search

ESCOMP / CMEPS

NUOPC Community Mediator for Earth Prediction Systems
https://escomp.github.io/CMEPS/
24 stars 79 forks source link

Restart frequency of CICE inconsistent with other components #459

Closed gustavo-marques closed 4 months ago

gustavo-marques commented 6 months ago

I am running out-of-the-box G cases using CESM_alpha17f. The issue can be reproduced using:

/glade/work/gmarques/cesm.sandboxes/cesm2_3_alpha17f/cime/scripts/create_newcase --case g.e23_a17f.GJRAv4.TL319_t232_zstar_N65.2024.056 --compset 2000_DATM%JRA-1p4-2018_SLND_CICE_MOM6_DROF%JRA-1p4-2018_SGLC_SWAV_SESP --res TL319_t232 --run-unsupported --driver nuopc

with the following modifications:

REST_OPTION  =   nyears
REST_N    =     5

CICE writes a restart and a rpointer every year, while the rest of the system writes them every 5 years. I only noticed this inconsistency because one of my cases was aborted in year 7 and when I re-submitted it, there was a mismatch in the clocks.

I'm cc'ing @dabail10 in case this issue should be moved to CICE.

dabail10 commented 6 months ago

Ah, I think I know the problem. The internal namelist restart parameters for CICE is still active. I thought these were reset in the cap, but it is doing a force_restart_now. I know the fix. If you want a temporary solution set the following in user_nl_cice:

dumpfreq = 'x','x','x','x','x'

dabail10 commented 4 months ago

This is fixed in cesm_cice6_5_0_12.