Technical advances in the study of ice-free parts of Antarctica can provide quantitative records that are useful for constraining and refining models of ice sheet evolution and behaviour. Such records improve our understanding of system trajectory, influence the questions we ask about system stability and help to define the ice-sheet processes that are relevant on different time-scales. Here, we illustrate the contribution of cosmogenic isotope analysis of exposed bedrock surfaces and marine geophysical surveying to the understanding of Antarctic ice sheet evolution on a range of time-scales. In the Dry Valleys of East Antarctica, 3He dating of subglacial flood deposits that are now exposed on mountain summits provide evidence of an expanded and thicker Mid-Miocene ice sheet. The survival of surface boulders for approximately 14 Myr, the oldest yet measured, demonstrates exceptionally low rates of subsequent erosion and points to the persistence and stability of the dry polar desert climate since that time. Increasingly, there are constraints on West Antarctic ice sheet fluctuations during Quaternary glacial cycles. In the Sarnoff Mountains of Marie Byrd Land in West Antarctica, 10Be and 26Al cosmogenic isotope analysis of glacial erratics and bedrock reveal steady thinning of the ice sheet from 10 400 years ago to the present, probably as a result of grounding line retreat. In the Antarctic Peninsula, offshore analysis reveals an extensive ice sheet at the last glacial maximum. Based on radiocarbon dating, deglaciation began by 17 000 cal yr BP and was complete by 9500 cal yr BP. Deglaciation of the west and east sides of the Antarctic Peninsula ice sheet occurred at different times and rates, but was largely complete by the Early Holocene. At that time ice shelves were less extensive on the west side of the Antarctic Peninsula than they are today. The message from the past is that individual glacier drainage basins in Antarctica respond in different and distinctive ways to global climate change, depending on the link between regional topography and climate setting.
The aim of this paper is to highlight recent advances in glacial geomorphology and glacial geology, achieved using cosmogenic isotope analysis, marine geophysics and high-resolution radiocarbon dating that can add a new quantitative dimension to the understanding of Antarctic ice sheet evolution. Prediction of the behaviour of the Antarctic ice sheet is important, since its response to climate change affects global sea level, ocean circulation and climate. Glaciological modelling is the primary way in which we can understand ice-sheet behaviour sufficiently rigorously to be able to run a series of ‘what if’ scenarios and thus judge the impact on the ice sheet of different levels of human impact on world climate. This is because the large scale of the ice sheet and its long response times rule out other means of experimentation. At present, as illustrated by other papers in this volume, we are making good progress in developing glaciological models that are constrained by glaciological field data. Also, effective use is increasingly being made of satellite and ground-based remote sensing to monitor current trends in such parameters as glacier extent and thickness. It is now possible to identify those parts of the ice sheet that are in decline and those parts that are increasing in thickness (Davis et al. 2005).
However, an impartial observer could argue that the science community is not effective in linking such studies of current processes and trends with those earth-science disciplines that study the past behaviour of the ice sheet. This historical perspective is an important part of the whole for three main reasons. First, geomorphic/geologic studies can provide evidence of past behaviour, e.g. the changes associated with a glacial cycle, which are a powerful way of testing and constraining glaciological models. It is clear that ice sheets are part of a world of complex systems that behave in a nonlinear way. Under such circumstances it is critical to ensure that models are constrained as tightly as possible by real world observations. Second, geomorphic/geologic studies can establish the background trajectory leading up to the present day. For example, the interpretation of a modern observation of ice-sheet thinning is changed by whether it is viewed as the continuation of a long-term trend over 10 000 years or whether it is a short-term perturbation from some longer-term equilibrium. Third, a longer-term perspective can help identify the main processes that control ice-sheet stability. Thus, it may be clear from the study of past ice sheets that calving of deep-water outlet glaciers is the primary control on ice-sheet thinning. In such a case, it directs attention to the key processes that need rigorous modelling if we are to predict the future behaviour of the Antarctic ice sheet with confidence.
An important priority is to create an archive of high-quality evidence of past ice-sheet behaviour against which models can be tested and refined. In this regard the more the archive is tuned to the needs of the modellers, the more useful it will be. The primary need is for information on past ice-sheet extents, ice thicknesses, flow directions, the age and duration of such changes, and ice-sheet trajectory as we come to the present day. Such information can be provided by cosmogenic isotope analysis on nunataks rising above the present ice-sheet surface and by marine geophysical and sedimentological studies of the former ice sheet margins offshore. Cosmogenic isotope analysis has the ability to measure rates of erosion on exposed rock surfaces and, if rates of erosion are known, the length of time that the surface has been exposed to cosmic radiation. It is difficult to overestimate the importance of the technique to geomorphology/geology. Previously, dating has depended on inferences from dateable deposits, which tend to be distributed patchily or, more commonly, offshore. Debates have often focused on the assumptions used to relate a particular deposit to a particular ice sheet event, as illustrated, e.g. by that concerning the stability or otherwise of the East Antarctic ice sheet (Webb et al. 1984; Denton et al. 1993). Cosmogenic isotope analysis allows the minimum age of a surface to be measured directly. It also has the potential to estimate the depth of erosion achieved in a glacial cycle and the duration of time that a surface has been buried (Gosse & Phillips 2001). Another recent advance has been the ability to view the landforms in shallow offshore areas in great detail, using swath bathymetry (Shipp et al. 1999; Canals et al. 2000; Anderson et al. 2002; Ó Cofaigh et al. 2002). It has been a revelation to be able to trace the evidence of formerly more extensive glaciers to their offshore glacial maxima. Coupled with sediment analysis of marine cores, it is possible to reconstruct past glacier extent and processes at a new level of detail. Finally, high-resolution analysis of organic remains can now help to overcome the problems that have hindered the use of radiocarbon dating in Antarctica. It is now possible to identify millennial-scale ice-margin fluctuations over the last 18 000 years or so.
This paper highlights the nature and quality of geomorphic/geologic evidence that is potentially available for modellers. Rather than attempt a review of what is known about the past behaviour of the ice sheet, we have decided to focus on three time-scales: millions of years, tens to hundreds of thousands of years and millennia. In each case, we show that it is possible to refine our knowledge of ice extent and thickness and also to throw light on the key processes that are involved on each of the time-scales. At a time-scale of millions of years we focus on cosmogenic isotope work in the Dry Valleys and show that there has been a stable, cold polar ice sheet and associated climate there for millions of years (figure 1). At the hundred-thousand-year time-scale of a glacial cycle, we use cosmogenic isotope analysis in West Antarctica to record past glacial cycles and deglacial processes operating over the last 10 000 years, while in the Antarctic Peninsula we show how marine geophysics and geology reveal a different pace and timing of deglaciation. Finally, at a millennial scale we use high-resolution radiocarbon dating to reveal contrasting ice-shelf behaviour and processes in the Holocene.
2. Millions of years: Dry Valleys, Southern Victoria Land, East Antarctica
In the Dry Valleys area the Transantarctic Mountains comprise an uplifted escarpment whose upper surface rises above the East Antarctic ice sheet and bears a remarkable suite of landforms related to subglacial meltwater outbursts (Denton & Sugden 2005). There are meltwater channel systems, some 30 km long and others, such as the Labyrinth, forming networks over an area of 5×5 km2 with individual channels over 300 m deep. Leeside slopes, including escarpments 200 m high, are often pockmarked with plunge pools and potholes, some 70 m across and tens of metres deep. In addition, there are areas of stripped and corrugated bedrock ranging in size from 100 m2 to 3 km2. The size of individual features and their extent points to exceptionally high meltwater discharges, while the pattern of the channels indicates that the flow was subglacial. The preservation of pre-existing deposits in the immediate vicinity suggests that the floods breached cold-based ice, at least over the mountains. The orientation of the channels and other glacial landforms shows that the mountains were overridden by an ice sheet flowing north-eastwards. The meltwater was derived from subglacial lakes accumulating beneath warm-based ice on the inboard flank of the mountains and discharged through the thin, cold-based ice sheet that covered the mountain rim.
In the Coombs Hills, at an altitude of approximately 1900 m, is an area of corrugated sandstone bedrock covered in places by flood deposits of dolerite clasts. Three approximately 10 cm diameter clasts were analysed for cosmogenic 3He concentrations in clinopyroxene. Details of the analysis are contained in Margerison et al. (2005) and reveal exposure ages between 8.63±0.09 and 10.40±0.04 Myr, the oldest yet recorded (figure 2). These are minimum ages in that they assume no erosion. If one takes erosion rates as low as 0.03–0.04 m Myr−1, then the exposure ages are increased to approximately 14 Myr (ago). Such low rates are matched only in the driest deserts on Earth (Summerfield et al. 1999). The implication is that the meltwater flood deposits have been exposed for approximately 14 Myr and that they represent outbursts beneath the maximum Antarctic ice sheet of the Mid-Miocene (Kennett 1977; Anderson 1999; Sugden & Denton 2004). The survival of the clasts at the surface for millions of years is evidence of extremely low erosion rates. In turn this implies that the climate must have remained in a dry polar state for millions of years since the deposits were exposed by thinning ice.
These cosmogenic results are independent evidence that supports the view, based on 40Ar/39Ar analysis of volcanic ash trapped in glacial till, that there has been limited change of the East Antarctic ice sheet in the Dry Valleys area since the Mid-Miocene (Marchant et al. 1993a,b). Since the local climate is dominated by the presence of the ice sheet, the implication of a persistent cold, dry climate is that the ice sheet has survived intact. Most changes have been related to the thickening of those outlet glaciers in contact with the sea as a result of sea-level lowering, as in the Pleistocene (Bockheim et al. 1989), or to thickening of landbound glaciers in warmer wetter periods, as in the case of Taylor Glacier in the Pliocene (Marchant et al. 1994).
3. Last glacial cycle (10 000–100 000s of years): West Antarctica
The margins of the West Antarctic ice sheet are dotted with mountain summits, which can be used as dipsticks monitoring past ice sheet behaviour. Cosmogenic isotope analysis on erratic boulders and bedrock has revolutionized the type of data that can be obtained. The technique is particularly valuable in allowing the identification of thinning during deglaciation, periods of burial by ice and exposure during previous interglacials. Also, when linked to detailed glacial-geomorphological analysis it has implications for former basal regimes and the erosive potential of a glacial cycle.
(a) Marie Byrd Land
An example of the application of cosmogenic isotope analysis to glacial geomorphology concerns the Sarnoff Mountains in the Ford Ranges of Marie Byrd Land. Here mountains rise to altitudes of 700–1100 m and 500–900 m above the Boyd outlet glacier (figure 3). Typically the lower slopes of the mountains are moulded by ice erosion while the summits are weathered and support tors and blockfields. Erratics are scattered over the summits. Analysis of 10Be and 26Al shows that there is a suite of samples with consistent exposure ages, which get younger with decreasing altitude (figure 4).
In these cases the 26Al/10Be ratio is close to the production ratio of 6, indicating continuous exposure. Exposure ages range from 10 400 years at an altitude of 700 m to approximately 2000 years at 200 m and reflect the steady thinning of the overriding ice throughout the Holocene. On the upper surfaces there are samples from boulders and bedrock with a significantly lower 26Al/10Be ratio indicating burial by ice. Thus a sample may have an exposure age of 100 kyr and a burial age of 400 kyr, implying that it has survived exposure and overriding for at least half a million years. The analysis is developed in more detail elsewhere (Sugden et al. 2005), and several scenarios can be modelled and compared with the data. From these results we conclude that over the last few glacial cycles, the complex burial history is best explained by summits being exposed for 50% of the time and lower slopes for less (Sugden et al. 2005). Further, the erratic boulders that have survived overriding by ice are often in fragile locations and are a dramatic witness of the power of cold-based ice to protect the underlying substrate, especially on upstanding mountains where it is thin and flow is divergent.
What is the significance of these results in the wider perspective of the West Antarctic ice sheet? The view that thinning and retreat of the ice sheet occurred during the Holocene is consistent with the view that the adjacent Ross ice shelf has experienced grounding line retreat since the Early Holocene (Conway et al. 1999). The latter result is based on the dating of glacier margin and raised beach landforms vacated by the retreating ice shelf and modelling of flow patterns in ice on Roosevelt Island. Probably, the thinning in the Sarnoff Mountains is likewise related to the retreat of grounding lines, in this case of the Boyd Glacier. It makes sense to argue that the retreat of the grounding line is responsible for glacier lowering; further, the freshness and instability of exposed deposits on bedrock beside the glacier suggest that the process may be continuing today. The burial history of the erratics suggests that a similar situation may have occurred during several previous interglacials.
The major drainage outlets from Marie Byrd Land flow into the Amundsen Sea and large bathymetric troughs on the continental shelf provided the pathways by which these outlets advanced across the shelf during the Quaternary (Anderson et al. 2002). In Pine Island Bay, marine geophysical and geological investigations show that during the last glacial maximum (LGM) the ice sheet was grounded on the outer shelf and may have reached the shelf edge (Lowe & Anderson 2002). Radiocarbon dates from marine sediment cores show that deglaciation of Pine Island Bay occurred in two distinct phases. Initial retreat from the outer shelf occurred prior to approximately 16 14C ka BP and was followed by rapid retreat to approximately the present ice sheet position (Lowe & Anderson 2002).
(b) Antarctic Peninsula
Recent marine geophysical and geological research from the western margin of the Antarctic Peninsula (figure 5) has resulted in significant advances to our understanding of the extent and behaviour of the Antarctic Peninsula ice sheet during the last glacial cycle, as well as the processes and conditions at the former ice-sheet bed. At the LGM the ice sheet was positioned at, or close to, the shelf edge. Large glacial troughs extend across the continental shelf, and sedimentary and geomorphic evidence from these troughs indicates that they were occupied by grounded palaeo-ice streams during, or immediately following, the LGM (Canals et al. 2000, 2002; Anderson et al. 2002; Ó Cofaigh et al. 2002, 2003, 2005a,b; Dowdeswell et al. 2004; Heroy & Anderson 2005).
In the Marguerite Bay region, grounded ice advanced to the continental shelf break at the LGM. The ice-sheet margin then retreated 70–100 km before stabilizing again, following which a large ice stream developed in Marguerite trough. Development of this ice stream may have been a glacio-dynamic response to regional deglaciation. The ice stream produced a soft deformation till, and moulded this into mega-scale glacial lineations that extend for up to 20 km across the shelf (figure 6). Macrosedimentological, micromorphological and geotechnical data indicate that pervasive till deformation beneath this ice stream took place over vertical sediment thicknesses of several metres, and that this was associated with significant advection of sediment towards the former grounding line (Ó Cofaigh et al. 2005a). An ice flux of approximately 20 km3 yr−1 is estimated for this ice stream, with an associated sediment flux of approximately 100–800 m3 yr−1 per metre of ice stream width (Dowdeswell et al. 2004).
Subglacial geology exerted a major control on ice stream development in Marguerite Bay (Ó Cofaigh et al. 2002). Streaming flow commenced over the crystalline bedrock by enhanced basal sliding, with the higher velocities occurring over the sedimentary substrate further downflow by subglacial deformation. This indicates spatial variation in the mechanism of rapid flow beneath individual ice streams. Subglacial meltwater channels eroded into crystalline bedrock in the inner parts of Marguerite Trough demonstrate drainage of meltwater beneath the ice stream (figure 7). Such features are probably the product of erosion over several glacial cycles and thus only provide maximum estimates of meltwater discharge.
Retreat of the Marguerite trough ice stream was underway by ca 14 500 cal yr BP (using a marine reservoir correction of 1300 years the uncorrected age of 13 490 14C yr BP is 12 190 14C yr BP; Pope & Anderson 1992; Anderson et al. 2002)1. Radiocarbon dates obtained on the acid-insoluble organic carbon fraction, thought to be derived largely from marine phytoplankton, indicate that open marine conditions were established on the mid-shelf by ca 13 200 cal yr BP (12 610 14C yr BP, Ó Cofaigh et al. 2005a). The latter authors suggest that the absence of grounding-zone wedges and recessional moraines on swath-bathymetric records implies that retreat of the ice stream was rapid. Moreover, major zones of cross-cutting landforms were not observed on the outer shelf. Such cross cutting would be expected had ice-sheet retreat occurred by progressive thinning. Evidence from raised beaches shows that the ice-sheet margin had retreated from Marguerite Bay by ca 10 000 cal yr BP (Bentley et al. 2005a). The timing of the final retreat of grounded ice has been constrained by recently acquired radiocarbon dates on foraminifera in sediments from a lake impounded by ice at the mouth of Moutonnée Valley on Alexander Island. These imply that Moutonnée Lake was free of ice by 9.5 ka cal yr BP (Bentley et al. 2005b). Thus, on its western side the Antarctic Peninsula ice sheet had retreated to a configuration similar to the present-day by the Early Holocene. Further, deglaciation was marked by phases of rapid retreat.
Geological and geophysical studies from the Weddell Sea margin of the Antarctic Peninsula provide evidence for an extensive ice sheet at the LGM (Evans et al. 2005), but they also highlight considerable differences in ice-sheet dynamics during deglaciation when compared to the western margin. Streamlined subglacial bedforms, including drumlins and mega-scale glacial lineations and associated deformation till, occur within cross shelf bathymetric troughs in the Larsen-A region and indicate that at the LGM the grounding-line extended to within less than 10 km of the shelf edge (Camerlenghi et al. 2002; Gilbert et al. 2004; Evans et al. 2005). In contrast to the rapid ice-sheet retreat in Marguerite Bay, however, ice-sheet retreat on the eastern margin was interrupted by a series of stillstands as indicated by the presence of grounding zone wedges on the mid- and inner-shelf (figure 8; Evans et al. 2005). Slower retreat is also supported by two further lines of evidence: (i) zones of cross-cutting lineations (e.g. in Robertson Trough) indicating switches in ice flow direction during deglaciation and (ii) thick sequences of deglacial sediments above till, as recovered in sediment cores. The latter point is in contrast to Marguerite Bay where deglacial sediments recovered in marine cores are thin, supporting the view of rapid retreat inferred from geophysical records (Ó Cofaigh et al. 2005a).
The timing of retreat on the eastern margin of the Antarctic Peninsula ice sheet is complicated by the presence of detrital and reworked carbon. However, recent dating of sediments on the inner Larsen-A ice shelf using geomagnetic palaeointensity indicates that the transition from grounded to floating ice was complete in this region by 10.7±0.5 cal ka BP (Brachfeld et al. 2003).
(c) Weddell Sea
Recent estimates of the ice volume lost from the Weddell Sea embayment since the LGM range from 2.35 to 2.7 m of global sea level equivalent (Bentley 1999; Denton & Hughes 2002) which is comparable to that lost from the Ross Sea. However, very little is known about the timing and rates of WAIS thinning in the Weddell Sea embayment (Bentley & Anderson 1998) and so there are few firm constraints against which models can be tested.
The first terrestrial constraints on deglacial history in the Weddell Sea have come from a small number of recent cosmogenic exposure ages. For example, dates from the Shackleton Range show that the east side of the Weddell Sea embayment has escaped significant overriding by ice for millions of years (Fogwill et al. 2004). A key question for understanding deglaciation of the Weddell Sea embayment revolves around the timing and rate of ice-sheet thinning in the Ellsworth Mountains. Preliminary data from Marble Hills in the Heritage Range suggest that there was some Holocene thinning (Todd & Stone 2003). There are also pre-LGM dates at higher altitudes but it is not yet clear if these indicate a thicker ice sheet that formed the main trimline mapped by Denton et al. (1992) or whether they are samples with complex exposure histories. If the latter is the case, then the trimline could be much older than the LGM.
(d) West Antarctic ice sheet: summary
Geological and geomorphological constraints on the past behaviour of the West Antarctic ice sheet are sparse, but already they point to two main conclusions. First, the dynamics of retreat vary from place to place. This is well illustrated by the contrast in retreat dynamics between the western and eastern flanks of the Antarctic Peninsula. The comparison shows that the rate of ice stream retreat was variable between different bathymetric troughs. Thus steady ice stream retreat, punctuated by stillstands, in the inner parts of the Larsen-A region of the East Antarctic Peninsula contrasts markedly with the Marguerite Bay ice stream on the western flank which experienced rapid retreat across much of its bed. The latter ice stream was active during a particular phase of retreat and implies that ice stream dynamics switched on and off during regional deglaciation. The slow and steadier pattern of retreat of ice streams in the inner Larsen-A area is similar to that in the Ross Sea sector where Shipp et al. (2002) showed a series of closely spaced grounding zone wedges and recessional moraines that record progressive retreat. This further contrasts with the two-step pattern of retreat in Pine Island Bay where an interval of initial steady recession was followed by rapid retreat across the inner shelf (Lowe & Anderson 2002).
Second, there is a clear difference in the timing of retreat from sector to sector. Thus, the ice sheet had withdrawn from offshore areas of the Antarctic Peninsula and approached its present configuration by the Early Holocene ca 9500 years ago. In the Ross Sea sector and Marie Byrd Land ice-sheet thinning has persisted throughout the Holocene. Clearly, such contrasts have important implications when, e.g. calculating the contribution of Antarctic deglaciation to sea level rise since the LGM. At present, the biggest gap in our knowledge is the behaviour of the ice sheet in the Weddell Sea embayment. We are not yet able to determine whether deglaciation of the Weddell Sea embayment was progressive and continued into the Late Holocene, or whether it occurred soon after the LGM. Current research in the embayment is aimed at better constraining deglaciation of the Ellsworth Mountains from cosmogenic isotope analysis at a number of sites throughout the range.
4. Millennia: Antarctic Peninsula ice shelves
The loss of ice shelves around the Antarctic Peninsula during recent decades is a dramatic response to the climatic warming that has been observed over the same period (Vaughan & Doake 1996). But it has been difficult to judge the significance of the changes in the absence of knowledge of past changes and the main processes involved. For example, are the ice shelves as restricted in extent as at any time since the LGM? Is collapse related more to surface temperature conditions (Doake et al. 1998) or to subsurface melting (Shepherd et al. 2003, 2004)? These are the sorts of questions that are amenable to input from the geological and geomorphological record.
George VI Ice Shelf flows westwards from the Antarctic Peninsula to Alexander Island where it impounds lakes (figure 9). Detailed analysis of sediments in one such lake reveals that the ice shelf was not present between 9500 and 8000 years ago and that the diatoms and foraminifera show that relatively warm ocean water occupied the site at that time (Bentley et al. 2005b). Loss of the ice shelf immediately postdates an early climatic optimum as recognized in Antarctic ice cores at ca 11 000–9500 years ago (Masson et al. 2000) and coincides with an influx of relatively warm Circumpolar Deep Water onto the Antarctic Peninsula shelf (Domack et al. 2001). The latter coincidence suggests that George VI Ice Shelf may have disappeared largely as a result of increased bottom melting due to oceanographic warming. These observations from the past reinforce conclusions based on modern studies (Shepherd et al. 2003; Payne et al. 2004) that ice shelves are susceptible to the effects of changing ocean currents on the rate of bottom melting. However, on the east side of the Antarctic Peninsula the picture is more complicated, with some ice shelves having disappeared in the Mid-Holocene (Pudsey & Evans 2001), but at least one ice shelf remaining intact until the late twentieth century (Domack et al. 2005). Once more, this demonstrates the importance of local factors in determining the response of an ice shelf to climatic forcing.
5. Wider implications
This paper has focused on the results obtained from technical advances in glacial geology and geomorphology that add a more refined quantitative dimension to the understanding of ice sheet evolution. It now seems that we are in a position to provide the sort of information, distributed over critical areas required to constrain models of ice sheet behaviour. This information has come from two main sources: cosmogenic isotope analysis and marine geophysics. Cosmogenic isotope analysis seems to have several particular advantages when viewed from the modelling point of view. It can produce firm chronological data at many points. The immediate potential is for data from any exposed surface or mountain summit protruding above the ice sheet. Furthermore, in the high latitude and slow weathering environment of Antarctica, the technique extends over time-scales from a few thousand years to tens of millions of years. In addition, the technique is able to contribute information about both rates of weathering and the age and burial history of a point. And there is also the exciting prospect of quantitative information from rock surfaces that are now buried beneath ice but can be reached through coring. Having stressed the potential it is important to stress an important caveat. The complexities of interpretation require a correspondingly detailed level of geomorphological analysis and rigour in sampling. Detailed field observations are critical. Marine geophysical and geological observations provide crucial information for ice sheet models in terms of ice sheet extent, palaeo-ice stream dimensions, and the nature and timing of ice sheet retreat. In addition, sedimentological and geotechnical investigations of marine sediment cores have allowed the reconstruction of former subglacial processes, especially the nature of the ice stream bed and the mechanisms of fast flow.
The main conclusion from this paper is that the response of the Antarctic ice sheet to global change is highly variable. At one extreme, some areas have not changed significantly in millions of years. At the other extreme ice shelves can break up in just a few weeks. The contrast in the processes and the timing of deglaciation between different sides of the Antarctic Peninsula shows the importance of local/regional factors such as topography/bathymetry in influencing how a glacier basin responds to change. If so, then a fruitful line of enquiry is to focus modelling on such glacier basins in an attempt to understand present processes and to identify future thresholds that may cause a glacier's behaviour to change. Such an approach will help solve outstanding problems such as the contribution of the Antarctic ice sheet to global sea-level rise since the LGM and the partitioning of any contribution within Antarctica (Clark et al. 2002). If the Antarctic contribution is less than suggested by some calculations, then the implication is that the missing ice is somewhere in the Northern Hemisphere. In turn this would change our understanding of Northern Hemisphere ice sheets.
A final question is to ask how well does the evidence of past changes in Antarctic ice sheet evolution match the trends and processes that are being identified by modern studies of changes in ice-sheet elevation. Any comparison is limited by the patchy distribution of geomorphological information and also by the resolution and coverage of satellite observations. Nevertheless, there are matches. The progressive thinning of the ice surface in Marie Byrd Land throughout the Holocene and its continuation to the present, as revealed by cosmogenic isotope analysis and field observations, coincides with an area of surface lowering in recent decades (Davis et al. 2005; Wingham et al. 2006), illustrating that this trend has been occurring for a long time, prior to the advent of satellite remote sensing.
With regard to process, marine geophysical records of deglaciation from the LGM suggests that Antarctic palaeo-ice streams have responded differently to external forcing such as climate/ocean warming or relative sea-level rise. In particular bathymetry acts as a first-order control in determining the way a palaeo-ice stream responds to rises in sea level and water temperature. These are the very processes that seem to be at work in those coastal parts of Antarctica that are experiencing thinning at the present day (Shepherd et al. 2001; Joughin & Tulacyzk 2002; Payne et al. 2004). Finally, we can point to analogies that mirror the increase in mass balance predicted to accompany climate warming in Antarctica. Modern measurements show that thickening is occurring in certain regions of East Antarctica (Davis et al. 2005). This agrees with evidence in the Dry Valleys that shows that outlet glaciers such as Taylor Glacier thickened and advanced during the Pliocene, a period of previous modest warming (Marchant et al. 1994).
The application of cosmogenic isotope analysis and marine geophysics have provided a new quantitative dimension to our understanding of Antarctic ice sheet evolution and the data generated with these techniques can be used to constrain and refine ice-sheet models. In this paper, we have highlighted the nature and quality of the geological evidence that is potentially available for modellers by focusing on three different time-scales. For each of these scales we have shown that it is possible to refine our knowledge of ice extent, ice thickness and key ice sheet processes.
In the Dry Valleys of East Antarctica, the survival of surface boulders for approximately 14 Myr, demonstrates exceptionally low erosion rates and points to the stability of the dry polar desert climate since the Mid-Miocene. In Marie Byrd Land, West Antarctica, cosmogenic isotope analysis of erratics and bedrock reveal steady thinning of the ice sheet from 10 400 years ago to the present. In the Antarctic Peninsula marine geophysical investigations reveal an extensive ice sheet at the LGM. Based on radiocarbon dating, retreat of this ice sheet began by 17 000 cal yr BP and was largely complete by 9500 cal yr BP. These data show that ice-sheet retreat occurred at different times and rates on the west and east sides of the Antarctic Peninsula, and was largely complete by the Early Holocene. At that time ice shelves were less extensive on the west side than today. Such geological data demonstrate that individual glacier drainage basins in Antarctica can respond in different and distinctive ways to climate change and, thus, that ice-sheet response to external forcing is not uniform.
More broadly, this long-term perspective needs to be integrated into ice-sheet models. Such geological studies provide evidence of past ice-sheet behaviour, which can be used to test and constrain models and can be used to establish the background trajectory leading up to the present day. They also facilitate the identification of the key processes that need rigorous modelling if we are to predict the future behaviour of the Antarctic ice sheet with confidence.
One contribution of 14 to a Discussion Meeting Issue ‘Evolution of the Antarctic Ice Sheet: new understanding and challenges’.
↵Radiocarbon dates on calcareous material (e.g. foraminifera) from Antarctica incorporate a reservoir age of 1300 years (see Anderson et al. 2002 for discussion). Radiocarbon dates obtained on total organic carbon are reported in uncorrected form. This is due to uncertainities in the precise marine reservoir correction to apply to this material and, in some cases, the fact that vibro- and piston-coring may not recover the seawater/sediment interface, so some fraction of the upper sediment column is lost and the core top dates erroneously old.
- © 2006 The Royal Society