Comparing a thermo-mechanical Weichselian ice sheet reconstruction to GIA driven reconstructions: aspects of earth response and ice configuration

Abstract. In this study we compare a recent reconstruction of the Weichselian ice-sheet as simulated by the University of Main ice-sheet model (UMISM) to two reconstructions commonly used in glacial isostatic adjustment (GIA) modeling: ICE-5G and ANU (also known as RSES). The UMISM reconstruction is carried out on a regional scale based on thermo-mechanical modelling whereas ANU and ICE-5G are global models based on the sea-level equation. The Weichselian ice-sheet in the three models are compared directly in terms of ice volume, extent and thickness, as well as in terms of predicted glacial isostatic adjustment in Fennoscandia. The three reconstructions display significant differences. UMISM and ANU includes phases of pronounced advance and retreat prior to the last glacial maximum (LGM), whereas the thickness and areal extent of the ICE-5G ice-sheet is more or less constant up until LGM. The final retreat of the ice-sheet initiates at earliest time in ICE-5G and latest in UMISM, while ice free conditions are reached earliest in UMISM and latest in ICE-5G. The post-LGM deglaciation style also differs notably between the ice models. While the UMISM simulation includes two temporary halts in the deglaciation, the later during the Younger Dryas, ANU only includes a decreased deglaciation rate during Younger Dryas and ICE-5G retreats at a relatively constant pace after an initial slow phase. Moreover, ANU and ICE-5G melt relatively uniformly over the entire ice-sheet in contrast to UMISM which melts preferentially from the edges. We find that all three reconstructions fit the present day uplift rates over Fennoscandia and the observed relative sea-level curve along the Ångerman river equally well, albeit with different optimal earth model parameters. Given identical earth models, ICE-5G predicts the fastest present day uplift rates and ANU the slowest, ANU also prefers the thinnest lithosphere. Moreover, only for ANU can a unique best fit model be determined. For UMISM and ICE-5G there is a range of earth models that can reproduce the present day uplift rates equally well. This is understood from the higher present day uplift rates predicted by ICE-5G and UMISM, which results in a bifurcation in the best fit mantle viscosity. Comparison of the uplift histories predicted by the ice-sheets indicate that inclusion of relative sea-level data in the data fit can reduce the observed ambiguity. We study the areal distributions of present day residual surface velocities in Fennoscandia and show that all three reconstructions generally over-predict velocities in southwestern Fennoscandia and that there are large differences in the fit to the observational data in Finland and northernmost Sweden and Norway. These difference may provide input to further enhancements of the ice-sheet reconstructions.