Conceptually, it would be great to improve the model such that the standard domain grid, from which boundary information like bedrock topopraphy in particular is obtained, could be a different (higher) resolution grid than the computational grid of the ice sheet model itself. This has the advantage of providing a lot of subgrid information to the ice-sheet model, such as bed roughness on the computational grid cell, sub-grid pinning points, grounded fraction etc. Thus even for low-resolution ice-sheet simulations, significant information about the domain in reality would be retained.
To do this would likely require some major changes to the yelmo derived type (at a minimum another sub-type), and possibly breaking changes to the API.
Conceptually, it would be great to improve the model such that the standard domain grid, from which boundary information like bedrock topopraphy in particular is obtained, could be a different (higher) resolution grid than the computational grid of the ice sheet model itself. This has the advantage of providing a lot of subgrid information to the ice-sheet model, such as bed roughness on the computational grid cell, sub-grid pinning points, grounded fraction etc. Thus even for low-resolution ice-sheet simulations, significant information about the domain in reality would be retained.