CMOS (complementary metal oxide semiconductor) device that has a maximum resolution of 50gm pixel -l The optical image is available to allow the operator to define the part of the sample to be analysed before scanning, and later serves to relate the radiographic and XRF data to visual colour features of the sediments. X-ray source andJocusing. The ITRAX uses a 3 kW X-ray generator and can be used with different tube anodes to obtain excitation for a range of elements. The current system uses a 3 kW molybdenum target tube that can operate up to 60 kV and 50 mA, but the actual voltagecurrent selected can be optimized for the elements required. In practice 30 kV and 30 mA are suitable for most elements. The X-rays emerging through the shutter of the tube turret are focused by means of a proprietary flatbeam optical device (not a collimator), generating a 20 x 0.2ram rectangular beam with its long axis perpendicular to the sample main axis. X-ray line camera. A digital X-ray line camera is used for recording the intensity of X-radiation transmitted through the sample. The camera, which consists of a linear arrangement of 1024 X-ray sensitive diodes, has a pixel resolution of approximately 20 gm and can operate with exposure times from 20 ms up to several seconds. The images produced by the ITRAX are 'radiographic positives', so that low-density areas appear light and higher density areas appear darker. The X R F detection system Reliable XRF analysis requires that the sampledetector distance is kept constant and the ITRAX achieves this by using a laser triangulation system to measure the topography of the sample surface. The XRF detector (Fig. 2), fitted to a vertical motorized stage, adjusts itself according to the topographic scan data previously acquired. The X-ray detector used is a Si-drift detector (SDD). These are proportional solid-state devices that have the advantage of providing high-energy resolution at high count rates but do not require liquid nitrogen cooling, unlike the alternative Si(Li) detectors (Lechner et al. 2001). The combination of the SSD, the associated electronics and the digital multi-channel analyser (MCA) allow

count rates up to 70kcps to be handled with only minor degradation in resolution. The SDD operates in a similar manner to a conventional drift chamber. Thus, the electron cloud generated by X-ray quanta absorbed at the surface in the SDD is guided onto a specific readout site on the device. In this way the SDD is analogous to an antenna for X-ray photons. The detector is also fitted with a pumped acrylic nozzle with a thin plastic entrance window to minimize the Xray attenuation that would otherwise occur with absorption in the air passage between the sample surface and detector.

Conventional X-radiography Until the advent of digital X-ray cameras, photographic film was used to record images. If the sample is thin and the X-ray source is small enough, each point in the recording device will correspond to a single point in the exposed sample (point-projection radiography). If, however, the source has a finite size there is no longer a simple point-to-point relationship between the recording device and the object. In principle, the structure in the exposed object will be affected by the source size. Structures in the object that are much smaller than the source size will appear blurred in the radiograph, but for structures that are much larger than the source the finite source size will manifest itself in a diffuse zone around the structure termed a penumbra (shadow). The penumbra phenomenon is normally considered to set the ultimate limit of resolution in point-projection radiography. The size of the penumbra is related to the source size and the ratio of the distances (object-camera)/(X-ray source-object). For thick objects, image distortions will also occur because of geometrical aberration generated from rays penetrating the objects far from the optical axis as these rays will penetrate the object off the perpendicular direction. Thus, density gradients that appear perpendicular to the optical axis will only appear sharp in the radiograph close to the optical axis, while further away they will become smoothed out due to this geometrical effect. These phenomena are significant limitations in recording radiographs from extended objects, particularly for objects extending in the same direction as the density gradient to be investigated. This might be the case for cores from trees or sediments where the main interest is to investigate the density variation along the length axis. Another source of distortion for thick objects is from forward-scattered Compton (incoherent) radiation that is also recorded with the transmitted radiation. For

THE ITRAX CORE SCANNER

t2

55

D

Fig. 3. Examples of ITRAX X-radiographic images from a variety of sediments: (A) laminated sediment core (200 mm length); (B) a 25 mm detail (left of centre) from (A); (C) a Moroccan laminite (3 cm in length); and (D) North Sea layered sandstone reservoir rock (5 cm in length). All images are 20 mm wide. thinner objects the problem is less severe as forward Compton scattering is substantially reduced, but the forward Rayleigh (coherent) scattering component will still degrade spatial resolution.

Benefits of ITRAX flat-beam radiography. One way to avoid the problems of geometrical aberrations and blurring from scattered radiation is to combine a narrow, parallel, high-flux X-ray beam with an X-ray line camera, and to record the radiograph by moving the object in a regular incremental manner. The use of a high-flux beam allows the line camera to record transmitted signals with relatively short dwell times per increment. The ideal movement direction of the object is parallel to the axis of the density gradient (e.g. perpendicular to sediment layering). Radiographic images are built up by successively adding the recorded radiographic information, line by line, while new areas of the sample move through the beam. By using a narrow (adjustable) slit between the object and the line camera the geometrical aberration and the contribution of scattered radiation can be reduced to practically zero, particularly in the vertical direction. The spatial resolution in the vertical direction is set by the width of the slit in front of the X-ray line camera, and the resolution in the horizontal direction is set by the size of the individual pixels in the diode array. ITRAX radiographic resolution. The resolution and contrast in radiographic images is related to the type of X-ray source used. The 3 kW Mo X-ray tube commonly used produces sufficient energy at 55 kV for transmission through splitcore samples of 10cm diameter. For thinner core samples lower X-ray energies from a Cu or Cr tube could also be used effectively for specific

applications. During factory set-up the radiographic resolution and contrast are evaluated using a 1 cm-thick parallel slab of finely laminated sediment, an 11 cm-thick layered sediment and an aluminium block with 15 sharp steps. The resolution is estimated by comparing the sharpness of the intensity rise corresponding to the each step in the aluminium block while the slit width was gradually reduced. When no further increase in sharpness is observed when reducing the slit width, the resolution corresponds to the actual slit size. Resolution is also recorded by estimating the minimum size of significant layers in the radiographs from the two sediment samples while reducing slit width in the same manner described above. The results show that for thin samples the ultimate resolution achievable is about 20 lam, while for approximately 11 cm-diameter split-sediment cores the ultimate resolution is about 100pro. Some examples of image quality and resolution are shown in Figure 3 for a range of sample types.

I T R A X X R F analysis X R F elemental detection limits are related to the type of X-ray source used, and the 3 kW Mo tube is well suited for working on split-core samples as it produces good excitation for a range of elements of environmental interest. Enhanced excitation for the lower Z elements may be achieved, at the cost of poorer excitation of the medium Z elements, by using a Cu or Cr tube. However, the regular changing of tubes is preferred to be avoided due to the inconvenience of dismantling and re-instating the water-cooling circuitry. In practice, a single Mo tube is adequate and a good range of elements can be obtained when operating at 30 kV and 30 mA.

56

I.W, CROUDACE ET AL.

I T R A X X R F sensitivity The sensitivity of the X R F system was evaluated using various international reference samples; United States Geological Survey (USGS) marine sediment MAG-1, USGS manganese nodules N O D - A - 1 and N O D - P - l , U S G S Green River Shale SGR-I and US National Institute of Standards and Testing (NIST) Borosilicate Glass 1411. Samples from these materials

(except the NIST glass) were prepared as dry, pelletized briquettes and measured on the I T R A X . The detection limits were defined as the elemental concentration corresponding to a peak area equivalent to three times the square root o f the average background at the peak position. For a given sample composition, the detection limits vary substantially across the Xray energy range according to tube anode, tube voltage, count rate and sample composition, as

Table 2. Compar&onbetween ITRAX and conventional WD-XRF ITRAX

Conventional WD-XRF

Services required

Three-phase power, water cooling

One- or three-phase power, water cooling

Typical X-ray tube used

3 kW Mo or 2.4 kW Cu

4 kW Rh

High resolution X-radiography provided

Yes

No

Optical image provided

Yes

No

X-radiographic spatial resolution (selectable)

_>100 ~tm

Not possible

Potential to add other sensors to the scanner

Yes

Unlikely

Time to acquire optical image X-radiograph data for a I m core at 200 lam resolution

0.5 h

Not possible

Sample treatment and preparation requirement

Non-destructive Flat exposed surface needed 6 ~tm polypropylene film used to inhibit drying during analysis

Semi-destructive (pellets) or destructive (beads) Requires 3 g or more of sample; involves subsample removal, drying, grinding and pelletization

Vacuum system required for XRF analysis

Limited option

Yes

Helium system option available for volatile or powder samples

No

Yes

Practicable scanning resolution

_>100 lam

5000 ~tm

Time to acquire data for a I m core at 200 lam scanning increments for selected elements (K, Ca, Fe, Sr)

2h

10 working days (for 100 sample at 1 cm resolution)

Time to acquire data for a 1 m core at 200 pm scanning increments (e.g. Si, AI, K, Ca, Ti, Fe, Mn, Zn, Sr, Zr)

15h

10 working days

Time to acquire data for a 1 m core at 200 t.tm 48 h scanning increments (e.g. Si, A1, S, CI, K, Ca, Fe, As, Pb, Zn, Br, Rb, Sr, Zr)

I0 working days

Elemental capability for an argillaceous sediment Elemental capability for a carbonate sediment

Si, AI, S, CI, K, Ca, Ti, Fe, As, Pb, Zn, Br, Rb, Sr, Zr, Ba CI, Ca, Ba, Fe, Zn, Br, Sr

Na-U Na-U

Nominal detection limits (100 s); see Table 3 Dependent on tube anode, excitation conditions, count-time, atomic number and sample composition

150 ppm for Ti 10 ppm for Sr

10 ppm for Ti 0.5 ppm for Sr

General analytical data quality for sediments

Good for elements named above and particularly for finegrained materials

High for most elements because of highly controlled preparation approach

THE ITRAX CORE SCANNER is well established in XRF analysis (Jenkins 1999). For indicative purposes the detection limits were recorded with a Mo-anode X-ray tube operating at two different voltages and current, optimized for medium-heavy elements and lighter elements, respectively (Table 2). It is noteworthy that wet and organic-rich sediment material may be associated with poorer detection limits caused by the less efficient excitation of elements because of increased scattering, but grain size and degree of compaction may also affect response.

Second-generation ITRAX The prototype ITAX core scanner output all the data as peak area integrals for elements. Further developments allowed quantitation of the count data, and comparison of these integrals with quantitative W D - X R F data (Figs 4 & 6, later) shows that there is a good correlation between the profile shapes obtained with the two techniques - except in some situations where X-ray absorption/enhancement effects become apparent with the ITRAX data. One clear example of this is shown in figure 4 of Thomson et al. (2006) where the sapropel layer (S1), which is rich in sea water and organic matter, shows a lower K/Ti than the main trend. This response is an X R F inter-element effect caused by the sea-water chlorine atoms absorbing the potassium K a X-rays. Had the count-rate data been converted to concentrations then the lower K/ Ti dip shown should not occur as inter-element corrections would have been applied. X R F e l e m e n t a l sensitivity

The second-generation ITRAX shows a significant improvement in the overall X R F elemental

57

Table 3. ITRAX detection limits for ITRAX instruments

Voltage Current A1 Si K Ca Ti Mn Fe Rb Sr

ITRAX (Prototype)

ITRAX (Second Improvement generation) factor

45kV 40mA not detectable 33 000ppm 450ppm 200ppm 120ppm 40ppm 60ppm 10ppm 10ppm

30kV 40mA 22000ppm 9000ppm 150ppm 100ppm 60ppm 25ppm 25ppm 5ppm 5ppm

4x 3x 2• 2x 2x 2x 2x 2•

The data are based on measurements using international geochemical reference samples USGSMAG1 and USGS-SGR1. Note that wet sediment may be less efficiently excited owing to less compaction, higher water content and grain-size issues. Measurement time is 100 s.

sensitivity, particularly for Si and above (Table 3). A1 sensitivity has also been improved but remains rather imprecise at the present time. The enhancement in response has been achieved by adjusting the orientation of the X-ray capillary optic to allow greater transmission of the primary bremmstrahlung, thereby providing improved excitation to the lower atomic number elements without detriment to the heavier elements. Additional improvements arise from developments with the detector preamplifier and counting electronics that allow higher count rates without significant loss of energy resolution.

Fig. 4. Comparison between WD-XRF and ITRAX (second generation) data for a laminated sediment from the Newport Deep (NPD7), Severn Estuary, UK. The dots represent contiguous samples 1cm thick analysed using WD-XRF and the continuous red line is the ITRAX profile obtained on the split core.

I.W. CROUDACE ET AL.

58

Table 4. XRF data determinedfor pelletized USGS reference materials using the ITRAX (second-generation) system

A1203 SiO2 K20 CaO TiO2 MnO Fe203 Ni Cu Zn Rb Sr

MAG-I Measured wt%

MAG-1 Recommended wt%

SGR-1 Measured wt%

SGR-1 Recommended wt%

16.6 51.1 5.6 1.45 0.79 0.11 7.23 0.005 0.004 0.014 0.017 0.015

16.37 50.36 3.55 1.37 0.75 0.098 6.8 0.0053 0.0030 0.0130 0.0149 0.0146

5.5 27.6 2.7 8.53 0.23 0.08 3.09 0.004 0.0076 0.0071 0.0093 0.043

6.52 28.24 1.66 8.38 0.264 0.034 3.03 0.0029 0.0066 0.0074 0.0083 0.042

MAG-1, muddy marine sediment with low carbonate content from the Wilkinson Basin, Gulf of Maine. SGR-1, petroleum and carbonate-rich shale from the Mahogany Zone, Green River Formation.

X R F analysis software A refined ' X R F f u n d a m e n t a l p a r a m e t e r s ' m o d e l is n o w used in Q-Spec to c o m p e n s a t e for inter-

e l e m e n t effects a n d to enable the c o n v e r s i o n o f c o u n t rates to c o n c e n t r a t i o n (see Table 4). T h e user is c u r r e n t l y responsible for p r o v i d i n g q u a n tification a n d is able to use suitable reference

Fig. 5. Integrated optical, radiographic and XRF data (Prototype ITRAX) for a laminated sediment core section from the Newport Deep (NPD7), Severn Estuary, UK. Note that levels of the anthropogenic pollutant elements Zn, Pb and Cu are higher in the upper 275 mm. The major elements Ca, K, Fe and Mn and trace elements Sr and Rb show much less variation throughout the core. The green line indicates a valid ITRAX measurement while the red broken lines are user-defined reference lines. Image generated using the program ItraX-Plot.

THE ITRAX CORE SCANNER samples to facilitate this process. This system uses the intensity of the scattered (Compton and Rayleigh) lines from the X-ray tube as a means of normalization, making the quantitative result independent of tube ageing or any other factor affecting the primary beam intensity. The ratio of the Compton and Rayleigh scattered intensities are used to estimate the variation of the average atomic number in the sample.

59

vertical reference lines. ItraX-Plot also allows temporary removal of suspect or meaningless data prior to final presentation but the original data are never modified. The images can be output or saved in high-quality formats as .PDF, .JPG or .BMP. Cox Analytical also produce an in-house package called RediCore for inspecting and displaying the downcore profiles.

Quantification issues in micro-XRF analysis Data visualization software The efficient visualization of the analytical data and images from the ITRAX is an important part of the evaluation process, and the National Oceanographic Centre has produced a flexible package called ItraX-Plot for examining and manipulating the core data (see, for example, Figs 5 & 6). The software opens the ITRAX files and allows the user to manipulate, re-size and optimize image files (optical and radiographic), as well as plotting up to 10 X R F elemental profiles at a time. Other options include adjustment of scaling, data smoothing, application of elemental divisors, inclusion of regions of interest, and the addition of horizontal and

The ITRAX geochemical data are normally output as counts and can be considered semiquantitative in nature, and as such need to be interpreted with caution. Errors may arise due to poor peak discrimination in the X-ray spectra, porosity changes, compaction or grain-size/ shape-related artefacts (recorded for K and Sr) and low count rates. Invalid data may be recorded when the X-ray detector is not in the correct position, particularly when the cut core surface is uneven or shows sudden variability from crack-related effects. Careful study of variation in the element integral profiles, the Compton scatter integral and the detector-sediment distance (validity) index can aid in identifying

Fig. 6. Comparison between WD-XRF data (red bars at 1cm resolution; scale at bottom) and ITRAX scan elemental integral profiles (Prototype ITRAX; continuous black lines; 200 lam resolution; scale at top) for Mediterranean sediment core (LC21 S#11A) containing the sapropel S1.

60

I.W. CROUDACE ET AL.

such invalid data. The ItraX-Plot data visualization software allows for ready examination of the data at various scales to consider whether they are spurious. Quantitative ITRAX X-ray microanalysis of natural samples can be successfully carried out, but usually involves a post-processing refinement session followed by a full batch re-analysis. Quantification uses 'XRF fundamental parameters' calculations and assumes compositional and physical homogeneity for the measured samples, and any deviation from this ideal state will introduce errors. The effectiveness of the quantitation will be impaired for some sample types where the small excitation volume, medium-coarse grain sizes and mineral effects combine. The analysis of natural samples containing discrete medium-coarse fragments (e.g. quartz, feldspar, rutile, ilmenite, zircon, biogenic carbonate clasts) may lead to analytical inaccuracies and will be most significant when measuring low Z elements or low energy X-ray energies. Where the unknown and reference samples have similar granular structures and elemental and mineral compositions then these problems can be reduced. Another effect requiring consideration is where pore-water solutes can dry out on a sediment surface, which will lead to spurious count rates. Frequently, this effect can be remedied by careful removal of a thin surface layer (using a glass slide, for example) prior to analysis. In spite of the potential problems noted above, the elemental profiles obtained using the ITRAX show that useful semi-quantitative and quantitative data can be obtained for a good range of elements. I T R A X m e a s u r e m e n t accuracy and precison

The analytical data shown in Figure 4 and Table 4 demonstrate that reasonable accuracy can be achieved. The poorer K20 data (Table 4) found when quantifying XRF data for the USGS reference samples may reflect unresolved interelement or particle-size effects. The next stage in the development of the ITRAX system is to develop and refine the conversion of count data to concentration data. The measurement precision of the ITRAX can be simply determined by calculating the square root of the count integral. Another means of estimating precision is from the general smoothness or spikiness of the XRF elemental profiles (e.g. Figs 4-6). Some variations reflect statistical noise or changes caused by compositional boundaries or cracks.

Examples of ITRAX applications The capability of the ITRAX core scanner has been evaluated by investigating diverse sediment types from wet, unconsolidated materials such as cores from the eastern Mediterranean Sea containing sapropels (Thomson et al. 2006) to heavy metal-polluted estuarine sediment. Other sample types such as peat, organic-rich lake sediment and mineralized rocks have also been examined. Some of the compositional parameters that can aid in routine sedimentological or lithostratigraphic analysis of marine sediment cores are summarized in Table 5. Further details of such information can be found in Rothwell et al. (2006) and Thomson et al. (2006). Two specific examples of ITRAX applications are given here. Polluted Severn E s t u a r y sediment

A subtidal sediment core collected from the Newport Deep, Severn Estuary (Fig. 5) was investigated to examine records of metal pollution from different industrial sources. Earlier studies by Allen & Rae (1986) and Allen (1988) had defined chemozones in sediment cores from the Severn Estuary. A recent study by the first author examined sedimentological, radiometric and geochemical data from the Newport Deep to determine the record of anthropogenic inputs in several sediment cores. The ITRAX was used to compare elemental profiles with those obtained using a Philips MAGIX-Pro WD-XRF spectrometer, also based at the National Oceanography Centre, Southampton. The latter method, though delivering highquality quantitative data, involved subsample collection and took approximately 8-10 days overall to acquire the final data. By contrast, ITRAX analysis only involves halving the core followed by measurement, which may take approximately 2 days if trace-element data are required (1 m core scanned at 200 ~m resolution with an XRF measurement time of 100s per increment). The ITRAX radiographs acquired for Newport Deep sediment cores NPD7 (Fig. 5) show distinctive mm-scale layering caused through variations in carbonate, clay and sand. Some elemental profiles show steps, very similar to those previously seen using the Philips WD-XRF. These steps are clearly evident for Cu, Zn and Pb in the ITRAX profiles (but are also evident for a wider set of elements with the WD-XRF, i.e. P, Cr, As, Sn, I and S) and are caused by anthropogenic inputs of heavy metals to the estuary over the last 50 years or more. Additional

THE ITRAX CORE SCANNER

61

Table 5. Examples of l T R A X output parameters that aid in sedimentological or lithostratigraphic analysis

Property measured

ITRAX detection efficiency

Comment on property

Compton scattering MoKinc

High

9 Relates inversely to the mean atomic number 9 Will vary with mineralogical composition, water and organic carbon content 9 Inflections may occur at bed boundaries 9 Will vary with sediment packing density

Ca/Fe

High

9 Indicative of biogenic carbonate:detrital clay ratio 9 May show strong correlation with sedimentary units 9 Ca/Fe profile is a good proxy for sediment grading and for assessing source relationships 9 Can distinguish foraminifer- or shell-rich layers

Sr/Ca

High

9 Enhanced Sr may indicate the presence of high-Sr aragonite which requires a shallow-water source 9 Affected by sediment packing/porosity and grain-size/shape variations

K/Rb

Moderate

K is commonly associated with detrital clay and may be enhanced in turbidite muds Potentially unreliable parameter as sea-water C1 atoms will absorb potassium X-rays, so apparent high K may reflect increased porosity

Zr/Rb Ti/Rb

Moderate

9 Zr and Ti are high in heavy resistate minerals and may be enhanced in turbidite bases Useful as sediment-source/provenance indicators

Si

Moderate-low

9 Important terrigenous or productivity indicator Normalization using detrital divisor can distinguish terrigenous or productivity origin May be useful as a sediment-source and provenance indicator

Fe/Rb Fe/Ti

Good

9 Commonly shows grain-size-related fractionation effects 9 Fe mobilized during redox-related diagenesis and elevated Fe commonly seen in oxic, or formerly oxic, parts of sediment Rb is an element commonly associated with detrital clay and may be enhanced in turbidite muds

Cu/Rb Cu/Ti

Moderate

9 Sharp Cu peaks are largely of diagenetic origin

As

Moderate

9 Commonly an indicator of pyrite which may be detrital or authigenic in origin

Mn/Ti

Good

9 Good indicator of redox-related diagenesis

Ba/Ti

Low-moderate

9 Important productivity indicator

Br/C1

Moderate-low

9 For marine sediments a constant ratio implies sea-water ratios. High ratios may indicate organic-rich layers as Br and S are high in organic-rich sediments

s/ci

The property refers to an element ratio or peak area integral. See Rothwell et al. (2006) and Thomson et al. (2006) for more specific discussion of the above parameters. cores e x a m i n e d from the Severn Estuary confirm the findings seen in N P D 7 .

M e d i t e r r a n e a n s e d i m e n t with s a p r o p e l A well-characterized subcore held in the BOSC O R F archive (Section l l A of core LC21), previously examined by M e r c o n e et al. (2000,

2001), was scanned (Fig. 6). This core contains an example of the organic-rich sedimentary units (sapropels) that form periodically in the eastern M e d i t e r r a n e a n basin. Sapropels are visually distinct because of their dark colour, but the I T R A X X - r a d i o g r a p h also reveals coincident physical property changes that result mainly from the lower sediment density and

62

I.W. CROUDACE E T AL.

high pore-water content in sapropels. I T R A X elemental profiles were compared with wavelength dispersive X R F data from discrete 1 cm samples taken through the most recent sapropel (S1). While recognizing that the measured X R F element integrals from the I T R A X do not have an exact constant relationship with element concentration over changing sediment types, the data show a good comparison (Fig. 6). There are plans to quantify the I T R A X data routinely. The I T R A X data (discussed in detail by Thomson et al. 2006) show that higher Ba/ Ti and Br/C1 ratios correspond with the high Corg content in the visual and oxidized sapropel. The thinning of the original sapropel thickness by post-depositional oxidation is revealed from Mn/Ti and Cu/Ti ratios, pyrite authigenesis in the residual visual sapropel from Fe/Ti and S/ C1 ratios and the As integral, and aragonite formation in and around the sapropel from the Sr/Ca ratio.

Conclusions The I T R A X multi-function core scanner has the capability to rapidly and conveniently acquire data for three important physical and chemical properties from split-sediment cores. A scanner run will provide a high-quality optical image file, a 16-bit X-radiography image file and multiple X-ray spectral files, and an immediate end-of-run analytical summary. Further postrun re-processing of X-ray is straightforward. This combined acquisition of physical and chemical data is of considerable attraction where initial non-destructive characterization of lake and marine sediment and rock cores is required. The I T R A X may be an acceptable substitute for traditional analytical methods in some cases and for some elements, but more generally it will be useful in acquiring indicative data for cores held in repositories before destructive sampiing for further detailed investigations. X R F scanner systems cannot be expected to deliver a quality of data comparable to that of systems such as W D - X R F s because of the small excitation volume used, the air path, and the effects of mineralogy, particle size, porosity and water content variations. Their high-resolution capability, however, considerably exceeds that achievable by conventional sampling methods and may lead to specific new applications (e.g. ombrotrophic peat analysis, examination of mineralized rock sections). The I T R A X instrument also provides a convenient platform for adding other sensors (e.g. Fourier Transform Infra Red spectroscopy, spectrophotometry)

that may provide helpful data such as have been used to extract palaeoenvironmental information from cores (Giosan et al. 2002a,b; Berg & Jarrard 2003). I.W. Croudace and R.G. Rothwell, and other colleagues and former colleagues at NOC, particularly Professors C. German, J. Thomson and P. Weaver, are grateful to the UK Office of Science and Technology via the Natural Environment Research Council for providing funds to purchase X-ray analytical instruments that included the ITRAX. The realization of the ITRAX core scanner is due to a co-operative venture between COX Analytical (A. Rindby, B. Stocklasser, P. Engstrom, S. Norder and J. Rudolfsson) and NOC (I.W. Croudace and R.G. Rothwell). We thank NOC colleagues R. Pearce for the loan of laminated sediment blocks and K. Davis for improvements to the technical artwork.

References ALLEN, J.R.L. 1988. Modern-period muddy sediments in the Severn Estuary (southwestern UK): a pollutant-based model for dating and correlation. Sedimentary Geology, 58, 1-21. ALLEN, J.R.L. & RAE, J.E. 1986. Time sequence of metal pollution, Severn estuary, southwestern UK. Marine Pollution Bulletin, 17, 427-431. BERG, M.D.V. & JARRARD,R.D. 2004. Cenozoic mass accumulation rates in the equatorial Pacific based on high-resolution mineralogy of Ocean Drilling Program Leg 199. Paleoceanography, 19, 1-12. PA2021, doi: 10.1029/2003PA000928. BERGSTROM, U., LINDEBERG, J. & RINDBY, A. 2001. Batch measurements of wood density on intact or prepared drill cores using X-ray microdensitometry. Wood Science and Technology, 35, 435-452. CROUDACE, I.W. & WILLIAMS-THORPE,O. 1988. A low dilution, wavelength-dispersive X-ray fluorescence procedure for the analysis of archaeological rock artefacts. Archaeometry, 30, 227-236. CROUDACE, |.W. & GILLIGAN, J. 1990. Versatile and accurate trace element determinations in iron-rich and other geological samples using X-ray fluorescence analysis. X-ray Spectrometry, 19, 117-123. GIOSAN,L., FLOOD,R.D. & ALLER,R.C. 2002a. Paleoceanographic significance of sediment color on western North Atlantic drifts: I. Origin of color. Marine Geology, 189, 25-41. GIOSAN, L., FLOOD, R.D., GRtJTZNER, J. & MUDIE,P. 2002b. Paleoceanographic significance of sediment color on western North Atlantic Drifts: II. Late Pliocene-Pleistocene sedimentation. Marine Geology, 189, 43-61. JANSEN, J.H.F., VAN DER GAAST, S.J., KOSTER, B. & VAARS, A.J. 1998. CORTEX, a shipboard XRFscanner for element analyses in split sediment cores. Marine Geology, 151, 143-153. JANSSENS, K., ADAMS, F. & RINDBY, A. (eds). 2000. Microscopic X-Ray Fluorescence Analysis. Wiley, Chichester.

THE ITRAX CORE SCANNER JENKINS, R. 1999. X-Ray Fluorescence Spectrometry, 2nd edn. Wiley, Chichester. LECHNER, P., FIORINI, C. ET AL. 2001. Silicon drift detectors for high count rate X-ray spectroscopy at room temperature. Nuclear Instruments and Methods, 458A, 281-287. MERCONE, D., THOMSON, J., ABu-ZrzD, R.H., CROUDACE, I.W. & ROHLING, E. 2001. High-resolution geochemical and micropalaeontological profiling of the most recent eastern Mediterranean sapropel. Marine Geology, 177, 25-44. MERCONE, D., THOMSON,J., CROUDACE,I.W., SIANI,G., PATERNE, M. • TROELSTRA, S.R. 2000. Duration of S1, the most recent eastern Mediterranean sapropel as indicated by AMS radiocarbon and geochemical evidence. Palaeoceanography, 15, 336-347. RINDBY, A., ENGSTROM, P. & JANSSENS,K. 1997. The use of a scanning X-ray microprobe for simultaneous XRF/XRD studies of fly-ash particles. Journal of Synchrotron Radiation, 4, 228-235. ROTHWELL, R.G., HOOGAKKER,B., THOMSON,J., CROUDACE, I.W. & FRENZ, M. 2006. Turbidite emplacement on the southern Balearic Abyssal Plain (western Mediterranean Sea) during Marine Isotope Stages 1-3: an application of ITRAX XRF scanning of sediment cores in lithostratigraphic analysis. In:

63

ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 79-98. STRELI, C., WOBRAUSCHEK,P., PEPPONI, G. & ZOEGER, N. 2004. A new total reflection X-ray fluorescence vacuum chamber with sample changer analysis using a silicon drift detector for chemical analysis. Spectrochimica Acta, B59, 199-203. THOMSON, J., CROUDACE, I.W. 8r ROTHWELL, R.G. 2006. A geochemical application of the ITRAX scanner to a sediment core containing eastern Mediterranean sapropel units. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 65-77. TSUJI, K., INJUK, J. & VAN GRIEKEN,R. (eds). 2004. Xray Spectrometry: Recent Technological Advances. Wiley, Chichester. V1NCZE, L., JANSSENS, K., ADAMS, F., RINDBY, A. & ENGSTROM, P. 1998. Interpretation of capillary generated spatial and angular distributions of Xrays: theoretical modelling and experimental verification using the European Synchrotron Radiation Facility Optical beam line. Review of Scientific Instruments, 69, 3494-3503.

A geochemical application of the ITRAX scanner to a sediment core containing eastern Mediterranean sapropel units J. T H O M S O N ,

I. W. C R O U D A C E

& R. G. R O T H W E L L

N a t i o n a l Oceanography Centre, Empress Dock, S o u t h a m p t o n S 0 1 4 3 Z H , U K Abstract: The ITRAX micro-X-ray fluoresence (XRF) core scanner has been applied in a

sediment geochemistry investigation. The core sections selected contain examples of the organic-rich sedimentary units (sapropels) that form periodically in the eastern Mediterranean basin. Sapropels are visually obvious from their dark coloration, but the ITRAX X-radiograph also reveals physical property changes that result mainly from the high pore-water content of sapropels. A consideration of wavelength-dispersive XRF data from discrete samples of the most recent sapropel (S1) was made along with the set of elements reported by the ITRAX instrument's energy-dispersive XRF system over core sections containing S1. This allowed selection of a suite of eight inter-element ratios or element integrals through which characteristic features of sapropel development and geochemistry were revealed. While recognizing that the measured XRF element integrals from the ITRAX do not have an exact constant relationship with element concentration over changing sediment types, this combination of ratios provides significant information for geochemical interpretation. These include evidence for: (i) the presence of high Corg contents in the visual sapropel from Ba/Ti and Br/C1 ratios; (ii) a thinning of the original sapropel thickness by post-depositional oxidation from Mn/Ti and Cu/Ti ratios; (iii) pyrite authigenesis in the residual visual sapropel from Fe/Ti and S/C1 ratios and the As integral; and (iv) aragonite formation in and around the sapropel from the Sr/Ca ratio. These same ratios were then used to interpret ITRAX data from a deeper section of the same core containing the older sapropel $3, where the same characteristics, including the relict post-depositional oxidative thinning of the original unit, could be identified with only minor differences of detail. Directions of supply of Fe, As and Cu into the sapropels could be inferred from profile shapes.

The often-continuous record of past environmental conditions and changes that is contained in deep-sea sediments makes them a valuable resource for palaeoceanographic and palaeoclimatological research. Detailed sampling for specialist investigations at marine core repositories is greatly facilitated if logging data are available that profile some aspect of the changes in sediment characteristics that occur downcore. Current logging techniques include gamma-ray attenuation, P-wave attenuation and X-radiography that give information on changes in sediment density, and digital photography, magnetic susceptibility and X-ray fluorescence that give information on changes in sediment composition. The I T R A X instrument, developed by Cox Analytical Systems of Gothenburg in association with Southampton Oceanography Centre (SOC, renamed as National Oceanography Centre, Southampton - NOCS - 1 May 2005), is a new micro-X-ray fluorescence (XRF) and micro-X-radiograph scanner designed for rapid, automatic, non-destructive characterization of the optical, density and chemical composition variations in split sediment cores (Croudace et al. 2006).

Compositional data from energy-dispersive X R F (ED-XRF) scanners, for example the C O R T E X system (Jansen et al. 1998), have mainly been applied to: 9

9

studies of climatologically driven cyclicities in sediment deposition that are reflected, for instance, in CaCO3 or Fe content fluctuations over time (e.g. Peterson et al. 2000; Palike et al. 2001); or sedimentological applications, where features such as ash layers, turbidite units or ice-rafted deposits may be recognized through their exotic compositions or their localized or organized bed characteristics (e.g. Hebbeln & Cortes 2001; Richter et al. 2001; Rothwell et al. 2006).

A further obvious application is to diagenetic geochemical studies, where the key is identification of unusually high contents of redox-sensitive elements in arrangements that are characteristic of some diagenetic process. Initial experience with the I T R A X system at the British Ocean Sediment Core Research Facility (BOSCORF) at SOC in such a diagenetic application is presented here. A wealth of published geochemical

From: ROTHWELL,R.G. 2006. New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 65-77.0305-8719/06/$15.00 9 The Geological Society of London 2006.

66

J. THOMSON E T AL.

i n f o r m a t i o n and interpretation is n o w available on eastern M e d i t e r r a n e a n sapropels, particularly on the most recent interval, sapropel S1. An example of this unit was first examined to ascertain which elements m e a s u r e d by the I T R A X core scanner yield i n f o r m a t i o n on the characteristic elemental distributions that are k n o w n to occur in S 1 from data m e a s u r e d on discrete samples by conventional techniques. Profile scans of the same elements from the next oldest sapropel ($3) that occurs deeper in the same core were then examined in the light of this experience with the S1 profile scans.

Material and methods Core LC21 is a 10cm-diameter giant (Calypso) piston core collected from a topographic high to the west of Karpathos Basin, NE of Crete, at 35~ 26~ in 1522m water depth (Rothwell 1995). The core is 13.5 m long and is stored at BOSCORF as 11 sections of 1.5 or 1.0m nominal length. Sapropel S I in particular from this core has been comprehensively investigated (Mercone et al. 2000, 2001; Casford et al. 2001, 2003; Rohling et al. 2002), and can be considered as a reference core for the Aegean area of the eastern Mediterranean during S1 times. Core SL104 is a 4.8 m-long, 11 cm-diameter gravity core collected SE of Crete on Meteor cruise M51-3 at 27~ in 2155m water depth 34~ (Hemleben et al. 2001). The subsections of the core examined in this work were contained in polyvinyl chloride (PVC) U-channels of 1 m length by 2 x 2cm cross-section that are often used in logging magnetic parameters (Weeks et al. 1993). A more detailed description of the prototype ITRAX core scanner used is presented elsewhere (Croudace et al. 2006), and only an outline description is presented here. The instrument is computer controlled and designed for automatic, high-resolution logging of sections of split-sediment cores up to 10cm in diameter and 1.8 m in length. Core sections are first scanned to ascertain whether a constant working distance for the X-ray detector can be achieved, using a laser distance finder to map the topography of the cut core surface. Regions where the detector will be retracted to avoid contact with the core surface are recorded. Simultaneously, a digital line camera captures a photographic image of the whole core surface. This process requires approximately 2 min for a core section I m long. After photography, the core is automatically returned to the zero position and a stepper motor drives the scan at a line resolution that can be as fine as 100 gm. At each measurement position a 2 cm-wide high-resolution X-radiograph is recorded using a charge-coupled device (CCD) line camera and an ED-

XRF spectrum is recorded using a Si-drift chamber detector. The time for each measurement, selected by the operator and dependent on the element detection sensitivity required, is typically 600-700ms and 1040s, respectively. The X-ray beam used to irradiate the core section is generated from a 3 kW Mo target (typically run at 55kV and 50mA) that is focused through a flat capillary waveguide. Each ED-XRF spectrum recorded is stored along with header data, but spectra are also deconvolved in live time to build up profiles of peak area integrals for individual elemental and the coherent (Rayleigh) and incoherent (Compton) X-ray scattering peaks. These elemental peak areas are roughly proportional to the concentrations of major and trace elements present in the sediment. For the core LC21 and SLI04 sections, scans were made every 0.5 mm with an excitation slit dimension of 20 mm x 200 gm and XRF analysis times of 35 or 45 s per measurement, so that processing each 1 m of core required 20-25 h.

Results and discussion Inorganic geochemistry

of sapropels

The sediments deposited in the eastern Mediterr a n e a n basin are mostly Corg and sulphide-poor clayey nannofossil oozes (subsequently referred to as 'marls'), but sapropels, sharply-defined dark sedimentary units with high contents of organic matter (Corg) and sulphide, occur as discrete beds within the sequence (Calvert 1983; W e h a u s e n & B r u m s a c k 2000; Calvert & F o n tugne 2001). Sapropels form only at times of m a x i m u m s u m m e r insolation in the n o r t h e r n hemisphere, with a frequency of 23 ka mainly set by variations in the Earth's orbital precession (Rossignol-Strick 1983; Hilgen 1991; Lourens et al. 1996; Tuenter et al. 2003). E a c h sapropel has a d u r a t i o n o f only a few t h o u s a n d years, and not every insolation m a x i m u m produces a sapropel (van Santvoort et al. 1997; Bard et al. 2002). Inorganic geochemical investigations of the compositional differences between the enclosing marls and sapropels have d e m o n s t r a t e d that m a n y redox-sensitive elements show large systematic enrichments in sapropels (e.g. Calvert 1983; T h o m s o n et al. 1995; Nijenhuis et al. 1999; W e h a u s e n & Brumsack 2000; Calvert & F o n t u g n e 2001; M e r c o n e et al. 2001). Sapropel S1 formed between 10 and 6 14C ka Bt' ( M e r c o n e et al. 2000). M e r c o n e et al. (2001) investigated its inorganic geochemistry in core LC21 with wavelength-dispersive X R F (WDX R F ) data and coulometric Corg and CaCO3 analysis (Figs 1 & 2). As in most sapropels, the CaCO3 content of sapropel S1 is lower than

ITRAX SCANNING OF SAPROPEL UNITS

67

Fig. 1. CaCO3, Corg and S concentration (wt%) and Al-normalized mass ratios of Si, Ti, Zr, Mg, Ba, Fe and Mn v. depth in core LC21 (after Mercone et al. 2001). All data are from WD-XRF except Corg and CaCO3 which were measured by coulometry. The depth axis is cm below sea floor, and section 11 of the core containing sapropel S1 is at 138.5-219.5 cm below sea floor. The lightly shaded zone indicates sapropel S1. those of the marls that enclose it, and, because biogenic CaCO 3 is a relatively pure phase, the concentrations of elements mainly present in the detrital aluminosilicate phase (e.g. Si, A1, Ti, K, Rb, Zr) must increase in the sapropel as a direct consequence of the decrease in CaCO3 content. Geochemical data are therefore often expressed as a ratio to A1 or some other element that is taken to be characteristic of the detrital phase in order to overcome this closed sum effect (Rollinson 1993; van der Weijden 2002). In contrast to the essentially constant element/A1 ratios found for the detrital elements (Fig. 1), many redox-sensitive elements when normalized to A1 show distinct enrichments within sapropel S 1 compared with the lower (detrital) levels observed in the enclosing marls (Figs 1 and 2). The S1 sapropel in core LC21 has two distinct lobes, although the central region still has a much higher Corg content than the marls above and below. Besides Corg , the elements showing the clearest enrichments in these two lobes are the Corg or sulphide-associated elements S, Fe, As, Mo and V. Other chalcophile elements (e.g. Cu, Ni, Pb and Zn) also show enrichments in the lobes of the sapropel,

although the magnitude of these enrichments in excess of the detrital element/A1 ratio is less marked than for As, V and Mo. In the enclosing marls, the concentrations of As and Mo are close to or below the detection limit of the W D - X R F instrument used, but both elements are readily detectable within the sapropel. The relationship between the concentration of an element in the sediment (as measured for example by W D - X R F ) and its peak area integral measured by E D - X R F excitation will not necessarily be constant along a core section of wet sediment (Jansen et al. 1998). Fluctuations in sediment mineralogy, particle size or water content, for example, will all influence the mean atomic number of the sediment under the X-ray excitation beam and, hence, affect the X-ray absorption and Compton and Rayleigh scattering. Thus, the higher water content of sapropels causes an increase in Compton scattering compared with the enclosing marls, and this alters the sensitivity of the system for different elements by different amounts. While recognizing the problems presented by such fluctuations in detection efficiency, a selection of parameters was made based on a consideration of: (i) the data of

68

J. THOMSON ET AL. Corg (wt.%) 1

VIAl (wt.)

2 0.001

100 !lL~lt~Ll~--I____

0.005 0.0, O0 ' ~ I ~ __~-

_ _ _

120 140 z (cm)160

Cu/AI (wt.)

Ni/AI (wt.)

0.0015 0.000

t

I

Zn/AI (wt.)

0.004 0.000

I h ~~_1

_-____l

0.001

0.002

LC21

180 2oo

k

_-

22o Mo/AI (wt.) 0.0000 100 Ll

120 140 Z

(cm)160 180

0.0004 0.0000 i i I I

CI (wt.%)

As/AI (wt.)

I__

0.0008

0

~'''I''

3

Br (ug/g) 0

150 0

I (ug/g)

25

50

'

i

200

220 --

f

Fig. 2. Corgand C1 (wt%), Br and I (pgg-1), Al-normalized mass ratios V, Cu, Ni, Zn, Mo, As v. depth in core LC21. All data except Corg are from WD-XRF. The depth axis is in cm below the sea floor. Lightly shaded zone indicates sapropel S1.

Figures 1 and 2 along with those elements that were recorded by the ITRAX instrument for the archive section of the core from which the data of Figures 1 and 2 were gathered (Table 1); and (ii) those elements recorded by the ITRAX instrument for the two sections of core SL104 containing sapropels (Fig. 3). The ITRAX response (scan integrals) to the concentration of an element in sediments varies greatly between different elements (Table 1), and some elements with both low mean concentrations and low mean scanner integrals (e.g. S, As) prove critical for interpretation of the localized elemental enrichments in the sapropel. On the following reasoning, seven element integral ratios and one direct integral from the element scans were selected to identify critical sapropel features. Ba/Ti integrals ratio. High Ba/AI levels have been shown to occur in sapropels where Corg values are high, due to the presence of biogenic Ba associated with organic matter, and these high Ba/A1 values appear to be preserved in sapropels even if Corg is oxidized at a later stage (Thomson et al. 1995, 1999; van

Santvoort et al. 1996, 1997). Aluminium was not detected efficiently with the ITRAX configuration used, and in the absence of A1 data the possible alternatives as the element for detrital aluminosilicate normalization among the elements detected by ITRAX (Table 1) are Ti, K and Rb. Although Lourens et al. (2001) have demonstrated that the Ti/A1 ratio in Pliocene eastern Mediterranean sediment sections containing sapropels shows a remarkable cyclicity with lower values in sapropels compared with the enclosing marls, this effect does not appear so marked in more recent sapropels (Fig. 1). Ba/Ti ratios were therefore selected on the expectation that they would behave similarly to Ba/A1 ratios. Br/Cl integrals ratio. Marked enrichments in Br content are found in sapropels (Ten Haven et al. 1987; Mercone et al. 2001) (Fig. 2), partly due to an enrichment of Br with Corg, and partly due to the higher porosity and, hence, higher sea-salt content in the pore waters of the sapropel compared with the enclosing marls. The latter factor also causes relatively high C1 contents in sapropels.

ITRAX SCANNING OF SAPROPEL UNITS

69

Table |. Mean response o f the I T R A X system over core LC21 section 11 compared with mean W D - X R F concentrations determined in individual samples from the same core section. The I T R A X data are the means o f 1590 element peak area integrals, each collected over a width o f 200 #m for 40 s at a spacing o f O.5 mm over 81 cm, while the W D - X R F data are 23 analyses o f 1 cm-thick samples. Note that the concentrations o f redox-sensitive elements in particular are highly variable in this section (Figs 1 & 2)

Element

ITRAX Mean integral

ITRAX SD on mean

ITRAX SD/mean%

WD-XRF Mean (gg g-l)

WD-XRF SD on mean

WD-XRF SD/mean%

Fe Ca Sr Zr Br Ni K Rb Co Ti Mn Zn Cu C1 Cr As V Pb Ba S

167646 96203 54567 28732* 3752 3600 3472 3171 3142 2592 2284 1055 973 923 622 483 127 122 95 24

19 087 13 549 5648 860* 1317 414 812 482 847 336 5243 173 214 197 213 543 113 87 63 64

11 14 I0 3* 35 11 23 15 27 13 230 16 22 21 34 112 89 71 67 264

28 418 171414 750 78 104 211 2402 n.a.* n.a.* 2285 1676 66 71 22 578 197 10 128 12 316 4343

2498 8516 68 6 33 19 786

9 5 9 8 32 9 33

135 1668 3 I0 7280 10 12 46 2 135 2313

6 99 5 14 32 5 117 36 14 43 53

Si A1 Mg

n.dfl n.d. ~ n.d. *

132028 41 739 29848

7228 2321 1265

5 6 4

*Prototype ITRAX Zr data compromised by a detector contamination. n.a.*: not available. n.d.~: not detected by the ITRAX system with the Mo target X-ray tube used.

9

9

-.

,

The Ba/C1 ratio is chosen to reveal the presence of additional Br associated with Corg in excess of the constant Br/CI sea-salt ratio. M n / T i integrals ratio. High Mn/A1 values in two separated peaks are commonly observed above the visual S 1 unit, as a result of oxidation of Mn 2+ in the sediment pore waters or sea water to MnOx by increased bottom water 02 levels that become available at the end of sapropel formation (Thomson et al. 1995, 1999; Calvert & Pedersen 1996; van Santvoort et al. 1996). This parameter is important in defining the extent of postdepositional oxidation of sapropel. C u / T i integrals ratio. This ratio was initially included because Cu, like several other chalcophile elements, generally shows an increase in sapropels (Fig. 2). The ratio also turned out to be a useful marker ofpost-depositional oxidation, as discussed below. F e / T i integrals ratio. Pyrite (FeS2) forms in sapropels (Passier et al. 1996, 1999), and the enrichment of Fe content causes an

increase in the Fe/Ti ratio over the assumed constant Fe/Ti detrital value. S / C l integrals ratio. The S/C1 ratio is chosen to reveal the presence of additional S associated with pyrite or Corg (Passier et al. 1999) in excess of the constant S/C1 sea-salt ratio. A s integral. Arsenic is strongly incorporated into FeS2 (Peterson & Carpenter 1986; Huerta-Diaz & Morse 1992). Arsenic data have not been normalized to Ti because As is generally close to its X R F limit of detection except where pyrite occurs (cf. Fig. 2). St~Ca integrals ratio. This ratio is monitored because the Sr/Ca ratio in sapropel units has been reported to be consistently higher than in the enclosing marls (Calvert 1983). The substantial CaCO3 content in eastern Mediterranean sediments (e.g. generally 4050%, Fig. 1) means that an increased Sr/Ca ratio requires the presence of some phase with an unusually high Sr/Ca ratio. This has been identified as aragonite that has been postulated to form as a consequence

70

J. T H O M S O N E T AL.

ITRAX SCANNING OF SAPROPEL UNITS of the alkalinity produced by sulphate reduction diagenesis within sapropels (Thomson et al. 2004). Inorganic geochemistry o f sapropel S1 as revealed by I T R A X in section 5, core S L I 0 4 Sapropel S1 is present as a dark band at 130230mm in the uppermost section 5 of core SL104. When submitted to ITRAX analysis, the dark band in the visual image is revealed by the X-radiograph to have a lower sediment density than the remainder of the section (Figs 3 & 4). This lower density of the sapropel sediments is due to their increased porosity, as revealed by the increased CI content from the increased pore-water content per unit sediment volume, and is the cause of increased Compton (incoherent) scattering in the sapropel (Fig. 4). This may be the reason for the local effect seen in the K/Ti ratio that was expected to be approximately constant from WD-XRF data on sapropel S1. High values of the Ba/Ti and Br/CI ratios are present through the 10cm-thick dark band, but while the Br/C1 ratio is high in exact coincidence with the dark coloration, high values of the Ba/Ti ratio continue for approximately 4cm immediately above it (90-130mm; Fig. 4). The Ba/Ti ratio profile also has a central minimum, similar to the two peaks in the Ba/AI ratio that follow the fluctuation of Corg in the two sapropel lobes in core LC21 (Fig. 1). Two clear peaks in the Mn/Ti ratio occur immediately above the visual sapropel, the upper of which has its maximum coincident with the point where high Ba/Ti values return to the detrital baseline, while the lower has its maximum on the upper face of the visual sapropel where Br/C1 values return to the salt baseline (Fig. 4). Such Ba, Mn and Br profiles are interpreted as due to the oxidation of Corg and sulphide from the upper reaches of the original sapropel S1 by bottom-water 02, with the upper Mn peak marking the end of sapropel formation and the lower Mn peak marking the progress of oxidation into the sapropel since that time (Thomson et al. 1995, 1999; van Santvoort et al. 1996). (Note that core

Fig. 3. Colour photograph, X-radiograph and intensity v. depth profiles for the 10 elements with the largest integrals reported by the ITRAX ED-XRF on scans along sections 5 and 3 of core SL104. The XRF data are displayed as three-point running means of 45 s XRF integrals and the depth scale is the depth in core in mm. Shaded and diagonal-hatched backgrounds indicate the zones of unoxidized (residual or visual) sapropel and the inferred oxidized sapropel, respectively.

71

LC21 (Figs 1 & 2) was selected for intensive study by Mercone et al. (2000, 2001) because the sediments of that core had accumulated so rapidly that these post-depositional oxidation effects on the S1 unit were minor or absent in that core.) The Br/C1 and Ba/Ti profiles in core SL104 section 5 are consistent with a loss of Br along with Corg on oxidation (Shimmield & Pedersen 1990), whereas Ba is retained after Corgoxidation (Thomson et al. 1995, 1999). In this example, the sapropel is interpreted as having been 14cm thick initially, with the upper 4cm oxidized since deposition and the lower 10cm remaining as the present visual sapropel. As the top of the visual S1 unit is only 13cm below the sea floor, this oxidation of the sapropel is expected to be active, and expected to continue either until the sapropel is fully oxidized or for as long as bottom-water O2 can diffuse down to the upper face of the sapropel. While the maximum of the lower Mn/Ti peak is located immediately above the visual sapropel, a clear peak in Cu/Ti occurs immediately below this Mn/Ti peak, exactly at the top of the visual sapropel (Fig. 4). In the raw data, high Mn values fall off at exactly the same level as the highest value on the upper face of the Cu peak rises to fall off over < 1 cm (Fig. 5). Other trace elements (Se, Hg and Ag) not detected by the ITRAX have also been shown previously to have large narrow peaks exactly at the top of the visual sapropel, i.e. on the anoxic side of the limit of oxidation (Mercone et al. 1999; Crusius & Thomson 2003). Such peaks are produced by the repeated mobilization of elements present at enhanced concentration in the original sapropel by oxidation with bottom-water 02 after sapropel formation, followed by their repeated immobilization by reduction in the anoxic conditions of the unoxidized sapropel. These peaks are expected to move downwards below the advancing oxidation front and continue to be augmented for as long as the oxidation front is active. The peak should remain in place in anoxic conditions as a relict marker of the oxidation process thereafter. This feature does not appear to have been reported previously in sapropel Cu profiles, probably because of the relatively small size of the Cu/Ti peak relative to the detrital background in the lower-resolution sampling undertaken for conventional analysis that is generally undertaken on sample increments no finer than 0.5-1 cm thick. The two Fe/Ti peaks in the sapropel are probably related to the two Ba/Ti peaks that should mark increases in Corg content which drives sulphate reduction, although the corroborative S/C1 and As evidence for pyrite formation is

72

J. T H O M S O N E T AL.

ITRAX SCANNING OF SAPROPEL UNITS

73

Fig. 5. Detail of the ITRAX Mn/Ti and Cu/Ti ED-XRF scans from section 5 of core SL104. These are the raw XRF count integral data plotted without smoothing. The level where Mn/Ti decreases and Cu/Ti increases (dotted line) marks the inferred location of the active oxidation front in this S 1 sapropel. now only present in the lower peak because of the effects of the inferred oxidation front (Fig. 4). The upper Fe peak with its maximum close to the upper face of the visual sapropel may have been initially present as FeS2, as in the example of Figure 1. This peak must have formed later than the lower Fe/Ti peak, and it contains a much lower As concentration. The Fe/Ti, S/C1 and As profiles all suggest that pyritization has been most intense towards the base of the visual S1 unit. In a reverse of the explanation proposed above for the Cu/Ti peak at the limit of oxidation, the shapes of the Fe/Ti and As profiles, with maximum values at the lower face of

Fig. 4. Colour photograph, X-radiograph and ratios of the ITRAX ED-XRF scan integrals for the Mo Compton scatter peak, K/Ti, Ba/Ti, Br/C1, Mn/Ti, Cu/Ti, Fe/Ti, S/C1 and Sr/Ca and the As integral as a function of depth in sections 5 and 3 of core SL104. The XRF data are displayed as three-point running means. The high water contents in the unoxidized S 1 sapropel at 130-230 mm in section 5 and in the visual $3 unit in section 3 at 2350-2430 mm cause local increases in Compton (incoherent) scattering, and the resultant changes in excitation efficiency probably affect the expected constancy of the K/Ti ratio. Shaded and diagonal-hatched backgrounds indicate the zones of unoxidized (residual or visual) sapropel and the inferred oxidized sapropel, respectively.

the sapropel and peak shapes that tail upwards into the sapropel unit, probably indicate the direction of supply of the extra Fe and As now found in FeS2. Berner (1969, 1984) has discussed how production of sulphide driven by organic matter oxidation with sulphate reacts with Fe z+ from the reduction of Fe oxyhydroxides in order to form diagenetic FeS2. The situation in the S1 unit in core SL104 section 5 is similar to Berner's (1969) case where supply of Fe z+ is sufficient to consume the HzS produced by sulphate reduction close to its locus of formation and thereby limit HzS diffusion. Passier et al. (1996, 1999) have demonstrated that a diffusive export of sulphide out of sapropel units can occur, in which case FeS2 forms below the sapropel, but in the case of the S1 unit in core SL104 the shapes of the Fe/Ti, S/C1 and As profiles suggest that most of the pyrite must have formed just inside the lower level of the sapropel. Consistent with this explanation, the lowest Fe/Ti values in this entire core section are found in the 20 cmsection immediately below the sapropel, which probably represents a loss of Fe oxyhydroxides that were reduced to Fe z+ that diffused upwards and is now immobilized in the sapropel as FeS 2. The colour of the sediment in this 20 cm-section has been altered from buff to grey, also suggesting localized reduction.

74

J. THOMSON ET AL.

The Sr/Ca profile (Fig. 4) has its maximum value close to the level of maximum pyrite formation. High Sr/Ca values in sapropels are indicative of the presence of aragonite (Thomson et al. 2004), and here these occur through both the residual and oxidized sections of the sapropel and for at least approximately 5 cm below. This is likely to reflect some diffusion of carbonate system species in solution before crystallization to aragonite.

Inorganic g e o c h e m i s t r y o f s a p r o p e l $ 3 as revealed by I T R A X

& section 3, core S L I 0 4

Sapropel $2 is rarely reported in the eastern Mediterranean sedimentary record and its formation is controversial, but sapropel $3 did form approximately 85ka ago (Bard et al. 2002). The $3 unit is clearly evident in both the visual image and the X-radiograph at 23502430mm depth in core in section 3 of core SL104 (Figs 3 & 4). High Br/CI values are exactly coincident with this 8cm-thick lower-density dark band, but high Ba/Ti values continue through the dark band and for approximately l l c m (2240-2350mm) above it. By analogy with the Br/CI and Ba/Ti profiles of sapropel S 1 in this core discussed above, post-depositional oxidation of the upper 11 cm of an $3 sapropel that was originally 19 cm thick is indicated. In this case the overall Ba/Ti profile is Gaussian in shape without the central dip in the profile seen in the S 1 sapropel (Figs 1 & 4), and the maximum Ba/Ti is up to 1.5 times that of the S1 sapropel (Fig. 4). Calvert & Fontugne (2001) have reported maximum Corg contents of 2.0 and 2.8wt% Corg in sapropels S1 and $3, respectively, suggesting that the Ba/Ti ratio may be broadly proportional to Corg content in core SL104. The sediments at the depths of section 3 in core SL104, almost 2m below the sea floor, are now in anoxic conditions. The post-depositional oxidation of this $3 unit could only have continued for as long as diffusional contact with bottom-water O2 was maintained, and all evidence of post-depositional oxidation of sapropel $3 must now be relict (van Santvoort et al. 1997). The Mn peaks at the tops of sapropels initially form as MnO x containing Mn[IV] and Mn[III], which is unstable under anoxic conditions, so that any solid-phase Mn enrichment formed above sapropel $3 is expected to have been reduced on burial. It might therefore be expected that the Mn 2+ produced by reduction would have diffused away in pore-water solution, but the largest solid-phase Mn/Ti peak in this core

section still appears to be in position above the sapropel (Fig. 4). This Mn/Ti peak is broad and much smaller than the corresponding Mn/ Ti peaks in sapropel S1, however, and lacks any evidence of the two sharp peaks that are characteristic of the active oxidation of sapropel S1 (e.g. Fig. 4). Another difference is that the excess Mn in the $3 unit appears to have penetrated slightly into the visible sapropel. One likely explanation is that diffusion of Mn 2+ away from the site of MnOx localization has been limited by sorption on to carbonate surfaces, a phenomenon that has been shown to be prevalent in carbonate-rich sediments (Boyle 1983; Thomson et al. 1986; Middelburg et al. 1987). A related possibility is the formation of a mixed Mn-Ca carbonate phase, a conversion process that forms authigenic carbonates as chemical laminations in the brackish sediments of Baltic Sea deeps after episodic oxygenation events (Huckriede & Meischner 1996). Kulik et al. (2000) have concluded that the critical requirement for Mn-Ca carbonate formation is the development of very high local pore-water Mn 2+ concentrations from MnOx reduction in the presence of alkalinity. As in the S1 profile in section 5, there is a peak in the Cu/Ti profile with a flat upper face and a downwards tail over approximately 20ram at the top of the visual sapropel at 2350mm (Fig. 4). This peak shape and position is consistent with the inferred limit achieved by post-depositional oxidation in this sapropel when active. This is neither the largest nor the only Cu/Ti peak in this section, however. Two larger peaks, both with sharp upper cut-offs, are present at 2300 mm in the oxidized sapropel and at 2660mm well below the sapropel. Localized enrichments with high Cu contents have been noted in deep-sea sediments elsewhere, although the mechanism for their formation and the source of the additional Cu remain enigmatic (Siesser 1976; van Os et al. 1993; Thomson et al. 1996). van Os et al. (1993) reported two different types of Cu enrichment, one that occurs in an association with other trace elements that appears similar to a MnOx association, and a second that occurs as thin purple-coloured bands that fade rapidly with time (oxidize?) and are apparently associated only with S. Such purple bands do not appear to have been reported hitherto from sapropels, but van Os et al. (1993) have suggested that they result from mobilization of an unidentified Cu-S species from organic-rich turbidite units. Another possibility is that Cu is mobile in an organo-S complex rather than as a Cu-S species (e.g. Skrabal et al. 2000).

ITRAX SCANNING OF SAPROPEL UNITS

Pyritization of the $3 unit is revealed by the Fe/Ti and S/C1 ratios and As values. Again, pyrite formation is most intense just above the base of the sapropel, and the mean value of the Fe/Ti ratio is again lower for tens of cm below the sapropel than above it, consistent with a loss of Fe from reduction of oxyhydroxides to supply the extra Fe found in the sapropel as FeS2 (Fig. 4). The cause of the second Fe peak in the oxidized section of sapropel $3 at 2310 mm is not clear. This peak is not associated with S or As, and it may represent solid-phase Fe(III) formed by a reduction and displacement of Mn(II) from Mn(III,IV)Ox by Fe 2+, a process discussed by Postma & Appelo (2000). The Sr/Ca profile in this section (Fig. 4) has its maximum in the visual sapropel but higher Sr/Ca values also occur approximately 30 cm above and about 20cm below, much more widely spread than the corresponding St/Ca peak in S1. It is not clear whether this represents the initial pattern of formation of aragonite in and around this sapropel, or whether some later dissolution of aragonite with an outwards diffusion of Sr 2+ has occurred.

Conclusions Several similar geochemical characteristics could be recognized in the most recent sapropel S 1 and in the older sapropel $3 through the combination of visual and X-radiograph images and XRF elemental ratio profiles obtained using the ITRAX core scanner. Element/Ti ratios of the semi-quantitative XRF integrals from the scanner appear to provide an acceptable alternative to the element/A1 ratios that are customarily used to define detrital aluminosilicate levels, and element/C1 ratios can be employed to indicate where S and Br increase markedly over their seasalt values due to pyrite formation or association with Corg , respectively. Evidence of the localized enrichments ofredox-sensitive elements associated with Corg or sulphide formation is readily identified in sapropels through the selected ratios, and profile trends reveal evidence of the oxidation that many sapropels undergo after deposition. The directions of movement of certain elements (Fe and As, Cu) during diagenesis can be inferred from profile shapes. The extra level of detail recorded by continuous high-resolution sampling with the finely collimated X-ray beam of the ITRAX provides a considerable advantage over conventional sampling and analysis, particularly for identification of narrow zones of unusually high concentrations of particular elements, such as the Cu localizations encountered in this study.

75

Data for this study were gathered with the prototype ITRAX system, but modifications and improvements to the primary X-ray excitation, X-ray detection and data processing of later ITRAX systems allow collection of quantitative data for a range of elements from A1 through to Zr (Croudace et al. 2006). Such data are achievable on favourable fine-grained sample types (e.g. compressed rock powder), but calibrated data for a larger set of elements can now also be obtained from split cores. We make the caveat that such data should not be expected to be truly quantitative in all cases, however. On unprocessed sediment in cores, XRF scanner systems cannot be expected to deliver a quality of data comparable to that of WD-XRF systems because of the small excitation volume used, the air path, and the effects of mineralogical, particle size, porosity and water content variations. The primary strength of scanner systems is their analytical rapidity and the potential of their high-resolution capability to provide insights not possible with conventional methods where cm-scale discrete samples are required to provide sufficient material for analysis. We gratefully acknowledge Professor C. Hemleben (Ttibingen University, Chief Scientist of Meteor cruise 51-3), Professor E. Rohling (NOCS) and BOSCORF for access to the core sections used in this study, Ms K. Davis (NOCS) for assistance with the illustrations, and Professor Steve Calvert (University of British Columbia) for referee comments on the original manuscript.

References BARD, E., DELAYGUE,G., ROSTEK, F., ANTONIOLI, F., SmENZl, S. & SCHRAG, D.P. 2002. Hydrological conditions over the western Mediterranean basin during the deposition of the cold Sapropel 6 (ca. 175 kyr BP). Earth and Planetary Science Letters, 202, 481-494. BERt,mR, R.A. 1969. Migration of iron and sulfur within anaerobic sediments during early diagenesis. American Journal of Science, 267, 19-42. BERNER,R.A. 1984. Sedimentary pyrite formation- An update. Geochimica et Cosmochimica Acta, 48, 605615. BOYLE, E.A. 1983. Manganese carbonate overgrowths on foraminifera tests. Geochimica et Cosmochimica Acta, 47, 1815-1819. CALVERT, S.E. 1983. Geochemistry of Pleistocene sapropels and associated sediments from the eastern Mediterranean. Oceanologica Acta, 6, 255-267. CALVERT, S.E. & FONTUGr,m, M.R. 2001. On the late Pleistocene-Holocene sapropel record of climatic and oceanographic variability in the eastern Mediterranean. Paleoceanography, 16, 78-94. CALVERT, S.E. & PEDERSEN,T.F. 1996. Sedimentary geochemistry of manganese: Implications for the

76

J. THOMSON E T AL.

MERCONE, D., THOMSON, J., CROUDACE, I.W., SIANI, environment of formation of manganiferous black G., PATERNE, M. & TROELSTRA, S. 2000. Duration shales. Economic Geology, 91, 36-47. of S 1, the most recent sapropel in the eastern MedCASFORD, J.S.L., ABu-ZIED, R. E:r AL. 2001. Mediterraiterranean Sea, as indicated by AMS radiocarbon nean climate variability during the Holocene. Medand geochemical evidence. Paleoceanography, 15, iterranean Marine Science, 2, 45-55. 336-347. CASEORD,J.S.L., ROHLING, E.J. ET AL. 2003. A dynamic MERCONE, D., THOMSON,J., CROUDACE,I.W. & TROELconcept for eastern Mediterranean circulation and STRA, S.R. 1999. A coupled natural immobilisation oxygenation during sapropel formation. Palaeomechanism for mercury and selenium in deep-sea geography, Palaeoclimatology, Palaeoecology, 190, sediments. Geochimica et Cosmochimica Acta, 63, 103-119. 1481-1488. CROUDACE,I.W., RINDBY,A. & ROTHWELL,R.G. 2006. MIDDELBURG,J.J., DE LANGE,G.J. & VANDERWEIJDEN, ITRAX: Description and evaluation of a new multiC.H. 1987. Manganese solubility control in marine function X-ray core scanner. In: ROTHWELL,R.G. pore waters. Geochimica et Cosmochimica Acta, 51, (ed.) New Techniques in Sediment Core Analysis. 759-763. Geological Society, London, Special Publications, NIJENHUIS,I.A., BOSCH,H.-J., SINNINGHEDAMSTI~,J.S., 267, 51-63. BRUMSACK,H.-J. & DE LANGE,G.J. 1999. Organic CRUSIUS,J. & THOMSON,J. 2003. Mobility of authigenic matter and trace element rich sapropels and black rhenium, silver and selenium during post-deposishales: a geochemical comparison. Earth and Planetional oxidation in marine sediments. Geochimica tary Science Letters, 169, 277-290. et Cosmochimica Acta, 67, 265-273. PALIKE, H., SHACKLETON, N.J. & ROHL, U. 2001. HEBBELN, O. & CORTES, J. 2001. Sedimentation in a Astronomical forcing in Late Eocene marine seditropical fjord: Golfo Dulce, Costa Rica. Geoments. Earth and Planetary Science Letters, 193, Marine Letters, 20, 142-148. 589-602. HEMLEBEN, C., HOERNLE, K., JORGENSON, B.B. & ROETHER,W. (eds). 2001. Ostatlantic- Mittelmeer PASSIER, H.F., MIDDELBURG, J.J., DE LANGE, G.J. & BOTTCHER, M.E. 1999. Modes of sapropel forma- Schwarzes Meet, Cruise No. 51, 12 Septembertion in the eastern Mediterranean: some constraints 28 December 2001. METEOR Berichte, Universitat based on pyrite properties. Marine Geology, 153, Hamburg, 03-1. 199-219. HILGEN, F.J. 1991. Astronomical calibration of Gauss PASSIER,H.F., MIDDELBURG,J.J., VANOS, B.J.H. & DE to Matuyama sapropels in the Mediterranean and LANGE, G.J. 1996. Diagenetic pyritisation under implication for the geomagnetic polarity time eastern Mediterranean sapropels caused by downscale. Earth and Planetary Science Letters, 104, ward sulphide diffusion. Geochimica et Cosmochi226-244. mica Acta, 60, 751-763. HUCKRIEDE, H. & MEISCHNER,D. 1996. Origin and PETERSON,M.L. & CARPENTER,R. 1986. Arsenic distrienvironment of manganese-rich sediments within butions in pore waters and sediments of Puget black-shale basins. Geochimica et Cosmochimica Sound, Lake Washington, the Washington coast Acta, 60, 1399-1413. and Saanich Inlet, BC. Geochimica et Cosmochimica HUERTA-DIAZ, M.A. & MORSE, J.W. 1992. Pyritisation Acta, 50, 353-369. of trace metals in anoxic marine sediments. GeoPETERSON, L.C., HAUG, G.H., HUGI-~N, K.A. & ROHL, chimica et Cosmochimica Acta, 56, 2681-2702. U. 2000. Rapid changes in the hydrologic cycle of JANSEN, J.H.F., VAN OER GAAST, S.J., KOSTER, B. & the tropical Atlantic during the last glacial. Science, VAARS, A.J. 1998. CORTEX, a shipboard XRF290, 1947-1951. scanner for element analyses in split sediment POSTMA, D. & APPELO, C.A.J. 2000. Reduction of Mncores. Marine Geology, 151, 143-153. oxides by ferrous iron in a flow system: Column KULIK, D.A., KERSTEN,M., HEISER,U. & NEUMANN,T. experiment and reactive transport modeling. Geo2000. Application of Gibbs Energy minimization to chimica et Cosmochimica Acta, 64, 1237-1247. model early-diagenetic solid-solution aqueousRICHTER, T.O., LASSEN,S., VAN WEERING,T.C.E. & DE solution equilibria involving authigenic rhodochroHAAS,H. 2001. Magnetic susceptibility patterns and sites in anoxic Baltic Sea sediments. Aquatic Geoprovenance of ice-rafted material at Feni Drift, chemistry, 6, 147-199. Rockall Trough: implications for the history of LOURENS, L.J., ANTONARAKOO,A., HILGEN, F.J., VAN the British-Irish ice sheet. Marine Geology, 173, HOOF, A.A.M., VERGNAUD-GRAZZINI, C. & ZACHARIASSE, W.J. 1996. Evaluation of the Plio37-54. ROHLING, E.J., MAYEWSKI, P.A., ABU-ZIED, R.H., Pleistocene astronomical timescale. PaleoceanograCASFORD, J.S.L. & HAYES, A. 2002. Holocene phy, 11, 391-413. atmosphere-ocean interactions: records from LOURENS, L.J., WEHAUSEN,R. & BRUMSACK,H.J. 2001. Greenland and the Aegean Sea. Climate Dynamics, Geological constraints on tidal dissipation and 18, 587-593. dynamical ellipticity of the Earth over the past ROLLINSON, H.R. 1993. Using Geochemical Data." three million years. Nature, 409, 1029-1033. Evaluation, Presentation, Interpretation. Pearson, MERCONE, D., THOMSON, J., ABU-ZIED, R.H., CROUHarlow, Longman NJ. DACE, I.W. & ROHLING, E.J. 2001. High-resolution ROSSIGNOL-STRICK, M. 1983. African monsoons, an geochemical and micropalaeontological profiling immediate climate response to orbital insolation. of the most recent eastern Mediterranean sapropel. Nature, 304, 46--49. Marine Geology, 177, 25-44.

ITRAX SCANNING OF SAPROPEL UNITS ROTHWELL,R.G. 1995. Cruise Report: Marion Dufresne Cruise 81. Mediterranean Giant Piston Coring Transect. NOL, Southampton. Unpublished report. ROTHWELL, R.G., HOOGAKKER,B., THOMSON,J., CROUDACE, I.W. • FRENZ, M. 2006. Turbidite emplacement on the southern Balearic Abyssal Plain (western Mediterranean Sea) during Marine Isotope Stages 1-3: an application of ITRAX XRF scanning of sediment cores to lithostratigraphic analysis. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 79-98. SHIMMIELD,G.B. & PEDERSEN,T.F. 1990. The geochemistry of reactive trace metals and halogens in hemipelagic continental margin sediments. Reviews in Aquatic Science, 3, 255-279. SIESSER,W.G. 1976. Native copper in DSDP sediment cores from the Angola Basin. Nature, 263, 308-309. SKRABAL,S.A., DONAT, J.R. & BURDIGE, D.J. 2000. Pore water distributions of dissolved copper and copper-complexing ligands in estuarine and coastal marine sediments. Geochimica et Cosmochimica Acta, 64, 1843-1857. TEN HAVEN, H.L., DE LEEUW, J.W., SCHENK, P.A. & KLAVER,G.T. 1987. Geochemistry of Mediterranean sediments. Bromine/organic carbon and uranium/ organic carbon ratios as indicators for different sources of input and post-depositional oxidation, respectively. Organic Geochemistry, 13, 255-261. THOMSON, J., CRUDELI, D., DE LANGE, G.J., SLOMt', C.P., ERBA, E., CORSELLI, C. t~ CALVERT, S.E. 2004. Florisphaera profunda and the origin and diagenesis of carbonate phases in eastern Mediterranean sapropel units. Paleoceanography, 19, 119. PA3003 doi:10.1029/2003PA000976. THOMSON, J., HIGGS, N.C. & COLLEY,S. 1996. Diagenetic redistributions of redox-sensitive elements in northeast Atlantic glacial/interglacial transition sediments. Earth and Planetary Science Letters, 139, 365-377. THOMSON, J., HIGGS, N.C., JARVIS, I., HYDES, D.J., COLLEY S. • WILSON, T.R.S. 1986. The behaviour of manganese in Atlantic carbonate sediments. Geochimica et Cosmochimica Acta, 50, 1807-1818.

77

THOMSON,J., HIGGS, N.C., WILSON,T.R.S., CROUDACE, I.W., DE LANGE, G.J. & VAN SANTVOORT, P.J.M. 1995. Redistribution and geochemical behaviour of redox-sensitive elements around S1, the most recent Eastern Mediterranean sapropel. Geochimica et Cosmochimica Acta, 59, 3487-3501. THOMSON, J., MERCONE, D., DE LANGE, G.J. & VAN SANTVOORT,P.J.M. 1999. Review of recent advances in the interpretation of eastern Mediterranean sapropel S1 from geochemical evidence. Marine Geology, 153, 77-89. TUENTER, E., WEBER, S.L., HILGEN, F.J. & LOURENS L.J. 2003. The response of the African summer monsoon to remote and local forcing due to precession and obliquity. Global and Planetary Change, 36, 219-235. VAN DER WEIJDEN,C.H. 2002. Pitfalls of normalization of marine geochemical data using a common divisor. Marine Geology, 184, 167-187. VANOS, B., VISSER,H.J., MIDDELBURG,J.J. & DE LANGE, G.J. 1993. Occurrence of thin, metal-rich layers in deep-sea sediments - A geochemical characterization of copper. Deep-Sea Research, 1, 40, 1713-1730. VAN SANTVOORT,P.J.M., DE LANGE, G.J., LANGEREIS, C.G., DEKKERS, M.J. & PATERNE, M. 1997. Geochemical and paleomagnetic evidence for the occurrence of 'missing' sapropels in eastern Mediterranean sediments. Paleoceanography, 12, 773-786. VAN SANTVOORT,P.J.M., DE LANGE, G.J., THOMSON,J., CUSSEN, H., WILSON, T.R.S., KROM, M.D. & STROHLE, K. 1996. Active post-depositional oxidation of the most recent sapropel (S1) in sediments of the eastern Mediterranean Sea. Geochimica et Cosmochimica Acta, 60, 4007-4024. WEEKS, R., LAJ, C. ET AL. 1993. Improvements in longcore measurement techniques: applications in palaeomagnetism and palaeoceanography. Geophysical Journal International, 114, 651-662. WEHAUSEN, R. & BRUMSACK, H.J. 2000. Chemical cycles in Pliocene sapropel-bearing and sapropelbarren eastern Mediterranean sediments. Palaeogeography, Palaeoclimatology, Palaeoecology, 158, 325-352.

Turbidite emplacement on the southern Balearic Abyssal Plain (western Mediterranean Sea) during Marine Isotope Stages 1-3: an application of ITRAX XRF scanning of sediment cores to lithostratigraphic analysis R. G U Y

ROTHWELL IAN

1 BABETTE

W. CROUDACE

HOOGAKKER 1 & MICHAEL

2, J O H N

THOMSON

1

FRENZ 1

1National Oceanography Centre, Empress Dock, Southampton S014 3ZH, UK 2Department of Earth Sciences, University of Cambridge, Downing Street, Cambridge CB2 3EQ, UK Abstract: The upper part (0-20 m) of a long piston core from the SE Balearic Abyssal Plain -

spanning the past 50 ka - has been studied using the ITRAX micro-XRF core scanner to obtain downcore elemental profiles. The Ca/Fe ratio was found to be an effective parameter to distinguish between turbidites and pelagites, because turbidites generally have higher Fe contents and lower Ca contents compared with pelagic intervals. Beds that were obscure when visually logged could be identified as turbidites or pelagites on their geochemical characteristics, allowing more complete subdivision of the sequence into genetic units. The ITRAX XRF data also provide useful information on textural grading, bioturbative mixing, identification of geochemically distinctive marker beds, indications of differences in provenance, and confirm or query the presence of early arrivals during turbidite emplacement. A chronostratigraphic framework for the core based on accelerator mass spectrometry (AMS) radiocarbon dating and correlation with oxygen isotope stages of pelagic intervals in other cores (using calcium carbonate stratigraphy) was also established. This shows that turbidite emplacement on this part of the Balearic Abyssal Plain has been modulated strongly by climate and sea-level change, with turbidite emplacement most frequent during the early Holocene when the rate of post-glacial sea-level rise was greatest. Deposition of the coarsest (i.e. sand and silt-based) turbidites at the core site was restricted to the full and Late Glacial (11-25ka). Turbidite emplacement during Oxygen Isotope Stage 3 was rare. Most of the turbidites at the site are distal, but some coarse-grained-based turbidites are characterized by higher Sr/Ca ratios (possibly indicating a higher aragonite content), higher Ca and lower Fe contents compared to other turbidites, and are interpreted as having a more proximal shelf source. Such turbidites are generally rare, however, and restricted to full Glacial and Younger Dryas time. There is little evidence for large-scale seismogenic turbidites (expected to be seen as randomly timed emplacement, seemingly independent of eustatic control) at the core site, despite proximity to the seismically active Algerian margin 100 km to the south. This suggests that seismogenic turbidites must largely bypass this part of the plain. Although the ITRAX core scanner provides a rapid and non-destructive means of characterizing downcore geochemical distributions in great detail, interpretation of the data requires caution and assessment from an informed standpoint. Analytical artefacts such as those caused by water or organic content, degree of compaction, grain-size and mineral effects, unevenness of the cut core surface and poor discrimination of closely spaced element XRF peaks need identification and elimination.

The development of non-destructive X-ray fluorescence ( X R F ) instruments that scan cut sedim e n t core sections to obtain high-resolution geochemical profiles has been a m a j o r technical advance for the study of sediment records. The first instrument for this type of analysis was the C O R T E X X R F scanner developed by The Netherlands Institute for Sea Research, Texel, The Netherlands, and n o w m a r k e t e d by A v a a t e c h Analytical X - R a y Technology, Texel (Jansen

et al. 1998; Richter et al. 2006). Other X R F core scanners n o w available are the I T R A X m i c r o - X R F core scanner with X - r a d i o g r a p h y , m a n u f a c t u r e d by Cox Analytical Systems of G o t h e n b u r g , Sweden ( C r o u d a c e et al. 2006) a n d a high-resolution X R F core scanner m a d e by R 6 n t g e n a n a l y t i k Messtechnik of Taunusstein, G e r m a n y (Haschke 2006). M o s t previous research using X R F core scanning has concentrated on p a l a e o c e a n o g r a p h i c studies (e.g.

From: ROTHWELL,R.G. 2006. New Techniquesin Sediment Core Analysis. Geological Society, London, Special Publications, 267, 79-98. 0305-8719/06/$15.00

9 The Geological Society of London 2006.

80

R. G. ROTHWELL E T AL.

Haug et al. 2001; Jahn et al. 2003; Kuhlmann et al. 2004; Lamy et al. 2004, and others). Relatively little work has been carried out on the application of continuous XRF profiling to deep-sea basin cores dominated by turbidites, which are typical of many marine basins, where the technique may have value in lithostratigraphic analysis and, perhaps, provenance studies. Determination of genetic units (i.e. whether beds are autochthonous or allochthonous in origin) within cored sediment sequences is a fundamental prerequisite for lithostratigraphic and facies analysis, particularly for determining depositional processes and basin architecture. Identification of genetic units is necessary to determine depositional histories, estimate autochthonous and allochthonous sediment fluxes, and for investigation of the frequency of, and potential controls on, mass-wasting of continental margins. In marine basin sedimentary sequences, individual beds are commonly formed either through pelagic settling of material derived from biological surface productivity and wind-derived material through the water column (to form pelagites), or through advective sediment movement, usually involving sediment gravity flows, commonly turbidity currents that result in the deposition of turbidites. Turbidites are the distal products of downslope sediment transport and may form 90% of the basin-fill in some deep marine basins (Rothwell et al. 1992). Turbidites can be characterized on a sequence of textural characteristics that reflect gravitative particle settling within the waning flow (Bouma 1962; Stow & Shanmugam 1980; Lowe 1982). Traditional views hold that movement of terrigenous sediment to the deep sea by turbidity currents is greater at times of low sea level when rivers discharge their sediment loads closer to the shelf edge (e.g. Vail et al. 1977; Shanmugam & Moiola 1982, 1984; Posamentier & Vail 1988; Muto & Steel 2002). Some authors, however, suggest that sea-level change is an important factor in turbidite emplacement, and even that the rate of change of sea level is more important than low sea level in the generation of turbidites in aseismic areas (e.g. Weaver & Kuijpers 1983; Marjanac 1996). Traditionally, the identification of turbidites in sediment sequences is based on a range of sedimentological and textural criteria, involving recording of visual descriptions of sedimentary structures and lithology, and specialized and sometimes time-consuming laboratory analyses such as grain-size determinations. Distinguishing turbidites in marine sediment cores can be straightforward if they show clear compositional, textural and colour differences from the

interbedded pelagites (e.g. Weaver & Rothwell 1987). Where the colour and texture of turbidites and pelagites are similar, however, for example where turbidites, derived from resuspension and downslope transport of pelagic sediment from topographic highs, pond in topographic lows or where both turbidites and pelagites are strongly coloured through hydrothermal staining (e.g. Rothwell et al. 1994), it may be difficult in practice to distinguish between pelagite and turbidite beds. Recognition may also be difficult when the turbidites are highly distal and consist entirely of mud, lacking textural features normally associated with graded bases. Further, it may be difficult to distinguish turbidite beds from pelagic layers if they are thin and extensively bioturbated. The presence of sharp bed bases and lack of obvious scattered foraminifers within the sediment (resulting in smooth structureless mud) are probably the best visual criteria for distinguishing turbidites from pelagic/hemipelagic muds. The upper few decimetres of turbidites are commonly bioturbated if the overlying sediment is pelagic or hemipelagic in origin. If bioturbation is moderate or intense, it can be difficult to ascertain the upper boundary of the turbidite bed with the overlying unit. This leads to difficulty in determining true thicknesses of pelagic/hemipelagic intervals and accurate assessment of pelagic accumulation rates and turbidite emplacement frequencies. We report here a high-resolution geochemical and sedimentological study of the upper part of a long piston core from the SE Balearic Abyssal Plain in the western Mediterranean Sea. This core (LC06) was taken by the research vessel Marion Dufresne in 1995 in 2845 m water depth on the floor of the abyssal plain at 38~ and 7~ l l 0 k m north of Cap de Fer, Annaba, Algeria, which lies 400km east of Algiers (Fig. 1). The core is 31m in length, although we report here on the succession of the upper 20m that spans the last 50ka. We have sought to answer a number of questions relating to the usefulness of high-resolution XRF scanning in lithostratigraphic analysis and regarding the late Quaternary depositional history of this part of the Balearic Basin. Specifically: 9 What are the best element or element ratio parameters, derived from high-resolution ITRAX scanning, for providing information on compositional contrasts, particle-size variation, provenance indications, diagenetic features and presence of exotic layers? 9 Does variation in element profiles and sediment property profiles mirror textural and sedimentological parameters and enable

ITRAX XRF SCANNING OF TURBIDITE CORES

81

present. High Fe relative to Ca, indicative of the detrital clay to biogenic carbonate ratio, was an important indicator of Balearic plain turbidite intervals in the ITRAX data. Most turbidites are varying shades of olive green in colour, and some beds had paler green upper parts due to oxidation by bottom-water 02 after emplacement (Wilson et al. 1986). Such dual-tone turbidites occur throughout the sequence studied, although some deeper turbidites show only one colour. Pelagic intervals were identified by colour (commonly pale fawn or brown) and the presence of scattered foraminifers and pteropod or shell debris throughout. Pelagic intervals commonly show slight-moderate bioturbation and burrowed lower boundaries. Sediment colour was measured using a Minolta CM-2002 spectrophotometer. Each section was independently logged by two of the authors (R.G. Rothwell and B. Hoogakker) with good agreement.

Dating of the sediment sequence

Fig. 1. Location of core LC06 on the Balearic Abyssal Plain in the western Mediterranean Sea. Position of comparator cores LC01 and LC04 are also shown. Bathymetric contour interval is 500 m. better assessment of distality/proximality relationships? What has been the turbidite emplacement frequency, has this changed over time, and how has turbidite emplacement on this part of the abyssal plain been modulated by climate and sea-level change?

Methods Core logging - discrimination of genetic units Archive core sections from the upper 20 m of the sediment core were re-faced using glass slides and visually logged by conventional core description (Ocean Drilling Program 1987). Individual sediment units were identified as pelagites or turbidites on textural and sedimentological criteria, and where problematic (for example, where turbidites were thin and bioturbated throughout) using ITRAX-generated geochemical proxy data. Generally, turbidites were identified on the presence of burrowed upper boundaries, the absence of scattered foraminifers within mud intervals, the presence of sharp planar or eroded bases, and normally graded silty and/or sandy bases if

Three accelerator mass spectrometry (AMS) radiocarbon ages were available, determined on hand-picked planktonic foraminifera samples from pelagic intervals within the upper 20 m of this core (Table 1). The lowermost age (37230 calibrated years BP) was disregarded as this is now believed to have been taken from a turbidite rather than a pelagic interval, so that the foraminifers analysed may have been derived from older deposits. Hoogakker (2003) demonstrated that the calcium carbonate stratigraphy of pelagic records from the Balearic Abyssal Plain closely mirror the oxygen isotope record, with high CaCO3 contents when planktonic foraminifera 6180 values are low. Variation in pelagic calcium carbonate can therefore be used as a proxy to identify known oxygen isotope events. Measurement of the calcium carbonate contents of all pelagic intervals identified in core LC06 was made by coulometry. Samples were oven dried at 100 ~ for 48h, the dried samples were ground and CaCO3 content measured from CO2 liberated by addition of 10% phosphoric acid after 6 min (Hoogakker 2003). The resulting calcium carbonate stratigraphy of core LC06 was compared with those obtained from pelagic intervals in two other long piston cores from the Balearic Abyssal Plain (LC01, 40~ 6~ water depth 2845m; and LC04, 38~ 6~ water depth 2800m) both within 250km of the LC06 core site (Figs 1 & 2). These cores had been calibrated by oxygen isotope measurements and this allowed identification of dated oxygen isotope events in LC06

82

R . G . ROTHWELL ET AL.

Table 1. Absolute dates for pelagic intervals" in core LC06 Depth (mbct)

Date type

14C age years BP • SD

Calibrated age (years BP) Age model

3.66-3.69* 13.46-13.65" 19.17 19.19" 19.87 20.46

AMS radiocarbon AMS radiocarbon AMS radiocarbon Oxygen Isotope Event 3.3 ~ Oxygen Isotope Event 3.31 ~+

7991 + 43 17730+ 190 34380• -

8381-8481 20 126-20 889 c. 37 230 + -

8400 20 500 c. 37 230 50 210 55 450

Source Hoogakker (2003) Rothwell et al. (1998) Rothwell et al. (1998) Hoogakker (2003) Hoogakker (2003)

* Depth in metres below core top is given as I m less than in the source works as we discount the uppermost 1 m core section (core section 31), which was empty and count core top as the top of core section 2. In the source works, the empty uppermost core section was included in depth calculations. t Calibrated using the magnetic curve of Laj et al. (1996), other radiocarbon dates calibrated using INTCAL98 tStuiver et al. 1998). § Inferred from calcium carbonate stratigraphy (see text).

(Table 1). The oxygen isotope analyses were m a d e on hand-picked, consistently sized specimens of the foraminifer Neogloboquadrina pachyderma (dextral variety) using a E u r o p a G E O 2020 stable isotope ratio mass spectrometer with individual acid-bath c a r b o n a t e p r e p a r a t i o n ( H o o g a k k e r 2003). Identification of these absolute age tie points within the pelagic record of LC06 allowed individual turbidites, identified t h r o u g h logging, to be dated to an a p p r o x i m a t e e m p l a c e m e n t time. This was done t h r o u g h estimation of pelagic sedimentation rates between the tie points and consideration of the thickness of pelagic intervals between individual turbidites.

E l e m e n t profiling using the I T R A X

dual elements that are nominally p r o p o r t i o n a l to concentrations of m a j o r and m i n o r elements within the sediment. Elements can be detected in the range from Si to U. A m o r e detailed description of the I T R A X instrument and the data generated is provided by C r o u d a c e et al. (2006). In the present study, the selected step size was 0.5 m m with a X R F c o u n t time at each step of 30 s. D a t a for AI, Si, Ca, K, Ti, V, M n , Fe, Ni, Cu, Zn, As, Br, Rb, Sr, and Zr are deconvolved from the X-ray spectra as the core runs t h r o u g h the I T R A X (elements lighter than A1 currently c a n n o t be m e a s u r e d using the I T R A X ) . In total, 40000 X R F spectra over 2 0 m of core were collected, providing a uniquely detailed geochemical characterization of part of a deepsea long piston core. The resulting X R F element

high-resolution X R F core s c a n n e r Archive core sections covering the u p p e r m o s t 20 m (0-50 ka BP, based on the dating evidence above) of core LC06 were run t h r o u g h the I T R A X m i c r o - X R F core scanner. This instrument is designed to u n d e r t a k e high-resolution energy-dispersive X R F ( E D - X R F ) measurements along the longitudinal axis of split sedim e n t cores. A n optical image of the full split core surface and a 2 0 m m - w i d e c o n t i n u o u s high-resolution X - r a d i o g r a p h are also obtained using optical and radiographic line cameras. The X-rays used to irradiate the core section are generated from a 3 k W M o target and focused t h r o u g h a flat glass capillary waveguide to allow very high-resolution m e a s u r e m e n t ( d o w n to 200 I,tm step size). The X-ray dwell time (for X-radiography), c o u n t time (for X R F ) and m e a s u r e m e n t step size are all user definable. Each E D - X R F spectrum is recorded and deconvolved to derive peak area integrals for indivi-

Fig. 2. Graphic logs (left) for cores LC01, LC04 and LC06. Turbidite intervals are shown as grey, pelagic intervals as black bands. The pelagic intervals are then summed to give total pelagic thickness (second column), which is then expanded by a factor of 8 to give 'Pelagic depth' (third column). This represents the pelagic sequence if no turbidites had been emplaced. Nannofossils zones for the pelagic sequence according to the scheme of Weaver (1983) are shown in column 4. The boundary between nannofossil zones 1 and 2 is dated to c. 50,000 years BP. Calcium carbonate profiles, derived through coulometry, for each core are shown together with oxygen isotope measurements for cores LC01 and LC04. Note the correlation between the calcium carbonate values and the oxygen isotope measurements in LC01 and LC04, showing the calcium carbonate stratigraphy to be a good proxy for oxygen isotope stages that can be used by analogy in core LC06. Emplacement of turbidites appears to cause little or no erosion of substrates. Position and dates of AMS radiocarbon dates are also shown.

ITRAX XRF SCANNING OF TURBIDITE CORES

83

84

R . G . ROTHWELL E T AL.

peak integrals are summarized and stored as .txt files that can be visualized and manipulated either in Excel or Itrax.PLOT - a custom-built program developed by Southampton Oceanography Centre and JF Computing (Stanford in the Vale, Oxfordshire, UK). Aluminium is at the limit of I T R A X detection and counts for this element are low. Therefore we could not use AI as a reference element and had to find an alternative detrital divisor element for normalization purposes. Grain-size analysis

To test whether variation in element ratio profiles determined by I T R A X X R F measurement mirrored textural parameters, one core section was analysed for grain size at 1 cm resolution. The grain-size analyses were carried out using a Malvern Mastersizer 2000 in combination with a Malvern Hydro G accessory unit and a 36pot Malvern auto-sampler. This instrument is capable of analysing a size range from 0.02 to 2000 gm. Approximately 0.25 (if mud) to l cm 3 (if sand) of bulk sediment was taken and dispersed in a 0.05% sodium polyphosphate (Calgon) solution by shaking overnight. For the analyses, a laser-target obscuration (sample concentration) of 10-20% was selected. Settings of particle refractive index to 1.52 and an absorption value of 0.1 provide average values for mixed mineral assemblage of marine sediments. Three measurements were carried out on each sample for which the averages were calculated and exported in 0.2 phi steps to calculate the arithmetic mean grain size according to Krumbein (1936).

Results and discussion A graphic log of the lithology of the upper 20 m of core LC06, derived from visual logging and ITRAX-measured geochemical proxies (principally the Ca/Fe ratio), is presented as Figure 3. A total of 52 turbidite beds were identified above the inferred depth for Oxygen Isotope Event 3.3 (indirectly dated at 50210 years BP: Martinson et al. 1987), which occurs at a depth of 19.87 metres below core top (mbct), giving an overall turbidite emplacement frequency of approximately one event every 960 years. Emplacement frequency has varied significantly over this time interval, however. Individual turbidites vary between less than 1 to 555cm in thickness, and most turbidites consist of Te division mud (Bouma 1962) or graded-ungraded T6-T8 muds (Stow & Shanmugam 1980). Beds

Fig. 3. Graphic log of the upper 20 m of core LC06 derived from visual and ITRAX logging. Beds are identified as pelagic or turbiditic in origin according to criteria described in the text. Core sections with illustrated XRF profiles are indicated. The megaturbidite is a large-volume deposit of basinwide extent, emplaced at 22 ka, during the last lowstand in sea level (Rothwell et al. 1998). Graded bases are indicated with arrows representing direction of fining.

with silty or sandy bases are generally rare and make up only about 15% of the total. Turbidites with basal sands make up 8 % of the total. Turbidites with visibly coarser-grained bases, whether silt and/or sand, only occur between 5.90 and 19.26mbct. Lamination is not seen in the coarser-grained bases, which commonly appear as massive, poorly sorted, normally graded sands and silts. The response of the I T R A X core scanner to different element concentrations is variable as is normally the case with X R F analysis. The magnitude of element integrals depends mostly on excitation efficiency produced by the particular X-ray tube primary radiation (in this case a Mo tube), the energy of the elemental X-rays and

ITRAX XRF SCANNING OF TURBIDITE CORES abundance of the element. For excitation of the sediments studied, Ca, Fe and Sr provide a good response, with K, Rb, Ti, Mn, As and Zn providing a moderate response. Br, Ni and Co can also provide useful information in organicrich sediments (Thomson et al. 2006). A clear distinction in pelagite/turbidite beds in LC06 is shown from plots of Ca/Fe. Visual logging identified a small number of thin (<10cm) heavily bioturbated beds of uncertain affinity. These lacked the sharp bases characteristic of turbidites, although patchy olive green colour and blebs of homogenous mud within these intervals suggested they may be thin turbidites bioturbated throughout their entire thickness. Examination of Ca/Fe ratios confirms this interpretation, showing Fe enrichment in all cases and enabling complete division of the sequence into genetic units, something that could not be done confidently on visual logging evidence alone. Element concentrations are commonly expressed as ratios to a detrital phase element to avoid closed sum effects due to variations in calcium carbonate content, which will automatically cause variations in element profile integrals (Rollinson 1993; van der Weijden 2002). Usually elements are normalized to A1, but as this element is at the limit of ITRAX measurement an alternative element divisor characteristic of the detrital phase was needed. Figure 4 shows profiles for MoK Compton scattering (the magnitude inversely relating to mean atomic number), Fe/A1, Fe/Ti, Fe/K, Fe/Rb and other ratios through LC06 section 30. In this section, two turbidites, at 120-590 and 590-960mm below section top, directly overlie one another. Compton scattering relates inversely to sediment mean atomic number so that atomic number increases where Compton scattering decreases. Visual logging recorded 2 mm of silt at 590 mm, but above this the unit appears as uniform homogenous olive green mud. Compton scattering decreases over a much longer interval (500590mm), revealing compositional grading not evident to the eye. Corresponding grading is not very obvious in the Fe/Ti ratio, except as a slight increase in Ti. A discrete peak at 380 mm, coincident with a narrow but conspicuous peak on the As integral profile, may indicate pyrite authigenesis. We consider K an unreliable divisor element as increases in K may relate to higher water content, as C1 from sea-water salt absorbs potassium K X-rays (Croudace et al. 2006). Rubidium typically shows a good signal on the ITRAX, and the Fe/Rb ratio shows the grading between 500 and 590 mm in the upper turbidite bed, seen as decreased Compton scattering values. We therefore consider Rb to be an

85

effective element divisor in the absence of precise A1, and this element was used in this study. Element integral ratios or integrals that we believe likely to be the most informative in aiding sedimentological study or lithostratigraphic analysis are summarized in Table 2 together with their sensitivity to ITRAX detection and indicator properties. We present these element ratios plotted against lithology for three representative core sections from core LC06 (Figs 5-7). These figures include core photographs, X-radiographs and profiles for Compton scattering, Ca/Fe, Sr/Ca, Fe/Rb, K/Rb, Ti/Rb, Zr/Rb, Cu/Rb, As and Si profiles throughout core sections 17 (12.91-13.78mbct), 15 (14.7815.78mbct) and 11 (18.78-19.78mbct). These figures illustrate the range of trends seen in these element ratios and the Compton, As and Si integrals. Lithology determined by visual logging and the geochemical profiles is also shown. The geochemical plots confirm the lithological subdivision showing clear 'steps' or inflections at bed boundaries, although bioturbative mixing leads to gradational profiles in the upper parts of turbidite beds. Section 15 (Fig. 6) consists entirely of ungraded homogenous mud and comes from the mud interval of a very thick (5.5 m) sand-based megaturbidite previously identified by Rothwell et al. (1998) as a large-volume last glacial lowstand deposit of basin-wide extent. The Mo Compton scattering and element ratio profiles show its uniform nature. Below we offer observations on the parameters measured by the ITRAX and their significance. M e a s u r e d integrals and ratios Compton scattering. The Compton scattering integral, which relates inversely to mean atomic number, decreases in the silt and sand layers due to size/density-related mineral fractionation. Grading, presumably due to mineralogical variation, can be also seen in this parameter, even when not visually obvious. Inflections in the profiles often correlate with bed boundaries. Sharp decreases in Compton scattering also correlate with discrete pteropod layers (e.g. Section 11, 79 cm, Fig. 7), possibly produced by current winnowing, as mean atomic number falls with looser sediment packing. Ca/Fe ratio. The Ca/Fe ratio reflects carbonate content and in core LC06 generally shows a strong correlation to sedimentary units. In general, turbidite sands, silts and muds, presumably sourced from shallower water, are richer in Fe and poorer in Ca than pelagic interbeds. The

86

R. G. ROTHWELL ET AL.

"~

~.~

=9o ~ "~

"~

~

0

~

o.~ c~ ~

~ o 0~'.~

~.~.~ ~.~ ~o~

..'~

~0

~

ro-~ ~= ~ . . .

~

"8~ 0

0

~ 0

. c~ ~

0

0

. ~ ".~ ~

~-~

~ o:.=

~ e~

~.-

.~o

:=

~ ~ ~ . ~

~-~ ~.=

.~

~.~ ~.~

~~ 9~

~.~

ITRAX X R F S C A N N I N G OF T U R B I D I T E CORES

87

Table 2. Property, element ratios or integrals believed to be informative in aiding sedimentological or lithostratigraphic analysis in clayey, sandy and carbonate-rich marine sediments together with their sensitivity to I T R A X detection (using a Mo X-ray tube) and indicator properties Property, element ratio or integral

Sensitivity to ITRAX detection

Compton scattering

High

9 Relates inversely to mean atomic number, commonly decreases in silt and sand layers due to size/density-related mineral fractionation 9 May show grading due to mineralogical variation 9 Inflections in profile commonly correlate with bed boundaries 9 Mean atomic number falls with looser sediment packing, so winnowing of sediment may be seen as decreased Compton scattering

Ca/Fe

High

9 Indicative of biogenic carbonate:detrital clay ratio 9 May show strong correlation with sedimentary units 9 Turbidites sourced from shallow water tend to be richer in Fe and poorer in Ca than pelagic interbeds 9 Ca/Fe profile is a good proxy for sediment grading, for identifying textural subdivisions within turbidites and for assessing source distality-proximality relationships 9 Ca/Fe profile within pelagites typically more variable than in turbidites, reflecting more heterogeneous sediment fabric 9 Ca peaks or their absence (commonly associated with increased Si) within turbidite bases distinguish foraminifer- or shell-rich and more terrigenous quartz-rich bases

Sr/Ca

High

9 Enhanced Sr may indicate presence of high-Sr aragonite which requires a shallow-water source 9 Affected by sediment packing/porosity and grain-size/shape variations

Fe/Rb

Good

9 Commonly shows grain-size related fractionation effects within turbidites 9 Fe mobilized during redox-related diagenesis and elevated Fe commonly seen in oxic, or formerly oxic, parts of turbidites 9 Rb is an element commonly associated with detrital clay and may be enhanced in turbidite muds

K/Rb

Moderate

9 K is commonly associated with detrital clay and may be enhanced in turbidite muds 9 Unreliable parameter as sea-water C1 absorbs potassium X-rays, so apparent high K may reflect increased porosity

Zr/Rb Ti/Rb

Moderate

9 Zr and Ti high in heavy resistate minerals and may be enhanced in turbidite bases 9 Sediment source/provenance indicators

Cu/Rb

Moderate

9 Behaviour of Cu poorly understood but Cu peaks largely of diagenetic origin

As

Moderate

9 Commonly an indicator of pyrite which may be detrital or authigenic in origin

Br/C1

Moderate-low

9 Indicator o f organic-rich layers as Br high in organic-rich sediments. For marine sediments a constant ratio implies sea-water ratio

Si

Moderate-low

9 Important terrigenous or productivity indicator 9 Normalization using detrital divisor can distinguish terrigenous or productivity origin 9 When terrigenous, useful as a sediment source and perhaps provenance indicator

Indicator properties

88

R. G. ROTHWELL E T AL.

-r~-o~

~ . ,...,

ITRAX XRF SCANNING OF TURBIDITE CORES

89

d 9~

o~

"~ r

o

oo .-~

.~"~

. ~"

~

~~

r ~0

0

c~

.~ ~0o

o

~o ~.. ~~

90

R. G. R O T H W E L L E T AL.

,--~

0

0

~

o.~

O ~

~.~ ~= ~ ~-~

~, .~'~ ~'~

~a

~ o~ ~ ~

o'~

ITRAX XRF SCANNING OF TURBIDITE CORES

91

Fig. 8. Grain-size profiles and Ca/Fe ratio through LC06 section 11, 15-97 cm (cf. Fig. 7). Note the close correspondence between the Ca/Fe profile and the mean grain size and the proportion of grains >32 ~tm in the sediment, suggesting that Ca content closely relates to the presence of foraminifers. The Ca/Fe ratio seems a good proxy for textural variation. Grain-size measurement was made at a resolution of 1 cm over the interval studied.

Ca/Fe profile shape within turbidites appears to mirror grading, or its absence (Fig. 8), and provide clear indication of the Piper (1978) and Stow & Shanmugam (1980) textural subdivisions without the need for grain-size analysis. Bioturbative mixing of pelagite into the upper parts of turbidites causes the Ca/Fe ratio to decrease in the tops of the turbidite units. Identification of textural subdivisions is important in assessing source distality-proximality relationships. Calcium peaks, or lack of Ca peaks (commonly associated with increased Si), within turbidite bases distinguish foraminifer- or shell-rich and more terrigenous quartz-rich bases. A small number of turbidites with sandy bases have relatively higher Ca and lower Fe compared to the others, and may have a more proximal source. The Ca/Fe profile within pelagites is typically more variable than in turbidites, reflecting the more heterogeneous bioturbated sediment fabric (Fig. 8). Sr/Ca ratio. Biogenic low-Mg calcite from surface ocean production has a limited Sr/Ca range (e.g. 1100 and 1900~tg Srg -1 CaCO3 in foraminiferal and coccolith calcites according to Stoll & Schrag 2000) and enhanced Sr may indicate

the presence of high-Sr aragonite, which requires a shallow-water source (Thomson et al. 2004). Shallow and lagoonal waters on subtropical and temperate continental shelves are commonly rich in Halimeda (a macroscopic green calcareous algae whose calcified segments are composed of very fine-grained aragonite). Halimeda-rich sediments are described from shallow temperate bays along Mediterranean coastlines (e.g. Fornos et al. 1992; Hieke & Werner 2000). Where it occurs Halimeda is an important contributor to neritic sediments. Many neritic invertebrates, such as bivalves, gastropods and corals, have aragonitic shells. Hence, neritic carbonate debris is richer in aragonite than ocean productivity derived low-Mg calcite of foraminifera and coccoliths, so the Sr/Ca ratio may reflect provenance. Turbidites with sandy bases, apart from the megabed (13.5-19mbct, Fig. 3), generally show higher Sr/Ca values than turbidites lacking coarsegrained bases, perhaps indicating a different, more proximal provenance (e.g. North African or Sardinian shelf). The Sr/Ca ratio, however, like the other parameters measured, is affected by sediment-packing/porosity and grain-size/ shape variations. Pelagic intervals containing abundant pteropods commonly register high

92

R. G. ROTHWELL ET A L.

Sr/Ca, although pteropod shells are composed of low-Sr aragonite (Krinsley & Bieri 1959; Kinsman 1969; Rutten et al. 2000) so, in these cases, the high Sr/Ca values must be a texturally related effect (Fig. 7). Fe/Rb ratio. The Fe/Rb ratio, like Sr/Ca, also commonly shows grain-size-related fractionation effects within turbidites. Fe is mobilized during redox-related diagenesis, so elevated Fe contents are seen within bioturbated (formerly oxidized) parts of turbidites that have been affected by post-deposition oxidation by bottom-water 02 diffusion into the turbidite after emplacement (e.g. section 11, 62.5-67 and 91.5-95cm, Fig. 7). Increased Rb within turbidite muds reflects greater clay contribution to the sediment. K/Rb ratio. The K/Rb ratio commonly shows low values in silts and sands compared to mud intervals reflecting the presence of K (like Rb) in clays and porosity-related artefacts associated with sea-water CI absorbing potassium X-rays (Croudace et al. 2006). Zr/Rb and Ti/Rb ratios. Zirconium and Ti contents are high in heavy resistate (i.e. resistant to chemical and mechanical erosion) minerals (de Meijer 1998) and these accessory minerals are commonly enriched in the bases of some turbidites due to gravitative settling. Differences seen in different turbidite bases may reflect distance from sediment source and provenance differences. In Figure 5, the thin silt turbidite between 38 and 41cm has a lower Zr content than the overlying sand at 29-38 cm, but is very similar to the overlying graded silt at 16-29 cm. The silt bed at 3 8 4 1 c m may, therefore, be an early arrival of the overlying graded turbidite at 6-38 cm (see discussion below). Cu/Rb. The behaviour of Cu in marine sediments is poorly understood. The Cu/Rb profile shows great variability with Cu peaks, apparently largely of diagenetic origin, occurring in both pelagic and turbidite intervals (e.g. Fig. 7). We believe that the discreet nature of the peaks discounts the possibility of any Cu artefact significantly affecting the data. As. Arsenic is efficiently excited using a Mo tube, so that coherent As signals are seen with the I T R A X even at low concentrations. Arsenic may be present as a detrital or authigenic phase and commonly indicates the presence of pyrite (FeS2). Indeed, As is a better indicator of this mineral than Fe, due to high background levels of Fe in detrital phases making it difficult to distinguish the relatively small additional amounts

of Fe in authigenic pyrite. In contrast, As has a low concentration in detrital phases but a high concentration in pyrite. Pyrite occurs in a wide range of grain sizes. Authigenic pyrite is widespread in marine sediments and can form in localized microenvironments where oxygen depletion and sulphate reduction occurs, such as within burrows and foraminifer test chambers. In core LC06 the As integral profiles suggests that detrital pyrite, seen within turbidite bases, is commonly of restricted grain size and more abundant in turbidite silts than sands (e.g. Fig. 5, 1541cm). This suggests either that sandsized pyrite has settled out more proximally, or that the pyrite present is of restricted grain size, perhaps occurring mainly as framboids (spheroidal grains composed of pyrite microcrystallites), the most common form of pyrite in marine sediments, formed by sulphate reduction within microfossil tests. Br/Cl. Bromine contents are high in organicrich sediments (Ten Haven et al. 1987; also fig. 4 in Thomson et al. 2006) causing the Br/C1 ratio to increase over the constant sea-water value. No organic-rich layers were identified in core LC06, however, and Br/CI ratios are consistently low in all sections. ITRAX

applications and limitations

ITRAX-acquired geochemical data are essentially semi-quantitative in nature and need to be interpreted with caution. Errors may arise due to poor peak discrimination in the X-ray spectra, porosity changes, compaction or grainsize/shape-related artefacts (recorded for K and Sr), low count rates and crack-related effects. Invalid data may be recorded when the X-ray detector is not in the correct position, particularly when the cut core surface is uneven or shows sudden variability. Careful study of variation in the element integral profiles, the Compton scatter integral and the detector-sediment distance index can aid in identifying invalid data. The data visualization program Itrax.PLOT can temporarily remove these suspect data prior to final presentation; the original data are never modified. The ITRAX profiles do allow useful primary textural- and depositional-related and secondary diagenetic features to be identified and their character studied in unprecedented detail. Early arrivals A further application of the I T R A X data may be in the identification of turbidite early arrivals.

ITRAX XRF SCANNING OF TURBIDITE CORES Early arrivals represent portions of a turbidity current that reach the point of deposition earlier than the main flow, usually due to flow division on the slope or rise. Such division may occur where canyons or channels divide or there is canyon overspill. Early arrival deposits are well documented from deep-sea basins (e.g. Weaver & Rothwell 1987) and commonly appear as basal silts and sands with thin mud caps, which are then directly overlain by the sand and/or silt of the main flow. The sand and silts of the early arrival and the main flow are usually very similar in composition. Several possible early arrivals were identified during visual logging of core LC06 on the basis of textural and colour similarities to the overlying turbidite. However, the ITRAX data showed that in many cases there were significant geochemical differences in the presumed early arrivals and overlying turbidites, possibly because the element/Rb values are different in different grain-size classes. It may be that some early arrivals had entrained other material on their passage downslope or, rather than being early arrivals, some beds may have different sources, representing distinct separate events. Comparison of element/Rb values in some 'early arrival' silts with those in the silts above overlying sands, however, did suggest some beds were probably related to the same turbidity current flow. Climatic modulation

Identification of individual turbidites, through visual and ITRAX logging, and placing them in a time stratigraphic framework allows assessment of any climatic modulation of emplacement (Fig. 9). This shows that turbidite emplacement at the LC06 core site has varied markedly with time over the last 50ka, with a significantly increased frequency of turbidite emplacement after 20 ka. Thouveny et al. (2000) demonstrated through magnetic susceptibility measurements of sediment fabric and sedimentation rate studies of Calypso piston cores that the upper 10-15 m of some, if not all, Calypso cores are up to 1.5-2 times longer than the same sequence recovered by conventional piston cores on the same site. This is interpreted as being due to syringing or oversampling of the sediments in the upper portion of these long cores due to cable rebound resulting in upward piston acceleration and a microfabric rotation into the vertical during the coring process. Skinner & McCave (2003) also show that the upper 5-15m of Calypso piston cores are affected by such oversampling on soil mechanics considerations. Turbidites within the upper 20m of the sediment sequence in core

93

LC06 are much thicker than those below this level (fig. 3 in Rothwell et al. 1998). Further, the time-stratigraphic framework suggests significant stretching of pelagic intervals within this part of the sequence. Uncorrected average pelagic accumulation rates are 11 c m k a -1 between 0 and 3.5mbct, 8 c m k a -1 between 3.5 and 13.5mbct, and approximately 1 cmka-1 below 13.5mbct. Therefore, stretching of the sequence is implicit over the interval 0-15 mbct. This stretching does not, however, negate the increase in turbidite frequency seen within the upper 20 m of the sequence. In Figure 9 turbidite thickness and basal sand/silt content are plotted against estimated emplacement time (see the Methods section for method of calculation). The times of emplacement for turbidites above the AMS radiocarbon datum of 8.4 ka are maximum ages only as it is unknown how much of the most recent sediment was lost during the coring process. It is apparent from the figure that the full and Late Glacial of the Late Devensian glaciation and the postglacial rise in sea level were significant times for turbidite emplacement. Between 26 and 45ka turbidite emplacement at the LC06 core site was rare with a frequency of one event every 89 ka, so that only three thin (from 2 to 11 cm in thickness) mud turbidites were emplaced over this 19ka time span. The full glacial (26-14ka) saw a marked increase in both frequency and magnitude of turbidite deposition. During this 12ka period, 18 turbidites (from >1 to 555cm in thickness) were deposited with a frequency of one event every 600-700 years, although emplacement frequency increases towards the Late Glacial (Fig. 9). The thickest turbidite present in LC06 was deposited at the Late Glacial Maximum at 22 000 years Be. This 5.5 m-thick bed is a megaturbidite of basin-wide extent that can be mapped as a conspicuous acoustically transparent layer on high-resolution seismic profiles (Rothwell et al. 1998). This bed, previously described from five long piston cores and highresolution seismic profiles across the abyssal plain from as far west as 5~ and as far north as 41~ has a massive sandy base that thickens and coarsens towards the north, suggesting a source on the southern European margin (Rothwell et al. 1998). All turbidites with silty and/or sandy bases in core LC06 were emplaced during full or Late Glacial times (Fig. 9). During the Younger Dryas and Holocene (144)ka) there was frequent emplacement of mud turbidites. During this time interval 26 turbidites are recorded in LCO6, giving an emplacement frequency of one event approximately every 500 years. The magnitude of Holocene turbidites

94

R . G . ROTHWELL E T AL. Turbidite thickness (metres) 0

1

2

ii

4

24

26

5

Younger l~yas

/

6

N <.1~

Bolling-Allemd - -

(1

~" 22

<~

3

.P

28

150

1O0

50

0

Sealevel curve (metres below present) Fig. 9. Emplacement time and bed thickness for turbidites identified in core LC06, 0-20 mbct. Black bars indicate individual mud turbidites, with blue and red bars indicating the thickness of silt and sand, respectively, within turbidite bases, where present. The sea-level curve of Shackleton (1987), and oxygen isotope and climatic boundaries are also shown. Dates for the Younger Dryas and Bolling/Allerod are taken from Hughen et al. (2000). Note that emplacement of sand and silt-based turbidites is largely confined to Oxygen Isotope Stage 2 and that frequency of turbidite emplacement strongly correlates to sea-level change, emplacement being most frequent when the rate of sea-level rise is greatest (during the early post-glacial).

is difficult to assess as their thicknesses recorded in the core are clearly stretched. Overall, it seems clear that turbidite emplacement frequency has increased during the Holocene, but that emplacement was most frequent when the rate of change (i.e. rise) of sea level was greatest (Fig. 9). A similar relationship has been reported for the recent and more distant geological record in aseismic areas. Wynn et al. (2002), for example, showed that turbidites within the Moroccan Turbidite system (NE Atlantic) were emplaced at Oxygen Isotope Stage boundaries during periods of

rapid sea-level change. Marjanac (1996) attributed the deposition of large turbidites in the Eocene-Miocene flysch of central Dalmatia (Croatia) to periods of sea-level rise. Turbidites on the Madeira Abyssal in the N E Atlantic are also reported to have been emplaced at Marine Isotope Stage boundaries at times of sea-level change (Weaver & Kuijpers 1983; Weaver et al. 1992). In the present study, low sea level seems to have been particularly important for the emplacement of large-volume turbidites (assuming thickness correlates with volume in this

ITRAX XRF SCANNING OF TURBIDITE CORES depositional setting), but increased turbidite emplacement frequency is essentially a Late Glacial-post-glacial phenomenon. Turbidite internal subdivisions

Most of the turbidites cored in the sequence are highly distal, many consisting entirely of ungraded E3 (Piper 1978) or T7-T8 mud (Stow & Shanmugam 1980). Indeed, the ITRAX geochemical profiles often provide an excellent proxy for determining grading within turbidite muds (Fig. 8) and identification of their textural subdivision according to the Piper and/or Stow and Shanmugam scheme, for example, allowing discrimination of graded and ungraded mud. In this way ungraded E3 and T7-T8 muds can be distinguished from graded E2 (Piper 1978) or T6 mud (Stow & Shanmugam 1980), a distinction that cannot be made with the human eye. The ITRAX data therefore allow the distality of individual turbidites to be assessed. S o u r c e o f turbidites

The pattern of allochthonous sedimentation seen at the LC06 core site suggests that climate and sea-level change are important, even dominant, factors both in affecting turbidite emplacement frequency and the deposition of large-volume beds. The core site is only 100 km distant from the North African margin. The Algerian margin is steep (slopes are 15~ or more), incised by canyons (Rehault et al. 1985; D6verch~re et al. 2005) and the hinterland mountainous. The African-Eurasian plate boundary runs in an east-west direction close to the Algerian margin. The African plate moves to the north towards the Eurasian plate in a NNW direction with a translation to the east resulting in seismogenic strike-slip faulting along the margin (Thomas 1976). Large offshore earthquakes on the continental shelf and slope have occurred (Ambraseys 1982; Harbi et al. 1999). Some historical earthquakes on the Algerian coast have been destructive. Over 95 earthquakes (including 47 with M > 5) have been recorded in NE coastal Algeria (between 35 ~ and 38~ 4~176 since 1357AD (Harbi et al. 1999, fig. 2). The main river discharging on the eastern Algerian margin is the River Soumam which enters the Gulf of Bejaia, 200km west of Annaba. The thick sedimentary accumulation offshore of the river mouth is affected by faulting which has generated submarine landslides moving sediments to the basin through the Bejaia Canyon (Harbi 1996; Harbi et al. 1999). One submarine earthquake on 21 August 1956 caused a tsunami

95

along the coast of Jijeli on the eastern coast of the Gulf of Bejaia (Roth6 1950). Seismicity and oversteepening of the shelf edge seawards of river mouths have long been held as potential triggers for turbidity currents. However, there seems little evidence for emplacement of southerly derived turbidites initiated by seismicity or shelf oversteepening at the LC06 core site - a surprising finding considering the known seismicity of the Algerian margin. This suggests that either few turbidites have been generated from the adjacent parts of the Algerian shelf and slope or that they have largely bypassed the LC06 core site. The water depth at the LC06 core site is 15m less than that recorded in the central Balearic Abyssal Plain (Rothwell 1995) and turbidity current pathways to the deep basin may bypass the LC06 location. Sand-based turbidites are generally rare in the sediment sequence - only four turbidites with sandy bases are recorded and all were emplaced during the full or Late Glacial. One of these is the basinwide megaturbidite identified by Rothwell et al. (1998), known to have a northern provenance. The other three show generally higher Sr/Ca ratios (although this may be a size-fraction effect) than the other turbidites and the geochemistry/mineralogy of the sands (typically high Ca, low Fe and, sometimes, increasing Si towards the base, and commonly containing rare to common rounded hematitic quartz) suggests a local shelf source. Potential sources for turbidites at core site LC06 are varied and include the southern European margin, Balearic Rise, Algerian shelf and slope, and the Sardinian margin. However, the paucity of proximal sandy turbidites with a neritic geochemical signature suggests that any locally derived allochthonous contribution to LC06 is small.

Conclusions The ITRAX core scanner provides rapid and non-destructive characterization of downcore geochemical distributions in unprecedented detail along with visual and X-radiographic images. Such data when gathered from turbidite-bearing sequences can provide significant information on grading and bioturbative mixing of turbidites, allow identification of geochemically distinctive marker beds, give indications of differences in provenance between beds, and confirm or query the presence of early arrivals. In the sediment sequence studied, variations in Ca/Fe generally provide a suitable proxy for the occurrence of pelagites and turbidites, allowing

96

9

9

9

9

9

R. G. ROTHWELL ET AL. discrimination of genetic units, if this parameter is coupled with others. This, together with a chronostratigraphic framework based on AMS radiocarbon dating and indirect correlation with oxygen isotope stages, demonstrates the dominance of climatic modulation of turbidite deposition in this area, despite proximity of a seismically active continental margin to the south. Grading in the Ca/Fe ratio within turbidite beds closely mirrors textural grading and allows discrimination of graded and ungraded turbidite muds without the need for grain-size determinations. Identification of fine texturally related detail, which may be invisible to the naked eye, suggests ITRAX data may allow assessment of turbidite distality in cored sequences and observations on the proximity of source areas. Semi-quantitative high-resolution geochemical profiling using the ITRAX micro-XRF core scanner has identified a few turbidite beds with distinctive geochemistry and these are potential marker beds for correlative purposes if found in other Balearic Abyssal Plain cores. The ITRAX micro-XRF core scanner data also provide useful information to confirm or query beds postulated as early arrivals of overlying turbidites. High-resolution geochemical profiling through sediment sequences containing turbidite beds provides a wealth of detailed geochemical information related to texture, mixing, provenance and diagenesis. These data need to be interpreted with caution and from an informed standpoint, however. Artefacts related to porosity, compaction and grain-size/shape changes, unevenness of the cut core surface, poor discrimination of close element peaks or instrument sensitivity issues relating to particular elements can complicate interpretation. The ITRAX data as presently recorded are best viewed as semi-quantitative in nature, and more studies are needed to compare and calibrate the records obtained with those measured using traditional destructive analysis (XRF or ICP-OES). Deposition of sand and/or silt-based turbidites on the SE Balearic Abyssal Plain during the last 50 ka is restricted to the full and Late Glacial. The thickest (largest volume) turbidites are also emplaced at this time. Turbidite emplacement frequency increases during the Holocene and is most frequent when the rate of post-glacial sea-level rise is greatest. Proximal turbidites with a southerly (Algerian margin) or northeasterly source (Sardinian

margin) are rare at the LC06 core site. Such turbidites, some perhaps seismically initiated, may bypass this location to reach the deeper parts of the Balearic Basin. The authors are very grateful to Drs A. Cattaneo and T. Courp for their reviews that significantly improved the paper. Professor P. Weaver is thanked for his editorial handling of the manuscript.

References AMBRASEYS,N.N. 1982. The seismicity of North Africa: the earthquake of 1856 at Jijeli, Algeria. Bollettino di Geofisica a Teorica ed Applicata, 24, 31-37. BOUMA, A.H. 1962. Sedimentology of Some Flysch Deposits. Elsevier, Amsterdam. CROUDACE, I.W., RINDBY, A. & ROTHWELL,R.G. 2006. ITRAX: description and evaluation of a new multifunction X-ray core scanner. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 51-63. DE MEIJER, R.J. 1998. Heavy minerals: from 'Edelstein' to Einstein. Journal of Geochemical Exploration, 62, 81 103. DI~VERCHI~RE,J., YELLES, K. E T AL. 2005. Active thrust faulting offshore Boumerdes, Algeria, and its relations to the 2003 Mw 6.9 earthquake. Geophysical Research Letters, 32, L04311, doi:10.1029/ 2004GL021646. FORNOS, J.J, FORTEZA, V., JAUME, C. ~: MARTINEZTABERNER, A. 1992. Present-day Halimeda carbonate sediments in temperate Mediterranean embayments: Fornells, Balearic Islands. Sedimentary Geology, 75, 283-293. HARm, A. 1996. La marge algdrienne orientale: resultants d'une etude par sismique r~flexion. C.R.A.A.G Publications (internal report), Algeria, 762. HARBI, A., MAOUCHE,S. & AYADI,A. 1999. Neotectonics and associate seismicity in the Eastern Tellian Atlas of Algeria. Journal of Seismology, 3, 95-104. HASCHKE, M. 2006. The Eagle III BKA system, a novel sediment core X-ray fluorescence analyser with very high spatial resolution. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 31 37. HAUG, G.H., HUGnEN, K.A., SIGMAN, D.M., PETERSON, L.C. & ROHL, U. 2001. Southward migration of the Intertropical Convergence Zone through the Holocene. Science, 293, 1304-1308. HIEKE, W. & WERNER,F. 2000. The Augias megaturbidite in the central Ionian Sea (central Mediterranean) and its relation to the Holocene Santorini event. Sedimentary Geology, 135, 205-218. HOOGAKKER, B.A.A. 2003. Climate control of allochthonous sedimentation in the deep sea. PhD thesis, University of Southampton. HUGHEN, K.A., SOUTHON, J.R., LEHMAN, S.J. & OVERPECK, J.T. 2000. Synchronous radiocarbon and climate shifts during the last deglaciation. Science, 290, 1951-1954.

ITRAX XRF SCANNING OF TURBIDITE CORES JAHN, B., DONNER, B., M/JLLER, P.J., ROHL, U., SCHNEIDER,R. & WEFER, G. 2003. Pleistocene variations in dust input and marine productivity in the northern Benguela Current: evidence of evolution of global glacial-interglacial cycles. Palaeogeography, Palaeoclimatology, Palaeoecology, 193, 515-533. JANSEN, J.H.F., VAN DER GAAST, S.J., KOSTER, B. & VAARS, A.J. 1998. CORTEX, a shipboard XRFscanner for element analyses in split sediment cores. Marine Geology, 151, 143-153. KINSMAN, D.J.J. 1969. Interpretation of Sr 2+ concentrations in carbonate minerals and rocks. Journal of Sedimentary Petrology, 39, 486 507. KRINSLEY, D. & BIERI, R. 1959. Changes in the chemical composition of pteropod shells after deposition on the sea floor. Journal of Paleontology, 33, 682-684. KRUMBEIN, W.C. 1936. Application of logarithmic moments to size frequency distributions of sediments. Journal of Sedimentary Petrology, 6, 35-47. KUHLMANN, H., FREUDENTHAL, Y., HEEMKE, P. & MEGGERS, H. 2004. Reconstruction of paleoceanography off NW Africa for the last 40,000 years: influence of local and regional factors on sediment accumulation. Marine Geology, 207, 20%234. LAMY, E., KAISER, J., NINNEMANN, U., HEBBELN, D., ARZ, H.W. & STONER,J. 2004. Antarctic timing of surface water changes off Chile and Patagonian ice-sheet response. Science, 304, 1959-1962. LAJ, C., MAZAUD, A. & DUPLESSY, 3.-C. 1996. Geomagnetic intensity and ~4C abundance in the atmosphere and ocean during the past 50kyr. Geophysical Research Letters, 23, 2045-2048. LOWE, D.R. 1982. Sediment gravity flows: II Depositional models with special reference to the deposits of high-density turbidity currents. Journal of Sedimentary Petrology, 52, 279-297. MARTINSON, D.G., PISIAS, N.G., HAYES, J.D., IMBRIE, 3., MOORE, T.C. & SnACKLETON, N.J. 1987. Age dating and the orbital theory of the ice Ages: Development of a high-resolution 0 to 300,000-year chronostratigraphy. Quaternary Research, 27, 129. MARJANAC, T. 1996. Deposition of megabeds (megaturbidites) and sealevel change in a proximal part of the Eocene-Miocene flysch of central Dalmatia (Croatia). Geology, 24, 543-546. MUTO, T. & STEEL,R.J. 2002. In defense of shelf-edge delta development during falling and lowstand of relative sea level. Journal of Geology, ll0, 421-436. OCEAN DRILLINGPROGRAM. 1987. Handbook for Shipboard Sedimentologists. ODP Technical Note, 8. PIPER, D.J.W. 1978. Turbidite muds and silts on deep sea fans and abyssal plains. In: STANLEY,D.J. & KEELING, G. (eds) Sedimentation in Submarine Canyons, Fans and Trenches. Hutchison & Ross, Stroudsburg, PA, 163-176. POSAMENTIER, H.W. & VAIL, P.R. 1988. Eustatic control on clastic deposition II - sequence and systems tract models. In: WILGUS,C.K., HASTINGS,B.S. et al. Sealevel Changes - An Integrated Approach. Society of Economic Paleontologists and Mineralogists, Special Publications, 42, i 25-154.

97

REHAULT,J.-P., BOILLOT,G. & MAUFFRETA. 1985. The Western Mediterranean Basin. In: STANLEY,D.J. & WEZEL, F.-C. (eds) Geological Evolution of the Mediterranean Basin, Raimondo Selli Commemorative Volume. Springer, Berlin, 101-129. RICHTER, T.O., VAN DER GAAST, S. ET AL. 2006. The Avaatech XRF Core Scanner: technical description and applications to NE Atlantic sediments. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 39-50. ROLLINSON, H.R. 1993. Using Geochemical Data: Evaluation, Presentation, Interpretation. Pearson Education, Upper Saddle River, NJ. ROTH~, J.P. 1950. Les s6ismes de Kherrata et la seismicit~; de l'Alg&ie. Publication Service, Cartes Geologique de l'Alg6rie, 4, 1-40. ROTHWELL, R.G. (ed.). 1995. Cruise Report: Marion Dufresne Cruise 81. Mediterranean Giant Piston Coring Transect. National Oceanographic Library, National Oceanography Centre, Southampton, UK. Unpublished report. ROTHWELL, R.G., PEARCE, T.J. & WEAVER, P.P.E. 1992. Late Quaternary evolution of the Madeira Abyssal Plain, Canary Basin. Basin Research, 4, 103-131. ROTHWELL, R.G., THOMSON, J. & K~,HLER, G. 1998. Low sealevel emplacement of a very large Late Pleistocene megaturbidite in the western Mediterranean Sea. Nature, 392, 377-380. ROTHWELL, R.G., WEAVER,P.P.E., HODK1NSON,R.A., PRATT, C.E., STYZEN, M.J. & HIGGS, N.C. 1994. Clayey nannofossil ooze turbidites and hemipelagites at Sites 834 and 835 (Lau Basin, Southwest Pacific). In: HAWKINS, J., PARSON, L. ALLAN, J. ET AL. Proceedings of the Ocean Drilling Program, 135, 101-130. RUTTEN, A., DE LANGE, G.J., ZIVERI, P., THOMSON,J., VAN SANTVOORT, P.J.M., COLLEY, S. t~ CORSELLI, C. 2000. Recent terrestrial and carbonate fluxes in the pelagic eastern Mediterranean; a comparison between sediment trap and surface sediment. Palaeogeography, Palaeoclimatology and Palaeoecology, 158, 197-213. SHACKLETON, N.J. 1987. Oxygen isotopes, ice volume and sea level. Quaternary Science Reviews, 6, 183190. SHANMUGAM,G. • MOIOLA,R.J. 1982. Eustatic control of turbidite deposition and winnowed turbidites. Geology, 10, 231-235. SHANMUGAM,G. & MOIOLA, R.J. 1984. Eustatic control of calciclastic turbidites. Marine Geology, 56, 273278. SKINNER, L.C. & MCCAvE, I.N. 2003. Analysis and modelling of gravity- and piston coring based on soil mechanics. Marine Geology, 199, 181-204. STOLE, H.M. & SCHRAG, D.P. 2000. Coccolith Sr/Ca as a new indicator for coccolithophorid calcification and growth rate. Geochemistry, Geophysics and Geosystems, 1, paper 1999GC000015 (http:// www.agu.org/pubs/crossref/2000.../ 1999GC000015.shtml; date of citation 25.2.05). STOW, D.A.V. & SHANMUGAM,G. 1980. Sequence of structures in fine-grained turbidites: comparison

98

R . G . ROTHWELL ET AL.

of recent deep sea and ancient flysch sediments. Sedimentary Geology, 25, 23-42. STUIVER, M., REIMER, P.J. ET AL. 1998. INTCAL98 radiocarbon age calibration 24,000-0 cal BP. Radiocarbon, 40, 1041-1083. TEN HAVEN, H.L., DE LEEUW, J.W., SCHENK, P.A. & KLAVER, G.T. 1987. Geochemistry of Mediterranean sediments: Bromine/organic carbon and uranium/organic carbon ratios as indicators for different sources of input and post-depositional oxidation respectively. Organic Geochemistry, 13, 255-261. THOMAS, G. 1976. Mise en evidence de d+crochements dextre Est-Ouest d'fige Quaternaire en Alg6rie Nord Occidentale. Comptes Rendues de l'Academie des Sciences, Paris, 283, 893-896. THOMSON,J., CROUOACE,I.W. & ROTHWELL,R.G. 2006. A geochemical application of the ITRAX scanner to a sediment core containing eastern Mediterranean sapropel units. In: ROTHWELL,R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 65-77. THOMSON,J., CRUDELI,D., DE LANGE, G., SLOMP, C.P., ERBA, E., CORSELLI C. & CALVERT, S.E. 2004. Florisphaera profunda and the origin and diagenesis of carbonate phases in eastern Mediterranean sapropel units. Paleoceanography, 19, PA3003 doi: 10.1029/2003PA000976. THOUVENY, N., MORENO, E., DELANGHE,D., CANDON, L., LANCELOT,Y. & SHACKLETON,N.J. 2000. Rock magnetic detection of distal ice-rafted debris: clues for the identification of Heinrich layers on the Portuguese margin. Earth and Planetary Science Letters, 180, 61-75.

VAIL, P.R., MITCHUM,R.M., JR ETAL. 1977. Seismic stratigraphy and global changes in sealevel. In: PAYTON, C.E. (ed.) Seismic Stratigraphy- Application to Hydrocarbon Exploration. AAPG Memoir, 26, 49-212. VAN DER WEIJDEN,C.H. 2002. Pitfalls of normalisation of marine geochemical data using a common divisor. Marine Geology, 184, 167-187. WEAVER,P.P.E. 1983. An integrated stratigraphy of the upper Quaternary of the Kings Trough flank area, NE Atlantic. Oceanologica Acta, 6, 451-456. WEAVER, P.P.E. & KUIJPERS,A. 1983. Climatic control of turbidite deposition on the Madeira Abyssal Plain. Nature, 306, 360-363. WEAVER, P.P.E. & ROTHWELL, R.G. 1987. Sedimentation on the Madeira Abyssal Plain over the last 300,000 years. In: WEAVER,P.P.E. & THOMSON, J. (eds) Geology and Geochemistry of Abyssal Plains. Geological Society, London, Special Publications, 31, 71-86. WEAVER, P.P.E., ROTHWELL, R.G., EBBING,J., GUNN, D. & HUNTER, P.M. 1992. Correlation, frequency of emplacement and source directions of megaturbidites on the Madeira Abyssal Plain. Marine Geology, 109, 1-20. WILSON, T.R.S., THOMSON,J., COLLEY,S., HYDES, D.J., HIGGS, N.C. & SORENSEN, J. 1986. Early organic diagenesis: the significance of progressive subsurface oxidation fronts in pelagic sediments. Geochimica et Cosmochimica Acta, 49, 611-822. WYNN, R.B., WEAVER,P.P.E., MASSON,D.G. & STOW, D.A.V. 2002. Turbidite depositional architecture across three interconnected deep-water basins on the northwest African margin. Sedimentology, 49, 669-695.

Colour logging as a tool in high-resolution palaeoceanography M. R O G E R S O N 1'2, P. P. E. W E A V E R 1, E. J. R O H L I N G 1, L. J. L O U R E N S 3, J. W. M U R R A Y 1 & A. H A Y E S 4

1National Oceanography Centre, European Way, Southampton S014 3ZH, UK 2present address. Department of Geography, University of Hull, Cottingham Road, Hull HU6 7RX, UK (e-mail: [email protected]) 3Faculty of Geoscience, University of Utrecht, Budapestlaan, 4, 3584 CD, Utrecht, The Netherlands 4Department of Geology, Royal Holloway College, Egham, Surrey TW20 OEX, UK Abstract: Colour and diffuse reflectance records can be used to develop astronomically tuned age models for long sediment cores. Here, we present high-resolution (1 mm) colour records from a sediment core from the western Gulf of Cadiz of SW Spain (D13892), spanning the last deglaciation, in parallel with stable isotope (~5180) and sea surface temperature (SST) proxy data. The age model is based on ~5180 stratigraphy complemented by five atomic mass spectroscopy (AMS) radiocarbon datings. We find good comparison between the colour record of D13892 and the GISP2 oxygen isotope series (R 2 = 0.81), which strongly suggests that the sediment colour reflects the state of the climate. As sediment colour variability has previously been found to be diagnostic of changes in mineralogical/chemical composition, we relate the causes of the colour variability in D13892 to changes in the local particle flux, and support these observations with data from core-logging X-ray fluorescence (XRF) analyses. The colour and XRF logger records for D13892 suggest that the last glaciation and Younger Dryas were characterized by an enhanced supply of terrigenous detritus into the western Gulf of Cadiz. Cyclicities with wavelengths of 607 and 1375 years are recognized in the colour records for the Holocene. This cyclicity also relates to variability in detrital supply, with an important eolian component implied by enrichment in hematite during cycle maxima.

Colour has been considered to be diagnostic of the composition of soils for several decades (Goddard et al. 1948), and this concept is now increasingly being applied to marine sediments (e.g. Mix et al. 1992). Geochemical investigation of sediment core material has shown that aspects of colour (e.g. lightness) are reliable indicators of important sedimentological components such as carbonate, free and bound iron, a variety of Fe minerals (e.g. hematite, pyrite, goethite) and clay. The dominant redox state of iron, be it within minerals or as ions (particularly within the lattice of clay minerals), can also be reliably estimated from colour (Mix et al. 1992, 1995; Nagao & Nakashima 1992; Balsam & Deaton 1996; Giosan et al. 2002). Greyscale (lightness) has successfully been used for initial chronostratigraphic model development within long sediment cores, especially those collected by the Ocean Drilling Program (Ortiz et al. 1999; Grutzner et al. 2002). It is rapidly and easily measured and dominantly controlled by the carbonate content, although total organic carbon (TOC) is also important, which is thought to be regulated by orbitally modulated insolation (Hays & Peruzza 1972; Volat et al.

1980; Ortiz et al. 1999). Recently, a combined investigation of geochemistry and sediment colour has shown systematic variability in mineralogical and carbonate content, both predicted by colour, on Milankovitch and sub-Milankovitch (c. 1.5ka) wavelengths (Grutzner et al. 2002). This indicates a potential for the use of colour as a means of developing initial chronostratigraphic models for high-resolution records from the late Quaternary. Here, we investigate this potential using sediment colour, geochemical, stable isotope and planktonic foraminiferal abundance records for a sediment core (D13892) from the western Gulf of Cadiz (SW Spain), with an independent age model based on five radiocarbon datings and 6180 isotope stratigraphy.

Study area

The Gulf of Cadiz (GoC) lies to the SW of the Iberian Peninsula, immediately west of the Strait of Gibraltar (Fig. 1). The exchange of Mediterranean and Atlantic water through the Strait of Gibraltar drives the formation of two currents that are present within the GoC

From: ROTHWELL, R.G. 2006. New Techniquesin Sediment Core Analysis. Geological Society, London, Special Publications, 267, 99-112. 0305-8719/06/$15.00 9 The Geological Society of London 2006.

100

M. R O G E R S O N ET AL.

9

9

o

e-. , ,...~ r o,o

9

9

9 ...o

8

NI .,..~

a

COLOUR LOGGING IN PALAEOCEANOGRAPHY (Bryden et al. 1988). The shallow, eastward component of this flow compensates for water lost from the Mediterranean by evaporation and by the deep, westward outflow (the Mediterranean Outflow, MO). Within the GoC, the Atlantic Inflow current flows along the shelf from NW to SE, elongating the sedimentary deposits formed at the mouths of the Iberian rivers along the northern and eastern margins of the GoC (Bryden et al. 1988; Lopez-Galindo et al. 1999). The subsurface Mediterranean Outflow (MO) is a plume of relatively warm and saline dense water that, after its exit from the Strait of Gibraltar, passes along the GoC slope between 500 and 1500m water depth. It interacts with the sea floor from Cape Spartel in the SE to Cape St Vincent in the N W (Kenyon & Belderson 1973; Ambar & Howe 1979; Iorga & Lozier 1999), where it profoundly influences the nature of sedimentation. Material is transported along the slope from the SE to the N W (Kenyon & Belderson 1973). In the SE, where the flow is strongest, the path of the current is characterized by abrasion surfaces and sand ribbons, which fine northwestwards as the flow decelerates, to become sandy and then muddy sediment drifts (Kenyon & Belderson 1973). Suspended material is transported in suspension out into the western GoC, to areas where the M O no longer interacts with the sea floor (Ambar et al. 2002). On the upper slope of the GoC, where the flow is at least partially in contact with the sea floor, the MO plume is characterized by significantly higher suspended particulate matter (SPM) content than the Atlantic water it is passing through (Ambar et al. 2002). This SPM comprises both mineral and biogenic grains, the biogenic component being dominantly calcareous. Further offshore, the SPM content of water in the MO plume is significantly decreased, and more comparable to Atlantic water than to the Mediterranean water found close to the shelf (Ambar et al. 2002). There, SPM is generally less than 10 lam in diameter and the biogenic component is dominantly siliceous (Ambar et al. 2002). This indicates that the majority of the SPM transported by the Mediterranean Outflow is deposited close to the region of supply where the plume interacts with the sea floor, and relatively little is transported into the western GoC today. Variability in the supply of sediment to the MO plume and in its capability to transport SPM (primarily its internal turbulence) would be reflected by variability in detrital material supply at the core location. The GoC region is arid and characterized by strong and stable winds, especially the easterly Levanter and northerly trade winds (Dorman et al. 1995). Today, Levanter winds trap air

101

pollution and dust close to the ground and cause visibility to decrease by as much as 30%. During the summer, the trade winds are strong, stable and active for more than 50% of the time at the city of Cadiz (Fiuza et al. 1982). It is likely that the western G o C experiences a significant supply of aeolian detritus via these wind regimes. If the local wind field has been variable in the past, it is anticipated that this would have some impact on sedimentation at the core location.

Material and methods Core D13892 is a 2m long kasten core recovered from the western Gulf of Cadiz, from a water depth of 1500 m, during RRS Discovery cruise 249. This location lies beyond the region where the MO interacts with the sea floor, and there is no indication of either current or turbidite activity in this core. Five radiocarbon datings have been performed on more than 6 mg of planktonic foraminiferal tests (Globigerina bulloides, Globigerinoides ruber, Globigerinoides sacculifer) picked from the >150/am fraction. The radiocarbon analyses were undertaken via the NERC Radiocarbon Laboratory (NERC-RCL), at the University of Arizona NSFAMS facility. For the planktonic foraminiferal abundance study, samples were disaggregated in demineralized water, washed and sieved to remove all material finer than 150 lain. Where necessary, samples were then split into suitable aliquots of at least 300 individuals for identification according to the taxonomy of Hemleben et al. (1989). The data are presented as percentages of total planktonic foraminiferal number. Planktonic foraminifera assemblages were analysed by Artificial Neural Network (ANN) to produce estimates of summer and winter sea surface temperature (SST) (Malmgren et aI. 2000). The training set used was based on core top samples from both the North Atlantic and the Mediterranean, as the GoC lies between these two basins. The specimens selected for stable isotope analyses are washed and sonicated in methanol to remove surface contamination. For each sample stable isotope analyses were carried out on between seven and 15 individuals of dextral coiling Neogloboquadrina pachyderma ranging in sizes from 150 to 212 ~tm. Stable isotope analyses were carried out using a Europa Geo 2020 mass spectrometer with individual acid-dosing method at the SOES facility in Southampton. The carbon and oxygen isotope ratios are expressed as ~5values, in per mils (%o), relative to the Vienna Peedee Belemnite standard (Coplen ! 988, 1994). Colour analysis and XRF scanning

Core-section surfaces were first smoothed with a glass slide and then imaged using a GEOSCAN colour

102

M. ROGERSON E T AL.

line-scan camera. This device uses a 3 x 1024 pixel CCD (charge-coupled device) array that measures absolute red, green and blue (RGB) colour intensities to produce a finely pixelated image of the sediment surface. A value of 255 in each wavelength band is calibrated to a standard white. Average RGB values are measured for each pixel (14~tm2), which are then combined into bitmap files for each core section. The bitmap files are compiled in the Geoscan software (www.geotek.co.uk/ site/scripts/module.php? webSubSectionID = 28) to form a continuous image for the whole core, into which subsurface depth data are incorporated. A panel of the image covering a whole core, core section or part of a section is then selected manually in the Geoscan software. Downcore resolution within this panel is set manually at any level up to a best resolution of 100 ~tm. Data for D13892 have been produced at 1 mm resolution. Average values for red (450490mm), green (500-550mm) and blue (590-660 mm) are automatically calculated for each I mm interval across the width of the selected interval. Numerical RGB data are then linked to the depth of the top of the measured interval from the core top in mm and complied as downcore reflectance records for each of the three wavelength bands. RGB reflectance values may be expressed either as mean reflectance in each of the wavelength bands ('absolute reflectance') or as the relative contribution of each wavelength band to the total reflectance at that depth ('relative reflectance'). Data are then exported from the Geoscan software as an ASCII file. The uppermost part of the core (subsequent to 10 ka BP) was selected for spectral analysis using the ARfit package (Paillard et al. 1996), to give insight into cyclicities underlying the colour variability. Additional analyses were performed using a Minolta Spectrophotometer. Ten samples were taken from the core, oven dried and ground to a fine powder. This powder was poured into a well cut into a sheet of cardboard and the surface smoothed prior to analysis, to remove errors due to differences in water content and surface unevenness. Spectral data were pretreated by differentiation to increase the variance (as proposed by Giosan et al. 2002) and is presented as curves showing the first differential of reflectance across the visible light spectrum. Here we use the lightness (L), chroma (C) and hue (H) system to parameterize colour (Giosan et al. 2002). Lightness has been shown in numerous studies to be sensitive to the carbonate composition (Mix et al. 1992, 1995; Nagao & Nakashima 1992; Giosan et al. 2002). Hue and chromaticness have been shown to be sensitive to the content and redox state of iron (Giosan et al. 2002). However, the response of colour to chemical and mineralogical changes is non-linear, and quantitative estimation of sediment composition is difficult. Here, colour data are supported by chemical data produced by X-ray fluorescence (XRF) analysis.

The core sections were analysed with the Cox Analytical Systems Itrax XRF core logger at Southampton Oceanography Centre. This system uses capillary optics to focus an X-ray beam produced from a molybdenum filament onto the surface of a split core. A stepper motor moves the core in 4001.tm increments, and at each increment the core is irradiated. The atomic fluorescences produced are collected by an analyser close to the surface of the core and the intensity of fluorescence peaks at registered wavelengths are recorded for each step. Thus, a 400 ~tm resolution record is produced of relative elemental abundances. Here, only data for Fe and Ca are reported. Spectral

analysis

Power spectra were obtained by using the CLEAN transformation of Roberts et al. (1987) in the ARfit package. The CLEAN technique is an iterative deconvolution method that involves trial fitting curves of a range of different frequencies to the data series. The height of spectral peaks therefore represent the probability that a significant periodicity of that wavelength exists in the data. For the determination of errors associated with the frequency spectra of the CLEAN algorithm we applied the Monte Carlo based method developed by Heslop & Dekkers (2002). The 97.5 and 99.5% significance levels for the Monte Carlo spectra were determined by: (1) 10% red noise addition (i.e. Control parameter = 0.1); (2) 'CLEAN gain factor' (Heslop & Dekkers 2002) of 0.1; (3) 500 CLEAN iterations; (4) the sample spacing for the interpolated data series (dt) was set at 5 years; and (5) 500 simulation iterations.

Chronostratigraphic framework T h e r a d i o c a r b o n dates were calibrated to calendar ages using the O X C A L p r o g r a m (www.rlaha.ox. a c . u k / o r a u / o x c a l . h t m l ) , a s s u m i n g a 400 year reservoir age. Figure 2 shows the position of the calibrated r a d i o c a r b o n dates, the probability distribution of their calibrations to calendar age a n d the 95% confidence interval o f possible a g e d e p t h correlations. A n initial c h r o n o s t r a t i g r a p h i c m o d e l is derived ( s h o w n in black), which represents the s m o o t h e s t line o f best fit between peaks o f high probability. It is very likely that m a j o r events, such as the w a r m i n g at the t e r m i n a t i o n o f the last glaciation a n d the start a n d end o f the Y o u n g e r Dryas, are s y n c h r o n o u s between this l o c a t i o n a n d the wider G o C a n d Iberian m a r g i n (Bard et al. 1987, 2000; Lebreiro et al. 1997; Cayre et al. 1999; Shacklet o n et al. 2000; T h o m s o n et al. 2000; C a c h o et al. 2001; M o r e n o et al. 2002; S c h 6 n f e l d et al. 2003), a n d so with the N o r t h Atlantic a n d Greenland in general. T h e c h r o n o s t r a t i g r a p h i c m o d e l

COLOUR LOGGING IN PALAEOCEANOGRAPHY

103

Fig. 2. Chronostratigraphy for D13892. Unbroken line indicates initial chronostratigraphic model, broken line indicates chronostratigraphy after tuning to GISP2 chronology (see Fig. 3).

for D13892 is therefore fine tuned by correlating these events in the D13892 6180 record to the GISP2 ~180 series, and these relationships are supported by the SST record. Figure 3 shows the SST and oxygen isotope data on this final chronostratigraphic model relative to the GISP2 isotope series, with the correlation points marked. The adjustment made to the chronostratigraphic framework during the fine-tuning process is generally small (individual corrections were less than 500 years - see Tables 1 and 2). The final chronostratigraphic model used is shown as the broken line in Figure 2.

Results 18

Figure 4d-f shows RGB and ~ ON.pachyderma(d) plots for D13892. As the red, green and blue plots track one another closely, the colour variation is mainly related to lightness (L). Figure 5 shows the L, C and H records for D13892. Close correspondence of L (Fig. 5b) and absolute

reflectance (Fig. 5a) in the D13892 is apparent. L in D 13892 correlates to the individual absolute reflectance records with R 2 in excess of 0.99 (N = 1814), which confirms that grey-scale reflectance is the underlying control on these data. Spectrophotometry (Fig. 6) indicates that most of the reduction in L between the minimum lightness (glacial) and maximum lightness (Holocene) parts of this core is attributable to a decline in reflectance in the wavelength band associated with carbonate (Giosan et al. 2002). The close relationship found between L and carbonate content in many other sediment cores (Nagao & Nakashima 1992; Mix et al. 1995; Balsam & Deaton 1996; Hughen et al. 1996; Ortiz et al. 1999; Giosan et al. 2002) can therefore be inferred here. The bulk colour changes in the core therefore represent the ratio between carbonate and detrital material supplied to this 1500m-deep location. The relative intensity of each of the wavelengths varies slightly along core (Fig. 4a-c), indicating that the colour variation also has a small hue/ chroma component. Figure 5d & e show chroma

104

M. ROGERSON E T AL.

O

o 0

o

o

N

,....,

3 g

0

r~

Z

Z

<

..~

COLOUR LOGGING IN PALAEOCEANOGRAPHY Table 1. Radiocarbon datings used in development of D13892 chronology. The OXCAL calibration system was used for conversion of conventional radiocarbon dates to calendar years Sample depth (cm)

Conventional radiocarbon age

l~r

Predicted calendar age

8 56.7 71 110 173

5647 10429 10868 10868 14353

45 56 59 58 76

6010 11 210 12040 12660 16800

and hue, respectively. Close correspondence is found between chromaticness (Fig. 5d) and the relative reflectiveness (Fig. 5c) of the red wavelengths (R 2 = 0.74, N = 1813).

Discussion Colour in D13892 as a tool f o r chronostratigraphy

There is good agreement between changes in 18 (~ ON. pachyderrl~(d) and changes in lightness, with increasing (~'~ .... (d) values generally correlating with relative darkening. This suggests that the history of sediment supply to this location is dominated by general North Atlantic climate changes. There is an even more striking agreement between the reflectivity and the ~5~80 GISP2 record which, in places, appear to correlate on millennial-centennial timescales. The L and GISP2 6180 curves are highly significantly correlated, with R 2 = 0.81 (N = 413). Colour logging therefore has great potential for the

105

development of initial sediment-core chronostratigraphies. However, there are discrepancies as well. For example, a discrepancy exists between the structures of the colour and GISP2 (~180 records concerning the expression of the warmest part of the B611ing-Allerod (c. 14.6 ka BP), which seems under-represented in the colour record of D13892. Origin o f colour variability

Figure 7 shows L for D13892, the GISP2 6~80 series, and chemical data produced by X R F logging of D13892 (Fe/Ca). Fe/Ca can be viewed as the detrital:carbonate ratio, and this curve agrees well with the reflectivity (L) record (R 2 = 0.65, N = 1814), giving strong support to our interpretation of the latter as a proxy for carbonate content. High peaks in Fe/Ca occur at the glacial terminations la and lb. In the case of T I a, the start of the Fe/Ca peak coincides with the deglaciation in 5180 of GISP2, whereas the deglaciation in the colour record of D13892 appears to be delayed. The sharp peak in Fe/Ca values in D13892 probably reflects elevated detrital supply during the deglaciation. A period of high detrital supply would be anticipated to cause depressed reflectivity values during its duration, as is recognized here at Tla. It is therefore probable that the apparent delay of T l a in the colour record of D13892 reflects the impact of ice melt, drainage basin rearrangement and/or changes in the behaviour of the local oceanic currents rather than the state of the climate. Where similar detrital pulses occur in other locations at the time of glacial terminations, the effect on the colour record will cause a potential error for age models developed from astronomical tuning. The Younger Dryas and glacial parts of the D13892 record exhibit lower lightness, higher

Table 2. Tie points used in the development ofD13892 chronology Tie point depth (cm)

Event

Calibrated radiocarbon age

Age in GISP2 chronology

33.1 61.7 100 136.7 146 152 166 170 176.3

Warming subsequent to 8.2 ka event Termination lb Termination of B611ing-Allerod Termination la Local 6180 minimum Local 61So maximum Local 6180 minimum Local 6180 minimum Local 6180 maximum

8577 11349 12594 14416 14971 15333 16171 16419 16 799

8 100 11700 13000 14600 15400 15800 16500 16850 17100

106

M. ROGERSON E T AL.

.~r.~ ZaZ

oo

t"q

O

~8

~,.=I r .,,,,a

0

~.~ o

o

~

.,..~

.-d~

og -~. m

COLOUR LOGGING IN PALAEOCEANOGRAPHY

107

Fig. 5. Raw and transformed colour data for D13892. (a) Absolute intensity in green wavelengths; Ca) reflectance (L*); (e) relative intensity (red); (d) chromaticness (C); and (e) hue (H~

Fe/Ca (Fig. 7) and significantly higher accumulation rates than the Holocene (Fig. 2). The number of planktonic foraminiferal tests per gram (dry wt) is low during periods with high rates of accumulation. In the Holocene, the number of tests per gram is of the order of 1000-1600, whereas during the latter part of the Younger Dryas, when deposition was most rapid (Fig. 2), the number of tests per gram is found to be as low as about 200. This suite of observations is more consistent with a variable rate of supply of detrital grains to the core location than with variable rate of carbonate production, which would be reflected in reduced accumulation rate during periods of low carbonate content. Cold periods thus exhibit higher detrital supply than warm periods in the D13892 record. Some variation in the rate of carbonate production cannot be ruled out, but the rate of carbonate supply on the Portuguese

margin is thought to have been relatively constant over the past 140 ka (Thomson et al. 1999). During the Holocene part of D13892, the > 150 lam sediment fraction consists almost exclusively of planktonic foraminiferal tests. Before the Holocene, and particularly during the Younger Dryas, abundant terrigenous grains are found in this fraction as well. These are mainly quartz and lithic clasts, some of them with dark coatings, with abundant mica and some black, oxidized wood fragments. Similar grain assemblages are found in samples from the lowermost part of the MO contourite drift (Gil Eanes drift, Fig. 1) (Habgood et al. 2003; Rogerson 2003). D 13892 lies further downstream in the MO pathway than this region, and the petrological similarity between the deposits reflect an important role played by the MO in supplying mineral grains to the western GoC. The largescale colour variablity in D13892 therefore

I08

M. ROGERSON ET AL. 0.06 "Background" Hotocene

//

o.05~ /

Enhanced haernatite 0.04

content

"-a .I

-a

/ /

0.03 - -

0.02 - Glacial

Holocene dilution layer

0.01 - -

Haematite

Carbonate

0

I 400

I

i

450

l 500

[

i

~

550 Wavelength (nm)

I 600

Fe2+/Fe3* in day

i

I 650

I

1 700

Fig. 6. First differential of reflectance of representative samples from D13892, derived from Minolta spectrophotometer measurements. Interpretation of primary controls on reflectance intensities follows Giosan et al. (2002). Characteristic peaks in reflectance spectra, identified by statistical analysis of sediment core material analysed by colour and geochemical methods, are found for carbonate (400-500 nm), hematite (500600 nm) and the reduced free iron (600-700 nm).

reflects variable supply of terrigenous grains to this location through time, with the MO probably playing a significant role as a transportation pathway. Colour variability in the H o l o c e n e

The Holocene parts of D 13892 exhibit a series of cyclic lightness minima (Fig. 4d-f). Spectral analysis shows spectral peaks at 1375 and 607 years in the Holocene part of this record (younger than 10 ka BP), which are significant at the 97.5 % confidence level (Fig. 8). These minima are found across all three absolute reflectance records (Fig. 4d-f), and thus in L, and coincide with maxima in the red and minima in the blue relative reflectance records (Fig. 4a-c). The high chromaticness (Fig. 5d) of these layers will reflect changes in the composition of the sediment itself, probably

the enrichment and redox state of the iron present. Several Holocene lightness minima coincide with local maxima in Fe/Ca (Fig. 7), confirming that these represent carbonate-depleted/iron-enriched layers. These lightness minima therefore reflect short duration (c. 100-500 years) periods of either increased supply of iron-rich mineral detritus or decreased carbonate productivity. Spectrophotometry indicates that these minima have increased reflectance in the wavelengths associated with hematite relative both to the background Holocene and the last glaciation (Fig. 6), which would generally be associated with increased eolian input (Balsam et al. 1995). This suggests that airborne dust may be an important factor governing the Holocene cyclic variability, implicating variability in either atmospheric turbulence or aridity in the region surrounding the GoC during this period.

COLOUR LOGGING IN PALAEOCEANOGRAPHY

109

r; 2 t"'t

O

5

O

0.2 el)

O

-6

l l0

M. ROGERSON E T AL.

I...........................................................................................................................................................................................................................

0.400 0.350 0.300 0.250 n

0.200 "~ O

--

m

1375

99.5%

607

0.150 0.100

-- 97.5% 0.050

10000

1000

100

0.000 10

Period (years) Fig. 8. ARfit spectral analysis of absolute intensity (green) record for the Holocene off D13892:97.5 and 99.5% confidence intervals are shown.

Conclusions

Colour logging is a useful tool that may be used for producing high-resolution chronostratigraphic frameworks by correlation with well-constrained climatic records. It has potential to be particularly useful as a method of developing initial chronostratigraphic information for a sediment core in order to develop a sampling strategy for more conventional analyses, as continuous colour logging is a cheap, rapid, non-destructive method of producing very high-resolution data for sediment cores. The western Gulf of Cadiz experienced an enhanced supply of detrital grains during the Younger Dryas and the latter part of the last glaciation. Greatly enhanced detrital supply is found in the western GoC during the glacial terminations (la and l b). The detritus is petrologically similar to sand found on the Gil Eanes Drift, indicating that it was probably supplied by the Mediterranean Outflow. Enhanced detrital supply was also found during cyclic periods within the Holocene, with spectral peaks at 607 and 1375 years. Increased hematite content in diluted layers implies increased aeolian dust input as one factor in controlling this cyclicity.

We would like to thank G. Rothwell and I. Croudace for allowing access to the BOSCORF Itrax scanner, and J. Thomson for his help in interpreting the data. S. Cooke, M. Cooper and M. Bolshaw are thanked for performing the stable isotope analyses. NERCRCL is thanked for providing funding for the radiocarbon dating (project 949.1201), and for performing the analyses. Reviewers A. Nederbragt and L. Giosan are thanked for their comments, which improved the manuscript considerably.

References

AMBAR, I. & How~, M.R. 1979. Observations of the Mediterranean Outflow II - The deep circulation in the vaicinity of the Gulf of Cadiz. Deep-Sea Research, 26A, 555-568. A ~ A R , I., SERRA,N. ET AL. 2002. Physical, chemical and sedimentological aspects of the Mediterranean outflow offIberia. Deep-Sea Research II, 49, 4163M177. BALSAM,W.L. & DEATON,B,C. 1996. Determining the composition of late Quaternary marine sediments from NUV, VIS and NIR diffuse reflectance spectra. Marine Geology, 134, 31-55. BALSAM, W.L., OTTOBLEISNER,B.L. & DEATON,B.C. 1995. Modern and last glacial maximum eolian sedimentation patterns in the Atlantic Ocean interpreted from sediment iron-oxide content. Paleoceanography, 10, 493-507.

COLOUR LOGGING IN PALAEOCEANOGRAPHY BARD, E., ARNOLD, A.J., MAURICE, P., DUPRAT, J., MOVES, J. & DUPLESSY,J.C. 1987. Retreat velocity of the North Atlantic polar front during the last deglaciation determined by 14C accelerator mass spectrometry. Nature, 328, (6133), 791-794. BARD, E., ROSTEK, F., TURON, J.L. & GENDREAU, S. 2000. Hydrological impact of Heinrich events in the subtropical Northeast Atlantic. Science, 289, 1321-1324. BRYDEN, H.L., BRADY, E.C. & PILLSBURY,R.D. 1988. Flow through the strait of Gibraltar. In: ALMAZAN, J.L., BRYDEN,H.L., KINDER,T. & PARILLA,G. (eds) Seminario sobre la Oceanografia fisica del Estracho de Gibraltar. SECEG, Madrid. CACHO, I., GRIMALT, J.O., CANALS, M., SBAFFI, L.,

SHACKLETON, N.J., SCH(~NEELD, J. & ZAHN, R. 2001. Variability of the western Mediterranean Sea surface temperature during the last 25,000 years and its connection with the northern hemisphere climatic changes. Paleoceanography, 16, 40-52. CAYRE, O., LANCELOT,Y., V1NCENT,E. & HALL, M.A. 1999. Palaeoceanographic reconstructions from planktonic foraminifera off the Iberian Margin: temperature, salinity and Heinrich events. Paleoceanography, 14, 384-396. COPLEN, T.B. 1988. Normalization of oxygen and hydrogen isotope data. Chemical Geology (Isotope Geoscience Section), 72, 293-297. COPLEN, T.B. 1994. Reporting of stable hydrogen, carbon, and oxygen isotopic abundances. Pure and Applied Chemistry', 66, 273-276. DORMAN, C.E., BEARDSLEY, R.C. & LIMEBURNER, R. 1995. Winds in the Strait of Gibraltar. Quarterly Journal of the Meteorological Society, 121, 19031921. FIUZA, A.F.G., MACEDO, M.E. & GUERREIRO, M.R. 1982. Climatological space and time variation of the Portuguese coastal upwelling. Oceanologica Acta, 5, 31-40. GIOSAN, L., FLOOD, R.D. & ALLER, R.C. 2002. Palaeoceanographic significance of sediment color on western North Atlantic drifts: I. Origin of color. Marine Geology, 189, 25-41. GODDARD,M.B., OVERBECK,R.M., ROVE,O.N., SINGEWALD, J.T. & TRASK,P.D. 1948. Rock-color Chart. National Research Council, Washington, DC. GRUTZNER, J., GIOSAN, L. E T AL. 2002. Astronomical age models for Pleistocene drift sediments from the western North Atlantic (ODP Sites 10551063). Marine Geology, 189, 5 23. HABGOOD, E., KENYON, N.H., MASSON, D.G., AKHMETZHANOV, A., WEAVER, P.P.E., GARDNER, J. & MULDER, T. 2003. Deep-water sediment wave fields, bottom current sand channels and gravity flow channel-lobe systems: Gulf of Cadiz, NE Atlantic. Sedimentology, 50, 1-27. HAYS, J.D. & PERUZZA, A. 1972. The significance of calcium carbonate oscillations in the eastern equatorial Atlantic deep-sea sediments for the end of the Holocene warm interval. Quaternary Research, 2, 355-362. HEMLEBEN,C., SPINDLER,M. & ANDERSON,0. R. 1989. Modern Planktonic Foraminifera. Springer, Berlin.

II l

HERNANDEZ-MOLINA,J., LLAVE, E. e r AL. 2003. Looking for clues to palaeoceanogrpahic imprints: A diagnosis of the Gulf of Cadiz contourite depositional systems. Geology', 31, 19-22. HESLOP, D. & DEKKERS,M. 2002. Spectral analysis of unevenly spaced climatic time series using CLEAN: signal recovery and derivation of significance levels using a Monte Carlo simulation. Physics of the Earth and Planetary' Interiors, 130, 103-116. HUGHEN, K.A., OVERPECK, J.T., PETERSON, L.C. & TRUMBORE, S. 1996. Rapid climate changes in the tropical Atlantic region during the last deglaciation. Nature, 380, 51-54. IORGA, M.C. & LOZIER, M.S. 1999. Signatures of the Mediterranean outflow from a North Atlantic climatology 1. Salinity and density fields. Journal of Geophysical Research, 104, 25,985-26,009. KENYON, N.H. & BELDERSON,R.H. 1973. Bed forms of the Mediterranean Undercurrent observed with side-scan sonar. Sedimentary Geology, 9, 77-99. LEBREIRO, S.M., MORENO, J.C., ABRANTES, F.F. & PFLAUMANN, U. 1997. Productivity and paleoceanography on the Tore Seamount (Iberian margin) during the last 225kyr: Foraminiferal evidence. Paleoceanography, 12, 718-722. LOPEZ-GALINDO, A., RODERO, J. & MALDONADO, A. 1999. Subsurface facies and sediment dispersal patterns: southeastern Gulf of Cadiz, Spanish continental margin. Marine Geology, 155, 83-98. MALMGREN, B.A., KUCERA, M., NYBERG, J. 8z WALLBROCK, C. 2000. Comparison of statistical and artificial neural network techniques for estimating past sea surface temperatures from planktonic foraminifera census data. Paleoceanography, 16, 520-530. MIx, A., HARRIS,S. (~ JANECEK,T.R. 1995. Estimating lithology from nonintrusive reflectance spectra: Leg 138. In: PISIAS, N.G., MAYER, L.A., JANECEK,T.R., PALMER-JULSON,A. & VAN ANDEL, T.H. (eds) Proceedings of the Ocean Drilling Program, Scientific Results, 138. Ocean Drilling Program, College Station, TX, 413-427. MIX, A.C., RUGH, W., PISIAS,N.G., VEIRS, S. & PARTY, L.S. 1992. Color reflectance spectroscopy: A tool for rapid characterization of deep-sea sediments. In: Proceedings of the Ocean Drilling Program, Initial Reports, 138. Ocean Drilling Program, College Station, TX, 67-76. MORENO, E., THOUVENY,N., DELANGHE,D., MCCAVE, I.N. & SHACKLETON,N. 2002. Climatic and oceanographic changes in the Northeast Atlantic reflected by magnetic properties of sediments deposited on the Portuguese margin during the last 340 ka. Earth and Planetary Science Letters, 202, 465-480. NAGAO, S. (J~ NAKASHIMA,S. 1992. The factors controlling vertical color variations of North Atlantic Madeira Abyssal Plain sediments. Marine Geology, 109, 83-94. ORTIZ, J., MIX, A., HARRIS, S. t~; O'CONNELL, S. 1999. Diffuse spectral reflectance as a proxy for percent carbonate content in North Atlantic sediments. Paleoceanography, 14, 171-186. PAILLARD,D., LABEYRIE,L. & YIOU, P. 1996. Macintosh program performs time-series analysis. Los Transactions of the American Geophysical Union, 77, 379.

112

M. ROGERSON ET AL.

ROBERTS, D.H., LEHAR, J. & DrUSHER, J.W. 1987. Time series analysis with clean - Part one Derivation of a spectrum. Astronomical Journal, 93, 968. ROGERSON, M. 2003. Palaeoceanographyand sedimentology of the Gulf of Cadiz; 30-0 ka BP. PhD Thesis, University of Southampton, Southampton. SCH6NFELD, J., ZAHN, R. & DE ABREU, L. 2003. Surface and deep water response to rapid climate changes at the Western Iberian margin. Global and Planetary Change, 36, 237-264. SHACKLETON, N., HALL, M.A. & VINCENT,E. 2000. Phase relationships between millennial-scale events 64,000~24,000 years ago. Paleoceanography, 15, 565-569.

THOMSON, J., NIXON, S. ET AL. 2000. Enhanced productivity on the Iberian margin during glacial/interglacial transitions revealed by barium and diatoms. Journal of the Geological Society, London, 157, 667677. THOMSON, J., NIXON, S., SUM~RHAYES, C.P., SCH6NFELD, J., ZAHN, R. & GROOTES,P. 1999. Implications for sedimentation changes on the Iberian margin over the last two glacial/interglacial transitions from (23~ systematics. Earth and Planetary Science Letters, 165, 255-270. VOLAT, J.-L., PASTOURET,L. & VERGNAUD-GRAZZINI, C. 1980. Dissolution and carbonate fluctuations in Pleistocene deep-sea cores: a review. Marine Geology, 34, 1-28.

Sediment colour analysis from digital images and correlation with sediment composition ALEXANDRA ANTHONY

J. N E D E R B R A G T

T. O S B O R N 1, A D R I A N & THOMAS

1 ROBERT

B. D U N B A R 2,

P A L M E R 3, J I ] R G E N

W. T H U R O W 1

WAGNER 4

1Department of Earth Sciences, University College London, Gower Street, London WC1E 6BT, UK 2Department of Geological and Environmental Sciences, Stanford University, Stanford, CA 94305-2115, USA 3Department of Geography, Royal Holloway, University of London, Egham, Surrey TW20 OEX, UK 4 Universitiit Bremen, Fachbereich 5 - Geowissenschaften, Postfach 330440, D - 28334 Bremen, Germany Abstract: Sediment colour records can be extracted from digital images of sediment core sur-

faces, which provide the spatial resolution needed to measure colour in laminated sediments. Digital cameras sample colour in three broad wavelengths, red, green and blue, which are subsequently translated into the CIE L*a*b* colour space. Methods to extract calibrated colour values are discussed in this paper. L*, a* and b* values are correlated with geochemical analyses ofcm-scale bulk sediment samples. The sediments are from a suite of laminated and homogenous sections containing organic matter, carbonate, biogenic opal and lithogenic material in variable proportions. Total organic carbon (TOC) content has the dominant effect on sediment colour. Results show that there is a strong correlation with lightness (L*) for TOC values between 0.5 and 10%, but that sediment lightness becomes saturated at higher TOC concentrations. Biogenic opal content cannot be resolved using the L*a*b* colour space. Biogenic opal in itself has a light colour but it tends to occur in darker coloured sediments because of a positive correlation between opal and TOC content. Carbonate content in the measured sections is generally less than 25%, at which values its effect on colour is obscured by the other sediment components.

Sediment colour is one of a suite of sediment properties that can be measured using fast and nondestructive techniques on marine and lacustrine cores. High-resolution colour records are useful for cross-correlation of adjacent overlapping cores (e.g. Kroon et al. 1998), as a proxy for sediment composition (Balsam et al. 1999; Helmke et al. 2002) or for wiggle-matching with other palaeoclimate records (e.g. Hughen et al. 1998). Quantitative colour measurements are now produced routinely, for example during Ocean Drilling Program (ODP) cruises. The commonly used photospectrometers can measure reflectance in small increments across a range of wavelengths, yielding a reflectance spectrum that can provide detailed information about lithological and chemical composition of the sediment (Mix et al. 1995; Balsam & Deaton 1996, and references therein). However, the maximum stratigraphic resolution is 1 or 2 m m at most. The higher spatial resolution needed to measure colour variation in finely bedded or laminated sediments can be obtained

from digital images (Schaaf & Thurow 1994; Merrill & Beck 1995; Nederbragt & Thurow 2001). In contrast to the photospectrometer, digital cameras summarize reflectance values in the range of visible light into three broad wavelength intervals (red, green and blue channels, RGB). On the other hand, a spatial resolution of 100 pm, or better, can be obtained easily with modern cameras and computers, allowing reconstruction of lamina-scale variation in colour. Such highresolution sediment colour time series have been used successfully to reconstruct palaeoclimatic records from laminated and varved sediments (e.g. Schaaf & Thurow 1998; Rodbell et al. 1999; Nederbragt & Thurow 2001). This paper first presents a overview of the technical requirements to obtain quantitative colour data from digital images, with special emphasis on images obtained with so-called 'prosumer' cameras (cameras in between the consumer and professional range of products). With ongoing advances in camera technology,

From: ROTHWELL,R.G. 2006. New Techniques 01 Sediment Core Analysis. Geological Society, London, Special Publications, 267, 113-128. 0305-8719/06/$15.00 ~) The Geological Society of London 2006.

114

A. J. NEDERBRAGT ET AL.

cameras from the upper end of the c o n s u m e r range can offer sufficient resolution to m a k e them attractive as an alternative to the m o r e expensive technical cameras or photospectrometers. However, c o n s u m e r cameras are not designed for technical purposes, and we will discuss the issues that need to be considered when using them for scientific purposes. As a second objective, colour data from a range of different sections are correlated with organic carbon, carbonate and biogenic opal content analyses on discrete samples to test if sediment colour can yield a (semi-)quantitative estimate for sediment composition 9 The sections that are selected for analysis are a c o m b i n a t i o n of (intermittently) laminated, as well as bioturbated, sequences mainly from the marine realm. One lacustrine core with highly variable organic c a r b o n content is a d d e d to further explore the effect of c a r b o n content on colour. The sections are representative for the type of sediments f o u n d in anoxic basins, given that the main strength of digital sediment colour analysis is the possibility to compile colour records in laminated sediments. Ideally, chemical analyses should have been p e r f o r m e d at laminascale samples to establish the correlation with lamina-scale variation in colour. However, the a m o u n t of material that can be collected accurately from single laminae is insufficient for s t a n d a r d geochemical methods. We therefore analysed cm-scale samples and c o m p a r e d results to the average colour values for the equivalent

interval, as a first step in identifying the main controls on sediment colour in a f o u r - c o m p o n e n t sediment system.

Material and methods S e d i m e n t colour d a t a

The colour data presented in this paper were collected for sediment cores from various localities (Fig. 1) using different camera systems. Colour data are relative, i.e. not calibrated against a colour standard, unless specified otherwise. ODP Site 893. Two overlapping holes were cored at ODP Site 893 in the Santa Barbara Basin, recovering an approximately 200 m-thick sequence of intermittently varved sediments spanning approximately the past 150 ka (Kennett et al. 1994). The laminated portions of the sequence consist of alternating light, diatom-rich laminae, which are deposited during the upwelling season in spring and summer, and dark, terrigenousrich laminae, which are deposited during autumn and winter (Soutar & Crill 1977; Bull & Kemp 1995). The varves are gnerally between 1 and 3 mm thick. Cores 893A-1H-893A-3H and 893B-1H-1-893B-4H-2, which contain a predominantly laminated sequence of Bolling/Aller6d-Holocene age, were photographed in the ODP core repository in College Station, TX, with the in-house LEAF still-camera system described by Merrill & Beck (1995). The resolution of the images is approximatelyl20 colour measurements per cm of sediment. A total of 172 samples were collected from the top of the

- 60ON

- 30~ /ii -

Site 1019 k Site 1017 ~I~ / ODP Site 893, Santa

0 o

e~Barbara _ :30 o Site

Basin

1016 ,,2.'.,& . . ~>,~ MD02 2515, ~ ' ~ t : ~ Guaymas

90~

60~

30 ~

- 60oS

ODP Site 1098, Palm_erD e e ~

120~

40~

30~

0~

125~

115~

105~

Fig. 1. Map showing localities discussed in text with an enlargement showing the locations of the NE Pacific sites.

D I G I T A L C O L O U R ANALYSIS OF SEDIMENT last glacial sediments to the top of the Holocene at a spacing of about 15 cm. O D P sites 1016-1021. A grid of 13 sites was drilled during ODP Leg 167 in the NE Pacific along depth and latitudinal transects, to reconstruct the Neogene palaeoceanographic history of the California Margin (Lyle et al. 1997). In general, accumulation rates of terrigenous material and biogenic silica increase towards the north, while carbonate content decreases (Lyle et al. 1997). Samples were collected at 5cm spacing in individual 1.5 m sections from the northern locations (sites 1016-1021; Fig. 1) for chemical analysis (Table 1), to correlate shipboard colour data with sediment composition. A longer geochemical record is available for about the past 260 ka at ODP Site 1018, which was sampled every 10cm (Lyle et al. 2000). Colour data were collected on board using a video system producing images with a resolution of 40 measurements per cm of sediment, and data were reprocessed after the cruise. Colour data were calibrated against four chips with known colours that were included in all images (Lyle et al. 1997; Nederbragt et al. 2000). O D P site 1098. This site in the Palmer Deep, Antarctica Peninsula, contains an approximately 50 m-thick Holocene sequence of intermittently laminated diatom oozes and diatom-bearing muds (Barker et al. 1999). Within the laminated intervals, light clay-rich sediments alternate with diatom-rich laminae that are generally darker than the surrounding sediment. The diatom-rich laminae range in thickness from 1 mm to more than 1 cm. They represent biogenic deposition during a single spring-summer season, but there are too many missing years for the sediments to be consistently varved (Nederbragt &Thurow 2001). Digital images of split-core surfaces were collected in the Ocean Drilling Program Core Repository in Bremen, Germany, using a Geotek line-scan camera system. The digital colour line-scan camera is fixed on a multi-sensor track, and scans the core while it is pushed along the rack by a motor at a nominal resolution of about 120 line scans per cm of sediment. However, at the time the Site 1098 cores were scanned the camera produced systematic noise and the colour data were filtered to one measurement per mm to remove the noise (Nederbragt &Thurow 2001). Samples for chemical analysis are from core 1098B-2H, taken at a sample spacing of 2.5 cm. MD02-2515. MD02-2515 in the Western Guaymas Basin (Gulf of California) is a 65 m-long giant piston core recovered during the IMAGES VIII (MONA) cruise with R.V. Marion Dufresne in July 2002. Most of the sediment is distinctly to faintly or intermittently laminated. A benthic foraminiferal oxygen isotope stratigraphy indicates that the sediment sequence ranges

115

from the mid-Holocene down into Marine Isotope Stage 3. Images of split-core surfaces were collected with a Canon EOS 300D with a 50mm macro lens. The (interpolated) resolution of the images is approximately 240 pixels per cm of sediment. The sediment colour data extracted from the images are calibrated against values measured for colour chips in a copy of the Geological Society of America (GSA) rock-color chart (Goddard et al. 1948), which was photographed under the same conditions as the sediment images. A set of 130 equally spaced samples was collected for preliminary geochemical analyses. Llyn Gwernan. This is a soft water lake (grid ref. SH 703 159) on the lower northern flanks of the Cader Idris mountain region in Wales, UK. The lake was formed around 14 000-13 000 14C years BP after retreat of the Devension ice sheet (Lowe & Lowe 1989). The data used here are compiled from a set of 1 m-long overlapping cores retrieved in 2003 with a Russian corer from the western edge of the lake. The sediment sequence at this site is approximately 14m-thick and consists of peat and organic lake muds, alternating with minerogenic clays and silts that were deposited during the Late-glacial period and the Younger Dryas. Colour data were collected for the deglacial sequence between 10 and 13m depth in core using a Nikon Coolpix 5700. The (interpolated) resolution of the images is approximately 250 pixels per cm of sediment. Colour data were calibrated using the GSA rock-color chart. Geochemistry

The abundance of total organic carbon (TOC), carbonate and biogenic opal content were measured for this study on discrete samples from ODP sites 893 and 1098, core MD02-2515, and on selected 1.5 sections from sites drilled during ODP Leg 167 (Table l). TOC and carbonate content were measured with different elemental analysers, a LECO-CS-300 (ODP Site 893, University of Bremen), a CarloErba NA1500 Series II (Site 1098, Stanford University) and a LECO-CS-200 (remaining samples, Wolfson Laboratory for Environmental Geochemistry at University College London). Total carbon content was measured on one aliquot of a powdered sample, another aliquot was acidified with HCI before measuring organic carbon content. Carbonate content is calculated as the difference between the two measurements. Biogenic silica content was measured on an aliquot of the same powder using the method of Mortlock & Froelich (1989). Measured silica values are multiplied by 2.4 to account for the water content of biogenic opal. Total organic content in the Llyn Gwernan core is estimated by loss-on-ignition (LOI). Powdered samples were weighed before and after furnacing at 550 ~ for 4 h. Published TOC and carbonate data are used for

116

A.

Iz

II

I

ET AL.

J. NEDERBRAGT

I

I

I

I 1 1 1

II

o~oo~ooo~oo~oooo~o~_~

~

t"q I

I

I

I

I

I

I

i

i l l

i l l

i ~ l i

ii

i

,...,..,

-~

D I G I T A L C O L O U R ANALYSIS OF S E D I M E N T ODP Site 1018 (Lyle et al. 2000). To allow correlation with the colour data, the original full-resolution colour data were integrated into one average value for the interval from which the corresponding chemical sample was collected.

Digital sediment colour data acquisition and processing The main requirement to collect sediment surface images that are suitable for subsequent extraction of quantitative colour data is that light conditions are kept as constant as possible during a scanning session (Nederbragt & Thurow 2004). A standard copy stand with two photographic lights is adequate for this purpose. Standard fluorescent lights, while relatively common, are not suitable as a light source. They cycle through a range of different colour temperatures at a rate of approximately 100 cycles s 1. With standard shutter speeds of 1/60s or faster, the colour of a series of images taken with a digital camera will vary noticeably depending on the temperature cycle of the fluorescent light source. The colour values registered by the camera can be calibrated using images of a colour reference card, which is imaged under the same conditions as the sediment (Nederbragt & Thurow 2004). If light conditions are indeed constant, all sediment images in a given set can be processed using one set of calibration values, which are estimated from the (known) colours in the reference card. However, the best results are obtained when a colour reference card is included in all sediment images. In practice, small fluctuations in light strength do occur, for example due to variable distance between sediment and camera when the core surface is not entirely flat, which require colour calibration for each image individually. The sediment surface has to be prepared to reduce the amount of noise in the final image due to reflections off wet sediments, or shadows or light areas produced around irregularities and scratches on the core surface. Soft sediment surfaces are scraped beforehand to expose the fresh sediment colour. When done carefully, most scratches and other surface irregularities can be smoothed away. A polarizing filter can be used to reduce reflections off wet sediment surfaces when a direct light source is used. Alternatively, the images can be taken with diffuse (indirect) light, by placing umbrellas in front of the light bulbs (Christensen & Bjorck 2001), or by deflecting the light via a mat white surface onto the split-core surface. Once the images are collected, colour data can be extracted along the stratigraphic axis using standard image analysis software. An essential part of the further processing of the extracted colour data is the application of a correction for uneven light distribution (Nederbragt & Thurow 2004). Even with the best

117

quality photographic lamps, the distribution of light across the area that is imaged is not completely uniform, with the result that colour values near the edge of the image are darker than those in the centre (or lighter, depending on the position of the lights). A detailed review of all the steps involved is given in Nederbragt & Thurow (2004). However, this review was based mainly on images as produced by a technical camera, although potential complications with nontechnical cameras were noted. Since then, we have tested two prosumer cameras to determine if they are suitable for scientific photography. We found that adequate results can indeed be obtained if the limitations of the camera are taken into account, both during photography and during processing of the data. The main points are image resolution and lack of control over image-processing parameters within the camera.

Resolution. Camera resolution is specified as the total number of pixels in the final image. However, for a standard digital camera this resolution is not real resolution, but it is interpolated. More correctly, the specified resolution applies to the number of lightsensitive photo diodes on the sensor array within the camera. A single pixel in the final image on the computer screen consists of three values, one for each of the three coiour channels, red, green and blue. A single photo diode in the camera can collect information about one colour channel only during a single exposure, yet the camera has to reconstruct a full colour image during this exposure. The sensor array is therefore composed usually of an alternation of one red, one blue and two green cells, although other arrays are possible. For the final photograph the colour information is interpolated in both directions across the array to get the three colour values needed for each pixel in the full-colour image. As a result, each block of 2 x 2 pixels in such an interpolated image is therefore roughly equivalent to a single pixel in a real-resolution image. For example, the Canon EOS 300D produces images of 3072 x 2048 (interpolated) pixels, but the amount of colour information is the same as in an image of 1534 x 1024 pixels real resolution. This means that four times as many pixels have to be sampled from an interpolated image than from a real-resolution image to obtain equally reliable colour estimates. Outputformat. The raw image data are generally processed within the camera before they are transferred to a storage medium. Most, but not all, prosumer and professional digital cameras offer the option to save the unprocessed data in some type of raw format; that is, the data as registered by the light sensor array without any processing. In theory, this offers the possibility to bypass the processing done by the camera or the accompanying software, and to find third-party software that allows more control over

118

A. J. N E D E R B R A G T E T AL.

the parameters used. To date, we have not found a package that delivers adequate results, but alternative software may become available as the digital camera market keeps growing. It is therefore important to select a camera that allows as much control as possible over processing parameters. The option to save the data in the uncornpressed T I F F format is essential, as loss of data will occur if the images are compressed. It should be also be possible to turn off features that modify the original colour, like contrast and/or colour enhancement. The most obvious problem, transformation into non-linear RGB, is the easiest to correct for and is discussed below.

Colour calibration. During processing in most, if not all, non-professional cameras, the original colour data, which are linearly proportional to the amount of light received by the camera, are transformed into a non-linear version of RGB or R'G'B'. In non-linear R'G'B', R' = Rmax(R/Rmax) h, in which b is a constant between 0 and 1, and Rmax is the maximum value that R can have (usually 255). The definitions for G' and B' are similar. Output on a computer screen is a non-linear function of the input energy. The above transformation, or gamma-correction, is therefore introduced to improve contrast and brightness of an image for viewing on a computer screen. However, equations to translate RGB colours into other colour systems are based on linera RGB. The non-linear R'G'B' values therefore have to be transformed back into linear colour data. A practical solution is to take an image of a suite of chips with known colours and to estimate the constants used in the gamma-correction. Suitable colour chips are provided by the Munsell soil colour chart or GSA rock-color chart, because they cover the range of colours that are found in sedimentss, and their colour values are known (The Munsell Color Science Laboratory 1999). Image analysis software can be used to measure the R'G'B' values of the various colour chips in such an image, after which a calibration function is estimated using the known colour values of those chips as the independent variable (see below). Note that a new calibration curve has to be determined every time that the photographic setting is changed (aperture and speed of the camera, position of the light source, distance between object and camera, etc.). As an alternative, a selection of dark and light coloured chips can be cut out of the colour chart, and pasted onto a smaller piece of cardboard, which can be placed alongside the core in all images. This offers the possibility of calibrating each image individually. The most commonly used colour co-ordinate system to describe sediment colour is meanwhile the CIE L*a*b* system (defined by the Commission Internationale de l'l~clairage). L* is lightness, and ranges from 0 (black) to 100 (pure white). The actual colour (hue) is expressed in a* (negative values are red, positive

is green) and b* (negative for blue and positive values for yellow). Translation from linear RGB into the L*a*b* system is by definition done via an intermediate step, the X Y Z tristimulus system. In this system, Y (luminance) represents grey-scale, and X and Z contain the actual colour information. In all three systems, pure black has the values (0, 0, 0) for all three variables. Pure white in RGB is (255,255,255), and in L*a*b* it is (100,0,0). However, in the X Y Z system, all colours are defined relative to a reference white. Different reference whites, which are defined by CIE, are used for different applications (Berns 2000). They define a colour that looks like pure white under specific light conditions. The two whites relevant here are CIE C and CIE D65, both viewed under a 2 ~ angle. D65 (2 ~ is the standard white used for television broadcasting and computer screens. It represents natural daylight, and has values X = 9 5 . 0 5 , Y = 1 0 0 and Z = 1 0 8 . 9 . The definition of Munsell colours is based on C(2~ which represents indirect sunlight and has values X = 98.074, Y = 100 and Z = 118.232. Linear scaling is the easiest way to convert between two different whites, by multiplying all X, Y and Z values by Xwl/Xw2, YwJ/Yw2 and Zwl/Zw2, respectively, in which w l and w2 refer to the values defined for the two reference whites. X Y Z is a linear transform of RGB:

-~

Yr

Yg

Yb

Zr

Zg

Zb

"

(1)

where Xr, Yr, Zr, etc., represent device-specific constants. Values for the constants can be estimated using the colours measured in an image of three chips with known colour values. The RGB values in the image and the known X Y Z values are inserted into equation (1) for each of the three known colours separately, yielding a set of three linear equations, which can be solved using standard mathematical techniques. In addition, a vector of constants can be added to equation (1) to allow for a non-zero intercept. In that case, a fourth colour chip is required to be able to solve the resulting set of four equations (Nederbragt et al. 2000). Default values for the constants in equation (1) are based on guidelines for colour representation in television broadcasting (ITU-R 2002; SMPTE RP 177-1993 1993):

--

0.2126

0.7152

0.0722

0/255 /

0.0193

0.1192

0.9505

B/255]

(2)

in which 255 represents the maximum value that R, G and B can have. The values in equation (2) are based on reference white CIE D65 (2~ Note that the resulting X, Y and Z values are scaled to unity (Y = 1.00) rather than to Y -- 100. By definition, Y (luminance)

D I G I T A L C O L O U R ANALYSIS OF S E D I M E N T

119

Fig. 2. Cross-plot of R, G and B values for 39 Munsell colour chips calculated from their defined X Y Z co-ordinates against the values measured in a series of images of those chips. Images of a GSA rock-color chart were collected with a Canon EOS 300D camera. Note that the offset in the blue channel is due to an imperfect white-balance, i.e. the images have a slight brownish cast. For further discussion see the text.

in this equation is the same as grey-scale. Grey-scale values are thus a weighted average of the three RGB colours. The reason is that, at the same light intensity, green is perceived as brighter than red, which in turn is brighter than blue. The inverse of the matrix multiplication in equation (2) is:

255

3.2405

-1.5373

-0.4985

-0.9693

1.8760

0.0416

0.0557

-0.2040

1.0573

Iil (3)

As in equation (2), the X, Y and Z values are scaled to unity. However, equations (1), (2) and (3) are valid only when the RGB space is linear. With the non-linear RrGtB r values obtained from prosumer cameras it is more convenient to combine colour calibration and correction for non-linearity into one step. In Figure 2 values obtained from a series of images of Munsell colour chips in a GSA color-rock chart are plotted against the theoretical RGB values, which are calculated from the known X Y Z values (The Munsell Color Science Laboratory 1999). Published X Y Z

values for Munsell colours are specified relative to reference white C (2~ All values were therefore first converted to D65 (2 ~ before they were inserted into equation (3). A best-fit function with shape Y = a + b X C can then be estimated to reconstruct the gamma-correction applied within the camera. The constant a allows for an offset from the zero point, that is when black as measured by the camera does not correspond to pure black. The inverse function is used to calibrate the RGB values. In Figure 3 there is an offset between the three colours. This is due to the white-balance setting of the camera not corresponding exactly to the colour temperature of the light source. As a result, all images taken under these conditions have a very slight brownish cast. However, the colour cast can be corrected at the same time as the non-linearity in RGB, by estimating a best-fit function for each of R, G and B separately. The regression equations are then applied to RGB values of the sediment in a series of images collected under the same conditions as the Munsell colour chips. This transforms the data into calibrated and linear RGB values, which are then translated into X Y Z tristimulus values using the default equation (2). The translation of X Y Z values into the L*a*b* co-ordinate system is defined by the

A. J. NEDERBRAGT ET AL.

120 following equations (Berns 2000):

Sediment colour and composition

L*= ll6(f(~n)--1@6)

]

x

Y b* = 2 0 0 ( f ( ~ n ) -

(4.1)

(4.2) f

Z ( ~-~) )

(4.3)

with f ( Y / Y n ) = (Y/Yn) V3 for Y/Yn > 0.008856 and f ( Y / Y n ) = 7.787(Y/Yn) + 16/116 for Y/Yn <_ 0.008856; f(X/Xn) and f ( Z / Z n ) are defined similarly. Xn, Yn and Zn refer to the tristimulus values of a reference white. The values to use in practice will be those obtained from equation (2) for pure white in the RGB system (255, 255, 255), that is the X Y Z values for D65 (2~ but scaled to unity. L*, a* and b* are thus a non-linear transform of the tristimulus values X, Y and Z, which are linearly proportional to the amount of light reflected from an object (Berns 2000). Theoretically, a mixing line of variable proportions of two components with different colours should therefore result in a linear relation between percentage of the components and colour expressed in the XYZ colour co-ordinate system, but not in the L*a*b* system. In theory, this is a disadvantage of the use of the L*a*b* to describe sediment colour, as it should lead to a non-linear relation between the colour values and compositional data. However, we have not observed any obvious nonlinearity generated by the conversion into the L*a*b* system (Figs 3-7).

Applications. Digital images not only have the resolution to describe lamina-scale fluctuations in colour, but also allow for reliable estimates of cm- to dm-scale colour fluctuations in laminated or finely mottled sediments. The disadvantage of the photospectrometre is that it has a fixed spot size, typically of 1 or 2 mm. Photospectrometre records therefore tend to show substantial scatter in sediments with fine-scale fluctuations in colour, when only one reading is taken every 1 or 2cm, due to undersampling of the full range of colour variation within the laminae or mottles. The continuous time series that can be obtained from image data have the advantage that colour can be integrated over a larger surface area, both laterally across the core surface or along the stratigraphic axis. Effectively, this allows better control on the spot size of the colour sample to produce stratigraphic trends that are more representative for cm-scale oscillations. For example, image-derived colour data for the laminated sediments at ODP Site 1098 yielded a smoother long-term trend that was easier to correlate with other core-logging data than the shipboard photospectrometre data (Nederbragt & Thurow 2001).

The marine sections analysed for this paper are all from near-shore high productivity areas with high sedimentation rates of both lithogenic and biogenic material. The resulting sediments, whether homogeneous or laminated, consist of variable amounts of four main components, the biogenic components TOC, carbonate and opal, in combination with lithogenic material (Figs 37). However, TOC content in these marine sections is generally less then 3%. The Llyn Gwernan section was therefore included in the dataset as the one lacustrine section, because of its extremely high and variable TOC content (Fig. 7). Linear correlation coefficients between the three colour variables L*, a*, b* and weight percentages of the biogenic components TOC, carbonate and biogenic opal are summarized in Table 1 for each section separately. Under ideal conditions, the three colour variables could yield proxy data to estimate three components. In practice L*, a* and b* are not independent, but are usually strongly correlated with each other. As a result, colour expressed in tristimulus values can resolve only one sediment component. Overall, the best results are found for lightness (L*) as a relative indicator for TOC content (dark) relative to all other sediment components (Figs 3-7).

Organic carbon content The coefficients in Table l show that a significant negative correlation between sediment lightness and TOC is found if the range of variation in TOC values is around 1% or higher. However, the correlation coefficients are all more than - 0 . 7 5 , that is, there is too m u c h scatter to allow L* to be used as a quantitative measure for TOC. In theory, inaccuracies in matching the location of the colour sample to the chemical sample could have contributed to the scatter within the data in laminated intervals. In practice, the effect appears to be minor, as the correlation between colour and composition in the h o m o g e n e o u s sections is not significantly better than in the laminated sediments (Table 1, Site 1018, Llyn Gwernan). Correlation coefficients of Ir] > 0.8 are required before one variable can yield a reliable estimate for another independent variable (e.g. Balsam & D e a t o n 1996). Yet, the stratigraphic plots of colour and T O C content in Figures 3-7 illustrate that L* is a good relative measure for TOC, in that the two variables show mirrored fluctuations with very few exceptions. The main feature of the relation between organic carbon content and L* in the Llyn

DIGITAL COLOUR ANALYSIS OF SEDIMENT

121

r~

=Z

r~

~<.~. o~ ~ ' ~

9

0

.'R~

~ o ~'~

. ~

122

A.J.

9

~

NEDERBRAGT

.,r

E T AL.

8

(j 0 io

o~

_s

v

'h~

,~."

0

0

,-Z

I

I

I

I

I

(sseulq6!q) ,7 ~

.

~s ~

.,,~

. ,...~

0

~ ,...~

os _~

~8

0 I,Li lm, -

r

N 0

cO --

,t

s

~

~~ ' ~

O~ v

0p.~

se6eis edo;os!

~1

~

i~1

o

i

0

0

o

i

0

~

i

0

i

0

1o

~

.

i

0

s~

(w) qidep el!sodwoo .-~

.eB

D I G I T A L C O L O U R ANALYSIS OF S E D I M E N T

123

0

"./-~

0~

01

r

~I

OI

O.

9

v

9

o 9 9

I

I

I

~"

9

A

I

I

~t"

00

9

I

I

~I"

I

~D

0 0 I--

I

CO

(sseulq6!q) .7 o Lo

o

~L o

~=~.

O

N O~ "e-

0

0

W

I--

r

m

j

0=~

O ',r--

',c-

0

O

I--

'0~

GO

~o

0

eJnlxel

lueLu!pas

I

IIIIIII

I

CO

IIIn IIIII I

I

IIIIIIIIIIIIIIIIIIIIIIIII I

0

T--

(w) qidep el!sodwoo

I

04 ,r--

III Ill I

IIII I

oo

~1" ,r--

..~ _~

124

A. J. N E D E R B R A G T ET AL.

o

i

9

9~

@ 9 9

9 ~

99

9

~

9 9o

9

9

v

9 o

-e4

9 o

0

@

9 ,~-i.." jfp~..,. , I

9 9 9 1 7~6

I

I

.

-

~

-~

'.~ I

I

I

(sseulq6!7) ,q 5

~

3

o

0

v

0

E = r~

o_ 0-e v o

.9

0

_

o_

~ ~=

m

~o_

04 CO ~

o

:~ ~=

~

,o-_

~'.~

~t~

o .

ON

-~2

aJn],xe~, :lueuJ!pes IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII IIIIIIII

o

o ~-

I

I

o 04

o co

IIIIIIIIIIIIIIIIIIIIIIIIIIIII

I d~ 9169

I o ~t"

(u J) q;dep pe;0en00-p!oA

o M3

o ~43

.~S

DIGITAL COLOUR ANALYSIS OF SEDIMENT

125

LOI (%) 0

10

20

30

40

105

=~

"e 9 o

U

..

9

9, 20 11.0 10

~"

11.5-

.~

0 ~

30

40

16 I

~'12 ~

12.0

9

.t'-

12.5 4

I

I

I

I

I

50

40

30

20

10

L* (Lightness)

9

~

oo

o

~

0

0

10

20

30

40

LOI (%)

Fig. 7. LOI and lightness (L*) in a core from Llyn Gwernan, Wales, plotted against stratigraphic depth, with a cross-plot of LOI against lightness (L*) and against luminance (Y). GS2 stands for Greenland Stadial 2 (last glacial). For further explanation see Figure 3. Gwernan section is that the correlation is clearly non-linear (Fig. 7). The possibility that this is due to L* (lightness) being a non-linear transform of Y (luminance or grey-scale) can be excluded, because the non-linear correlation is even more pronounced in a cross-plot of Y against LOI (Fig. 7). Apparently, sediment colour starts to become saturated when TOC content reaches values around 10%, the point at which the slope of L* v. TOC starts to decrease noticeably. Lightness values are essentially constant for samples with greater than 20% TOC (Fig. 7). A similar pattern can be expected in other sediments with high TOC content. Extrapolation of regression lines for the marine sections (Figs 3-6) would give an intercept with L * = 0 (pure black) at TOC contents of 16.9% (Site 893), 15.2% (Site 1018), 9.2% (Site 1098) and 9.7% (MD02-2515). Such extrapolation outside the range of measured values is statistically unwarranted, but the exercise illustrates that L* can be expected to become saturated if more TOC were added at values that are roughly similar to those found for the Llyn Gwernan dataset.

The effect of lithology on a* and b*, i.e. the hue of the sediment, is generally weaker than that on L*, and varies from section to section. Out of seven sections that have a significant negative correlation between L* and TOC, five show a significant positive correlation of a* and/or b* with TOC, i.e. dark and organic-rich sediments are more red, orange or yellow. The correlation of both a* and b* with TOC is negative in two sections, MD02-2515 and Llyn Gwernan (Table 1). The sign of the correlation is most probably dependent on the oxidation state of iron, although further chemical work would be required to substantiate this hypothesis. Bulk sediment inorganic chemical data, which were collected for part of the samples, are inconclusive because they cannot show in what form the iron is present in the samples. However, stratigraphic fluctuations in a* in other regions have been related to the presence of iron minerals. For example, in the northern North Atlantic, variation in a* can be explained by variable concentrations of detrital hematite (Giosan et at. 2002; Helmke et al. 2002). The

126

A. J. NEDERBRAGT ET AL.

marine samples here are all from sections either in basins with fully anoxic conditions or in areas with an intense Oxygen Minimum Zone (Site 1018). The variable correlation between a*/b* and TOC, from negative to positive, may reflect differences in the type of anaerobic degradation of organic matter, which affect the redox conditions that iron is sensitive to. A positive correlation between TOC and a* and/or b* (TOC-rich sediments are more reddish, orange or yellow) could result from the presence of pyrite which is formed during sulphate reduction. Carbonate content

The effect of carbonate on colour in the organicrich sediments included in this study is minimal, even for samples that contain more than 25% carbonate (Table 1). A high positive correlation between L* and carbonate content is found only for section 1020C-9H-4. In this section carbonate content is low (7.6% at most), but TOC content is essentially constant (Table 1). A significant positive correlation is also found for Site 1018, but the correlation coefficent r = 0.33 is far too low to use L* to predict carbonate content. The lack of correlation in the other sections could be due to TOC having a stronger effect on colour than carbonate. In most sections carbonate content fluctuates independently from TOC, resulting in a correlation between TOC and carbonate that does not differ significantly from r = 0. At Site 893 and in the sections from holes 1017D and 1019C, however, carbonate content shows a positive correlation with TOC and biogenic silica content (Table 1), suggesting that sediment composition in those sections is largely determined by variable dilution with siliciclastic material of the three biogenic components together. Lack of correlation between carbonate and colour must be attributed to a combination of low carbonate content and the presence of multiple components within the sediment. Colour expressed in tristimulus values has been used successfully as a measure for carbonate content in cores from the Atlantic Ocean, where sediments consist mainly of a mix of light-coloured carbonate with dark-coloured lithogenic material (Balsam et al. 1999; Ortiz et al. 1999). A clear correlation between carbonate content and grey-scale (luminance) is also found in mid-Cretaceous organic rich chalks from the Tarfaya Basin in Morocco (Kuhnt et al. 2005). Those sediments also represent a two-component system, in that they consist predominantly of carbonate (light) and organic

matter (dark) in variable proportions. In the sediments from high productivity regions studied here the effect of carbonate on sediment colour is apparently overshadowed by the fluctuations in TOC content, and possibly also in opal. B i o g e n i c opal c o n t e n t

The varved sediments from ODP Site 893 and from core MD02-2515 have been described as consisting of light diatom-rich laminae alternating with dark lithogenic laminae (Soutar & Crill 1977; Pike & Kemp 1996). However, biogenic opal content measured in discrete samples, which represent an average of several varves, does not show such a positive correlation with lightness (Table 1). If anything, diatom-rich sediments tend to be dark. Correlation coefficients between opal and L* in ODP Site 893 and core MD02-2515 are weakly negative but do not differ significantly from zero. Data for the laminated sediments from ODP Site 1098 show a significant negative correlation with lightness, i.e. high opal content is found in TOC-rich, laminated intervals, which are darker than surrounding homogeneous sediments (Fig. 5). The other sections show similar results, giving correlation coefficients between opal and L* that range from weakly to strongly negative. The dark colour of opal-rich sediments is surprising, given that a pure diatom ooze is nearly white. Three factors that may contribute to the absence of a positive correlation between L* and opal are carbonate content, water content and, especially, the distribution of TOC. Opal and carbonate content can be distinguished from each other in the near-infrared range of the light spectrum, as opal is more reflective in this range than carbonate. Using photospectrometers that measure reflectance outside the range of visible light, Mix et al. (1995) and Balsam & Deaton (1996) found a strong correlation between opal content and the shape of the reflectance spectrum, allowing estimation of opal content as well as of carbonate and TOC content. However, pure chalk and diatom ooze, when dry, are both nearly white, and probably virtually indistinguishable in colour in the visible range. The presence of both components in the sections studied here may contribute to a lack of correlation between L* and each component separately. Water content also plays a role at least for the laminated sediments in the Palmer Deep. Thick laminae of diatom-ooze occur at Site 1098 (Fig. 5), which are generally darker than the surrounding sediments (Barker et al. 1999; Nederbragt & Thurow 2001) partly because the

DIGITAL COLOUR ANALYSIS OF SEDIMENT surrounding clay-rich sediments are unusually light in colour and partly because the high water content of the highly porous diatom ooze produces a darker colour. Balsam et al. (1998) recommended that colour is measured on dry sediments, to avoid problems with variable water content. However, drying the sediment is not really feasible if the aim is to establish colour records for entire cores. The effects of water content are therefore inherent to colour acquisition using non-destructive techniques. The main reason that opal-rich sediments tend to have a dark colour is presumably due to a positive correlation between TOC and opal content in most sections (Table 1), with the result that any effect of opal on lightness is overshadowed by the effect of TOC. In general, both the production of diatoms and of organic matter increase when marine productivity increases in the upwelling areas in the Eastern Pacific, while the response of carbonate production and preservation is more variable (e.g. Gardner et al. 1997; Mortyn & Thunell 1997). A positive correlation between TOC and opal content may therefore be inherent to the type of sediments that were included in this study, either due to glacial-interglacial changes in productivity or to variable dilution with terrigenous material. A negative correlation between carbonate and opal may point to temporal variation in productivity as the main cause for variation in the biogenic components (Table 1, core MD02-2515 and ODP sections 1016B-2H-3 and 1021B-1H-4), while a positive correlation between all three biogenic components is probably due to dilution by siliciclastic material (ODP Site 893). Further work will be needed to resolve the discrepancy between the dark colour of opalrich sediments at the cm-scale and their inferred light colour at the lamina scale. We expected initially to find a positive correlation between L* and opal content in at least some of the sections, based on the visual observation that diatom-rich laminae are light in the varved sediments in the Guaymas Basin (core MD02-2515) and the Santa Barbara Basin (ODP Site 893). The discrepancy is potentially due to de-coupling of diatom and TOC fluxes at the seasonal scale. Increased sedimentation of diatoms or their resting spores can occur at the end of the productive season (Leventer et al. 1996; Kemp et al. 2000). This could lead to deposition of discrete organic-rich and diatom-rich laminae within a single varve, even if TOC and opal content covary over a period of several years (several varves), which is represented in the cm-scale samples analysed here.

127

Thanks are due to P. Francus and J. Helmke for their suggestions to improve the manuscript, and to the Ocean Drilling Program and the IMAGES programme for making the material available. H. Kneale provided the chemical analyses for the Llyn Gwernan section. The help of J. Beck, W. Hale, G. Rothwell, P. Rumford and A. Wuelbers when scanning cores in the core repositories in College Station, TX, Bremen, Germany, and Southampton, UK, was invaluable. The research for this paper was supported by research grants from NERC and Leverhulme.

References BALSAM,W.L. & DEATON,B.C. 1996. Determining the composition of late Quaternary marine sediments from NUV, VIS, and NIR diffuse reflectance spectra. Marine Geology, 134, 31-55. BALSAM, W.L., DEATON, B.C. & DAMUTH,J.E. 1998. The effects of water content on diffuse reflectance spectrophotometry studies of deep-sea sediment cores. Marine Geology, 149, 177-189. BALSAM, W.L., DEATON, B.C. & DAMUTH,J.E. 1999. Evaluating optical lightness as a proxy for carbonate content in marine sediment cores. Marine Geology, 161, 141-153. BARKER, P.F., CAMERLENGHI,A. ET AL. 1999. Proceedings of the Ocean Drilling Program, Initial Reports, 178. Ocean Drilling Program, College Station, TX (CD-ROM). BERNS, R.S. 2000. Billmeyer and Saltzman Principles of Color Technology. Wiley, New York. BULL,D. ~ KEMP,A.E.S. 1995. Composition and origins of laminae in late Quaternary and Holocene sediments from the Santa Barbara Basin. In: KENNETT, J.P., BALDAUF,J.G. ~ LYLE, M. (eds) Proceedings of the Ocean Drilling Program, Scientific Results, 146. Ocean Drilling Program, College Station, TX, 77-88. CHRlSXENSEN,J.Q. & BJORCK,S. 2001. Digital sediment colour analyses, DSCA, of lake sediments; pitfalls and potentials. Journal of Paleolimnology, 25, 531-538. GARDNER,J.V., DEAN, W.E. & DARTNELL,P. 1997. Biogenic sedimentation beneath the California Current system for the past 30 kyr and its paleoceanographic significance. Paleoceanography, 12, 207-225. GIOSAN, L., FLOOD, R.D. & ALLER, R.C. 2002. Paleoceanographic significance of sediment color on western North Atlantic drifts; I, Origin of color. Marine Geology, 189, 2541. GODDARD, E.N., TRASK, P.D., DE FORD, R.K., ROVE, O.N., SINGEWALD,J.T. & OVERBECK,R.M. 1948. Rock-color Chart. Geological Society of America, Boulder, CO. HELMKE,J.P., SCHULZ,M. & BAUCH,H.A. 2002. Sediment-color record from the Northeast Atlantic reveals patterns of millennial-scale climate variability during the past 500,000 years. Quaternary Research, 57, 49-57. HU6HEN, K.A., OVERPECK,J.T. ET AL. 1998. Deglacial changes in ocean circulation from an extended radiocarbon calibration. Nature, 391, 65-68.

128

A . J . NEDERBRAGT ET AL.

ITU-R. 2002. Recommendation BT.709-4: Parameter Values for the HDTV Standards for Production and International Programme Exchange. International Telecommunication Union - Radiocommunication Sector, Geneva. KEMP, A.E.S., PIKE, J., PEARCE, R.B. & LANGE, C.B. 2000. The 'fall dump': a new perspective on the role of a shade flora in the annual cycle of diatom production and export flux. Deep-Sea Research H, 47, 2129-2154. KENNETT,J.P., BALDAUF,J.G. ETAL. 1994. Proceedings of the Ocean Drilling Program, Initial Reports, 146. Ocean Drilling Program, College Station, TX. KROON, D., NORRIS, R.D. ET AL. 1998. Proceedings of the Ocean Drilling Program, Initial Reports, 17lB. Ocean Drilling Program, College Station, TX. KUHNT, W., LUDERER, F., NEDERBRAGT, A.J., THUROW, J. & WAGNER, T. 2005. Orbital-scale record of the Late Cenomanian-Turonian Oceanic Anoxic Event (OAE-2) in the Tarfaya Basin (Morocco). International Journal of Earth Sciences, 94, 147-159. LEVENTER, A., DOMACK, E.W., ISHMAN, S.E., BRACHFELD, S., MCCLENNEN, C.E. & MANLEY, P. 1996. Productivity cycles of 200-300 years in the Antarctic Peninsula region: Understanding linkages among the sun, atmosphere, oceans, sea-ice, and biota. Geological Society of America Bulletin, 108, 1626-1644. LOWE, J.J. & LOWE, S. 1989. Interpretation of the pollen stratigraphy of Late Devensian iateglacial and early Flandrian sediments at Llyn Gwernan, near Cader Idris, North Wales. New Phytologist, 113, 391-408. LYLE, M., KOIZUMI, I. ET AL. 1997. Proceedings of the Ocean Drilling Program, Initial Reports, 167. Ocean Drilling Program, College Station, TX. LYLE, M., MIX, A., RAVELO, A.C., ANDREASEN, D., HEUSSER, L. & OLIVAREZ, A. 2000. Millennialscale CaCO 3 and Corg events along the northern and central Californian margins: stratigraphy and origins. In: LYLE, M., KOIZUMI, I., RICHTER, C. & MOORE, T.C., JR (eds) Proceedings of the Ocean Drilling Program, Scientific Results, 167. Ocean Drilling Program, College Station, TX, 163-182. MERRILL, R.B. & BECK, J.W. 1995. The ODP color digital imaging system: color logs of Quaternary sediments from the Santa Barbara Basin, Site 893. In: KENNETT, J.P., BALDAUE, J.G. & LYLE, M. (eds) Proceedings of the Ocean Drilling Program, Scientific Results, 146. Ocean Drilling Program, College Station, TX, 45-59. Mix, A.C., HARRIS, S.E. & JANECEK, T.R. 1995. Estimating lithology from noninclusive reflectance spectra: Leg 138. In: PISIAS, N.G., MAYER, L.A., JANECEK, T.R., PALMER-JULSON, A. & VAN ANDEL,T.H. (eds) Proceedings of the Ocean Drilling Program, Scientific Results, 138. Ocean Drilling Program, College Station, TX, 413-427. MORTLOCK, R.A. & FROELICH, P.N. 1989. A simple method for the rapid determination of biogenic opal in pelagic marine sediments. Deep-Sea Research, 36A, 1415-1426.

MORTYN, P.G. & THUNELL, R.C. 1997. Biogenic sedimentation and surface productivity changes in the Southern California Borderlands during the last galcial-interglacial cycle. Marine Geology, 138, 171-192. NEDERBRAGT, A.J. & THUROW, J. 2001. Sediment colour variation and annual accumulation rates in laminated Holocene sediments (ODP Site 1098, Palmer Deep). In: BARKER, P.F., CAMERLENGHI, A., ACTON, G.D. & RAMSAY,A.T.S. (eds) Proceedings of the Ocean Drilling Program, Scientific Results, 178. Ocean Drilling Program, College Station, TX, 1-20 (online: http://www-odp.tamu.edu/ publications/178_SR/chap_03/chap_03.htm). NEDERBRAGT, A.J. & THUROW, J. 2004. Digital sediment colour analysis as a method to obtain high resolution climate proxy records. In: FRANCUS, P. (ed.) Image Analysis, Sediments, and Paleoenviroments. Developments in Paleoenvironmental Research, 7. Kluwer, Dordrecht, 105-124. NEDERBRAGT, A.J., THUROW, J. & MERRILL, R. 2000. Data Report: Color records from the California Margin (ODP Leg 167): Proxy indicators for sediment composition and climatic change. In: LYLE, M., KOIZUMI, I., RICHTER, C. & MOORE, C., JR (eds) Proceedings of the Ocean Drilling Program, Scientific Results, 167. Ocean Drilling Program, College Station, TX, 319-324. ORTIZ, J., MIX, A.C., HARRIS, S. & O'CONNELL, S. 1999. Diffuse spectral reflectance as a proxy for percent carbonate content in North Atlantic sediments. Paleoceanography, 14, 17 l-186. PIKE, J. & KEMP, A.E.S. 1996. Records of seasonal flux in Holocene laminated sediments, Gulf of California. In: KEMP, A.E.S. (ed.) Palaeoecology and Palaeoceanography From Laminated Sediments. Geological Society, London, Special Publications, 116, 157-169. RODBELL, D.T., SELTZER, G.O., ABBOTT, M.B., ENFIELD, D.B. & NEWMAN,J.H. 1999. An approximately 15,000-year record of E1 Nifio-driven alluviation in southwestern Ecuador. Science, 283, 516-520. SCHAAE, M. & THUROW, J. 1994. A fast and easy method to derive highest-resolution time-series data sets from drillcores and rock samples. Sedimentary Geology, 94, 1-10. SCHAAE,M. & THUROW, J. 1998. Two 30000 year highresolution greyvalue time series from the Santa Barbara basin and the Guaymas Basin. In: CRAMP, A., MACLEOD, C.J., LEE, S.V. & JONES, E.J.W. (eds) Geological Evolution of Ocean Basins. Results from the Ocean Drilling Program. Geological Society, London, Special Publications, 131, 101-110. SMPTE RP 177-1993. 1993. Derivation of Basic Television Color Equations. Society of Motion Picture and Television Engineers, White Plains, NY. SOUTAR, A. & CRILL, P.A. 1977. Sedimentation and climatic patterns in the Santa Barbara Basin during the 19th and 20th centuries. Geological Society of America Bulletin, 88, 1161-1172. THE MUNSELL COLOR SCIENCE LABORATORY. 1999. http://www.cis.rit.edu/research/mcsl/.

Sediment mineralogy based on visible and near-infrared reflectance spectroscopy R I C H A R D D. J A R R A R D & MICHAEL D. V A N D E N B E R G 1 Department o f Geology and Geophysics, UniversiO' o f Utah, Salt Lake City, UT84112, USA 1present address." Utah Geological Survey, Energy and Minerals Section, 1594 W. North Temple, Salt Lake City, UT84114, USA (e-mail.'[email protected])

Abstract: Visible and near-infrared spectroscopy (VNIS) can be used to measure reflectance spectra (wavelength 350-2500 nm) for sediment cores and samples. A local ground-truth calibration of spectral features to mineral percentages is calculated by measuring reflectance spectra for a suite of samples of known mineralogy. This approach has been tested on powders, core plugs and split cores, and we conclude that it works well on all three, unless pore water is present. Initial VNIS studies have concentrated on determination of relative proportions of carbonate, opal, smectite and illite in equatorial Pacific sediments. Shipboard VNIS-based determination of these four components was demonstrated on Ocean Drilling Program Leg 199.

Rock colour is one of our oldest clues to composition, and spectroscopy instruments now surpass the eye in providing quantitative compositional information. The experimental technique is straightforward and non-intrusive: simply illuminate a rock surface, record the spectrum of reflected light and extract any spectral responses that are sensitive to mineralogy. Reflectance spectroscopy can be used on powders, core plugs and split cores, and measurements take only seconds. These advantages make the technique particularly attractive for depleted intervals of core (e.g. many basalts) and for climate proxies (requiring thousands of high-resolution measurements). Spectroscopy is the measurement of light intensity, as a function of wavelength, reflected from a solid, liquid or gas (Hapke 1993; Clark 1995). The reflected or scattered photons can be detected and measured with a spectrometer. Some photons are absorbed by minerals at and near the surface of a rock or sediment due to both electronic and vibrational processes (Hunt 1977; Clark et al. 1990; Burns 1993; Clark 1995). These processes create absorption troughs in the reflecting spectrum that can be used for mineral identification. Water, Mg-OH, A1-OH, Fe OH and CaCO3 absorption bands, identifiable in the near-infrared (Fig. 1), as well as overall trends in the spectrum are particularly useful for mineral identification. Some absorption troughs are present in more than one kind of mineral (e.g. Fig. 1 and http://speclab.cr.usgs. gov), but with different trough depths. Clark (1995) gives a concise overview of the physics of reflectance spectroscopy, and Hapke (1993)

provides a more detailed review. The majority of previous work in reflectance spectroscopy has involved remote sensing (e.g. Goetz et al. 1983; Clark & Roush 1984; Clark et al. 2003). Using laboratory reflectance spectra as ground-truth, scientists convert airborne (airplane or satellite) reflectance data into mineral maps of the surface of the Earth or other planets (Clark et al. 1991, 1993, 2003). A bandwidth of 350-2500nm is often used to study the reflectance spectra of minerals in a laboratory setting (Hunt & Salisbury 1970, 1971; Hunt et al. 1971a,b, 1972, 1973; Adams 1974; Hunt 1977, 1979; Singer 1981; Clark 1983; Swayze & Clark 1990). The United States Geological Survey Spectroscopy Laboratory in Denver, CO has created an extensive online digital library of mineral reflectance spectra (http://speclab.cr. usgs.gov).

Previous work: spectroscopy at visible wavelengths The ultimate goal of all sediment spectroscopy techniques is to provide accurate, qualitative or quantitative estimates of sediment composition. Many studies have shown that different marine sediment types have distinctive spectral features within the visible and very near-infrared region of the electromagnetic spectrum. Mix et al. (1992, 1995) developed and used a prototype split-core analysis track for automated core scanning of reflectance (455-945nm) on Leg 138. Their goal was to estimate biogenic calcite, biogenic opal and non-biogenic contents from

From: ROTHWELL,R.G. 2006. New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 129-140. 0305-8719/06/$15.00 (C The Geological Society of London 2006.

130

R. D. JARRARD & M. D. VANDEN BERG

Fig. 1. Reflectance spectra from four climatically sensitive minerals. Previous marine sediment studies measured the mostly visible bandwidth in the highlighted area. The additional spectral information present in the expanded near-infrared region greatly improves identification of minerals affected by palaeoclimatic changes. recovered cores. Their estimates were best for biogenic calcite, but opal and non-biogenic material were not always distinguished reliably. A revised split-core analysis track instrument with an improved signal-to-noise ratio and wider frequency band (250-950 nm) was used on Legs 154, 162 and 167 (Harris et al. 1997; Shipboard Scientific Party 1997; Ortiz et al. 1999). Starting with Leg 154, shipboard core scanning with the Minolta spectrophotometer (400-700nm) became a routine measurement and used as a proxy for various minerals, particularly calcite (e.g. Shipboard Scientific Party 1995). Balsam et al. (1999) evaluated the use of visible-light (400-700 nm) spectroscopy, or optical lightness, for determining carbonate content of marine sediments in five Atlantic piston cores and Ocean Drilling Program (ODP) Hole 997A. Their carbonate estimates were good, but they warned that optical lightness is strongly affected by the composition of the non-carbonate fraction, such as clay. Balsam & Deaton (1996) used 250-850 nm spectra to estimate concentrations of carbonate, opal and organic carbon in various Atlantic piston cores and at ODP Site 847 (East Pacific Rise). They found that the character of downcore changes in mineralogy was well determined, but systematic offsets were sometimes evident for individual mineral concentrations. Additional visible-range spectral research, summarized by Balsam & Damuth (2000), includes use of the slope of the reflectance

curve to estimate chlorite and organic carbon (Balsam & Deaton 1991) or hematite and goethite (Deaton & Balsam 1991) and factor analysis to extract indicators of up to five components (Balsam & Damuth 2000).

Visible and near-infrared spectroscopy In contrast to the studies above employing mainly visible light, visible and near-infrared spectroscopy (VNIS) measures spectra for wavelengths of 350-2500nm, corresponding to the visible, near-infrared and part of the nearultraviolet regions of the electromagnetic spectrum. This range of wavelengths is also commonly examined by remote-sensing spectroscopy studies. The expanded near-infrared region provides a suite of additional spectral features that offer the potential of improved identification of minerals affected by palaeoclimate (Fig. 1). By using local ground-truth to calibrate specific spectral responses, one can develop algorithms for calculating concentrations of calcite, opal, smectite and illite. Instrumentation

Our VNIS studies use the Analytical Spectral Devices FieldSpec Pro FR Portable Spectroradiometer (Hatchell 1999). The light source is a high-intensity light probe with a quartz halogen

VISIBLE AND NEAR-INFRARED SPECTROSCOPY bulb and a DC current stabilizer for uniform light intensity. We chose the FieldSpec because it is also useful for Integrated Ocean Drilling Program (IODP) fieldwork; the Analytical Spectral Devices LabSpec Pro, with internal light source, is more convenient for I O D P use. Spectral data are collected on an attached microcomputer. Before any reflectance measurement is taken, spectral response of the light source needs to be normalized to a 100% reflectance level at all measured wavelengths. This is accomplished by using a Spectralon -~: white calibration plate. This calibration step is repeated every 3-5 min during routine measurement sequences. The Analytical Spectral Devices spectroradiometer contains three spectrometers, each measuring a different bandwidth. Subtle sensitivity differences among the three spectrometers generate vertical offsets at the boundaries between adjacent spectrometer wavelengths. These offsets are readily removed with a spreadsheet algorithm. Spectral noise is sometimes recorded between 350 and 400nm, caused by fluctuations in the spectral signature of the light source. This noise does not interfere with the mineral identification technique.

Illumination geometry Figure 2 shows the instrument, along with illustrations of two different instrumentation setups used to generate VNIS measurements on ODP core samples. The first instrument configuration (Fig. 2a) is used when the samples are relatively flat and large enough to cover the entire 2.5 cm-diameter opening in the bottom of the high-intensity light probe (e.g. split cores or powders). The fibre-optic detector cable, which conveys the reflected light to the spectrometer, is inserted within the light probe. The light probe is placed directly onto the sample and reflected light is transmitted via fibre-optic cable to the spectrometer. The light probe must be cleaned between measurements to prevent sample contamination. This set-up results in a more reproducible total reflectance value, a useful parameter for mineral concentration calculations. The second method (Fig. 2b) is used for smaller or irregularly shaped samples. The light probe is placed at the same angle and distance in relation to the sample as the fibre-optic sensor, to maximize detected reflectance. Nevertheless, total reflectance is lower than for the first method, and some extraneous light may be recorded. This method requires movement of only the sample and involves no contamination issues, resulting in faster measurement time.

131

Figure 3a compares the spectrum from an irregularly shaped core plug with that of the same sample after grinding to powder. Both techniques work well.

Effect o f water Water has a strong spectral signature in the near infrared: a large water absorption trough at 1900nm and an OH trough at 1400nm. These characteristics make VNIS a sensitive indicator of subtle variations in concentrations of hydrous minerals (e.g. smectite in basalts and sediments, opal). However, if samples are wet, the nearinfrared (950-2500nm) spectral signature of pore water dominates spectral features of mineralogical origin (Fig. 3b), particularly those of hydrous minerals such as smectite and opal. Also, wet sediments are typically darker than dry sediments, which affects total reflectance readings (Balsam et al. 1998). Consequently, VNIS is less adaptable to continuous-core measurements on saturated sediments than are visible-light techniques such as the Minolta spectrophotometer measurements. For visible-wavelength measurements, wet or dry cores can be analysed with minimal sample preparation: slight surface scraping or covering with clingfilm may be needed. In contrast, we routinely use VNIS only on dried c. 10cm 3 sediment samples or dry non-ODP cores. For example, we determined clay mineralogy of the mid-Tertiary CIROS-1 core from Antarctica (Vanden Berg & Jarrard 2001) and hydration of ODP basalts (Kerneklian & Jarrard 2006) by placing the VNIS light probe, with its internal fibre-optic detector, directly on the split-core surface. It may be feasible to analyse sediments lacking hydrous minerals without drying by removing the water spectrum. Remote-sensing VNIS analyses face an analogous problem and use a different technique: effects of atmospheric water vapour on the spectra are avoided by notching out portions of the spectrum (e.g. Clark et al. 2003).

Pilot studies The pilot studies described below were performed to evaluate the usefulness of VNIS for compositional determination and to investigate sources of error.

Mineral standards Four of the most common minerals in pelagic sediments are opal, calcite, smectite and illite.

132

R. D. JARRARD & M. D. VANDEN BERG

Fig. 2. VNIS instrumentation set-up. The first set-up (A), with the light probe placed directly on the sample, is used for split cores and powdered samples. The second set-up (B) is used when the sample is small or oddly shaped. (C) Photograph of the instrument, showing its components. We prepared end-point standards of these minerals, along with 50 : 50 mixes of pairs of these end points. VNIS response of some minerals is sensitive to particle size (Clark et al. 1990), so we

selected standards with grain sizes comparable to those in target pelagic sediments. Our 'opal' standard is a diatom ooze, and our 'calcite' standard is a nannofossil ooze; both are from O D P

VISIBLE AND 0 .

.

,

~A

.

.

NEAR-INFRARED .

' ~

,

.

.

.

.

,

"ooi~r

SPECTROSCOPY .

.

" '

.

.

133

. . . .

,

,

0.1 1.2 I~ ' lip 1.0

-

. . . . . . . .

'

-

. . . . . . . . .

.

8 o.~ 0.6 0.4 0.2 0.0

.8

0

.

ii I

2

. . . . .

9

,

~

. . . .

,

. . . .

,

Diatom ooze

........

/ / ~

80.8 D

~ 0.7 n-

0.6 0.~ 0.41

,

. . . .

,

. . . .

~ ,

1s

,

.

,

.

,

.

.

.

E

.

.

.

.

, .

.

,

~ !

.

,

,

r

-

9

. . . .

depths: 192

~0.6

"0.4

,

0.'.

0.0

~(Chert) "

"'"

i

500

.

" ............

.

.

.

,

1000

.

"

.

.

.

! " " "

,

1500 Wavelength (nm)

,,,"",

19~]''"''"

,

,

-

,

.

,

2000

. . . .

""

2500

Fig. 3. V N I S s p e c t r a f o r pilot studies o f sources o f v a r i a n c e . (A) C o r e plug v. p o w d e r o f s a m e s a m p l e . (B) D r y v. wet calcite-rich s a m p l e . (C) lllite v. s m e c t i t e f r o m the e q u a t o r i a l Pacific. (D) R a d i o l a r i a n o o z e v. d i a t o m ooze. (E) O p a l - A v. o p a l - C T .

134

R.D. JARRARD & M. D. VANDEN BERG

intervals that are moderately pure (less than 2% other components). The smectite standard is Namontmorillonite (Swy-1) from the Clay Mineral Society, and the illite standard is from Ward's Natural Science Establishment, Inc. (46 E 4100). Figure 1 shows VNIS spectra for the four endpoint mineral- or sediment-type standards. As expected, the four end points have distinctively different spectra, each with diagnostic features. For example, calcite has a diagnostic absorption trough at approximately 2350nm and has the highest total reflectance. Illite displays a small water trough at 1900nm and has the lowest total reflectance. Both smectite and opal have a large 1900nm water trough, but smectite also has several absorption troughs at 22002300nm, whereas opal has a much deeper and broader absorption trough at 1400nm than the other minerals. As expected, based on VNIS results for other minerals (Clark et al. 1990), the 50:50 mixes have generally intermediate spectral characteristics, but the spectral responses are non-linear. Non-linear responses necessitate use of several intermediate mixes for calibration of spectral responses.

Equatorial Pacific sediments Our second pilot study (Vanden Berg & Jarrard 2002) was a reconnaissance of quantitative concentration determination for palaeoclimatically significant minerals. To a first approximation, near-equatorial Pacific sediments can be thought of as consisting of calcite (nannofossils and/or foraminifers), opal (radiolarians and/or diatoms) and terrigenous components. We took 24 samples from ODP Site 846 (Leg 138), an eastern equatorial Pacific site in which the dominant components are calcite and opal. Samples were chosen at the same locations (+3cm) as ones previously analysed by Mix et al. (1995) for calcite, opal and 'other' (100%-calcite % - o p a l %). Opal concentrations may be slightly underestimated because of incomplete opal dissolution (Farrell et al. 1995). These ground-truth measurements were used to quantify the responses of individual spectral features. We determined calcite based on the calcite absorption band at approximately 2350nm. For opal, we used three spectral characteristics (depth of the 1900nm water trough, drop between 1300 and 900nm, and drop between 900 and 400nm), recognizing that each could also be affected by a different component, and would therefore be most accurate and useful in different environments. This feasibility study demonstrated that VNIS can achieve an accuracy of +10% for

calcite and opal determination at this site. This accuracy is not directly portable to sites with quite different concentrations of calcite, opal and other components, but we expect the same approach of local ground-truth for VNIS results to succeed in many other environments.

Low-latitude Pacific clays Clay minerals deposited in the Pacific Ocean are mainly derived from wind-blown continental dust (Rea 1994). Wind patterns that tap distinct source areas deposit different clay-mineral assemblages: illite-rich clay minerals from arid regions in Asia are brought into the North Pacific by the Westerly winds that are dominant above 25~ whereas smectite-rich clay minerals from the volcanic regions of Central and South America are carried into the North Pacific by the Northeast Trades that dominate latitudes between 25~ and the inter-tropical convergence zone, which is near the equator (Merrill et al. 1994; Rea 1994). A small pilot study was conducted to confirm that Pacific deep-sea clay mineralogy could be identified by the VNIS technique. Three high clay-content samples were chosen from the site survey cores for ODP Leg 199. Xray diffraction (XRD) and VNIS analyses were performed (Vanden Berg 2003). The XRD analyses showed that two samples contain dominantly illite with lesser amounts of kaolinite, chlorite and smectite. In contrast, the third sample contains dominantly smectite with lesser amounts of illite and kaolinite, but no chlorite. VNIS analyses of these samples (Fig. 3c) indicate a definite spectral difference between the smectite-rich sample and the two illite-rich samples. This outcome confirms the results from our first feasibility study, which indicated that smectite and illite should be readily distinguishable with VNIS. Because the spectral signatures of both smectite and illite are compositionally dependent (Clark et al. 1990), this study is more diagnostic of VNIS usefulness for smectite v. illite for the equatorial Pacific. VNIS could not recognize the small amounts of chlorite or kaolinite, but the dominant clay mineral was easily identified. This study shows that VNIS can be a useful tool for rapid identification of clay minerals in cores, aiding reconstruction of Pacific wind patterns.

Opal Our first study involving opal was to determine whether diatom-derived opal has a different spectral signature than radiolarian-derived

VISIBLE AND NEAR-INFRARED SPECTROSCOPY

135

Fig. 4. Reconnaissance study using VNIS for determination of organic carbon percentage. (A) Calibration of VNIS to conventional chemical method. (B) Spectral effects of organic carbon at short wavelengths.

opal. There appears to be a subtle difference between the two biogenic opal sources (Fig. 3d), perhaps due to hydration differences or trace-mineral contaminants. Next, we investigated whether the spectral signature of opal changes at the opal-A-CT boundary. Seven samples were chosen from Deep Sea Drilling Project Site t66, which is located in the Central Pacific Basin (3~ 175~ The upper five samples are late Miocene-middle Eocene radiolarian oozes and the two deeper samples are middle Eocene cherts (Winterer et al. 1973). Radiolarians were separated from the clays by sieving. Even though the crystallinity of the two zones is different, the spectral signature of opal-A radiolarian oozes is very similar to that of the opal-CT cherts (Fig. 3e). The total reflectance difference is attributable to the measuring technique (powdered ooze v. chert pieces) and does not reflect true opal-A-CT changes. These results suggest that VNIS-based opal determination may not be adversely affected by the opal-A-CT transformation. Organic matter

We undertook a small pilot study of the feasibility of using VNIS to estimate organic carbon content. Fifteen samples, containing 0.3-8.0% organic carbon (M. Lyle pers. comm. 2000), were analysed from sites 882 and 1014. Reflectance in the portion of VNIS spectrum below about 1800 nm wavelength is sensitive to organic carbon percentage. The drop in the interval 1300-900 nm provides a simple and moderately robust estimate of organic carbon, because here the organic carbon effect is largest and the effects of other common pelagic components are

relatively small. This reconnaissance study suggests that organic carbon should be considered in VNIS analyses whenever its abundance is more than about 2%, and that the accuracy of resulting organic carbon estimates is about +2% (Fig. 4). This initial accuracy is much less than has already been obtained from the mainly visible portion of the spectrum (250-850nm) (Balsam & Deaton 1991, 1996), so future VNIS-based analyses of organic carbon should be capable of surpassing the accuracy of our pilot study.

Mineral calculation VNIS is routinely used in remote sensing and mineral exploration as a mineral identification technique, not for calculation of mineral concentrations. The United States Geological Survey Spectroscopy Laboratory has led this effort, culminating in development of the Tetracorder software package for automatic identification of several spectrally dominant minerals from each VNIS spectrum, based on detection of diagnostic spectral absorption features (Clark et al. 2003). IODP objectives for VNIS differ from remote-sensing ones, however: we know what minerals are likely to be present within a site, and we seek estimates of their concentrations. To calibrate the VNIS response to local mineralogy, we use geochemical analyses of samples from the same region. An alternative approach is to use mixes of pure mineral standards. However, VNIS response is sensitive to both grain size and - especially for clays subtle compositional variations (Clark et al. 1990). An additional advantage of using local ground-truth samples is that spectral features

136

R . D . J A R R A R D & M. D. V A N D E N BERG

A

100

'

'

'

I

'

'

,

I

'

'

'

I

'

'

'

I

' ~

Calcite

Opal

=,

60 o~ 80 i

rE

60

9

"O ID

9

I l l

40

i i

9

~. 40 o9 z

9

m i

9

l

l

l

9

20 9

20

9

9

"

I

R = 0.970 ,

,

I

06

~

20

L

,

I

,

~

,

1

,

,

L

I

t

,

0i

J

O0

40 60 80 G r o u n d - t r u t h mineral %

~,

~

~

,

0

i

,

J

,

i

RMSE = 4.2% ,

,

,

h

20 40 Ground-truth mineral %

, , , r , , , , , , , ] , , , a '

100

,

'

i

,

60

,

Smectite 8O e,_

E 60

"O

| "O

~. 40 9

i

S

9

z 2O ~"

R = 0.951 R M S E = 6 . 1 %

i f

0

,

I

0

,

,

L

20

[

f

,

,

40

I

r

,

,

6O

I

~

,

80

~_

100

G r o u n d - t r u t h mineral %

B

lOO

'

'

~

'

~

'

I

i

,-,

i=,

,

---

,

Smectite o~

,

100

,

, , , , , , , , , , , , , , - ~ , , Iliite

II m

8O

80

60

6O

/

..

: /

~

~ 9

r" ._

E

"O

"6

"O

~

40

z 20,

/

4~I 20

R=o: 9

R = 0.978 0 -

'

0

'

'

I

20

,

~

,

I

40

,

,

,

I

60

~

,

,

J

80

Ground-truth mineral %

, ~

,

100

0

20

40

60

80

100

G r o u n d - t r u t h mineral %

Fig. 5. Cross-plots comparing VNIS predicted mineralogy with ground-truth mineralogy for Leg 199 sediments. (A) Three mineral (calcite, opal, smectite) determination. (B) Determination of smectite v. illite.

VISIBLE AND NEAR-INFRARED SPECTROSCOPY sensitive to other minor components (e.g. organic matter) can be identified and avoided. Our first large-scale ODP application of VNIS was on Leg 199, which cored an equatorial Pacific transect, in order to study equatorial palaeoceanography and palaeoclimate of the last 55Ma (Lyle et al. 2002). Our goal was to determine high-resolution mass accumulation rates of opal, calcite, smectite and illite for the eight Leg 199 sites. We achieved this high resolution by determining mineralogy first from VNIS and then using the VNIS-based mineralogy as ground truth for conversion of multi-sensor track logs to continuous mineralogy logs. Cores from the eight Leg 199 sites were sampled at a spacing of approximately 0.75 m, then all 1343 samples were dried, crushed to powder and their reflected spectra were recorded. Our VNIS analysis begins with evaluation of which spectral features are sensitive to mineral concentrations, followed by quantifying each relationship (including any non-linearities). Onboard Leg 199, mineralogy was calculated from VNIS spectra (Lyle et al. 2002) by using the algorithm of Vanden Berg & Jarrard (2002), based on multiple regression of spectral features on ground-truth mineralogy, followed by matrix inversion. Post-cruise analyses, however, demonstrated that significantly better agreement between predicted and ground-truth mineralogy could be obtained by just using stepwise multiple regression, with each ground-truth mineral percentage as a dependent variable and with the suite of spectral features as independent variables (Vanden Berg & Jarrard 2006). Additional improvement was obtained by applying different solutions for clay intervals (smectite v. illite) than for biogenous intervals (calcite, opal and smectite). The calculations sometimes predict slightly negative concentrations for a mineral component; these are converted to zero and then all concentrations are adjusted to total 100%. Figure 5 compares known mineral concentrations to VNIS-predicted mineral concentrations, for the multiple regression method of Vanden Berg & Jarrard (2006). The correlation coefficients range from a high of 0.99 for illite to a low of 0.95 for the three-mineral smectite solution; root-mean-square (rms) errors are 4-8% (Fig. 5). We note, however, that the determinations of relative percentages of smectite v. illite are much less accurate than their excellent correlation coefficients suggest, because of the paucity of intermediate mixes (Fig. 5b). The accuracy of VNIS-based calcite concentrations can also be evaluated by comparison to independent coulometer calcite concentrations (Fig. 6). The correlation between the two datasets is very good (VNIS predicted calcite is • of

137

coulometer calcite). Calcite was measured by couiometer on Leg 199 at a sample spacing of approximately 4.6 m, whereas VNIS-based calcite concentrations were measured at a spacing of about 0.75 m.

Discussion and conclusions When multiple regression is used to determine transforms relating mineral concentration to geophysical properties, a potential pitfall is that some predictive variables are merely chance (spurious) correlations. This problem is greatest when the number of analysed independent variables is large and the number of independent variables retained in a multiple-regression solution is large. For example, Balsam & Deaton (1996) undertook multiple regression on 61 'independent' variables (reflectances at 10nm spectral spacing), and their final regression equations contained between eight and 13 of these independent variables. One test for this problem is to split the ground-truth dataset into calibration and verification portions (e.g. Mix et al. 1992; Balsam & Deaton 1996). Another approach is to reduce the number of independent variables. For example, Handwerger & Jarrard (2003) used principal components analyses to reduce the number of variables employed in their mineralogy regression. Our approach begins like the Tetracorder method of remote-sensing spectral analysis (Clark et al. 2003) by identifying and quantifying a few distinctive spectral characteristics, such as depth of the 1400nm OH- absorption trough. With this relatively small number of potential independent variables, we then use stepwise multiple regression to confine the final predictive equations to only the most significant terms. Any empirical transform, such as those relating VNIS or other spectroscopy data to mineralogy, is subject to an additional potential pitfall: lack of generality. A calibration dataset from a single IODP site can provide excellent transforms for measurements within that site, yet it may generate only fair mineralogy predictions for nearby sites and poor mineralogy predictions for other oceans. Balsam & Deaton (1996) discussed this problem in detail, noting the possible influences of regional variations in matrix effects (other background minerals) and in composition of analysed minerals. They found, for example, that they could predict carbonate concentration based on 250-850nm spectroscopy with a rms error of 6% within Site 847 but 11.6% for circum-Atlantic data. For comparison, VNIS-based carbonate results

138

R.D. JARRARD & M. D. VANDEN BERG 100

'

'

'

]

'

'

'

I

'

'

'

I

'

'

'

I

'

'

Site 1215

'

'

I

'

'

'

I

'

'

'

I

'

'

'

I

'

Site 1218

9

'0'

9149

80

m 60 "_~ CO

z>

40 O

O

9

20

R -- 0.982 ,

0 100

,

'

,

'

i

'

I

'

L

L

'

I

'

I

'

I

h

'

I

'

I

'

L

J

'

l

, ~

l

'

I

'

I

,

,

,

1

~

,

1

,

,

,

1

~

,O

Site 1219

Site 1220 9

v

80

j

60

9149

'6 ~o

%

CO

z

>

40

20 /~ 0

:9~

0

9 ,

,

R = 0.992

R = 0.980 I

20

,

,

,

J

,

~

~

I

,

~

L

1

L

L

~

40 60 80 coulometer calcite (%)

,

100 (~

,

,

I

20

,

,

,

[

,

,

,

I

,

,

,

I

40 60 80 coulometer calcite (%)

100

Fig. 6. VNIS-based calcite concentrations compared with calcite concentrations measured by coulometer for

four Leg 199 sites. for our four sites of Figure 6 have rms errors of 4-6%. Individual studies deal with this problem by assuring that their ground-truth, or calibration, dataset is representative of their unknowns. We conclude that absorption spectroscopy is a rapid and non-destructive way to determine concentrations of several minerals using only their spectral signature. Our own studies, which were centred on devising a simple, routine method for determining concentrations of calcite, opal, smectite and illite, indicate that VNIS can provide accurate, high-resolution mineralogy for many IODP sites. Because the technique is so rapid, pilot studies to investigate other potential VNIS uses can be performed without sacrificing much time.

The primary limitations of the VNIS technique are that the mineral of interest must have a distinctive spectral signature, the mineralogy must not be overly complex (four or five main minerals at maximum) and the cores or core samples should be dry. In contrast, visible-light spectroscopy is much less sensitive to pore water but relies on a much narrower bandwidth. For water-saturated cores, a combination of continuous-core light spectroscopy and discrete-sample VNIS (e.g. Vanden Berg & Jarrard 2006) can be fruitful. A detailed comparison is needed of the results of light spectroscopy and VNIS on the same sample suite, to assess the incremental value of adding spectral information from the near-infrared. Reflectivity in the near-ultraviolet (c. 250-400 nm) is also mineralogically significant

VISIBLE AND NEAR-INFRARED SPECTROSCOPY (Balsam & Denton 1996), so ideally the comparisons would encompass the entire spectral range of 250-2500 nm. We thank M.W. Lyle for help throughout this project, and we thank W. Balsam for his constructive review. This research used data provided by the Ocean Drilling Program (ODP). O D P is sponsored by the US National Science F o u n d a t i o n and participating countries under m a n a g e m e n t of Joint Oceanographic Institutions, Inc. Funding for this research was provided to M.D. Vanden Berg and R.D. Jarrard by the US Science Support Program.

References ADAMS, J.B. 1974. Visible and near-infrared diffuse reflectance spectra of pyroxenes as applied to remote sensing of solid objects in the Solar System. Journal of Geophysical Research, 79, 4829-4836. BALSAM,W.L. & DAMUTH,J.E. 2000. Further investigations of shipboard vs. shore-based spectral data: implications for interpreting Leg 164 sediment composition. Proceedings of the Ocean Drilling Program, Scientific Results, 164 [Online]. Available from World Wide Web: http://www.odp-tamu. edu/publications/164_SR/chap_31/chap_31 .htm. BALSAM, W.L. & DEATON, B.C. 1991. Sediment dispersal in the Atlantic Ocean: Evaluation by visible light spectra. Reviews in Aquatic Science, 4, 411417. BALSAM,W.L. & DEATON, J.E. 1996. Determining the composition of late Quaternary marine sediments from NUV, VIS, and NIR diffuse reflectance spectra. Marine Geology, 134, 31-55. BALSAM, W.L., DEATON, B.C. & DAMUTH, J.E. 1998. The effects of water content on diffuse reflectance spectrophotometry studies of deep-sea sediment cores. Marine Geology, 149, 177-189. BALSAM, W.L., DEATON, B.C. & DAMUTH J.E. 1999. Evaluating optical lightness as a proxy for carbonate content in marine sediment cores: Implications for marine sedimentation. Marine Geology, 161, 141-153. BURNS, R. 1993. Mineralogical Applications of Crystal Field Theory. Cambridge University Press, Cambridge. CLARK, R.N. 1983. Spectral properties of mixtures of montmorillonite and dark carbon grains: Implications for remote sensing minerals containing chemically and physically adsorbed water. Journal of Geophysical Research, 88, 10,635-10,644. CLARK, R.N. 1995. Reflectance spectra. In: AHRENS, T.J. (ed.) Rock Physics and Phase Relations. A Handbook of Physical Constants. American Geophysical Union, Washington, DC, 178-188. CLARK, R.N. & ROUSH,T.L. 1984. Reflectance spectroscopy: Quantitative analysis techniques for remote sensing applications. Journal of Geophysical Research, 89, 6329-6340. CLARK, R.N., KING, T.V.V., KLEJWA,M., SWAYZE,G. & VERGO,N. 1990. High resolution reflectance spec-

139

troscopy of minerals. Journal of Geophysical Research, 95, 12,653-12,680. CLARK, R.N. SWAYZE,G.A. & GALLAGHER,A. 1993. Mapping minerals with imaging spectroscopy. U.S. Geological Survey. Office of Mineral Resources Bulletin, 2039, 141-150. CLARK, R.N., SWAYZE, G.A., GALLAGHER,A., GORELICK, N. & KRUSE, F. 1991. Mapping with imaging spectrometer data using the complete band shape least-squares algorithm simultaneously fit to multiple spectral features from multiple materials. Proceedings of the Third Airborne Visible/InJiared Imaging Spectrometer (A VIRIS) Workshop. Jet Propulsion Lab Publication, 91-28, 2-3. CLARK, R.N., SWAYZE, G.A.F.T aL. 2003. Imaging spectroscopy: Earth and planetary remote sensing with the USGS Tetracorder and expert systems. Journal of Geophysical Research, 108, 5131, doi: 10.1029/2002JE001847. DEATON, B.C. & BALSAM, W.L. 1991. Visible spectroscopy - a rapid method for determining hematite and goethite concentrations in geological materials. Journal of Sedimentary Petrology, 61, 628-632. FARRELL, J.W., RAFFI, I. ET AL. 1995. Late Neogene sedimentation patterns in the eastern equatorial Pacific Ocean. Proceedings of the Ocean Drilling Program, Scientific Results, 138, 717-753. GOETZ, A.F.H., BARRETT,N.R. & ROWAN, L.C. 1983. Remote sensing for exploration: An overview. Economic Geology, 78, 573-590. HANDWERGER, D. & JARRARD, R.D. 2003. Neogene changes in Southern Ocean sedimentation based on mass accumulation rates at four continental margins. Paleoceanography, 18, 1081, doi:10.1029/ 2002PA000850. HAPKE, B. 1993. Introduction to the Theory of Reflectance and Emittance Spectroscopy. Cambridge University Press, New York. HARIUS, S.E., Mix, A.C. & KING, T. 1997. Biogenic and terrigenous sedimentation at Ceara Rise, western tropical Atlantic, supports Pliocene-Pleistocene deep-water linkage between hemispheres. Proceedings of the Ocean Drilling Program, Scientific Results, 154, 331-345. HATCHELL, D.C. (ed.). 1999. Analytical Spectral Devices, Inc. Technical Guide, 3rd edn. Analytical Spectral Devices, Inc., Boulder, CO. HUNT, G.R. 1977. Spectral signatures of particulate minerals in the visible and near-infrared. Geophysics, 44, 1974-1986. HUNT, G.R. 1979. Near-infrared (1.3-2.4 jam) spectra of alteration minerals - Potential for use in remote sensing. Geophysics, 44, 1974-1986. HUNT, G.R. & SALISBURY,J.W. 1970. Visible and near infrared spectra of minerals and rocks, I, Silicate minerals. Modern Geology, 1,283-300. HUNT, G.R. & SALISBURY,J.W. 1971. Visible and near infrared spectra of minerals and rocks, II, Carbonates. Modern Geology, 2, 23-30. HUNT, G.R., SALISBURY,J.W. & LENHOFF,C.J. 1971a. Visible and near infrared spectra of minerals and rocks, III, Oxides and hydroxides. Modern Geology, 2, 195-205.

140

R . D . J A R R A R D & M. D. VANDEN BERG

HUNT, G.R., SALISBURY,J.W. & LENHOFF, C.J. 1971b. Visible and near infrared spectra of minerals and rocks, IV, Halides, arsenates, vanadates, and borates. Modern Geology, 3, 1-14. HUNT, G.R., SALISBURY,J.W. ~; LENHOFF, C.J. 1972. Visible and near infrared spectra of minerals and rocks, V, Oxides and hydroxides. Modern Geology, 3, 121-132. HUNT, G.R., SALISBURY,J.W. t~ LENHOFF, C.J. 1973. Visible and near infrared spectra of minerals and rocks, VI, Additional silicates. Modern Geology, 4, 85-106. KERNEKLIAN, M.J. & JARRARD, R.D. 2006. Data report: Visible and near-infrared spectroscopy as an indicator of hydration in the upper oceanic crust of ODP Site 1256. Proceedings of the Ocean Drilling Program, Scientific Results, 206, in press. LYLE, M.W., WILSON, P.A. ET AL. 2002. Proceedings of the Ocean Drilling Program, Initial Reports, 199 (CD-ROM). MERRILL, J., ARNOLD, E., LEINEN, M. & WEAVER, C. 1994. Mineralogy of aeolian dust reaching the North Pacific Ocean 2. Relationship of mineral assemblages to atmospheric transport patterns. Journal of Geophysical Research, 99, 21,025-21,032. MIX, A.C., HARRIS, S.E. & JANECEK, T.R. 1995. Estimating lithology from nonintrusive reflectance spectra: Leg 138. Proceedings of the Ocean Drilling Program, Scientific Results, 138, 413-427. Mix, A.C., RUGH, W., PISIAS,N.G., VIERS, S., LEG 138 SHIPBOARD SEDIMENTOLOGISTS& LEG 138 SCIENTIFIC PARTY. 1992. Color reflectance spectroscopy; a tool for rapid characterization of deep-sea sediments. Proceedings of the Ocean Drilling Program, Initial Reports, 138, 67-78. ORTIZ, J.D., O'CONNELL, S. & MIX, A. 1999. Data report: spectral reflectance observations from recovered sediments. Proceedings of the Ocean Drilling Program, Scientific Results, 162, 259-264. REA, D.K. 1994. The paleoclimatic record provided by eolian deposition in the deep sea: The geologic

history of wind. Reviews of Geophysics, 32, 159195. SHIPBOARDSCIENTIFICPARTY. 1995. Synthesis. Proceed-

ings of the Ocean Drilling Program, Initial Reports, 154, 421-442. SHIPBOARDSCIENTIFICPARTY. 1997. Explanatory notes.

Proceedings of the Ocean Drilling Program, Initial Reports, 167, 15-39. SINGER,R.B. 1981. Near-infrared reflectance of mineral mixtures: Systematic combinations of pyroxenes, olivine, and iron oxides. Journal of Geophysical Research, 86, 7967-7982. SWAYZE, G.A. & CLARK, R.N. 1990. Infrared spectra and crystal chemistry of scapolites: Implcations for Martian mineralogy. Journal of Geophysical Research, 95, 14,481-14,495. VANDEN BERG, M.D. 2003. The use of light absorption

spectroscopy as a mineral identification tool."Implications.for the study of Cenozoicpaleoclimate. Master's thesis, University of Utah. VANDEN BERG, M.D. & JARRARD, R.D. 2001. Light absorption spectroscopy as a paleoclimate and correlation technique for the CRP and CIROS-1 drill cores, McMurdo, Sound, Antarctica, Los, Transactions of the American Geophysical Union, 82, (47), Fall Meeting Supplement, abstract PP51A-0543. VANDEN BERG, M.D. & JARRARD, R.D. 2002. Determination of Equatorial Pacific mineralogy using light absorption spectroscopy. Proceedings of the Ocean Drilling Program, Initial Reports, 199, 1-20 (CD-ROM). VANDEN BERG, M.D. & JARRARD, R.D. 2006. Data report: High-resolution mineralogy for Leg 199 based on reflectance spectroscopy and physical properties. Proceedings of the Ocean Drilling Program, Scientific Results, 199, 1-23 [Online]. Available from World Wide Web: http://www.odp-tamu.edu/ publications/199_S R/volume/chapters/203.pdf. WINTERER, E.L., EWING, J.I. E:r AL. 1973. Site 166. Initial Reports of the Deep Sea Drilling Project, 17, 103-144.

Applications of confocal macroscope-microscope luminescence imaging to sediment cores A. C. R I B E S 1, F. R. R A C K 2, G. T S I N T Z O U R A S

3, S. D A M A S K I N O S

4

& A. E. D I X O N 4

1DALSA, 605 McMurray Road, Waterloo, Ontario, Canada N 2 V 2E9 (e-mail: [email protected]) 2Joint Oceanographic Institutions, 1201 New York Avenue, N W , Suite 400, Washington, DC 20005, USA (e-mail: [email protected]) 3Christie Digital Systems Inc., 809 Wellington St. N., Kitchener, Ontario, Canada N2G 4 Y7 4Biomedical Photometrics Inc. & GeneFocus, 550 Parkside Drive, Unit A12, Waterloo, Ontario, Canada N2L 5V4 Abstract: We demonstrate the successful application of a novel, confocal scanning laser macroscope-microscope (cslM/m) system for non-invasive imaging of samples taken from lake and ocean sediment cores. Advantages of the macroscope-microscope system over other macroscopic luminescence imaging techniques, such as scanning electron microscopy-based cathodoluminescence and scanning-stage laser imaging, are highlighted and the implications for new core analysis techniques are explored. The macroscope-microscope can image specimens ranging from 25 • 25~m up to 7.5 • 7.5 cm in under 10 s using reflected light or photoluminescence as contrast mechanisms. Macroscope mode is used to rapidly survey the specimen and provide a photoluminescence 'roadmap'. Microscope mode is used to provide ultra-high-resolution images of microfossils or areas of interest. Laser scanning is non-invasive and does not require any preparation of the specimen. Photoluminescence and fluorescence imaging results are shown for an entire core section recovered from Lake Huron by the Geological Survey of Canada (GSC). Photoluminescence images are shown for Ocean Drilling Program samples of a diatom mat and a radiolarian microfossil within the sample, a laminated interval of sediment from the Santa Barbara Basin (Site 893) and specimens from the Scotian Shelf (collected by the GSC).

Luminescence studies (Coyne et al. 1989; Barker & Kopp 1991; Senesi et al. 1991) of geological specimens have proven to be useful in identifying their chemical and structural make-up. Spatially resolved luminescence or luminescence imaging has been carried out extensively at microscopic (p,m) levels (Van Gijzel 1967; Petford & Miller 1992; Montemagno & Gray 1995; O'Connor 1996) but has seen limited use at macroscopic (cm) scales. We have imaged selected sediment and rock samples from a group of 30 specimens on loan from the international Ocean Drilling Program (ODP) using a confocal scanning laser macroscope-microscope (cslM/m) instrument (Dixon et al. 1995, 1996; Ribes et al. 1995, 1996). The macroscope-microscope can image specimens ranging in size from 25 • 25 p.m up to 7.5 z 7.5cm in less than 10s using either reflected light (RL) or photoluminescence (PL). This non-destructive measurement technique can easily resolve submillimetre features in cores, which allows for investigation of laminated sediment sequences at temporal resolutions of 1-100 years (e.g. human timescales).

The goal of this research is to explore the capabilities of the macroscope-microscope in order to potentially develop a wide range of geological applications. We illustrate the capabilities of the macroscope-microscope by imaging samples with a broad range of textures, compositions and sizes, ranging from an entire 1.5m-long, 7.5cm-diameter sediment core, down to a 25 lam-sized fragment of a microfossil skeleton.

Previous work There has been an extensive amount of imaging performed of geological specimens with confocal (Petford & Miller 1992; O'Connor 1996) and conventional (Van Gijzel 1967; Montemagno & Gray 1995) fluorescence microscopes. Fluorescence macroscopy, however, has seen little or no published research in the area of geology or sedimentary petrology. Flannagan & Huang (Huang & Bradford 1992; Flannagan et al. 1995) have used a largearea footprint (tens and hundreds of centimetres)

From: ROTHWELL,R.G. 2006. New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 141-150. 0305-8719/06/$15.00 9 The Geological Society of London 2006.

142

A. C. RIBES E T AL.

portable laser scanner to measure soil topography on site; however, this apparatus is closer to a scanning laser range finder than an imaging system. A successful attempt has been made by scanning electron microscopists to produce images spanning up to 1 cm in size using electron-induced autofluorescence (Coyne et al. 1989, chapter 11), also known as cathodoluminescence. Mn 2+doped carbonates and rare-earth-doped materials commonly exhibit fluorescence in the visible spectrum of light at room temperature. Cathodoluminescence (CL) imaging involves scanning a beam of highly energetic electrons onto a specimen (Holt & Joy 1989; Yacobi & Holt 1990). As these electrons traverse the specimen they will loose energy via ionization, which serves as the excitation mechanism for resident fluorophores. Once the fluorophores are in an excited state, light emission occurs via the same mechanism as with photon-excited luminescence. The advantage of CL imaging is that it mimics photoluminescence with a very short wavelength excitation; therefore, a complete autofluorescence emission spectrum is generated. Detection with a charge-coupled device (CCD) camera yields spectacular true-colour luminescence images. Unfortunately, there are several disadvantages associated with large-area CL imaging. First and foremost is the fact that a single macroscopic CL image is generated by assembling a large number of microscopic images into a collage. This technique, referred to as tiling, involves a large total scan time and a significant amount of post-processing. In many cases each image can vary significantly with its neighbours, as a function of background intensity, resulting in a very nonuniform total image. Most geological specimens are non-conducting; therefore CL on a scanning electron microscope requires the specimen to be coated with a conductor, such as a thin layer of gold. In many cases other forms of specimen preparation are required, such as polishing and cutting the specimen into thin sections. Even Fig. 1. (a) The confocal scanning laser macroscopemicroscope. This scanning-beam system can provide lateral resolutions ranging from 0.25 gm in microscope mode to 10 gm in macroscope mode. Images can be obtained in less than 10 s using reflected light or photoluminescence over fields of view ranging from 25 x 25 lam up to 7.5 x 7.5 cm. (b) A photograph of the macroscope with the microscope arm absent. In this configuration the instrument is optimized to image specimens up to 7.5 x 7.5 cm in size with 10 tam resolution. A key component of the macroscope is its imaging objective, which needs to have a diameter larger than the specimen being imaged as the scanning beam is always at right angles to the said specimen (it is a telecentric lens).

with cold-cathode CL techniques, where conductive specimens are not required, damage from highly ionizing electrons can result. The

macroscope-microscope

A schematic diagram of a combined confocal scanning laser macroscope-microscope developed by the Scanning Laser Microscopy Laboratory at the University of Waterloo (Ontario, Canada) is shown in Figure la (see also Table 1). This

(a)

(b)

LUMINESCENCE IMAGING TO SEDIMENT CORES

143

Table 1. A glossary of terms used in confocal microscope and macroscope imaging Beam splitter

An optical device used for separating an incident beam of light into two components. In photoluminescence imaging it is common to use a dichroic beam splitter that preferentially transmits light emitted from the specimen (e.g. red) and reflects light used to excite the specimen (e.g. blue)

Confocal Scanning Laser Microscope (CSLM)

An optical microscope in which a focused laser beam is scanned along the x and y axes of a specimen in a raster pattern. The reflected light or fluorescence emitted from the specimen is sensed by a photomultiplier tube and displayed as pixels on a computer screen. Light emitted away from the focal plane is blocked by a pinhole located in a plane confocal with the specimen. This technique allows the specimen to be optically sectioned along the z axis

Fluorescence emission spectrum

The spectrum of wavelengths emitted by the specimen after it has been excited by light. Typically, the average wavelength of emission is longer than the average wavelength of excitation. Also, the wavelength distribution of the fluorescence emission spectrum is generally independent of the excitation wavelength

Fluorescence

The process by which an atom or molecule, which is excited by absorption of radiation (usually ultraviolet or visible light), releases the absorbed energy as light having a wavelength longer than the absorbed radiation

Longpass (LP) filter

An optical interference or absorptive glass filter that reflects or absorbs shorter wavelengths and transmits longer wavelengths. In photoluminescence imaging a longpass filter is used in front of the detector to pass the fluorescence emission from the specimen and attenuate the reflected emissions. Without the longpass filter in the beam path, the reflected emission from the specimen will dominate and, hence, a reflected-light image can be obtained

Luminescence

The emission of light from any substance. Typically, fluorescence is associated with short emission lifetimes (ns) while phosphorescence can last seconds. The generation of luminescence can occur through excitation by light (photoluminescence), chemical energy (chemiluminescence), electron beams (cathodoluminescence), heat (thermoluminescence), biochemical energy (bioluminescence) or mechanical energy (triboluminescence)

Macroscope

Similar to a CSLM but with the ability to image specimens up to 7.5 x 7.5cm in size with a single scan. A combined macroscope-microscope becomes a very powerful reflectedlight and fluorescence imaging instrument

Photomultiplier tube (PMT)

An electronic device designed to collect and amplify light signals. Incoming photons strike the cathode of the photomultiplier to liberate electrons, which are accelerated onto a dynode that in turn liberates more electrons thus creating an amplification effect. Several dynodes are arranged in series to produce a measurable current pulse given the original photon. The PMT is effectively a photon to current converter and amplifier

instrument is a confocal (Pawley 1995) scanning laser system able to image in two distinct ways: (1) macroscope mode; and (2) microscope mode. In microscope m o d e the instrument operates like a conventional confocal microscope providing fields of view ranging from 25 x 25 ~tm up to about 4 x 4 m m in size, depending upon the microscope objective used. Resolutions (lateral) are also objective dependent, and range from 0.25 ~tm for a 0.9 numerical aperture (NA) objective to about 2~tm for a 0 . 1 4 N A objective. F r a m e times typically vary between 5 and 30s, and imaging can be carried out using reflected light or photoluminescence. In macroscope mode (a p h o t o g r a p h of a macroscope-only instrument is shown in Fig. lb), resolution is sacrificed in exchange for a m u c h larger field of view. Lateral resolution is limited to 10 lam while axial resolution

is equal to 300 lam in contrast to microscope m o d e where axial resolutions are on the order of microns (t~m). A detailed description of the m a c r o s c o p e microscope system follows. As shown in Figure l a, light from a laser is expanded to a 2 c m - d i a m e t e r collimated beam. A variety of laser sources can be used, such as h e l i u m - n e o n (633 nm, red), d i o d e - p u m p e d solid state (532nm, green), and a r g o n - i o n (488nm, blue). Laser light provides a high-quality, m o n o chromatic beam that can be easily focused d o w n to a diffraction-limited point. The expanded beam strikes a dichroic beam splitter that preferentially reflects the excitation wavelength and transmits the PL emission from the specimen. F o r reflected light imaging a 50/50 beam splitter is c o m m o n l y used. Two closely spaced galvano m e t e r scanning mirrors pivot the collimated

144

A. C. RIBES E T AL.

beam about two orthogonal axes. The beam pivots, approximately, about a point equidistant from each mirror, which in turn coincides with the front focal point of a telecentric f-theta laser scan lens. These custom-made lenses are designed to accept a collimated beam pivoting at their front focal point and to generate a focusing cone of light that is always at right angles to the focal plane regardless of scan angle (the lens is telecentric). Telecentricity promotes uniform images by ensuring that the incident and emitted beams travel along the same path and therefore avoid clipping and vignetting effects at large scan angles. The f-theta property ensures the laser spot on the specimen always moves in proportion to the scan angle even for large angles (>5~ What is commonly referred to as the macroscope lens provides a 7.5 x 7.5 cm field of view, a 10 ~tm lateral resolution and an effective NA equal to 0.04. This is an extremely large NA, given its field of view, considering the closest a microscope objective comes to the macroscope lens is a 0.025 NA, 1.7 x 1.7cm field of view Mitutoyo 1 x microscope objective. Light reflected or emitted from the specimen passes back through the macroscope lens, is descanned (that is, the beam returns along the same path it entered) by the scanning mirrors, passes through the dichroic beam splitter and is focused by the detector lens onto a confocal pinhole. Light passing through the confocal pinhole is detected by a photomultiplier tube. The confocal pinhole ensures that only light from the focal point of the macroscope lens will reach the detector. Light emitted by the sample originating from either above or below the focal plane of the imaging lens will be blocked by the confocal pinhole. Only objects near the focal plane will be imaged. How near an object must be to the focal plane for it to be imaged is defined by the axial resolution of the confocal system. Axial resolutions range from a few microns to about 0.5 ~tm for high NA microscope objectives in air. The macroscope lens has an axial resolution equal to 3001am. The primary advantage of confocal systems in both modes is the elimination of out-of-focal-plane light when imaging in PL. Photoluminescence from above and below the focal plane tends to blur small features in nonconfocal imaging, which effectively reduces the imaging system's resolution. Confocal photoluminescence imaging produces crisp, detailed images of features in the focal plane of interest. In order to create a two-dimensional image of a specimen, an imaginary grid can be superimposed on the specimen and the laser spot must

be moved to each point on this grid. Light reflected or emitted from each point on the grid is measured and recorded as a pixel intensity value forming part of a digital image. An image from the macroscope is simply a collection of single-point measurements. Instead of just measuring photoluminescence intensity at each point, a time-resolved or wavelength spectrum can also be recorded. For example, having recorded a photoluminescence spectrum at each point, one can plot peak intensity v. position on the specimen for a particular chemical species, thereby spatially resolving chemical composition along, or across, the sample. Each pixel value is generated by moving the specimen under a stationary beam (scanningstage method) or by raster scanning the beam under a stationary specimen (scanning-beam method). Traditionally, the scanning-stage method has been used to image large (> 1 x 1 cm) specimens because inexpensive, high-quality, onaxis optics can be used throughout the instrument. The maximum field of view is limited only by the stage motion. Two main disadvantages associated with this technique result from the fact that the specimen has to be moved. Many confocal applications have been developed for imaging biological specimens, such as cells immersed in a liquid solution. These types of specimens must remain stationary while imaging. Another disadvantage with moving stages is that objects can only be scanned slowly, such that vibrations are kept to a minimum and high resolution is maintained. Scan times for large specimens can therefore span several minutes. It is for these two reasons that commercial confocal microscopes have, almost exclusively, adopted scanning-beam systems. The macroscope was developed in the mid-1990s to extend fast (<10s) scan times associated with scanning-beam systems to macroscopic (wide-area) fields of view. The macroscope-microscope is a versatile instrument able to provide an image with a wide variety of contrast mechanisms. The combined macroscope-microscope instrument provides a zoom factor of 3000 for PL and RL images. In addition to total PL and RL imaging, it can also generate optical beam induced current (OBIC) images. An OBIC image represents a two-dimensional current map of a device such as a solar cell or photodetector. At the time this article was written the University of Waterloo (UW) Scanning Laser Microscopy Laboratory had four imaging systems in operation. Although in the past, a combined macroscope-microscope existed in the laboratory, this instrument has been split up into two

LUMINESCENCE IMAGING TO SEDIMENT CORES

145

separate instruments due to excessive demand. Most of the images presented herein were taken using either the macroscope or the laboratory's homemade confocal scanning laser microscope. The laboratory also has a high-resolution macroscope able to provide 0.1-0.2 NA over a 1-2 cm square field of view. The fourth instrument is a macroscope coupled to a 100-fs Ti:sapphire laser for carrying out 2-photon photoluminescence imaging.

Results The confocal scanning laser macroscope-microscope is an instrument that can provide fast scan rates, with a resolution of at least 10 gm laterally over a field of view spanning 25 x 25 gm to 7.5 x 7.5cm, a zoom factor of 3000. The 7.5 x 7.5 cm field of view can be scanned in one shot using PL or RL imaging without moving or preparing the specimen in any way. This represents a tremendous advantage over scanningstage systems, where the entire specimen must be moved, and scanning electron-beam CL systems, where extensive tiling needs to be performs for cm-sized specimens. A collection of 30 samples, representing a broad range of lithologies and textures recovered by the ODP, were used to evaluate the PL imaging capabilities of the macroscopemicroscope. These specimens included nannofossil and siliceous oozes, diatom mats, glacial diamicts, volcanic ash, organic-rich sapropels, corals, basalt, sands, silts, clays, etc. We show imaging results for a small set of these specimens. Relative fluorescence intensity emissions are tabulated for 18 specimens. A PL macroscope image (488 nm excitation) of a 1.5m-long section of core from Lake Huron sample (Lewis et al. 1994; Rea et al. 1994) is shown in Figure 2. The imaged core section (core 94800-001, interval 722-922cm) is from NW Lake Huron and has been fully described by Godsey et al. (1999). Among other characterization and imaging techniques, Godsey et al. used a colour digital CCD camera to map out varve thickness using reflected light. Our imaging technique complements this by providing PL images. The numbers on each side indicate the subbottom depth in centimetres. The composite core image in Figure 2, which is divided into two parts, was assembled from a total of 25 separate images taken at approximately 6cm intervals along the core. Only the central 6 cm of the 7.5cm-wide core is shown, due to x-position drift as the core was moved in the z-direction for each image. The entire composite core

Fig. 2. Photoluminescence image of an entire 1.5 mlong core from Lake Huron (side images). The numbers alongside denote depth in cm and the total core image consists of 25 individual merged images. The central image has been contrast enhanced and smoothed to highlight the banding features.

image, as shown on the right- and left-hand sides of Figure 2, consists of only the raw data and has not been enhanced in any way. Rhythmic banding or laminations are evident throughout most of the core section; however, the dynamic range of the PL emissions are such that the first 50cm of the core section appears very dark. The central image in Figure 2 shows the same interval of core, from about 790 to 810cm, imaged using a higher gain setting. This composite image consists of four separate 512 x 512 pixel images that were contrast enhanced and passed through a 3 x 3 median filter in order to bring out the horizontal banding pattern. Assuming a 10gin lateral resolution, extremely clear images with 1 x l cm fields of view are well within the resolving capabilities of the macroscope. Laminations, which are observed at millimetre or smaller spatial scales and, hence, represent sedimentological and palaeoenvironmental cycles on human timescales, are of great interest to palaeoclimatologists. These features are easily resolvable with the macroscope. The macroscope may be ideally suited as a shipboard assembly-line-type imaging device. When properly synchronized with the movement

146

A.C. RIBES E T AL.

(a)

2 cm

(b)

Fig. 3. (a) Photoluminescence macroscope image of a 'diatom mat" recovered from the Equatorial Pacific Ocean during ODP Leg 138 at Site 850. (b) Macroscope image of boxed area in (a). (e) Several fragments of siliceous microfossils imaged with a 0.7 NA microscope objective in PL. (d) Magnified view of a siliceous (radiolarian) fragment from boxed area in (c). of a split core section along a track in the z-direction, the entire 1.5m section can be imaged in about 10 min. While the macroscope is rugged enough to be used on a ship, this is not true of the highresolution confocal microscope system. A combined macroscope-microscope is therefore better suited for detailed analysis of cores at a repository, where environmental conditions and vibrations can be controlled and a large number of samples can be imaged. Scanning-stage microscopes are poorly suited for imaging lake and ocean cores due to the core's excessive size (i.e. 1.5m long). The macroscope adds at least an order of magnitude to the maximum imaging size obtained via tiling when compared to CL imaging, that is a larger initial field of view leads to larger composite images. A photoluminescence macroscope image (633 nm excitation) of a 4 • 6cm ODP specimen from the eastern Equatorial Pacific (sample 138850B-10H-7, 20-30cm) is shown in Figure 3a representing a 'diatom mat' comprised of microfossil skeletons and other materials. Banding is clearly evident, probably due to bioturbation and variations in sediment lithology. The origin of PL is most probably due to varying amounts of biogenic microfossils. Banding becomes somewhat obscured by the specimen's graininess at the

submillimetre level (Fig. 3b). A much higher resolution is needed to resolve the microfossils making up this diatom mat. A 200 • 200 ~tm PL image taken with a 0.7NA microscope objective (Fig. 3c) shows fragments from a radiolarian skeleton and an intact specimen in the upper portion of the image. Figure 3d shows one of the fragments at a higher magnification. The confocal effect is evident in this image because only the microfossil is visible while the surrounding background is black or empty as it lies outside the objective's axial resolution. This illustrates one of the macroscope-microscope's key strengths: the ability to quickly sample a large area and then zoom in on an area of interest much like changing magnification on a conventional microscope. Two images of the same specimen were obtained using different excitation wavelengths (Fig. 4). The specimen is a 7 • 2cm interval of a laminated sediment recovered from the central Santa Barbara Basin by ODP (sample 146-893A-1H-3, 8-18cm). This sample exhibits a significantly different PL emission when excited with blue (488nm) light (Fig. 4b) than when excited with red (633 nm) light (Fig. 4a). With 488 nm excitation, certain areas throughout the specimen luminesce with a much higher efficiency than the surrounding areas. 633nm excitation creates a much more uniform emission since

LUMINESCENCE IMAGING TO SEDIMENT CORES

(a)

147

(b) 4 cm

Fig. 4. (a) Laminated sedimentary sequence recovered from the central Santa Barbara Basin during ODP Leg 146 at Site 893, imaged in PL with 633 nm excitation light. (b) Same specimen imaged under 488 nm excitation. Notice the difference in autofluorescence character when the excitation wavelength is changed. autofluorescence does not seem to dominate in any particular area. The whiter bands running vertically through the specimen are rich in biogenic calcium carbonate, primarily skeletons of foraminifers. Multi-wavelength probing, such as this, can be used to characterize and spatially resolve the lithological composition and chemical signature of geological specimens. In addition

to having distinct emission spectra, varying chemical species will also have distinct absorption spectra, hence their wavelength-dependent response. Imaging with two excitation wavelengths can be achieved by co-linearizing two different laser beams, as shown in Figure 5. Similar results can be achieved by using a laser with various

Fig. 5. Spectrally resolved version of the Confocal Scanning Laser macroscope. The specimen can be excited with different wavelengths including broad-band excitation such as white light. Reflected light or photoluminescence from the specimen can be spectrally resolved by rotating the diffraction grating allowing the user to view an image at a particular emission wavelength or to scan through emission spectra at various points on the specimen.

148

A. C. RIBES ET AL.

Fig. 6. Three specimens (left, silty clay; centre, silty clay with sand; right, fine-grained clay) recovered from the Scotian Shelf were imaged simultaneously using the macroscope. This allows for direct comparisons of relative intensities between samples. emissions lines where the user chooses an excitation wavelength by inserting an appropriate laser-line filter. Figure 5 also shows how the emissions from a specimen, be it PL or RL, can be resolved into its spectral components. Light reflected or emitted from a specimen is steered towards a diffraction grating whose angle of reflection varies continuously with wavelength. The user can obtain an image of the entire specimen at a particular emission wavelength by keeping the grating stationary or can look at the range of wavelengths emitted by the specimen at a particular point. The scanning system can also be automated to produce spectral maps of the entire specimen by scanning the grating over many or all points on the specimen. A spectrally resolved macroscope can achieve the same imaging results, and more, as a reflectedlight spectrally resolved CCD camera system. Three Scotia Shelf specimens (Fig. 6) collected by the Geological Survey of Canada (K. Moran, pers. comm.) were simultaneously imaged using the macroscope in PL (633 nm excitation). The specimens, from left to right, represent a silty clay, a silty clay with sand and a terrigenous clay, respectively. This type of simultaneous imaging lends itself well to direct comparisons between specimens. Laser intensity, detector sensitivity and gain drift are all minimized by taking a single image of the entire group of samples. Instrument-related non-uniformity is commonly present in tiled CL images due to drifting imaging parameters. This kind of parallel imaging works also very well for time-resolved and spectrally resolved measurements, in which comparison with a time or spectral standard is essential for quantitative measurements. In this case, the fluorescence intensity is most probably related to the texture of the sediment and the relative percentage of microfossils in the sample; i.e. the more microfossils, the stronger the luminescence. Single excitation wavelength

fluorescence response can be used as a discriminator for different sediment types or to quantify the relative percentage and distribution of (calcareous) microfossils in a sample. The relative fluorescence intensity response was determined for 18 specimens, ranging from nannofossil ooze to volcaniclastic sand (Table 2). The measurements were taken under identical conditions (except for amplifier gain) using a 633 nm excitation source. The relative intensity scale is in arbitrary units ranging from 0 to 100. Samples at the top of the scale tend to be biogenic oozes rich in calcareous nannofossils. Microfossil-poor sediments, such as volcaniclastic sand and basalt, gave weak fluorescence response. Sample 144880A- 1H-3, 70-80 cm (volcaniclastic lithoclast sand) exhibited no fluorescence response within layers of pure volcanic ash (glass). As the ash layer is mixed with microfossil skeletons (calcareous nannofossils) the fluorescence reappears. Fluorescence response was evident in samples at many stages of lithification, ranging from Holocene unconsolidated sediments (Santa Barbara Basin) to Eocene laminated sediments from the east Greenland Continental Margin. Higher fluorescence intensity in samples with higher water content seems to be a secondary factor compared to lithological and textural variations. The macroscope-microscope has proven to be a very effective instrument in rapidly surveying large samples and providing a fluorescence roadmap to guide further investigations. The high zoom capability of this instrument allows for detailed analysis of interesting intervals at 'nested' scales of resolution. Microscope mode can be used for ultra-high-resolution confocal imaging of sub-millimetre features. Three-dimensional (3D) reconstructions of various microfossils are also possible using microscope mode.

Conclusions and s u m m a r y

The confocal scanning laser macroscope-microscope (cslM/m) has been successfully applied to the imaging of geological specimens, namely laminated sediment sequences in lake cores and a wide range of lithologies commonly recovered in ocean cores. The cslM/m provides rapid, high-resolution images with field of view ranging from 25 x 25 ~tm up to 7.5 x 7.5 cm. This imaging technique is non-destructive and requires no preparation of the specimens. Both PL and RL images can be acquired and entire sections of ocean/lake cores can be imaged in minutes via tiling of individual images to form composites. In macroscope mode, lateral resolutions as high as 10 pm can be obtained, which is well

LUMINESCENCE IMAGING TO SEDIMENT CORES

149

Table 2. Relative fluorescence intensity Jbr 18 marine sediment and rock samples recovered by ODP. Imaging conditions were identical for all specimens (633 nm excitation) except for amplifier gain ODP sample ID (leg site/hole core/type section, interval)

Depth (metres below sea floor)

Location

Lithological description

Relative fluorescence intensity I00

130-805B-2H-7, 53-63 cm

16.23-16.33

Ontong Java Plateau, western Pacific

Nannofossil ooze

146-893A-1H-3, 8-18cm

3.10-3.20

Santa Barbara Basin, offshore California

Laminated diatom nannofossil ooze with clayey silt

76

Eastern Equatorial Pacific

Nannofossil diatom

52

138-850B-10H-7, 20-30 cm

97.70-97.80

ooze

122-763A-4H-6, 80-90 cm

32.20-32.30

Exmouth Plateau, NW Australia

Foraminifer ooze

37

144-876A-4R- 1, 102-109 cm

34.32-34.39

Wodejebato Guyot, western Pacific

Rudstone (Cretaceous)

33

144-876A-7R- l, 77-85 cm

62.97-63.05

Wodejebato Guyot, western Pacific

Rudstone (Cretaceous)

30

146-893A-2H- 1, 55-60 cm

7.05-7.10

Santa Barbara Basin, offshore California

Diatom nannofossil silty clay (grey bed)

30

128-799A-7H-4, 75-85 cm

54.35-54.45

Sea of Japan, Yamato Rise

Siliceous claystone

29

128-798C-8H-5, 90- 100 cm

68.56-68.66

Sea of Japan, Yamato Basin

Clayey diatom ooze

27

130-804C-24X-1, 45-55 cm

216.95-217.05

Ontong Java Plateau, western Pacific

Nannofossil ooze with radiolarians

25

151-913B-26R-6, 93-100cm

489.53-489,60

Northeast Greenland Continental Margin

Clayey siliceous ooze (Eocene)

14

Peru Continental Margin

Diatomaceous foraminifer mud

12

112-679B-1H-5, 10-20cm

6.10-6.20

138-845-13H-1, 70-80cm

117.80-117.90

Eastern Equatorial Pacific

Clayey radiolarian ooze

9

122-759B-39R-4, 80-90 cm

303.80-303.90

Wombat Plateau, NW Australia

Silty claystone (Triassic)

6

144-880A-1H-3, 70-80 cm

3.70- 3.80

Takuyo- Daisan Seamount

Volcaniclastic lithoclast sand

4

160-967B-3H-7, 17-44 cm

23.97-24.22

Eastern Med., south of Cyprus

Sapropel with pyric clay layers

3

Takuyo-Daisan Seamount

Volcaniclastic sand

2

Juan de Fuca Ridge

Fine-grained basalt

1

144-880A- IH-2, 120-130cm

2.70-2.80

139-858G-10R-l, 55-65 cm (Pc. 12)

365.47-365.55

beyond what is needed to image submillimetre thick sediment layers. The macroscope is both rugged and compact e n o u g h to be used on a ship for on-site imaging following core recovery. While the microscope m o d e o f the combined instrument is not as well suited for shipboard use, it serves as an ultra-high-resolution (submicron) c o m p a n i o n to the macroscope for

studying, for example, the individual microfossils that comprise the sediments. Confocal imaging allows for crisp, highly detailed photoluminescence images o f specimens and allows for 3D rendering and surface t o p o g r a p h y analysis. A n imaging survey of several O D P samples has shown that fluorescence response can be used to discriminate between different sediment

150

A. C. RIBES ET AL.

types, and may be used to describe the distribution and relative percentage of microfossils in sediment samples. All specimens surveyed have shown fluorescence response regardless of age, type or water content. Calcareous sediments show the highest response, while increasing quantities of clay minerals minimize the intensity. Although the interpretation of these images remains qualitative at this point, future experiments employing time and spectral resolving capabilities will lead to quantitative identification and characterization of luminescing materials. Recently the macroscope-microscope has been commercialized by Biomedical Photometrics (BPI), a developer, manufacturer and marketer of widefield, confocal scanning laser imaging systems for the pathology, medical research, genomics and general microscopy markets. The company's focus is on tissue and tissue array imaging, genetic microarray imaging, and confocal microscopy. For more information go to www.confocal.com. We would like to thank the National Science and Engineering Research Council (NSERC) and Photonics Research Ontario (PRO) for the various grants that have supported the development of the macroscopemicroscope systems. We also thank the curator and staff of the Ocean Drilling Program for providing the samples used in this study. This paper is a contribution of the Canadian Climate System History and Dynamics Program (CSHD) supported by funding from NSERC. This work was conducted at the Confocal Microscopy Laboratory at the University of Waterloo, Waterloo, Ontario, Canada.

References BARKER,C.E. & KOPP, O.C. (eds). 1991. Luminescence Microscopy and Spectroscopy." Qualitative and Quantitative Applications. SEPM Short Course 25. SEPM, Dallas, TX. COYNE,L.M., MCKEEVER,S.W.S. & BLAKE,D.F. 1989. Spectroscopic Characterization of Minerals and their Surfaces. American Chemical Society Symposium Series, 415. American Chemical Society, Washington, DC. DIXON,A.E., DAMASKINOS,S. & RIBES,A. 1996. Confocal scanning beam laser microscope/macroscope. Applications in Fluorescence." Society of PhotoOptical Instrumentation Engineers, 2705, 44-52. DIXON, A.E., DAMASKINOS,S., RIBES, A. & BEESLEY, K.M. 1995. A new confocal scanning beam laser macroscope using a telecentric f-theta laser scan lens. Journal of Microscopy, 178, 261-266.

FLANNAGAN, D.C., HUANG, C., NORTON, L.D. & PARKER, S.C. 1995. Laser scanner for erosion plot measurements. Transactions of the American SocieO' of Agricultural Engineers, 38, 703-710. GODSEY,H.S., MOORE,T.C., REA, D.K. & SHANE,C.K. 1999. Post-Younger Dryas seasonality in the North American midcontinent region as recorded in Lake Huron varved sediments. Canadian Journal of Earth Sciences, 36, 533-547. HOLT, D.B. & JoY, D.C. (eds). 1989. SEM Microcharacterization of Semiconductors. Academic Press, New York. HUANG, C. & BRADFORD,J. 1992. Applications of a laser scanner to quantify soil microtopography. Soil Science Society of America Journal, 56, 14-21. LEWIS, C.F.M., MOORE, T.C., REA, D.K., DETTMAN, D.L., SMITH, A.M. & MAYER,L.A. 1994. Lakes of the Huron Basin: their record of runoff from the Laurentide Ice Sheet. Quaternary Science Reviews, 13, 891~922. MONTEMAGNO,C.D. & GRAY, W.G. 1995. Photoluminescent volumetric imaging: a technique for the exploration of multiphase flow and transport in porous media. Geophysical Research Letters, 22, 425-428. O'CONNOR, B. 1996. Confocal scanning laser microscopy: a new technique for investigating and illustrating fossil radiolaria. Micropaleontology, 42, 395-402. PAWLEY, J.B. (ed.). 1995. Handbook of Biological Col~focal Microscopy, 2nd edn. Plenum, New York. PETFORD, N. & MILLER, J.A. 1992. Three-dimensional imaging of fission tracks using confocal scanning laser microscopy. American Mineralogist, 77, 529 533. REA, D.K., MOORE,T.C., LEWIS,C.F.M., MAYER,L.A., DETTMAN, D.L., SMITH,A.J. & DOBSON,D.M. 1994. Stratigraphy and paleolimnologic record of Lower Holocene sediments in northern Lake Huron and Georgian Bay. Canadian Journal of Earth Sciences, 31, 1586-1605. RIBES, A.C., DAMASKINOS, S. & DIXON, A.E. 1995. Photoluminescence imaging of porous silicon using a confocal scanning laser macroscope/microscope. Applied Physics Letters, 66, 2321 2323. RIBES, A.C., DAMASKINOS,S., TIEDJE,H.F., DIXON,A.E. & BRODrE, D.E. 1996. reflected-light, photoluminescence, and OBIC imaging of solar cells using a confocal scanning laser macroscope/microscope. Solar Energy Materials and Solar Cells, 44, 439-450. SENESI, N., MIANO, T.M., PROVENZANO, M.R. & BRUNETTkG. 1991. Characterization, differentiation, and classification of humic substances by fluorescence spectroscopy. Soil Science, 152, 259-271. VAN GIJZEL, P. 1967. Palynology and fluorescence microscopy. Reviews of Palaeobotany and Palynology, 2, 49-79. YACOBI, B.G. & HOLT, D.B. 1990. Cathodoluminescence Microscopy of Inorganic Solids. Plenum Press, New York.

Pressure coring, logging and subsampling with the HYACINTH system P. J. S C H U L T H E I S S 1, T. J. G. F R A N C I S 1, M. H O L L A N D 2, J. A. R O B E R T S 1, H . A M A N N 3, T H J U N J O T O 3, R. J. P A R K E S 4, D. M A R T I N 4, M . R O T H F U S S 5, F. T Y U N D E R

6 & P. D. J A C K S O N 7

1Geotek Ltd, 3 Faraday Close, Drayton Fields, Daventry N N l l 8RD, UK 2Department of Geological Sciences, Arizona State University, Tempe, Arizona, USA 3Technische Universitiit Berlin, Maritime Technik, Berlin, Germany 4School of Earth, Ocean and Planetary Sciences, University of Cardiff, UK 5Technische Universitiit Clausthal, Institut filr ErdOl- und Erdgastechnik, Germany 6Fugro Engineers B. V., Leidschendam, The Netherlands 7British Geological, Survey, Keyworth, Nottingham, UK Abstract: The HYACINTH suite of equipment has been developed to investigate the pressure sensitive behaviour of sedimentary formations up to 250 bar (25 MPa). It does this by collecting pressure-preserved samples from boreholes that can be retrieved, subsampled and analysed in controlled conditions in the laboratory. This paper reviews the development of the system, how it originated from the need to better understand the nature and distribution of gas hydrates beneath the sea bed, and its achievements to date. While gas hydrates continue to be the major scientific and commercial impetus for using, and further developing, this pressure-sampling technology, other important scientific driving forces, including the growing interest in the deep biosphere beneath the sea floor, are playing an important role. We review the downhole tools, the transfer system and the suite of different pressure chambers that are required to make a complete working system. Non-destructive logging of cores contained in pressure chambers, using existing gamma- and X-ray techniques, is discussed, as are future logging techniques that will have sensors embedded within the pressure chambers. Subsamples can now be taken at full pressure and transferred into specialized chambers where intrusive measurements and experiments can be performed (e.g. inoculation chambers for microbiology). The versatile philosophy behind the integrated systems will enable future developments to be made by third parties who want to obtain subsamples at in situ pressure from the HYACINTH system. We conclude by reviewing some of the highlights of the HYACINTH operations on ODP Leg 204 where the downhole tools retrieved cores containing gas hydrates (up to 40% by volume) that were subsequently logged on board in the laboratory. These data have already contributed to the scientific understanding of the nature and distribution of gas hydrates beneath the seabed in one area on the Oregon Margin off the USA.

All core samples recovered from beneath the land or sea floor are disturbed to some extent. One important form of disturbance occurs when in situ stresses (lithostatic or hydrostatic pressure) are released during core retrieval. This form of disturbance is of little importance for many studies, but for others it is crucial. In geotechnical engineering, the mechanical properties of soils can be significantly altered as a result of stress relief during sampling. Sediment porewater chemistry can change if pressure is released because the solubility of minerals and gases is dependent on pressure. Some sediment microorganisms from the deep sea require high pressures to grow (obligate barophiles) and may not survive depressurization. Some 'pressuresensitive science' can be performed with in situ sensors or samplers: geotechnical engineers use

many downhole test methods, pore-water chemists extract and retrieve pore water only using in situ samplers, and deep-sea microbiologists rely on in situ growth experiments. However, there is a balance between the cost of designing in situ experiments and recovering samples at in situ pressures. Many types of measurement and experiment cannot be adapted to in situ methods, and pressure core sampling is the only solution. The need to conduct geotechnical tests on shallow samples of gassy sediments led Johns (1992) to take shallow-water pressure cores and perform the subsampling and testing in a hyperbaric chamber. This very practical approach can, however, only be achieved at low pressures and hence has only limited applications. All the pressure-coring tools discussed in this paper only preserve the hydrostatic stresses within a sample;

From: ROTHWELL,R.G. 2006. New Techniquesin Sediment Core Analysis. Geological Society, London, Special Publications, 267, 151-163. 0305-8719/06/$15.00 9 The Geological Society of London 2006.

152

P. J. SCHULTHEISS E T AL.

they do not attempt to preserve the lithostatic stresses that certainly will effect some geophysical and geotechnical measurements. Nowhere is pressure-related core disturbance more evident than when sampling sediments containing methane hydrate. Methane hydrate is an important reservoir of methane in deep-sea sediments, and its role in global climate change, as a geohazard, and as a possible energy resource is still poorly understood (Kvenvolden 1993; Collett 2000). The study of methane hydrate has been hampered by the lack of suitable pressure tools, both for recovery of hydrate-bearing core and for preservation and study of hydrate-containing samples in the laboratory. Sediment cores containing hydrate where pressure is released during retrieval suffer dramatic disturbance (Fig. 1). Not only is methane hydrate unstable at atmospheric pressure, but it only forms when pore waters are saturated with methane. Free methane gas may also occur within the hydrate matrix (disequilibrium). On pressure release, any free methane immediately expands; the dissolved methane exsolves to form bubbles that also rapidly expand; and the hydrate dissociates, slowly releasing water and gas, causing the cores to become soupy. Not only is the distribution and nature of the hydrate lost but the fabric of the core is also ruined, and the only indication that gas hydrates existed in the core is from pore-water anomalies which show freshening. The increasing interest in marine gas hydrates and their highly ephemeral nature has driven the development of pressure-coring tools in recent years. Sea-floor hydrates can be recovered at in situ pressures using systems such as the OMEGA Multi Autoclave Corer (MAC) and the Dynamic Autoclave Piston Corer (Hohnberg et al. 2003). These systems, both developed by the Maritime Technology group of the Technical University of Berlin, can be deployed from a research vessel and retrieve cores that retract into pressure-retaining autoclaves. The MAC system has been successful in retrieving short (50 cm-long) gas hydrate cores from the seabed. However, sampling hydrates over the complete gas hydrate stability zone (GHSZ) that may extend for several hundred metres below the sea floor (depending on water depth and in situ temperature) requires drilling technology and associated sampling devices. To this end a number of downhole, wireline-operated, pressure coring tools have been developed. The first successful wireline pressure-coring tool, the Pressure Core Sampler (PCS), was developed by the Ocean Drilling Program (ODP) (Pettigrew 1992; Graber et al. 2002). It was used extensively in hydrate-bearing formations in

Fig. 1. ~Moussy' core from ODP Leg 204 containing chunks of methane hydrate: an illustration showing what happens when sediment containing gas hydrate is recovered using conventional (unpressurized) coring systems. The chunks of solid hydrate visible in the core are shown in the diagram above the photograph. 1995 and 2002, at Blake Ridge on ODP Leg 164 (Paull et al. 1996) and at Hydrate Ridge on ODP Leg 204 (Tr6hu et al. 2003). The PCS takes a 43 mm-diameter, 1 m-long unlined core in formations that range from soft sediments to firm clay, although it can apparently core massive hydrate as well (Tr6hu et al. 2003). The core is cut by a pilot bit extending ahead of the main drill bit, driven by rotation of the whole drill string. After the core has been cut, the core tube (containing the core) is retracted into the sample chamber and the pressure (up to 690bar or 69MPa) is maintained by closing a ball valve at the lower end of the chamber. On retrieval the chamber can be removed from the tool for scientific investigations (e.g. controlled degassing of hydratebearing core: Dickens et al. 2000; Milkov et al. 2003). Following ODP Leg 164, it was apparent that the ODP PCS had a number of intrinsically limiting factors. The PCS is excellent for capturing a volume of sediment plus all associated gases, as proved by the 30 successful runs (out of 39 deployments) on ODP Leg 204, which also proved the PCS to be a 'working' tool. However, the PCS core diameter is limited by using a ball valve, the core quality is limited in many circumstances because the complete drill string rotates

SUBSAMPLES FROM THE HYACINTH SYSTEM during coring, and the core cannot be removed from the system without releasing the pressure from the sample. Hence, two new wireline tool developments were initiated in response to the perceived limitations of the PCS on Leg 164: the Japanese Pressure-Temperature Coring System (PTCS), which is designed to take longer, larger cores than the PCS and to provide an element of temperature control; and the European-funded HYACE ('Hydrate Autoclave Coring Equipment') project, whose wireline pressure corers are designed to recover cores in liners that can subsequently be transferred into other pressure vessels. This paper describes the advances that have been made to the HYACE system during the more recent HYACINTH project, illustrating the unique ability to obtain pressure cores in a wide range of lithologies, transfer cores under in situ pressure into other pressure chambers, and to perform non-destructive measurements and invasive experiments at full/n situ pressure. For simplicity, the combined developments of the HYACE and HYACINTH projects are referred to in this paper as the HYACINTH system.

been hitherto impossible due to the inability to obtain undisturbed (pressure-preserved) samples. The extent of this deep microbial habitat, thought to account for some 10% of the total global biomass, has only recently been recognized (Parkes et al. 1994, 2000). Furthermore, deep marine sediments are thought to account for over 60% of all bacterial biomass on Earth (Whitman et al. 1998). Understanding the subsea-floor biosphere may help to define the limits of life on Earth and allow identification of biotechnologically useful organisms. The study of hydrate, the deep biosphere and a wide range of other scientific problems require the development of equipment for handling and making measurements on pressure cores and subsamples, which is one of the primary goals of the HYACINTH project. Another potential application of pressure coring that has recently emerged is for research into CO2 sequestration; the HYACINTH system could be used to sample rocks from the deep formations underground where CO2 has been sequestered, allowing quantitative study of the fluids that they contain. The objectives of the HYACINTH project as given in the original proposal are: 9

The HYACINTH system In late 1997, the European Commission began funding HYACE under its MAST III programme. It was primarily an engineering project aimed at developing tools for the recovery and handling of gas hydrates under pressure in offshore drilling. Two types of wireline pressure corer were developed within HYACE: the Fugro Pressure Corer (FPC) and the HYACE Rotary Corer (HRC), both of which are described below. The FPC and the HRC were designed to mate to laboratory pressure chambers, allowing any hydrate recovered to be transferred, preserved and studied at in situ pressures. By the end of the HYACE project, prototype versions of both the FPC and the HRC had undergone testing on land in Germany and briefly on the J O I D E S Resolution on Leg 194 off NE Australia. The HYACINTH ('Deployment of HYACE tools In New Tests on Hydrates') project was designed to create a fully operational pressure coring and transfer system using the HYACE coring tools. HYACINTH was funded in 2001 by the European Commission under its Framework Five programme. While the HYACINTH system will still be used extensively for gas hydrate research, the 'deep biosphere' has provided a second scientific driving force for the programme. The study of pressure-sensitive micro-organisms from high-pressure regimes in the subsurface has

153

9 9

9 9

9

Recovery of good-quality gas hydrate core under pressure with both the FPC and HRC. Refinement and tuning of both the FPC and HRC. Transfer of core into laboratory pressure chambers for storage, evaluation and subsampling. Geophysical logging of core under pressure. Development of technologies for subsampling pressurized cores for chemical, microbiological and petrophysical study. Microbiological analysis of uncontaminated sediment samples.

D o w n h o l e tools

Two types of wireline pressure-coring tool were developed in the earlier HYACE project, a percussion corer and a rotary corer, that were designed to cut core in a wide range of lithologies where gas-hydrate-bearing formations might exist (Fig. 2). Although the tools are quite different, they include a number of important common features: they recover lined cores and mate to a common transfer system, they use 'flapper valve' sealing mechanisms rather than ball valves to maximize the core diameter and downhole drive mechanisms to ensure high core quality. Both corers were designed to be compatible with the bottom hole assembly and drill string used on the drillship J O I D E S Resolution operated by the internationally funded Ocean Drilling Program.

154

P. J. SCHULTHEISS E T AL.

Fig. 2. Diagram showing the two wireline-operated pressure corers. The Fugro Pressure Corer (FPC) uses a downhole hammer mechanism to drive the core barrel into the sediment. The HYACE Rotary Corer (HRC) uses a downhole motor to rotate the core bit and cut the core. With both corers, the core is pulled into the autoclave on recovering the corer, preserving the core at in situ pressure.

The percussion corer was developed by F u g r o Engineers BV and is k n o w n as the F u g r o Pressure C o r e r or FPC. The F P C uses a water h a m m e r , driven by the circulating fluid p u m p e d d o w n the drill pipe, to drive the core barrel into the sediment up to 1 m a h e a d of the drill bit. The

core diameter is 57 m m (liner outer diameter is 63 mm). On completion of coring, the recovery of the corer with the wireline pulls the core barrel into the autoclave, in which the pressure is sealed by a specially designed flapper valve. The F P C is designed to retain a pressure of up

SUBSAMPLES FROM THE HYACINTH SYSTEM to 250 bar (25 MPa). It is suitable for use with unlithified sediments ranging from stiff clays to sandy or gravelly material. In soft sediments it acts like a push corer. The rotary corer was developed by the Technical University of Berlin and the Technical University of Clausthal, and is known as the HYACE Rotary Corer or HRC. The HRC uses an Inverse Moineau Motor, driven by the circulating fluid pumped down the drill pipe, to rotate the cutting shoe up to 1 m ahead of the roller cone bit. The cutting shoe of the HRC uses a narrow kerf, dry auger design with polycrystalline diamond (PCD) cutting elements. This design allows the core to enter into the inner barrel before any flushing fluid can contaminate the material being cored. The core diameter is 51 mm (liner outer diameter is 56 mm). On completion of coring, the recovery of the corer with the wireline pulls the core barrel into the autoclave, in a similar manner to the FPC, and the pressure is sealed by a specially designed flapper valve. The HRC is designed to retain a pressure of up to 250 bar (25 MPa) and is suitable for use in sampling lithified sediment or rock. It is essential that the pressure-coring tools can be used from different drilling platforms. Originally the tools were configured to operate on a scientific drilling vessel. Deploying the H Y A C I N T H pressure corers on a small geotechnical drillship is quite different from deploying them on a large drillship, such as the JOIDES Resolution, because of the much smaller derrick and restricted workspace. On ODP Leg 204, the 12-14m-long tools were assembled horizontally, lifted into the derrick and lowered vertically into the drill pipe. For operation on a geotechnical drilling vessel, such as the Fugro Explorer, the coring tools have been shortened and the assembly procedures modified. The HYACE Rotary Corer has been modified at the Technical University of Clausthal so that it is about 2 m shorter than the tool deployed on ODP Leg 204, can be split in two and the two halves assembled/disassembled vertically.

HYACINTH

transfer and chamber systems

The ability to manipulate cores, take subsamples and make measurements, all at in situ pressures, are major objectives of the H Y A C I N T H project. To ensure that future developments can be adapted to the system, some common principles and parts were established. To enable different chambers to be easily attached to each other in series, without the need for using either screwthreaded pipe joints or flange nuts, a 'quick fit' flange was designed. This allows quick couplings

155

to be made between pressure chambers without requiring rotation. To move cores between different pressure chambers a 'Manipulator Chamber' has been designed that can be used to either push core samples from one chamber to another, or to latch onto cores (using a latching arrangement) and pull the sample into a chamber. All pressure chambers that maintain pressure in isolation from the rest of the system carry a specially designed large-bore (65mm) ball valve. This allows cores taken with either the FPC or the HRC to move freely between chambers. To help reduce costs (and weight) the sealing mechanism at the opposite end of a chamber to the ball valve is an inverted cone seal that can be connected to the Manipulator. However, a chamber could be designed with a ball valve at each end, allowing the core to move completely through the chamber if necessary. To move a core under pressure from one chamber to another, the chambers are connected, the Manipulator is attached to the end of the appropriate chamber (pushing v. pulling), and the empty chambers are filled with water and pressurized until the pressure is equal to that in the chamber containing the core. The ball valves are opened and the core is moved from one chamber to the other with the Manipulator, which is retracted before closing all valves and depressurizing the empty chambers. Pressure alone cannot stabilize gas hydrates. Even if the pressure is maintained at in situ values, methane hydrate will begin to dissociate as the temperature increases, destroying the sample and leading to possibly dangerous pressure increases inside the chamber. To keep the sample in pristine condition, in situ temperatures should be maintained as well as in situ pressures. For safety reasons, it is also important to ensure that all pressure chambers are equipped with relief valves, burst discs or other pressure-relief mechanisms in the unhappy event of temperature excursions. To help maintain in situ temperatures, most ocean cores containing methane hydrate should be quickly cooled (in an ice bath) when they reach the surface and all subsequent handling operations should take place in cold (2-6 ~ laboratories. This includes the core transfer system, which has the additional requirement of using chilled sea water as the pressurizing fluid.

Transfer systems. When the core is first recovered it is contained in the corer autoclave, which is then disconnected from the main tool. The pressure core in the FPC or HRC autoclave is similar to a conventional piston core in that the piston remains within the top of the liner; however, this

156

P. J. SCHULTHEISS ET AL.

Fig. 3. Diagram of the HYACINTH transfer system showing the arrangement of the manipulator, Shear Transfer Chamber (STC) and ball valve. On recovery, the pressure core is transferred from the corer autoclave into the STC, the technical and geological portions of the core are sheared apart, and the geological portion then transferred into a storage chamber. The whole process is conducted at the in situ pressure of the core. pressure core piston assembly contains many more technical components than a conventional piston. The core and piston/sensor mechanism (referred to as the Catch Assembly, c. 1.6m long) consists of a 'technical portion' (c. 0.6m long) and a 'geological portion' (c. 1.0m long). Only the geological portion needs to be retained under pressure for study. The technical portion needs to be extracted from the high-pressure environment as soon as feasible in order for it to be recycled for further corer deployments. The first transfer is designed to extract the Catch Assembly from the autoclave and to separate it into its two components (technical and geological). To achieve this, the autoclave is connected in series to a Shear Transfer Chamber (STC) and a Manipulator (Fig. 3). After equalizing the pressures, the Manipulator is extended through the STC and mates to the core in the autoclave. The core is withdrawn into the STC where the technical and geological parts of the Catch Assembly are separated by shearing the liner between them. At this stage the corer autoclave is isolated, removed and replaced by another chamber (logging or storage). The pressures are again equalized and the Manipulator pushes the geological portion of the core into the logging or storage chamber. The technical portion of the core (still attached to the Manipulator) is pulled back into the STC. The ball valve on the logging or storage chamber is closed and the chamber removed for scientific measurements. The technical portion of the core is recycled for subsequent coring operations. As the FPC and the HRC cores have slightly different diameters, the STC is supplied with two cutting boxes adapted to the specific core dimensions. The cutting boxes can be quickly interchanged by means of quick-fit flange-clamp connections. Once the core has been isolated in a sealed chamber, its nature

Logging/storage chambers.

can be examined by geophysical logging. A H Y A C I N T H Logging Chamber has been developed that is sealed at one end with a ball valve and at the other with the conical seal fitting to which the Manipulator can be attached for making transfers. The cylindrical part of the Logging Chamber is manufactured from glassfibre reinforced plastic (GRP). An outer steel tube, with longitudinal windows cut into it, allows access for the geophysical sensors and holds the steel end caps in place when the chamber is pressurized. The use of GRP was originally intended to enable acoustic velocity to be measured on the core from outside the pressure vessel. Unfortunately, it has been found that with the thickness of GRP required this is not easily possible. However, accurate density logs can be achieved using a gamma attenuation densitometer (see the section 'Pressure coring for scientific applications on ODP Leg 204' later). The Storage Chamber is a simple cylindrical pressure vessel, sealed at one end with a ball valve and at the other with the conical seal fitting to which the Manipulator can be attached for making transfers (Fig. 4). The cylindrical part of the Storage Chamber is manufactured from stainless steel or high-strength aluminium alloy. The Storage Chamber is designed to preserve a core under pressure for periods of hours, days or, possibly, years. While in the Storage Chamber, density profiles can be obtained using the gamma attenuation densitometer (see V-MSCL below), although measurements will take longer through steel than through the lower-density aluminium alloy or the GRP of the Logging Chamber. The aluminium storage chambers are suited to X-ray scanning for three-dimensional (3D) computed tomography (CT), as well as gamma densitometry, and these chambers can be manufactured at different thicknesses (different maximum working pressures) to maximize the X-ray transmissivity. Detailed structural

SUBSAMPLES FROM THE HYACINTH SYSTEM

157

has been replaced with a gas in order to reduce contamination. All parts of the tools exposed to the core are sterilized so that the axial subcore remains pristine. The axial subcore is extruded into a purpose-built Microbiological Sampling Chamber, where the subcore is visually inspected and slivers of the subcore are transferred into separate chambers filled with growth medium (Fig. 5). These growth experiments will be the first to be carried out on micro-organisms from deep sediment layers that have not experienced pressure disturbance. Other systems, still under development, will allow samples to be taken from this enriched medium in order to isolate and incubate specific species for physiological and characterization studies.

Fig. 4. HYACINTH Storage Chamber. The stainless steel or aluminium pressure chamber has a large ball valve at one end that can be opened/closed with the handle shown. information is necessary for many types of scientific study, as well as to provide information to guide subsampling. Subsamplingfrom H Y A C I N T H cores. Subsampling under pressure, and the subsequent transfer of subsamples into specialized chambers for further measurements and experiments, is the culmination of the H Y A C I N T H project. The H Y A C I N T H subsampling system is applicable to a wide range of possible subsampling requirements and scientific applications, including gas analysis, pore-water analysis, biological experiments, and petrophysical and geotechnical measurements. Within the H Y A C I N T H project itself, the emphasis is on providing sediment subsamples for microbiological experiments, but many of the tools and procedures are generally applicable. One subsampling tool that has been developed can be used to cut out a short length of whole core (10cm) anywhere along the 1 m FPC or HRC core and transfer it into a small Core Section Chamber. (The exact location of this subsection might be chosen after examining core log data taken under pressure.) Another subsampling tool can extrude an axial subcore from the whole-core subsample into a user-defined chamber. When these tools are used for microbiological experiments, the initial subsectioning takes place after the normal pressurizing medium (sea water)

Instrumented chambers. Very few of the properties of gas hydrates can be measured through the steel or aluminium walls of storage chambers. Consequently, other types of chamber are under development where the measurement system is located inside a pressurized instrumented chamber. Within the HYACINTH project a Resistivity Imaging Chamber is under development, containing an array of electrodes that will provide an electrical resistivity image of the core. The presence of gas hydrate in a core has a significant effect on its electrical resistivity and is used to infer the presence of gas hydrates from downhole measurements. Electrical resistivity cannot be measured on cores within the existing H Y A C I N T H chambers because of the high electrical conductivity of the metal used in their construction. The Resistivity Imaging Chamber does not require any ball valves, as it will connect in line between two standard pressure chambers. Measurements will be acquired as the core is passed from one chamber to another under full pressure. The resistivity measurements are complicated by the need to develop and use porous core liners; a suitable liner has been developed and is in the testing phase. A pressurized core-logging system using other geophysical and geotechnical sensors is under development. It will consist of a Measurement Chamber (MC) containing the sensors, an Extension Chamber and many standard H Y A C I N T H components (Fig. 6). The pressure core will be brought to this system in a H Y A C I N T H Storage Chamber, coupled to the other chambers, the pressures equalized and the core stepped past the measurement sensors in the MC by means of push rods in the manipulators at each end. Standard H Y A C I N T H manipulators are manually operated but these will be motor driven, allowing the push rods to be computer controlled so that the core can be moved with accuracy in

158

P.J. SCHULTHEISS E T AL.

Fig. 5. Diagram showing how small subsamples from a pressurized core are cut and transferred into microbiological experimental chambers: (A) the Core Section Chamber (CSC), carrying a whole-round section from the whole core under pressure, is connected to the other chambers; (B) when all chambers have been pressurized, the large ball valves of the CSC are opened and the whole-round core section is pushed over the extrusion tube. Slivers from the axial subcore are taken with the rotary cutter, and pushed into a microbiological chamber; and (C) an aseptically collected subsample is then available for microbiological experiments. either direction to predetermined positions. One set of sensors that will be used will be for Pwaves. This will provide accurate velocity profiles, which are particularly important for the interpretation of seismic reflection profiles in hydrate-bearing lithologies, where significant velocity excursions might be expected. The

velocity data can also be used in concert with the density data to provide an acoustic impedance profile for the generation of synthetic seismograms. The microbiological subsampling chamber is the first instrumented subcore chamber to be developed. The aseptic, repetitive sectioning of

Fig. 6. Diagram of the pressurized core-logging system illustrating how whole cores can be logged with various sensors by passing them through an instrumented measurement chamber. The system under development will use computer-controlled, linked push rods to move the core past the sensors under full pressure.

SUBSAMPLES FROM THE HYACINTH SYSTEM the subcore under visual inspection is a complex operation and serves as a good example as to what could be possible. A host of specialized instrumented subcore chambers might be used to measure everything from chemical to geotechnical properties and is limited only by a lack of imagination.

Pressure coring for scientific applications on ODP Leg 204 It was realized early on that the success of the H Y A C I N T H project would depend on close collaboration with ODP. A formal agreement between the H Y A C I N T H partners and ODP was signed in November 2001. Participation in Leg 204 was the fruition of this collaboration. ODP Leg 204 was a multidisciplinary investigation into the dynamics of gas hydrates in an accretionary complex known to contain massive hydrate off the coast of Oregon, USA. A number of holes were drilled during the summer of 2002 at sites on and around the southern summit of Hydrate Ridge, a topographic high in the accretionary complex of the Cascadia subduction zone, located approximately 80km west of Newport, OR (Tr6hu et al. 2003). Previous studies of southern Hydrate Ridge had documented the presence of sea-floor gas vents, outcrops of massive gas hydrate and a 50 m-tall 'pinnacle' of authigenic carbonate near the summit (Torres et al. 1999; Johnson et al. 2003). A high priority within Leg 204 was the retrieval of cores at in situ pressure to quantify the amount of methane present in the sediment, to determine the decimeter to centimetre-scale structure of the methane hydrate within the sediment and to collect geophysical data to aid in the interpretation of seismic profiles. Together, the total methane concentrations and physical nature of the hydrate are critical for understanding the dynamics of hydrate formation and the physical properties of methane-saturated sediments. Accurate knowledge of the in situ concentration of methane in the sediment formations enables the equilibrium phase (aqueous, free gas, solid hydrate or a combination thereof) of the methane to be calculated. The ODP PCS has proven effective for estimating in situ gas concentrations and was used extensively for this purpose during Leg 204. However, the ODP PCS is not useful for collecting structural or physical data on hydrate-bearing sediment. The H Y A C I N T H systems were employed on ODP Leg 204 to measure the density structure of the cores at in situ pressures. Total methane quantification by depressurization was also performed on H Y A C I N T H pressure cores, and during the

159

slow depressurization and dissociation of the hydrate structural information was collected, allowing confirmation of subcentimetre-scale hydrate-bearing structures. ODP Leg 204 was a test cruise for the HYAC I N T H tools, but it also provided the opportunity for the system to yield important scientific data. Cores at in situ pressures were recovered with both the HRC and the FPC, transferred into laboratory chambers without loss of pressure and geophysically logged. The total depth from which the cores were acquired ranged from 785 to 1553m below sea level. This was the first time that laboratory measurements have been made of the physical properties of natural hydrates at subsea-floor pressures, with the in situ pressure being maintained throughout. It was found that the gamma density logs were the most useful and that they could be acquired from pressure cores contained in either the GRP-walled Logging Chamber or the steelwalled Storage Chamber. Vertical Multi-Sensor Core Logger

The equipment used to measure gamma density profiles on the cores was a purpose-built Vertical Multi-Sensor Core Logger (MSCL-V), using similar sensor design to that of the standard Geotek Multi-Sensor Core Logger. Since depressurization of H Y A C I N T H cores (and gas collection) while repeatedly gamma profiling was to be an important part of the ODP Leg 204 science agenda, a vertical orientation was chosen to allow gases to be bled from the top of the core. This procedure was designed to be similar to the depressurization experiments that had been performed on PCS cores on ODP Leg 164 (Dickens et al. 2000). Compared with a standard MSCL, which measures a number of parameters on the whole core, the only measurements that can be easily made through a pressure tube are P-wave velocity and gamma density. In practice, the Pwave velocity proved unreliable even in the GRP pressure tube. The next P-wave velocity measurements wilt take place in an instrumented chamber (as d e s c r i b e d above), along with measurements of electrical resistivity. However, on ODP Leg 204, the high-resolution gamma density profiles did contribute important new data to the science effort (see below). Gamma density is calculated by measuring the attenuation of a collimated beam of gamma rays. With the Geotek system, using a 5mmdiameter collimator, the along-core spatial resolution is about 1 cm. The density calibration was performed using aluminium and water standards in a similar procedure to that used

160

P. J. SCHULTHEISS ET AL.

routinely with the standard Geotek MSCL (Schultheiss et al. 2004). Despite the fact that there are two core liners to consider, the measurements of density were estimated to have an accuracy of + 0 . 0 3 g c m -3. To acquire the total gamma counts necessary to achieve this accuracy with measurements on cores in the steel Storage Chamber, the counting time for each measurement was 30 s.

Dissociation experiments Depressurization/dissociation experiments were carried out on H Y A C I N T H cores in a similar manner to those experiments performed on the O D P PCS. As the pressure on the core is reduced, methane and other gases present in the core as free gas, dissolved gas or as gas hydrate are released at a rate that depends on the pressure, temperature and the physical state. Dissolved gas quickly exsolves, but gas hydrate is comparatively slow to dissociate when the temperaturepressure conditions are outside the gas hydrate stability zone. Gases were collected during depressurization through a pressure manifold into a 11 bubbling chamber (inverted measuring cylinder) to determine the quantity of gas evolved. The composition of this gas was measured by gas chromatography onboard ship, as was routinely done during depressurization experiments with the O D P PCS. After each incremental pressure release and gas collection (10-15 rain), another gamma density profile was collected (30 rain) and the process repeated. One of the difficulties encountered on Leg 204 was the lack of a suitable cold laboratory space for the MSCL-V (Fig. 7). Ideally, the MSCL-V would be located in a cold room (0-6 ~ but on the JOIDES Resolution this equipment was located in the scientific hold (22-25 ~ Consequently, when the hydrate-bearing cores were logged, the pressure chambers warmed up slowly (despite all efforts to insulate them) and had to be periodically (every 3-4 h) returned to the refrigerator to cool down. This was done to ensure that the pressure-temperature state of the core remained within the gas hydrate stability zone.

Summary o f results The H Y A C I N T H tools proved effective at hydrate recovery, both in massive hydrate formations and sediments with subcentimetre-scale ephemeral hydrate. O D P Core 204-1249F-2E ( H R C 4) was collected using the H R C and was recovered at full pressure (80 bar; 8 MPa) from only 8m below the sea floor at the southern

Fig. 7. The Vertical Multi-Sensor Core Logger being used to log pressure cores on board the JOIDES Resolution during ODP Leg 204. summit of Hydrate Ridge (777 m water depth). This core, which contained decimetre-scale chunks of massive hydrate, was logged repeatedly in the H Y A C I N T H logging chamber over the next 3 days in stages while it was depressurized and the gas subsampled for compositional analysis. A total of 101.51 of gas were collected and 24 gamma density logs were measured along the length of the core. The mass balance calculation showed that the core consisted of approximately 38% methane hydrate (by volume) and 62% fine-grained sediment (by volume), with an average bulk density of 1.3 gcm -3 (67% porosity). ODP Core 204-1244E-8Y (FPC 9) was collected using the FPC and was recovered under full pressure from 50m below the sea floor, within the gas hydrate zone to the NE of the southern summit. This core was also logged repeatedly in the H Y A C I N T H logging chamber during the depressurization process over a 12-h period, during which a total of 3.81 of gas were collected. Three gamma density profiles are shown in Figure 8. Density profile A is at in situ conditions and shows a distinct zone of low density, interpreted as a thin hydrate layer. Profile B, which was recorded during the depressurization/degassing process, shows gas layers developing from the hydrate and the general reduction in density

SUBSAMPLES FROM THE HYACINTH SYSTEM

161

20

30

40 A

E u e-

5o

"o 60 c o') o') 70 o _1

80

90

100 1.5

1.6

1.7

1.8

1.9

1.5

1.6

Gamma

1.7

1.8

1.9

1.5

1.6

1.7

1.8

1.9

density (g/cc)

Fig. 8. Density profiles of ODP Core 204-1244E-BY (FPC9) obtained during the depressurization process. The profiles shown are: (A) before depressurization (at in situ pressure); (B) at an intermediate stage in the depressurization/degassing process; and (C) at atmospheric pressure.

caused by gas exsolving from solution. Profile C shows the last density profile, which was collected after all the gas has been removed from the sediment. Calculations show that only 0.2% of the core (by volume) was gas hydrate and that the density anomaly interpreted to be a hydrate in density profile 1 can be accounted for by 5 ram-thick layer of hydrate (density 0.92 g cm -3) in the 57 mm-diameter core. The lower gas layer, which formed at the base of the core, indicated the presence of another discrete hydrate zone within the core catcher (just below the lowest point of density measurement). After all the gas had been released, the core was X-rayed, and the X-ray images showed two steeply dipping features at the same location as the density anomalies. Three interstitial water samples were taken from the core to test for pore-water freshening due to hydrate dissociation; the two core locations containing the dipping features had fresher pore waters than the middle of the core, which showed no freshening. (See chapter 3, Site 1244, pp. 15 and 111 in Tr~hu et al. 2003.) Thin layers of hydrate like those in Core 204-1244E-8Y are unlikely to be preserved using conventional coring techniques, but they can be observed and studied using the H Y A C I N T H pressure coring tools.

While there was no subsampling under pressure on O D P Leg 204, samples of massive hydrate were transferred out of the H Y A C I N T H pressure chambers in a relatively intact state by freezing the core prior to releasing the pressure. O D P Core 204-1249G-2E was recovered at full pressure from a hole adjacent to Core 2041249F-2E and at a depth of 13 m below the sea floor. It was initially logged in a H Y A C I N T H Storage Chamber, which revealed massive hydrate structures. To provide scientists with the best hydrate sample possible at one atmosphere, this H Y A C I N T H core was depressurized rapidly after freezing. To preserve the sample, the sea-water pressurizing fluid was first replaced by helium and the core was frozen for more than 24h at - 1 0 ~ When the Storage Chamber was depressurized and the ball valve opened, the core was found to be composed of large intact pieces of massive hydrate that were immediately preserved in liquid nitrogen for scientific study. These are the most pristine hydrate samples to be collected from more than 13 m below the sea floor. One particular H Y A C I N T H pressure core was able to contribute a significant insight into a scientific controversy. Free methane gas cannot

162

P. J. SCHULTHEISS E T AL. was again logged, showing that this reference core could be successfully transported for subsequent studies.

20

30

Conclusions 40

50 A

E ca 60 t--

"0

70

c

80 O

_J 90

100

110

120 0.6

0.8

1.0

1.2

1.4

1.6

1.8

Gamma density (g/r Fig. 9. Density profile of ODP Core 204-1249H-2Y (FPC 10) obtained under in situ pressure showing free gas inside a region of massive hydrate. exist in equilibrium with the dissolved methane gas in sea water and methane hydrate in the hydrate stability zone, yet in hydrate samples collected at the sediment-water interface with a grab sampler (Suess et al. 2001), a free gas bubble structure within the massive hydrate structure had been observed. It was not known if the free gas within this structure could exist in situ or whether it was an artifact caused by the pressure release during sampling. Core 204-1249H-2Y was recovered from 13.5 m below the sea floor near the southern summit of Hydrate Ridge. It was recovered at full pressure, transferred and logged in a stainless steel H Y A C I N T H Storage Chamber. The density profile (Fig. 9) shows massive hydrate layers and one interval where the average cross-core density is only about 0.75gcm -3. This measurement indicates quite conclusively that free gas can exist in situ within the sediment/hydrate structure. After being transported back to ODP in Texas it

The H Y A C I N T H project is reaching a successful conclusion, providing a powerful new set of tools for the study of sediments and rocks preserved at in situ pressures. It has been demonstrated that sub-bottom samples of gas-hydrate-bearing sediments can be recovered and then transferred, without loss of pressure, into a variety of chambers for non-destructive testing at in situ conditions and during depressurization and hydrate dissociation. Measurements on the samples recovered to date have already demonstrated the value of obtaining pressure-preserved core material and have contributed to the scientific assessment of the nature and distribution of gas hydrates at one location. Subsampling equipment has been developed that will allow more elaborate biological, chemical, geophysical and geotechnical measurements to be made in the future. Other institutions and scientists will be needed to help develop more sophisticated equipment to extend the utility of the H Y A C I N T H coring, transfer and subsampling equipment. This will happen as new questions about deep environments emerge and provide the impetus to investigate sediments and rocks that have pressure-sensitive characteristics. For example, research into CO2 sequestration could benefit from the availability of pressure cores. The H Y A C I N T H system itself can also be further developed, assuming that there is the scientific or commercial demand to extend the working range of the tools beyond the current 250 bar (25MPa). To achieve these higher pressures, additional engineering and more exotic materials would be needed, but there is much scope for further development if scientific curiosity or other demands are strong enough. The authors would like to thank the Ocean Drilling Program and the technicians, drilling crew and Shipboard Scientific Party of ODP Leg 204. The HYACINTH project is funded by the European Commission Framework Five Programme, Contract Number EVK3-CT2001-00060.

References COLLETT,T.S. 2000. Natural gas hydrate as a potential energy resource. In: MAX, M.D. (ed.) Natural Gas Hydrate in Oceanic and Permafrost Environments'.

Kluwer, Dordrecht.

SUBSAMPLES FROM THE HYACINTH SYSTEM DICKENS, G.R., WALLACE,P.J., PAULL, C.K. & BOROWSKI, W.S. 2000. Detection of methane gas hydrate in the pressure core sampler (PCS): Volume-pressure-time relationships during controlled degassing experiments. In: PAULL, C.K., MATSUMOTO, R., WALLACE, P.J. & DILLON, W.P. (eds) Proceedings of the Ocean Drilling Program, Scient~c Results, 164. Ocean Drilling Program, College Station, TX, 113-126. GRABER, K.K., POLLARD,E., JONASSON,B. ~ SCHULTE, E. (eds). 2002. Overview of Ocean Drilling Program Engineering Tools and Hardware. Ocean Drilling Program Technical Note, 31 (online). Available from World Wide Web: (cited 1 June 2004). HOHNBERG, H.J. ET AL. 2003. Pressurized Coring of Near-surface Gas Hydrate Sediments on Hydrate Ridge: The Multiple Autoclave Corer, and First Results from Pressure Core X-ray CT Scans. Geophysical Research Abstracts, 5. European Geophysical Society, Nice. JOHNS M. 1992. Geotechnical properties of Mississippi River Delta sediments utilizing in situ pressure sampling techniques. In: Handbook of Geophysical Exploration at Sea, 2nd edn. Hydrocarbons. CRC Press, Boca Raton, FL, 352401. JOHNSON, J.E., GOLDFINGER,C. & SUESS,E. 2003. Geophysical constraints on the surface distribution of authigenic carbonates across the Hydrate Ridge region, Cascadia margin. Marine Geology, 202, 79-120. KVENVOLDEN, K.A. 1993. Gas hydrates - geological perspective and global change. Reviews of Geophysics, 31, 173 187. MILKOV,A.V., CLAYPOOL,G. ETAL. 2003. In situ methane concentrations at Hydrate Ridge offshore Oregon: new constraints on the global gas hydrate inventory from an active margin. Geology, 31,833-836.

163

PARKES, R.J., CRAGG, B.A. ET AL. 1994. Deep bacterial biosphere in Pacific Ocean sediments. Nature, 371, 410413. PARKES, R.J., CRAGG, B.A. & WELLSBURY,P. 2000. Recent studies on bacterial populations and processes in subseafloor sediments: a review. Hydrogeology, 8, 11-28. PAULL, C.K., MATSUMOTO,R. ET AL. 1996. Proceedings of the Ocean Drilling Program, Initial Reports, College Station, TX. Ocean Drilling Program, 164. Ocean Drilling Program, College Station, TX, 5-12. PETTIGREW,T. 1992. Design and Operation of a Wireline Pressure Core Sampler (PCS). Ocean Drilling Program Technical Note, 17. SCHULTHEISS, P.J., ROBERTS, J.A. & CHAMBERLAIN,R. 2004. GEOTEK Multi-Sensor Core Logger Manual (online). Available from World Wide Web: < http://www.geotek.co.uk/ftp/manual.pdf > (cited 1 June 2004). SUESS, E., TORRES,M.E. Er AL. 2001. Sea floor methane hydrates at Hydrate Ridge, Cascadia margin. In: PAULL, C.K. & DILLON, W.P. (eds) Natural Gas Hydrates: Occurrence, Distribution, and Detection. American Geophysical Union, Geophysical Monograph, 124, 87-98. TORRES, M.E., BOHRMANN,G. ET AL. 1999. Geochemical Observations on Hydrate Ridge, Cascadia Margin. Oregon State University Data Report, 174, 99-3. TRI~HU, A.M., BOHRMANN,G. ET AL. 2003. Proceedings of the Ocean Drilling Program, Initial Reports, 204 (CD-ROM). Available from: Ocean Drilling Program, Texas A&M University, College Station, TX 77845-9547, USA. WHITMAN,W.B., COLEMAN,D.C. & WIEBE,W.J. 1998. Prokaryotes: The unseen majority. Proceedings of the National Academy of Sciences of the United States, 95, 6578-6583.

On-site geological core analysis using a portable X-ray computed tomographic system BARRY

M. F R E I F E L D 1 T I M O T H Y

J KNEAFSEY 1 & FRANK

R. R A C K 2

1Earth Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA (e-mail: [email protected]) 2joint Oceanographic Institutions, 1201 New York Avenue, N W , Washington, DC 20005, USA Abstract: X-ray computed tomography (CT) is an established technique for nondestructively characterizing geological cores. CT provides information on sediment structure, diagenetic alteration, fractures, flow channels and barriers, porosity and fluid-phase saturation. A portable CT imaging system has been developed specifically for imaging whole-round cores at the drilling site. The new system relies upon carefully designed radiological shielding to minimize the size and weight of the resulting instrument. Specialized X-ray beam collimators and filters maximize system sensitivity and performance. The system has been successfully deployed on the research vessel JOIDES Resolution for Ocean Drilling Program's legs 204 and 210, at the Ocean Drilling Program's refrigerated Gulf Coast Core Repository, as well as on the Hot Ice #1 drilling platfornl located near the Kuparuk Field, Alaska. A methodology for performing simple densiometry measurements, as well as scanning for gross structural features, is presented using radiographs from ODP Leg 204. Reconstructed CT images from Hot Ice #1 demonstrate the use of CT for discerning core textural features. To demonstrate the use of CT to quantitatively interpret dynamic processes, we calculate 95% confidence intervals for density changes occurring during a laboratory methane hydrate dissociation experiment. The field deployment of a CT represents a paradigm shift in core characterization, opening up the possibility for rapid systematic characterization of three-dimensional structural features, and leading to improved subsampling and core-processing procedures.

Radiographic imaging of geological samples has been used since 1947 (Boyer et al.) to determine oil saturation in cores and to investigate fluid flow (Morgan et al. 1950). The development of X-ray computed tomography (CT) by Hounsfield (1973) sparked a revolution in X-ray imaging by enabling the non-destructive estimation of density with high spatial resolution, based on the reconstruction of a series of radiographs. Currently, X-ray CT is used to non-destructively characterize geological samples and investigate dynamic processes. Computed tomographic images reveal sediment structure, diagenetic alterations, fractures and flow paths - and permit a large range of other properties to be inferred from quantitative densiometry. X-ray CT also allows the intelligent selection of appropriate subsampling locations by revealing the internal structure of a whole core. Early pioneering X-ray CT work for soil science applications was performed by Petrovic et al. (1982), looking at bulk soil density, and Hainsworth & Aylmore (1983), studying water saturation distribution. In the petroleum engineering field, Vinegar (1986) and Wellington & Vinegar (1987) laid a comprehensive foundation for performing petrophysical characterization of

oil-bearing units, enhanced oil recovery experiments and multi-phase relative permeability measurements using X-ray CT. To date, most of the work performed on geological core characterization has been performed using medical scanners, which are designed for imaging the human body and not optimized for core characterization. The portable CT system described here has been designed specifically for core characterization. Unlike medical systems that require the user to exit the room during system operation, a novel radiological shielding arrangement minimizes the weight and volume of the system and permits operation of the X-ray system in close quarters. To produce the highest-quality threedimensional images, we have designed an X-ray compensator, similar in purpose to the compensators used in early medical systems, to reduce image saturation beyond the edge of the core and improve system sensitivity. A comparison between common specifications for a state-of-the-art medical CT system and the portable core analysis system is shown in Table 1. The new portable CT system has been used on the JOIDES Resolution during the Ocean Drilling

From: ROTHWELL,R.G. 2006. New Techniques in Sediment Core Analysis, Geological Society, London, Special Publications, 267, 165-178. 0305-8719/06/$15.00 ~ The Geological Society of London 2006,

166

B.M. FREIFELD

E T AL.

Table 1. A comparison between the portable X-ray CT and a state-of-the-art multi-slice medical C T system

Scanning speed Resolution Footprint Weight Power consumption Portability Capital costs Maintenance costs

Portable core scanner

Medical CT system

500 slices/1 min 100 gm 1.37 x 0.61 x 2.03 m 250 kg 800 W Yes Moderate Low

32 slices/0.4 s 400 lam 5 x 7 m (manufacturer recommended room layout) 20OOkg 15000W No (tractor-trailer mounted units exist) High High

Program's (ODP) Leg 204 and Leg 210 to image whole-round cores, as well as at the ODP's refrigerated Gulf Coast Core Repository. The system was also used at Hot Ice #1, a methane hydrate research drilling project on the North Slope of Alaska. At the Hot Ice #1 drill site, the system was operated in a room at subfreezing temperatures to stabilize permafrost cores and reduce the rate at which any recovered hydrate would decompose. This paper details the portable X-ray CT system design, shows examples of acquired data and uses a methane hydrate dissociation experiment to demonstrate the system's quantitative capabilities. Suggestions on future research activities and areas needing attention for improving X-ray CT core analysis are presented in the paper's conclusions.

System description A schematic layout and photograph of the portable CT installed on the Bridge Deck of the J O I D E S Resolution are shown in Figure 1. The core is installed vertically, typically within a core holder constructed of aluminium or other weakly attenuating material. The core holder is rotated around its vertical axis. A horizontal gantry holding the X-ray source and detector is raised and lowered by a belt-driven actuator to facilitate the imaging of selected regions of the core. The X-ray source has a tungsten target and a 250 lam-thick beryllium window, delivering up to 130 kV at 0.5 mA, with a variable focal spot size that increases from 5 lain at 4 W to 100 tam at 65W. Computer control permits adjustment of both anode voltage and current. To be able to efficiently capture the cone-beam projection of the core, we used a dual-field 100mm/150mm cesium iodide image intensifier. The image intensifier exhibits some geometrical distortion (commonly referred to as pin-cushioning) and lacks the dynamic range of available solid-state X-ray detectors, but rapidly acquires images and has high sensitivity. Rapid imaging is important for

imaging large numbers of cores and for performing transient studies. Resolution of the CT system is dependent on numerous system parameters, including X-ray source focal spot size, image intensifier phosphor properties, location of the core between the Xray source and the image intensifier, chargecoupled device (CCD) camera resolution and the image intensifier field of view. For the CT images shown in this paper, X-ray spot size varied from 60 to 100 gm depending on power setting. The image intensifier input window was kept fixed at 150mm in diameter, and the camera resolution was 768 x 494. The automatic brightness control on the image intensifier's CCD was disabled to minimize noise. With these settings, pixels on the acquired radiographs were approximately 200 jam2, resulting in reconstructed voxels (2001am on a side) that have a volume of 8 picolitres (pl, x 10 -121). To produce the highest quality images, careful adjustment of the CCD camera focal length and f-stop were performed. To reduce noise and increase the dynamic range of the system, great care was taken to optimize the X-ray beam path for core imaging. Figure 2 shows the primary components of the X-ray imaging system. A beam collimator mounted on the X-ray source was used to minimize albedo, which degrades image quality. To reduce camera noise, multiple frames are acquired and averaged without moving the object, leading to a noise reduction proportional to the square root of the number of frames. Ten frames acquired over a 0.4-s period was considered a reasonable trade-off between speed and image quality for the images acquired in this paper, although that number can be adjusted to match the requirements of the user's particular application. As part of the beam collimator, a computercontrolled copper shutter mounted in front of the beryllium exit window can optionally be used to reduce the soft X-ray content of the beam. This is normally only performed at

GEOLOGICAL CORE ANALYSIS USING X-RAY CT

167

(a)

(b) Fig. 1. (a) Schematic diagram showing the components of the portable X-ray CT system. (b) The CT system installed in the Bridge Deck JOIDES Resolution core laboratory during ODP Leg 210. energies above 100kV to reduce the influence that the effective atomic number has on X-ray attenuation, and lead to more precise density estimates. Below 100kV, there is a significant component of photoelectric absorption, which is proportional to effective atomic number. Reducing the X-ray voltage (softer X-rays) results in increased photoelectric absorption and serves to highlight contrasting media and differences in fluid-phase saturation. This also leads to greater uncertainty in the density estimates because of increased beam-hardening effects (filtering of soft X-rays). A concise discus-

sion of the effects of spectral energy, including implications for core analysis, is provided by Wellington & Vinegar (1987). An aluminium compensator (Fig. 2), installed between the sample and X-ray imager, reduces the dynamic range of the X-rays incident upon the image intensifier and increases sensitivity to attenuation variations within the sample. Because of the geometry of passing a coneshaped beam through a cylindrical object, the X-rays passing beyond the outside edge of the core holder are not attenuated by the object; thus, the compensator is thickest in this region.

168

B. M. FREIFELD E T AL.

Fig. 2. A schematic diagram showing the portable X-ray CT's X-ray beam path. The X-rays are first hardened through a filter and collimated before they pass through the geological core. The acylindrical compensator reduces the variation in intensity of the X-rays striking the image intensifier that results from the cylindrical geometry of the core. By greatly attenuating the X-ray beyond the edge of the core holder, the aluminium compensator eliminates image blooming, an image defect that involves pixels on a CCD approaching their electron charge capacity and spilling over into nearby pixels, washing out the image. Where the core is thickest, in the centre, the attenuator is thin, providing minimal additional attenuation of the X-rays. The result is an image of fairly uniform intensity striking the

image intensifier. The aspherical compensator was designed using a fan-beam X-ray path simulation, assuming a core of uniform density. The slight variation in attenuation that occurs along the vertical axis of the core (as X-ray conebeam angle increases) was ignored. While the qualitative benefits of using the compensator have been mentioned above, Figure 3 provides a quantitative example of how density resolution can be improved by

Fig. 3. Radiographs taken with and without the aluminium X-ray compensator. Histograms for the mid-region of each image show how the compensator permits use of greater X-ray energy, resulting in a broader distribution of pixel values. This translates into CT reconstructions with greater density resolution.

GEOLOGICAL CORE ANALYSIS USING X-RAY CT using the compensator, without having to resort to cameras with increased dynamic range. For this discussion, it should be mentioned that every radiograph is composed of pixels with an intensity spanning a range that can be digitally represented, i.e. 10 or 12 bits. Similarly, a dynamic range, or bit depth, can be assigned for any region of interest. The dynamic range of a region of interest will be less than or equal to that of the whole image. As noted previously, without the compensator in place a large variation in intensity occurs from the outside of the core image to the inside. With the compensator in place the dynamic range across the image is reduced, and greater contrast can be seen in any particular region of interest. Figure 3 shows radiographic images of sandstone cores taken with the X-ray energy set below the threshold at which significant image blooming occurs, along with a histogram of pixel intensity for the region of interest enclosed in the white rectangles. The image taken without using the compensator was taken with the X-ray source set to 90 kV and 250 laA, whereas the image with the compensator was taken at l l0kV and 300laA. For the region of interest without the compensator, the histogram reveals an image bit depth of 6 bits. The image with the compensator has a bit depth of 7.7 bits, representing a more than threefold increase in the X-ray attenuation resolution. The increased performance provided by the compensator is significant. The aluminium compensator also contains milled flats that extend beyond the edges of the acylindrical region, used for correcting for fluctuations in X-ray tube intensity. By normalizing each image by the average image intensity in the reference region, one source of errors in density estimates is eliminated. This is particularly important for experiments that span long periods, such as petroleum core floods, which can span several weeks in duration. Note that large variations in X-ray current can be corrected for using this method, but, because of photoelectric absorption, changes in X-ray voltage produce non-linear changes in the image intensity. Thus, X-ray tube current can vary over a broad range, and all of the images can be normalized to each other. Any change in the voltage setting, however, will require an independent system calibration for each selected voltage. The image reconstruction software Imgrec (developed at Lawrence Livermore National Laboratory) was employed to perform fanbeam convolution back projection (CBP) and Feldkamp reconstruction of the acquired conebeam radiographs (Feldkamp et al. 1984). The CBP algorithm is used for rapid reconstruction

169

of acquired radiographs to verify correct parameter settings (only a few seconds are required to reconstruct a single horizontal slice), but does not account for the divergent cone-beam projection geometry. This limits CBP to the region near the radiograph's mid-plane. The Feldkamp algorithm corrects for the divergent cone-beam geometry and can be used to accurately reconstruct images with X-ray projection angles up to 6 ~ or 7 ~ However, beyond that point, the approximations that make the Feldkamp algorithm computationally tractable result in noticeable geometric distortions. The images shown in this paper have used cone-beam angles from 6~ to 10~ The reconstruction of 180 radiographs into a 10cm three-dimensional (3D) volume dataset takes approximately 10 rain on a 3 GHz PC. For either the CBP or the Feldkamp algorithm, it is important to acquire both a dark image from the camera (an image with the X-ray beam off) and a background image (where no object is in the beam path), which is subtracted from subsequent images of the object to account for X-ray intensity variations across the imaging plane. Each radiographic image contains an array of pixels representing the linear attenuation coefficient, #(0, ~p), measured along a beam path, L, from X-ray source to detector location (0, ~). For a monochromatic X-ray source Lambert's law,

I = Ioe- Ji, ~(o,~.~,)dl, relates the measured intensity I incident on the image intensifier to the X-ray source intensity, I0. For a material with a bulk density p and a constant electron density to mass ratio in a monochromatic X-ray field, the relationship #/p is a constant. Since our X-ray source is polychromatic and the object contains materials with various electron density to mass ratios, the relationship between p and # becomes non-linear. We can use this to our advantage in acquiring data by using lower X-ray voltages to improve the contrast between dissimilar materials, or higher voltages with filters to remove the softer X-rays to improve density estimates. To convert X-ray attenuation to density, we perform the Feldkamp 3D CT reconstruction on a sequence of radiographic projections. The CT reconstruction converts the sequence of measured #(0, ~) to a 3D linear density coefficient #(x,y,z). An experimentally determined calibration, relating # to p, is used to convert #(x,y,z) to density, p(x,y,z) for the imaged object. A reconstructed image of a calibration standard containing a variety of materials of

170

B. M. FREIFELD E T AL.

(a)

(b) Fig. 4. (a) A single horizontal slice from a CT reconstruction of a cylindrical density reference standard containing rods of aluminium, wax, air, nylon, high-density polyethylene (HDPE), PVC and acrylic. (b) A linear relationship between density and attenuation is established using the calibration standard shown in (a). The PVC data point is not used in the regression because the high effective atomic number of the chlorine atoms increases the X-ray attenuation.

known density is shown in Figure 4a. It should be noted that, whereas a radiographic image displays less dense regions as brighter, the common convention for CT reconstructed images, followed here, is to show less dense regions as darker and more dense regions as brighter. The density standard consists of a 7.62 cm-diameter PVC cylinder with a series of vertical holes, each of which con-

tains a rod of different materials. For this standard, Figure 4b shows density v. attenuation, along with regression analysis of the data. The data for PVC are not used for creating the calibration curve, because the high atomic number of the chlorine contributes to a significant photoelectric absorption component in the X-ray attenuation, resulting in an overestimated density.

GEOLOGICAL CORE ANALYSIS USING X-RAY CT The reconstructed CT image of the calibration cylinder displays artefacts and aberrations that are worth noting. Geometrically, because of image intensifier pin-cushioning, there is some visible elongation of the reference rods, which should appear circular. This aberration can be removed by remapping the pixels on each radiograph to eliminate the distortion prior to performing CT reconstruction. Quantitatively, there are negligible changes in the estimated material densities due to geometrical distortion. Beam hardening, due to the polychromatic nature of the X-ray beam, is a reconstruction artefact responsible for the polyvinyl chloride (PVC) cylinder appearing brighter in the centre than at the edges. Unlike typical beam hardening, which make the outer edges of an object appear more dense (since it attenuates soft X-rays) and the centre appear less dense (due to the interaction with harder X-rays), the aluminium compensator adds a counterintuitive beam-hardening aberration. This is because the background correction image used to normalize for intensity variation across the image intensifier suffers from beam-hardening effects, making the image appear disproportionately dark at the edges and bright at the centre. The background corrected CT core images thus appear bright (less dense) at the edges, and dark (more dense) in the centre. There are several ways to eliminate or correct for beam-hardening effects when using the compensator. The simplest method is to perform a polynomial correction to the radiographs that removes the beam-hardening trend. This is the method that has been applied to the images in this paper. The second method, which will be carried out in the future, is to design the compensator using a polychromatic ray-path simulation. Several innovations in the CT system make it both portable and radiologically safe for use in a core laboratory. The key to transportability is minimizing the volume enclosed within lead shielding. The usual shielding method for a fixed system, encapsulation of the entire unit or room within a lead enclosure, would have resulted in a unit of limited portability. We minimized the shielding required to enclose the X-ray path by forming a cross that translates along the core axis (Fig. 1). One arm of the cross encompasses the main X-ray beam, while the other arm reduces radiation scattered along the core axis. To permit loading and unloading of the core, we split the vertical arm of the cross, allowing it to open and close by telescoping back over the horizontal arm. The entire system is designed to meet the United States radiological requirements for a

171

cabinet safe system (21CFRw US Food & Drug Administration 2004). Redundant safety interlocks are located both on the access door and the shielding, to prevent energizing the Xray unit while personnel are manipulating the core.

Deployment on Ocean Drilling Program Leg 204 The portable X-ray CT system was first deployed on the JOIDES Resolution for Ocean Drilling Program Leg 204, Drilling for Hydrates at Hydrate Ridge, Cascadia Continental Margin, approximately 50 nautical miles from the Oregon Coast. A radiation safety check was performed to verify that no hidden damage occurred to the radiation shielding during transport. Training of JOIDES Resolution technical staff was conducted. Both the portability and ease of use of the CT system were demonstrated by the CT system's set-up and operation within a 12-h period. During Leg 204, the CT system was used in two different modes: (1) to linearly scan the core, taking two images at 90 ~ orientation to each other at 10cm intervals along the length of the core; and (2) in CT mode, which acquires 180 images with a 2 ~ core rotation between each image. In less than 2 min, the linear imaging mode captures gross structural features along the entire length of a 1.5 m core. These data can be used to perform high-resolution densiometry. Over 12000 radiographs from more than 500 cores were acquired during Leg 204, with the images available for both real-time viewing and electronic archival storage. The CT imaging mode takes between 1 and 3 min to image a 10 cm core section, depending on the final desired resolution. The radiographic images provide quantitative densiometric measurements with high spatial resolution. An alternative methodology for performing systematic density measurements from whole cores is CS 137 gamma-ray densiometry (GRD), which is part of the JOIDES Resolution's multi-sensor core-logging system (Blum 1997). The G R D provides an integrated average density along the mid-line of the core over a 1 cm 2 area, with a typical sample frequency of one data point every 2.5cm. The spacing of G R D data points is limited because it takes several seconds (e.g. 4s for a 7cm-diameter core) to acquire each measurement. The X-ray radiographs provide a 2D dataset with a resolution of 200 gm. By acquiring two images at 90 ~ orientation to each other, the X-ray provides some indication as to gross structural features that may be ambiguous from a single image.

172

B. M. FREIFELD E T AL.

Fig. 5. Radiograph of ODP Leg 204 1251B15 siltyclay sulphide-rich bioturbated core. The bright horizontal features result from disturbance of the core during recovery caused by exsolution of gas. The dark deposits are sulphide infilling of burrows. The radiographs taken during linear scanning reveal more than just gross structural features and densiometric data. Small heterogeneities are apparent, particularly if the supporting matrix is fairly uniform. As an example, a radiograph taken from ODP Leg 204 Site 1251B showing dark sulphide deposits filling in former bioturbation features is shown in Figure 5. The bright horizontal fractures are created by gas evolution during core recovery, which results in significant changes in the core fabric. To estimate density using the X-ray radiographs, three processing steps are performed. The first step is to correct for fluctuations in the energy emitted from the X-ray source using the reference region from the aluminium compensator. Second, geometrical effects (scanning through the cylindrical core and compensator using a divergent cone beam) are corrected by normalizing each acquired radiograph by an ideal image generated by a homogeneous core. This image can either be calculated theoretically or obtained experimentally. For this paper, an experimentally obtained image from a homogeneous sediment core was used to normalize the radiographs. Finally, the normalized data are converted from corrected attenuation to density, using a linear calibration like the one established in Figure 4b. Figure 6 shows radiography data from ODP Leg 204, Site 1251 B37 compared to G R D and logging-while-drilling (LWD) density logs taken in

nearby ODP Leg 204 1251A (Tr6hu et al. 2003). Density estimates using the X-ray radiography, G R D and LWD data are plotted every 2ram, 2.5 cm and 15cm, respectively. The X-ray radiography density estimate is calculated using the average attenuation measured for the 100 pixels contained within a 2 x 2 mm region. Both X-ray radiography and G R D are able to infer a dense carbonate-rich feature (corroborated by a petrologists visual core description) a few centimetres in length at 305.45 m below sea floor (mbsf). Only the X-ray radiographs are able to reveal the lateral extent of this feature across the core, as well as the presence of another thin dense feature at 305.32 mbsf. Since the LWD was not taken in the same borehole, a direct comparison for this particular feature is not possible. The radiographic data (i.e. between 306.4 and 306.5mbsf) show very-well-defined areas of core disturbance interspersed between regions of intact core. Because of the limited spatial resolution of the G R D measurements, areas of the core that are heavily disturbed by gas exsolution appear as increased noise in the data, whereas the radiographs shows the subhorizontal cleavages in the core. The LWD measurement does not display the density fluctuations, typical of the disturbed core, caused by gas exsolution during core recovery. It is encouraging that all of these methods offer very good overall agreement for both density trends and absolute estimated value.

Hot Ice #1 Methane Hydrate Research Well Methane hydrate is a naturally occurring clathrate compound, commonly found within deep oceanic or permafrost regions. In methane hydrate, crystalline water encages a gas molecule. Interest in naturally occurring gas hydrates is increasing because of their energy resource potential, as well as their role in climate variability (Kvenvolden 1988). An exploratory hydrate drilling programme, Hot Ice #1, targeting the Sagavanirktok Formation was drilled down to 427 m in April 2003, and completed in February 2004 to a total depth of 701m. The portable X-ray CT was operated as part of a mobile laboratory to non-destructively image cores recovered during drilling operations. The room in which the X-ray CT was operated was kept at temperatures of between - 1 0 and - 5 ~ to prevent deterioration of the permafrost cores and to minimize dissociation of any recovered hydrates. Dissociation of hydrate is the phasechange transformation from the crystalline hydrate to distinct water (either liquid or solid) and gas phases.

GEOLOGICAL CORE ANALYSIS USING X-RAY CT

173

Fig. 6. A comparison of different methods of estimating bulk density for ODP Leg 204 1251B 37. The graph shows data from X-ray radiography, Cs 137 gamma-ray densiometry and logging-while-drilling (LWD) density measurements. Prior to the initial mobilization to Hot Ice #l, hydrate dissociation studies were conducted using the X-ray CT scanner to spatially and temporally track the conversion of methane hydrate to methane and water ice using synthetic methane hydrate samples. Figure 7 shows a simplified schematic of the hydrate dissociation experiment. The synthetic hydrate was manufactured by the method detailed in Stern et al. (1996) and combined with water ice and sand in a sealed pressure vessel. Hydrate dissociation, stimulated by allowing the room air to warm the sample in the pressure vessel, was monitored by measuring the temperature and pressure and periodically acquiring CT data (Freifeld & Kneafsey 2004). The goal was to determine how sensitive CT imaging is to spatial and temporal changes that occur during the hydrate dissociation process, so that these measurements can be applied to

natural hydrate-bearing samples recovered during the Hot Ice #1 coring operation. Figure 8 shows a sequence of reconstructed CT slices tracking the progression of dissociation of the synthetic methane hydrate. Differential images, created by subtracting the baseline image from subsequent images, are used to highlight relatively subtle changes in density as dissociation progresses. The difference in the images taken prior to the start of the dissociation process, and after dissociation is complete (as verified by the pressure in the vessel), yields estimates of total hydrate saturation and spatial distribution. Visually, the dissociated hydrate appears brighter as methane leaves the system. Quantifying hydrate dissociation changes in the CT data depends on accurate system calibration, and high-resolution densiometry. The three rectangular highlighted regions shown in the

174

B. M. FREIFELD ET AL. 61 min difference image (Fig. 8) were used to calculate changes in density. Since any one voxel has considerable noise, confidence in density estimates can be increased by averaging multiple voxels, assuming that the region averaged over is homogeneous, and any variation in measured attenuation results from normally distributed noise. The 95% confidence interval for an estimate can be expressed as o"

P0.95(X) = p(x) • t0.95,n X/~

Fig. 7. A simplified schematic of a hydrate dissociation experiment. Samples of ice and hydrate have been mixed in a sand matrix. The pressure and temperature within the sample vessel are monitored during thermally induced dissociation of the hydrate. The X-ray CT periodically acquires images to spatially resolve the location of the dissociating front.

where n is the number ofvoxels, o- is the standard deviation of density estimates and t is Student's t distribution. Water ice is used as a reference material in this experiment, since it melts at a much higher temperature than the stability point for methane hydrate ( - 8 0 ~ at 1 atm), resulting in a region with constant density throughout this experiment. The 95 % confidence interval for estimating the density change for the water ice is shown in Figure 9a as a function of the length of a cubical region of interest. As the cubic region of interest is increased to a few mill• in length, the uncertainty declines to • gcm -3. Figure 9b shows the 95% confidence intervals for hydrate density changes during dissociation as a function of cubic region of interest, taken

Fig. 8. X-ray CT images of a synthetic hydrate dissociation experiment. The baseline image shows the location of hydrate and water ice in a sand matrix. The difference images reveal the progression of dissociation within the hydrate nodules. The grey scale bar only applies to the difference images. The rectangular highlighted regions in the difference image acquired at 61 min were used to calculate changes in hydrate density.

GEOLOGICAL CORE ANALYSIS USING X-RAY CT

(a)

175

0.15

~ " 0.10

E

0.05

>,

~ _ _ . _ _ _ e-

0.00

10

,,"g'~176 e-.

f

J

J

I

J

L) -0.10

-0.15 1

(b)

2 Cube Length (mm)

3

0.10

8 min 35 min - - - 44 min'

0.05

0.00

V

>,

e-

f

-0.05

. , . , . , .--. --, ;-..--. :-'. :-..-"

10

C

. m

O) C

-0.10 ...-

1-

r

-0.15

-0.20 0

1

2

3

4

C u b e L e n g t h (ram) Fig. 9. (a) The estimated change in density for water ice used as a reference material in a hydrate dissociation experiment, showing the upper and lower 95% confidence intervals as a function of region of interest size. The density changes were calculated using the difference between the baseline CT dataset and a dataset acquired after 44 min. The expected change is 0.0 g cm 3. (b) Time progression for changes in density of hydrate as dissociation progresses, expressed with upper and lower 95% confidence intervals as a function of region of interest size. The final dataset, taken at 44 min, was acquired after dissociation was determined to be complete by independent pressure measurements. at three different times after the start of the experiment. As the hydrate dissociates and methane gas evolves, the density of the hydrate is seen to decrease. At the start of the experiment the density of the methane gas encaged in the porous hydrate is calculated to be 0.840 g cm -3, based on the known stoichiometric ratio between water and methane. The hydrate density decreased by 0.088gcm -3 after dissociation concluded, as calculated from the X-ray difference image at 44

min. The estimated density reduction relies upon a CT calibration using regions of interest containing sand, hydrate and ice, and their known densities. Figure 9b shows that even though Xray CT system resolution may be 200 gm, if the objective is to determine small changes in hydrate saturation then the smallest region of interest that yields reasonable confidence has a minimum spatial dimension of approximately a 2.5mm cube, or roughly 2000 voxels.

176

B. M. FREIFELD E T AL.

Fig. 10. X-ray CT image of an interbedded sandstone from Hot Ice #1. The upper detail shows wispy submillimetre laminae of fine sand, while the lower detail shows interbedded organic laminae and small claystone clasts. The dark subhorizontal feature located approximately 10 cm below the top of the core is an ice lens. The sedimentary structure at Hot Ice #1 contains thick sequences of conglomerates, sandstones, mudstones and coals. Coring operations recovered no hydrates, which was corroborated by standard geophysical logs and the general absence of finding significant gas-bearing formations. X-ray CT images were acquired of more than 200 1 m-long core tubes. Operationally, the only problem to note from work at the Arctic Platform Mobile Laboratory, in a space maintained at - 5 to - 1 0 ~ was the failure of a coating on one of the objective lenses in the image intensifier. The coating appeared to delaminate, causing a slight distortion in one portion of the image. This problem did not

occur when the system was operated at the warmer, 2-6 ~ temperatures within the Ocean Drilling Program's Gulf Coast Core Repository. A C T image of permafrost sandstone recovered at Hot Ice #1, 210.7m below ground surface (mbgs), is shown in Figure 10. The slice shown from the 3D dataset is a mid-core vertical plane, resulting from averaging five 200 lain-thick slices. The core displays numerous interbedded laminae, ranging from wispy millimetre-sized fining layers, as shown in the upper detail, to organic interbeds containing small claystone clasts. The dark subhorizontal feature near the top of the core is an ice lens that has formed near a set of fine sand laminae. Changes in bedding strike and dip are

GEOLOGICAL CORE ANALYSIS USING X-RAY CT clearly visible and can be used to define strike set thicknesses for various sedimentary layers. Figure 11 shows vertical and horizontal CT images taken of three different core samples collected during Hot Ice #1. Figure 1 la shows vertical CT images of an unlithified sand (570.3 mbgs) with interspersed, rounded mollusk-shell fragments. The dense (bright) region (last image on the right) near the top of the core is an iron claystone concrete. Figure 1 l b shows horizontal CT images of a conglomerate cemented in a sand-ice matrix bearing numerous ice lenses

177

(dark features). Individual gravel-sized fragments and small claystone clasts (bright rounded features) are visible. In Figure l lc horizontal CT images reveal abundant mussel fossils contrasting with the fine sand matrix.

Conclusions X-ray CT imaging of recovered core adds significantly to the amount of information that can be systematically obtained in the field. Structural information, porosity and phase saturation can

(a)

(b)

(c)

Fig. 11. X-ray CT images of core recovered at Hot Ice #1. (a) Vertical slices through an unlithified sands containing abundant fossil shells. The large bright region shown in the image to the far right is a clay ironstone concrete. The rounded dense objects are quartz pebbles. (b) Horizontal images of a permafrost conglomerate cemented together by a sand ice matrix. (c) A sandstone core containing abundant mussel fossils.

178

B.M. FREIFELD ET AL.

be obtained. High-resolution images of wholeround core permit intelligent subsampling locations to be chosen, allowing intersection or avoiding specific features. The portable system detailed here offers significant improvements in image quality over previous X-ray CT systems by incorporating specially designed collimators and filters to optimize the X-ray beam path. By minimizing the sample volume enclosed by radiation shielding, carefully selecting rugged components and limiting infrastructure requirements, the X-ray CT we have developed is easily transported and operated. The vertical core orientation minimizes the footprint of the CT, facilitating the instrument's inclusion into confined laboratory space. By performing imaging at the drilling location, the highest-quality data can be obtained before transport and storage lead to core degradation. This is especially important when looking at ephemeral properties, such as gas hydrate saturation. While the portable CT system shown here represents a paradigm shift from previous systems, numerous areas are open for continued development. These areas include incorporation of dual-energy techniques for mineral identification and improved density estimation, datamining software for identifying particular features or structures and integration of the CT system into multi-property measurement systems. The interface for accessing the enormous volume of data acquired in high-resolution 3D imaging needs improvement, requiring advances in data handling and interpretation software. This work was supported by the Assistant Secretary of the Office of Fossil Energy, Office of Natural Gas and Petroleum Technology, US Department of Energy under Contract No. DE-AC03-76SF00098. The authors would like to thank D. Schneberk for making available to us the CT image-processing code, Imgrec, and J. Pruess for assistance in data analysis. In addition, we would like to thank L. Tomutsa, D. Maloney and I. Croudace for their careful review.

References BLUM, P. 1997. Gamma-ray densiometry (Chapter 3). In: Physical Properties Handbook: A Guide to the Shipboard Measurement of Physical Properties of Deep-sea Cores. Ocean Drilling Program Techology Note, 26 (online). Available from World Wide Web:

http://www-odp.tamu.edu/publications/tnotes/tn26/ INDEX.HTM (accessed 7 January 2005). BOYER, R., MORGAN, F. & MUSKAT, M. 1947. A new method for measurement of oil saturation in cores. Petroleum Transactions, American Institute of Mining Engineers, 170, 15-33. FELDKAMP, L.A., DAVIS, L.C. & KRESS, J.W. 1984. Practical cone-beam algorithm. Journal of the Optical Society of America, A, 1, 612-619. FREIFELD, B.M. & KNEAFSEY,T.J. 2004. Investigating methane hydrate in sediments using X-ray computed tomography. In: Advances in the Stud)' of Gas Hydrates. Kluwer Academic/Plenum Publishers, New York. HAINSWORTH,J.M. & AYLMORE,L.A.G. 1983. The use of computer assisted tomography to determine the spatial distribution of soil water content. Australian Journal of Soil Research, 21,435-443. HOUNSFIELD, G.N. 1973. Computerized transverse axial scanning (tomography): Part I, Description of system. British Journal of Radiology, 46, 10161022. KVENVOLDEN,K.A. 1988. Methane hydrate - a major reservoir of carbon in the shallow geosphere. Chemical Geology, 71, 41-51. MORGAN, F., McDOWELL, J. & DOTY, E. 1950. Improvements in the X-ray saturation technique of studying fluid flow. Petroleum Transactions, American Institute of Mining Engineers, 170, 183184. PETROVlC,A.M., SIEBERT,J.E. & RIEKE,P.E. 1982. Soil bulk density in three dimensions by computed tomographic scanning. Soil Science Society of America Journal, 46, 445-450. STERN, L.A., KIRBY, S.H. & DURHAM, W.B. 1996. Peculiarities of methane clathrate hydrate formation and solid-state deformation, including possible superheating of water ice. Science, 273, (5283), 1843-1848. TRI~HU,A.M., BOHRMANN,G. ET AL. 2003. Proceedings of the Ocean Drilling Program, Initial Report, 204 (CD-ROM). Available from Ocean Drilling Program, Texas A&M University, College Station, TX 77845-9547, USA. US FOOD & DRUG ADMINISTRATION. 2004. 21CFRw Cabinet X-ray Systems. Revised 1 April 2004. Center for Devices and Radiological Health, US Food and Drug Administration. United States General Printing Office, Title 21, 8, 629-632. VINEGAR, H.J. 1986. X-ray CT and NMR imaging of rocks. Journal of Petroleum Technology, 38, 257 259. WELLINGTON, S.L. & VINEGAR, H.J. 1987. X-ray computerized tomography. Journal of Petroleum Technology, 39, 885-898.

Nuclear magnetic resonance pore-scale investigation of permafrost and gas hydrate sediments R. L. K L E I N B E R G Schlumberger-Doll Research, Ridgefield, C T 06877, U S A Abstract: Permafrost is a ubiquitous feature of arctic landmasses, and natural gas hydrate occurrence is widespread in the arctic and beneath the sea floor on continental slopes at all latitudes. The mechanical, thermal and hydraulic properties of the subsurface are profoundly modified by ice and hydrate. Nuclear magnetic resonance is a relatively recent addition to the measurement methods used to characterize recovered samples. This review shows how magnetic resonance has been used in two field studies to quantify frozen and unfrozen components of the sediment pore space, to understand the growth habit of ice and hydrate in rock and sediments, and to estimate hydraulic permeability.

It is well known that frozen soils cover a substantial fraction of the land surface of the Earth. It is less well known that vast areas of submarine continental slopes are also frozen due to the presence of gas hydrate, an ice-like solid with a melting point well above 0~ at the elevated pressures found beneath the bottom of the ocean (Kvenvolden 1993). Permafrost- and gas-hydrate-affected sediments have considerable practical importance. Cold region and deep-sea engineering require knowledge of mechanical and transport properties. The presence of gas hydrate effects sea-floor slope stability. This paper describes how nuclear magnetic resonance (NMR) is used to assay ice or hydrate in cores. Moreover, it provides estimates of relative permeability to water, and furnishes unique information about the pore-scale interaction of ice or hydrate with host sediments. First, the properties of permafrost and gas hydrate are briefly reviewed, and the basic principles of N M R are laid out. Then the applications of N M R to the study of frozen core are described in detail. Finally, two field studies are used to illustrate the information available from magnetic resonance measurements.

Permafrost Permafrost is continuously frozen soil present in arctic regions (Anderson & Morgenstern 1973). The upper limit of permafrost is typically within a few metres of the land surface, and the lower limit is controlled by the long-term mean surface temperature and the geothermal gradient (Lachenbruch et al. 1982). The base of permafrost is not necessarily at the 0 ~ isotherm, and may in fact be substantially shallower (Collett & Bird 1988; Collett et al. 1993).

Not all water in permafrost is frozen. The amount of unfrozen water depends on temperature, pressure, water salinity, and the mineralogy and specific surface area of the soil (Dillon & Andersland 1966; Anderson & Tice 1972; Anderson & Morgenstern 1973). Unfrozen water has substantial effects on mechanical and transport properties, including the strength of the sediment, speed of sound, thermal conductivity and permeability to water flows. Thus, unfrozen water content is important to both civil engineers and geophysicists. Sampling of permafrost formations requires some care since it is desirable to maintain the core below 0 ~ Coring subsurface formations normally requires the use of a drilling fluid, which will melt ice. However, if the operation is fast enough, only a thin surface layer of the whole core (typical diameter 10cm) will be affected. Plugs can be drilled from the frozen centre of the core in a chilled laboratory. Numerous methods can be used to quantify unfrozen water in sediment. Classical methods such as dilatometry, calorimetry and differential thermal analysis are reviewed by Anderson & Morgenstern (1973). In recent years non-destructive methods have predominated. Nuclear magnetic resonance (Tice et al. 1982; Smith & Tice 1988) is fast and accurate. The N M R assay is based on a large contrast in magnetic-relaxation times of water in solid and liquid phases, as explained below. Dielectric constant techniques (Smith & Tice 1988), which are based on differing electromagnetic propagation speeds through liquid and solid, have similar advantages.

Gas hydrates Gas hydrates are clathrate compounds in which individual small molecules (which are commonly,

From: ROTHWELL,R.G. 2006. New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 179-192. 0305-8719/06/$15.00 9 The Geological Society of London 2006.

180

R.L. KLEINBERG

but not exclusively, in the gas phase at room temperature and pressure) occupy sites within a crystalline matrix of water molecules (Sloan 1998). There are three structural forms, distinguished by crystal structure and the sizes of the cages, in which the guest molecules reside. The most common form in nature is Structure I, which is exemplified by methane hydrate. The composition is Gx'(H20)6, where G is the guest species and x is approximately 1 for Structure ! hydrate. Gas hydrates form at elevated pressure and reduced temperature, conditions that are satisfied in both marine and terrestrial environments (Kvenvolden 1993). In the ocean, hydrate can exist below the depth at which the pressuretemperature profile crosses the phase boundary, typically around 500m. It does not exist in the ocean proper because it is less dense than water. However, it is readily trapped in sediments and is stable down to the depth at which the geothermal gradient of the solid earth recrosses the phase line, typically several hundred metres below the sea floor. Hydrates are also stable in a band of depths below the surface in arctic regions, overlapping and below the range of permafrost stability. The creation of gas hydrate deposits requires a source of gas. Seeps of natural gas, consisting mostly of methane, are common in many parts of the world, and so are hydrate accumulations. The total amount of hydrocarbon gas trapped in hydrates is immense, and has significant implications for sourcing of fossil fuel and for global climate change (Kvenvolden 1993). Collection and preservation of hydrate core samples is a significant problem. Whereas permafrost only needs to be kept at subsurface temperature to maintain a core sample in its original state, hydrate-affected samples must be kept under pressure, a significantly more difficult problem (ODP Shipboard Scientific Party 2003; Schultheiss et al. 2006). Moreover, it is essentially impossible to reconstitute a hydrate core sample to its original state: once the core exits the temperature and pressure ranges over which hydrate is stable, evolved gas tends to redistribute both itself and the melt water, and can even disaggregate the sample. There are a number of methods of quantifying the amount of gas hydrate in rock and sediment core samples. The most commonly used technique is indirect and destructive of the sample. Hydrate decomposition releases fresh water, thereby changing the pore-water salinity. Comparing the observed pore-water salinity to an assumed background depth profile permits computation of the original hydrate content (Dickens et al. 1997).

The formation of hydrate in unconsolidated sediment has a stiffening effect, so measurement of the speed of sound would appear to be a good technique for determining hydrate content. However, these estimates of hydrate concentration depend on whether the hydrate preferentially cements grains at their contacts, coats grains uniformly, partially supports the frame, floats freely in the pore space, or forms isolated lenses and nodules. As the microgeometry of hydrate disseminated in porous media is unknown, quantitative determinations of hydrate content from wave-speed measurements are uncertain. Other parameters of the interpretation models include the amounts and acoustical properties of the various minerals composing the matrix and their degree of cementation, which may or may not be known. Because saline pore water is a good conductor of electricity and hydrate is an insulator, electrical resistivity measurements have also been used to characterize hydrate core samples. However, interpretation of these measurements depends on empirical models with poorly constrained parameters that depend on sediment properties and on the hydrate saturation itself (Spangenberg 2001). They also require knowledge of porewater salinity, which is a problem when the formation or dissociation of hydrate creates pore-water anomalies. Dielectric constant measurements avoid many of the problems of acoustic and low-frequency electrical measurements. The dielectric constant has some sensitivity to pore-water salinity and hydrate microgeometry, but these are secondary effects. Dielectric constant has been used to measure the presence of hydrate in several investigations (Boissonnas et al. 2000; Wright et al. 2002).

Nuclear magnetic resonance measurements Nuclear magnetic resonance (more briefly, magnetic resonance or NMR) is well known for being able to provide information about molecular structure and motion, and images of the human body. Less well known is its ability to investigate porous media (Kleinberg 1999). N M R is now widely used in the oil and gas industry to characterize the productive potential of hydrocarbon reservoirs (Kleinberg 1996a; Kleinberg & Flaum 1998). Basic principles

To make an N M R measurement in the laboratory, a sample is placed inside a magnet and a coil. Magnetic nuclei are first aligned by the

NMR ASSAY OF PERMAFROST AND HYDRATES static magnetic field B 0, generated by a large electromagnet or superconducting solenoid. Then the nuclei are irradiated by the coil, which transmits pulses with a carrier frequency f0 = (~'/2rr)B0, where 7 is the gyromagnetic ratio of the nucleus. -y is different for each type of nucleus. Hydrogen is the most commonly probed nuclear species, and for hydrogen nuclei 7/2rr = 42.58 M H z T -1. The pulses reorient the magnetic moments of the nuclei. Measurements of magnetic relaxation have proved to be particularly helpful in porous media studies. The transverse relaxation (T z) measurement consists of a series of pulses. The first pulse rotates the nuclei 90 ~ from the B0 direction. This is followed by a long series of 180 ~ pulses. When irradiated with this series of pulses, a nuclear spin system will return a series of equally spaced spin echoes, one after each 180 ~ pulse. The echo spacing, TE, is typically a fraction of a millisecond. The transverse magnetization decay is monitored by measuring the amplitudes of the echoes during the sequence. An entire decay curve is acquired during one echo train, which makes this measurement very efficient. The characteristic decay time of echo amplitude, 7"2,is the transverse relaxation time. The initial amplitude of the decaying proton N M R signal is proportional to the hydrogen content of the sample. The signal from hydrogen locked in solid minerals decays orders of magnitude faster than the signal from hydrogen in liquids. Thus, the signal from the solid is easily excluded from the amplitude measurement by, for example, making the echo spacing (and therefore the time before the first echo) longer than the solid signal relaxation time. Ice and both host and guest components of gas hydrates respond like solids, so they are invisible in N M R measurements designed to be insensitive to solids.

Thus, water in small pores relaxes rapidly, while water in large pores relaxes more slowly. The surface-relaxivity coefficient P2 is a characteristic of magnetic interactions at the fluid-solid interface (Kleinberg et al. 1994; Foley et al. 1996). In sands and analogous materials it is dominated by paramagnetic centres at grain surfaces. There are two other N M R relaxation mechanisms. One of them occurs in bulk liquids, and is independent of the presence of porous media. Another is associated with molecular diffusion through magnetic field gradients. These processes operate in parallel, so the observed magnetization decay of fluid in a single pore is: a4(t) = M0exp

E'] -Y2

(2)

where M0 is the spin-echo amplitude extrapolated to the start of the measurement sequence and 1

1

1

1

: T2B q- T2---7 q-- ~T2

(3)

TzB is the bulk relaxation time. Tzo is a relaxation time connected with diffusion, and is discussed further below. Rocks and sediments generally have very broad distributions of pore sizes, and therefore the magnetization decays can be expressed as a sum of exponential decays (Gallegos & Smith 1988):

M(t) = E m i e x p i

-T2/

(4)

where mi is proportional to the volume of fluid relaxing at the rate 1/ Tei. The sum of the volumes is proportional to the fraction of the material occupied by liquid, the porosity 4~NMR:

M~ = E mi ~ ~bNMR"

N M R of porous media

(5)

i

The N M R relaxation rate of fluids in porous media is controlled in part by relaxation at the pore-grain interface. Molecules in a fluid diffuse, eventually reaching a grain surface where their nuclear spins can be relaxed. In sediments and sandstones, the rate-limiting step is the relaxation process at the surface, not the transport of unrelaxed spins to the surface (Kleinberg et al. 1994). The decay rate due to surface processes, 1/T2s, does not depend on pore shape but only on the surface-to-volume ratio, S / V , of the pore: r2s - p2

181

pore

(1)

To analyse the measurements, the monotonic but non-exponential magnetization decays are fitted to equation (4), where M(t) typically represents the amplitudes of thousands of spin echoes equally spaced in time, and the Tzi are typically 30-50 preselected time constants, equally spaced on a logarithmic scale over the range from tenths of milliseconds to seconds. The number of terms in the summation is somewhat arbitrary, since the exponentially decaying basis functions are not linearly independent. In fact, there are far less than 30 independent pieces of information in a typically noisy decay. Therefore, the set of mi(T2i) are found using a

182

R.L. KLEINBERG

regularized non-linear least-squares technique that renders the results smooth and stable in the presence of random noise (Butler et al. 1981). The function m(T2) is conventionally called a T2 distribution. Bulk relaxation depends only on the temperature and pressure of the fluid and is easily accounted for. The mechanism connected with diffusion through magnetic field gradients poses more important problems. When this process is significant compared to the first two, the relaxation-time distributions are difficult or impossible to interpret. Magnetic field gradients exist in sediments and rocks due to the magnetic susceptibility contrast between diamagnetic pore fluids and paramagnetic mineral grains, which in common sands are about 1% iron. The effect depends strongly on the applied magnetic field B 0. It is very desirable to employ N M R apparatus using low magnetic field (less than about 0.1 T) for core studies (Kleinberg & Horsfield 1990; Kleinberg 1999). At low field, for most sandstones and sand sediments, both the bulk and magnetic field gradient relaxation processes are negligible compared to the surface relaxation process. Then equation (1) can be used to estimate pore size from N M R relaxation time. A pore-space model is required to find pore diameter from the surface to volume ratio. A convenient model is a network of interconnected cylindrical tubes; this is the model used (often implicitly) for mercury porosimetry analysis of natural earth materials. In this model, the surface-to-volume ratio of a pore is S / V = 4/D, where D is the pore diameter. Although this transform from relaxation time to pore size is not rigorous, it has been found to be widely useful. The NMR-derived pore-size distribution of a typical sandstone is shown in Figure 1. The area under the curve is proportional to the liquidfilled porosity 0NMR. The lower axis is the relaxation time plotted on a logarithmic scale. The upper axis shows the corresponding pore sizes, calculated assuming the pores are cylindrical. Thus, the NMR-determined pore diameter is related to the relaxation time T2 by: D = 4pzT2.

(6)

The surface relaxivity of Berea sandstone, obtained by comparing mercury porosimetry to N M R data (Kleinberg 1999), is P2 = 11 p.ms -l. This value is within the usual range of relaxivities of oil reservoir sandstones, which average about 5 gill S-1 . Water associated with clay minerals has a distinct signature in the T2 distribution. It is

Pore Diameter 0Jrn) 1

10

. . . . . . . . . . . . . . . . . . . . . . . . . .

,

T

Berea 3

0.01

0.1

1

T 2 (sec)

o2o429-olb

Fig. 1. NMR relaxation-time distribution for a typical sandstone. The area under the curve is proportional to the liquid-water-filled porosity 0NMRThe corresponding pore-diameter distribution (upper axis) is computed using equation (6) with P2 = l l l . t m s

1.

commonly observed that T2 distributions of sandstones have a foot at T2 < 3 ms (Kleinberg 1996b), which is correlated with the quantity of clay-bound water (Straley et al. 1994). The distinctiveness of clay-bound water in the T2 distribution indicates that it does not exchange with pore water on the timescale of the N M R measurement (Kleinberg 1999). Further information on the capabilities and limitations of N M R measurements of rocks and sediments are available in a review (Kleinberg 1999).

Applications of N M R to the study of frozen core Unfrozen water

There are a number of reasons why water in soil remains unfrozen below 0 ~ including salinity and capillary pressure effects. The depression of the freezing point by capillary forces is (Adamson 1976; Morishige & Kawano 1999): AT =Tm

4%----L-1 Qrps D

(7)

where Tm is the normal melting point, %1 is the surface tension of the solid-liquid interface, Qf is heat of fusion per gram, Ps is the density of

NMR ASSAY OF PERMAFROST AND HYDRATES the solid and D is the diameter of the cylindrical capillary. For the water-ice transition, Tm = 2 7 3 . 1 5 K , Q f = 3 3 4 k J k g -l and P s = 916.2kgm -3. The surface tension of the waterice interface is uncertain; Franks ~1982) cites values ranging from 19.7 to 44 mJ m - (excluding one value at 6.4mJm-2), which average to "Ysl= 28.3mJm-2. Using these values we find A T = (0.1 tam-K)/D. In the permafrost experiments described below, the lowest temperature to which the core samples were exposed was - 1 4 ~ This is just cold enough to freeze water in 0.007 p.m-diameter pores, which are the smallest pores to which the N M R measurements were sensitive. As the temperature increased, ice in successively larger pores melted. Water intimately associated with clay constitutes a distinct fraction of the water found in rocks and soils. As noted in the subsection on ' N M R of porous media' earlier, clay-bound water is correlated with N M R relaxation times in the range T2 < 3ms. Differential thermal analysis shows that some clay-associated water does not freeze until the temperature is lowered to the range - 6 0 ~ < T < - 3 5 ~ (Anderson & Tice 1971). A fraction of water in coal is unfrozen down to -125 ~ (Mraw & Naas-O'Rourke 1979). Thus, at least some of the water associated with clay and coal is not expected to be frozen in naturally occurring permafrost.

Quantitative assay of ice or hydrate N M R spectrometers sensitive to hydrogen in solids have been used to measure structure and properties of gas hydrate in bulk (Davidson & Ripmeester 1978; Ripmeester & Ratcliffe 1988). However, in order that the electronic dead time be shorter than the relaxation times of nuclei in solids, high-frequency (high B0, e.g. 4T) measurements are necessary. However, as pointed out above, core studies require very low applied magnetic fields, 0.1 T or less. Thus, measurements that are most useful for probing pore-scale effects in sediments and rock cores are blind to ice and hydrate. N M R would not appear to be helpful in quantifying the frozen constituents of cores. However, when N M R is combined with a measurement of porosity that does not distinguish between frozen and unfrozen pore water, reliable and quantitative measurements of ice or hydrate saturation are possible. The difference between total porosity and the N M R indication of liquid-water content gives the amount of frozen material present in the measurement volume.

183

There are a number of methods of determining total porosity. For consolidated rock, the sample is thawed and the porosity determined by the standard Archimedes method. Unconsolidated sediments present more challenges. If a permafrost sample is contained in a non-metallic sample tube, or retains cohesion, the thawed sample can be remeasured by N M R and the result corrected for the 8% shrinkage of water upon melting. Hydrates in sediment require further care, as water can be expelled from the sample when gas is liberated upon thawing. A useful technique for both permafrost and hydrate samples is to measure the mass density of the unthawed sample using a gamma-ray densitometer (Blum 1997). The bulk density Pb is

pb=(1-d?)pm+~SwPw+O(1-Sw)Pi

(8)

where Pm, Pw and Pi are the mass densities of the mineral matter, the liquid water and the frozen (ice or hydrate) phase, respectively, 4~ is the porosity, Sw is the water saturation (fraction of pore space occupied by liquid water), and 1 - Sw = Si is the saturation of the frozen phase. The apparent N M R porosity is gbNMR = (~S w

(9)

SO

4 = ,NMR (Pw -- Pi) _}_ (Pm -- Pb) ( " m - Pi) ("m -- "i) ~NMR Si = 1 - - -

(10)

(11)

The density of methane hydrate, which is not stoichiometric, depends on its composition and is generally in the range of 9 0 0 k g m - 3 (Collett 1998).

Pore-scale distribution of frozen components Nuclear magnetic resonance has the unique capability of determining the sizes of pores in which liquid water resides. When invisible to NMR, the presence of ice or gas hydrate diminishes the integrated amplitude and changes the shape of the apparent (i.e. liquid-filled) pore-size distribution. These changes depend on where the solid resides within the pore space, and on the magnetic coupling between pore water and the solid. Therefore, if the pore-size distribution of the thawed sample is determined, an N M R measurement of the same material containing

184

R . L . KLEINBERG

ice or hydrate will indicate whether they partially or fully occupy small pores, large pores or both. The following discussion applies equally to ice and gas hydrate. If ice coats grain surfaces, its relaxivity to liquid water, P2 (water-ice), replaces P2 (water-rock) in equation (1). In principle, pore-size information will be retained, but the transformation between T2 and pore diameter is changed. On the other hand, if the ice grows in the interior of pores, its surfaces add to the silica grain surfaces and equation (1) must be generalized to account for the simultaneous influence of two different surfaces. The relaxivity of the ice-water surface is unknown, but the following end-member scenarios can be identified:

on thousands of sandstones, for which there is order-of-magnitude agreement with conventional water or gas flux measurements (Straley et al. 1994). However, the technique has not been systematically tested in unconsolidated sediments, nor in sediments consolidated by ice or hydrate. The empirical correlation that connects hydraulic permeability, k0, to porosity and a one-parameter measure of relaxation time, T2LM, is

9

C is a coefficient that depends on mineralogy. A large number of measurements on water-saturated clean and clay-rich sandstones showed that typically C = 4000D s -2 (Straley et al. 1994) (1 darcy (D) ,,~ 0.987 x 10-12 m2). Just as the presence of ice or hydrate has an unknown effect on surface relaxivity, it will similarly influence the N M R permeability estimate in as-yet-unknown ways. If hydrate magnetically shields grain surfaces, or strongly relaxes water, at very least the coefficient of equation (12) will be affected. Relative permeability is the permeability of the sediment to a single fluid when two or more constituents occupy the pore space. In a rock or sediment containing oil, water and/or or gas, each of these fluids will have a different relative permeability, which depends on the saturations. Here we use the term to describe the hydraulic permeability when the pore space is partially filled with ice or hydrate:

Ice coats grain surfaces and P2 (water-ice) << P2 (water-rock): in all but the smallest pores, the bulk relaxation process will dominate, and the relaxation-time distribution m(T2) will tend to pile up at the bulk relaxation time T2B. The bulk relaxation time depends on temperature and pressure, but not salinity unless dissolved paramagnetic ions are present; for water at 0~ and latin, T2B = 1.59 s (Simpson & Carr 1958). 9 Ice coats grain surfaces and P2 (water-ice) >> P2 (water-rock): the relaxation-time distributions snap to short relaxation times at the first appearance of ice, then shrink as ice fills the pores. ~ Ice preferentially fills the largest pore spaces and P2 (water-ice) << P2 (water-rock): water relaxes at grain surfaces while being excluded from large pore bodies. The distributions gradually shrink and move smoothly to shorter relaxation times as freezing proceeds. 9 Ice preferentially fills the largest pore spaces and P2 (water-ice) >> P2 (water-rock): water relaxes at both grain and ice surfaces, so the relaxation-time distribution is concentrated at short values of T2. In reality, the N M R response might result from a combination of these effects, for example if the grains are partially coated with ice.

Hydraulic permeability The hydraulic permeability of a porous medium depends generally on the square of the crosssectional dimension of the flow channels (Scheidegger 1960). The sensitivity of N M R measurements to the pore size makes it a good permeability indicator for sandstones. The N M R estimate of permeability has been tested

4

2

ko --- CqSNMRT2LM

(12)

where T2cM is the logarithmic mean value of the T2 distribution T2LM ---- 10[(1/0)~ imi(Tz')l~176

krw -

k(Sw) k0

(13)

(14)

where k0 is the permeability of the fully liquidwater-saturated sediment, and k(Sw) is the permeability at water saturation Sw, with the remaining pore space filled with ice or hydrate at saturation Si = 1 - Sw. Relative permeability may be found when permeability measurements of both fully watersaturated sediment and the same sediment partially saturated with ice or hydrate are available. This is the case in permafrost thawing studies, where the N M R properties of a sample are followed over a range of water saturations. It is also the case at the base of permafrost or hydrate, to the extent that sediment properties are uniform over the transition zone. Water

NMR ASSAY OF PERMAFROST AND HYDRATES saturation is found from SW z

~NMR(Sw) ~NMR (Sw = 1)

(15)

where qSyMR(Sw) and qSNMR(Sw = 1) are apparent N M R porosities in the partially and fully water-saturated sediment, respectively. Using this and equations (12) and (14), the N M R estimated relative permeability is:

-

k(Swl ko

- S4w

\T2LM (1.o)J

(16)

The mineralogy dependent coefficient C does not appear in this ratio. As noted above, this approach has some important limitations: the permeability estimate is based on an empirical correlation, not a flow measurement, and it assumes that the correlation is as valid for ice-affected sediment as it is for water-filled rock. A subsidiary assumption is that N M R relaxation of liquid water at an ice or hydrate surface is no stronger than at a mineral grain surface. It also assumes that differences of the microgeometrical distribution of water in unfrozen and partially frozen rock do not invalidate the correlation.

Core measurements using an N M R borehole logging tool Apparatus Most modern N M R laboratory instruments operate at high magnetic field, B0, and correspondingly high frequency, f0- However, the magnetic susceptibility of ordinary sands and sandstones makes this a poor strategy when investigating fluid in the pore space of rocks, as explained in the subsection on ' N M R of porous media'. Therefore specialized low-field N M R equipment is used for core studies, for example the Resonance Instruments M A R A N , and the N U M A R CoreSpec instruments. These instruments, which use compact permanent magnets, are smaller and simpler to deploy to field locations than large conventional laboratory machines, some of which use superconducting magnets. The Schlumbcrger Combinable Magnetic Resonance Tool (CMR) can also be used to measure core samples, and it is particularly useful for fieldwork. The C M R is an oilfield wireline logging tool rated to survive and operate in arctic, tropical, desert and marine environments. The C M R is basically cylindrical and has the unusual capability of making measurements on

185

compact external samples (Kleinberg et al. 1992). Permanent magnets project a static magnetic field into an external volume. This field is relatively homogeneous over a volume approximately 15cm long and 4cm 2 in cross-sectional area, centred 2.5 cm from the face of the instrument. The static field strength in this volume is B0 = 52mT. An antenna transmits pulses of radiofrequency (RF) with a carrier frequency of f0 = 2.2 MHz to perform transverse relaxation measurements; the same antenna receives the spin echo signals, used to monitor the magnetization decay of water protons. The effective dead time is 200 las. The apparatus is sensitive only to the region in which the static field is relatively homogeneous. N M R cannot make measurements through metal core barrels, but is insensitive to the protons in plexiglas core barrels.

Sea-floor m e a s u r e m e n t s o f g a s - h y d r a t e - s a t u r a t e d core s a m p l e s Location and procedures. Experiments conducted at the sea floor provide opportunities to observe creation of gas hydrates under conditions that in some ways mimic natural processes (Brewer et al. 1997). For this reason, a remotely operated sub-sea vehicle (ROV) was equipped with the C M R and carried out a series of controlled experiments at the sea floor offshore of central California. Measurements were made at various points in the Monterey Canyon between depths of 1000 (36~ 122~ and 3000m (36~ 122~ (Kleinberg et al. 2003a,b). Sea-floor conditions fall easily within the specifications of the CMR, which was designed to operate at temperatures between - 2 5 and +175~ pressures up to 140MPa and water salinities up to the saturation limit of NaC1. The weight of the CMR, 133 kg, was not a serious problem for ROV deployment. However, at 4.3m, the tool was much too long to be mounted on the vehicle. Therefore, the C M R C electronics cartridge was mounted horizontally at the bottom of the ROV. The CMRS sonde (sensor) section was mounted vertically at the front of the ROV, where its magnet and antenna were easily accessed by a manipulator arm and video cameras. A special L-coupling mated the C M R C and CMRS. Rocks and sediments were transported and treated in 76mm outside diameter Plexiglas tubes. Valves were provided which allowed the introduction of methane at the bottom and venting at the top. The sample tubes used for sediments were 600 mm long and had an inside diameter of 70mm, which included 99.3% of

186

R.L. KLEINBERG

the resonated volume of the N M R instrument. The sample tubes used for rocks were 300 mm long and had an inside diameter of 54mm, which included 92.5% of the resonated volume. These tubes were individually machined to minimize dead volume: the resonated volume not within the sample was entirely within the walls of the tubes, which contributed no signal. There was no evidence for signal from bulk sea water inside or outside the sample tubes. Hydrate was grown in porous media by various methods that were hypothesized to approximate natural mechanisms. In some experiments, gaseous methane was bubbled through sea-watersaturated rocks or sediments while within the gas hydrate stability zone of the ocean. In other experiments, methane was dissolved in sea water and allowed to slowly transform to hydrate over a period of weeks.

Hydrate assay and growth habit. In the sandstones the presence of hydrate decreased the apparent N M R porosity, as expected. In one group of experiments samples were allowed to equilibrate at the sea floor at a depth of 1000m for 1 month. The temperature was approximately 4~ and the pressure was approximately 10MPa. At this pressure the methane hydrate decomposition temperature is 12 ~ in sea water (Henry et al. 1999). Thus, the core samples were well within gas hydrate stability conditions, and no free gas was expected to be present; however, there was no independent check for the presence of gas. Results for Berea sandstone are shown in Figure 2. Accuracy of the N M R measurement was established by comparing the sea-floor measurement of sandstone porosity when fully saturated with sea water (top bar) with later laboratory measurement (bottom bar). In the presence of hydrate, the N M R signal (light shaded part of middle bar) was reduced by an amount corresponding to 0.012 of the rock volume. This agreed well with the hydrate volume estimated from the quantity of gas evolved from the sample as it was transported above the hydrate stability zone (dark shaded part of middle bar). Estimates of hydrate volume from the quantity of collected gas should be treated cautiously. Although degassing was carried out well above the hydrate stability boundary, the thermal lag of the rock and remaining gas bubbles in the pores may lead to a systematic underestimate of hydrate volume. For a hydrate-bearing sand pack, the N M R results strongly sugw that no hydrate was formed in two 60 cm volumes of investigation; visual examination revealed hydrate was localized in nodules and lenses outside these volumes.

Fig. 2. NMR liquid-water-filled porosity measurements (light bars) of Berea sandstone at the sea floor fully saturated with sea water (top), at the sea floor partially saturated with hydrate (middle) and in the laboratory fully saturated with salt water (bottom). As hydrate is invisible to the CMR, the apparent porosity is reduced in its presence. The reduction agrees with a hydrate assay based on gasevolution measurements made when the sample was transported above the hydrate stability zone (dark bar). By contrast, in the rock samples, the loss of liquid water in the volume of investigation was proportional to the hydrate content of the rocks as a whole. Thus, the measurements were consistent with generally uniform dissemination of hydrate throughout the sandstone samples. The observation of large nodules and lenses in the sand pack, contrasted to the uniform dissemination of hydrate in the consolidated rock, is central to understanding natural gas hydrate deposits. Bulk hydrate can only form when sand and silt grains are excluded from a growing mass of hydrate. In rock already consolidated, or cemented by the first appearance of hydrate, bulk hydrate cannot form. The presence of hydrate lenses and nodules in drill core (see, for example, Uchida et al. 1999) can only be explained if hydrate does not promptly cement unconsolidated sediments.

Drilling rig measurements o f permafrost Location and procedures.

N M R measurements of permafrost core were made on the Anadarko Hot Ice 1 drilling rig, located approximately 35km south of the Arctic coast of Alaska

NMR ASSAY OF PERMAFROST AND HYDRATES (70~ 150~ (Kleinberg & Griffin 2005). Continuous 7.6cm-diameter core was taken from surface to 428 m during March and April 2003. The lithologies in the drilled interval are unconsolidated fine sands, clays or muds, ice lenses typically thinner than 1 cm, and coals. A mobile arctic core laboratory was installed on the drilling rig, and special techniques were used to preserve the integrity of the permafrost core. After drilling each 3 m interval, a wirelineretrieved core barrel was recovered from the well. The core was extracted, cut into 1 m lengths, and immediately delivered to a core analysis trailer adjacent to the rig floor for petrographic description and N M R measurement using the CMR. The wellbore was maintained at approximately - 3 ~ the rig floor and cutting shack were typically - 1 4 ~ and the core analysis trailer was maintained near - 3 ~ Generally speaking, approximately 60 min elapsed from the time the core was recovered at the wellhead to the start of N M R measurements, which took approximately 5 min each. Typically, each 1 m-long core was measured at one location; hence, the coverage was 15 cm per metre of core length. Some attempt was made to minimize selection bias. However, grossly washed-out sections and conglomerates were both undersampled as measurements of those intervals would be meaningless in any event. Visible ice lenses, which comprised a very small fraction of the recovered core, were generally excluded from the measured volumes as the goal of the investigation was to understand how permafrost interacts with the pore space of the sediment. The C M R is sensitive to electromagnetic interference at 2.2MHz, and to broadband noise sources in general. To minimize measurement noise, a 38 cm-diameter, 85 cm-long openended wood-frame copper screen was used to shield the tool antenna and core sample. It was found that when core protruded from the end of the shield it conducted significant interference to the antenna.

Ice assay and growth habit. The apparent permafrost N M R porosity (liquid-water-filled pore space as a fraction of total sediment volume) was generally less than 0.1, considerably lower than would be expected for shallow unfrozen liquid-water-saturated sediments. In contrast, coals were characterized by unfrozen water contents of about 0.20-0.25 of total volume, in qualitative agreement with previous work. At the base of permafrost, which occurred within a reasonably massive sand body, NMR-determined porosity increased rapidly over an interval

187

0,4 ..........................

I-.

~-,

0

i

0.3

,'

/ "''M'u'd w_e -a"Je'~;

&,1" / ~

..........

Sand Sand Wellhead Wellhead

_

2

-~

/ Sand

! -''jl

~:z -o o.1 i ~ = : " : : ' - ~ 5~

0

~ l~z ~_ -l "! ".

/

," Mud ,'

"~ 1 Er 0$' 2

~

4

6

Elapsed Time (hr)

8

10 o3o4o9-o3b

Fig. 3. Thawing of mud and sand cores monitored by NMR porosity measurements. The cores were precooled to -14 ~ then warmed in an 18 ~ room. The horizontal lines indicate the porosity measurements immediately after retrieval of the core from the borehole at approximately -3 ~ When fully thawed, 92% of the pore space is occupied by liquid water and the remainder is occupied by air.

of 4m. A systematic increase in NMR-determined unfrozen water content was an earlier indication of the base of permafrost than was petrographic examination. Two typical 30cm lengths of core, a sand (from depth 211m) and a mud (from depth 202m), were removed from storage at - 1 4 ~ and allowed to thaw in an 18 ~ room. N M R measurements were made on the cores approximately once per h for 9 h. The unfrozen water porosities are plotted as a function of time in Figure 3. The sand started with significantly less unfrozen water than the mud. The mud completely thawed in 4.5 h, and the sand in 6 h. The data reflect conditions in a volume centred 2.5cm from the surface of the 7.6cm-diameter cores, as described in subsection on 'Apparatus'. In both cases significantly less unfrozen water volume was found at - 1 4 ~ than when the core was first removed from the well at a temperature of approximately - 3 ~ The wellhead values of apparent N M R porosity are denoted by horizontal lines in Figure 3. As the density of ice is 920 kg m -3, the pore space was not completely water-saturated upon thawing so, in the absence of sediment shrinkage, 8% of the pore space was occupied by air after all the ice had melted. Relaxation-time distributions are shown in Figure 4 for the mud and in Figure 5 for the sand. The pore-space model used to convert relaxation time, lower axes, to pore diameter,

188

R.L. KLEINBERG Pore Diameter (gm) 0.01

0.1

0.01

1

0.1 . . . . . . . .

,-N,.

~-

/

0

.~

~

\ ', 9

~,,

\

-

" -

~

!

--~'-- 7:24 hrs wellhead

;

i

,* 0:00 ~ N - - 3:16 9 - - v - - 4 24 i --&-- 5:23

- - - 5 : 4 1 hrs

~' ,

!

0:00 hrs 9 1:26 hrs

",,:

~

i

- , ~ q r

,

hrs hrs hrs hrs

T

. . . . . .

i

100 . . . . . .

/ll~ ~

- 0:,3.rs

/

\

-

\

% '. N

,a

10 T 2 (ms)

-

|

', 1

i

Pore Diameter ([am) 1 10

100 030409-01b

Fig. 4. Relaxation-time distributions of mud during thawing from -14 to 18 ~ Frozen samples are characterized by low NMR amplitudes and short relaxation times. After 1.5 h of warming (solid squares) the core returned to the state in which it was removed from the wellbore (heavy solid line). During the melting process the amount of NMR-visible (liquid) water increased, and successively larger pore spaces thawed. By 7.5 h (outermost line) the permafrost had completely melted, revealing the relaxation-time distribution of the water-saturated sediment. The corresponding pore-diameter distribution (upper axis) is computed using equation (6) with P2 = 5 gm s 1. upper axes, is the network of interconnected cylindrical tubes described in the subsection on ' N M R of porous media'. The assumed relaxivity parameter is P2 = 5 g m s -I . For both sand and mud, the frozen cores are characterized by low-amplitude distributions at T2 < 3 ms. Some of this water could be in small pores, in which the freezing point is depressed due to capillary forces, see equation (7). Much of the unfrozen water is very probably claybound water, described in the subsections on ' N M R of porous media' and 'Unfrozen water', which does not freeze at naturally occurring temperatures. As the cores thawed, amplitude increased, especially at the longest relaxation times. In both Figure 4 and Figure 5 the topmost curves represent the thawed 92% liquid-watersaturated pore-size distributions. These data are not due to a thermal gradient within the N M R volume of investigation, 4 cm 2 in cross-section centred 2.5cm from the core surface. Temperature uniformity can be tested by calculating the Blot number, Bi = hL/k, the

1

10

100

1000 030409-02b

T 2 (ms)

Fig. 5. Relaxation-time distributions of sand during thawing from -14 to 18 ~ Frozen samples are characterized by low NMR amplitudes and short relaxation times 9During the melting process the amount of NMR-visible (liquid) water increased, and successively larger pore spaces thawed. By 6.25 h (topmost line) the permafrost had completely melted, revealing the relaxation-time distribution of the water-saturated sediment. The corresponding porediameter distribution (upper axis) is computed using equation (6) with P2 = 5 lam s . dimensionless ratio of the surface heat transfer coefficient, h, the thermal conductivity of the solid, k, and the volume to surface area ratio, L (Ozisik 1980). In the present case h = 5 . 4 W m -2 K -1, determined from the rate of thawing. For a cylinder the volume to surface area ratio is one-quarter of the diameter so L = 0.0188m. The thermal conductivity, computed using a simple mixing rule (Hearst et al. 2000), is k = 4 . 6 4 W m - 1 K -1 for the unthawed core and k = 3 . 2 0 W m -j K -1 for the thawed core. For both unthawed and thawed core Bi < 0.1. Thus, the temperature inside the cores is uniform during thawing (Ozisik 1980). The data shown in Figures 4 and 5 allow us to narrow the range of possible growth habits of ice in sediments, as described in the subsection on 'Pore-scale distribution of frozen components'. In doing so, we assume that freezing and thawing of water in sediment is reversible, an assumption these experiments have not tested. The data are not consistent with scenario (1). If ice coated grain surfaces and P2 (water-ice) << P2 (waterrock), the low-temperature relaxation-time distribution would be concentrated around T2 = 1.59 s, which it is not. Since the relaxation-time

NMR ASSAY OF PERMAFROST AND HYDRATES

0.01 0.1 1.4 , ........

N t'O

~-~ ~

Pore Diameter (~tm) 1 10 .

.

.

.

.

.

.

.

.

.

.

.

.

.

.

189

100

'

. . . . . .

1.2

--

0:00 hrs 1:18 hrs

.....

3:16 hrs 4:24 hrs

i

-

5:23 hrs

i'~

-- -- 6:13 hrs

E - o 0.8 -55

~ 0.6 "~>, ..oll "5 >" 0.4 o.r

-i

0.2

1

10

1 O0

1000 041110-01

T 2 (ms)

Fig. 6. Data from Figure 5 (except with data for I h 18 min of thawing instead of wellhead data). Each T2 distribution has been normalized by the T2 distribution found after 6 h 13 min of thawing. distributions do not suddenly move to longer relaxation times at the end of the thawing process, scenarios (2) and (4) are also excluded. Thus, we conclude that liquid water is in contact with the grain surfaces, consistent with common observation (Anderson 1967; Churaev et al. 1993). We also conclude that the water-ice interface does not strongly relax the nuclear spins of hydrogen in liquid water. The pore-size control of the melting of ice is illustrated by Figure 6, in which the 7"2 distributions of Figure 5 are normalized by the distribution of the fully thawed sample. A temperature measurement in the interior of the core was not available, but we surmise that the curves of Figure 6 constitute an indirect thermometer. At first, ice melted in successively larger pore spaces as the temperature increased. After about 4 h of thawing, ice in pores smaller than 0.3~tm had melted, indicating that the internal temperature was within about 0.3 ~ of the bulk melting temperature, according to equation (7). In larger pores the effect of pore size on melting point is negligible, and these pores thaw uniformly. Relative permeability. Fortuitously, the base of permafrost, at a depth of 384.0 m, lay within a reasonably massive and homogeneous sand body, which constituted a natural laboratory for exploring the development of frozen sediment. The transition from fully liquid-watersaturated sediment to permafrost occurs over a depth interval of 4 m. These data, together with those from the thawed samples, were used to

determine the saturation dependence of the relative permeability using equation (16). Relative permeability estimates for the thawed samples, and for the sands at the base of permafrost are shown in Figure 7. All data follow a common trend as a function of liquid-water saturation. This is surprising as k0 differs widely among these three datasets. The permeability of the thawed sand was 53mD, and that of the thawed mud was 0.21 mD. The permeability of

10 0

>, 10 .2

E ~. 10 -4

[]

>= 10-6

9

Mud

9

Sand Permafrost Base

10 ~ 0

J

J

I

i

0.2

0.4

0.6

0.8

Permafrost Saturation CMR 030526-07f

Fig. 7. Relative permeability to water computed using equation (16) for thawing mud, thawing sand and cores from the base of permafrost.

190

R. L. KLEINBERG

the sediment just below the base of permafrost was 14 mD. These estimates were found using equation (12) with C = 4000 D s -2.

Conclusions Nuclear magnetic resonance methods used for evaluation of oil and gas reservoirs are also useful in understanding pore-scale properties of permafrost and hydrate-bearing formations. The quantitative assay of unfrozen water content is model independent and does not depend on any adjustable parameters. However, an independent measurement of porosity is required. In the case of permafrost, this can be an N M R measurement of water content after the core has thawed. When dealing with hydrate in unconsolidated sediments, a measurement of bulk density before hydrate decomposition is desirable. The pore-scale distribution of unfrozen water can also be obtained using N M R measurements. At the lowest temperatures attained in this study, unfrozen water appears to be associated with clay. The results of the permafrost study suggest that the silt and clay sediments investigated remain liquid-water-wet in the presence of ice, and that the water-ice interface is not effective in relaxing nuclear spins. In agreement with capillary pressure theory, ice melts first in small pores, and melts in large pores only near the end of the thawing process. Although the permeabilities of the various sediments, as estimated by N M R , vary widely, their relative permeabilities to water as a function of ice saturation are remarkably uniform. However, the use of N M R to estimate permeability of frozen soil or rock is unproven. It would be very desirable to make fluid flux permeability measurements of samples that are also characterized by N M R . D.D. Griffin participated in all phases of the experiments described here. P.G. Brewer and R.F. Sigal motivated the work and made available opportunities for fieidwork.

References ADAMSON,A.W. 1976. Physical Chemistry of Surfaces. Wiley, New York, chap. VII-2. ANDERSON, D.M. 1967. Ice nucleation and the substrate-ice interface. Nature, 216, 563-566. ANDERSON,D.M. & MORGENS~RN,N.R. 1973. Physics, chemistry, and mechanics of frozen ground: A review. In: Second International Conference on Permafrost. National Academy of Sciences, Washington, DC, 257-288.

ANDERSON,D.M. & TICE, A.R. 1971. Low-temperature phases of interfacial water in clay-water systems. Proceedings of the Soil Science Society of America, 35, 47-54. ANDERSON, D.M. & TICE, A.R. 1972. Predicting unfrozen water contents in frozen soils from surface area measurements. Highway Research Record, 393, 12-18. ANDERSON, D.M., PUSCH, R. & PENNER, E. 1978. Physical and thermal properties of frozen ground. In: ANDERSLAND,O.B. & ANDERSON, D.M. (eds) Geoteehnical Engineering for Cold Regions. McGraw-Hill, New York, section 2.4. BLUM,P. 1997. Physical PropertiesHandbook." A Guide to the Shipboard Measurement of Physical Properties of Deep-sea Cores. Ocean Drilling Program Technical Note, 26, chap 3. Online at: http://www-odp.tamu. edu/publications/tnotes/tn26/ (accessed 22 October 2004). BOISSONNAS, R., GOLDBERG, D. & SAITO, S. 2000. Electromagnetic modeling and in situ measurement of gas hydrate in natural marine environments. In: HOLDER,G.D. & BISHNOI,P.R. (eds) Gas Hydrates: Challenges for the Future. Annals of the New York Academy of Sciences, 912, 159-166. BREWER, P.G., ORR, F.M., JR, FRIEDERICH,G., KVENVOLDEN, K.A., ORANGE, D.L., MCFARLANE, J. & KIRKWOOD, W. 1997. Deep ocean field test of methane hydrate formation from a remotely operated vehicle. Geology, 25, 407410. BUTLER, J.P., REEDS,J.A. & DAWSON,S.V. 1981. Estimating solutions of first kind integral equations with nonnegative constraints and optimal smoothing. SIAM Journal of Numerical Analysis, 18, 381397. CHURAEV,N.V., BARDASOV,S.A. & SOBOLEV,V.D. 1993. On the non-freezing water interlayers between ice and a silica surface. Colloids and Surfaces, A79, 11-24. COLLETT,T.S. 1998. Well Log Characterization of Sediment Porosities in Gas Hydrate-bearing Reservoirs. Society of Petroleum Engineers Paper, 49298. COLLETT, T.S. & BIRD, K.J. 1988. Freezing-point depression at the base of ice-bearing permafrost on the North Slope of Alaska. In: Fifth International Conference on Permafrost, Volume 1. Tapir, Trondheim, 50-55. COLLETT, T.S., BIRD, K.J. & MAGOON,L.B. 1993. Subsurface temperatures and geothermal gradients on the North Slope of Alaska. Cold Regions Science & Technology, 21,275-293. DAVlDSON,D.W. & RIPMEESTER,J.A. 1978. Clathrate ices -recent results. Journal of Glaciology, 21, 33-49. DICKENS, GR., PAULL, C.K. & WALLACE, P. 1997. Direct measurement of in situ methane quantities in a large gas hydrate reservoir. Nature, 385, 426~28. DILLON, H.B. & ANDERSLAND,O.B. 1966. Predicting unfrozen water contents in frozen soils. Canadian Geotechnical Journal, 3, 53-60. FOLEY, I., FAROOQUI,S.A. & KLEINBERG, R.L. 1996. Effect of paramagnetic ions on relaxation of fluids at solid surfaces. Journal of Magnetic Resonance, A123, 95-104. FRANKS, F. 1982. The properties of aqueous solutions at subzero temperatures. In: FRANKS, F. (ed.)

NMR ASSAY OF PERMAFROST AND HYDRATES

Water, A Comprehensive Treatise, Volume 7." Water and Aqueous Solutions at Subzero Temperatures. Plenum, New York. GALLEGOS, D.P. & SMITH, D.M. 1988. A NMR technique for the analysis of pore structure: Determination of continuous pore size distributions. Journal of Colloid and Interface Science, 122, 143-153. HEARST,J.R., NELSON,P.H. & PAILLET,F.L. 2000. Well Logging for Physical Properties. Wiley, Chichester. HENRY, P., THOMAS, M. & CLENNELL, M.B. 1999. Formation of natural gas hydrates in marine sediments 2. Thermodynamic calculations of stability conditions in porous sediments. Journal of Geophysical Research, 104, (BI0), 23 005-23 022. KLEINBERG,R.L. 1996a. Well logging. In: Encyclopedia of Nuclear Magnetic Resonance, Volume 8. Wiley, Chichester, 4960~4969. KLEINBERG, R.L. 1996b. Utility of NMR T2 distributions, connection with capillary pressure, clay effect, and determination of the surface relaxivity parameter P2. Magnetic Resonance Imaging, 14, 761-767. KLEINBERG, R.L. 1999. Nuclear magnetic resonance. In: WONG, P.-Z. (ed.) Experimental Methods in the Physical Sciences, Volume 35, Methods in the Physics of Porous Media. Academic Press, San Diego, chap. 9. KLEINBERG, R.L. & FLAtJM, C. 1998. Review: NMR detection and characterization of hydrocarbons in subsurface earth formations. In: BLUMLER,P., BLUMICH, B., BOTTO, R. & FUKUSHIMA, E. (eds) Spatially Resolved Magnetic Resonance. Wiley-VCH, Weinheim, chap. 54. KLEINBERG,R.L. & GtUEEIN,D.D. 2005. NMR measurements of permafrost: Unfrozen water assay, pore scale distribution of ice, and hydraulic permeability of sediments. CoM Regions Science and Technology, 42, 63-77. KLEINBERG, R.L. & HORSFIELD,M.A. 1990. Transverse relaxation processes in porous sedimentary rock. Journal of Magnetic Resonance, 88, 9-19. KLEINBERG, R.L., FLAUM,C., GRIFFIN, D.D., BREWER, P.G., MALBY, G.E., PELTZER, E.T. & YESINOWSKI, J.P. 2003a. Deep sea NMR: Methane hydrate growth habit in porous media and its relationship to hydraulic permeability, deposit accumulation, and submarine slope stability, Journal of Geophysical Research, 108, (B10), 2508, doi:10.1029/ 2003JB002389. KLEINBERG, R.L., FLAUM, C. ET AL. 2003b. Seafloor nuclear magnetic resonance assay of methane hydrate in sediment and rock. Journal of Geophysical Research, 108, (B3), 2137, doi:10.1029/ 2001JB000919. KLEINBERG, R.L., KENYON,W.E. & MITRA, P.P. 1994. Mechanism of NMR relaxation of fluids in rock. Journal of Magnetic Resonance, AI08, 206-214. KLEINBERG, R.L., SEZGINER, A., GRIFFIN, D.D. & FUKUHARA, M. 1992. Novel NMR apparatus for investigating an external sample. Journal of Magnetic Resonance, 97, 466-485. KVENVOLDEN, K.A. 1993. Gas hydrates - Geological perspective and global change. Reviews of Geophysics, 31, 173-187.

191

LACHENBRUCH,A.H., SASS,J.H. ET AL. 1982. Permafrost, heat flow, and the geothermal regime at Prudhoe Bay, Alaska. Journal of Geophysical Research, 87, (BI 1), 9301-9316. MORISHIGE, K. & KAWANO, K. 1999. Freezing and melting of water in a single cylindrical pore: The pore size dependence of freezing and melting behavior. Journal of Chemical Physics, 110, 48674872. MRAW, S.C. & NAAS-O'ROURKE, D.F. 1979. Water in coal pores: Low temperature heat capacity behavior of the moisture in Wyodak coal. Science, 205, 901902. ODP SHIPBOARDSCIENTIFICPARTY. 2003. Site 1249. In: TR~HU, A.M., BOHRMANN,G. ET AL. (eds) Proceedings of the Ocean Drilling Program, Initial Reports, 204, Ocean Drilling Program, College Station, TX, 21-28. http://www-odp.tamu.edu/publications/204_IR/VOLUME/CHAPTERS/ IR204_08.PDF (accessed 22 October 2004). OzlslK, M.N. 1980. Heat Conduction. Wiley, New York. RIPMEESTER, J.A. & RATCLIFFE, C.I. 1988. Low temperature cross polarization/magic angle spinning 3C NMR of solid methane hydrates: Structure, cage occupancy, and hydration number. Journal of Physical Chemistry, 92, 337-339. SCHEIDEGGER,A.E. 1960. The Physics of Flow Through Porous Media. Macmillan, New York. SCHULTHEISS, P.J., FRANCIS, T.J.G. ET AL. 2006. Pressure coring, logging and subsampling with the HYACINTH system. In: ROTHWELL, R.G. (ed.) New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 151-163. SIMPSON, J.H. & CARR, H.Y. 1958. Diffusion and nuclear spin relaxation in water. Physical Revew, 111, 1201-1202. SLOAN, E.D. JR. 1998. Clathrate Hydrates of Natural Gases, 2nd edn. Marcel Dekker, New York. SMITH, M.W. & TICE, A.R. 1988. Measurement of the Unfrozen Water Content of Soils." Comparison of NMR and TDR Methods. US Army Corps of Engineers Cold Regions Research & Engineering Laboratory Report, 88-18. SPANGENBERG,E. 2001. Modeling of the influence of gas hydrate content on the electrical properties of porous sediments. Journal of Geophysical Research, 106, (B4), 6535-6548. STRALEY, C., ROSSINI,D., VINEGAR, H., TUTUNJIAN,P. & MORRISS, C. 1994. Core Analysis by Low FieM NMR. Society of Core Analysts Paper, SCA9404. TICE, A.R., OLIPHANT, J.L., NAKANO, Y. • JENKINS, T.F. 1982. Relationship Between the Ice and Unfrozen Water Phases in Frozen Soil as Determined by Pulsed Nuclear Magnetic Resonance and Physical Desorption Data. US Army Corps of Engineers Cold Regions Research & Engineering Laboratory Report, 82-15. UCHIDA, T., DALLIMORE, S.R., MIKAMI, J. & NIXON, F.M. 1999. Occurrences and X-ray computerized tomography (CT) observations of natural gas hydrate, JAPEX/JNOC/GSC Mallik 2L-38 gas

192

R . L . KLEINBERG

hydrate research well. In: DALLIMORE, S.R., UCHIDA, T. & COLLETT,T.S. (eds) Scientific Results from JAPEX/JNOC/GSC Mallik 2L-38 Gas Hydrate Research Well, Mackenzie Delta, Northwest Territories, Canada. Geological Survey of Canada, Ottawa, Bulletin, 544.

WRIGHT, J.F., NIXON, F.M., DALLIMORE,S.R. & MATSUBAYASHI,O. 2002. A method for direct measurement of gas hydrate amounts based on the bulk dielectric properties of laboratory test media. In: Proceedings of the Fourth International Conference on Gas Hydrates, Yokohama, Volume 2, 745-749.

Quantitative magnetic resonance imaging methods for core analysis QUAN

C H E N 1, F R A N K

R. R A C K 2 & B R U C E J. B A L C O M 1

1 M R I Centre, Department o f Physics, University of New Brunswick, P.O. Box 4400, Fredericton, N B E3B 5A3, Canada (e-mail." [email protected]) 2joint Oceanographic Institutions, 1201 New York Avenue, N W , Suite 400, Washington, D C 20005, USA Abstract: The majority of sedimentary rocks have significant paramagnetic impurities, which lead to magnetic resonance signal lifetimes too short to be detected by clinical magnetic resonance imaging (MRI) methods. Quantitative information is the ultimate goal for rockcore analysis. The SPRITE (single-point ramped imaging with Tl enhancement) imaging technique has proven to be a very robust and flexible method for the study of a wide range of systems with short signal lifetimes. As a pure phase-encoding technique, SPRITE is largely immune to image distortions generated by susceptibility variations, chemical shift and paramagnetic impurities, unlike clinical magnetic resonance imaging methods. It enables systems with transverse lifetimes as short as tens of microseconds to be successfully visualized. Our experimental results show that most sedimentary rocks have a single exponential transverse magnetization decay for T~, which suggests that quantitative imaging of local fluid content can be easily obtained. Some examples of MRI techniques are represented that reveal internal sedimentary characteristics and heterogeneity. In addition, the application of quantitative MRI techniques to examine flow mechanisms in rock cores is outlined.

Petrophysical analysis is very important to evaluate sedimentary formations. Fundamental petrophysical parameters that one might hope to determine through direct measurement include porosity, saturation and permeability. Porosity is the fraction of rock volume available to bear the formation fluids. Saturation is the ratio of the fluid volume of one phase to the total pore volume. Permeability is a measurement of the ability of these formation fluids to flow through the rock. Traditional methods for core analysis are based on bulk measurement, which must remain largely insensitive to any heterogeneity within the core. However, the heterogeneous nature of the sample will strongly influence the fluid flow characteristics of the medium. Thus, resolution of the sample heterogeneity is a key issue for sedimentary formation description and evaluation. Magnetic resonance (MR) and magnetic resonance imaging (MRI) are promising experimental methodologies that are well suited to addressing such issues. The intention of this review article is to review the state of the art in MRI for rock-core analysis. We focus our review on the SPRITE (single-point ramped imaging with Tl enhancement) class of imaging methods, developed in our laboratory, which now appears to be the most general and robust methodology. The first M R experiments were reported in 1946 (Bloch et al. 1946; Purcell et al. 1946). In 1956 M R was applied to study porous media

for the first time (Brown & Fatt 1956). With the advent of MRI (Lauterbur 1973), a wide variety of applications for this method have been explored, particularly in the area of clinical diagnostic imaging. Applications to porous media and rocks have been limited by technical challenges associated with these samples. Many physical parameters can be spatially resolved, in principle, by MRI, such as spin density of a variety of nuclei (ill, 19F, 23Na, 31p, etc.) and signal lifetimes, as well as chemical shifts (Blackband et al. 1986; Vinegar 1986), which may be employed to determine porosity (Osment et al. 1990; Lucas et al. 1993), poresize distribution changes with pressure (Chert et al. 2002b) and pore geometry (Chen & Song 2002), as well as saturation (Chen et al. 2002a), respectively. The investigation of fluid flow through porous media by MRI (Chen et al. 2002a) has been an active topic in many fields, such as petroleum engineering (Chen et al. 1996), groundwater hydrology (Chen & Kinzelbach 2002), soil and environmental science, etc. MRI can also be employed to directly determine the molecular motion associated with diffusion and flow. This provides an opportunity to determine local velocities within porous media or rocks by MRI (Guilfoyle et al. 1992; Mansfield & Issa 1994, 1996; Kutsovsky et al. 1996). Previous applications of MRI to porous media and rocks were based on spin echo

From: ROTHWELL,R.G. 2006. New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 193-207. 0305-8719/06/$15.00 9 The Geological Society of London 2006.

194

Q. CHEN ET AL.

methods, which are very successful in clinical MRI due to the very long transverse MR signal lifetime of biological tissue. However, these traditional clinical MRI methods fail in actual rocks with paramagnetic impurities, since transverse signal lifetimes of these rocks are too short to be detected by these MRI methods. Even if some rocks may have a longer transverse signal lifetime, it is still very difficult to obtain a quantitative image by these methods. Meanwhile, these frequency-encoding MRI methods also suffer from image distortion due to susceptibility variations, chemical shift and paramagnetic impurities. Quantitative information remains the imperative in MRI applications to core analysis.

Magnetic resonance theory In this section we will briefly outline some basic principles of MR. Additional information, and more complete descriptions, may be found in a variety of introductory or advanced reference books (Stark & Bradley 1988; Sanders & Hunter 1989; Callaghan 1991; Hennel & Klinowski 1993; Canet 1996; Bovey & Mirau 1996).

Magnetic m o m e n t The basis of M R is the quantum mechanical property possessed by many nuclei of spin angular momentum. The spin angular momentum is determined by the spin quantum number I. The spin quantum number is related to the spin angular momentum of the nucleus and thus to the magnetic moment, #, through the following equation:

# = 7Ih/27r

(1)

m =-1/2, pertaining to a low- and highenergy state, respectively. The low-energy state is occupied preferentially according to the Boltzmann distribution described in the next section. When exposed to the static magnetic field, the magnetic moment vectors experience a torque. As a result, there is a precession of the magnetic moments around the axis of the static magnetic field at the Larmor frequency, defined by equation: co=TB o

(2)

where a~ is angular frequency in radians per s, B0 is the static magnetic field. The variable 7 is a constant characteristic of the nucleus. For t H, the value of the 7 is 42.58 MHz Y -1 . The precession of the magnetic moment in the static magnetic field is analogous to a spinning top in the Earth's gravitational field.

Macroscopic magnetization The phases of a large group of precessing spins are random. The individual magnetic moment vectors, #, define the surface of a double cone (Fig. 1), with their vector sum yielding a macroscopic magnetic moment, M0, often termed the sample magnetization. The net magnetic moment is aligned parallel to the axis of the static magnetic field (B0). The macroscopic transverse magnetization, Mxy (i.e. magnetization perpendicular to B0), of individual spins is cancelled by symmetry due to the random phases.

a

~ ,Bz

_

where 7 is the gyromagnetic ratio that is a constant characteristic of the nucleus and h is Plank's constant.

Nuclei in a static magnetic f i e M When subjected to a static magnetic field, B0, the magnetic moments of individual nuclei interact with the static field and the energy of the magnetic moments are split into 21 + 1 different energy states corresponding to the 2 1 + 1 possible values of the z component of the magnetic moment. The B 0 field direction defines the z axis. For I H (the major isotope of hydrogen), which has spin quantum number I = 1/2, there are two allowable basic states with orientations 'parallel' (or spin up) and 'anti-parallel' (or spin down) to the applied static field corresponding to magnetic quantum numbers m = + 1 / 2 and

Y

Fig. 1. Macroscopic magnetization, M0, is created by an assembly of magnetic moments, #, with spin quantum number of 1/2. The individual magnetic moment vectors, #, make up the surface of a double cone, their vector sum generates a macroscopic magnetic moment, M0, or sample magnetization, which is aligned parallel to the static magnetic field, B0, and the z-axis.

MRI METHODS FOR CORE ANALYSIS M R as a type of spectroscopy is fundamentally the result of transitions between different energy levels, which for a nucleus with spin quantum number of 1/2 corresponds to the magnetic quantum numbers m = + l / 2 and m = - l / 2 . The energy required to generate a transition in MR is the energy difference between the two levels. In MR, the resonance phenomenon occurs if the radiofrequency (RF) energy is applied at the Larmor frequency, and results in the spins transiting from the low-energy state to high-energy state, and vice versa. The energy difference between the two levels is El~2 - E_l/2 = "yhBo/27r. Two populations can be defined for this twostate quantum system, the low-energy state, or spin up, and the high-energy state, or spin down. The populations of these two states are given by the Boltzmann distribution, the ratio of the spin-down population (Ndown) to spin-up population (Nup) is give by equation: Nd~

Nup

-- e (-~'h~~

(3)

where k is the Boltzmann's constant, 1.38085 • 10 -23 J K -1, and T is temperature in K. Since the energy difference between the two states (El~2 - E_1/2 = 7hBo/27r) is small compared to the thermal energy, k T , the population difference is extremely small. M R is usually considered an insensitive technique due to this unfavourable population ratio. Population differences of just a few parts per million (ppm) are not uncommon. This net magnetic moment, which is the observable sample magnetization, M0, for the spin quantum number I = 1/2 case, is given by equation (4):

Mo =

N'y2(h/27r)2B 0 4kT

(4)

where N is the number of nuclei. Note that the sample magnetization, M0, and ultimately the M R signal will increase with the quantity of material, and the magnetic field strength, but will decrease with increasing temperature. The hydrogen nucleus, I H, is the most commonly studied, and the most sensitive for M R because of its favourable 7 and the ubiquity of 1H, particularly in water-bearing or water-saturated samples. Radiofrequency field

In a conventional M R experiment sample excitation is detected when transverse magnetization is generated. Transverse magnetization will be

195

produced if a small RF field of amplitude B1, rotating synchronously with the precessing spins, is applied. When this RF field acts in a direction perpendicular to the static magnetic field, B0, the macroscopic magnetization experiences a torque of the RF field, and is rotated away from its equilibrium state (macroscopic magnetization aligned parallel to the z axis) to create transverse magnetization. The RF flip angle, 0, is expressed as: 0 = 7B1 tpulse where Bl is the amplitude of RF field and tpulse is its duration. Therefore we can control the RF flip angle by changing the amplitude and/or duration of the RF field. Following the RF pulse at tpulse, the magnetization is subjected to the effect of static magnetic field only and thus precesses about B 0. The precessing transverse magnetization will induce a voltage in a receiver coil, the result is usually termed a free induction decay (FID). The excitation B1 pulse is generated by a RF probe that also functions as the receiver coil. Typically, both real and imaginary components of the FID signal are detected in a quadrature detection scheme. This time-domain signal may be Fourier transformed to produce a frequency spectrum. R e l a x a t i o n times

The effect of an RF pulse is to perturb the spin system from thermal equilibrium. Equilibrium will be recovered by a process known as spinlattice relaxation. The process involves an exchange of energy between the spin system and surrounding thermal reservoir, known as the 'lattice'. The phenomenological description for spinlattice relaxation is written:

d~ dt

-

( ~ - M0) T1

(5)

where M, is the longitudinal component of the sample magnetization and M0 is the equilibrium magnetization, which is in the direction of a longitudinal magnetic field B 0. Thus, the recovery to equilibrium, also known as longitudinal relaxation, is governed by a time constant, 7"1, the spin-lattice, or longitudinal, relaxation time constant. The solution of this equation is given by equation (6): M. = M:(O)exp(-t/Tj)

+ M0[1 - e x p ( - t / T l ) ]

(6)

Q. CHEN ET AL.

196

where Mz(0) is the value of longitudinal magnetization at t = 0, while t in this case is time following RF excitation. The transverse relaxation or spin-spin relaxation, which is characterized by the time constant T2, is the process whereby spins reach thermal equilibrium among themselves. Spin-spin relaxation is the result of the loss of coherence between the nuclear spins. The observed net transverse magnetization decreases as the spins dephase. This process continues until the equilibrium state of random-phase spin precession and zero net transverse magnetization is achieved. 7"2 is known as the spin-spin or transverse relaxation time, which describes the lifetime of transverse magnetization (M,y) due to the application of an RF pulse. The phenomenological description of spinspin relaxation is expressed by equation (7): __dM-~Y_ dt

Mxy 7"2

(7)

The solution of this equation is given by:

Mxy(t) = Mxy(O)exp(-t/r2).

(8)

The C P M G method (Carr & Purcell 1954; Meiboom & Gill 1958 - C P M G represents the authors' names of these two papers. It is the most common method of measuring T2. It consists of one 90 ~ pulse followed by a series of 180 ~ pulses. The time interval between two 180 ~ pulses is the echo time, TE, and the time between two repeated measurements for signal averaging is the recovery time, TR. TR is usually chosen to be long enough to ensure the magnetization recovery to equilibrium is complete. The transverse sample magnetization, following an RF excitation, will process and dephase due to the inhomogeneity of the static magnetic field. The sample magnetization can be 'refocused' following an additional 180 ~ pulse as there is no molecular displacement. As the nuclei rephase, they generate a signal in the receiver coil - a spin echo (Hahn 1950). Multiple 180 ~ pulses can be applied (number of n) to produce a series of echo trains. The echo amplitude decays as a function of the echo time as given by:

Mry(nTE) = M o exp(-nTE/T2).

(9)

The exponential description of relaxation is employed in the case where the interaction terms corresponding to spin-spin relaxation are weak. This regime was described by Bloembergen et al. (1948), which works well for nuclear spins within liquid state molecules.

However, for liquid-filled porous rocks, the relaxation process and the underlying mechanisms are much more complex. Relaxation theory in porous media is discussed in the next section.

Relaxation theory in porous media For fluids confined in porous rocks the T1 and T2 values will be shorter than that of the bulk fluid if the fluid interacts with the rock surface, which promotes M R relaxation (Korringa et al. 1962; Kenyon 1992). There are different relaxation mechanisms, that operate in parallel, which contribute to the overall relaxation rates l/T1 and 1/ T2 (Kleinberg & Horsfield 1990; Chen et al. 2002b):

1/Tl = 1/T1B + p l S / V .

(10)

The two terms on the right-hand side of equation (10) represent two mechanisms of spin-lattice relaxation: molecular motion in bulk fluids and surface relaxation at the pore wall, respectively (Fukushima & Roeder 1981; Cohen & Mendelson 1982). TlB is the bulk spin-lattice relaxation time, Pl is the spin-lattice surface relaxivity, S / V is the surface-to-volume ratio of the pore fluid, while 7 is the gyromagnetic ratio:

l/T2 = 1/T2B + p 2 S / V + ((TGiTE)2Do)/12. (11) The three terms on the right-hand side of equation (11) represent three mechanisms of spinspin relaxation due to molecular motion in bulk fluids, surface relaxation at the pore wall and molecular diffusion in internal magnetic field gradients (the magnetic field gradient due to B0 inhomogeneity is very small, and can be neglected in porous rocks). T28 is the bulk spin-spin relaxation time, P2 is the spin-spin surface relaxivity, Gi is the internal magnetic field gradient, TE is the echo time, while Do is the self-diffusion coefficient of the liquid. The bulk relaxation time is a property of the fluid only. Because the relaxation time of liquid in rocks is much shorter than the relaxation time of bulk liquid, the bulk terms in equations (10) and (11) can be neglected. The surfaceinduced relaxation is due to an interaction between the fluid and the solid surface. Equation (10) may thus be approximated by

1/T l = p l S / V .

(12)

The diffusion-related relaxation of equation (11) is caused by the internal magnetic field

MRI METHODS FOR CORE ANALYSIS gradients resulting from the magnetic susceptibility contrast between the rock grains and porefilling fluids. This phenomenon affects 7"2, not Tl. In rocks, the internal gradients are a complicated function of microscopic pore geometry. A rough approximation to the internal gradients (Gi) inside a homogeneous static external magnetic field B 0 may be developed from a simple geometrical estimate of the grain size and the magnetic susceptibility difference between the fluid and matrix: a i ,.~ A B i / d g ~ A ~ B o / d g .

(13)

ABi is the magnetic field broadening induced by the magnetic susceptibility difference, AX, between the rock grains and the pore-filling fluid, while dg is the mean grain diameter of the rock. Under conditions of a low magnetic fiend strength (i.e. Gi is also small) and short T E (Kleinberg & Horsfield 1990), the enhancement in T2 decay due to diffusion in the inhomogeneous internal magnetic fields is negligible compared to the surface relaxation mechanism. Therefore, the measured 7'2 values are given by:

1 / T 2 = p2S/V.

(14)

Latour et al. (1992) showed that the relaxation time is independent of temperature over the range 25-175 ~ and verified that the relaxation time of fluid in a wide variety of rocks is within the 'fastdiffusion' (Brownstein & Tarr 1979) or 'surfacelimited' (Belton et al. 1988) regime in which the relaxation at the surface is slower than the transport of the hydrogen nuclei to the surface. Thus, the spins experience a rapid exchange of environments so that the local fields in each region of a pore are averaged to their mean value. As a consequence, a single exponential decay is observed for a given pore, and the rate of magnetization decay depends solely on the surface-to-volume ratio (Kleinberg et al. 1994). Equations (12) and (14) form the basis of MR core analysis and log interpretation, where the relaxation times T 1 or T2 are proportional to V/S, which in turn is proportional to pore size. In small pores relaxation is faster than in large pores. The total signal is a superposition of the signals coming from all pores within the measurement volume. It is manifested as multi-exponential decay in a CPMG measurement of transverse magnetization decay:

M(t) = Z

ai(O)exp(-t/Tzi)" i

(15)

197

The overall decay is the sum of the individual decays that reflect the pore-size distribution. An inverse Laplace transform of data following equation (15) will yield a 7'2 relaxation time distribution. In this case the Tzi belong to a preselected basis set of relaxation time constants and Ai (0) is proportional to the number of 1H nuclei with relaxation time T2i (Bulter et al. 1981). Because 7"2 depends linearly on pore size in the case for weak diffusion coupling and strong surface relaxation, the 7"2 distribution corresponds to a pore-size distribution. Similarly, the 7'1 distribution, equation (16), may be obtained following inversion-recovery measurement of Tl and inverse Laplace transform:

M(t) = Z

Ai(0)[1 - 2exp(-t/Tli)].

(16)

i The effective spin-spin relaxation time (T~) may be obtained by analysing the free induction decay (FID). The overall FID decay rate (1/T~) may be represented as the sum of the contributions due to the spin-spin relaxation rate (1/7"2) and contributions due to the heterogeneity of the magnetic field due to the magnet (1/T2m) and susceptibility difference (Ax) between the pore-filling fluid and the solid matrix (1/T2i). The 1/ T2m term is usually insignificant in realistic porous media:

1/T~ = 1/Ta + 1/Tam + 1/T2i

(17)

In sedimentary rocks, our experimental results, and numerical simulation, show that the overall FID decay rate (1/T~) is dominated by the term of 1/T2i, which results in a typical single exponential T~ decay or a Lorentzian line shape in the MR spectrum. This occurs even when the fundamental T2 decay observed in a CPMG measurement is a multiple exponential. This single exponential T~ decay appears to be true even for rocks partially saturated with water and air. The decay rate of the FID and MR line width (1/TrT2i) for rocks may be estimated by the following equation: 1 7ABi ----CAxf 7rT2i -- 27r

o

(18)

where f0 = 7B0/2~r is the Larmor frequency, AB i is the magnetic field broadening induced by the magnetic susceptibility difference, AX, between the rock grains and the pore-filling fluid, and the parameter C is a proportionality constant. While an apparently simple result, the single exponential T~ decay is a very profound observation, as will be outlined in the later sections.

Q. CHEN ET AL.

198

Magnetic resonance imaging theory

100 -!

~L

Spatial encoding

._~ 10

In order to produce an M R image from the M R signals, the experimental signal must be spatially encoded. Spatial information is incorporated into an M R signal through the application of magnetic field gradients (Lauterbur 1973). The magnetic field gradient, for example Gx = dBo/dx, imposes a linearly varying magnetic field across a sample. The spectrum that results from an M R measurement in the presence of a magnetic field gradient should recovery the geometry of the sample, often weighted by the relaxation times, in the direction of the applied magnetic field gradient through a spatially varying Larmor resonance frequency. This modified precessional frequency is given as a function of position, x, in equation (20):

e~ z

0

100

200

300

400

500

600

t (US)

Fig. 2. A semi-logarithmical plot of the free induction decay for a fully water-saturated Berea sandstone following a 90 ~ RF pulse. The best-fit line is a single exponential with a decay time constant T~ of 127 gs.

The M R experiments were performed in a 2.4T, 32cm inside diameter (i.d.) horizontal bore superconducting magnet (Nalorac Cryogenics Inc., Martinez, CA), with one watercooled 7.5 cm i.d. gradient set (maximum gradient strength 100 gauss per cm (G cm-l)) and another water-cooled 20cm i.d. self-shielded gradient set (maximum gradient strength 10Gcm-l). Both gradient sets were driven by Techron (Elkhart, IN) 8710 amplifiers. Bird-cage style 1H probes were employed for M R experiments. The probes were driven by a 2 k W A M T (Brea, CA) 3445 RF amplifier. The M R acquisition was undertaken with an Apollo console (Tecmag Inc., Houston, TX) controlled by N T N M R software. Figure 2 below illustrates the single exponential T~ decay in a Berea sandstone. The semilogarithmical plot of Figure 2 was generated in an experiment using a 90 ~ RF pulse for a fully water-saturated Berea sandstone with diameter of 2.5cm and a length of 5.2cm. The data were fit to the equation:

w(x) = 7(B o + xGx)

In this manner, spatial information is recorded in the M R spectrum and an image should be reconstructed from this spectrum. Modern MRI methods employ a fast fourier transform (FFT) to reconstruct the experimental MRI data into images. A simple consistent mathematical formalism, k-space, may be used to systematically understand image reconstruction methodology for the vast majority of M R I techniques (Callaghan 1991). Because of the fundamental importance of the k-space formalism to modern MRI, we will briefly introduce this notation in one dimension. Consider the resonance frequency of a particular isochromate of spins, at a chosen position in the x-direction displaced from the magnet centre. In a frame of reference rotating at the Larmor frequency, the observed resonance frequency is given by:

a;(x) = TxOx S = Moexp(-t/T~)

(20)

(21)

(19)

where S is the M R signal intensity, M0 is the equilibrium sample magnetization and the MR signal intensity at t = 0. Variable t is the time following the RF pulse, while T~ is the effective spin-spin relaxation time. The fit T~ was 127 ~s. The F I D has a single exponential decay over two orders of signal intensity. Experiment shows that the T~ of the Berea sandstone samples varied from 114 to 127~ts when the water saturation was varied from 9.1 to 100% (Chen et al. 2003), but remained single exponential T~ decay, with multi-exponential Tl and T2 behaviour. We have observed single exponential F I D decay, with T~ largely insensitive to water and air saturation, in many sedimentary rocks.

where the frequency aJ depends on position x, gradient Gx and the gyromagnetic ratio 7. The phase of the M R signal may be represented by a complex exponential, exp[ia~(z)t]. The signal detected from a collection of isochromates distributed in the x-direction will be a summation of the individual signals. For a large number of isochromates, the summation may be written as an integral over position x. We introduce the proton density p(x) to represent that the number of ~H nuclei as a function of position x. A reciprocal space, k-space, k x is defined by equation (22), which has units of m m - l :

1

kx = ~-~Gxt

(22)

MRI METHODS FOR CORE ANALYSIS Thus, the observed signal S(kx) may be written as shown in equation (23):

S(kx) = J p(x)e i27rxk"dx

90

rf

I

(23)

Note that S(kx) and p(x) are related by an analytical Fourier transform. Therefore, the generation of an M R I requires acquisition of the M R signal in k-space, S(kx), in one-dimensional (1D) image, then reconstruction of an image, p(x), by an inverse Fourier transform of the S(kx) through equation (24). Two- and threedimensional (2D and 3D) imaging requires spatial encoding of orthogonal spatial directions. This is equivalent to sampling additional dimensions of k-space:

p(x) = J S(kx)e -i2~x~xdkx.

199

(24)

The k-space signal is sampled with equally spaced data points, and with a number of data points set to a power of 2 in order to employ a F F T for imaging reconstruction. The definition of k-space, equation (22), shows that it may vary either as a function of time for a constant magnetic field gradient or due to a change of the magnetic field gradient for a constant time. The first method is known as frequency encoding where the M R signal is detected as a function of time. The second method is termed phase encoding, and requires changes in the strength of the magnetic field gradient for a specified period of time. The most common form of MRI, known as Fourier imaging, employs frequency encoding in the x direction and phase encoding in alternate directions, as shown in Figure 3. A 90 ~ RF pulse is employed to generate transverse magnetization. A 180 ~ RF pulse is applied to produce a spin echo. Two programmable phase-encoding gradients, Gy and Gz, are employed. A frequency-encoding gradient, Gx, also called a read gradient, is applied to create the positional dependence of frequency during sampling the spin echo signal, the spin echo signal is sampled with time during a constant gradient, Gx. TE is the time interval between the 90 ~ pulse and the centre of the echo, the time interval between the 90 ~ pulse and 180 ~ pulse is equal to the time interval between the 180 ~ pulse and the echo centre. The experiment is repeated at an interval of time, TR, with an increment in the phaseencoding gradients. The signal intensity for any region of space in the resulting M R image, following Fourier transform of the k-space data, is given in

Gx read Gy phase1 Gz phase2

180

echo

I ^^^^^ v

/

v

V

~

\

~-~

Fig. 3. Schematic description of a typical 3D FT imaging method. A 90~ RF pulse is applied to create a transverse magnetization. Two phase-encoding gradients (variable amplitude) are applied along the y-direction (G~.)and the z-direction (Gz). At the same time, a gradient Gx is applied along the readout direction, this gradient is employed to compensate for the phase imparted by the readout gradient for frequency encoding of echo signal. A 180~ RF pulse is applied to form a spin echo. During the echo formation, the read or frequency-encoding gradient (G,-) is applied, which causes the spins to be frequency encoded along the x-direction. equation (25):

where M0 is the sample magnetization at equilibrium, which is proportional to the proton density. Equation (25) also includes two exponentially weighted terms determined by the ratio of TE to /'2 and the ratio of TR to T1. In many systems, such as sedimentary rocks with paramagnetic impurities, a spin-spin relaxation time of less than 1 ms is common. In this case, spin-echo-based imaging methods, for example Figure 3, are not feasible. The transverse signal lifetime, T2, is much shorter than the smallest echo time, TE, which is limited by the multiple switched magnetic field gradients (typical gradient switching times of hundreds of microseconds). Even if the transverse signal lifetimes for some rock samples are longer than the echo time, it is very difficult to obtain spindensity (quantitative) imaging by these M R I methods, due to the fundamental multi-exponential T2 decay in rocks. Quantitative measurement is the ultimate goal for core analysis, unlike clinical M R I methods where relaxation time contrast is generally sufficient. Meanwhile, these frequency-encoding M R I methods also suffer from image distortion due to susceptibility variations, chemical shift and paramagnetic impurities. Quantitative M R I methods for short signal lifetime systems, such as rocks, are discussed in the next section.

Q. CHEN ET AL.

200

S P R I T E imaging technique Single point imaging (SPI) MRI methods have been proven to be useful imaging modalities for studies of short relaxation time systems (Emid & Creyghton 1985; Gravina & Cory 1994). An imaging method with modified magnetic field gradient waveforms known as SPRITE (SinglePoint Ramped Imaging with T1 Enhancement) (Balcom et al. 1996) was proposed by one of the co-authors to decrease the acquisition time for imaging. The SPRITE imaging method is illustrated in Figure 4. Repetitive excitation and acquisition is performed in the present of a ramped or stepped primary phase-encoding gradient, Gx in this case. A single short-duration (a small flip angle, 0) RF pulse is used to generate transverse magnetization after the magnetic field gradients have been switched and allowed to stabilize for each step. The purpose of employing a short-duration RF pulse is to ensure a uniform excitation of overall spins in the sample. As the RF pulse is applied in the presence of a magnetic field gradient, its duration must be short enough to irradiate the overall distribution of frequencies introduced by the gradient. After a fixed duration phase-encoding time, tp, a single complex k-space datum point is acquired on the FID. After each repetition

RF P'l'l'l'i'i.l'l'l'l'l'l.l'l'l I-4

time, TR, the value of the applied primary gradient, Gx, is incremented for 1D k-space sampling, which leads to much more rapid sampling for k-space data and minimizes gradient vibration due to impulsive Lorentz forces. Meanwhile, the gradients on the second and third dimension, in this case Gy and Gz, are incremented following each cycle of Gx leading to simple Cartesian sampling of 2D and 3D kspace data. Three-dimensional imaging can be achieved with the SPRITE method in minutes for favourable samples. Unlike frequency encoding, where the time evolution of the sample magnetization is measured, the SPRITE method is a pure phaseencoding MRI technique. Owing to the fixed phase-encoding time, tp, this method eliminates image distortions due to internal gradients induced by magnetic susceptibility variations, chemical shift and other unwanted effects involved with time evolution of the MR signal (Balcom 1998). Meanwhile, the excitation and acquisition are carried out in the presence of phase-encoding gradients, which enables the systems with a very short T~ lifetimes (on the order of tens of microseconds) to be successfully visualized by SPRITE. The SPRITE technique has been successfully used to visualize a wide variety of samples with short signal lifetime including gases, semi-rigid polymers and a wide range of porous media (Rack et al. 1997; Gingras 2002a,b; Balcom et al. 2003). The local image intensity for a SPRITE experiment is expressed as:

TR S=M0exp

~ ~

I I

Fig. 4. Schematic description of the 3D SPRITE (Single-Point Ramped Imaging with T1 Enhancement) method. The primary phase-encoding gradient, Gx in this case, is stepped with an RF pulse excitation at each gradient level. Sixty-four steps, each on the order of 1ms duration, are typically employed. The second and third gradients, Gy and Gz, are amplitude cycled to phase encode the Y and Z dimensions. A single point on the FID signal is acquired at an encoding time, tp, after each RF pulse. A short duration, low-flip angle (0) pulse assures a uniform excitation of spins throughout the sample. The RF pulses are applied at intervals of TR.

I

-T~

1-exp(-TR/T1)

]sin0

(26)

where M 0 is the local sample magnetization, which is proportional to the local 1H density, tp is the phase-encoding time, and TR is the time between RF pulses. The angle 0 is the RF pulse flip angle. T1 is the spin-lattice relaxation time, while T~ is the effective spin-spin relaxation time. For the SPRITE method the repetition time, TR, is often shorter than T1, which results in a complex image-intensity equation with a T1 contrast due to partial saturation of longitudinal magnetization. One limitation of the standard SPRITE method is the complexity of the image-intensity equation through the terms in square brackets of equation (26) involving the longitudinal relaxation time, Tl, and the

MRI METHODS FOR CORE ANALYSIS

201

repetition time, TR. In addition, partial saturation of the sample magnetization at the centre of the k-space leads to reduced signal-to-noise ratio. This image-intensity equation can be simplified significantly, with improved signal-tonoise ratios, through the application of a centric scan strategy that commences data sampling at the k-space origin. A version of the SPRITE technique, with a centric scan strategy yielding rapid images and with improved signal-to-noise ratio, Spiral-SPRITE and Conical-SPRITE (Halse et al. 2003, 2004), will be outlined in the next section.

Centric scan S P R I T E imaging technique Centric scan SPRITE techniques are also pure phase-encoding M R I methods. They differ from the standard SPRITE methodologies in the pattern in which k-space is sampled. Standard SPRITE collects lines of k-space data, sampling from one extreme, through the centre, to the opposite extreme. For centric scan SPRITE techniques, data acquisition commences at the k-space origin and proceeds to the extremities of k-space. The local image-intensity equation is: S = M 0exp

('p) -~

sin0.

(27)

From this local intensity equation, as long as !p << ir~, the only factor governing this equation is the equilibrium magnetization, M0, which is proportional to spin density and the flip angle, 0. TI and TR simply become resolution parameters, not image-intensity parameters (Halse et al. 2004). The advantages of centric scan SPRITE are profound: simplified image contrast is achieved faster with better signal-to-noise ratio and less stress on the equipment. It is better, stronger, faster and easier. The single exponential T~ decay feature for rocks is crucial to the quantitative nature of the experiment. True density imaging may be achieved for samples with short T~ by acquiring a series of centric scan SPRITE images with different tp, and then fitting equation (27). If T~ is constant, as in the earlier example, the centric scan SPRITE is directly and simply yielding spin-density imaging, which is local fluid-content imaging. The spin-density imaging is crucial to quantitative spatially resolved rock-core analysis. The Spiral-SPRITE method employs a modified Archimedean spiral k-space trajectory, which is illustrated in Figure 5. The SpiralSPRITE pulse program is illustrated in Figure 6.

Fig. 5. An Archimedean spiral k-space trajectory. The plot represents a 64 x 64 2D data matrix covered by the spiral trajectory. Only 2564 grid points were measured over a Cartesian grid of size 642. Data points in the four corners of k-space were neglected, since these points have very low signal-to-noise ratio and make little contribution to the overall image. The Conical-SPRITE methodology is a 3D centric scan technique, employing a series of spiral trajectories along conical surfaces to sample the k-space cube. The Conical-SPRITE method is illustrated in Figure 7. Gx and Gy are sinusoidally ramped with a linear increase of G_. The first k-space point on each cone is sampled in the absence of a magnetic field gradient, all trajectories commence at the k-space origin. Compared with frequency-encoding methods, SPRITE imaging experiments are considered inefficient as only a single datum point is acquired after each step of RF excitation. It is

TR

RF

G~

O.H-n.H,H,O,O,O.n, H.O.O.O-O. H.O.].H.H.U.O,n.O.n. tp y

~-~-L_L~~L rr # ~J

Gy

Fig. 6. Spiral-SPRITE method. The gradients of Gx and Gy are sinusoidally ramped through a spiral trajectory. This diagram illustrates the first few dozen points on a spiral trajectory. RF pulses are applied at each gradient level with a single data point sampled at time tp following each pulse.

202

Q. CHEN E T AL.

tp RF I.l.l.l.l.l.l.l+l.l.l.l.l+l.l.l.l.l.l.l.t.l.l-t.l+l.l.[-I.l.l.l-I.l.l.I. TR H

_

Gz

_

I

Fig. 7. The 3D Conical-SPRITE MRI method. Three phase-encoding gradients (Gx, G~. and G=) are employed. Oscillating X and Y gradients, as well as a linearly ramped Z gradient, define a conical trajectory in k-space. An RF pulse is applied at each gradient level. The repetition time (TR) is the time between successive RF pulses, a single FID point is sampled at a time tp after each RF pulse. This figure illustrates the first few dozen points on a single conical trajectory.

sensible to increase the efficiency of SPRITE imaging methods by acquiring multiple FID points following each R F pulse. A new version of SPRITE technique with multiple F I D points acquisition has been published (Halse et al. 2004). These acquired FID points may be employed for signal averaging to increase signal-to-noise ratio or for T~ mapping and true spin-density mapping.

Visualizing internal structures of sedimentary rocks by M R I For natural sedimentary rocks, such as hydrocarbon reservoir rocks, the available samples are normally derived from cores obtained from limited locations in the whole sedimentary body. As such, they already represent, in same sense, a spatially resolved view of the properties of interest. However, fluid dynamics in these systems is often controlled by small-scale heterogeneity, which may be obtained by MRI. Carbonate rocks are generally highly inhomogeneous, typically characterized by a wide presence of fractures and vugs. Some sandstone rocks may have shale bedding structure and other heterogeneous properties. All of these inhomogencous characteristics significantly influence the dynamics of fluids in the media. M R I mcasuremcnts can reveal these internal sedimentary

Fig. 8. An MR image of a brine-saturated limestone sample. A slice was selected from a 3D SpiralSPRITRE MRI dataset. The image shows a dual porosity system (bright zones, high porosity; dark zones, low porosity). The grey patch in the marked white circle region is a macroscopic fossil. The image field of view (FOV) was 17 cm. The total acquisition time was 40 min. structures and improve our knowledge of the spatial scale and extent of heterogeneity in the formation. Some MR1 images for sedimentary rocks are illustrated. An M R image of a brine-saturated limestone sample is shown in Figure 8. The Tyndall Limestone rock is from a Saskatchewan natural gas field. This image was extracted from a 3D Spiral-SPRITE dataset. In the image the bright zones represent highporosity regions, while the dark zones represent low-porosity regions. The image shows an obvious dual porosity system. The grey patch in the marked white circle region is a macroscopic fossil. A cut-away view of a 3D image of a brinesaturated sandstone is shown in Figure 9. The diameter of the cylindrical core was 2.5 cm. The image shows that the shale bedding architecture traverses this sample. The existence of fractures and their distribution are very important to hydrocarbon reservoir evaluation, and have a significant effect on multiphase flow in a reservoir. A fracture distribution in a hydrocarbon reservoir limestone core is shown in Figure 10. The slice was selected from a 3D M R I dataset. The diameter of the cylindrical full-size core was 7 cm. Examination of the inhomogeneous nature of sedimentary rocks is very important to formation

MRI METHODS FOR CORE ANALYSIS

203

suited to laboratory studies of fluid dynamics in porous media. We illustrate this potential through a brief discussion of recent experiments examining the imbibition flow mechanism. The dynamic processes of fluid flow in rocks can be determined by M R I non-intrusively and non-destructively. A study of fluid flow in rock is illustrated in the next section.

Spontaneous imbibition with M R I

Fig. 9. An MR image of a brine-saturated sandstone core sample. The particular image displayed was selected from a 3D dataset. The cut-away section reveals a shale bedding architecture traversing the sample. evaluation. However, the dynamic processes of fluid flow are even more important, since the ultimate motivation of formation evaluation is often to determine the dynamic characteristics of fluid flow in the formation. M R I is a laboratory measurement, which, because it is noninvasive and non-destructive, seems ideally

Fig. 10. An MR image of a brine-saturated hydrocarbon reservoir limestone core. A slice was selected from a 3D MRI dataset. The diameter of the cylindrical core was 7 cm. The image shows a fracture system distribution.

Spontaneous imbibition is a process by which a non-wetting fluid displaces a wetting fluid in a porous medium by capillarity. Co-current and counter-current imbibition are defined as wetting and non-wetting fluid flow in the same and opposite directions, respectively. The imbibition phenomenon has a significant effect on oil recovery from hydrocarbon reservoirs, especially for fractured reservoirs. The processes of spontaneous co-current and counter-current imbibition were monitored by 3D Conical-SPRITE M R I (Chen et al. 2003).

Experiments Berea sandstone core was used for the experiments. Its porosity was 18.6% and its permeability 0.18 Itm2. The cylindrical sample was 25 mm in diameter and 52 mm in length. Spontaneous imbibition experiments were carried out with distilled water. For co-current imbibition, the bottom of the rock sample was kept in touch with a bulk water reservoir, air in the core was displaced by water due to capillary action. For countercurrent imbibition, the whole sample was immersed in a distilled water reservoir. The water was drawn into the centre of the core from the surface by capillary forces. At different saturation states, 3D Conical-SPRITE images were measured with the sample sealed by Teflon tape to avoid evaporation. The sample was repetitively dried, and the procedure repeated several times at different saturation states. For Conical-SPRITE imaging the imaging matrix was 64 • 64 x 64, the field of view (FOV) was 7 • 7 • 7 cm, the flip angle was 13 ~ a phase-encoding time (tp) of 40 Its and a repetition time (TR) of 2ms, with 39 discrete cones sampled for a single-scan imaging time of only 2.5 min.

Co-current imbibition The & situ water distribution during spontaneous co-current imbibition is shown in the image series of Figure 11. The broad bright base in the series

204

Q. CHEN ET AL.

Fig. 11. The dynamic processes of co-current imbibition of the Berea sandstone monitored by MRI. A series of images (a)-(tl) show 2D longitudinal slices extracted from 3D Conical-SPRITE MRI datasets. The time interval between the successive images was 10.5 min. The images indicate a piston-like water-penetrating process. The flat base in the images is a water reservoir, the weak signal at left of the images is background signal from a Plexiglass component in the RF probe head. Field of view (FOV) was 7 x 7 cm.

images is a water reservoir. T h e water distrib u t i o n s in the Berea s a n d s t o n e core exhibit a r e c t a n g u l a r shape, a n d s h o w a piston-like water f r o n t m o v e m e n t . B e h i n d the a d v a n c i n g waterfront, no f u r t h e r increase o f water c o n t e n t was a p p a r e n t . W a t e r flows with a piston-like p a t t e r n d u r i n g c o - c u r r e n t imbibition.

Counter-current imbibition F i g u r e 12 shows 2 D slices selected f r o m a 3D C o n i c a l - S P R I T E d a t a s e t after i m b i b i t i o n for 8s. T h e overall water s a t u r a t i o n was 26.3%. T h e water p e n e t r a t e d towards, b u t h a d n o t yet reached, the centre o f the core in this stage. A

Fig. 12. (a) Longitudinal and (b) transverse 2D slices extracted from a 3D Conical-SPRITE MRI dataset of Berea sandstone core during counter-current imbibition for 8 s. The average water saturation was 26.3%. The penetrating waterfronts have not yet reached the centre of the sample. The FOV was 7 x 7 cm, and slice thickness was 1 mm.

MRI METHODS FOR CORE ANALYSIS

205

Fig. 13. (a) Longitudinal and (b) transverse 2D slices extracted from a 3D Conical-SPRITE MRI dataset of Berea sandstone core after 15 s of counter-current imbibition. The overall water saturation was 30.9%. The penetrating waterfronts have just merged, and a homogeneous water-content distribution is observed in the images. The FOV was 7 x 7 cm, with a display slice thickness of 1mm. uniform water-content distribution behind the waterfront was observed. Figure 13 shows 2D slices extracted from a 3D Conical-SPRITE dataset after imbibition for 15s with an overall water saturation of 30.9~ The penetrating waterfronts have just reached and merged at the centre of the sample. The images show a uniform water-content distribution. The 3D Conical-SPRITE MRI results indicate that water content increases globally after the penetrating waterfronts have merged. These results reveal a film-type penetration mechanism for counter-current imbibition. There are two main mechanisms for watersaturation evolution during counter-current imbibition, i.e. a water-film-front advance mechanism and water-film-thickening mechanism. Before waterfronts reach the centre part of the sample, water penetrates along the corner and surfaces of the pore network where capillary pressure is higher, while air flows along the centre of the pore space. After the waterfronts merge at the centre region of the sample, water flows continuously with a film-type mechanism making the pre-existing water film thicker. This results in a more homogeneous water distribution with an increase of water saturation during countercurrent imbibition.

Conclusion The majority of sedimentary rocks have paramagnetic impurities, which result in M R transverse signal lifetimes that are too short to be detected by clinical MRI methods. Even for rocks that may have longer MR signal lifetimes, it is still very difficult for traditional MRI methods to obtain a quantitative image. Traditional

methods are based on T2 contrast, and the T2 decay curve is very complex in rocks. Traditional MRI methods also suffer from image distortion due to susceptibility variations, chemical shift and paramagnetic impurities, rendering them non-quantitative. Quantitative information is the ultimate goal for rock-core analysis. The SPRITE imaging technique has proven, over the last 8 years, to be a very robust and flexible method for the study of a wide range of systems with short signal lifetimes. In particular, these techniques appear near ideal for direct fluid-content imaging in porous media. As a pure phase-encoding technique, SPRITE is largely immune to image distortion. It enables the systems with T~ lifetimes as short as tens of microseconds to be successfully visualized. The SPRITE imaging methodologies are still rapidly developing (Halse et al. 2004). Our experimental results show that most sedimentary rocks have a single exponential T~ decay, which ensure that quantitative fluid-density images can be easily obtained by exponential fitting. MRI measurements on test cores described in this review reveal the internal sedimentary structure, fossils, dual porosity system and fracture distribution in carbonate rocks, as well as shale bedding structure in sandstone rocks. All of these inhomogeneities significantly influence the dynamics of fluids in the media. Dynamic water imbibition into air-filled Berea sandstone was studied using the ConicalSPRITE MRI technique. These measurements provide direct evidence for differences in the pore-filling mechanisms for co-current imbibition and counter-current imbibition in Berea sandstone. A piston-like mechanism for cocurrent imbibition and a film-like mechnism for counter-current imbibition have been observed by MRI.

206

Q. CHEN ET AL.

B.J. Balcom thanks NSERC of Canada for operating and equipment grants. B.J. Balcom also thanks the Canada Chairs programme for a Research Chair in MRI of Materials (2002-2009). The UNB MRI Centre is supported by an NSERC Major Facilities Access grant. B.J. Balcom and Q. Chen thank the Atlantic Innovation Fund for project support. We also thank B. MacMillan for his help, and R.P. MacGregor for his technical assistance.

References BALCOM,B.J. 1998. SPRITE imaging of short relaxation time nuclei. In: BLUEMLER,P., BLUEMICH,B., BoTro, R. & FUKUSHIMA,E. (eds) Spatially Resolved Magnetic Resonance." Methods. Materials, Medicine, Biology, Rheology, Geology, Ecology, Hardware. Wiley/VCH, Weinheim, Germany. BALCOM, B.J., BARRITA, J.C., CHOI, C., BEYEA, S.D., GOODYEAR, D.J. & BREMNER, T.W. 2003. Singlepoint magnetic resonance imaging (MRI) of cement based materials. Materials and Structures, 36, 166-182. BALCOM,B.J., MACGREGOR, R.P., BEYEA,S.D., GREEN, D.P., ARMSTRONG, R.L. & BREMNER, T.W. 1996. Single Point Ramped Imaging with Ti Enhancement (SPRITE). Journal of Magnetic Resonance, Series A, 123, 131-134. BELTON, P.S., HILL, B.P. & RAIMBAUD,E.R. 1988. The effects of morphology and exchange on proton NMR relaxation in agarose gels. Molecular Physics, 63, 825-842. BLACKBAND,S., MANSFIELD,P., BARNES,J.R., CLAGUE, A.D.H. & RICE, S.A. 1986. Discrimination of crude oil and water in sand and bore cores using MR imaging. Society of Petroleum Engineers Formation Evaluation, l, 31-34. BLOCH, F., HANSON, W.W. & PACKARD, M. 1946. Nuclear induction. Physical Review, 69, 127. BLOEMBERGEN, N., PURCELL, E.M. & POUND, R.V. 1948. Relaxation effects in nuclear magnetic resonance absorption. Physical Review, 73, 679-712. BOVEY, F.A. & MIRAU, P.A. 1996. NMR of Polymers. Academic Press, New York. BROWN, R.J.S. & FATT, I. 1956. Measurement of fractional wettability of oil field rock by the nuclear magnetic relaxation method. Petroleum Transactions, American Institute of Mining, Metallurgical and Petroleum Engineers, 207, 262-264. BROWNSXEIN,K.R. & TARR, C.E. 1979. Importance of classical diffusion in NMR studies of water in biological cells. Physical Review A, 19, 2446-2453. BULTER, J.P., REEDS,J.A. & DAWSON, S.V. 1981. Estimating solution of first kind integral equations with nonnegative constraints and optimal smoothing. SIAM Journal on Numerical Analysis, 18, 381-397. CALLAGHAN,P.T. 1991. Principles of Nuclear Magnetic Resonance Microscopy. Clarendon Press, Oxford. CANET, D. 1996. Nuclear Magnetic Resonance, Concepts and Methods. Wiley, Chichester. CARR, H.Y. & PURCELL,E.M. 1954. Effect of diffusion on free precession in nuclear magnetic resonance experiments. Physical Review, 94, 630-638.

CHEN, Q. & KINZELBACH,W. 2002. An N M R study of single- and two-phase flow in fault gauge filled fractures. Journal of Hydrology, 259, 236-245. C~tEN, Q. & SONG, Y.-Q. 2002. What is the shape of pores in natural rocks? Journal of Chemical Physics, 116, 8247-8250. CHEN, Q., GINGRAS, M.K. & BALCOM, B.J. 2003. A magnetic resonance study of pore filling processes during spontaneous imbibition in Berea sandstone. Journal of Chemical Physics, 119, 9609-9616. CHEN, Q., KINZELBACH, W. & OSWALD, S. 2002a. Nuclear magnetic resonance for studies of flow and transport in porous media. Journal of Environmental Quality, 31,477-486. CHEN, Q., KINZELBACH,W., YE, C. & YUE, Y. 2002b. Variations of permeability and pore size distribution of porous media with pressure. Journal of Environmental Quality, 31, 500-505. CHEN, Q., WANG, W. & CAI, X. 1996. Application of NMR imaging to steam foam flooding in porous media. Magnetic Resonance Imaging, 14, 949-950. COHEN, M.H. & MENDELSON, K.S. 1982. Nuclear magnetic resonance and the internal geometry of sedimentary rocks. Journal of Applied Physics, 53, 1127-1135. EMID, S. & CREYGHTON,J.H.N. 1985. High resolution NMR in solids. Physica, 128B, 81-83. FUKUSHIMA, E. & ROEDER S.B.W. 1981. Experimental Pulse NMR - A Nuts and Bolts Approach. Addison-Wesley, Reading, MA. GINGRAS,M.K., MACMILLAN,B. & BALCOM,B.J. 2002a. Visualizing the internal physical characteristics of carbonate sediments with magnetic resonance imaging and petrography. Bulletin of Canadian Petroleum Geology, 50, 363-369. GINGRAS, M.K., MACMILLAN, B., BALCOM, B.J., SAUNDERS,T. & PEMBERTON,S.G. 2002b. Using magnetic resonance imaging and petrographic techniques to understand the textural attributes and porosity distribution in Macarinichnus-burrowed sandstone. Journal of Sedimentary Research, 72, 552-558. GRAVINA, S. & CORY, D.G. 1994. Sensitivity and resolution of constant time imaging. Journal of Magnetic Resonance, Series B, 104, 53~51. GUILEOYLE, D.N., MANSFIELD,P. & PACKER, K. 1992. Fluid flow measurement in porous media by echoplanar imaging. Journal of Magnetic Resonance, 97, 342 358. HAHN, E.L. 1950. Spin echoes. Physical Review, 80, 580-594. HALSE,M., GOODYEAR,D.J., MACMILLAN,B., SZOMOLANVI, P., MATHESON,D. & BALCOM,B.J. 2003. Centric scan SPRITE magnetic resonance imaging. Journal of Magnetic Resonance, 165, 219-229. HALSE, M., RIOUX,J. ETAL. 2004. Centric scan SPRITE magnetic resonance imaging: optimization of SNR, resolution and relaxation time mapping. Journal of Magnetic Resonance, 169, 102-117. HENNEL, J.W. & KLINOWSKI,J. 1993. Fundamentals of Nuclear Magnetic Resonance. Longman, Harlow, Essex. KENYON, W.E. 1992. Nuclear magnetic resonance as a petrophysical measurement. Nuclear Geophysics, 6, 153-171.

MRI METHODS FOR CORE ANALYSIS KLEINBERG,R.L. & HORSFIELD,M.A. 1990. Transverse relaxation progresses in porous sedimentary rock. Journal of Magnetic Resonance, 88, 9-19. KLEINBERG, R.L., KENYON,W.E. & MITRA, P.P. 1994. Mechanism of NMR relaxation of fluids in rock. Journal of Magnetic Resonance, Series A, 108, 206-214. KORRINGA, J., SEEVERS,D.O. & TORREY, H.C. 1962. Theory of spin pumping and relaxation in systems with a low concentration of electron centers. Physical Review, 127, 1143-1150. KUTSOVSKY, Y.E., SCRIVEN, L.E., DAVIS, H.T. & HAMMER, B.E. 1996. NMR imaging of velocity profiles and velocity distributions in bead packs. Physics of Fluids, 8, 863-871. LATOUR, L.L., KLEINBERG,R.L. & SEZG1NER,A. 1992. Nuclear magnetic resonance properties of rocks at elevated temperatures. Journal of Colloid and Interface Science, 150, 535-548. LAUTERBUR,P.C. 1973. Imaging formation by induced local interaction. Examples employing nuclear magnetic resonance. Nature, 242, 190-191. LUCAS,A.J., PEYRON,M. erAL. 1993. A Rigorous Evaluation of the Experimental Requirementsfor Perjorming Quantitative Porosity Measurements by NMR. Society of Petroleum Engineers, SPE-2742.

207

MANSFIELD,P. & ISSA, B. 1994. Study of fluid in porous rocks by echo-planar MRI. Magnetic Resonance Imaging, 12, 275 278. MANSFIELD,P. • ISSA,B. 1996. A microscopic model of fluid transport in porous rocks. Magnetic Resonance Imaging, 14, 711-714. MEIBOOM, S. & GILL, D. 1958. Modified spin-echo method for measuring nuclear relaxation times. Review of Scientific Instruments, 29, 688-689. OSMENT, P.A., PACKER, K.J. ET AL. 1990. NMR imaging of fluids in porous solids. Philosophical Transactions of the Royal Society of London, Series A, 333, 441-452. PURCELL, E.C., TORRY, H.C. & POUND, R.V. 1946. Resonance absorption by nuclear magnetic moments in a solid. Physical Review, 69, 37-38. RACK, F.R., BALCOM, B.J., MACGREGOR, R.P. & ARMSTRONG,R.L. 1997. Magnetic resonance imaging of the Lake Agassiz-Lake Winnipeg transition. Journal of Paleolimnology, 19, 255-264. SANDERS, J.K. & HUNTER, B.K. 1989. Modern NMR Spectroscopy. Oxford University Press, Oxford. STARK, D.D. & BRADLEY,W.G. 1988. Magnetic Resonance Imaging. C.V. Mosby, St Louis, MO. VINEGAR, H.J. 1986. X-ray CT and NMR imaging of rocks. Journal of Petroleum Technology, 38, 257-259.

Rapid non-contacting resistivity logging of core P. D. J A C K S O N 1, M. A. L O V E L L 2, J. A. R O B E R T S 3, P. J. S C H U L T H E I S S 3, D. G U N N l, R. C. F L I N T 1'4, A. W O O D 5, R. H O L M E S 6 & T. F R E D E R I C H S

7

1British Geological Survey, Nottingham NG12 5GG, UK (e-mail: [email protected]) 2Department of Geology, University of Leicester LE1 7RH, UK 3Geotek Limited, 3 Faraday Close, Daventrv, Northampton N N l l 5RD, UK 4present address." Department of Aeronautical & Automotive Engineering, Loughborough University LEll 3TU, UK 5Adrian Wood Associates, Danehill, Brookhill Road, Copthorne, Crawley RHIO 3PS, UK 6British Geological Survey, Edinburgh EH9 3LA, UK 7Department of Geosciences, University of Bremen, P.O. Box 330 440, D-28334 Bremen, Germany Abstract: We demonstrate a non-contact approach to whole-core and split-core resistivity measurements, imaging a 15 mm-thick, dipping, conductive layer, producing a continuous log of the whole core and enabling the development of a framework to allow representative plugs to be taken, for example. Applications include mapping subtle changes in grain fabric (e.g. grain shape) caused by variable sedimentation rates, for example, as well as the wellknown dependencies on porosity and water saturation. The method operates at relatively low frequencies (i.e. low induction numbers), needing highly sensitive coil pairs to provide resistivity measurements at the desired resolution. A four-coil arrangement of two pairs of transmitter and receiver coils is used to stabilize the measurement. One 'coil pair' acts as a control, enabling the effects of local environmental variations, which can be considerable, to be removed from the measurement at source. Comparing our non-contact approach and independent traditional 'galvanic' resistivity measurements indicates that the non-contact measurements are directly proportional to the reciprocal of the sample resistivity (i.e. conductivity). The depth of investigation is discussed in terms of both theory and practical measurements, and the response of the technique to a variety of synthetic 'structures' is presented. We demonstrate the potential of the technique for rapid electrical imaging of core and present a whole-core image of a dipping layer with azimuthal discrimination at a resolution of the order of 10mm. Consequently, the technique could be used to investigate different depths within the core, in agreement with theoretical predictions.

Electrical resistivity is used as a primary method for reservoir characterization, being related to porosity, porosity style and the nature of the fluids in the pore space (e.g. oil, water or gas). Where sediment is fully saturated the resistivity changes will reflect changes in porosity and fabric, whereas changes in pore fluids may be dominant over these. Traditionally, electrical resistivity measurements on core have used the galvanic approach where metal electrodes are attached to plugs of rock subsampled from full or half-round core. These galvanic resistivity measurements on core have benefits of potentially very-fine resolving power and an independent depth of investigation (e.g. Jackson et al. 1995; Lovell et al. 1995). Their disadvantages stem from the need for direct electrical contact between electrodes and core sample, and electrode polarization and contamination, for example.

While cross-contamination between soft sediment cores is a major issue in ocean research (e.g. in the Ocean Drilling Program), non-contact resistivity imaging at the well site was identified as needed in a study of research and development requirements in core analysis in the Petroleum Industry (Department of Energy 1991).

Electrical induction logging of core Following the successful introduction of galvanic electrical resistivity logging of boreholes in the 1930s, induction logging became established in the 1950s (e.g. Doll 1949) as it was capable of operation in oil-based muds. Non-contact electromagnetic (EM) conductivity measurement of the terrestrial land surface was introduced in the 1980s (McNeil 1980), proving to be a rapid

From: ROTHWELL,R.G. 2006. New Techniquesin SedimentCoreAnalysis. Geological Society, London, Special Publications, 267, 209-217. 0305-8719/06/$15.00 @, The Geological Society of London 2006.

210

P. D. JACKSON E T A L .

Fig. 1. Principle of the electromagnetic resistivity measurement at low induction number (skin depth >> coil spacing); transmitter (Tx) and receiver (Rx) coils aligned to create vertical magnetic dipoles (left) and horizontal magnetic dipoles (right) (after McNeil 1980). highly successful mapping technique. Moving a pair of coils at fixed separation over the surface of interest, we have adopted McNeil's approach to non-contact resistivity measurements on core samples. The method is summarized in Figures 1 and 2, where the cumulative response curves and the associated two different coil orientations can be seen. For a transmitter-receiver coil pair placed at the surface of a conducting 'Earth', if

the frequency is low enough, the traditional ~skin depth' (the tendency of alternating current to flow near the surface of a conductor) becomes large compared to the coil spacing, and the mutual induction of the secondary currents can be neglected (McNeil 1979) in what is called the 'low induction number' criterion. Here the magnetic field, caused by the electric currents induced in the 'Earth' is 90 ~ out of phase with the primary

Fig. 2. Cumulative responses from which theoretical measurements can be calculated for both horizontal dipole (lower curve, red) and vertical dipole (upper curve, blue) (after McNeil 1980).

RAPID NON-CONTACT RESISTIVITY CORE LOGGING

211

Fig. 3. Photograph of single pair of coils in the test rig.

Fig. 5. Relationship between transmitter and receiver signals. range of the core measurements to be substantially improved. Figure 5 illustrates the relationship between the signals obtained at a transmitter and received coil pair, showing the 90 ~ phase shift and the greater noise on the far loweramplitude received signal.

Calibration Fig. 4. Photograph of twin-paired coils. magnetic field exciting the transmitter, and is inversely proportional to the ground resistivity. In addition, the current flowing in each layer within a horizontally layered Earth would be independent of one another, while the contribution of each layer to the measurement is the inverse of the product of the layer's conductivity and the difference in cumulative response function between its upper and lower bounds (Fig. 2).

Design To improve resolution, pairs of small coils were developed for use at low induction numbers near the surface of core samples with the plane of the coils horizontal (vertical magnetic dipoles), as shown in Figure 2. Initially, the sensors consisted of two 15mm-diameter coils placed 42mm apart within transparent plastic holders (Fig. 3). Subsequently, stability was increased using an additional pair of 'control' coils, situated 200mm away responding to air alone (Fig. 4). This simple device, unavailable for downhole or conventional terrestrial prospecting, has enabled both the sensitivity and

The performance of the system is verified in Figure 6 where the output measurements can be seen to be directly proportionately to fluid conductivity. Similarly, this non-contact EM method is compared with both traditional galvanic measurements and published values for NaCI solutions of differing concentrations (Fig. 7). These results confirm the performance of both the new non-contact method and conventional four-electrode galvanic measurements. The performance of the non-contact resistivity approach for core measurements was investigated using a suite of synthetic samples of known geometry and properties. Each synthetic core was constructed inside a thick-walled (10mm) 90 mm-diameter (inside diameter) Perspex tube. Using the vertical dipole arrangement of coils (Fig. 1), the measurement was relatively insensitive to the presence of the Perspex wall. The response to simulated resistive layers within a uniform host rock is shown in Figure 8, where the core is moved past the coil pair. This ability to map structure within a core is illustrated further in Figures 9 and 10, where a wedge and thin dipping layer are presented. Figure 9 shows an electrical image (made by mapping successive profiles along the core) for a resistive saturated sandstone surrounded by pore fluid.

212

P.D. JACKSON ET AL.

Fig. 6. Calibration curves for both vertical and horizontal dipole configurations. Note the sandstone core is cut with a sloping or dipping face to the left and a perpendicular face to the right. In Figure 10 a dipping conductive fracture within a saturated sandstone formation is simulated. The dipping layer case, being 15ram thick, demonstrates the potential of the method to assess structure almost three times finer than the coil separation.

Depth of investigation Separate to this investigation of the ability of the method to image features along the core, we investigated the depth of penetration of the signal into the core by the non-contact measurement. This is illustrated in Figure 11 for a range of material resistivities. Here a 100mm-diameter

Fig. 7. Comparison of galvanic and non-contact measurements of resistivity of salt solutions with published data (Carmichael 1982).

RAPID NON-CONTACT RESISTIVITY CORE LOGGING

213

Fig. 8. Measurement response to an insulating 'resistive' layer; the top figure shows increased layer thickness.

Fig. 9. Unwrapped image of the test arrangement: resistive saturated sandstone bounded by conductive water.

214

P. D. JACKSON E T AL.

Fig. 10. Unwrapped image of a simulated conductive fracture bounded by saturated sandstone.

Fig. 11. Response curves for varying water resistivities (see the text for a full description).

R A P I D N O N - C O N T A C T RESISTIVITY C O R E L O G G I N G

Fig. 12. Non-contact resistivity formation factors: cored off Morocco in 1072 m water depth.

Fig. 13. Non-contact resistivity formation factors: vibrocore samples from the upper slope west of the Hebrides. The magnetic susceptibility peak is due to a thin layer of magnetite-rich grains.

215

216

P. D. JACKSON ET AL.

Fig. 14. Non-contact resistivity formation factors: 'vibrocore' samples from the upper slope west of the Hebrides indicating possible slope instability, very irregular log responses, perhaps a debris flow overlain by homogeneous muds.

plastic tube, sealed at one end with the sensor coil pair situated beneath, was filled with varying depths of water for a range of fluid resistivities. The results show that 50% of the contribution to the measurement comes from the closest 10mm of material, demonstrating that the technique will respond to structure within the core rather than averaging values over a given depth interval.

Case examples Application of the non-contacting technique in scientific studies has recently been widely enabled through the integration of the non-contact methodology described here into the GEOTEK Multi-Sensor Core Logger (MSCL). Two different environments are described here. Figure 12 shows resistivity formation factors acquired on core recovered off the Moroccan coast at about 31~176 water depth 1072m. When dealing with discontinuous core sections it is important to be aware of end effects; here in this example this results in a significant loss of data at each

section limit due to a (small) air gap that was unavoidable between two sections measured successively. A second application is shown in Figures 13 and 14. Both samples are from the upper slope west of the Hebrides in areas of seabed with structured sidescan sonar backscatter indicating possible Holocene slope instability. Figure 13 shows core samples made predominantly of mud (57-72% particles <2 gm) with unconfined shear strengths averaging approximately 4 kPa. A peak of magnetic susceptibility just below approximately 2.5m is thought to originate from a thin layer of magnetite-rich rock granules. Figure 14 is mud (approximately 65% particles <2 gm) with average unconfined shear strengths of approximately 10 kPa in the upper part. This overlies a very irregular log response associated with a mud, which on the basis of the sidescan image is possibly a debris flow (approximate average 46% particles <2pm). This lower interval appears to be slightly more cohesive, with unconfined shear strengths averaging approximately 14 kPa.

RAPID NON-CONTACT RESISTIVITY CORE LOGGING

217

Conclusions

References

We have described a new non-contact m e t h o d for the rapid electrical logging of core. Experimental results on synthetic core and on saline solutions demonstrate the success of this approach, while results for core from offshore M o r o c c o and offshore N W Europe show the applicability in ocean-floor studies of saturated sediments. Wider application to h y d r o c a r b o n core has yet to be demonstrated, but where conductive paths exist in the rock it may yield useful additional core logs on which to base further sampling and special core analysis.

CARMICHAEL,R.S. 1982. Handbook of Physical Properties of Rocks. CRC Press, Boca Raton, FL. DEPARTMENTOF ENERGY. 1991. Study of Research and Development Requirements in Core Analysis.

The development of the non-contact resistivity method for use on core samples arose from a recommendation from the Department of Energy Study of Research and Development Requirements in Core Analysis (HMSO 1991). Research and development was co-funded by the DTI Offshore Supplies Office. Dr T. Frederichs of the University of Bremen is thanked for kindly providing the data for Figure 12. P.D. Jackson, D. Gunn and R. Holmes acknowledge this paper is published with the permission of the Director, BGS. Core collection was partly funded by the Western Frontiers Association during a British Geological Survey regional programme of sampling in the Rockall Trough in 2001.

Geoscience Report to the Department of Energy (OSO), October 1991. HMSO, London. DOLL, H.G. 1949. Introduction to induction logging and application to logging of wells drilled with oil base mud. Transactions of the American Institute of Mining and Metallurgical Engineers, 186, 148162. JACKSON, P.D., LOVELL, M.A. ET AL. 1995. Electrical core imaging I: A new technology for high resolution investigation of petrophysical properties. Scientific Drilling, 5, 139-151. LOVELL, M.A., HARVEY, P.K. ET AL. 1995. Electrical core imaging II: investigation of fabric and fluid flow characteristics. Scientific Drilling, 5, 153-164. MCNEIL, J.D. 1979. Interpretive aids for use with electromagnetic (non-contacting) ground resistivity mapping. In: Forty-first European Association of Exploration Geophysicists Meeting, Hamburg, June 1979. GEONICS Limited, Ontario, Canada. MCNEIL, J.D. 1980. Electromagnetic Terrain Conductivity Measurement at Low Induction Numbers. GEONICS Limited, Ontario, Canada, Technical Note, TN-6.

Logging-while-coring - new technology for the simultaneous recovery of downhole cores and geophysical measurements D. G O L D B E R G

l, G. M Y E R S l, G. I T U R R I N O 1, K. G R I G A R 2,

T. P E T T I G R E W 2'3 & S. M R O Z E W S K I 4'5

1Lamont-Doherty Earth Observatory, Borehole Research Group, Rte 9W, Palisades, N Y 10964, USA (e-mail: [email protected]) 2Texas A & M University, Ocean Drilling Program, 1000 Discovery Drive, College Station, TX 75845, USA 3Present address: Stress Engineering Services, Inc., 13800 Westfair East Drive, Houston, TX 77041, USA 4Schlumberger Drilling and Measurements, 135 Rousseau Road, Youngsville, LA 70592, USA 5present address: Lamont-Doherty Earth Observatory, Borehole Research Group, Rte 9W, Palisades, N Y 10964, USA

Abstract: A newly developed logging-while-coring system was deployed during Ocean Drilling Program legs 204 and 209 off the coast of Oregon and near the Mid-Atlantic Ridge. The system consists of two existing devices modified to be used together - a Schlumberger Resistivity-atthe-Bit* tool, and a Texas A&M University wireline-retrieved core barrel and latching tool. The combination allows for precise core-log depth calibration and core orientation within a single borehole, and without a pipe trip. These tests, conducted in clay-bearing sediments (Leg 204) and in crustal peridotite and gabbroic rocks (Leg 209), mark the first simultaneous use of coring and logging-while-drilling technologies. Sediment cores were recovered with 33% recovery, on average, and as high as 68% to 75 m depth below the sea floor. Core recovery in crustal rocks was only 1-2%, however, penetrating to 21 m depth below sea floor, which is attributed to a problem with the core catcher. High-resolution logs were recorded in the downhole tool memory over the entire drilled intervals at both test sites. It is anticipated that logging-while-coring systems will be utilized more routinely where rig time constraints may otherwise preclude coring in difficult drilling environments.

Merging state-of-the-art wireline coring and logging-while-drilling (LWD) technologies provides two vital datasets without sacrificing time or adding risk associated with longer open hole times. Until now it has not been possible to continuously collect large diameter core and in situ logging data simultaneously. Logging-while-drilling and wireline logging measurements are typically made following coring in Ocean Drilling Program (ODP) holes. Continuous wirelineretrievable coring is routine in nearly all ODP drill holes, whereas industry coring programmes are often limited in key intervals due to time and cost constraints. The O D P drills holes up to 2000 m (c. 6500 ft) deep without a riser in water depths ranging from 300 to 6000m (c. 100020 000 ft). Sea water is utilized at high pressure to clear the hole of cuttings. Following the coring operations, the hole is logged with conventional wireline tools. In cases where drilling is difficult and wireline log quality is anticipated to be poor, L W D technologies are employed. A

dedicated L W D hole is often the only alternative to collect in situ log data in such difficult drilling environments. This new development of logging-while-coting technology achieves two primary objectives: 9 9

to reduce the time required to log after drilling and coring has been completed in a hole; to make in situ measurements using L W D over the same cored interval in a particular hole.

The system construction, testing and deployment were conducted jointly by the Ocean Drilling Program groups at the Lamont-Doherty Earth Observatory and Texas A&M University and by Schlumberger Drilling and Measurements.

System design and testing A schematic layout of the logging-while-coring system is depicted in Figure 1. To make this

From: ROTHVCELL,R.G. 2006. New Techniquesin Sediment Core Analysis. Geological Society, London, Special Publications, 267, 219-228. 0305-8719/06/$15.00 9 The Geological Society of London 2006.

220

D. GOLDBERG E T AL. A typical RAB-8 battery ordinarily occupies the annular space in the tool. The RAB-8 battery was redesigned to retain the annular space, allowing the M D C B to pass through. In addition, the standard O D P bit size is 25 cm (9 7/8 inches), considerably smaller than conventional bits used with the RAB-8 collar. A new resistivity button sleeve and slick stabilizer were fabricated to accommodate the ODP bit. The tool standoff from the borehole wall for the modified RAB tool is nominally 4.7ram (0.185 inches) for this O D P configuration. Following the fabrication of all required M D C B and RAB-8 parts, the logging-whilecoring system was assembled and tested at the Schlumberger Genesis testing facility in Sugar Land, TX (Fig. 2a). This ensured that the components mated properly and assembly could be accomplished in the field. A coring test through low-grade cement was also conducted using the Genesis rig at this location and successfully recovered core through the RAB-8 (Fig. 2b). Both tests were conducted prior to deployment of the system at sea during Ocean Drilling Program Leg 204 on Hydrate Ridge off the coast of Oregon.

Ocean Drilling Program tests

Fig. 1. The logging-while-coring system developed by ODP and Schlumberger. The motor-driven core barrel depicted in yellow passes through a modified Schlumberger RAB-8 collar to allow the acquisition of resistivity and gamma-ray data while collecting a full core. concept a reality, an existing core barrel was selected to fit through the throat of a modified Schlumberger Resistivity-at-the-Bit* (RAB-8*) tool (e.g. Lovell et al. 1995). Only the Motor Driven Core Barrel (MDCB) among the ODP's coring systems is sufficiently narrow to fit within the 87.5mm (3.45-inch) annulus of the RAB-8. Minor modifications of the MDCB were required to accommodate the tool length and latching mechanism.

The logging-while-coring system was deployed on the D/V J O I D E S Resolution for use on O D P Leg 204, offshore Oregon, in July 2002 (Bohrmann et al. 2003). The test was conducted in 777m (2550ft) water depth at the crest of southern Hydrate Ridge at O D P Site 1249 (Fig. 3). Drilling proceeded ahead to 30m (98.4ft) below sea floor, where coring operations began with sequential 4.5m, then 9m-long (15 and 30 ft) cores recovered through gas-hydrate-bearing clay-rich sediments to 74.9 m (245 ft) depth. Starting the coring operation below 30m (98.4 ft) depth allowed for more consistent control of weight-on-bit and torque during drilling in these shallow and soft sediments. A standard O D P 25 cm (9 7/8 inch) diameter four-cone bit was used and the rotation rate increased from 15 to 45rpm with depth. Average penetration rate was approximately 8 m h -1 (c. 25 ft h-l). Eight cores were recovered from Hole 1249B with 32.9% recovery, on average, through a 45m (147.5ft) interval. Cores recovered using plastic liners have a slightly narrower diameter (59 mm or 2.35 inch) than standard ODP cores, yet recovery as high as 67.8% was reached (Table 1). Note that over the same interval in an adjacent hole, the average recovery using standard O D P coring equipment was 77%.

LOGGING-WHILE-CORING

221

(a)

(b) Fig. 2. (a) Testing the logging-while-coring system at the Schlumberger Genesis rig site. (b) Successful Genesis rig test yielding first core through a RAB-8 tool. Two 9 m (30 ft) cores were also taken using the MDCB without liners and recovered 65mm (2.56 inch) diameter cores with up to 42.3% recovery. The cores were intact and processed normally through the ODP sediment laboratory. Without liners, however, the cores were extruded from the barrel, and further core handling and

processing of the disturbed sediments was limited. All eight cores were processed and archived normally on board the D/V JOIDES Resolution (Ocean Drilling Program 1999). Figure 4 illustrates the first core recovered from Hole 1249B prior to measurement and processing. Core

222

D. GOLDBERG E T AL. Table 1. Coring summary in ODP Hole 1249B

Fig. 3. Location map of the logging-while-coring experiments on legs 204 and 209.

measurements including density and magnetic susceptibility were made onboard the D/V J O I D E S Resolution using a multisensor track. Bulk density, porosity and grain density core measurements were made on discrete sample plugs. The occurrence of gas hydrates in the core material and their rapid dissociation precluded the measurement of natural gamma-ray activity, which requires longer measurements times.

Core #

Advance (m/ft)

1W 2A 3A 4A 5A 6A 7A 8A 9A Total

29.9 ' 9 8 . 4.5 ' 1 4 . 7 4.5 '14.7 4.5 ' 1 4 . 7 4.5 '14.7 4.5 '14.7 4.5 '14.7 9.0 '29.5 9.0 '29.5 45 '147.6

Recovery (m/ft)

% recovered

-

3.05/10.0 1.92/6.3 1.15/3.77 0.93/3.05 2.06/6.76 0.57/1.87 3.81/12.5 0.52/1.71 14.01/45.9

67.78 42.67 25.55 20.67 45.78 12.67 42.33 05.78 Average 32.9

High-resolution logs and image data were recorded in the downhole memory of the RAB8 tool over the entire 74.9m (245ft) drilled interval in Hole 1249B. The RAB-8 tool was also calibrated post-deployment in salt-water calibration tanks at Sugar Land, TX. The tool functioned properly during this test and the calibration showed that the field data are reliable. During Leg 209, one hole, Hole 1275C, was drilled using the RAB-8 at water depths twice that of the Leg 204 drilling targets (Kelemen et al. 2004). Typically, 2.0-7.0m (6.5-23ft) of core were recovered over this shallowest interval in other holes drilled through gabbro and peridotire using conventional coring systems during Leg 209. In particular, Hole 1275D was drilled with rotary coring technology and recovered 6.4 m (21 ft) of core over the uppermost 22.6m

Fig. 4. Core recovered using the logging-while-coring system onboard the Ocean Drilling Program drill ship, J O I D E S Resolution.

LOGGING-WHILE-CORING

223

p :~..o

E~'.~

',..~.,

=~ ~:~11 e..)

'-~

O~

o

0

o

<

;>-,

g~

~-~

o ~

. . ~ "~

224

D. GOLDBERG E T AL.

(74 ft) of penetration, yielding an average recovery of 28.2%. However, in Hole 1275C, just a few metres away, the RAB-8 coring system recovered only 17cm (6.7 inches) of core from 20.8m (68.2 ft) of penetration below the sea floor, yielding an average recovery of 1.3%. A problem with the one-way core valve opening, often referred to as a core catcher, was identified as the most likely reason for the poor recovery in this test. Variable control of weight-on-bit while coring over the

shallowest interval beneath the sea floor may also have contributed to the poor core recovery in this environment.

Results Figure 5 shows a summary of the primary core and drilling data acquired in Hole 1249B including resistivity images, and the resistivity and

Fig. 6. Downhole log data acquired using the GVR-6 and VDN tools from Hole 1249A, adjacent to Hole 1249B, as indicated. Depth measurements for time after bit, average bit rate of penetration and bit rotation (rpm) are also shown. Depth is noted in metres below the sea floor.

LOGGING-WHILE-CORING gamma-ray logs from the RAB-8. Core measurements of discrete samples from Hole 1249B are presented at discrete depths from 29.9 to 75.0 m (98-246 ft) below sea floor, as well as multisensor track core measurements. Core measurements have a depth accuracy of + 0 . 5 m (+l.64ft). As core recovery averages only 32.9% in this hole, depth matching between core and log measurements may be somewhat imprecise at specific depths. Ties are made using density, magnetic susceptibility and gamma-ray data, and, for example, all three measurements increase near 60m (146.8ft) below the sea floor, indicating a change in lithological content. Downhole drilling

225

parameters recorded during coring in Hole 1249B are also indicated in Figure 5. Hole 1249B was drilled to maintain a rate of penetration of 2 0 m h -1 (c. 6 5 f t h -1) over each cored interval. Weight-on-bit ranged widely, however, as it was difficult to control precisely in these shallow and soft sediments. The time after bit (of the RAB-8 measurements) varies due to the time required to drill and recover each core, and substantially more time than standard drilling or LWD operations is required. The difference between drilling ahead and coring time may introduce some uncertainly in the core to log depth correlation.

Fig. 7. A comparison of responses between the RAB-8 to the GVR-6 tools in adjacent holes. Side-by-side resistivity images and resistivity and gamma-ray logs are plotted from Hole 1249A and Hole 1249B. Depth is noted in metres below the sea floor.

226

D. GOLDBERG E T AL.

Core photographs of core 204-1249B-5A (43m or 141ft depth) indicate a gas-hydraterich core that largely dissociated creating a 'mousse'-like sediment fabric (Fig. 5). The reflective areas are an indication of where the gas hydrate existed. Core 204-1249B-6A (49m or 161 ft depth) indicates a change in the composition of the cored material. The core recovery in these materials is high given that the MDCB core barrel is designed for use in harder rocks. The MDCB system cuts core by rotation, filling the barrel slowly as the bit advances. A pistontype core barrel is more conducive to high recovery of low-strength materials. The MDCB core barrel will be modified in the future to shorten the core length and reduce friction as the core enters the barrel. These are important changes aimed at improving core recovery with this system. A comprehensive suite of LWD data was acquired in nearby Hole 1249A using GVR-6* and VDN* tools (Fig. 6). The lateral offset between holes 1249A and 1249B is 40m (131 ft). A difference of approximately 0.5m (1.64ft) in water depth exists between the two sites. The logs from Hole 1249A show the rate of penetration and time after bit curves are lower than in Hole 1249B and remain relatively constant for the drilled interval (Fig. 6).

The RAB-8 data collected in Hole 1249B are compared with GVR-6 data from nearby Hole 1249A in Figure 7, which shows important similarities and differences. The large increase in resistivity in the upper interval in both holes corresponds to the presence of gas and gas hydrate. Some variation in the images between the holes may be associated with the greater time after bit for the RAB-8 measurement (e.g. coring v. drilling operations). The gamma ray shows a linear trend with an offset that may be attributed to the difference in standoff between the RAB-8 and GVR-6 tools. In general, the image data in holes 1249A and 1249B correlate well, with differences due to environmental conditions and lateral variations in geological heterogeneity between the two sites. Resistivity images acquired from Leg 209 are depicted in Figure 8. The resolution of the images is high, with high dynamic range and full data recovery with no data spikes. Gamma data are measured for the purpose of depth correlation between logs and core, if available. Although no direct comparison with cores is possible due to the low core recovery, the images provide sufficient information to orient borehole features that are apparent in this highly resistive crystalline formation. While the coring result in this hole was disappointing, the image data in this environment

Fig. 8. Resistivity image data acquired during ODP Leg 209 with the logging-while-coring system. The depth interval shown penetrates space 8 m (26.3 ft) into crystalline oceanic rocks. The natural gamma-ray log (GR), shown as overlay curves, is recorded for the purpose of depth correlation between logs and core. Depth is noted in metres below the sea floor.

LOGGING-WHILE-CORING and the operational experience are valuable for further development of the technology.

Summary Geologists and geophysicists collect subsurface data in order to predict the location and characteristics of the formation, strata and structural features (e.g. fractures). In recent years, technology has advanced to allow the collection of geophysical and geological data as a well is being drilled. Such LWD measurements are typically made prior to or following coring in a separate borehole, if coring has been undertaken, because these technologies could not be deployed together. Two holes must be drilled: a first hole for obtaining core samples and a second hole for obtaining LWD data. Conventional wireline logging is typically undertaken in the same cored borehole, but only if hole conditions are good. In cases where drilling is expected to be difficult, wireline logging data may be acquired in a separate dedicated borehole, also requiring two holes to be drilled. The logging-whilecoring system offers the significant advantage of collecting core and log data simultaneously and over the same interval in a single borehole. Merging state-of-the-art wireline coring and LWD technologies provides these two vital datasets without the possible detrimental effect of holeto-hole offsets on the data, even if only several metres of lateral distance separate the drill holes. Correlation accuracy depends on the yield recovery of the core and sample/data match-up. Uncertainty in the correlation of core and log depths may cause significant errors. A common usage of core and logging data in geophysical applications is for stratigraphic correlation to a seismic section at a particular drilling site, where the tie to the seismic section depends strongly on the accuracy of the depth-to-time conversion. When the logging data and a core sample are obtained using the logging-while-coring system, their accurate depth correlation is assured. Furthermore, time requirements for the logging-while-coring system are the same as for coring operations alone. No additional time is required to log after drilling the core hole. That time, needed for refitting the drill string with the LWD assembly and to drill a second hole, adds to the overall duration of the on-site activities. The efficiency of using the logging-whilecoring system thus saves rig costs without adding the risk associated with borehole failure over longer open-hole exposure times. When continuous coring operations are required, as is

227

common in scientific drilling projects, borehole failure often occurs after the core hole is completed thereby restricting or eliminating the possibility of acquiring log data in that hole. The logging-while-coring system can increase the likelihood that logging data and core will both be successfully recovered at a particular drill site. The deployment of a new logging-while-coting system in ODP recorded resistivity and gammaray logs and resistivity images while simultaneously collecting core at two deep-water sites, each with different geological compositions in sedimentary and crystalline rock environments. Measurements on recovered core were correlated directly with log data over the same interval. LWD data from both conventional and whilecoring operations at a nearby site agree well. The experience of Leg 204 underscores the utility and importance of the logging-while-coring system in sedimentary environments, while making apparent several areas for technical improvements. The experience of Leg 209 suggests that hard-rock drilling with the logging-while-coring system has promise, although it requires design improvements for coring in this environment and with coring in the interval directly below the sea floor in deep water. Looking ahead, the loggingwhile-coring system will be deployed in both soft- and hard-rock formations in the future using a core barrel designed specifically for each application. A. Janik (LDEO; now at EXXON Mobil) synthesized all the log and core data figures. We thank the Leg 204 and Leg 209 shipboard parties, Transocean Sedco Forex and crew, Schlumberger Sugar Land Product Center, and technical staff for their efforts to develop and deploy this new technology. This project was supported and funded by the National Science Foundation and the Department of Energy/National Energy Technology Laboratory. This paper was expanded and adapted from an article published in Petrophysics (2004). Lamont-Doherty Earth Observatory contribution number 6706.

References BOHRMANN,G., TR~HU,A.M. ETAL. 2003. Proceedings of the Ocean Drilling Program, Initial Reports, 204 (on line). Available from the World Wide Web: http:// www-odp.tamu.edu/publications/204_IR/204ir.htm. GOLDBERG,D., MYERS,G., ITURRINO,G., GRIGAR,K., PETTIGREW, T., MROZEWSKI,S. & ODP LEG 209 SHIPBOARD SCIENTIFIC PARTY. 2004. Loggingwhile-coring: First tests of a new technology for scientific drilling. Petrophysics, 45, (4), 328-334. KELEMEN,P., KIKAWA,E. ET AL. 2004. Proceedings of the Ocean Drilling Program, Initial Reports, 209 (on line).

228

D. GOLDBERG ET AL.

Available from the World Wide Web: http://wwwodp.tamu.edu/publications/204_IR/204ir.htm. LOVELL,J.R., YOUNG, R.A., ROSTHAL,R.A., BUFFINGTON, L. & ARCEr,~AUX, C.L., JR. 1995. Structural interpretation of resistivity-at-the-bit images. In: Transactions of the Society of Petrophysics and

Well Log Analysts, 36th Annual Logging Symposium. Society of Petrophysics and Well Log Analysts Inc., Houston, Texas, USA, paper TT. OCEAN DRILLING PROGRAM. Publications Policy Appendix B. Available from the World Wide Web: http://www-odp/tamu.edu/publications/policy.html.

Integration of the stratigraphic aspects of very large sea-floor databases using information processing C H R I S J E N K I N S l, J I M F L O C K S 2 & M A R K K U L P 3 IINSTAAR, University of Colorado, 1560 30th St, Boulder, CO 80309-0450, USA (e-mail: chris.jenkins@ colorado.edu) 2 USGS Center for Coastal & Watershed Studies, 600 Fourth Street South, St Petersburg, FL 33701, USA 3pIES, University of New Orleans, 2000 Lakeshore Drive, New Orleans, LA 70148, USA

Abstract: Information-processing methods are described that integrate the stratigraphic aspects of large and diverse collections of sea-floor sample data. They efficiently convert common types of sea-floor data into database and GIS (geographical information system) tables, visual core logs, stratigraphic fence diagrams and sophisticated stratigraphic statistics. The input data are held in structured documents, essentially written core logs that are particularly efficient to create from raw input datasets. Techniques are described that permit efficient construction of regional databases consisting of hundreds of cores. The sedimentological observations in each core are located by their downhole depths (metres below sea floor - mbsf) and also by a verbal term that describes the sample 'situation' - a special fraction of the sediment or position in the core. The main processing creates a separate output event for each instance of top, bottom and situation, assigning top-base mbsf values from numeric or, where possible, from word-based relative locational information such as 'core catcher' in reference to sampler device, and recovery or penetration length. The processing outputs represent the sub-bottom as a sparse matrix of over 20 sediment properties of interest, such as grain size, porosity and colour. They can be plotted in a range of core-log programs including an in-built facility that better suits the requirements of sea-floor data. Finally, a suite of stratigraphic statistics are computed, including volumetric grades, overburdens, thicknesses and degrees of layering.

The sea-floor database structure dbSEABED (Jenkins 1997, 2002) is an integration of multitudes of datasets that have been collected by marine surveys over the decades. It currently holds over one million attributed (sea-floor describing) sites worldwide, including over 400 000 such sites for US waters in what constitutes the USGS usSEABED project (Williams et al. 2003). Software behind dbSEABED is directed to maximizing the usefulness of this large data collection for research and society. The principal intention of the project is to extract as much useful information as possible from the data, which were collected in diverse geological settings, to widely varying standards, using different techniques and for various survey goals. Naturally, the extraction of information from the inputs is never complete. While most of the data pertains to surficial observations of the top 10cm of the sea floor (e.g. grabs, photographs, inspections), a significant part describes what lies below the surface of the sea floor, for instance from core and drillhole descriptions and electronic geophysical probes. As marine technological sophistication grows an increasing number of projects require

this kind of sub-bottom information, such as earth history research, naval acoustics, offshore resources, coastal replenishment, foundations engineering and contaminant studies. Because of the miscellaneous nature of the data, orthodox methods of databasing or visualization that depend on having highly disciplined data inputs have limited usefulness. Most commercial core-log programs assume standards of data content that are agreed in engineering industries (e.g. LAS for geophysical well-log data: CWLS 1998). In contrast, sea-floor datasets require extensive filtering before a standardization is reached. The software methods of that filtering as applied to stratigraphic aspects of sea-floor data are the subject of this paper. Although the methods are discussed in terms of dbSEABED, they will be applicable in other projects that seek to integrate sea-floor datasets.

The input data

Data content As in most other geological databases, the data in dbSEABED can be divided into four main

From: ROTHWELL,R.G. 2006. New Techniquesin Sediment Core Analysis. Geological Society, London, Special Publications, 267, 229-240. 0305-8719/06/$15.00 9 The Geological Society of London 2006.

230

C. JENKINS E T AL.

Table 1. Example of the written vibrocore core-log format of dbSEABED (fictitious data). The data are a mix of numerical analysis results, visual descriptions, sedimentary units and point data. Note that some described units may overlap (0~).05, 0-2.01), so that there may be two descriptions applying to a given depth in the core. Descriptive terms #1 dbSEABED are rendered as abbreviations for practical reasons o/readability, computability and distinguishing from metadata

1,SFS,Core,99.28378,-99.90885,2.268,Institute X,,,Vibrocorer, 5.59,,,8May2000,15:20, Louisiana ...... 4.62 0,SFS, < site > ,ProjId:CoreId:FileName = BSS:34:bss00 34.txt 1,LTH,,0.00,0.05,som/shl_frgs + dora/fsnd,,,planr- beds burws 1,LTH, ,0,2.01,, ,snd/90% + slt/10% + cly/0% +minr/shl_cntnt 1,COL .... gry 1,LTH ..................... 40 o~/biotrbn 1,LTH . . . . . . . lenticlr_bed/0% x_bed/0% 2,LTH,,0.05,0.10,som/shl_frgs + dom/sit 2,LTH,,0.10,0.15,som/shl_frgs + dom/slt 3,LTH,in burw,0.10,0.15,mud 2,LTH,,0.15,3.20,som/shl_frgs + dom/slt 2,TXR .... 5,55,40, ,3.6

Site Data

Site metadata Surficial unit description Broader surficial unit Broader surficial unit (colour description) Broader surficial unit (visual bioturbation intensity) inclnd_bed/0% hznl_lamin ,/100% Broader surficial unit (bedding type) + dom/fsnd ,,,planr- beds burws Subsurface unit description (material and bioturbation) + dom/fsnd ,,,planr- beds burws Subsurface unit description (material and bioturbation) Special situation in subsurface unit: description (material in bioturbation) Subsurface unit description + dom/fsnd ,,,sml- burws (material and bioturbation) Subsurface unit (gravel : sand :mud and median grain size)

sections: (i) positioning and site information, such as geographic co-ordinates, navigational accuracy, date/time of sampling, vessel and sampler type; (ii) sub-bottom locating information such as sample top and bottom depths (in m b s f metres below sea floor), and any special situation an observation is under (e.g. 'of acid residue', 'inside pebble', 'from core catcher'); (iii) descriptive parts of the data (e.g. lithology, colour, grain size, geotechnics); and (iv) other metadata (e.g. methodologies, known faults and doubts, personnel). The character of the positioning, descriptive and metadata (i, iii and iv) are dealt with in other publications. Section (ii) is the focus of this paper. Data arrangement

The data layout of an illustrative core is shown in Table 1. dbSEABED holds its data in 'Data Documents' which are ASCII, comma delimited and both human- and machine-readable. These documents constitute written core logs and are processed by dbSEABED programs to generate a set of tables that are capable of easy import into a wide range of client geographic information systems, core-log programs, mathematical processes and relational databases. The Data Documents are ideal for data legacy and, because they hold data more or less in its original form without manual reclassification, they leave the

possibility open of applying different processing methods in the future. Each document holds the data of an institution or industry sector, so that data that are similar in format, tradition and methodology are associated. Each separate instance of site/time/top/base/ situation is a data 'event'. All the data from such an event appear as one line in the final output tables, so that a multi-attribute three-dimensional sparse matrix of the sea-floor's top several metres is compiled in X-, Y- and Z-directions. At data storage level, dbSEABED handles downcore stratigraphic issues by associating each observation with three fields: (i) sub-bottom top in mbsf; (ii) sub-bottom base in mbsf; and (iii) situation provided as word-based descriptive data. There are also items of information at the site level that may help to constrain issues of sub-bottom location. The type of sampler used for instance, and its barrel length or bite are limitations on the sub-bottom depth. The observed penetration and recovery of a device are also constraints (but note possibility of overpenetration and partial recovery: Skinner & McCave 2003). It is important to note that the sea-floor data that are available for integration are always incomplete in some respects. Some will not have an explicit specification of sampler type, analysis methods or units of measurement. dbSEABED responds to this in three ways: (i)

INTEGRATION OF SEA-FLOOR DATABASES

231

Fig. 1. Schematic diagram showing issues facing integration of stratigraphic aspects of sea-floor datasets. (A) Description in terms of sedimentary units, the units based on various parameters not coinciding. The code 'Nn' is a Munsell designation for shades of grey. (B) Description in terms of mixed unit and point observations. GSM reports gravel:sand:mud percentages. The symbols represent shell remains. (C) A rendering of the integrated data as point observations. Note that data gaps are preserved. abort the processing of that observation (or site) if the problem is serious, meaning a greater than 5% impact on results; (ii) report the problem and/or uncertainty and still make outputs; or (iii) proceed under some reasonable default assumption that has little impact on results (5% rule) and report the assumption and/or uncertainty.

Unit v. point data types Most sea-floor core records contain a mixture of unit and point observations. Units are defined by a top-bottom range in mbsf, whereas point observations are at a single depth (Fig. 1). But also, unit observations are generally wordbased descriptions, while point observations tend to be numeric analysis results. In this respect sea-floor data are unlike downhole logging or oceanographic CTD (Conductivity Temperature Depth) profiles, which are machine-produced, standardized and abundantly populated series of numeric point measurements. In general, commercial core-log programs plot unit and point observation data separately, labelled contrastingly as 'lithological units' or 'integrative data', on the one hand, and as 'curves', on the other (CWLS 1998; Rockware Inc. 2004). A good characterization of a cored sequence should use both types of data together, particularly

since methods now exist to bring word-based and numeric attribute data into conformance (Jenkins 1997, 2002). Also, neither form of data is sufficient to describe a core by itself. Descriptions in terms of units suppress issues of variability in each unit, while point analyses under-sample the variations in the core unless they are abundant downcore. A combination of point and unit data types will give the best output result in terms of statistical data analyses and visual core logs fit for stratigraphic interpretation.

Sub-bottom location and situation The downcore location of an observation can be given as an absolute distance from the surface or core top, or as a relative location such as 'top of core' or 'below red layer'. Interpretation of the numeric mbsf is straightforward, but word descriptions of location can be parsed in only a small number of cases. Tractable cases include 'core catcher' and 'top third of core' when there is information on core recovery length and/or penetration. Otherwise, the word sub-bottom reference must be relegated to situation data. The 'situation' of an observation refers to its location, context and representativeness. It is usually conveyed by word, and it can be quite diverse, placing an observation in a sediment fraction (textural, physical and/or chemical), in

232

C. JENKINS E T AL.

a mesoscale feature of the core or at an uncertain top-base mbsf. Examples are: 'in nodule', 'of coarse fraction', 'for pore water', and for downcore location 'approx. 3mbsf' and 'just below red layer'. It is important that situation data continue to be associated with data through to the final outputs of an information-processing system to prevent observations on specific fractions being misinterpreted as applying to the bulk sediment.

Importing graphical core logs Downhole data, unless collected initially in a database format, is difficult and time consuming to import in quantity. Most of the data are available in reports and publications as graphical core logs. Common methods of importing these data include operators typing in or digitizing the unit depths and characters. To import data in large quantities we investigated two other techniques: (i) dictation; and (ii) over-drawing followed by batch processing of the drawing file. The first is often more efficient than typing as hands are free to measure the depth ranges of patterned unit boxes. The second is the most efficient and allows batches of similarly formatted core logs to be imported into a database at a time. Because efficient data import is so important to database construction, more information on the process is given in Appendices A and B.

The data processing General method The overall function of the software system is 'Information Processing' (the 'systematic performance of operations upon data such as handling, merging, sorting and computing': ATIS 2004), aided specifically by 'Data Mining' (a 'non-trivial process of identifying valid, novel, potentially useful and ultimately understandable patterns in data': Fayyed et al. 1994) and 'Language Parsing' (see Winston 1984, chap. 9). The Data Mining proceeds first by the discovery of relationships using special information-rich sections, followed by application of those relationships over the entire data collection (where possible). There is also a statistical process of calibration and validation of those relationships using information-rich parts of the data. In the relationships we include statistically established meanings of word-based terms such as 'core catcher' and 'shipek grab', which are added to a thesaurus that is employed in the Language Parser. The meanings are conveyed in terms of parameters

like penetration, footprint, sample quality. As more relationships are determined and applied, the amount of information that proceeds to output rises even as the amount of input data stays constant. The processing is performed by a computer program, which successively addresses each dataset, site and sample. The process responds according to the content and style of the data that are available at each site and sample event. It is important that the design of the processes optimizes the amount of data that contribute to outputs - while maintaining specified quality standards. Occasionally, this entails a departure from normal practices. For example, output of data are made even where an absolute mbsf cannot be calculated. Displaying such data as narrative footnotes on core logs is not standard practice. The Language Parsing is performed on structure language noun-phrases, a level of grammar that applies to sedimentological/technical wordbased descriptive data, especially when it is divided into fields based on theme (such as by sampler type, analysed phase, amount of recovery, etc.). Note that the parsing used in this paper is much simpler than the fuzzy set theory parsing made for sediment descriptions in dbSEABED (Jenkins 1997, 2002).

Specific methods Using sampler type. An initial indication of the stratigraphic extent of a sampling is provided by the type of sampler. Sampler details can appear in the data at the level of dataset, data subset or site. (When different samplers are used in tandem they are treated as two separate events.) In the processing the description of sampler type is converted to lower case and searched for key syllables or words, such as: 'piston' and 'core' or 'kullen' and 'core' for Kullenberg piston core; 'manip' and 'arm' for (submersible) manipulator arm. The focus on distinctive syllables, words, abbreviations and alternatives is needed because many survey datasets use diverse abbreviations, lower/upper case and punctuation to record sampler type. In one example - piston core - all these are encountered: 'Piston Core', 'Piston Corer', 'PistonCore', 'PistCore', 'Piston', 'P.Core', 'PistonC' and 'PC'. The search is based on a thesaurus that holds searched terms 'pist'+'core', 'pc', 'piston' + 'c' and others. If a word in the description is not accounted for (e.g. 'pstn'), then a warning is produced at run-time and put into the diagnostics report for fixing later. A sampler described as 'corer' is treated as highly ambiguous in outputs.

INTEGRATION OF SEA-FLOOR DATABASES A good match produces a senior synonym ('PistonCore') and proposes a sampler penetration and footprint where possible. Obviously, in the case of a piston core, barrel length and hence penetration are unknown. But it may be assumed initially, and unless there is contrary information, that the recovery starts at the surface, so an initial whole-core sub-bottom range for the entire core would be top = 0.0 mbsf and b a s e = ' u n k n o w n ' . On the other hand, for a ShipekGrab the assumed sampler sub-bottom range is top = 0.0 mbsf and base = 0.10 mbsf. Sampler footprint is treated similarly. The final outputs for sampler information consist of the senior synonym name with the penetration and footprint appended, as in: 'ShipekGrab [Pen = 0.05 FtptL :W = 0.3 : 0.15]'. The penetration and footprint values can be extracted by later dbSEABED modules, namely the core-log generator and seabed roughness estimators. Sampler information that is specified by survey may be superceded later in the processing of the dataset by information specified per site. Furthermore, the sub-bottom depth ranges based on sampler type may be superceded by the penetration and recovery lengths that were actually observed at the site, and perhaps also by top-base values on observed units and analyses in the core itself. In this way, the most specific and credible information goes to output.

Assigning sub-bottom locations. To specify the location of each event in a core the processing uses the top, base and situation data of the event. Specifically, it: (i) extracts the straightforward numeric mbsf depths; (ii) where possible, it transforms (parses) descriptive locations into numeric depths; and (iii) in the remaining cases it reports location only in terms of situation. The overwhelming number of observations have a top and base mbsf specified that is simply passed to output in process (i). If the top-base input data are not numeric (e.g. 'approx. 1 mbsf') they are added to the string situation data for possible parsing. The parsing first attempts to associate a metric depth to the descriptive situation data based on the composition of the description (e.g. 'approx 3mbsf'), or by association with core recovery or penetration lengths (e.g. in case of 'core catcher' or 'top third of core'). Terms in the description are recognized in the same way as samplers were recognized - with their attendent diverse abbreviations, cases, punctuations (e.g. 'CC', 'core catcher', 'catcher', 'fingers'). Some of these operations can only be done on a statistical basis, and then they introduce an uncertainty into outputs (Jenkins

233

submitted). In the cases where no top and base mbsf can be specified reliably, the described location is output entirely as situation data. Note that every change of situation triggers a separate output - even when the observation top and base mbsf values are the same as before. Thus, observations at top : base : situation = '0.3, 0.5' and '0.3, 0.5, top' result in separate outputs from dbSEABED, with the situation data reported verbatim to output tables. This caution, however, leads to subsequent problems because industry software is usually unable to allow for special analysis situations and treat all data as applying to the bulk cored materials.

Dealing with the uncertainties In any integrative project, the scope for uncertainty in the outputs needs to be considered. For the common parameters that are used to describe the sea floor Jenkins (submitted) has examined this issue in detail. A similar process may be used to consider the reliabilities of the stratigraphic operations performed by dbSEABED and described here. For example, when core recovery differs from penetration, the sub-bottom location of the observations made on the recovered core materials is not determined more accurately than the disparity between recovery and penetration. This applies also to the materials at core top and in a core catcher. It is very important, therefore, that core penetration be recorded during survey and that the data be carried through processing to final coreqog and stratigraphic compilation. Another important factor in uncertainty is the relationship of numeric and word-based information bearing on the sub-bottom location of an observation. Obviously, in this case numeric data (mbsf) will be the more precise and, as discussed earlier, in some cases the sub-bottom level cannot be deciphered from a relative wordbased description. In cases where a unit or point mbsf could not be assigned properly, dbSEABED programs output a diagnostics report that warns users. Opportunities then exist for reviewing the original data and either making corrections or decommissioning the data, with effect in the future on runs of the processing.

Stratigraphic outputs Once the information processing is complete, the sea-floor data are in an information-rich standardized format that is amenable to visualization, statistical analysis, relational databasing and use in numerical models. The outputs consist of

C. JENKINS ET AL.

234

tables that are a 'lowest common denominator' comma delimited ASCII format, capable of being imported into a wide range of commercial and research applications. The tables are ready for import into GIS (geographical information system) and RDB (Relational Database) and are somewhat normalized. The data contain positional, audit-trail and attribute information types: latitude, longitude, water depth, topbottom-situation, relational keys by dataset and event, sampler type, audit-trail data, and a host of sedimentary descriptors including texture, composition, colour, consolidation, grain types and features. Detailed metadata are closely associated.

Core-log generator Requirement. The ideal way to display and check the stratigraphic results after information processing is to output core logs in a single format to facilitate visual comparisons of results. Although specially formatted output files compatible with Rockware ':~; and other proprietary packages are generated in the course of processing, a core-log generator that could treat thousands of cores in one project and which could satisfy other special requirements was also implemented as a part of the processing. The same module performs error detection and volumetric (3D) statistical stratigraphic calculations across core datasets. The technical requirements that prove difficult for commercial core-log display packages include: 9

9

9 9

9 9

9

mixed unit and point observations need to be fully integrated and to contribute together to the visualizations and statistics; several conflicting values for a parameter may occur at levels in cores (see Table 1, 00.05 and 0-2.01 mbsf) and must be displayable; stratigraphic units that overlap or envelop each other must be treatable and displayable; provision for displaying uncertainties in downhole depths of units and on parameter values; a footnote to be presented when data could not be placed in terms of numeric mbsf; realistic-colour plotting of sediment colours, preferably directly from Munsell Codes; the logs should still be visually effective when the core has a sparse number of measurements or observations.

Many of these requirements suggest that graphical logs for sea-floor cores need to have a strong symbolic component based on point data.

Intensely graphical 'curve' traces are rarely applicable, needed only in instances where densely measured data exist as, for instance, with Multi-Sensor Core Logger outputs (e.g. Geotek Ltd 20O4).

Implementation.

The generator writes the core logs as Postscript ~: files, which has several benefits for the project: (i) the core logs can be examined individually, and can also be combined into poster-sized stratigraphic fence-diagram compilations involving virtually limitless numbers of cores; (ii) with minor changes to the program, different core-log designs can be produced now and in the future, responding to particular research applications and improvements in the dbSEABED information-processing sequence; (iii) uncertainties on both the plotted parameters and on the sub-bottom mbsf can be expressed; (iv) the core logs can be imported into many different applications (e.g. Adobe Illustrator ':K;') and can be superimposed on seismic profiles; (v) after transformation to an image format (usually TIF) the core logs in their user-specified design may be interactively linked or plotted in GIS and 3D visualization applications; and (vi) complex word-based descriptive data can be associated with the cores in various ways, including footnotes. Figure 2 illustrates one format of core log that is produced in this way. The outputs are orthodox, except perhaps in the emphasis on symbolic point-based rendering of the data. Symbols are easily introduced by the user specifying either a bitmap or a simple Postscript ~: drawing, and those being linked to the program. Sediment colour is rendered either as written Munsell codes, those codes rendered in their approximate colour or as coloured units (the last two using Postscript" HSV colour system). To achieve integration of unit and point data the core-log generator subsamples the units as sets of points and renders them alongside point analyses (Fig. 1C). This is necessary particularly in projects that require maximal integration of a parameter, such as sand content, from point analyses and unit descriptions. Because units are reduced to a curve for each parameter, plotting in this way also helps deal with the fact that units recognized by different parameters such as grain size and colour may not coincide in range through a core. It also allows conflicting data values to be shown (Fig. 1B, C) and for data that are not fully representative, because of their situation, to be plotted (brackets; Fig. 1C). The rendering of units in this way is performed subsequent to processing, at a scale suited to the plotted core log.

INTEGRATION OF SEA-FLOOR DATABASES

235

Fig. 2. Example of a computed core log. The core logs can be drawn to very largeformat Postscript | compilations, thereby enabling interpretation work by stratigraphers on poster-sized paper sheets or on a computer. That interpretation includes fence-diagram correlations, thickness measurements and isopach estimations. In some areas hundreds of cores may be viewed and worked on at once in a single presentation. Control over which core logs will be included and over the chart internal co-ordinates is exercised from a set-up file which may be composed in an interactive GIS environment such as ArcView | At a basic level, the stratigraphers' work proceeds by drawing on the work sheet but, alternatively, can proceed by drawing on the postscript directly using an application such as Adobe Illustrator ':R~.

Statistical stratigraphy During the process of compiling the core logs, statistical parameters of significant applied and research interest can be generated for a region or

locality. Because the calculations are implemented as small and easily programmable modules, there is unlimited choice of calculation methods. When the data for a core are judged to be too sparse the program may reject a calculation result as unreliable. The outputs are presented in either: (i) GISready formats representing the core collection areally; or (ii) as the more detailed 3D XYZ sub-bottom matrices. Specific calculations which have been implemented to date include: (i) mud overburden above grades of sand exceeding 50% sustained over I m thickness; the calculation is restricted to the top 10 m of the seabed; (ii) the depth-integrated grade (per cent) of sand in the top 1 m of the seabed; (iii) vertical variability of grain sizes in the top 4 m of the seabed, expressed as a standard deviation. Figure 3 illustrates an example of overburden calculations, displayed in a GIS in conjunction

236

C. J E N K I N S E T AL. d

O

o~-~

o~

'.~ r~

~

~

.

INTEGRATION OF SEA-FLOOR DATABASES with other mapping information for the area: bathymetry, the core distribution and surficial sediment patterns. The calculation o f d o w n c o r e variability (iii, above) warrants further comment. To maximize the a m o u n t of data available in each core and from the cores across an area, it is calculated using both point and unit values. Usually, variance methods assume a population of discrete samples (points). The stratigraphic variance was calculated as: V Z : (xit i - X T ) 2 / ( T ( r l

- l)///)

where xi and ti (i = 1,n) are the point or unit parameter values and thicknesses, X is the thickness-weighted parameter average ( ~ ( x i t i ) ) over the total thickness of observations T ( ~ , ti). T will usually be less than the full stratigraphic range of the observations. Point analyses are assigned a m i n i m u m t o f 0.05 m, in effect rating them as a unit of that thickness. Confidence on the variance may be tested using Chi-squared methods. The probability of the variance differing, by say, 10% is c o m p u t e d using the Chisquare probabilities Q and P (Press et al. 1989, chap. 6, 14). The higher this probability, the less reliable is the statistic. The variances can also be rejected when n is too small (approximately less than 8).

Conclusion W h e n large numbers of diverse and informationrich environmental datasets are integrated using information-processing techniques, significant new opportunities for research and application arise. In this case the 3D structure of the seabed over m a n y parameters is m a d e available for regions where considerable core information exists. In d b S E A B E D integrative m e t h o d s that work across mixed word and numeric data have been implemented for the stratigraphic sub-bottom aspects of sea-floor core data, leading not only to the mass production o f the useful core logs, but also to wide-area compilations of core logs that can be interpreted by stratigraphers. An added facility is the calculation o f stratigraphic statistics for multitudes o f cores at a time and 3D portrayals of the subbottom. I thank S.J. Williams of the USGS (Woods Hole) and I. Overeem of INSTAAR for assistance with datasets, technical guidance and review of the paper. USGS Coastal and Marine Geology division, Louisiana Department of Natural Resources (LDNR), University of New Orleans, ONR (award N-00014-01-0376),

237

NGDC (NOAA), INSTAAR and Duke University kindly provided funding for the work. Drs J. Harff and L. Poppe are thanked for their constructive reviews.

Appendix A. Efficient graphical data entry of core data Computer-aided drawing over scanned graphical core logs, followed by batch processing of the drawing files may be the most efficient method of entering stratigraphic data into databases. The drawing may be done in many software applications, but in this project the GIS MapInfo ~' is used because it provides appropriate co-ordinate systems and drawing export formats. The graphical core log is scanned to a TIF image, imported and geo-registered with core tops at y - - 0 and downcore depths scaled in metres. Numbers of core logs may be included in a single graphic, suitably arranged. The drawing is composed by: (i) typing information over the graphic (e.g. for general core information and notes); or (ii) drawing a line object representing the downhole ranges units and giving the lines an attribute (e.g. lithological character). Each major section of data is assigned a different colour and/or font in the drawing so the batch processing can distinguish its type. In the example (Fig. A1) general core information such as site name, position, elevation, date and sampler type was written in cyan/regular font; D10 grain size values in yellow/ bold-underlined; water contents in yellow/regular font. The line objects are also drawn in colours representing data type: lithological units were drawn red; consolidation units in magenta; colour units in green. For each line object a label was typed into the browser table, verbatim for the labels that appear on the core log, for instance: lithology 'CL SIS' (a USACE code), consolidation 'vSo' and colour 'dGr'. Objects can be drawn in any order and different data types can be interspersed. The drawing can be saved and can be worked on over a period of time or updated if necessary. At conclusion the whole drawing for each core is selected and exported to Mapinfo ~ Interchange format - ASCII MID/MIF file pairs. Batch processing proceeds using a dedicated program (in QuickBASIC ~) and the number of core logs that can be processed is limitless. The program scans for the general data, then lithology and other attributes, arranges them by units in stratigraphic order, and outputs to the dbSEABED and tabular spreadsheet formats. Errors and inconsistencies are reported, for instance out-of range sub-bottom depths, and line objects without labels. Once the dbSEABED data are available they are pasted into one of the structured documents and are ready as contributions to GIS mappings, volumetric visualizations, standardized core logs and numerical stratigraphic analysis.

238

C. JENKINS E T AL.

Fig. AI. Example of the efficient process of data entry: a drawn-over core log from USACE sand studies of Louisiana, USA. The overall vertical scale is validated in the process using the scaled black line; general core information at the top is typed over, as are D10 and water-content values to left of the lithological column. Lithological, consolidation and colour units (right of column) are line objects, with verbatim labels assigned in the associated brower data table.

Appendix B. Digitizing methods for core-log data For use in database, spreadsheet and GIS applications, the usual format of paper copies of core descriptions requires conversion to a digital format. A method for

converting the information from sediment core description sheets to spreadsheet format is described below (see also Flocks 2006). A core description sheet is laid out in a horizontal (x) dimension representing the attribute columns and a vertical 0') dimension representing depth in the

I N T E G R A T I O N OF SEA-FLOOR DATABASES

239

Fig. B1. Output-flow using sand per cent curve from a core description sheet. (A) Points digitized from profile; (B) raw x , y digitizer-output values; (C) conversion to depth v. sand per cent; and (D) resulting sand per cent plot.

core. The range in x that each attribute column occupies can be defined in an index table. The manual digitizing process (digitizing tablet or on-screen) reduces the information on the sheet to x, y values (Fig. B1) so that presence or magnitude of an attribute or feature are recorded v. depth (y). At this point in the process the type of data on the form is irrelevant. Subsequently, the index table is used to establish the link between the digitized value of x and the column or form attribute, which can be divided into four basic types: symbols, curves, sections and text. The index table contains the x-value, attribute name and type of each feature on the sheet. The type of attribute (symbol, curve, section or text) determines what routine will be used to process the data. For example, with curve data such as sand per cent (Fig. BI), the left edge of the column equals 0% presence and the right edge usually 100%. The percentage of sand per core depth interval is calculated in proportion to the x offset between the x-values of the 0% and 100% column edges. It is possible to interpolate values for downcore intervals between the digitized points, which are usually marked at changes of symbols and inflections of curves. Text and symbol descriptions require extra referencing, which can be assisted in

part by using a symbolic menu. The process converts all of the available information, graphic, symbolic and text, rather than just to stratigraphic units. Once numerous core description sheets have been digitized, the digital files are batch-processed to greatly increase conversion speed. Subsequently, graphical display of the data allows the user to compare the computed values to the actual core description sheet (Fig. B1). Finally, the output of the data can be customized to fit database requirements.

References ATIS. 2004. Telecom Glossary 2K. Alliance for Telecommunications Industry Solutions. (Web document: www.atis.org/tg2k/_information_processing. html; download date 2 April 2004.) CWLS. 1998. Log ASCII Standard (LAS) Software. Canadian Well Logging Society. (Web document: http://www.cwls.org/las_info.htm; download date 6 March 2004.) FAYYED, U., PIATESTKU-SHAPIO,G. & SMYTH, P. 1994. Knowledge discovery and data mining: Towards a unifying framework. In: Advances in Knowledge Discovery and Data Mining. AAAI/MIT Press, Cambridge, MA, USA, 1-36.

240

C. JENKINS ET AL.

FLOCKS, J.G. 2006. Converting Analog Interpretative Data to Digital Formats for use in Database and GIS Applications. USGS Open File Report, 2004-1070. GEOTEK LTD. 2004. Multi-Sensor Core Logger. (Web document: www.geotek.co.uk/site/scripts/module. php?webSubSectionID = 26; download date 1 April 2004.) JENKINS, C.J. 1997. Building Offshore Soils Databases. Sea Technology, 38, 25-28. JENKINS, C.J. 2002. Automated digital mapping of sediment colour descriptions. Geo-Marine Letters, 22, 181-187. JENKINS,C.J. 2006. Uncertainty in geological seafloor mappings using point-site datasets. Continental Shelf Research, submitted. PENLAND, S., SUTER, J.R., MCBRIDE, R.A., WILLIAMS, S.J., KINDINGER, J.L. & BOYD, R. 1989. Holocene sand shoals offshore of the Mississippi River Delta Plain. Gulf Coast Association of Geological Societies Transactions, 39, 471-480. PRESS, W.H., FLANNERY, B.P., TEUKOLSKY, S.A. & VETTERLING,W.T. 1989. Numerical Recipes." The

Art of Scientific Computing (FORTRAN Version). Cambridge University Press, Cambridge. ROCKWARE INC. 2004. Rockworks2002. (Web document: www.rockware.com; download date 1 April 2004.) SKINNER, L.C. & MCCAVE, I.N. 2003. Analysis and modelling of gravity-and piston coring based on soil mechanics. Marine Geology, 199, 181-204. WILLIAMS, S.J., JENKINS, C. ET AL. 2003. New digital geologic maps of US continental margins; insights to seafloor sedimentary character, aggregate resources and processes. In: Coastal Sediments '03 Conference Proceedings, Fifth International Symposium on Coastal Engineering and Science of Coastal Sediment Processes, Crossing Disciplinary Boundaries, Clearwater Beach, FL, 18-23 May 2003. CD-ROM published by World Scientific Publishing Corporation and East Meets West Productions, Corpus Christi, Texas, USA. WINSTON, P.H. 1984. Artificial Intelligence, 2nd edn. Addison-Wesley, Reading, MA.

Core data stewardship: a long-term perspective CARLA

J. M O O R E

& R A Y E. H A B E R M A N N

US Department of Commerce, National Oceanic and Atmospheric Administration, NOAA Satellite and Information Service, National Geophysical Data Center, E/GC3 325 Broadway, Boulder, CO 80305-3328, USA (e-mail: [email protected]; [email protected]) Abstract: The US National Oceanic and Atmospheric Administration (NOAA) National

Geophysical Data Center (NGDC) and collocated World Data Center for Marine Geology and Geophysics, Boulder, CO, USA provides scientific data stewardship for many environmental datasets, including geological information derived from sea-floor samples. The essence of NGDC's stewardship philosophy is that data management practices must evolve and incorporate new technologies in order to keep data interoperable with complementary data streams and maintain their usefulness in a changing research environment. The Index to Marine and Lacustrine Geological Samples database exemplifies NGDC's evolutionary and collaborative approach to data management. The most recent version of the Index is a geospatially enabled relational database, providing data discovery and delivery via an interactive map on the Web. Geospatial databases and Internet mapping tools are an integral part of NGDC's current centre-wide systems architecture. Environmental data management in the 21st century tends to focus on standardized, digital data produced in terabytes by remote sensing systems, largely because of their coverage and ready availability for numerical analysis and environmental modelling. In contrast, geological data from the sea floor are diverse, relatively small in volume and sparse in coverage due to the extremely high cost of sample collection. The diversity of these data, both in content and form, tends to make them less directly 'available' for many software applications. At the same time, the scarcity of sea-floor samples increases the demand for historical data, introducing further diversity over time through changing measurements and technologies, and making

effective long-term data stewardship of data about these samples absolutely essential. This paper addresses the role of data centres in core data stewardship, and specifically the evolution of the Index to Marine and Lacustrine Geological Samples database, maintained by the National Geophysical Data Center (NGDC) on behalf of a consortium of primarily North American curators of geological sample material. The purpose of the Index is to assist scientists in locating sea-floor and lakebed geological sample material for use in their research, and to permanently archive information about samples in the collections of the participating repositories. Figure 1 illustrates the geographic distribution of the more than 260000 sea-floor samples

Fig. 1. Sample locations in the Index to Marine and Lacustrine Geological Samples database, December 2004.

From: ROTHWELL,R.G. 2006. New Techniquesin Sediment Core Analysis. Geological Society, London, Special Publications, 267, 241-251. 0305-8719/06/$15.00 9 The Geological Society of London 2006.

242

C . J . MOORE & R. E. HABERMANN

Table 1. Collaborating institutions and data contributed, December 2004 Cruises in database North American participants Antarctic Research Facility, Florida State University Bedford Institute of Oceanography, Geological Survey of Canada Deep Sea Drilling Project Lamont-Doherty Earth Observatory, Columbia University National Lacustrine Core Repository, University of Minnesota National Oceanic and Atmospheric Administration Ocean Drilling Program* Oregon State University, College of Ocean and Atmospheric Sciences Scripps Institution of Oceanography, University of California, San Diego Smithsonian Institution, National Museum of Natural History University of Hawaii, School of Ocean and Earth Science Technology University of Miami, Rosenstiel School of Marine and Atmospheric Sciences University of Rhode Island, Graduate School of Oceanography University of Southern California University of Texas, Institute of Geophysics University of Wisconsin US Geological Survey Woods Hole Oceanographic Institution European affiliates Alfred-Wegener-Institute for Polar and Marine Research, Germany IFM-GEOMAR Leibniz-Institute for Marine Sciences, Germany Southampton Oceanography Centre, UK International Marine Global Change Study (samples stored at other facilities) Total (including samples in review)

Samples in database

Intervals in database

Records with lithology

Records with age

Data or image links to

61

4047

9099

8906

2392

252

8781

8781

0

0

96 93

20 323 11 046

96 939 25 933

91 119 19762

94209 17414

7

281

510

509

491

19

997

2418

1693

58

107 236

631 5133

631 5289

0 4850

0 2476

Facility NGDC

156

6210

10 746

10187

8978

NGDC

1416

59 628

59 628

59426

0

NGDC

80

919

1626

1524

786

86

1780

1780

0

0

115

2972

3765

3593

857

391 10

7472 205

7472 508

5591 508

5704 470

1 331 330

349 16 575 4023

349 18 233 4024

349 7708 3040

348 8335 3197

45

1820

1820

83

2416

2416

43

581

581

4

189

189

3962

156378

262737

NGDC

NGDC NGDC & Facility NGDC & Facility

NGDC

Facility NGDC

Facility 218765

145715

* 'Samples in database' is number of sites, not cores for the Ocean Drilling Program.

represented in the database as of D e c e m b e r 2004. Table 1 summarizes database contributions by participating institution or agency. The US N a t i o n a l Science F o u n d a t i o n (NSF) funds several institutions to contribute i n f o r m a t i o n on their collections, and requires that sea-floor samples collected with N S F funds be described in the database (National Science F o u n d a t i o n

2004). Continually u p d a t e d data and i n f o r m a t i o n are available on the curators' c o n s o r t i u m W e b

site at U R L h t t p : / / w w w . n g d c . n o a a . g o v / m g g / curator/. D u r i n g 2004, over a quarter of a million W e b pages and data listings, consisting of nearly four gigabytes o f data a n d information, were served to users worldwide from the curators' W e b site and database. Twenty-five per cent of users were from the US and 20% were from 91 other identifiable countries. The top 10 nations accessing the site last year were: Japan, the U K , N o r w a y , C a n a d a , France, Italy, Switzerland, Brazil, Mexico and the Netherlands. The

CORE DATA STEWARDSHIP country of origin of the remaining 55% of users is uncertain due to shared domain name extensions (.com,.net,.org) or unresolved domain addresses.

The role of national and world data centres in core data management National and world data centres play a vital role in preserving the public availability of sea-floor data by partnering with individual researchers, institutions and research programmes to provide continuing stewardship for the data they create (National Research Council 2002, p. 62, Sidebar 4-2). As with any data type, standards-compliant archival and documentation are the key to maintaining scientific validity of data derived from sea-floor samples. Secure storage keeps data in existence, but their utility depends on their availability in a meaningful form. The National Geophysical Data Center is committed to incorporating new technologies and protocols into its systems architecture to ensure the enduring availability of data in its archives. NGDC's goal is to encourage the free and open exchange of data among scientists worldwide and to promote excellence in data management through international cooperation. Over the past nearly three decades, N G D C has collaborated extensively with the US National Science Foundation, archiving data for many NSF-supported marine geological programmes. Examples include the curators' consortium, the International Decade of Ocean Exploration, and international drilling programmes spanning the Deep Sea Drilling Project, Ocean Drilling Program and new Integrated Ocean Drilling Program. Partnerships with individual researchers, institutions and research programmes around the world are the foundation on which NGDC's data management programs are built.

The Index to Marine and Lacustrine Geological Samples: an example of stewardship

Collaborative inception of the database In 1977, a curators' consortium was formed, consisting of representatives from several US oceanographic institutions and NGDC. NSF funded two workshops where the group met to discuss ways to more effectively share data and information about their sea-floor sample collections with the scientific community (McCoy 1977). As part of this process, they designed an 80-column fixed-field data format and asked

243

N G D C to manage the resulting database. The original format decided upon by the group included station location, codes for sample collection information and storage methods, codes for summary lithology and texture, geological age, and a 15-character comments field. Initially, the curatorial repositories handcoded data onto paper forms and sent them to N G D C where they were key-punched, then archived on magnetic tape. Incoming data were subjected to a series of F O R T R A N programs that checked latitudes, longitudes, and fields with standardized content such as device, lithology, texture and age. Then, as now, irregularities were resolved iteratively with the contributor, and all entries were reviewed and approved by the responsible curator before incorporation into the database. By the end of 1978, nearly 5000 sea-floor samples were described in the database. Four years later, in 1981, the database included information on 49 691 cores, grabs and dredges. This represented a Herculean effort on the part of the participating institutions considering the laborious nature of the hand-coding and key-entry required. Data for the 9067 cores from the Deep Sea Drilling Project were an exception, because they were algorithmically generated and digitally submitted. Initially, researchers accessed the database by conveying their search criteria via telephone, fax or post to NGDC. A rudimentary command-line interface was made available late in 1979 enabling the curators to perform simple text searches from a remote terminal. Regardless of whether searches were run remotely, or by N G D C staff, the resulting sample listings or sample location plots were printed out at N G D C and posted to the requestor.

Co-operative evolution of the database Composition of the curators' consortium has changed over time, as has the form and content of their database. All modifications to the database are approved by the consortium. The group has met approximately every 2 years for the last 28 years and corresponds via a list server between meetings (Mix et al. 2003). Techniques used to access the Index have also evolved (Fig. 2). During the mid-1980s, NOAA headquarters implemented the System 2000 hierarchical database management system (where data are stored in a tree structure or hierarchy) on a central mainframe. A useful suite of data access and manipulation tools accompanied the system. Translation of the Index to Marine Geological Samples to System 2000 enabled flexible

244

C.J. MOORE & R. E. HABERMANN

1978

1984

Fixed-field ~rmat

1990

Hierarchical dbms*

dbms* tools

1994

Relational dbms*

l e m m i n g | i n | l i e |

Custom FORTRAN tools

1991

2001

Geospatial

~

C tools dbms* tools +Web access/Jav~map +interactive maps

*commercial database management software/tools used for data ingest and access

Fig. 2. Evolution of software used to access the Index to Marine and Lacustrine Geological Samples. ad hoc queries on the database using its built-in 'natural language' capability. Access continued to be provided to users primarily through N G D C staff. In the late 1980s, when NOAA decentralized its data systems and decommissioned System 2000, the Index was briefly reconstituted as an ASCII flat file with custom C programs used for maintenance and access. N G D C then acquired the Oracle | relational database management system (Oracle Corporation, Redwood Shores, CA) and implemented the Index as a series of related tables. The entire suite of Oracle | database management tools became available for data ingest and quality control, and database maintenance and access. In particular, the relational structured query language (SQL) enabled flexible queries and made database corrections and system modifications much simpler. Built-in database constraints and cross-links between tables provided additional automatic error detection. During this same timeframe, the curators began work on a new set of parameters to better describe rock dredge samples in the database. A set of standardized descriptors for rock lithology, mineralogy, glass content, weathering and metamorphism was created by curators at the University of Rhode Island and Oregon State University, and circulated for review. At their 1991 meeting, the curators' consortium approved the final rock dredge extensions and voted to abandon the fixed-field exchange format in favour of a more flexible, tab-delimited, variable-length record. The group also requested a spreadsheet-based data entry form. To implement the changes, N G D C redesigned the relational structure of the database and programmed an Excel | (Microsoft Corporation, Redmond, WA) entry form to include dropdown lists of descriptors that the curators could use to automatically populate many fields. This database entry scheme, with minor modifications and additions, is still in use, as are the quality control programs that pre-process incoming tab-delimited data. Although several repositories still use the spreadsheet for data

entry, other institutions use alternative methods to produce equivalent data records in tabdelimited form.

Internationalization of the database During 1993 N G D C established a Web site and, by 1994, the Index to Marine Geological Samples was fully searchable online through the N G D C Web server (http://www.ngdc.noaa.gov/mgg/ curator/). The online search interface was developed using Oracle's '~ Internet Application Server software and PL/SQL, a proprietary programming language based on the structured query language common to most modern relational database management systems. A parallel, peer-review, Web interface was implemented to allow the curators to visually inspect sample information before approving it for addition to the database. For the first time, users were able to freely access the database over the Internet, and to construct custom queries through a Web forms interface (Fig. 3). N G D C enhanced the interface with JavaTM-based maps (Sun Microsystems Inc., Santa Clara, CA) for interactive entry of latitude and longitude limits, and display of search results in map form. Database access became not only free, but globally available and nearly instantaneous (Moore 1996). The Web interface sparked international interest in the Index. Representatives from Canada, Germany and the United Kingdom joined the curators' group. The Intergovernmental Oceanographic Commission, Committee on International Oceanographic Data and Information Exchange endorsed the Index to Marine Geological Samples database in resolution IODE-XIV.2 (Intergovernmental Oceanographic Commission 1996). Soon afterwards, a European equivalent, EUROCORE, was created for sea-floor samples collected by members of the European Union and/or held by European sample repositories. The EU-SEASED online data access system (http://www.eu-seased.net/) provides extensive metadata on collections described in EUROCORE, although it does not, as yet, provide the

CORE DATA STEWARDSHIP

245

Fig. 3. The Oracle g PL/SQL Web interface to the Index (non-extended version).

capability to directly download descriptive or analytical data derived from sea-floor samples. EU-SEASED and the Index cross-reference one another on their respective Web sites.

Addition of new types of data and interconnectivity to the database At the 1998 meeting of the curators' consortium, the Index to Marine Geological Samples officially became the Index to Marine and Lacustrine Geological Samples with the addition of lakebed data from the National Lacustrine Core Repository at the University of Minnesota. A free-form description field was added to accommodate the new lacustrine data and the group voted to include information about province and principal investigator in the database,

as well as a flag indicating whether sample material was exhausted. The curators also agreed to add cruise reports, core photographs, graphic core descriptions and other data to the database. The additions were to be either as contributions to the N G D C archive or implemented as links to data and images maintained at participating facilities (Table 1), according to the preference of each curator. It was agreed that, should the institute no longer be capable or willing to provide these ancillary data to users, then the data were to be transferred to N G D C and the collocated World Data Center for Marine Geology and Geophysics, Boulder, CO, where the data would be permanently archived and remain in the public domain. Expansion of the database to include viewing and download of data and imagery corresponding

246

C.J. MOORE & R. E. HABERMANN

Fig. 4. Examples of ancillary data and imagery accessible through the Index. to sea-floor and lakebed samples introduced a new round of system developments at NGDC and additions to the Web interface (Fig. 4). Several institutions, including the National Lacustrine Core Repository, the University of Rhode Island, Scripps Institution of Oceanography and Oregon State University, have contributed core photographs and logs, scanned cruise reports and other data to the database. Many of these are unpublished and were not previously available online. Beginning in 2005, NOAA is funding a new project to rescue and digitize images and data from several participating institutions (National Geophysical Data Center 2005). The multi-year effort is to scan more than 50000

core photographs, 7500 core X-rays, 20 000 seafloor photographs and 14000 pages of paper documents, and to key-enter information from 5000 pages of core logs and descriptions. Resulting digital data and information will be provided to the contributing repository for their use, permanently archived by N G D C and integrated with the Web interface to the Index for online access. The Index also acts as a portal to large quantities of auxiliary data and information available through the Web sites of participating facilities. For example, entries for Ocean Drilling Program cores link directly into the Janus database at Texas A&M University (Ocean Drilling Program 2005). Many Lamont-Doherty Earth Observatory

CORE DATA STEWARDSHIP entries point back to that institution's own online core search for more information (LamontDoherty Earth Observatory 2005), and rock dredge entries in the Index from multiple institutions link to related geochemical analyses in the Petrological Database of the Ocean Floor (Lehnert et al. 2000). Interoperability with more external systems is planned.

Using the database for improved access to related data archives at N G D C Discrepancies in latitudes, longitudes and naming conventions used by different investigators for a single geological sample are problematic when attempting to locate all data available for that sample, especially when the information is dispersed across multiple publications, laboratory reports and/or digital databases. Because information in the Index is created and verified by the responsible repository, the Index is a definitive source of correct locations and nomenclature not only for archived samples, but also for related data archived by NGDC. N G D C is using the Index to create a common cross-reference to additional data in its archives. For example, the full suite of core data from the

247

Deep Sea Drilling Project archived by N G D C is now searchable through the Index. Index entries for samples archived by the Smithsonian Institution collected during hydrographic surveys by NOAA's National Ocean Service provide a pathway to all related information at N G D C for that survey, including descriptive reports, smooth sheets, hydrographic soundings and other geophysical data. Other candidate data in the NGDC archives include roughly 70000 pages of scanned cruise and data reports in Adobe :~ Portable Document Format (PDF) (Adobe Systems Inc., San Jose, CA) and several hundred megabytes of digital data contributed by researchers and institutions worldwide over the past three decades.

Geospatially enabling the database In 2001, because the database was already in a relational system, N G D C was able to rapidly implement an online interactive map interface to the Index to Marine and Lacustrine Geological Samples using the Earth Science Research Institute (ESRI t~:) ArclMS T M Internet mapping software (Environmental Systems Research Institute Inc., Redlands, CA) (Fig. 5) (Moore

Fig. 5. ESRI c~ ArcIMS T M interactive map Web interface to the Index.

248

C.J. MOORE & R. E. HABERMANN

Fig. 6. Western Pacific and Indian Ocean sea-floor sample locations viewed on a background layer of shaded bathymetry.

et al. 2001). Initial data layers included individual institutional collections, a combined layer for searching across repositories, continents, lakes and colour shaded relief from satellite imagery. The interactive map cross-references the Oracle ~ forms interface shown in Figure 3. Standard ArcIMS TM options are available for data discovery, including zoom, rectangular and polygonal selection, identify, find, query and export tools. Results returned include station and collection information, summary lithology and age, and links to full data and information for the samples selected. An advantage of the map interface is the capability to add new layers. In 2004, N G D C added background imagery derived from satellite images over land areas and N G D C ' s 'ETOPO2' global 2-minute gridded elevation database (National Geophysical Data Center 2001) over ocean areas as a map layer to provide depth context for the display of sample locations (Fig. 6). N G D C is working on a layer of polygons representing the boundaries of named oceans and seas.

In addition to being flexible, visual and extensible the interactive map interface enables export of selected data as shapefiles, ready for direct upload into a geographic information system (GIS) on the user's desktop. Shapefiles include not only those components necessary to use the data with GIS software, but also a data table in dBASE TM format (dataBased Intelligence Inc., Vestal, NY) that is importable into many commercial spreadsheet or database software packages.

Into the future: NGDC's overall systems strategy and architecture Implementing an integrated approach to data management N G D C is simplifying its data management architecture by focusing development on commercial off-the-shelf software for metadata management, geospatial database systems and Internet mapping tools (Fig. 7). The goal is to build multiple

CORE DATA STEWARDSHIP

249

Fig. 7. The NGDC environmental data architecture test bed.

pathways for migrating data into and out of these systems and multiple connections between them. The system is being developed by the centre-wide geospatial data services group. The geospatial group is also active NOAA-wide in designing and implementing metadata and data management solutions.

The role of metadata in data management The metadata system N G D C is building exemplifies this flexible, interoperable approach. The system is based on metadata content standards developed by the US Federal Geographic Data Committee (FGDC) and the International Standards Organization. There are multiple paths into the metadata system (Fig. 7, lower right). Data managers can enter metadata using a variety of Web interfaces or they can upload metadata from extensible mark-up language (XML) files on their desktop computers. Metadata can also be imported from and exported to other metadata systems, i.e. the Global Change Master Directory (http://gcmd. nasa.gov/) and F G D C Clearinghouses. N G D C

is developing interfaces into the system directly from data-processing systems using open source standards and protocols. Metadata in the system are searchable through widely used document search protocols such as the US Library of Congress Machine-Readable Catalog Standard, the international Dublin Core standard and the International Standards Organization Z39.50 protocol (National Information Standards Organization 2004). Results can be displayed through a variety of Web interfaces or as XML. Standards-based development ensures continued interoperability with evolving systems worldwide (Gatenby 2004). We are in the process of integrating all of NGDC's existing data-tracking schemes with the metadata system. This approach will ensure that standards-compliant metadata are created immediately for all data that come into the centre, and will automatically update the metadata records as the data are processed. Comprehensive documentation is fundamental to the long-term value of scientific data and, as a national and world data centre, NGDC's mandate includes maintaining a strong datametadata connection.

250

C. J. MOORE & R. E. HABERMANN

Geospatial databases." a data stewardship foundation The heart of the N G D C system is a geospatial database. N G D C is using both the Oracle ~' and IBM "~-) Informix | (International Business Machines Corporation, White Plains, NY) relational database management systems. These systems support inclusion of geospatial objects (points, lines and polygons) directly in the database as a custom data type. They also support a wide variety of operations on these objects (intersections, point-in-polygon, nearest neighbour, etc). This enables integration of all N G D C geospatial data into a single foundation while significantly decreasing the custom programming required. Like in the metadata case, the emphasis is on migrating many legacy formats into the databases, a process supported by both commercial off-the-shelf software and in-house tools. Once the data are in the database they are accessible for Web-based text and spatial searches and retrievals using an interface database that holds fragments of structured query language and hypertext mark-up language. The data are also easily integrated into a variety of desktop geographic information systems (GIS) and Internet mapping tools. The combined metadata and tracking system is also tightly integrated with NGDC's spatial databases making the information directly accessible to desktop GIS and other search and display programs.

Internet mapping: bringing it all together The Internet mapping component of the system is supported by ArcIMST~r and the OpenGIS '~ Consortium compliant Web Feature Service and Web Map Service mapping tools (Open Geospatial Consortium Inc., Wayland, MA) (http://www.opengis.org/). The Web Feature Service specification allows communication between clients and servers at the feature (point, line and polygon) level, while the Web Map Service specification supports creation of map-like views of data gathered from multiple, diverse and/or remote sources. These tools support client- and server-side access, analysis and mapping applications, and are integrated with a number of existing and on-going data access and analysis projects in NOAA and N G D C (atmospheric trajectory models, custom data objects, etc.). N G D C works with software vendors like ESRI ~2~, IBM c~ (InformixC'~), Oracle '~ and Blue Angel Technologies Inc. to extend their tools to

better accommodate the needs of data managers and data users. The centre acts as a beta test site for relevant software, providing input on new features and capabilities, and making suggestions for interface improvements. This co-operation leverages the vendors' development efforts, ensuring that their products better meet the needs of our users.

Incorporating partnerships in systems design Our experience at N G D C clearly indicates that on-going partnerships with data providers are a crucial element in end-to-end scientific data stewardship. The Reference Model for an Open Archival Information System (OAIS) is an international standard for archives developed recently by the Consultative Committee for Space Data Systems (Consultative Committee for Space Data Systems 2002). The OAIS model also stresses the importance of partnerships between data providers and archives, and proposes agreement frameworks for developing and describing them. N G D C supports this concept and is developing prototype agreements with several data providers.

Benefits to legacy databases of the new systems architecture Importing existing databases such as the Index to Marine and Lacustrine Geological Samples into centre-wide geospatial database systems allows immediate application of standard quality assessment and improvement tools to the data as well as their integration into desktop and Internet mapping applications. Our goal is to make these tools available on the Web to N G D C data managers as well as to data providers, supporting interactions between these groups that improve stewardship of the data. This standards-based, interoperable approach to data management promotes global use of data archived by NGDC and ensures its long-term usefulness to researchers worldwide. The Index to Marine and Lacustrine Geological Samples database was conceived through the vision of F.W. McCoy, D. Heinrichs of NSF, and the other original members of the Curators' Consortium: W.R. Riedel, R.W. Roberts, D.S. Gorsline, M. Perlmutter, D.A. Johnson, H.J. Schrader, J.V. Gardner, T.C. Moore, W.M. Ferrebee, F. Theyer, D. Clark, K. McMillen, D.S. Cassidy and R. Combellick. There are many other curators, too numerous to list, who have helped the database grow and evolve in form, content and capability over the past nearly three decades. The Index continues to be a valuable long-term resource for the global scientific community due to the support of NSF, and the dedication of these curators and their respective institutions.

CORE DATA STEWARDSHIP

References CONSULTATIVE COMMITTEE FOR SPACE DATA SYSTEMS.

2002. Reference Model for an Open Archive Information System (OAIS). CCSDS 650.0-B-1. World Wide Web address: http://www.ccsds.org/docu/dscgi/ds. py/Get/File- 143/650x0b I .pdf. GATENBY, J. 2004. Internet, Interoperability and Standards, Filling the Gaps. National Information Standards Organization White Paper. World Wide Web address: http://www.niso.org/press/whitepapers/ Gatenby.html. INTERGOVERNMENTAL OCEANOGRAPHIC COMMISSION.

1996. Committee on International Oceanographic Data and Information Exchange Resolution 10DEXIV.2. World Wide Web address: http://ioc.unesco. org/iode/. LAMONT-DOHERTY EARTH OBSERVATORY. 2005. Deep Sea Sample Repository. World Wide Web address: http://www.ldeo.columbia.edu/res/fac/CORE_ REPOSITORY/RHP 1.htrnl. LEHNERT, K., Su, Y., LANGMUIR,C.H., SARBAS, B. NOHL, U. 2000. A Global Geochemical Database Structure for Rocks. G 3. Technical report, 1999GC000026, ISSN: 1525-2027. World Wide Web address: http://www.petdb.org/. McCoY, F. 1977. At Scripps marine curators gather. GEOTIMES, 22, (120), 26-28. Mix, A., CONRAD, B. ET aL. 2003. Curators of sea floor and lakebed samples celebrate 25 years of service. LOS, Transactions of the American Geophysical Union, 84, (20), 191-192. MOORE, C.J. 1996. Curators of marine geological samples gather at NGDC; Twenty years of cooperation results in worldwide access to global ocean floor samples. Earth System Monitor, 6, (4), 10-11.

251

MOORE, C.J., MIRCHANDANI,A.R. & STEVENS,T. 2001. RDBMS and GIS as a framework for marine geological data discovery and access at NGDC. Los, Transactions of the American Geophysical Union, 82, (47), Fall Meeting Supplement, Abstract OSl IB-0356, f591. NATIONALGEOPHYSICALDATA CENTER.2001. ETOP02 Global 2-Minute Gridded Elevation Data Set. World Wide Web address: http://www.ngdc.noaa.gov/ mgg/global/global.html. NATIONAL GEOPHYSICALDATA CENTER. 2005. Digitization of Marine and Lacustrine Records of Climate Change, a Project of the US Department of Commerce, National Oceanic and Atmospheric Administration, National Environmental Satellite, Data, and Information Service, National Climatic Data Center Climate Data Modernization Program. World Wide Web address: http://www.ncdc.noaa.gov/oa/climate /cdmp/cdmp.html. NATIONAL INFORMATION STANDARDS ORGANIZATION. 2004. Understanding Metadata. National Information Standards Organization Press. World Wide Web address: http://www.niso.org/standards /resources/UnderstandingMetadata.pdf. NATIONAL RESEARCHCOUNCIL. 2002. Geoscience Data and Collections: National Resources in Peril. National Academy Press, Washington, DC. NATIONALSCIENCEFOUNDATION.2004. Division of Ocean Sciences Data and Sample Policy. NSF, 04-004. World Wide Web address: http://www.nsf.gov/pubs /2004/nsf04004/. OCEAN DRILLING PROGRAM. 2005, JANUS Web Database. Ocean Drilling Program, College Station, TX. World Wide Web address: http://iodp.tamu. edu/database/.

The Janus database: providing worldwide access to O D P and I O D P data RAKESH

MITHAL

& DAVID

G. B E C K E R

Texas A & M University, Integrated Ocean Drilling Program, 1000 Discovery Drive, College Station, T X 77845, USA (e-mail: [email protected]) Abstract: In 1997, the Ocean Drilling Program began using a relational data management system and applications called Janus to store and retrieve data collected on the drill ship JOIDES Resolution. Since then many new web-based data access queries have been added, and some modifications to the database have been made. The database modifications were needed to allow the migration of pre-Janus data to the database, to incorporate modifications in the calibration procedures, to improve database performance, to add storage and access to various digital images, and to add access to scanned digital images of original paper data. These modifications were made by staff at Texas A&M University and reflect enhancements required by the Ocean Drilling Program (ODP) scientific community. With the start of the Integrated Ocean Drilling Program (IODP) an opportunity exists to make additional enhancements to the existing Janus database. As such, two major areas are identified: developing a core description program(s) and providing data visualization tools for improved analysis. Efforts to use Java-based programs to build these enhancements are cited. Geographic information systems (GIS) technology is being explored as a viable approach to providing these enhancements. Building an understanding of the Janus database is best accomplished by visiting the Web site http://iodp.tamu.edu/database. Available data can be reviewed using the online 'Database overview' option. A list of all queries that are available in Janus can be acquired through the online 'Data search' option.

This paper describes the Janus database system (as of October 2004) that was developed and enhanced during the Ocean Drilling Program (ODP), and offers suggestions regarding potential enhancements for use in the new Integrated Ocean Drilling Program (IODP). The Ocean Drilling Program was an international partnership of scientists and research institutions that focused on drilling deep earth geological samples beneath the oceans of the world. The scientific deliverables for which Texas A&M University was responsible included acquisition of data on the riserless drill ship JOIDES Resolution, database management, data quality, data access and data migration. The legacy of ODP is the collected data, the publications and the actual cores, which are stored at various repositories. Of importance to this paper are the data that are available online in the Janus database. The Integrated Ocean Drilling Program (IODP) is an international research programme that explores the history and structure of the Earth as recorded in sea-floor sediments and rocks. I O D P builds upon the earlier successes of two international drilling programmes, the Deep Sea Drilling Project (DSDP) and the Ocean Drilling Program (ODP) (Ocean Drilling Program 2002), which revolutionized our view

of Earth history, climate change, natural hazards and microbiology. I O D P greatly expands the reach of the previous programmes by using multiple drilling vessels, including riser, riserless and missionspecific platforms, to achieve its scientific goals. These goals follow the themes outlined in the IODP Initial Science Plan (Integrated Ocean Drilling Program 2001): 9 9 9

the deep biosphere and the subsea-floor ocean; environmental change, processes and effects; solid earth cycles and geodynamics.

A riser vessel, Chikyu, will be supplied and operated by Japan. The riser vessel will allow for long-term expeditions lasting up to 1 year in a single location. Mission-specific platforms, operated by the European Consortium for Ocean Research Drilling, will be used to study ice-covered and shallow-water regions. The United States will operate a riserless vessel for 2-month expeditions around the globe.

The Janus database

The original Janus database The initial development of the Janus database occurred from 1994 to 1997, and it was first

From: ROTHWELL,R.G. 2006. New Techniques in Sediment Core Analysis. Geological Society, London, Special Publications, 267, 253-259. 0305-8719/06/$15.00 9 The Geological Society of London 2006.

254

R. MITHAL & D. G. BECKER

used in production mode on O D P Leg 171. Three groups were involved in the development of the Janus database: one representing the final users of the system (the Janus Database Steering Committee), a second representing the final supporters of the system (Texas A&M University) and a third, the developers of the system (Tracor Applied Sciences Inc., Austin, TX). Each group contributed to the success of the final database. Leaving any one of the three groups out of the development work could have resulted in a lower quality product. The scientific community decided that the new data management system should be based on: ~ 9 ~

relational database management system (RDBMS); Unix servers; support all (PC, Mac and Unix) client computers.

At that time the popular R D B M S ' were Oracle, Sybase and Informix, and the popular Unix servers were made by Hewlett-Packard (HP), Digital Equipment Corporation (DEC), SUN Microsystems and Silicon Graphics Inc. (SGI). Tracor, in association with the Janus Database Steering Committee and Texas A&M University, chose Oracle RDBMS and DEC Unix servers. The

choice of Oracle as the RDBMS and DEC as the Unix servers worked out very well. Oracle is a very powerful database engine that keeps a detailed log of every user activity and provides expanded functionality for disaster recovery due to accidents or user errors. The database design provides an opportunity to create and implement constraints for data integrity. The DEC Unix servers were also quite powerful. A cluster of two servers was set up to ensure redundancy and if one server failed the other would pick up all services automatically. Janus has been in operation for about 7 years (1997-2003) - there have been no losses of data and the servers have been up 100% of the time. To support multiple client computers Tracor used third-party software, Neuron Data, to deal with the inherent differences between the PC, Mac and Unix platforms, and built the Janus applications on top of it. The database was originally designed to house the basic core information and the numeric scientific data associated with those cores. All science data are attached to either a core section or to a sample identifier. Therefore, the basic information regarding leg, site, hole, core, section, sample and its depth forms the backbone of all the scientific data in the database. Table 1 shows the priorities

Table 1. Janus development priorities* Group 1. Core log Leg/site/hole Sample Chemical samples Group 2a. Gamma-ray attenuation bulk density P-wave velocity Magnetic susceptibility Natural gamma Colour reflectance Palaeomagnetics Group 3.

Group 2b. Palaeontology Age profile Smear slide

Downhole temperature tools (Adara, WSTP) Thermal conductivity Sonic velocity Shear strength Moisture and density

Group 4a. Hydrocarbon safety (Rock Eval) Carbon/carbonates Gas chromatography Interstitial water X-ray fluorescence/X-ray diffraction Group 5.

Hard rock description Thin section description

Group 6.

Tensor/sonic core monitor Underway geophysics Seismics Core photographs

* Data type priorities in which the Janus database was developed.

Group 4b. Sediment description Structural description

JANUS DATABASE of data types (Janus Database Steering Committee unpublished minutes 1994) for the development of the Janus database. Tracor started with the development of data types in Group 1. After its completion the development team was divided into two teams. Team 1 started work on Group 2a and Team 2 started work on Group 2b. After completion of Group 2a the Team 1 continued with Groups 3, 4a, etc. Team 2 completed work on Group 2b and then started on Group 4b. Most of the data types were successfully incorporated into the Janus database either originally by Tracor or later by the Ocean Drilling Program at Texas A&M University (ODP/TAMU).

Data flow through Janus The data flow through the Janus database is shown in Figure 1. The central Janus database consists of more than 450 Oracle tables. The Janus data model is divided in 25 subject areas. The Entity-Relationship diagrams and associated attribute definitions are available on the Web at http://iodp.tamu.edu/database/janusmodel.html. The data model is a very important part of the database. It serves as a graphic representation of the database, and provides a language in which all database design work is done. The Janus data model has been maintained using the data modelling software, ERwin, developed by Computer Associates, Islandia, NY. The ODP/TAMU database staff has kept the data model and the database in full synchronization. As of October 2004 the Janus database contained all data from IODP Expedition 301, ODP legs 171-210 and most of the data from ODP legs 101-170. The size of the Janus database, not including digital images, is about 12GB. The images from ODP legs 198-210 occupy about l134GB of disk space and are stored outside the Janus database. The data are collected either by direct data entry or by instruments. The Keyboard Data Entry Applications are used for direct data entry. The Instrument Control Applications produce data in ASCII files. The Data Uploaders read the ASCII files produced by Instrument Control Applications and enter the data into the Janus database. The Data Editing applications are needed for data clean-up. Some major applications used for Janus data are shown in Figure 1. The numbers after the names of the applications are their version numbers. For example, Janus 206 was last modified during ODP Leg 206. It is used to enter basic operations and core information. JSample is used to enter sample information. Paleojava, PaleoDatum and Agemodel are used to enter micro-palaeontological data. In data

255

uploader applications Janus is the original one, and is still used to upload chemistry and thermal conductivity data. The Generic Uploader is used to upload Multi-Sensor-Track (MST) and physical properties data. The JavaUploader is used to upload magnetic susceptibility and X-ray diffraction (XRD) data. The data editing applications are used to correct errors in the database. The Janus application is used to edit MST and physical properties data, CryoEdit to edit palaeomagnetic data, RSCdit to edit colour reflectance data and AdaraEditor to edit downhole temperature tools data. The SectionEditor is useful to correct core sections. The data can be retrieved from the Janus database in many different ways. Most of the data are retrieved on the Web using the Janus web queries written by developers at ODP/ TAMU. Some data can be retrieved and displayed using the CompositeLogViewer. This application was originally developed by Central Computer Services Co. Ltd for Japan Marine Science & Technology Center (JAMSTEC) and later modified by ODP/TAMU. The Gas Analysis Program (GAP) is used to display gas data for safety purposes. Commercial applications such as SQL*Plus and Microsoft Access can also be used to extract data from the Janus database.

Enhancements to Janus The database group at ODP/TAMU made several enhancements to the original Janus database from 1997 through to the end of ODP in 2003. The migration of older ODP data required many modifications to the original Janus database. For example, the Janus database was designed to store calibration information and raw data - the results were calculated from both of them. However, older ODP data had no calibration information. Therefore, the Janus database had to be modified to store the results as well. The changes in the calibration procedures of some instruments in the laboratories aboard the JOIDES Resolution, for example P-wave velocity data acquisition systems, required corresponding changes in the Janus database. Some modifications were made to the database to improve its performance. As an example, the redesign of the natural gamma-ray (NGR) spectra tables resulted in a more compact Oracle table allowing faster data retrieval. New data types were added to the Janus database; for example, point source susceptibility (ms2f) and non-contact resistivity (ncr). A very important enhancement to the Janus database was the addition of digital images and their access through the database. The images

256

R. M I T H A L

& D. G. B E C K E R

l-i 01

o

o n

9 ,,,,

0 " - -

f r 09

rr~Xc

'

0"~ 13_ ~ ~ - ~

,,~~

~

0

~- ~

0.--

~--o

~ ~ ~=

~-

~

~0r

,^ ~1

~ ~

~.

~.~ 0 0 "~ ,,~ - r . ~

~

Q_

.~ r ~

~- 0

~"

~- ~ ' =

9 o=

6

~1 .'*~--

o ~

0~.- ,o

F-0~

.,o

,', ~-.~

o0~ >

~

~.~

o "o 0

0

~ ~o ~ ~ , ~ ~ ~,

.,,i-,

ID

\

E

o~,

~ LLI LLI 0

/

.~ ~.~

co o Q.

tO

d ~~

0

iii

.--

~.~ 0

~0

~, ~-~

o 0 ~ E0~

~

o~>~

o 0

.~o~

0

o

~

,~

0 0

~-~

~

"~ 'r_.

o~ ~

~'

~

~00

_

~

~ ~

~

N~

N

JANUS DATABASE are much larger in size than the numerical data and analysis showed that it was not cost-effective to include images in the Oracle database. Therefore, the images were placed on a server outside the Janus database. Within the database, only the path names to the location of the image files and an index of images were included. By using this methodology, a search function to the images was added without adversely affecting the performance of the database. As a consequence, access to images of core photographs, core sections and prime paper data were added as enhancements. Every core recovered during the Ocean Drilling Program was photographed on 4 x 5 inch film. Imaging specialists are in the process of scanning these core photographs and making them available on the Web. Almost 20 000 scanned images from O D P legs 151-210 are currently available through the Janus database. The remaining photographs are currently being scanned. Since O D P Leg 198, digital images of each core section were produced on the research vessel using a digital imaging system (DIS). All those core section images are available on the Web through the Janus database. Some information on all ODP legs (101-210) was collected on paper only, e.g. the hand-written, detailed core description of core sections by scientists aboard the ship. While compact core descriptions in graphic form (one page for each core) are available in the Initial Reports volumes of the Proceedings of the Ocean Drilling Program as P D F files, corresponding, detailed, hand-written core descriptions (one page for each core section) were available only on paper, and not in a digital format. These paper data have been scanned and are being made available via the Janus database.

Potential enhancements to the Janus database Core description. While some modifications and enhancements have been made to the original Janus database and Janus applications, several additional enhancements could be made. The first is providing visual core descriptions in a searchable format. Currently, the Janus core description information is only searchable by Leg/Site/Hole/Core/Section. It is not searchable by the contents of the core descriptions themselves. For example, one cannot search core sections where basic igneous rock was found or where current ripples were found. They cannot be searched for by combinations of lithology such as situations where sediments were found

257

underlying an igneous rock. Such searches are possible only if the contents of the core descriptions are in the database, not just the images of core descriptions. Previous efforts to include full core descriptions in the Janus database yielded unsatisfactory results. Consequently, providing access to this information remains a challenge to the Integrated Ocean Drilling Program (IODP). The Central Computer Services Co. Ltd in Tokyo, Japan is currently developing (for JAMSTEC) a Java-based application for core description. This application reads a digital image of the core section to be described and allows the user to add core description on top of the image. The newly entered core description information is saved in the Oracle database for future retrieval and/or editing. The application was developed to be mobile and used on tablet PCs. Geographic information systems (GIS) technology could also be used for core description if one thought of a cut section of core as a twodimensional (2D) map. The core could be scanned and the digital image placed into a 2D space, using x and y co-ordinates where one axis could be depth. Geological features on the image could be described (or mapped) using points, lines and polygons, the basic topological characteristics of a GIS. These described features would be stored in the Janus database while the topology would be stored in the GIS database. Researchers could view the core image, its GIS features and associated data (including descriptive information).

Data visualization. Another area of challenge is providing good data visualization tools. In the past, some efforts were made to provide tools to view safety data, such as gases in the cores, and to view more widely used numerical data including MST and Cryomagnetic data. However, providing a comprehensive Janus data visualization tool remains a challenge. As with core description, the Central Computer Services Co. Ltd is engaged in developing a Java-based data visualization application. This application provides a list of all cruises for which data exist in the database. The user selects a cruise and the application provides all data types that are available for that cruise. U p o n selecting the data types the user gets a visual plot of all those data at the same depth scale on their screen. The depth and data scales can be easily changed and new data columns can be added or deleted as the user wishes. The visual plot of data and images can be sent to a printer or plotter as a final printed product. This

258

R. MITHAL & D. G. BECKER

Ocean

Drilling

Program

Data O v e r v i e w - as of Oct. 7, 2004

ANALYSIS

Total in 210 209 208 Janus Site/Hole Summary (metres cored) 222685 828 357 3589 Hole/Core Summary (cores) 36440 115 218' 426 Core/Section Summary (sections) 192915 706 394 2987 Corelo 9 (samples) 2182440 6972 3416 38694 GRA Bulk Density (sections) 0' 2558 135878 571 Magnetic Susceptibility (sections) 135988 571 372 2575 Natural Gamma Radiation (sections) 73008 571" 372" 2404 P-Wave Velocity (Whole Core) (sections) 59546 0 0 1208 P-Wave Velocity (Split Core) (samples) 64751 580 149 638 Moisture Density (samples) 92011 586 145' 613 Thermcon (samples) 195 36920' 119 239 Shear Strenqth (samples) 26081' 0: 0 0 Colour Reflectance (sections) 83 0 2872 63264 Point Susceptibility - MS2F (sections) 2929' 590' 0 178 Downhole Temperature - Adara (sample,, 1219' 0 0 0 Splicer (tie points) 4372 0 0 349 Tensor (cores) 2534' 1~ 0 293 Cryomagnetics (sections) 100034 5991 336 2569 Palaeontological Investigation (samples) 27777 49: 0 0 Range Table (taxa) 451 0 0 20146 Ageprofile (Datum list) o o o 4573 Depth-Age Model 0 0" 0 2135 i X-Ray Diffraction (samples) 29 14219 i 188 120 X-Ray Diffraction Images (samples) 0 0 0 12176 X-Ray Fluorescence (samples) o o, o 3952 Inductively Coupled Plasma (samples) 1320 148 116 0 Chemistry: Rock Eval (samples) 0 0 7652 199 Chemistry: Carbonates (samples) 65700 768 110 1024 Chemistry: Gas Elements (samples) 20527 235' 0 175 Chemistry: Interstitial Water (samples) 12742 2 0 132 Smear Slides (samples) 44 0! 7 11927 Sedimentary Thin Sections (samples) 118 o o o Hard Rock Thin Sections (samples) 0 339 0 1606 Visual Core Description (sections) 36225 676 0 435 410 Core Photographic Images 24098 109 ! 215 Section Photographic Images 21410 689: 392 2886 Closeup Information 4943 206 342 88 ,

J

i

,

Notes: (1) Pre-204 information excluded for brevity. (2) For details and current information ~ _

,

207

206. 205

204

to

101 3122 515 281 504 123' 70 2714 450 268 31546 42531 3001 2268' 405 ! 220 2270 405 233 2240 405 230 14 37 0 1 8 8 7 366:100 12251 338! 309 13 ~ 93: 78 2 0! 0 2604' 162. 254 421 116 233 0 Oi 0 157 0! 0 3 17 0 1805 3601 253 1413! 01 0 823' 0! 0 o. ol o 0

81 0

o 0 157 860 334: 150 948 o 123 1000 4741 2649 254

0:

3066 526 2685 23049 1966 1929 0

35 99 1399 422 224 431 121 0 0

0 0 280 246 o

0

0

441 112 0 0

377 0

o~

o

o

43 51 0 0 23 51 105 56 172 28! 5 81 103 29 562 01 119 1188 o o o 174 89 0 0 0 0 116 60 337 447 265 2427 162 108 226

_

Fig. 2. An overview of the data in the Janus database as of May 2004. The database contains data from all ODP and IODP cruises (ODP legs 101-210 and IODP Expedition 301); however, only legs 204-210 are shown in this figure for brevity. application is still under development and is intended for production use on the riser ship Chikyu. Its implementation on the riserless ship would depend on the similarity of the database structures on the two ships.

A n o t h e r possible data visualization enhancem e n t is to use an electronic m a p to display geological and geophysical data. This can be done by incorporating GIS technology into the Janus database. A GIS focuses on the organization of

JANUS DATABASE data by geography/location. Throughout the Janus database there already exists GIS-ready information for sites and holes, namely geographic locations. Organizing these data by their geography allows one to access the data via an electronic map or plot the locations of sites and holes that meet certain predefined criteria. This new dimensionality to Janus would allow researchers to model geological information across space and time.

Accessing the Janus data The best way to gain an understanding of the current Janus database system is to go to its Web site (http://iodp.tamu.edu/database) and try it. There are two ways to access the data: 'Database overview' and 'Data search'. In the 'Database overview' option, one can see all available Janus data as a function of data type and leg number. By clicking on a leg number one gets a list of data available for all sites for that leg. By clicking on a site number, one gets a list of data available for all holes at that site. This view provides the number of records for each data type at each hole. By clicking on the number of records, one gets all of those data records. The queries can be easily modified for additional options by the researcher. In the 'Data search' option, the researcher gets a list of all queries. On choosing a particular query, one gets all options for that query. New users of Janus are encouraged to access the data through the 'Database overview'. Figure 2 shows an overview of the data in the Janus database as of October 2004. The database contains data from all O D P and I O D P cruises (ODP legs 101-210 and I O D P Expedition 301); however, only legs 204210 are shown in this figure for brevity.

Conclusions The Janus database is one of the most successful scientific databases in the world. It already contains most of the O D P data collected from

259

1985 through to 2003, and it is configured to start collecting data for the IODP. The credit for this goes primarily to the way that the Janus database was designed and developed. Three groups were involved: one representing the final users of the system, a second representing the final supporters of the system and a third, the developers of the system. The authors recommend that future database projects use a similar three-group approach to the design and development of their database systems. The methodology used to handle digital images turned out to be quite good and efficient. Keeping the images on a server outside the database, while retaining the index and path of those images in the database, proved to be efficient and fully serves the needs of the scientific community both on ship and on shore. Lastly, there are two areas in which the authors believe significant enhancements can be made to the Janus database: a better database structure and related application(s) to store and retrieve Visual Core Description (VCD) information and better data visualization tools. The opinions expressed in this paper are those of the authors and do not reflect the opinions of ODP, IODP or Texas A&M University. The Janus database was developed with funds provided by the Ocean Drilling Program. The ODP was sponsored by the US National Science Foundation (NSF) and participating countries under management of Joint Oceanographic Institutions (JOI) Inc. J. Fox and A. Klaus reviewed this paper and provided useful comments.

References INTEGRATED OCEAN DRILLINGPROGRAM.2001. Earth, Oceans and Life, Scientific Investigation of the Earth System Using Multiple Drilling Platforms and New Technologies, Integrated Ocean Drilling Program Initial Science Plan, 2003-2013. IODP, College Station, TX. OCEAN DRILLINGPROGRAM. 2002. Achievements and opportunities of scientific ocean drilling: the legacy of Ocean Drilling Program. JOIDES Journal, 28, 1.

Index Note: page numbers in italics denote Figures, while those in bold indicate Tables.

acoustic rolling contact (ARC) transducers 9 alkenones 20 applications CORTEX scanner 5 digital imaging 5 Eagle III BKA system 35-7 geochemistry 65-77 ITRAX core scanner 6, 60-2, 92 multi-sensor core loggers 4 non-imaging optical systems 5 porosity logging 4 X-ray computed tomography 6 X-ray fluorescence scanners 5-6 ARC see acoustic rolling contact transducers archives see databases arsenic 92 Artificial Neural Network SST 101,104 Atlantic sediments 39-50 autofluorescence image 147 Avaatech XRF core scanner 17, 39-50 CORTEX scanner comparison 40 NE Atlantic sediments 43-8 technical details 40 XRF logging 40-3 Balearic Abyssal Plain 79-98 biogenic opal content 115, 121, 123-4, 126-7 biomarkers 20-1 BOSCORF see British Ocean Sediment Core Research Facility boundaries Cretaceous-Tertiary 12, 14, 16 Palaeocene-Eocene 12, 13, 14 Br/C1 ratio 92 British Ocean Sediment Core Research Facility (BOSCORF) 66 bulk density estimates 173 Ca/Fe ratio 85-91 calcite see calcium carbonate calcium carbonate coralline aragonite 47-8 VNIS 130, 131-4, 136, 138 XRF core scanning 43-5 cameras 12, 101-2, 113-14, 117 carbonates cold-water mounds 47-8 colour analysis 115-17, 121, 124, 126 MRI visualization 202-3 see also calcium carbonate Cariaco Basin, offshore Venezuela 35 cathodoluminescence (CL) ~.maging 142

centric span SPRITE imaging technique 201-2 Challenger expedition 1-2 chamber systems, HYACINTH 156-7 chemical fossils 20-1 chronostratigraphy 102-3, 105 CL see cathodoluminescence clathrates 179-80 see also gas hydrates clay minerals 130, 131-4, 136 climate modulation 93-5 proxy 106 sensitive minerals 130 CMR see Combinable Magnetic Resonance Tool co-current imbibition 203-4 cold-water carbonate mounds 47-8 collaboration database 243 research 23-4 colour analysis see digital colour analysis colour logging 99-112 chronostratigraphy 102-3, 105 colour variability 105-9 D 13892 chronology 105 spectral analysis 102 uses 14, 15 XRF scanning 101-2 Combinable Magnetic Resonance Tool (CMR) 185, 187 compensators 167-8 Compton scattering 85, 86, 87, 8 8 - 9 0 computed tomography (CT) 18, 19, 165-78 see also portable X-ray computed tomographic system confocal macroscope-microscope luminescence imaging 19-20, 141-50 instrumentation 142-5 scan results 145-8 terminology 143 Conical-SPRITE methodology 202, 202 coralline aragonite 47-8 core analysis 21 core data digitizing methods 238-9 efficient graphical data entry 237, 238 end users 2 national and world data centres 243 participating institutions 242 stewardship 241-51 core recovery 21 core-logging systems development 7-8

262 core-logging systems (cont.) generator 234-5 HYACINTH 151-65 multi-sensor loggers 4, 8-12 non-destructive techniques 4-6, 7-20 CoreWall data visualization system 23-4 correlation coefficients 116 CORTEX XRF scanner 5, 39, 40, 42, 79 counter-current imbibition 204-5 Cretaceous-Tertiary boundary 12, 14, 16 CT scanning see computed tomography Cu/Rb profile 92 currents, Gulf of Cadiz 99-101 D13892 chronology 99-112 chronostratigraphy 103 GISP2 isotopes 109 tie points 105 Dansgaard-Oeschger cycles 45-6 data, accessibility 21-3 data management 248-50 data mining 232 data visualization 23-4 databases 22-3 computed core log 235 dbSEABED 229-40 EU-SEASED 244-5 EUROCORE 244-5 geospatial 250 Index to Marine and Lacustrine Geological Samples 241,243-8 integration 229-40 JANUS 253-9 stewardship 241-51 dating of sediments 81-2 dbSEABED database 23,229-40 data processing 232-3 input data 229-32 stratigraphic outputs 233-7 uncertainties 233 deep biosphere 21 Deep Sea Drilling Project (DSDP) 3, 7-8, 22 deep-sea exploration history 1-3 design Eagle III BKA system 32-4 logging-while-drilling 219-20 rapid non-contact resistivity logging 211 detection limits 42 diatom mat image 146 Dibden Bay, Southampton Water 19 digital colour analysis 113-28 applications 5 data acquisition and processing 117-20 materials and methods 114-20 scope and techniques 12-15, 18 sediment colour and composition 120-7 digitizing methods 238-9

INDEX dissociation experiments HYACINTH system 160 methane hydrate 172-7 disturbance from core retrieval 151-2 drilling Deep Sea Drilling Project 3, 7-8, 22 rig measurements, permafrost 186-90 see also Integrated Ocean Drilling Program; logging-while-drilling; Ocean Drilling Program DSDP see Deep Sea Drilling Project Dynamic Autoclave Piston Corer 152 Eagle III BKA system 31-7 analytical conditions 33-4 applications 35-7 design 32-4 specification 34 efficient graphical data entry 237, 238 electrical induction logging 209-17 electrical resistivity 10, 209 electromagnetic conductivity measurements 209-11 element ratios sapropels 68-74 turbidites 85-92, 87 elements, ITRAX scans 70, 72, 82-8 end users 2 equatorial Pacific sediments 134, 146 EU-SEASED database 244-5 EUROCORE database 244-5 Faeroe Drift, NE Atlantic 45 Faeroe-Shetland Channel 46-7 fast fourier transform (FFT) 198-9 Fe/Rb ratio 92 Feni Drift, NE Atlantic 44-5 FFT see fast fourier transform fluorescence see X-ray fluorescence scanners fluorescence intensity 149 FPC see Fugro Pressure Corer France, deep-sea exploration 2-3 Fugro Pressure Corer (FPC) 154-5 galvanic induction 209, 211,212 Gamma Ray Attenuation Porosity Evaluator (GRAPE) 4, 8 gamma-ray densiometry (GRD) 171-2, 173 gas hydrates assay and growth habit 186 nuclear magnetic resonance 180-5 occurrence 21 pressure logging 151-65 properties 179-80 quantitative assay 183 sea-floor measurements 185-6 see also HYACINTH system; methane hydrate genetic units determination 80, 81

INDEX geochemistry 1TRAX core scanner 65-77 NE Pacific sediments 115-17 sapropels 66-75 GEOSCAN camera 12, 101-2 geospatial databases 250 GEOTEK multi-sensor core logger applications 4 configuration 10 integration 216 vertical logging 9, 10, 159-60 Geowall project 23 Germany, deep-sea exploration 2 G1SP2 chronology 104 global accessibility 21-3 GoC see Gulf of Cadiz grain-size analysis 84, 91 GRAPE see Gamma Ray Attenuation Porosity Evaluator graphical core logs databases 232, 237, 238 turbidites 83, 84 GRD see gamma-ray densiometry grey reflectance 15 greyscale 99 see also lightness Guaymas Basin sediments 114, 115, 124 Gulf of Cadiz (GoC) sediments 99-112 GVR-6 tool 224, 225, 226 hardgrounds 47-8 high-resolution colour logging 99-112 core-logging systems 7 digital colour analysis 113-28 X-ray fluorescence analyser 31-7 historical development core-logging systems 8 deep-sea exploration 1-3 Holocene, colour variability 105-8 Hot Ice #1 methane hydrate research well 172-7 HPC see Hydraulic Piston Corer HRC see HYACE Rotary Corer HYACE see HYdrate Autoclave Coring Equipment HYACE Rotary Corer (HRC) 155 HYACINTH system 21, 151-65 components 158 development 153 dissociation experiments 160 downhole tools 153 ODP Leg 204 159-62 subsampling 157, 158 transfer and chamber systems 155-9 HYdrate Autoclave Coring Equipment (HYACE) 21, 153 see also HYACINTH system hydraulic permeability 184-5 Hydraulic Piston Corer (HPC) 7-8

263

ice assay and growth habit 187-9 quantitative assay 183 illite 130, 131-4, 136 illumination geometry 131 IMAGES see International Marine Past Global Changes Study imbibition 203-5 in situ pressures 21, 151 Index to Marine and Lacustrine Geological Samples 241,243-8 induction logging 209-17 information processing 229-40 see also databases infrared (IR) thermal imaging 14-15 see also visible and near-infrared reflectance spectroscopy inorganic geochemistry, sapropels 66-75 input data 229-32 institutions collaborating on databases 242 instrumentation ITRAX 54 macroscope-microscope system 142-5 non-destructive core logging 3-7 nuclear magnetic resonance 185 VNIS 130-1,132 Integrated Ocean Drilling Program (IODP) 3 VNIS 135, 137, 138 worldwide access to data 253-9 internal structures visualization 202-3 International Marine Past Global Changes Study (IMAGES) 3 internationalization of databases 244-5 Internet mapping 250 IODP see Integrated Ocean Drilling Program IR imaging see infrared thermal imaging ITRAX multi-function X-ray core scanner 17, 51-63 applications 6, 60-2, 92 conventional WD-XRF comparison 56, 57, 59, 60, 69 limitations 92 output parameters 61 sapropel geochemistry 65-75 materials and methods 66 sapropel SI 71-4 sapropel $3 74-5 scanning procedure 52-3 second-generation 57-60 specifications 51-7 turbidite emplacement 79-98 applications and limitations 92 element profiling 82-4 JANUS database 22-3,253-9 access 259 data flow 255, 256 data overview 258 development 253-5 enhancements 255-9

264 JOIDES

INDEX Resolution

logging-while-drilling 220-2 Ocean Drilling Program 3 portable CT scanner 165-6, 167, 171-2 pressure corers 153, 155, 160 K/Rb ratio 92 Lake Huron core 145 language parsing 232 light intensity see reflectance spectroscopy lightness 99, 103, 1 2 1 - 5 , 126 lithostratigraphy, Walvis Ridge 16 Llyn Gwernan, Wales 114, 115, 125 logging chambers 156, 160 tools 185 see also core-logging systems logging-while-drilling (LWD) 219-28 density measurements 173 ODP tests 219, 220-4 system design and testing 219-20 low-latitude Pacific clays 134 luminescence imaging 141-50 LWD see logging-while-drilling MAC see OMEGA Multi Autoclave Corer macroscope-microscope system 141-50 macroscopic magnetization 194-5 magnetic moment 194 magnetic resonance imaging (MRI) 19, 193-207 rock structures visualization 202-3 spontaneous imbibition 203-5 theory 194-202 see also nuclear magnetic resonance magnetic susceptibility cold-water carbonate mounds 48 mud volcano sediments 46 multi-sensor logging 9-11 Marine Isotope Stages 1-3, 79-98 MDCB see Motor Driven Core Barrel measurement chambers 157 Mediterranean Outflow (MO) 101 Mediterranean sediments sapropels 61-2, 65-77 turbidites 79-98 Meerfelder Maar core 36 metadata 249-50 methane hydrate core disturbance 152 dissociation experiments 172-7 occurrence 21 see also gas hydrates microscopes see confocal macroscope-microscope luminescence imaging Milankovitch variability 43-5 minerals reflectance spectroscopy 135-7

standards, VNIS 131-4 Minolta spectrophotometer 130 MO see Mediterranean Outflow molecular stratigraphy 20-1 Motor Driven Core Barrel (MDCB) 220, 221,226 MR see magnetic resonance MRI see magnetic resonance imaging MSCL-V see Vertical Multi-Sensor Core Logger MSCLs see multi-sensor core loggers mud volcano sediments 46-7 multi-function X-ray core scanners 17, 51-63 multi-sensor core loggers (MSCLs) 4, 8-12 national data centres 243 National Geophysical Data Center (NGDC) 241,243, 244-50 National Oceanic and Atmospheric Administration (NOAA) 243-4, 246 NE Atlantic sediments 39-50 NE Pacific sediments 114, 115, 122 NGDC see National Geophysical Data Center NMR see nuclear magnetic resonance NOAA see National Oceanic and Atmospheric Administration non-contact resistivity logging 209-17 non-destructive core logging techniques 4-6, 7-20, 51 non-imaging optical systems 5, 15-17 nuclear magnetic resonance (NMR) 19, 179-92 basic principles 180-1 gas hydrates 185-90 hydraulic permeability 184-5 logging tools 185 permafrost 182-5 porous media 181-2 see also magnetic resonance imaging Ocean Drilling Program (ODP) core data 22 data overview 258 development 3, 7 fluorescence intensity data 149 Hole 1249A 224, 225 Hole 1249B 223, 225 Leg 138 8 Leg 164 159 Leg 171 254 Leg 201 14 Leg 204 15, 159-62, 171-2, 173, 220-4 Leg 207 13 Leg 209 226 Leg 210 167 logging-while-drilling 219, 220-4 NE Pacific sites 114-15 portable X-ray CT scanner 159-62 pressure coring 152-3, 159-62 Site 1262 16 VNIS 137 worldwide access to data 253-9

INDEX OMEGA Multi Autoclave Corer (MAC) 152 on-site geological core analysis 165-78 opal biogenic content 115, 121, 123-4, 126-7 VNIS 130, 131-5, 136 optical systems, non-imaging 15-17 organic carbon colour analysis 115-17 digital colour analysis 120-7 VNIS 135 output-flow 239 OXCAL calibration program 102-3 oxygen isotope series 103, 104 Pacific sediments clays 134, 146 colour analysis 114, 115, 122 geochemistry 115-17 ODP sites 114-15 Palaeocene-Eocene boundary 12, 13, 14 Palmer Deep, Antarctic Peninsula 114, 115, 123 partnerships, systems design 250 PCS see Pressure Core Sampler pelagic sediments 82, 131-4 permafrost drilling rig measurements 186-90 nuclear magnetic resonance 179-92 relative permeability 189-90 permeability hydraulic 184-5 relative 189-90 petrophysical analysis 193 photography see cat, :ras; digital imaging photoluminescence ?L) imaging 145-6, 148 piston corers 2-3. -8 pixel values 14a PL see photolu .nescence planktonic for" ainiferal tests 101, 107, 109 point data 23" pore-scale in stigation 179-82 porosity Io~, ng applications 4 porous me. NMR I J-2 relaxat ,n theory 196-8 portable (,-ray computed tomographic system 165-78 medic ,1 CT system comparison 166 ODP Leg 204 171-2 system description 166-71 Pressure Core Sampler (PCS) 152-3, 159 pressure coring 151-65 see also HYACINTH system prosumer cameras 113-14 P-wave data 9 quantitative magnetic resonance imaging 193-207 quantitative micro-XRF analysis 59-60 RAB-8 see Resistivity-at-Bit tool

265

radiofrequency field 195 radiographs densiometric measurements 171-2, 173 with/without compensators 168-9 rapid non-contact resistivity logging 209-17 case examples 216 depth of investigation 212-16 design and performance 211-12 formation factors 215-16 galvanic induction comparison 209, 211,212 records of multi-sensor core loggers 11 reflectance spectroscopy 15-17, 129-42 see also visible and near-infrared reflectance spectroscopy relative permeability 189-90 relaxation theory 196-8 relaxation times 195-6 remotely operated sub-sea vehicle (ROV) 185-6 resistive layers 211-12, 213 resistivity imaging chambers 157 resistivity logging, non-contact 209-17 Resistivity-at-Bit (RAB-8) tool 220, 222-4, 225, 226 Rosemary Bank, NE Atlantic 43-4 ROV see remotely operated sub-sea vehicle sampling Avaatech XRF core scanner 40-1 data processing 232-3 Eagle III BKA system 32-3 HYACINTH system 157, 158 Santa Barbara Basin sediments 114, 121, 147 sapropels 61-2, 65-77 inorganic geochemistry, S 1 71-4 inorganic geochemistry, $2 74-5 ITRAX scanning methods 66 scanners see multi-sensor core loggers; portable X-ray computed tomographic system; X-ray fluorescence scanners Schlumberger Combinable Magnetic Resonance Tool 185, 187 Genesis rig 221 Resistivity-at-Bit tool 220, 222-4, 225, 226 Scotia Shelf specimens 148 sea surface temperature (SST) 101,104 sea-floor sediments data end users 2 exploration 1-3 gas hydrate measurements 185-6 large databases 229-40 second-generation ITRAX 57-60 sediments colour analysis 113-28, 129 composition 113-28, 129 mineralogy 129-42 see also Atlantic sediments; Pacific sediments; seafloor sediments Severn Estuary sediments 60-1 Shear Transfer Chamber (STC) 156

266 sidescan sonar data 100 single point imaging (SPI) 193, 200-2 smectite 130, 131-4, 136 software database 244 ITRAX 52-3, 58-9 southern Balearic Abyssal Plain 79-98 spatial encoding 198-201 spectral analysis, D 13892 102 spectrophotometer 10, 130 spectroscopy see reflectance spectroscopy Spiral-SPRITE methodology 201,202 SPM see suspended particulate matter spontaneous imbibition 203-5 SPRITE (single-point ramped imaging with Tl enhancement) 193, 200-2, 205-6 Sr/Ca ratio 91-2 SST see sea surface temperature static magnetic field 194 statistical stratigraphy 235-7 STC see Shear Transfer Chamber storage chambers 156 stratigraphy chronostratigraphy 102-3, 105 Cretaceous-Tertiary boundary 12, 14, 16 data 229-40 dbSEABED database 233-7 moleculer 20-1 multi-sensor core loggers 11-12 Palaeocene-Eocene boundary 12, 13, 14 statistics 235-7 Walvis Ridge 16 stress relief 151 sub-bottom location 231-2, 233 suspended particulate matter (SPM) 101 Sweden, deep-sea exploration 2 techniques, non-destructive 4-6, 7-20 telecentricity 144 terminology, macroscope-microscope imaging 143 thawing of permafrost 188-9 Ti/Rb ratio 92 time requirements, logging-while-drilling 227 titanium Eagle Ill BKA system 35 XRF analysis 17-18 total organic carbon (TOC) 115-17, 120-7 transfer systems 155-6 turbidite emplacement 79-98 climatic modulation 93-5 dating the sequence 8 I-2 early arrivals 92-3 element concentration 84-5 element profiling 82-8

INDEX emplacement time and bed thickness 93-5 grain-size analysis 84, 91 graphic logs 83, 84 internal subdivisions 95 measured integrals and ratios 85-92 sources 95 UK, deep-sea exploration 1-2 uncertainties 233 unfrozen water 182-3 unit data 231 variability of colour 105-9 VariSpot X-ray focusing system 34 VDN tool see VISION Density Neutron tool vertical logging 9 Vertical Multi-Sensor Core Logger (MSCL-V) 159-60 vibrocore core-log format 230 visible and near-infrared reflectance spectroscopy (VNIS) 129-42 illumination geometry 131 instrumentation 130-1,132 mineral calculation 135-7 pilot studies 131-5 water effect 131 VISION Density Neutron (VDN) tool 224, 226 Walvis Ridge, SE Atlantic 16 water content Palmer Deep sediments 126-7 reflectance spectroscopy 131 wavelength dispersive X-ray fluorescence (WD-XRF) 56, 57, 59, 60, 69 wireline logging 227 wireline-operated pressure corers 153-4 world data centres 243 X-ray computed tomography applications 6, 19 development 18 portable CT core scanner 165-78 X-ray fluorescence scanners 17-18 applications 5-6 Avaatech scanner 39-50 colour analysis 101-2 CORTEX scanner 39, 40 Eagle III BKA system 31-7 ITRAX multi-function scanner 17, 51-63, 67-75, 79-98 Younger Dryas, D13892 chronology 99-112 Zr/Rb ratio 92

New Techniques in Sediment Core Analysis Edited by

R.G. Rothwell

Marine sediment cores are the fundamental data source for information on seabed character, depositional history and environmental change. They provide raw data for a wide range of research including studies of climate change, palaeoceanography, slope stability, oil exploration, pollution assessment and control, seafloor survey for laying cables, pipelines and construction of seafloor structures. During the last three decades, a varied suite of new technologies have been developed to analyse cores, often non-destructively, to produce high-quality, closely spaced, co-located downcore measurements. These techniques can characterize sediment physical properties, geochemistry and composition in unprecedented detail. Palaeoenvironmentally significant proxies can now be logged at decadal, and in some cases, annual or sub-annual scales, allowing highly detailed insights into climatic history and associated environmental change. These advances have had a profound effect on many aspects of the Earth Sciences and our understanding of the Earth's history. In this volume, recent advances in analytical and logging technology and their application to the analysis of sediment cores are presented. Developments in providing access to core data and associated datasets, and advances in data mining technology in order to integrate and interpret new and legacy datasets within the wider context of seafloor studies are also discussed.

Visit our online bookshop: http://www.geolsoc.org.uWbookshop Geological Society web site: http://www.geolsoc.org.uk

Thanks to Cox AnalyhcaLSystemsof Gothenberg, Sweden for generouscontrlbuhons towards colouf pnnhng costs

Cover illustration: A virtualrealityvisualizationof sedimentcoresin the Ship Shoalarea, Louisiana,USA,generatedby dbSEABED(seethis volume).The topographic surfaceextendsfrom coastalwetlandsto 40 m waterdepth (green to blue shading).Unitsin the coresare classedgravel,sandor mud (red, yellow,green).Suchvisualizationassistssand resources mapping for Louisianacoastalprotection.Changesto the shapeof the seabedbetweensurveys,or datum differences,causesomecoresto protrude abovethe griddedseabedtopography. Courtesyof Universities of Coloradoand New Orleansand the US GeologicalSurvey.