Whether fast ice streaming is accommodated by hard-bed sliding or till deformation, at the most basic level it is enabled by the presence of subglacial water at pressures close to the overburden pressure (e.g. Engelhardt & Kamb, 1997; Engelhardt et al.,
1990b; Kamb, 1991, 2001). Theoretical calculations of basal melting/freezing rates indicate that, at least in the Siple Coast region, basal melting predominates inland, particularly beneath ice-stream tributaries, where it takes place at rates of several millimetres per year (Joughin et al., 2003a, 2004; Vogel et al., 2003). Calculations for ice-stream trunks are more ambiguous and suggest that these experience either basal melting or freezing at rates of a few millimetres per year. Basal freezing is particularly likely just upstream of grounding lines, near which ice streams may be susceptible to stoppage unless there is a sufficient supply of water from upstream (Bougamont et al., 2003a,b). The spatial distribution of melting and freezing rates predicted by best available models strongly suggests a need for a regional subice-stream water transport, which should move subglacial water from the inland areas of production toward the marginal zone. This is in agreement with borehole observations, which included the discovery of a water-filled subglacial cavity in the palaeomargin of Kamb Ice Stream (Engelhardt & Kamb, 1997; Kamb, 2001; Carsey et al., 2002). However, these recent results are inconsistent with the proposition of Tulaczyk et al. (2000b) that ice-stream beds may have no organized regional drainage (their 'undrained bed model').
It is somewhat unfortunate that understanding subglacial water generation, storage and transport is key to understanding ice-stream flow rates and mass balance because it is difficult to make observations of subglacial water storage and flow at length-scales over which ice flow rates are determined (several ice thicknesses or more). The best direct data, thus far, come from borehole experiments, which sample relatively short scales (fraction of a metre to dozens of metres). Nonetheless, even extensive borehole investigations conducted over a dozen of years on the Whillans, Kamb, and Bindschadler Ice Streams did not produce an unequivocal picture of the physics of water flow beneath soft-bedded ice streams (Engelhardt & Kamb, 1997; Kamb, 2001). What can be positively concluded from these borehole studies is that subglacial water is present beneath ice streams at high pressure, either as till pore water or as free water occurring at the ice-till interface. Although the water pressure is close to the overburden pressure, it is not quite established how close. Borehole water-level measurements suggest effective stress (overburden less water pressure) of less than ca. 100kPa whereas till properties indicate less than ca. 10kPa (Tulaczyk et al., 2001a). The presence of a relatively widespread subglacial water film with a thickness of the order of millimetres or more (e.g. Weertman & Birchfield, 1983; Alley et al., 1986a, 1987b, 1989) is not supported by borehole observations (Engelhardt & Kamb, 1997; Kamb, 2001). Inflow of borehole water into the subglacial system was able to create new accommodation space in the form of a water gap, which overprinted the 'pristine' subglacial water system. As has been observed on mountain glaciers, the presence of boreholes perturbs subglacial conditions. In addition, measurements made in an individual borehole may be only locally representative. These complications hinder the use of borehole experiments in constructing models of subice-stream water flow useful in simulations of ice flow (e.g. Harper & Humphrey, 1995; Harper et al., 2002).
However imperfect borehole observations of subglacial water systems are, not too many viable alternatives for studying subglacial water flow and storage exist. A geophysical technique, which could provide regional-scale constraints on subglacial water systems, is desirable. However, thus far, geophysical techniques provide either only qualitative constraints or approximate quantitative constraints on selected properties (e.g. Blankenship et al., 1986,1987; Anandakrishnan & Alley, 1997a,b). For instance, ice-penetrating radar can successfully map out areas of'wet' and 'frozen' bed (Bentley et al., 1998) but it is much more difficult to quantify spatial changes in subglacial water abundance within the 'wet' zones from radar surveys alone (Gades et al., 2000; Catania et al., 2003). Seismic surveys have been also valuable in showing spatial variability of subice-stream beds but have difficulty with quantifying properties of subglacial water systems (Anandakrishnan & Bentley, 1993; Atre & Bentley, 1993; Anandakrishnan & Alley, 1997a,b; Smith, 1997a,b; Vaughan et al., 2003b).
Recent advances in ground-based, airborne and satellite geodetic techniques offer the tantalizing possibility that regional-scale subglacial water storage (length-scales of several ice thicknesses or greater) may be monitored through accurate measurements of ice-surface elevation changes. It has been long known that changes in subglacial water storage may result in large (ca. 0.1-1 m) variations in ice-surface elevation on mountain glaciers (e.g. Iken et al., 1996; Fatland & Lingle, 2002). However, no similar observations have been made on ice streams and ice sheets until recently, when Grey et al. (2005) reported oval-shaped inflation and deflation of the ice surface on several Siple Coast ice streams based on satellite radar interferometry. These localized elevation changes of between ca. 0.5 and ca. 1 m occurred over a time period of 24 days. Because of their high magnitude and localized nature, the observed ice-surface changes are best explained by filling and drainage of subglacial lakes, which are several kilometres across (Grey et al., 2005). In the upper part of the Kamb Ice Stream, the data imply drainage of ca. 20 million m3 from the subglacial lake within the observation period. This indicates that a significant focusing of water flow into subglacial lakes occurs in the region because this much water is produced in 24 days over an area ca. 100,000km2 if one assumes a reasonable basal melting rate of several millimetres per year (Joughin et al., 2003a, 2004; Vogel et al., 2003). This has significant implications for models of subice-stream water flow because until now their basic assumption was that no large changes in subice-stream water storage occur over such temporal scales and that subglacial water inputs and outputs are more-or-less in balance at all times (Alley et al., 1989; Kamb, 1991, 2001; Walder & Fowler, 1994; Ng, 2000b). It is as yet unknown whether the observed variability in subglacial water storage and flow has its reflection in changes in ice-stream velocity. Nonetheless, the observations reported by Grey et al. (2005) call for a paradigm shift in understanding and quantitative treatment of subice-stream water flow to account for the influence of water storage in subglacial lakes on the rate and physical nature of subice-stream water drainage.
Was this article helpful?