I have been working with the "fake coupler" from Cecilia Bitz to get ROMS and CICE working together. In the name of "release early, release often", I have just pushed this code out to my svn branch and to github. What is the fake coupler, you ask? It is for running both models on the same grid with the same time-step. I include a Python script for getting a Netcdf CICE grid from a ROMS grid file. The two models use the same processes (same number of processes, obviously), but don't need to be on the same tiling.
You can still use the old ICE_MODEL option if you like. I took out the sfwat variable since it was bogus and put in a non-working MELT_PONDS option. Do not use the MELT_PONDS...
For the CICE coupling, see the README.CICE file. Basically, you should download the latest code (Version 5.0.4) from Los Alamos and put it somewhere. You need to apply a patch to CICE, mostly to allow it to read the Netcdf grid file. You also need to tell ROMS where the CICE files are, both source and compiled.
My recent tests have been in a Bering Sea domain originally designed as a COAWST domain - I have grids for WRF, ROMS and CICE for this thing (Rob Cermak ran WRF, I did not). I start in September with no ice and see what happens. Ice grows with both ice models, but then doesn't all melt back. More ice remains in CICE than in the other model - tuning to be done. Let me know what works for you if you decide to play too.
Meanwhile, the Norwegians are working on a more professional coupling with MCT. You can wait for that if you prefer.
CICE and ROMS
Re: CICE and ROMS
After a post in another thread, I want to say a few things about this coupling. In science one typically hangs onto code until the paper is written and all the problems have been solved. On the other hand, the Open Source way is to "release early, release often". My code is in this second camp. That means you can play too, but you might need the tools for solving whatever problems remain. There will be a more robust code and there will be papers, but not yet. The run I am doing now has Atlantic water melting too much ice - and I'm not funded to figure it out, this is just a side project. Another feature of the Open Source model is that if you find bugs, you have a responsibility to report them - and share fixes.
Using COAWST to play with this with WRF too has still more traps. For one thing, WRF isn't talking to CICE directly unless someone is working on that without telling me. For another thing, I haven't been sending John Warner an email with each update I push. Is he keeping up? I haven't checked.
Using COAWST to play with this with WRF too has still more traps. For one thing, WRF isn't talking to CICE directly unless someone is working on that without telling me. For another thing, I haven't been sending John Warner an email with each update I push. Is he keeping up? I haven't checked.