Summary: | The main objective of this thesis was to develop a method to link landform data with numerical ice sheet models, to improve the testing and refinement of ice sheet models and to allow better reconstructions of ice sheet extent, chronology and subglacial processes. Results from APCA margin matching indicated the model agreed best with the Norwegian moraines, and also provided evidence for ice margin stability along the coast of Norway. In contrast, the model had difficulty reproducing the southern and eastern moraines of the ice sheet, such that good correspondence with one moraine typically decreased the correspondence with another. The mean residual between model output and sets of glacial lineations reflected changes in ice sheet configuration during ice sheet growth and decay and indicated periods of best correspondence with field data during distinct periods within the last 100kya. Thus it is possible to evaluate when and under what dynamic conditions during a glacial cycle particular landform patterns are generated. Within the Kiruna, Sweden study area, the model corresponded with the southeasterly flow set during 76kya, 48kya, and 12kya, while corresponding with the northerly flow sets during 85kya, 68kya, 40kya, and 20kya. The model agreed best with a northerly flow set, which obtained mean residuals of only 9° at 68kya and 40kya. This fluctuation results from a shifting ice divide during ice sheet growth and decay, and the presence of predicted areas of frozen bed conditions. APCA and flow orientation analysis results were compiled to allow each model run to be ranked according to the level of correspondence with field evidence; this allowed for the isolation of the climate and input parameters that produced output that best matched geomorphologic and chronological evidence. The best results generated from the model used in this study required accentuated precipitation along the western margins (Norway), an enhanced smoothing of topography, and high Weertman sliding values. Results from this research can ...
|