Numerics

Our group has co-founded the Ice-sheet and Sea-level System Model in partnership between Dartmouth and the NASA Jet Propulsion Laboratory. ISSM is a large scale, high-resolution, massively parallelized finite element model dedicated to ice sheet modeling and is our primary tool. Our work on improving numerical methods includes:

Data assimilation

Inverse methods, that combine ice sheet modeling and surface observations, provide the tools to tackle these questions. We have used inverse methods to investigate the patterns of basal friction under grounded ice using surface velocities derived from Satellite interferometry (Morlighem et al. 2011). More recently, we have applied these techniques at the scale of Antarctica (Morlighem et al. 2013) and we discovered that basal sliding is widespread beneath the Antarctic Ice Sheet. This suggests that coastal perturbations may be transmitted further inland than expected. More recently we have been working on automatic differentiation (with ADOLC, TAPENADE and CoDiPack) to invert for any model input parameter.

Mesh adaptation

A high mesh resolution is critical in some key regions such as grounding lines, ice fronts, or shear margins. In order to maintain a good model accuracy while minimizing the computational cost of large scale models, we have been developing:

• Static anisotropic mesh refinement (through the development of BAMG)
• Adaptive mesh refinement (AMR)

Implementation of subelement physical processes

It has been shown that tracking the grounding line at the subelement level relaxes the requirements on mesh resolution. We have been investigating the impact of subelement parameterizations on basal friction and ocean-induced melt in the vicinity of the grounding line in order to establish the most efficient strategies for accurate modeling of grounding line dynamics.

Linear solvers

The linear solver is the main bottleneck in large scale simulations. We are working on testing new generation solvers to improve the performance of the model.