{{ message }}
Repository navigation
Feature request: CICE model interface cleanup #1191
Description
Use case
The CICE model interface involves scripting and datm stream files that are out of date. Scripts for CESM2.0 set up the datm model to drive CICE with CAM4 reanalysis files. The official release of CESM2 with CICE5 is CESM2.1 and should use CAM6+DART reanalysis files or JRA55. CESM3 will use CICE6, which is currently configured in CESM2.3.
Is your feature request related to a problem?
The CICE user community is fairly dispersed and involves lots single-user scripting, imo partly because the DART scripting is out of date / related to model versions that are not common releases.
Describe your preferred solution
- Update CESM 2.0 scripting to CESM2.1, which uses CESM2 framework and CICE5, with suggested default ensemble atmospheric forcing
- Provide CESM scripting for CESM2.3, which is a CESM2 framework that uses CICE6, with suggested default ensemble atmospheric forcing.
- Remove docn and CAM4 reanalysis forcing files from the cice model directory.
- Remove CESM scripts for versions of the model that were run on the yellowstone machine.
Describe any alternatives you have considered
- Leaving everything the way it is and letting people do their own thing
Activity
kdraeder commentedon Oct 12, 2026
Here are a few thoughts, while they're fresh in my mind from the DAlooza.
@mollymwieringa and I have discussed copying the/her perturbed JRA55 30-member ensemble
from scratch to campaign storage. That would make them a more reliable option for atmospheric forcing.
The set is 14 Tb, so there's room on DAReS' campaign for more members, if those would be useful enough.
There are benefits and drawbacks to both the JRA and CAM6+DART atmospheric forcing sets.
Choosing one to be the default forcing for sea ice assimilations should get some discussion.
The existing CESM2_3_setup_ensemble has a lot of redundant andor impenetrable text in the creation
of the user_nl_datm_streams files. It would be more readable if each file were listed on a separate line,
which can be done with continuation characters. Or a lot of the text could be moved into a template file
like the one from clm/shell_scripts/cesm3_0 which I've attached. The template would have member numbers, year,
andor ... inserted by loops over a
sedcommand in CESM2_3_setup_ensemble.user_nl_datm.CPLHISTForcing.complete.txt
Or maybe these should be developed in python to be appealing to the new generation.
Will Icepack (the single column model used by CICE) have a separate DART model interface? See #475
Will a shared code module between those be useful?
Some of this may be developed in the Sea Ice DAlooza Project Ideas page