Dear all,
At our institution we plan to set up ROMS with 4DVar for operational forecasting. Since our domains cover the Arctic we will need to take sea-ice into account, but there is no dynamic ice-model with AD/TL-code included in the official branch. So far we consider two alternatives:
1) Using the ICE_SHELF option, a short assimilation window, and frequent updates of the ice sheet. The sea ice cover we can get from the national ice service (or other sources). In this way we can stick to official ROMS code, but we will at least need to adjust model hydrography due to freezing/melting etc.
2) Use a nonstandard forward nonlinear model that is coupled to a dynamical sea-ice model. At present we are using the model from Kate Hedstrom's branch, but coupling to e.g. CICE is also an option.
We want to avoid a duplication of efforts and would very much like to learn what other people have been doing. This post lists some options for sea-ice/ocean model coupling, but I assume there has been some development since then?
Many thanks in advance!
Kai Christensen
Norwegian Met. Institute