The dynamical characteristics of marine ice sheets.

1. Resumé

The marine nature of the West Antarctic Ice Sheet suggests that it is potentially unstable, but our lack of understanding means its state of equilibrium remains unknown. Our goal is to clarify the complex interactions between ice sheet, ice stream, ice shelf, atmosphere, ocean and lithosphere which determine the state of equilibrium and control the mass balance of the West Antarctic Ice Sheet so that its future behaviour and response to climate change can be more reliably predicted.

We shall use a balanced approach of observation and modelling. The current state of the ice sheet will be determined by a mixture of satellite remote sensing, airborne geophysical surveys and ground based studies such as seismic, radar and GPS surveying. A hierarchy of numerical models will be used to assimilate and analyse observations, to investigate the mechanics of marine ice sheets, and to describe the evolution and variability of the Antarctic Ice Sheet. By focusing attention on the West Antarctic Ice Sheet, especially the Ellsworth Land and Filchner Ronne Ice Shelf sectors, we shall be complementing programmes being undertaken by other national and international agencies. However, we shall also be carrying out related studies on other ice sheets, in Antarctica and elsewhere.

By the end of the project, we hope to have achieved a considerable improvement in our understanding of the critical processes responsible for determining the state of equilibrium of marine ice sheets and in describing the past and current state of the Antarctic ice sheet, its present-day mass balance and rate of change. This will give us a strong basis for being able to give more reliable predictions about the future behaviour of the ice sheet and the potential variability of its response to climate change over the coming decades and centuries.

2. Scientific background

Global mean sea level would rise on average by 5 to 6 m if the ice in the West Antarctic Ice Sheet (WAIS) were to melt or start floating. Called a marine ice sheet because it rests on a bed well below sea level, the WAIS has long been thought to be uniquely vulnerable to decay. Fast flowing ice streams drain the ice sheet into the two largest ice shelves in the world (Ross and Filchner-Ronne), forming a complex and dynamic coupled ice sheet/stream/shelf system. It is sensitive not only to external climate forcings such as sea level, ocean temperature, surface temperature and accumulation, but also to the mechanisms and processes acting at the margins and base of the ice sheet which could introduce "internally" generated periodicities to the ice sheet cycle of growth and decay.

Clear evidence of ice sheet extent in the past is lacking and it is not certain when the WAIS was smaller than it is today (Burckle, 1993). It is likely that the ice sheet grew to the edge of the continental shelf during the last glacial period, and has retreated to its present position between 12,000 and 6,000 years ago.

Numerical modelling of the ice shelf/stream interaction suggests that time scales for disturbances to propagate from ice shelf to ice sheet may be of the order of hundreds to thousands of years. An analysis of the qualitative mechanics of marine ice sheets (Hindmarsh, 1993a) shows that they are in neutral equilibrium if the transition region between grounded and floating ice is sharp, and may still be in neutral equilibrium if the transition region is smooth. Ice streams represent a smooth transition region and may help to stabilise the ice sheet.

Fast flowing ice streams drain up to 90% of the West Antarctic Ice Sheet, mainly into ice shelves such as Filchner-Ronne and Ross. The ice streams act as elongated transition zones, often more than 100 km in length, between the grounded, slowly moving inland ice sheets and the floating ice shelves (Frolich and MacAyeal, 1991). They are recognised as playing an important, though not fully understood role, in controlling the state of equilibrium of marine ice sheets (Hindmarsh, 1993b). The large ice streams are influenced at their downstream ends by the ice shelves into which they flow. At the upstream end, there is a change from sheet mechanics, where the ice is deforming by shear, to streaming flow, where fast flow behaviour such as basal sliding begins. This upstream transition zone is less well understood than the downstream one, as several processes could contribute to the onset of fast flow. Lateral margins of ice streams may be bounded either by ice from the surrounding ice sheet or by rock from a bed trough constraining the flow.

Ice stream flow is restrained by stresses transmitted across the bed and the lateral (shear) margins. A variety of results exist about their relative importance. Frolich and Doake (1988) found an equal split for Rutford Ice Stream. Results from the Siple Coast differ on Ice Stream B from no resistance from the base (Whillans and van der Veen, 1993) to equal resistive drag from base and sides (Echelmeyer and Harrison, 1993), while there is more unanimity that the majority of the resistive drag on Ice Stream E may come from the base (Bindschadler et al, 1993; MacAyeal et al, 1993). Although a phenotype ice stream would contain restraint from both sides and base, each ice stream that has been studied so far has considerable individual characteristics.

Hindmarsh (1993a) has shown that the mechanics of the ice sheet-stream-shelf transition can be described by considering the variation in basal traction, decreasing from the ice sheet towards the ice shelf. A scale analysis of the governing equilibrium (Stokes) equations shows that the transition zone can be considered as three rheological layers, depending on the relative importance of shear and longitudinal stresses. In two dimensions, the upper layer is where the longitudinal stress dominates, the middle later where longitudinal and shear stresses are of equal magnitude, and the lower layer where shear stress dominates. As basal traction decreases towards the shelf the upper layer expands to fill the whole region, pinching out the lower layers. In three dimensions, a similar scale analysis can be carried out. However, one significant area where the analysis may not be valid because the traction is varying too rapidly in space is the ice stream lateral margin. As many earlier ice stream type models (e.g. Budd and Jensen, 1975; Alley and Whillans, 1984; Van der Veen, 1987; Muszynski and Birchfield, 1987) can be shown to be equivalent to first order both to each other and to a rationally derived reduced model, it is important to be able to include the effect of side shear in a rational way.

The importance of understanding processes in the margins of ice streams, coupled with the difficulty of making measurements there due to crevassing, imply that satellite observations can make a valuable contribution. The lateral shear margins on Rutford Ice Stream are zones of high strain rate in which crevassing is often apparent on visible imagery. ERS-1 SAR images show that the shear margins have a high backscatter coefficient (0); this may be a result of high strain rates causing microfracturing in the surface layers, and may in turn be an observable precursor of crevasse formation. SAR images of the head of Rutford Ice Stream, Antarctica show an area of bright lineaments that marks the onset of streaming flow where crevassing is visible on Landsat imagery. Similar features have been observed on SAR images of Pine Island Glacier (Lucchitta et al, 1994) suggesting that they may prove to be definitive markers for the onset of streaming flow.

3. Research and methodology

Numerical modelling

The 1980s saw the construction of several `all-purpose' models of the whole Antarctic ice sheet, which had particular aims of simulating the evolution of Antarctica through glacial cycles and of constructing global change prognoses. The three ice-sheets of Antarctica - the Peninsula, the West Antarctic and the East Antarctic all provide different scientific challenges which have concomitant effects on modelling strategies and the technical requirements of their execution. Moreover, each of these ice sheets interacts with the rest of the earth in different ways, meaning that as the scope of modelling broadens, different modelling strategies are being adopted for the different ice sheets for reasons that are essentially practical. These strategies should nevertheless recognise the ultimate aim of modelling must be reliable short and long term prediction and retrodiction of the glaciological characteristics of the Antarctic Ice Sheet.

The Peninsula

Specific objectives will be to:

Construct (linearised) models of the Peninsula which have the aim of producing short term prognoses, and which take specific account of uncertainty in the state of balance of the ice sheet, the influence of basal topography, and which maximise use of data from this heavily surveyed area. Inverse, stochastic and statistical (Bayesian) techniques could profitably be used in this area. This has links with ICD1.

Initiate modelling of the Peninsula with the aim of trying to relate ice sheet evolution to the offshore record. This will link in with the Geosciences Division Programme on Antarctic Paleoenvironmental Change Studies (APECS). APECS is studying the Quaternary changes in sediment source and transport related to grounded ice sheets on the continental shelf (specifically on the continental rise off the Pacific margin of the Antarctic Peninsula). Another goal of APECS is to carry out numerical modelling of terrigenous sediment transport under a glacial regime on the Antarctic Peninsula.

The Peninsula has a very short relaxation time for glaciological equilibration (1000 to 2000 years) compared with the other Antarctic ice sheets. Its glaciological response to global change is much more likely to be of significant magnitude than those of the other ice sheets. An appropriate modelling response (see ICD2) is the use of linearisation techniques (Oerlemans and van der Veen, 1984; Hindmarsh, 1992) which permit the use of a wider range of statistical and stochastic techniques (e.g. Van der Veen, 1992). The approaches used in this work can be made rather more sophisticated and rigorous, in particular by incorporating the effects of spatial variation, and in consequence more precise estimates of the influence of data uncertainty can be made.

Observationally, the grounding line in the Peninsula seems to occur where the ice thickness is more or less zero, which is presumably at least in part a reflection of the rugose terrain of the Peninsula. The technical modelling issues are related to the modelling the influence of the topography on the flow of ice - how it affects stress fields, and how well the basal topography has to be characterised before the dominating error in ice sheet prediction stems from our ignorance of other glaciological processes.

The Peninsula probably responded rather passively to sea level changes during glacial cycles, expanding to the shelf edge when sea levels dropped, and retreating when they rose again. If one assumes that the Peninsula remained to all intents and purposes uncoupled from the West Antarctic Ice Sheet, then modelling of the expansion and retreat during a glacial cycle would be reasonably straightforward. More sophisticated modelling would await the construction of a satisfactory West Antarctic Ice Sheet model.

West Antarctica

Specific objectives will be to:

Improve the numerical modelling of marine ice sheets to the extent that we are confident we are producing solutions with controlled accuracy.

Carry out further research into the dynamical characteristic of ice stream switching.

Initiate modelling of the West Antarctic sedimentary basins with emphasis on pore fluid movement, tectonism, isostasy and eustasy.

The West Antarctic ice sheet is a marine ice sheet drained by fast flowing ice streams. The conventional argument is that this makes it particularly sensitive to perturbation and liable to produce a rapid and significant effect on global sea level. West Antarctica provides one of the most difficult areas in modelling ice sheet dynamics. We have insufficient understanding of ice deformation and basal processes to be confident that we are representing the processes of ice sheet deformation successfully and prognoses of ice sheet behaviour which depend critically upon knowledge of these processes are essentially conjectural at the moment.

There is increasing evidence that the numerical problems associated with modelling marine ice sheets - the grounding line, pulsatory behaviour of ice streams, and even the way marine ice sheet models are modelled - present rather greater a challenge than has hitherto been appreciated. There is not even a consensus view as to what constitutes the basic dynamical response of marine ice sheets to forcing.

Specific issues are:

The lateral regions of ice streams provide boundary conditions to shelf models which are not being properly resolved and which result in severe numerical problems (EISMINT Workshop, 1994).

The switching of ice streams (by some as yet unexplained process) may cause kinematical shock waves to propagate into ice shelves, again producing a severe numerical challenge (MacAyeal and Barcilon, 1988).

One argument put forward regarding the different prognoses of marine ice sheet models is that the numerical techniques are failing to represent an anomalous dynamical feature of marine ice sheets, neutral equilibrium (Hindmarsh, 1993a). According to this argument, this particular dynamical feature is quite general, and stream-shelf interactions essentially modulate it.

All of these are research problems which must be tackled before we can begin to make long term prognoses of marine ice sheet behaviour.

East Antarctica

Specific objectives will be to construct models of the East Antarctic ice sheet:

through glacial cycles, with the aim of illuminating the climate record in polar ice cores;

through the Cainozoic, with the aim of providing constraints on circulation during the Tertiary and on loading of the Antarctic continent during the same period.

East Antarctica has significance as having the oldest ice and thus the longest climate record. It is not essentially a marine ice sheet, although it does have some large drainage basins where the base is substantially below sea level. Like the Peninsula, the feeling is that is has probably responded rather passively to the cyclical changes in sea level which occurred in the upper Pleistocene. The major scientific issues are related to when a large ice sheet formed, and as to whether it grew during the mid-Tertiary only to decay during the `greenhouse' Pliocene.

These issues are of particular concern to climatologists, as the East Antarctic ice sheet is one of the earth's major topographic features and must significantly affect circulation patterns. It seems clear from ice sheet modelling experiments (Huybrechts, 1994) that the East Antarctic ice sheet is stable even to severe perturbations of meteorological conditions, and that putative instabilities in the Pliocene must have been climatological and probably related to topography, or due to internal processes in the ice sheet which are not operating today.

The Transantarctic Mountains, which are a key area for data gathering, are in an active rift zone, and basin modelling techniques have begun to be applied to these areas (e.g. van der Beek and others, 1994). Much of the data for and against a stable East Antarctic ice sheet comes from these areas, and rapid tectonic movements over short time periods have been advocated as an explanation for apparent pathologies in the data. Modelling of the integrated ice-sheet/lithosphere system might help resolve some of the uncertainties.

Ice sheet modelling can help constrain the climate record by identifying flow lines and modelling changes during the glacial cycle. This has links with ICD3. As with the Peninsula, changes may be small enough to be describable by linearisations which permits a much wider range of techniques to be used in the analysis - inverse, statistical and stochastic.

Field observations

Seismic sounding

Specific objectives will be to:

Determine the nature of the bed of Rutford Ice Stream and other glaciers;

Investigate the seismic anisotropy on Rutford Ice Stream;

Obtain spot depths beneath ice shelves.

Seismic techniques will be used for investigating basal conditions of ice streams and for determining bathymetry beneath ice shelves. Seismic methods complement radar ones, as radio waves cannot significantly penetrate conductive materials such as sea water. A combination of both methods allows other useful parameters such as ice density profiles to be determined.

Work begun on Rutford Ice Stream will be extended to other ice streams to determine the variability of the basal hydraulic regime. Seismic anisotropy measurements give an indication of the water content and properties of material at the glacier bed, important parameters for understanding the fast flow of glaciers. The bathymetry around the southern region of Filchner Ronne Ice Shelf will be measured to extend existing data and used in tidal modelling of the Weddell Sea region as well as in modelling the evolution of the ice sheet.

Ground surveys

Specific objectives will be to:

Position radar reflectors and determine local velocity fields;

Investigate nature of ice stream margins;

Obtain tidal flexure profiles across grounding lines;

Support accurate positioning of seismic and radar profiles.

Accurate positioning is required for determining the movement of markers on the ice surface. GPS and Transit receivers are being used, backed up by more traditional survey techniques such as microwave distance measuring, levelling, and using theodolites. GPS methods require a fixed reference base station for accurate absolute positions, else only strain figures can be measured. Reflectors for calibrating SAR imagery will be positioned by GPS and the local velocity field determined. Tidal displacements can be measured by GPS across grounding lines, giving bending profiles accurate to a few cm.

Ice sounding radar (radio-echo sounding)

Specific objectives will be to:

Investigate the nature of the basal and reflecting surfaces on Filchner Ronne Ice Shelf;

Study conditions at the base and margins of ice streams on Orville and Zumberge coasts;

Obtain ice thickness and basal topography (Ellsworth Land, Dronning Maud Land, Antarctic Peninsula);

Link internal layers to physical and chemical properties of ice;

The 150 MHz radar will be developed to become a versatile instrument with various operating modes, e.g. high power pulse, chirp, PCM, etc. allowing the optimum mode to be selected for the particular requirements of the task in hand. Competing requirements of high power (for depth penetration) and short pulse length (for high resolution) means that a compromise has often to be selected. The Network Analyser offers a wide frequency range for studying frequency dependent phenomena such as polarisation and internal layering. The projects on which the radars will be used range from ice thickness sounding (e.g. in Ellsworth Land, Antarctic Peninsula and Dronning Maud Land), echo strength surveys (e.g. of marine ice layers on Filchner-Ronne Ice Shelf and basal conditions of ice streams), polarisation measurements (e.g. across shear zones and grounding lines and on ice shelves) and internal layering.

The link between chemistry of ice and the electrical conductivity has now been firmly established. Techniques will be used to turn electrical profiles from ice cores into predictions of radar internal layer profiles. These will then be compared with field data. These are available from the GRIP site in Greenland, and from shallow cores in Antarctica. If it can be established that the layers (or at least some of them) are isochrones, then they can be used for ice dynamics and mass balance studies. There are also physical parameters such as density and ice fabric that affect the generation of internal layers and therefore need to be studied.

Satellite remote sensing

Specific objectives will be to:

Help produce a baseline elevation model of Filchner Ronne Ice Shelf from satellite altimetry;

Determine topography to support mass balance studies;

Develop an operational SAR interferometry system and apply to WAIS;

Monitor ice sheet to detect changes;

Investigate fracture of ice;

Produce thematic maps and initiate the Antarctic bed map project.

Satellite remote sensing from earth observing platforms such as Landsat, SPOT, ERS-1, allows glaciologically important parameters such as surface morphology, elevation, slope, velocity and strain rate to be determined, and features related to flow and structure recognised and mapped. Ground data are still often needed to help verify and calibrate the satellite derived data; in some instances, once point data have been obtained, continuous fields can be derived. Campaigns will be undertaken to support particular objectives.

We shall continue to interpret visible-band imagery for glaciological parameters as well as for monitoring changes. Synthetic Aperture Radar (SAR) data from ERS-1 and similar planned missions will be used for all-weather coverage and will be interpreted for glaciological features which complement those seen in visible imagery. Evidence of fracturing and crevassing can be seen in visible and SAR imagery, allowing the conditions for initiation, growth and decay to be interpreted (Vaughan, 1993). We shall continue to develop SAR interferometric techniques for deriving continuous fields of surface displacements and locating grounding line positions (Goldstein et al., 1993). Radar altimetry data from platforms such as ERS-1 will be used for determining surface elevations, slopes and characteristics. Other aspects, such as using passive microwave data, are detailed in project ICD2.

4. Wider implications

Observations from this work will complement those from other regions of the West Antarctic Ice Sheet, allowing a broader view to be taken. The methods developed to analyse data and model ice sheets will improve our understanding of how ice sheets behave and help to indicate how to predict the variability of ice sheet forecasts.

By addressing global phenomena with enormous economic consequences (a sea level rise of 1 m could cause 1010 worth of damage to London alone), the project provides a risk assessment (wealth protection) for a paltry premium.

5. References

Alley, R.B. and I.M. Whillans. 1984. Response of the East Antarctic Ice Sheet to sea-level rise. J. Geophys. Res. 89(C4), 6487-6493.

Beek, P. van der, S. Cloetingh and P. Andriessen. 1994. Mechanisms of extensional basin formation and vertical motions at rift flanks: constraints from tectonic modelling and fission-track thermochronology. Earth Planet. Sci. Lett., 121, 417-433.

Bindschadler, R.A., M.A. Fahnestock, T.A. Scambos and D.D. Blankenship. 1993. The identification of "sticky spots" on Ice Streams D and E, West Antarctica. VISAG, Abstract No. 124.

Budd, W.F. and D. Jensen. 1975. Numerical modelling of glacier systems. IAHS Publication 104, 257-291.

Burckle, L.H. 1993. Is there direct evidence for late Quaternary collapse of the West Antarctic ice sheet? J. Glaciol., 39(133), 491-494.

Echelmeyer, K.A. and W.D. Harrison. 1993. The role of the margins in ice stream dynamics. VISAG, Abstract No. 3.

EISMINT Workshop. 1994. Model Intercomparision Workshop held in Bremerhaven, June 1994.

Frolich, R.M. and C.S.M. Doake. 1988. Relative importance of lateral and vertical shear on Rutford Ice Stream, Antarctica. Ann. Glaciol. 11, 19-22.

Frolich, R.M. and D.R. MacAyeal. 1991. Numerical modelling of Rutford Ice Stream, Antarctica and its catchment area. IAHS (Leningrad, Sep. 1990).

Goldstein, R.M., H. Engelhardt, B. Kamb and R.M. Frolich. 1993. Satellite radar interferometry for monitoring ice sheet motion: application to an Antarctic ice stream. Science, 262, 1525-1530.

Hindmarsh R.C.A. 1992. Estimating ice sheet response to climate change. In The contribution of Antarctic Peninsula Ice to Sea Level Rise. Eds. Morris, E.M.

Hindmarsh, R.C.A. 1993a. Qualitative dynamics of Marine Ice Sheets. NATO ASI Series I12, Ed. Peltier, W.R. 67-99.

Hindmarsh, R.C.A. 1993b. Modelling the dynamics of ice sheets. Prog. Physical Geog. 17(4), 391-412.

Huybrechts, P. 1994. Glaciological and climatological probabilities and improbabilites of alternative glaciological models of Antarctica. In Landscape evolution in the Ross Sea Area, Antarctica. Ed. Wateren, F.A.M van der, A.L.L.M. Verbers and F. Tessenbohn. Rijks Geologische Dienst, Haarlem. 107-112.

Lucchitta, B.K., C.E. Smith, J.A. Bowell and K.F. Mullins. 1994. Velocities and mass balance of Pine Island Glacier, West Anatrctica, derived from ERS-1 SAR images. In Space at the service of our environment. Proceedings of 2nd ERS-1 Symposium. ESA SP-361, 147-151.

MacAyeal, D.R. and V. Barcilon. 1988. Ice shelf response to ice stream discharge fluctuations. I. Unconfined ice tongues. J. Glaciol. 34, 121-127.

MacAyeal, D.R., T.A. Scambos and R.A. Bindschadler. 1993. The basal stress regime of Ice Streams D and E, Antarctica. VISAG, Abstract No. 49.

Muszynski, I. and G.E. Birchfield. 1987. A coupled marine ice-stream-ice-shelf model. J. Glaciol. 33, 3-15.

Oerlemans, J. and C.J. van der Veen. 1984. Ice Sheets and Climate. Reidel pp. 217.

Van der Veen, C.J. 1987. Longitudinal stresses and basal sliding: a comparative study, In Dynamics of the West Antarctic ice sheet. Eds. Van der Veen, C.J. and J. Oerlemans. Dordrecht, Reidel, 223-248.

Veen, C.J. van der. 1993. Interpretation of short-term ice-sheet elevation changes from satellite altimetry. Climatic Change 23(4), 383-405.

Vaughan, D.G. 1993. Relating the occurrence of crevasses to surface strain rates. J. Glaciol., 39(132), 255-266.

Whillans, I.M. and C.J. van der Veen. 1993. Mechanics of Ice Stream B, Antarctica. VISAG, Abstract No. 111.