Budgell vs. CICE sea ice model with ROMS

[ecjoliver]

I'm developing a regional ocean-ice model for a subarctic fjord on the Labrador coast (Hamilton Inlet). I'm interested in getting the seasonal and interannual circulation in the fjord-system as well as the ice conditions - most notably how the circulation/atmospheric conditions affect areas of thicker/thinner ice. If possible, I would like to include a landfast ice parameterisation as well (e.g. Lemieux). I have a ROMS ocean-only (no sea ice) setup that I am happy with, and am now moving to coupling this with sea ice.

I see two main options out there for this.
1. The Budgell sea ice model as incorporated in the ROMS code itself
2. The CICE model using e.g. Kate Hedstrom's branch, or MetRoms, or COAWST.

The questions I have for those on this forum more knowledgeable that I about these options are:
a. Is there a general recommendation of the Budgell model vs. CICE? Is one vastly better than the other, or are the differences more subtle and so depends on the application?
b. In the case of the Budgell model is there a difference in how this is implemented across the different ROMS branches e.g. Kate Hedstrom's branch vs. the branch available via myroms?

I've already started implementing my setup using the ROMS-CICE model from Kate's branch, and it's going OK (aside from boundary condition problems) but now I'm re-considering if the Budgell model may serve my needs as well, and in a simpler way. The latter's clean integration with the rest of ROMS is attractive, while CICE running separately via a coupler sometimes feels a bit clumsy.

[kate]

CICE is a much fancier model, with compile time choices of the number of ice categories, choices for melt ponds, you name it. I don't know your domain well enough to know if you need the extra features. For our needs, we ran the ROMS ice because it's cheaper. Now we're moving away from that for the next generation of Arctic modeling. However, we're changing ocean models as well, but there's a lot of work to do to have everything that's in ROMS (tides, for instance). Another feature of the ROMS ice is some attempt at an open boundary given values from a larger ice-covered domain so that I have successfully run (offline) nested ice-ocean domains in ROMS.

My recent code is on github, not myroms.

[wilkin]

Some enhancements to the Budgell ice code are reported here:
"A high-resolution coupled ice-ocean model of winter circulation on the bering sea shelf. Part I: Ice model refinements and skill assessments", by Durski and Kurapov, 2019, Ocean Modelling
https://www.sciencedirect.com/science/a ... Uy1EynBFfQ

[ecjoliver]

Thanks both. The biggest difficulty I find with CICE (technically speaking) is that it seems not simple to have open boundary conditions with some lateral boundary forcing of ice conditions.

[nicholass]

If it helps, MET Norway has been running MetROMS (with CICE) relaxing to TOPAZ on the lateral boundaries.
There's a version with CICE5 (https://github.com/metno/metroms/pull/21), and CICE6 to come.

Any thoughts on whether it's been tried or feasible to adapt the open boundaries for ice in ROMS to CICE?

[kate]

There's a CICE forum at https://bb.cgd.ucar.edu/cesm/forums/cic ... rtium.146/ which would be another place to ask this sort of question.

[pduarte]

Regarding the comment about CICE: "it seems not simple to have open boundary conditions with some lateral boundary forcing of ice conditions". This issue has been addressed already. I am currently running a coupled ROMS+CICE model for a large region around Svalbard that uses time-varying sea-ice boundary conditions from a larger scale model. This is similar to what Met Norway is using. The main modifications were in ice_restoring and ice_forcing routines. Details may be provided upon request.

[detraceyb]

@ecjoliver, I would caution you to first dig into some references on the effects of resolution on ice models. Your study area is small, and typically sea ice models have been developed for climate applications, and resolutions at which the assumption of a continuum (of thicknesses and areas) holds true. I am out of touch with the current state of the art in marginal ice zone modeling, but it used to be pretty poor. There is an ocean modeling group at the Bedford Institute of Oceanography that primarily uses NEMO and CICE. PM me if you want me to ask around for you about any regional ice modeling in progress.