MODELING FORMER ICE SHEETS AS A TOOL FOR UNDERSTANDING GLACIAL LANDSCAPES

We use a time-dependent, two-dimensional ice flow model along four flowlines of the southern Laurentide Ice Sheet and two transects of the Scandinavian Ice Sheet to explore growth of the ice sheets and conditions at the ice-bed interface during the last interglacial-glacial cycle. The model is mainly driven by climate and incorporates flow divergence, permafrost development, and calving. Temperature input is derived from Greenland d18O data, adjusted to local paleoclimate records.

The relations between the geologic record and ice flow modeling results are threefold: Some geologic features are used as model input or help constrain input conditions. Other landscape features are used for model validation. And finally, we use the ice flow model as an exploratory tool to constrain environmental conditions during formation of glacial features whose genesis is still poorly understood and for examining regional ice sheet behavior where the geologic record yields controversial evidence.

The usefulness of ice flow modeling for glacial landscape interpretation is demonstrated here using two examples. 1) Our model results for the Scandinavian Ice Sheet favor the "nunatak theory" for western Norway, suggesting that the ice sheet did not completely cover all mountain peaks. This is in agreement with block fields and trimlines, but contrary to evidence from till fabrics and striation patterns. 2) Modeling of the Green Bay Lobe of the southern Laurentide Ice Sheet produces ice margins that were stable for at least one thousand years, high basal shear stresses, and permafrost depths of 80-200 m associated with the phases that produced extensive drumlin zones; during deglaciation, basal melt water and sliding become more important.

Some challenges remain: The model in its current form seems to be able to reproduce only first-order ice flow behavior that is climate-driven. Fast ice flow, however, which is probably responsible for short-term ice margin retreats and readvance events during later stages of deglaciation and is also related to landform development is not yet well represented in the model.