Figure 7.7: Feedback loops associated with land ice masses. Changes in their volume affect the salt balance at the sea surface and may influence the thermohaline circulation (THC). While there is palaeo-
climatic evidence that this has happened often during the last Ice Age, the above processes are unlikely to play a major role for future climate change
Ice stream instability, ice shelf break-up and switches in routing and discharge of glacial melt water present themselves as mechanisms for altering the surface salinity of oceans and inducing changes in the pattern and strength of the THC. There are positive and negative feedbacks associated with changing land ice masses (Figure 7.7). In coupled climate models, all these cryospheric processes are represented simply as surface fresh water inputs to the ocean (see Chapter 8, Section 220.127.116.11).
Ice sheets are continental scale masses of fresh water ice formed by the burial and densification of snow. They can be divided into areas that are grounded on the land surface, either below or above sea level, and areas that are afloat. Grounded parts of ice sheets exhibit slow and fast modes of flow. The slow sheet-flowing component is the more prevalent but the fast stream-flowing component can account for most of the ice discharge. Switching between slow and fast modes of flow may occur thermally, mechanically or hydrologically and is suggestive of "surging", a known cyclic instability of certain glaciers (Kamb et al., 1985; Bindschadler, 1997). The slow-flow process for ice sheets is internal creep and the fast-flow processes relevant to ice streams are bottom sliding, enhanced creep and sub-glacial sediment deformation (Alley, 1989; Iken et al., 1993; Engelhardt and Kamb, 1998). The factors controlling onset, discharge and width of ice streams are subject to poorly understood geological, topographic, thermal and hydrological controls (Clarke and Echelmeyer, 1996; Anandakrishnan and Alley, 1997; Bindschadler, 1997; Anandakrishnan et al., 1998; Bell et al., 1998; Jacobson and Raymond, 1998).
Although in GCMs the albedo of land ice is typically fixed, satellite observations could be used to remove this limitation; the melt area, as identified from satellite passive microwave observations (Mote et al., 1993; Abdalati and Steffen, 1995), and AVHRR-derived albedo estimates can now be mapped (Stroeve et al., 1997). Land ice dynamics and thermodynamics, ignored in current coupled GCMs, respond to changes in the temperature and balance of accumulation and melt at upper and lower surfaces. These boundary conditions involve couplings to the atmosphere, ocean and lithosphere. Ice sheet models typically represent atmospheric boundary conditions using simple elevation-based parametrizations. However, results obtained for Antarctica using this approach (Huybrechts and Oerlemans, 1990; Fastook and Prentice, 1994) compare favourably with those using more comprehensive AGCM-derived boundary conditions (Thompson and Pollard, 1997). Predictive models of future evolution must incorporate past ice-mass variations because ice sheets continue to respond to climate change for several thousand years. Time-scales complicate the inclusion of land ice in existing coupled climate models, and it is usual in these models to regard land ice as passive, providing only a static boundary for the atmosphere. Models of ocean circulation seldom incorporate explicit treatments of melt water inputs from land ice.
Models simulate the coupled evolution of ice sheet flow, form and temperature. The flow law in Glen (1955) is commonly adopted but such models must employ a poorly justified flow enhancement factor to correctly capture the height-to-width ratios of ice sheets. In general, these models perform well in inter-comparison exercises (Huybrechts et al., 1996; Payne et al., 2000), but there is uncertainty in the predicted thermal structure and bottom melting conditions. This constitutes a potentially serious shortcoming because several models of ice sheet instability invoke thermal trigger mechanisms (MacAyeal, 1992). Ice stream flow models are limited by the current knowledge of the controlling physical processes. Outstanding issues involve the relative importance of bottom drag and lateral drag (Whillans and van der Veen, 1993; MacAyeal et al., 1995) and the representation of fast-flow and water-transport processes (Fowler and Schiavi, 1998; Hindmarsh, 1998; Tulaczyk et al., 1998). Recently, coupled models of the evolution of ice streams within ice sheets have been developed (Marshall and Clarke, 1997) which incorporate the crucial processes of ice stream onset (Bell et al., 1998) and margin migration (van der Veen and Whillans, 1996).
Floating margins of ice sheets are called ice shelves, the largest of which are found in West Antarctica and mediate the transfer of ice between fast-flowing ice streams and the Southern Ocean. Break-up of one or both of the large West Antarctic ice shelves, either associated with ice stream instability or independent of it, cannot be discounted and would be considerably more probable than complete disintegration of the West Antarctic ice sheet (Bentley, 1998; see also Chapter 11, Section 11.3.3). Such an event would be accompanied by a large increase in iceberg flux to the Southern Ocean and uncertain effects on the production of AABW and ocean circulation. Different ice shelf models compare well (MacAyeal et al., 1996); however, the stability of grounding lines remains an issue and could significantly influence the behaviour of numerical models (Hindmarsh, 1993). The suggestion that the grounding line is inherently unstable (Thomas and Bentley, 1978) is not supported by more recent modelling studies (Muszynski and Birchfield, 1987; Huybrechts and Oerlemans, 1990; Hindmarsh, 1993). However, this issue has yet to be resolved.
Megafloods of water stored beneath ice sheets have been postulated (Shaw et al., 1996) but the subject is controversial. The estimated total volume of lake water beneath Antarctica is equivalent to 10 to 35 mm of sea level rise (Dowdeswell and Siegert, 1999) but the simultaneous release of all this stored water is unlikely. The timing would be unpredictable and climate impacts, if any, would be associated with the effect of the fresh water pulse on AABW formation. Modelling such phenomena is beyond the scope of existing ice dynamics models. Although progress has been made in the understanding of ice stream processes, there are still many unanswered questions and representation of ice stream and grounding line physics in land-ice dynamics models remains rudimentary.
Other reports in this collection