On the Identification of the Basal Drag Parameter in Ice Sheet Models Using L‐Curves

Abstract The unknown and not observable basal drag distribution underneath the glaciers strongly impacts ice flow speeds and, with that, the ongoing mass loss of the Antarctic Ice Sheet. Therefore, basal drag is required for precise ice sheet modeling and accurate projections of future sea‐level rise. This can be achieved by applying an inverse method based on observed ice surface velocity data. The forward model equations, including boundary conditions, represent the ice dynamics in an approximated way. The ice‐base boundary condition is the main focus here, as it describes a non‐linear friction law. This law depends on the unknown basal drag parameter determined by the inversion and utilizes an effective pressure from a subglacial hydrology model. The inverse method minimizes a cost function consisting of the sum of a term quantifying the misfit between simulated and observed surface velocities and a regularization term to penalize unrealistic oscillations in the basal drag parameter. An L‐curve analysis determines the optimal weighting of both cost function terms. Here, we perform inversions for three different domains of Antarctica, comprising about 9.4 Mio , to compare the variability of the resulting basal drag and the L‐curves. The results present a low basal drag, as well as a low variability, predominating over large parts of the interior of EAIS and WAIS. In contrast, some fast‐flowing glaciers reveal a patchy pattern of alternating high and low basal drag. In addition, parts of the grounding line exhibit a high basal drag, which potentially affects the future retreat behavior of the ice sheets.