Comparison with previous studies

It is of interest to make a cornparison of these model sirnulations with results obtained in previous studies. Estirnates of the Antarctic contribution to sea level lowering at maximum glaciation have been made earlier in numerical studies by Oerlemans (1982b) and Budd and Srnith (1982). These ice flow rnodels operated on a coarser 100 km grid and excluded therrnodynamics and ice shelf flow. In comrnon with the present result, both studies predicted a large increase in the volume of the West Antarctic ice sheet if sea level was lowered by 100-1 50 m. Ice volume increases with respect to the observations turned out to correspond to sea-level falls of between 27.5 m (Oerlernans, 1982b) and 33 rn (Budd and Srnith, 1982). However, these rnodels were unable to sirnulate the Ronne-Filchner ice shelf in the reference state and strongly overpredicted the ice Cover over the Antarctic Peninsula. As a result, calculated ice volumes for present-day conditions were already far in excess of the observations by 25-50%.

A sirnilarly high estirnate for the late Wisconsin contribution to global sea level was also reported by Nakada and Lambeck (1988). They believe that Holocene retreat of the Antarctic ice sheet could have added as much as 35 rn to the world's oceans. This result was obtained in an indirect way, by subtracting the equivalent of the ice volurne of fully-grown northern hemisphere ice sheets from the 130 m of total sea level change and matching the resulting rnelting histories with calculated and observed vertical movernents of the Earth's rnantle.

Although modelling approaches are fundamentally different, a further cornparison can be made with the steady state reconstruction of the CLIMAP- group (Stuiver et al., 1981; cf. fig. 3.5). In their approach, ice sheet margins were not generated by the model, but 'known' locations of domes, flowlines and rnargins served as constraints to calculate forrner elevations and flowline profiles. This was done by ernploying plastic ice flow theory, where the yield Stress for each point was calculated in a cornplicated way from such elernents as basal thermal conditions (slidingl no sliding) and transverse flowline

geometry. However, their reconstructed ice thicknesses were also substantially larger than those produced in our simulation (fig. 6.2), although the ice sheet extent agrees more closely. In particular over the West Antarctic ice sheet, the CLIMAP surface elevations are consistently higher by more than 500 m. The total ice volume during the LGM would in their reconstruction have increased to 37.1 X 106 km3. This is equivalent to 24.7 m of sea-level fall, of which 16.3 m originales from West Antarctica and 8.4 m from East Antarctica.

Using a rock-to-ice density ratio of 3 instead of their preferred value of 4 would result in even larger volumes, yielding corresponding figures of 29.2 m, 17.7 m and 11.5 m (Stuiver et al., 1981).

In contrast to these studies, we obtained a value of only 16 m. This can be explained by a more refined treatment of ice mechanics, especially in the basal layers, and the incorporation of thermo-mechanical coupling. Also the inclusion of lower accumulation rates during the LGM, as indicated by the Vostok ice core, represents a significant contribution to our result. These additional factors were not included in any of the studies discussed above.

6.3. SUMMARY

In this chapter, we discussed results from a number of steady state experiments with the complete model, in which all internal degrees of freedom were taken into account. This approach helped to disentangle the complex interaction of environmental controls (accumulation rate, ice temperature and sea level) On the behaviour of the ice sheet. The main points that anse of this study can be listed as:

(i) first, it appears that the model is able to account for the major characteristics of the ice sheet's flow Pattern, including a correct simulation of the main ice divides, ice domes and major ice shelves. The calculated grounding line under present-day environmental conditions is generally close to its observed position. Somewhat larger deviations include a slight recession (order of magnitude: 100 km) along the most overdeepened outlet glaciers in East Antarctica and the incorporation of several small ice-covered off-shore islands into the main ice sheet in West Antarctica. These imperfections may

respectively be attributed to a possible overprediction of basal sliding and the coarse representation of deep bedrock channels.

(ii) in line with glacial-geological evidente, the rnost pronounced ice sheet fluctuations occur in the West Antarctic ice sheet. Most of the implied spreading of grounded ice across the shallow ROSS and Weddell Seas during glacial periods can be attributed to lower global sea levels.

(iii) lower ice ternperatures also lead to expanded ice Cover and larger ice thicknesses, but the associated accumulation rate reduction largely counteracts this effect. In the full-glacial state, surface elevations over the East Antarctic plateau do not change much.

(iv) grounded ice in the ROSS and Weddell Seas is always at the pressure rnelting point and basal sliding occurs. This results in srnaller surface slopes compared to the East Antarctic ice sheet.

(V) according to our steady state reconstruction, the Antarctic ice sheet may contribute sorne 16 m (relative to the reference run) to global sea level lowering at rnaximum glaciation. This is substantially iess than other figures put forward in previous studies. In particular, our ice thicknesses do not Support the CLIMAP reconstruction of Stuiver et al. (1981), even though the ice sheet extent agrees better. This is because of the more sophisticated modelling, the incorporation of thermornechanical coupling and the inclusion of a more realistic clirnatic forcing.

(vi) the low accurnulation rates prevailing over the ice sheet lead to long response time scales. The model needed up to 30000-40000 years to relax to a full-grown West Antarctic ice sheet after a sudden change in sea level was applied.

An implication of (vi) is that a steady state approach may not really be appropriate to reconstruct the ice sheet's history during a glacial cycle. In other words, since environmental conditions in the Antarctic have not been stable during the long times needed to reach equilibrium, it is unlikely that the ice sheet ever reached a stationary state. This would reduce the figure of 16 m at the Last Glacial Maximum obtained here even further. Consequently, the

problem should be envisaged as time-dependent and an attempt should be made to simulate the ice sheet during a complete glacial cycle. This is the subject of the next chapter.