A deglacial model for Antarctica: geological constraints and glaciological modelling as a basis for a new model of Antarctic glacial isostatic adjustment

We present a new reconstruction of the Antarctic Ice Sheets between 20 ka BP and the present day. Our reconstruction is derived using a numerical model to generate a physically-consistent ice surface across the whole of the continent. We define the extent of the ice sheet at five time slices; 20, 15, 10, 5 and 0 ka BP, assuming an equilibrium state for the 20 ka BP time slice, and a transient state for the deglacial time slices. The evolution of the ice sheet within the numerical model is driven by variations in temperature, accumulation rate, and relative sea level. In order to reconstruct the concave profile of the ice sheet in marine-grounded regions, such as the Ross and Weddell Seas, we force our model to develop channels of faster flow by defining greater basal sliding along the trajectory of former ice streams. We find a strong dependence upon the basal sliding parameters, and also the position of the grounding line. We use an extensive data base of geological and glaciological data to constrain our ice-sheet reconstruction. Grounding-line extent is prescribed from marine geological data and we test ice-sheet thickness against onshore geological data at 62 sites. Of the five time slices considered, our 20 ka BP reconstruction is the best constrained by data and has an RMS misfit of 147.6 m when compared to observations of ice thickness change between 20 ka BP and the present day. Across all time slices there are large regions of the ice-sheet which are poorly constrained, especially after 20 ka BP. We estimate the spatial distribution of uncertainty in our ice-sheet reconstruction, and note that the solutions are least reliable in regions of complex topography. We predict that the Antarctic Ice Sheets contributed 9 +/- 1.5 m of eustatic sea level to the global ocean between 20 ka BP and the present, and our reconstruction with minimum misfit contributes similar to 8 m eustatic sea level during this period. These values, which we argue are an upper bound, are lower than many previous estimates. The reconstructed pattern of ice unloading can serve as a new input for glacial isostatic models. (C) 2011 Elsevier Ltd. All rights reserved.