Fast Methods for Large Scale Ice Sheet Modelling

Sea level rise, as a consequence of climate change, is predicted to have vast adverse impacts on the economy and society. Pragmatic counter measures, such as infrastructures for coastal protection, will require massive investments.

Yet, our current ability for long-term planning relies on sea level rise predictions that are highly uncertain, due to inaccurate models of expected meltwater contributions from the ice sheets on Greenland and Antarctica.

Ice sheets models contain large errors because it is too computationally costly to accurately represent ice dynamics near the coastlines.

In this proposal, we will develop efficient numerical methods for ice sheet simulations with accurate descriptions of coastal dynamics.

We will mitigate numerical instabilities, which are now a bottleneck that limits the simulation time-step sizes resulting in a non-linear problem being solved unnecessarily often.

The improved stability will further allow to apply an adaptive approximation method, so that a complex physical model is used along the coastline of the ice sheet, while a faster model can be employed elsewhere.

We expect that our methods will speed-up the simulations at least 40 times, while they will also reduce memory consumption. This higher efficiency can be used to improve e.g. spatial resolution. We demonstrate the efficiency of our approach on the Antarctic Ice Sheet. The project will be performed by the applicant, a PhD student and a postdoc, with support from collaborators.