E3SM
E3SM copied to clipboard
Switch data ice-shelf melt rates to Paolo et al 2023
This data set is an Antarctic melt-rate climatology covering the years 1992-2017.
The new datasets is an improvement because:
- it is more accurate than the previous Adusumilli et al. (2020)
- it includes rerouting of fluxes that are not under ice shelves in the MPAS-Ocean mesh
- the fluxes are carefully renormalized so that the total flux is identical to the original Paolo et al. dataset
- fixes the sign of the heat flux (see https://github.com/MPAS-Dev/compass/pull/781)
https://github.com/E3SM-Ocean-Discussion/E3SM/pull/78 contains a small amount of relevant discussion. The new dataset was created in Compass following https://github.com/MPAS-Dev/compass/pull/778.
This PR is expected to be bit-for-bit for all current test configurations. It will be non-BFB for the DISMF test added in https://github.com/E3SM-Project/E3SM/pull/6229 if that were to get merged before this PR.
Testing
@darincomeau ran a 5-year B-case with these changes. Analysis from that run are here: https://web.lcrc.anl.gov/public/e3sm/diagnostic_output/ac.xylar/analysis/20240223.v3.LR.CRYO1850-DISMF.paolo.chrysalis/
The results show the expected melt rates and melt fluxes, as discussed in https://github.com/MPAS-Dev/MPAS-Analysis/pull/987.
Simulation using new DISMF file is running here
/lcrc/group/acme/ac.dcomeau/E3SMv3/20240227.v3.LR.CRYO1850-DISMF.paolo.chrysalis
MPAS-Analysis results for a 10 year B-case are here: https://web.lcrc.anl.gov/public/e3sm/diagnostic_output/ac.xylar/analysis/20240227.v3.LR.CRYO1850-DISMF.paolo.chrysalis/yrs_01-10/
@xylar -- I don't think we expect this to be BFB for current tests with DISMF?
@jonbob yes, I noted the test expected to be non-BFB above, so I guess we should change the BFB tag?
Thanks @darincomeau -- I missed that! But I'll change the BFB flag and add something to the PR description
Passes:
- SMS_Ld1.T62_oQU240wLI.GMPAS-IAF-DISMF.chrysalis_intel (dataset not changed for this resolution)
- ERS_Ld5.T62_oQU120.CMPASO-NYF.chrysalis_intel
merged to next
The only test this PR should impact is already failing on chrysalis, since it specifies debug:
SMS_D_Ld1.ne30pg2_r05_IcoswISC30E3r5.CRYO1850-DISMF.chrysalis_intel
This test also runs on pm-cpu and currently passes, so we'll expect it to have DIFFs
merged to master and expected DIFFs blessed
Thank you @jonbob! This is a really nice addition to have.
Thanks @darincomeau and @cbegeman for your reviews and other help on this work!