Comparison of Stokes Ice Sheet Models for Marine Ice Sheet Experiments
Stokes-flow ice sheet models are commonly used to generate solutions of reference for ice sheet model intercomparison test cases. To date, a single model, Elmer-Ice, has been used by the community for generating these reference solutions. We conduct a detailed comparison with a second Stokes model, FELIX-S, and show that, despite significant differences in methedologies between the two models, the two sets of model solutions converge with increasing grid resolution, as expected. This lends confidence to the practice of treating Stokes model solutions as reference solutions in model intercomparison exercises.
-
grounding line positions in Stokes models are keenly sensitive to the implementation of grounding line physics
-
despite this, solutions from two Stokes models with different numerics and different physics converge under mesh refinement
-
agreement between multiple models at adequate spatial resolution lends confidence to the practice of treating Stokes model solutions as reference solutions in model intercomparison exercises
Stokes-flow ice sheet models are commonly used to generate solutions of reference for ice sheet model intercomparison test cases. These test cases are critical for informal validation of ice sheet models as applied to particular ice dynamical problems of interest (e.g., grounding line motion) and for testing the accuracy of computationally cheaper, and numerically less complex "shallow" approximations to the Stokes-flow equations of motion. To date, a single model, Elmer-Ice, has been used by the community for generating these reference solutions. In this work, we conduct a detailed comparison between Elmer-Ice and a second Stokes model, FELIX-S, as applied to a set of marine-ice sheet model benchmark test cases. We show that, despite significant differences in methedologies between the two models, the two sets of model solutions converge with increasing grid resolution, as expected. This lends confidence to the practice of treating Stokes model solutions as reference solutions in model intercomparison exercises.