Merge b4bdev 20241220

[slevis-lmwg]

For details, see the ChangeLog for tag ctsm5.3.017.

For reference, this resolves #2869.

[slevis-lmwg]

I saw Joseph's message that izumi had been restored to service, so I submitted aux_clm this morning. It seemed as though tests were going to work, but a large number "failed to initialize" in the SHAREDLIB_BUILD phase. I tried resubmitting three of these, and I get "Read-only file system" errors.

I will try again on Monday. If I see the same behavior, I will return to the original plan of merging this PR without izumi testing.

[ekluzek]

I like that this has a nice namelist control from the user perspective with the author's name attached. Although in other cases we don't use just the name -- but also the year for the reference paper. That enables different years for the same author to apply when/if that's needed. And it also makes it easy to figure out what the reference paper is.

The most important problem here is that there isn't a test for Tom's scheme here, and minimally we should make sure we have a test for both options.

I'm also glad that the basic change is nicely captured in it's own function. This can allow unit or funcitonal testing for it.

I didn't catch this when it came to b4b-dev, but there are changes I'd like to see, in terms of modularity for making things as local as possible

As such I suggest a bunch of changes, that I do think are worth doing. However, if there is a tag needed for simulations with this, I realized the changes I suggest could be done later, and they are all things that could go on b4b-dev later.

[slevis-lmwg]

I am getting new errors on izumi today, rather than the error that I reported yesterday. I have submitted aux_clm again to see what happens. Meanwhile Ithough I will move on with this merge, because I get the sense that izumi is unreliable.

[slevis-lmwg]

Submitting aux_clm after the code revisions.

[slevis-lmwg]

aux_clm

• on derecho OK same results as before
• on izumi NOT OK but this computer seems to have serious problems; e.g. all 13 gnu tests:

    FAIL ERS_Ld1200_Mmpi-serial.1x1_numaIA.I2000Clm50BgcCropQianRs.izumi_gnu.clm-cropMonthlyNoinitial GENERATE exception
    FAIL ERS_Ld600_Mmpi-serial.1x1_smallvilleIA.I1850Clm50BgcCrop.izumi_gnu.clm-cropMonthlyNoinitial GENERATE exception
    FAIL ERS_Ly3_Mmpi-serial.1x1_smallvilleIA.IHistClm50BgcCropQianRs.izumi_gnu.clm-cropMonthOutput GENERATE exception
    FAIL ERS_Ly5_Mmpi-serial.1x1_smallvilleIA.I1850Clm50BgcCrop.izumi_gnu.clm-ciso_monthly--clm-matrixcnOn GENERATE exception
    FAIL ERS_Ly5_Mmpi-serial.1x1_smallvilleIA.I1850Clm50BgcCrop.izumi_gnu.clm-ciso_monthly GENERATE exception
    FAIL SMS_D.f10_f10_mg37.I2000Clm60BgcCrop.izumi_gnu.clm-crop GENERATE exception
    FAIL SMS_D_Ld1_Mmpi-serial.f45_f45_mg37.I2000Clm50SpRs.izumi_gnu.clm-ptsRLA GENERATE exception
    FAIL SMS_D_Ld1_Mmpi-serial.f45_f45_mg37.I2000Clm50SpRs.izumi_gnu.clm-ptsROA GENERATE exception
    FAIL SMS.f10_f10_mg37.I2000Clm50BgcCrop.izumi_gnu.clm-crop GENERATE exception
    FAIL SMS_Ld5_Mmpi-serial.1x1_brazil.IHistClm60Bgc.izumi_gnu.clm-mimics GENERATE exception
    FAIL SMS_Ln9.f10_f10_mg37.I1850Clm45Bgc.izumi_gnu.clm-clm45cam4LndTuningModeZDustSoilErod GENERATE exception
    FAIL SMS_Ly3_Mmpi-serial.1x1_numaIA.I2000Clm50BgcDvCropQianRs.izumi_gnu.clm-ignor_warn_cropMonthOutputColdStart GENERATE exception
    FAIL SMS_Ly5_Mmpi-serial.1x1_smallvilleIA.IHistClm60BgcCropQianRs.izumi_gnu.clm-gregorian_cropMonthOutput GENERATE exception

Similarly with most nag and intel tests, plus a few tests say "SHAREDLIB_BUILD failed to initialize"

UPDATE: I think the "GENERATE exception" tests arise from lack of disk space on /fs/cgd/csm and we have ended up without a new baseline. In any case, izumi feels like an unnecessary time sink right now...