A brief review on modelling sediment erosion, transport and deposition by former large ice sheets

MARTIN J. SIEGERT*

*School of GeoSciences, Grant Institute, University of Edinburgh, West Mains Road, Edinburgh EH9 3JW, UK (e-mail: m.j.siegert@ed.ac.uk)

ABSTRACT

Although numerical ice sheet models have been used often to predict the size and dynamics offormer ice sheets, few exercises have utilised the geological record to fully constrain model out-put. Geological evidence can be used to build detailed hypotheses regarding ice sheet and climatehistory, which can be tested by computer models provided ice flow is coupled to sediment erosion,transport and deposition. Here three examples of how geological data have been used in con-junction with coupled ice-sheet/sediment modelling to comprehend large-scale glacial history arediscussed. The first describes how numerical reconstructions of the late glacial Eurasian Ice Sheethave benefited from marine geophysical surveys quantifying sediment fans along the former icemargin. The second reviews how models have been used to determine the likely accumulationsof sediments from the initiation and growth of the Antarctic Ice Sheet. The third discusses develop-ments in modelling techniques that allow detailed predictions of sediment deposits and their evolution through time. In addition, a note is provided on how modelling can be used to provideprocess information concerning former ice sheet instabilities, which can lead to marine sedimentaryrecords such as Heinrich layers.

Keywords Ice sheet, numerical modelling, sediment, subglacial, erosion.

INTRODUCTION

Glaciers and ice sheets have long been recogn-ised as important agents of erosion and deposition.Much of our understanding of ice sheet behavi-our comes from the study of glacial geology (e.g.Hambrey, 1994). While quantification of glacialgeologic processes has been considered for sometime (Drewry, 1986), the ability to fully comprehendlarge-scale glacial sedimentary systems and, fromthis, glacial history was restricted until the adventof numerical modelling techniques (and recentadvances in computer technology that make mod-elling of large systems possible).

Ice sheet modelling has progressed over the last decade into a sub-discipline of glaciology withconsiderable cross-disciplinary importance (e.g.with contributions to the IPCC reports on climatechange and to integrated models of the Earth system). During the 1990s, the European Science

Foundation (ESF) European Ice Sheet ModellingInitiative (EISMINT) programme allowed inter-comparison between ice sheet models and set upa series of ‘benchmarks’ to which models can betested (Huybrechts et al., 1996). The outcome of EISMINT is an appropriate level of regulation in ice sheet modelling activity. This service to the scientific community has been developed furtherwith the generation of freely available ice-sheetmodelling software, such as GLIMMER (GENIELand Ice Model with Multiply Enabled Regions,http://glimmer.forge.nesc.ac.uk).

Many ice sheet models incorporate calcula-tions of basal temperature and water production. Such models, when coupled with models of sedi-ment erosion and transfer beneath ice sheets, arecapable of testing glaciological hypotheses regard-ing ice sheet history, derived from geological data.In this review some of the recent advances in ourknowledge of ice sheet dynamics and histories,

GSP_4_C05.qxd 11/5/07 13:21 Page 53

Glacial Sedimentary Processes and Products Edited by Michael J. Hambrey, Poul Christoffersen, Neil F. Glasser and Bryn Hubbard © 2007 International Association of Sedimentologists. ISBN: 978-1-405-18300-0

54 M.J. Siegert

made possible through coupled ice-sheet/sedimentmodels, are highlighted. The aim of the paper isto collate and compare previous research and toidentify ways forward for future activities.

BACKGROUND TO SEDIMENT MODELLING IN ICE SHEETS

Ice sheets are capable of eroding, transporting and depositing huge quantities of sediment. Forexample, marine geological evidence shows that the Bear Island Fan, on the continental shelf breakwest of the Barents Sea, received material at a rateof over 1 m per 1000 years during the last glacia-tion as a consequence of sediment delivery by theEurasian Ice Sheet (Laberg & Vorren, 1996). Themechanism by which the material was transportedis likely to be through the deformation (and there-fore motion) of the glacier bed as other processes,such as entrainment within basal ice, are difficultto reconcile with the volumes and rates measured.The process of subglacial bed deformation as ameans by which ice sheets may flow has beenstudied for around thirty years, since this ‘paradigmshift’ was first detailed by Boulton and Jones (1979),and later quantified by Boulton & Hindmarsh(1987). The notion that large ice sheets can growon sedimentary material which, if water saturated,has little strength caused many glaciologists toconsider the very stability of such ice masses (e.g.MacAyeal, 1992; Clark, 1994). These studies madethe link between glacial process and the geologicalrecord critical to evaluating quantitative ice sheethistories. The following is a simple method, usedin several ice sheet models (and as a basis formore sophisticated approaches), by which ice sheetmotion is linked with basal sediment transport.

Most ice sheet models are centred on the con-tinuity equation for ice (Mahaffy, 1976), where thetime-dependent change in ice thickness is associatedwith the specific net mass budget:

= bs(x, t) − ∇.F(u, H) (1)

where F(u,H) is the net flux of ice (m2 yr−1) (the fluxof ice being the product of ice velocity, u, and icethickness, H). The depth-averaged ice velocity, u(m s−1), is calculated by the sum of depth-averaged

dHdt

internal ice deformation and basal motion (slidingand/or bed deformation). The specific mass bud-get term bs, is usually a function of a number ofprocesses including ice sheet surface mass balance(accumulation and ablation) and, where necessary,iceberg calving and ice shelf basal melting. The con-tinuity equation can be applied to a grid of cells(in the case of finite difference schemes) or a net ofnodes (in the case of finite elements). In either case,the equation is used to calculate the net change in ice thickness about an ice sheet through under-standing ice flow between cells or nodes.

Alley (1990), in one of the first modelling studiesof ice-sediment interaction and from which manysubsequent models have been based, described thevelocity due to the deformation of water-saturatedbasal sediments, ub (m s−1), as:

ub = hbKb (2)

where Kb is a till deformation softness, hb is thedeforming till thickness (m), and N is the effectivepressure (Pa). The till yield strength, τ*, is:

τ* = N tan(φ) + C (3)

where C is a till cohesion coefficient, and tan(φ) is a dimensionless glacier-bed friction parameter.Equation 3 assumes that subglacial sediment be-haves as a plastic substance (Boulton & Hindmarsh(1987) refer to it as a nonlinear Bingham fluid). Fieldobservations of sediment deformation have rein-forced the opinion that basal sediments deformplastically (e.g. Clarke, 2005), but incorporatingsuch flow effectively into ice flow models is prob-lematic, due to the nonlinearity between stressand strain rates. Choosing an appropriate rheo-logy for subglacial sediments is, clearly, critical to coupled ice-sheet/sediment models, however, andBougamont & Tulaczyk (2003) and Bougamont et al. (2003) have produced ice-sheet models inwhich a plastic rheology is used. Such modellingmay well mark a way forward for future work.

Sediment can be eroded as well as transported,and Alley (1990) considered this to be controlled by:

t = Kt N (4)ub

hb

(τb − τ*)N2

GSP_4_C05.qxd 11/5/07 13:21 Page 54

A brief review on modelling sediment erosion, transport and deposition by former large ice sheets 55

where t is the thickness of till produced in a yearand Kt an abrasion softness (3 × 10−9 m Pa−1). One keyelement to the calculation is the effective pressure,which can be assumed as the difference between iceoverburden pressure and the basal water pressure(i.e. as water pressures increase, the effective pressuredecreases, increasing sediment deformation).

The strength of subglacial deforming sedimentneeds to be known for two reasons. First, it dictatesthe rate of flow of material and, second, it controlsthe depth to which deformation can occur. Thisinformation is needed to understand the transportof sediment. Provided the sediment has at leastsome integral strength, its deformation will lead toa flux of material in the direction of ice flow. Thereare two ways in which this flux can be calculated.The simplest is to assume a Eulerian system (Presset al., 1989), in which sediment flow can be calcu-lated by a continuity equation applied to the samegrid as used to determine ice flow. Such a systemis easy to comprehend and simple to implement(e.g. Dowdeswell & Siegert, 1999). A more sophis-ticated system involves a Lagrangian approach(Press et al., 1989), where the ice flow model is used to establish sediment flow vectors, which arethen used independently to the ice flow model topredict sediment movement. This technique has an advantage of ensuring mass conservation andavoiding numerical diffusion (e.g. Hagdorn &Boulton, 2004), and can be used to track individualpackets of material through time and space.

Ice sheets can also erode and transport sedimentby entrainment of material into their basal layers.The processes controlling sediment entrainment arecomplex, however, and not easily quantified. Con-sequently, very few numerical modelling studieshave attempted to build a comprehensive under-standing of glacial sedimentary processes (e.g. Tully,1995; Hildes et al., 2004). One way forward on thisissue has been established by Christoffersen et al.(2006), who have shown how ice accretion at IceStream C, West Antarctica, is a dominant mechan-ism by which sediment is incorporated into the icesheet and, subsequent, transported.

MODELLING SUBGLACIAL SEDIMENTS

Coupled ice-sheet/sediment models have beenused to help understand the connection between

ice sheet history and the geological record in sev-eral regions. Three examples of how ice-sedimentmodels have been used to study former ice sheetsare described below: the first investigation providesan assessment of how the geological record can beused to constrain ice sheet history (Dowdeswell &Siegert, 1999); the second shows how modelling canbe used to determine where sediment build up maybe expected (Pollard & DeConto, 2003); and the thirddemonstrates a way in which the practice can bedeveloped in future (Hildes et al., 2004).

Eurasian Ice Sheet

Since the Late Cenozoic (2.5 million years ago), aseries of submarine sedimentary fans has developedalong the continental margins of the Norwegian-Greenland Sea (Dowdeswell et al., 1996; Vorren et al., 1998). The fans are distinct depocentres, beinglocated at the mouths of bathymetric troughs andseparated from each other by essentially sediment-free zones (Fig. 1). They are also very large; the Bear Island Fan (the largest) being in excess of280,000 km2 in area. The major source of sedi-ments to the fans is from large continental ice sheetsthat exist during periods of glaciation. Marinegeophysics has demonstrated that sediment isdelivered to the tops of the fans, from where theyare distributed across the wider fan surface bygravity-driven sliding and slumping processes (e.g.Dowdeswell et al., 1996). The rate of sedimenta-tion is restricted to periods when the ice sheet terminates at the shelf break (about 10% of each100,000 year glacial cycle). Rates of sediment supply are thought to be very high during theseactive phases (Laberg & Vorren, 1996, estimate over100 cm per 1000 years across the entire Bear IslandFan). There is an order of magnitude less sedi-mentation when the ice sheets are either restrictedto the continental shelf interior or, as is the case dur-ing interglacials such as now, absent.

Dowdeswell & Siegert (1999) adapted Alley’s(1990) simple ice-sheet/sediment model to predictice sheet history by ‘matching’ model results to geophysical data from the continental margin fanson the eastern side of the Norwegian-Greenland Sea. In this case, the ‘match’ between model resultsand observations was undertaken qualitatively.Model output was generated and compared againstknown records of ice sheet size (such as moraine

GSP_4_C05.qxd 11/5/07 13:21 Page 55

200 km

POLAR

NORTH

ATLANTIC

BearIsland

Fan

NorthSeaFan

Deep oceandelivery(km3/kyr)

500 5000

Sediment flux(km3/kyr/km

ice front)

0.37 0.70

300300300

2100

2100

2100

240024002400

180018001800

1500

1500

1500

900900900

000

1200

1200

1200

600600600

300300300

800800800

600600600

400400400

800800800

500500500

Fig. 1 Measurement and modelling of sediment supply to the western Eurasian margin. (A) The Eurasian Ice Sheet atthe LGM. Contours are ice thickness (in m) and arrows denote flow directions, with ice margin velocities noted (inmetres per year). Taken from Siegert et al. (1999). (B) The positions of major bathymetric trough mouth fans as measuredby side-scan sonar, chirp sonar and seismic methods. (C) Rates of sediment supply as calculated by numerical modellingand as measured from geophysical data. The location of (D) the Bear Island Trough Mouth Fan and (E) the North SeaFan are provided on page 57, in relation to enhanced ice flow as calculated by the ice sheet model illustrated in (A).

GSP_4_C05.qxd 11/6/07 9:05 Page 56

A brief review on modelling sediment erosion, transport and deposition by former large ice sheets 57

positions and uplift data). Once the results showedsome agreement with these records (gauged througha sensitivity analysis of climate input parameters)a comparison between output and independentgeological data (e.g. sediment fans) could then bemade. The model calculated the spatial distribu-tion of subglacial sediment over the Eurasian HighArctic, the volume of sediment which accumulatesat the shelf edge at the mouths of bathymetrictroughs, and the time-dependent variations in therate of sediment supply to the continental shelf and margin.

The modelling revealed that the most recentglaciation of the Eurasian High Arctic occurred after28,000 years ago, and that ice streams within bathy-metric troughs were active by around 25,000 yearsago. The maximum ice thickness over the BarentsSea exceeded 1400 m at around the LGM. At thistime, ice extended to the shelf break along both thewestern Barents Sea margin and the Arctic Oceanmargin north of the Barents and Kara seas. Icestreams draining the Barents and Kara seas werepresent within most major bathymetric troughsduring full-glacial conditions. A sedimentation rateof 2–4 cm yr−1 was predicted along the mouth of theBear Island Trough between 27,000 and 14,000 yearsago (Fig. 1). This is equivalent to 0.07 to 0.13 cmyr−1 averaged over the entire fan. Similarly, highdelivery rates of 2–6 cm yr−1 of glacial sediments(equivalent to 0.2–0.6 cm yr−1 averaged over the fan)were predicted between 27,000 and 12,000 years ago at the mouth of the Storfjorden Trough southof Spitsbergen. The modelled volumes of sedi-ment that accumulate at the continental margin of

the Bear Island and Storfjorden troughs (4,600 km3

and 900 km3) are similar to the volumes of LateWeichselian sediment measured over the respect-ive fans using seismic methods (4,200 km3 and 700 km3) (Fig. 1). The model also predicted thatmajor glacierfed fan systems would have built up on the northern, Arctic Ocean margin of theBarents and Kara seas, particularly on the contin-ental slope adjacent to the St. Anna and Franz-Victoria troughs.

Ice-sheet decay affected the marine portions of the ice sheet after 15,000 years ago, leaving a northern ice mass between Svalbard and FranzJosef Land which decayed after 13,000 years ago.The transition from a continental-wide ice sheet toone restricted to the northern Barents Sea resultedin a series of sea-floor sediment deposits, thathave been measured by marine geophysical surveysaround the northwestern shallows.

The modelling demonstrated that ice streams inthe Eurasian Arctic were influenced by basal motion(in this case a deforming sediment layer, but it isacknowledged that sliding may also be importantto ice stream development), and that the process ofsediment deformation led to the delivery of hugequantities of material to the shelf break (Fig. 1).Given the similarity between the Eurasian Ice Sheetand the West Antarctic Ice Sheet, and that deform-ing sediment is thought to control ice streamdynamics, one might also expect large volumes of sediment to be sent to ice stream mouths and,when the ice sheet expanded during full glacials,to the shelf break. Such analogy may be import-ant when attempting to understand glacial history

km/yr

0

1

100 km

3000 m2200 m

1500 m

200 m

BearIsland

Fan

BjørnøyaTrough

500 m

km/yr

0

1

100 km

1500 m

500

m

3000 m

200

m

NorthSeaFan

NorwegianTrough

300

2100

2400

1800

1500

900

0

1200

600

300

800

600

400

800

500

Fig. 1 (cont‘d)

GSP_4_C05.qxd 11/6/07 9:05 Page 57

58 M.J. Siegert

from the marine geological record in the Ross Sea,such as in the forthcoming Antarctic drilling pro-gramme (ANDRILL, http://andrill.org/). Clearly,ice-sheet/sediment modelling, of the type under-taken in the Eurasian Arctic, can provide valuableinsights into marine ice sheet dynamics respons-ible for the accumulation of continental shelf andshelf break sedimentary records.

In recent years, Hagdorn & Boulton (2004) havedeveloped a semi-Lagrangian model in whichsediment erosion, transport and deposition inScandinavia can be tracked over consecutive glacialepisodes. The modelling is particularly powerfulas sediments from individual glaciations can be monitored through initial transportation anddeposition, to reworking and redistribution. Theresults from such work can be used to understandthe glacial history within complex glacial geologicsequences involving sedimentary material frommultiple origins.

Antarctic Ice Sheet

The formation of the Antarctic Ice Sheet representsa key episode in the climate history of the Earth.DeConto & Pollard (2003) used a numerical ice-flow model, semi-coupled with a GCM climatemodel, to demonstrate that the onset of the ice sheet,known to be around 34 million years ago, was an inevitable consequence of atmospheric CO2

lowering. This result is in contrast to the tradi-tional view of ice sheet initiation in conjunction with the opening of the Drake Passage, so isolat-ing the Antarctic continent climatically with thedevelopment of the circumpolar current. While thisopening provided a ‘trigger’ for ice sheet build-up,according the model subsequent CO2 decline meantthat Antarctic Ice Sheet formation was inevitable.Modelling the early glacial history of Antarctica was continued by Pollard & DeConto (2003),through the incorporation of an Alley (1990) typesediment model to their existing ice-climate model,so allowing the erosion, transport and deposition ofglacigenic material to be predicted. As the infantAntarctic Ice Sheet would have had an ablating margin, their model also included algorithms todescribe fluvial transport and deposition to the con-tinent margin. The results of the model reveal howthe presence of deforming basal sediments causethe glacial-interglacial signal of ice sheet initiation,

growth and retreat, to be amplified due to the reduc-tion in basal drag and consequent increases in iceflow velocities (so resulting in more rapid andcomplete decay). The sediment component of themodel allowed a determination of where to expectmajor deposits of material (Fig. 2), which might be useful in planning future data acquisition pro-grammes. Most material collects at just a few loca-tions, at the mouths of topographic and bathymetrictroughs. Consequently, the development of largefans, similar to those observed and modelled in theEurasian Arctic, should be expected in Antarctica(Fig. 2). Given the period of Antarctic glaciation,and the rates of sedimentation predicted (in somecases over 10 m per 1000 years) such depocentresmay be considerable in size.

One place where Pollard and DeConto predictsignificant build-up of material is around Prydz Bay (offshore of the Lambert-Amery system). Toinvestigate the sedimentological and ice sheet his-tory of this region further, Taylor et al. (2004) usedan ice-sheet model to evaluate the glaciologicaland topographical requirements for the formationof this sedimentary fan. The Lambert-Amery sys-tem is the largest drainage pathway of the EastAntarctic Ice Sheet, from which a large amount of onshore and offshore geological evidence isavailable. The records show that an ice stream wasestablished in Prydz Bay during late Miocene–early Pliocene time, in association with the majoroffshore depocenter. The ice sheet subsequentlywithdrew to its present position, leaving the AmeryIce Shelf to cover much of the trough. Taylor et al.(2004) revealed, using numerical modelling, that bed morphology change was probably responsiblefor driving changes in both ice-sheet extent anddynamics in the Lambert-Amery system at PrydzBay. Changes in bathymetry, caused by sedimenterosion is required for shelf-edge glaciation and correlates well with the Prydz Channel fan sedi-mentation history inferred by geophysical methods.This association suggests a feedback between sedi-ment erosion and glaciation, whereby the currentgraben is cut to encourage an ice stream responsiblefor the point-sourced fan development.

Laurentide Ice Sheet

The Late Quaternary ice sheets of North Americahave been investigated for many decades because

GSP_4_C05.qxd 11/5/07 13:21 Page 58

A brief review on modelling sediment erosion, transport and deposition by former large ice sheets 59

of the well preserved glacial landforms and pro-glacial lake sequences that now litter the land sur-face. A comprehensive analysis of these geologicaldata allowed Dyke & Prest (1987) to form a con-ceptual model of Late Quaternary ice sheet history,involving maps of ice sheet extent through time, andan appreciation of ice flow direction. Subsequent

analysis of satellite imagery of glacial landforms has added to the identification of former flowdirections and their development in time (Boulton& Clark, 1990; Stokes & Clark, 2001). The acknow-ledgement that much of these former ice sheets layover unconsolidated sediment (Alley, 1991) led tothe finding that much of the southern margin of the

(A)400 ka

400 ka

Sed

imen

t Flu

x (m

/ka)

2010

30

meters

2500

2000

1500

1250

1000

750

500

250

100

50

0

a

(B)

Fig. 2 Modelling sediment transportand deposition by the Antarctic IceSheet during a 400,000 year period of its Eocene-Oligocene phase oforbitally-forced growth and decaycycles (~34 million years ago). (A) Average ice thickness of, and (B)sediment deposited by, the ice sheetover the period of the model run.Taken from Pollard & DeConto(2003).

GSP_4_C05.qxd 11/5/07 13:21 Page 59

60 M.J. Siegert

ice sheet was subject to unstable flow (e.g. Clark,1994). In addition, discrete layers of ice rafted debrisdiscovered in the North Atlantic by Heinrich (1988)have been interpreted by many scientists to beevidence for periodic fast flow of the Laurentideice sheet in the Hudson Bay region (Bond et al., 1992;Alley & MacAyeal, 1994; Dowdeswell et al., 1995).Each of these insights to the history and dynamicsof the Laurentide ice sheet, gained from analysis ofglacial geology, has been supplemented by numer-ical modelling investigations.

Much has been learnt about the dynamics of theLaurentide and Cordilleran ice sheets from numer-ical modelling. Marshall et al. (2002) produced themost comprehensive series of model experiments,which allowed an appreciation of ice sheet volumethrough time (with a conclusion that as much a 80 m worth of global sea-level change was storedin the Laurentide ice sheet, making it the world’slargest ice sheet at the LGM), the production anddistribution of meltwater (supraglacial, englacial and subglacial) and the potential influence of basalwater on ice sheet dynamics (Marshall & Clarke,1999). These ice sheet modelling studies formed thebackground to Hildes et al.’s (2004) highly detailednumerical investigation into glacial sediment trans-port and distribution in glacial North America.Rather than simply relating the erosion of sedi-mentary material to basal motion, and the transferof this material through a deforming basal layer (as in most previous models of sediment transferin ice sheets), Hildes et al. (2004) designed a modelto account for a full variety of glacial sediment-ary processes. They included erosion of sedimentby following and developing the model of Tully(1995), which accounts for the process of ice sheetabrasion (englacial basal clasts scratching the sub-glacial bed) over a variety of lithologies and by theexcavation of blocks through the quantification of crack propagation in basal rock. Supraglacial sedimentary processes were ignored in the model,but this is justifiable given the fact that large icesheets have very little exposures of rock abovetheir surface from which material may fall. As inMarshall et al. (2002) the model included a ther-modynamic ice flow component, which allowedcoupling between the ice sheet and hydrology(including sheet basal flow and groundwatermovement). An important input to the model wasa depiction of the surface geology (both hard rock

and loose material), from which the model was ableto predict the development of sediment sources and sinks, and their effect on ice dynamics (Fig. 3).While the model of Hildes et al. (2004) is certainlythe most comprehensive attempt to quantify glacialerosion, transport and deposition, the results failedto correspond well with the known distribution ofglacial material. Nonetheless, the model representsa benchmark in attempts to fully quantify glacialsedimentary processes, and is a new base fromwhich further studies can be developed.

MODELLING THE SEDIMENTOLOGICALCONSEQUENCES OF UNSTABLE ICE FLOW

Marine sedimentological investigations have re-vealed a series of ice rafted debris deposits acrossthe North Atlantic corresponding to the 7000 yearperiodic production of huge volumes of icebergs,named Heinrich layers (Heinrich, 1988; Bond et al.,1992). Heinrich layers are thought to have formedby the periodic unstable flow of the Laurentide IceSheet during the last glaciation (MacAyeal, 1993),as follows. From a relatively stable configuration,the ice sheet slowly builds up over an essentiallyfrozen base. The ice sheet becomes larger and subglacial temperatures rise as a consequence ofextra insulation from the cold air on the ice sur-face and extra heating induced by the deformationof ice. Eventually the basal temperatures reach thepressure melting. As the ice base becomes wet, subglacial sliding and the deformation of subglacialsediments can occur, which for the Laurentide Ice Sheet means enhanced rates of ice flow to themargin. Much of the ice is drained through a single outlet, the Hudson Strait, which becomes the focal point for iceberg release to the NorthAtlantic. The drainage of so much ice depletes thereserves in the parent ice sheet and it becomes thinner. As it does so, heat is lost from the ice sheetbase and, eventually, the ice becomes refrozen tothe bed, so reducing the flux of ice to the margin.Once enhanced ice velocities have been curtailedice-sheet re-growth occurs.

Payne (1995) and Payne & Donglemans (1997)showed, using numerical modelling, how thebuild-up and decay of large ice sheets has a strongcontrol on the subglacial thermal regime. Under relatively stable external forcing conditions, an ice

GSP_4_C05.qxd 11/5/07 13:21 Page 60

A brief review on modelling sediment erosion, transport and deposition by former large ice sheets 61

00

00

00

15001500

15001500

15001500

25002500

25002500

25002500

350035003500

350035003500

350035003500

450045004500

450045004500

450045004500

Initial sediment depth

Normalized water pressure

Mean total erosion = 0.41

0 5 10 15

0 2 3 41 5 6 7

1

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0

95°W 85°W75°W

65°W

55°W

45°W

77°N72°N

67°N62°N

57°N52°N

47°N42°N145°W

135°W

125°W115°W

105°W

110°W90°W

70°W

70°N

60°N

50°N

40°N

150°W

130°W

100°W110°W 90°W80°W

70°W

60°W

50°W

70°N

75°N

80°N

60°N

65°N

50°N

55°N

40°N

45°N

140°W

150°W

120°W

130°W

Fig. 3 Numerical modelling of glacial sediment erosion in NorthAmerica between 60,000 and 10,000years ago. (A) Sediment distributionprior to ice sheet growth (in metres).(B) Distribution of basal water at theLGM (the figure denotes normalizedwater pressure). (C) Total amount ofsediment erosion after glaciation (in metres). The approximate position of modelled ice thickness contours(given in metres) are superimposedon each panel. Adapted from Hildeset al. (2004).

GSP_4_C05.qxd 11/5/07 13:21 Page 61

62 M.J. Siegert

sheet may oscillate between periods of ice growthwhen the ice is cold-based, and ice decay caused bythe attainment of warm-based thermal conditionsand the consequent initiation of rapid basal motion.By utilising a thermo-coupled time-dependent icesheet model, Payne (1995) highlighted the import-ance of basal sliding to the oscillatory behaviourof ice sheets. In his model, Payne showed how theaccumulation rate of ice governed the periodicityof ice sheet oscillations. For low accumulation rates,such as those experienced over the Laurentide icesheet, periods of between 6000 and 7000 yearswere predicted.

One problem associated with the notion ofunstable oscillations being responsible for forma-tion of Heinrich Layers is the method of sedimententrainment of material within the ice sheet dur-ing the enhanced flow phase, when most icebergsare produced. The problem is that a surge-type situation will involve the ice sheet effectivelydecoupled from the glacier bed, which may notallow the take-up of material into the basal ice layers. However, Alley & MacAyeal (1994) estab-lished a model that allows both entrainment andsurging to take place, which results in the produc-tion of sediment-laden icebergs. The model assumesthat the start of the purge phase involves enoughfrictional heat to melt the ice base, lubricating thesediments and causing them to deform. As this happens the ice stream begins to thin, which resultsin colder temperatures at the ice base and sowater and sediment freeze onto the base of the icestream. Beneath the refrozen basal ice the sedimentis still warm and water-saturated, however, and so the ice stream remains active and fast flowing.When all the water-soaked sediment has frozen onto the ice stream base it becomes ‘cold-based’ and the purge phase ends. This soft-bed model pro-duces a volume of sediment within the ice sheetthat is larger than required to account for theHeinrich layers. However, a hard-bed model wouldproduce far less. So, the presence of both hard and soft bed conditions would seem to make themodel fit the measurements. Although this modelremains unproven, Alley and MacAyeal (1994)suggest a number of good reasons as to why alternative entrainment mechanisms are less likely.For example, if the process of sediment entrain-ment was through ice-tectonics such as foldingand cavitation, a rate of uptake of material far in

excess of that calculated and measured in modernice sheets and glaciers would be required to formthe Heinrich Layers. Further, a pressure inducedbasal freezing process results in two orders ofmagnitude less sediment than is required.

SUMMARY AND FUTURE DEVELOPMENTS

Coupled ice-sheet/sediment models have beenused to understand the glacial history of severalformer ice sheets, both during the last glaciationand in deeper time. The results of these exerciseshave been compared, with differing success, with thegeological record of past glaciation. The simplestmodels, in which subglacial sediment transfer isrelated to basal motion, applied to marine settings,predict the build-up of large volumes of materialin front of former ice streams. On the margin of theBarents Sea, there is good geological evidence insupport of such deposition. When the sophistica-tion of the model increases, and is applied to a ter-restrial setting, the success in matching the knownglacial geology is reduced. The problem of such an approach may be that the processes by whichmaterial is entrained and moved by ice sheets iscomplex and not easily transferred to traditional ice sheet models. Nevertheless, as pointed out byHildes et al. (2004), the objective of comparing com-puter model results with the sedimentary recordis essential for ‘advancing Quaternary science’.

One area in which this advance may take placeinvolves investigations into glacial sedimentarylandform development. By comprehending glacio-logical requirements for the formation and pre-servation of known landforms, information on pastice sheets, at a local scale, could be formed. Muchwork has been undertaken on the identification andmeasurement of glacial landforms, and qualitativeassessment of their implications for ice sheet his-tory (e.g. Stokes & Clark, 2001). Furthermore, someattempts have been made to model the small-scaleprocess responsible for landform development (e.g.Hindmarsh, 1999). Future work may involve usingthis relatively small-scale evidence, in combinationwith other data, to form detailed appreciation oflarge-scale glacial history.

In addition, models capable of a time-dependentanalysis of sediment erosion, transport and deposi-tion, over more than one glacial cycle, provide

GSP_4_C05.qxd 11/5/07 13:21 Page 62

A brief review on modelling sediment erosion, transport and deposition by former large ice sheets 63

a valuable means by which geological features,formed and influenced by more than one glacialevent, can be assessed. The work of MagnusHagdorn in the development of such models hasbeen particular important on this issue (e.g. Boultonet al., 2003).

Several new scientific programmes are con-figured to assist the development of these modelsand to maximise their potential in understand-ing ice-sheet processes. For example, the AntarcticClimate Evolution (ACE, www.ace.scar.org) pro-gramme of the Scientific Committee on AntarcticResearch (SCAR) aims to use the predictive abilityof numerical models to test hypotheses developedfrom the geological record. In doing so, ACE willlink the numerical modelling community with geo-logists. Clearly model development is required in conjunction with data collection. The next fewyears, which will see the International Polar Yearand well-established scientific programmes (suchas the Antarctic drilling programme, ANDRILL, andthe integrated ocean drilling programme, IODP) islikely to witness this dual phase advance. Finally,as a follow up to EISMINT, a new ‘ice sheet modelintercomparison project’ (ISMIP) has been estab-lished. One of ISMIP’s three themes concerns theincorporation of sediments models, and the role ofsediment on unstable ice flow. Results from this project are expected over the next few years.

ACKNOWLEDGEMENTS

Dr. Chris Stokes and Dr. Poul Christoffersen are thanked for their helpful and constructive recommendations.

REFERENCES

Alley, R.B. (1990) Multiple steady states in ice-water-tillsystems. Ann. Glaciol., 14, 1–5.

Alley, R.B. (1991) Deforming bed origin for southernLaurentide till sheets? J. Glaciol., 37, 67–76.

Alley, R.B. and MacAyeal, D. (1994) Ice-rafted debris asso-ciated with binge/purge oscillations of the LaurentideIce Sheet. Palaeoceanography, 9, 503–511.

Bond, G., Heinrich, H., Broecker, W., Labeyrie, L.,McManus, J., Andrews, J., Huon, S., Jantschik, R.,Clasen, S., Simet, C., Tedesco, K., Klas, M., Bonani, G.and Ivy, S. (1992) Evidence for massive discharges

of icebergs into the North Atlantic ocean during thelast glacial period. Nature, 360, 245–249.

Bougamont, M. and Tulaczyk, S. (2003) Glacial erosionbeneath ice streams and ice-stream tributaries: con-straints on temporal and spatial distribution of ero-sion from numerical simulations of a West AntarcticIce Stream. Boreas, 32, 178–190.

Bougamont, M., Tulaczyk, S. and Joughin, I. (2003)Response of subglacial sediments to basal freeze-on:II. Application in numerical modeling of the recentstoppage of Ice Stream C, West Antarctica. J. Geophys.Res., 108, 2223, doi: 10.1029/2002JB001936.

Boulton, G.S. and Hindmarsh, R.C.A. (1987) Sedimentdeformation beneath glaciers: rheology and sedimento-logical consequences. J. Geophys. Res., 92, 9059–9082.

Boulton, G.S. and Clark, C.D. (1990) A highly mobileLaurentide Ice Sheet revealed by satellite images ofglacial lineations. Nature, 346, 813–817.

Boulton, G.S. and Jones, A.S. (1979) Stability of temper-ate ice caps and ice sheets resting on beds withdeformable sediment. J. Glaciol., 24, 29–42.

Boulton, G.S., Hagdorn, M. and Hulton, N.R.J. (2003)Streaming flow in an ice sheet through a glacialcycle. Ann. Glaciol., 36, 117–128.

Christoffersen, P., Tulaczyk, S., Carsey, F. and Behar, A.(2006) A quantitative framework for interpretation of basal ice facies formed by ice accretion over sub-glacial sediment. J. Geophys. Res., F01017, doi: 10.1029/2005JF000363.

Clark, P.U. (1994) Unstable behaviour of the Laurentideice sheet over deforming sediment and its implica-tions for climate change. Quatern. Res., 41, 19–25.

Clarke, G.K.C. (2005) Subglacial processes. Annu. Rev.Earth Planet. Sci., 37, 247–276.

DeConto, R.M. and Pollard, D. (2003) Rapid Cenozoicglaciation of Antarctica triggered by declining atmo-spheric CO2. Nature, 421, 245–249.

Dowdeswell, J.A. and Siegert, M.J. (1999) Ice-sheetnumerical modeling and marine geophysical meas-urements of glacier-derived sedimentation on theEurasian Arctic continental margins. Geol. Soc. Am.Bull., 111, 1080–1097.

Dowdeswell, J.A., Kenyon, N.H., Elverhøi, A., Laberg,J.S., Hollender, F.-J., Mienert, J. and Siegert, M.J. (1996)Large-scale sedimentation on the glacier-influencedPolar North Atlantic margins: long-range side-scansonar evidence. Geophys. Res. Lett., 23, 3535–3538.

Dowdeswell, J.A., Maslin, M.A., Andrews, J.T. andMcCave, I.N. (1995) Iceberg production, debris raft-ing and extent and thickness of Heinrich layers (H-1, H-2) in North Atlantic sediments. Geology, 23,301–304.

Drewry, D.J. (1986) Glacial Geologic Processes. EdwardArnold (Publishers) Ltd., London.

GSP_4_C05.qxd 11/5/07 13:21 Page 63

64 M.J. Siegert

Dyke, A.S. and Prest, V.K. (1987) The late Wisconsin andHolocene history of the Laurentide ice sheet. Géog.Phys. Quatern., 41, 237–263.

Hagdorn, M. and Boulton, G.S. (2004) Simulating sub-glacial sediment transport using a semi-lagrangianmethod. Geophys. Res. Abstracts, 6, 05153.

Hambrey, M.J. (1994) Glacial Environments. UCL Press,London. 296 pp.

Heinrich, H. (1988) Origin and consequences of cyclicice rafting in the Northeast Atlantic Ocean during thepast 130,000 years. Quatern. Res., 29, 143–152.

Hildes, D.H.D., Clarke, G.K.C., Flowers, G.E. andMarshall, S.J. (2004) Subglacial erosion and englacialsediment transport modelled for North Americanice sheets. Quatern. Sci. Rev., 23, 409–430.

Hindmarsh, R.C.A. (1999) Coupled ice-till dynamicsand the seeding of drumlins and bedrock forms.Ann. Glaciol., 28, 221–30.

Huybrechts, P., and Payne, A. and the EISMINT Inter-comparison Group (1996) The EISMINT benchmarksfor testing ice-sheet models. Ann. Glaciol., 23, 1–12.

Laberg, J.S. and Vorren, T.O. (1996) The Middle and Late Pleistocene evolution of the Bear Island TroughMouth Fan. Global Planet. Change, 12, 309–330.

MacAyeal, D.R. (1992) Irregular oscillations of the WestAntarctic ice sheet. Nature, 359, 29–32.

MacAyeal, D.R. (1993) Binge-purge oscillations of theLaurentide Ice Sheet as a cause of the North Atlantc’sHeinrich events. Palaeoceanography, 8, 775–784.

Mahaffy, M.W. (1976) A Three Dimensional NumericalModel of Ice Sheets: Tests on the Barnes Ice Cap,Northwest Territories. J. Geophys. Res., 81, 1059–1066.

Marshall, S.J. and Clarke, G.K.C. (1999) Modeling northAmerican freshwater runoff through the last glacialcycle. Quatern. Res., 52, 300–315.

Marshall, S.J., James, T.S. and Clarke, G.K.C. (2002)North American Ice Sheet reconstructions at the LastGlacial maximum. Quatern. Sci. Rev., 21, 175–192.

Payne, A.J. (1995) Limit cycles in the basal thermal regimeof ice sheets. J. Geophys. Res., 100, 4249–4263.

Payne, A.J. and Dongelmans, P.W. (1997) Self organ-isation in the thermomechanical flow of ice sheets. J. Geophys. Res., 102, 12219–12234.

Pollard, D. and DeConto, R.M. (2003) Antarctic ice andsediment flux in the Oligocene simulated by a climate-ice sheet-sediment model. Palaeogeogr. Palaeoclimatol.Palaeoecol., 198, 53–67.

Press, W.H., Flannery, B.P., Teukolsky, S.A. andVetterling, W.T. (1989) Numerical Recipes (The art ofScientific Computing). Cambridge University Press,702 pp.

Siegert, M.J., Dowdeswell, J.A. and Melles, M. (1999) LateWeichselian glaciation of the Eurasian High Arctic.Quatern. Res., 52, 273–285.

Stokes, C.R. and Clark, C.D. (2001) Palaeo-ice streams.Quatern. Sci. Rev., 20, 1437–1457.

Taylor, J., Siegert, M.J., Payne, A.J., Hambrey, M.J.,O’Brien, P.E., Leitchenkov, G. and Cooper, A.K.(2004) Late Miocene/early Pliocene changes in sedi-mentation paths, Prydz Bay, Antarctica: changes inice-sheet dynamics? Geology, 32, 197–200.

Tully, M.J.C. (1995) Numerical modeling of erosion and deposition beneath Quaternary ice sheets. Ph.D.thesis, University of Cambridge.

Vorren, T.O., Laberg, J.S., Blaumme, F., Dowdeswell, J.A.,Kenyon, N.H., Mienert, J., Rumohr, J. and Werner, F.(1998) The Norwegian-Greenland Sea continentalmargins: Morphology and late Quaternary sediment-ary processes and environment. Quatern. Sci. Rev., 17,273–302.

GSP_4_C05.qxd 11/5/07 13:21 Page 64