Large scale modeling of Antarctica ice flow using inverse control methods on ice rheology and basal drag.

Large scale modeling of Antarctica and Greenland ice flow is a difficult problem that needs to be addressed if we want to be able to understand the present and future evolution of these continent's mass balance. A good first approximation for modeling ice flow is to work within the shallow ice approximation, using for example the forward model developped by MacAyeal (1989). Implementing a finite element solution to this model, and working with parallel cluster technologies, we have been able to develop a model capable of modeling the entire Antarctica ice flow within a reasonable degree of accuracy. Every ice sheet/ice shelf subsystem that constitue Antarctica are taken into account Realistic modeling can only be carried out if the rheology of ice and the basal drag at the bedrock interface are known. These parameters cannot be measured, and we have to resort to inverse control methods using satellite measurements of velocities, to infer those parameters. We have developped a control method that simultaneously inverts for ice rheology and basal drag. This is important as those two parameters often play similar roles in constraining ice flow. Using this dual control method, we can target subsystems of Antarctica and gradually improve the quality of our forward model. We have so far implemented this method on the entire Pine Island and Thwaites basin, where new data for surface and thickness elevation are avaialble. Results show a realistic pattern of distribution for ice rheology and basal friction, and improvements in the forward model for the entire Antarctica continent are significant. From those results, patterns can be extrapolated for areas of Antarctica where ice rheology and basal friction are totally unknown.