To assess the amount of ice volume stored in glaciers or ice caps, a method to estimate ice thickness distribution is required for glaciers where no direct observations are available. In this study, we use an existing inverse method to estimate the bedrock topography and ice thickness of the Renland Ice Cap, East Greenland, using satellite-based observations of the surface topography. The inverse approach involves a procedure in which an ice dynamical model is used to build-up an ice cap in steady state with climate forcing from a regional climate model, and the bedrock is iteratively adjusted until the modelled and observed surface topography match. We validate our model results against information from airborne radar data and satellite observed surface velocity, and we find that the inferred ice thickness and thereby the stored total volume of the ice cap is sensitive to the assumed ice softness and basal slipperiness. The best basal model parameters for the Renland Ice Cap are determined and the best estimated total ice volume of 384 km3 is found. The Renland Ice Cap is particularly interesting because of its location at a high elevation plateau and hence assumed low sensitivity to climate change.

The isothermal, non-sliding shallow-ice approximation, combined with mass conservation, is a fundamental model for ice-sheet and glacier flow. It determines the ice extent, geometry and velocity by the solution of a free-boundary problem. In this paper, the steady-state form of this problem is solved directly, without time-stepping, thereby demonstrating a fully implicit scheme with no stability restrictions. The classical Mahaffy (1976) finite difference method is first reinterpreted as a ‘finite volume element’ scheme that has both an everywhere-defined approximate thickness function and an approximation of the conservation equation in flux integral form. From this reinterpretation an improved scheme is built by using better quadrature in the integral and upwinding on that part of the flux which is proportional to the bed gradient. The discrete equations are then solved by a parallel Newton scheme which respects the constraint that ice thickness is non-negative. The results show good accuracy on both flat-bed and bedrock-step exact solutions. The method is then applied at high resolution to model the steady-state geometry of the Greenland ice sheet, using only bedrock elevation and present-day surface mass balance as input data.