alpha-ctsm5.4.CMIP7.12.ctsm5.3.075: Add support for CLM_CMIP_ERA (= "cmip6" or "cmip7")

[slevis-lmwg]

Description of changes

Definition of DONE in the issue #3471

Specific notes

Contributors other than yourself, if any:
@ekluzek

CTSM Issues Fixed (include github issue #):
Resolves #3471
Resolves #3311
Resolves #1733
Resolves #3250
Half of #3050

Are answers expected to change (and if so in what way)?
No.

Any User Interface Changes (namelist or namelist defaults changes)?
The CLM_CMIP_ERA flag became necessary because we will not have SSP and NOANTHRO pft raw datasets for CTSM5.4. This means that SSP and NOANTHRO cases default to cmip6 fsurdat/landuse files, while other cases default to cmip6 fsurdat/landuse files.

Does this create a need to change or add documentation? Did you do so?
May need to update #3364

Testing performed, if any:
So far...

./create_test SMS_Ld5_Mmpi-serial.1x1_brazil.IHistClm60Bgc.derecho_gnu.clm-mimics
./create_test SMS_D_Ld5.f10_f10_mg37.ISSP245Clm60BgcCropCrujra.derecho_intel.clm-default
./create_test SMS_D_Ld5.f09_t232.ISSP245Clm60BgcCropCrujra.derecho_intel.clm-default

• The first two PASS. They correctly set cmip_era = "cmip7" and "cmip6" respectively, but they do not use that information because these grids do not have cmip7 1850 fsurdat files, yet.
• The third incorrectly picks up the cmip7 fsurdat file because I need corrections in CLMBuildNamelist.pm.

[slevis-lmwg]

@ekluzek this is how far I got today. The next two days will be very busy, and I may take Friday off. I would benefit from your corrections and/or comments on where to focus next, as this perl/xml interface is not my forte :-)

[slevis-lmwg]

Submitting the same three tests as above.

[slevis-lmwg]

Submitting the same three tests as above.

Same behavior in the first two, as I expected and hoped.
Behavior corrected now in the third, also as expected and hoped.

• aux_clm on izumi and derecho
• ctsm_sci retest on derecho (see checkbox below):

./run_sys_tests -s ctsm_sci --skip-compare -g ctsm_sci-alpha-ctsm5.4.CMIP7.12.ctsm5.3.075
./run_sys_tests -s aux_clm -c alpha-ctsm5.4.CMIP7.11.ctsm5.3.075 -g alpha-ctsm5.4.CMIP7.12.ctsm5.3.075

• aux_clm FAIL RUNs to address:

ERP_D_Ld3_PS.f09_g17.I2000Clm50Sp.derecho_intel.clm-prescribed

floating divide by zero
899 CanopyFluxesMod.F90
Did it pass before the fsurdat changed? If I were to switch back the fsurdat, I think I would need to re-test the ERI case below.

ctsm_sci FAIL RUNs to address:

ERI_Ld9.f09_t232.I2000Clm60BgcCropCrujra.derecho_intel.clm-default
SMS.f09_g17.I1850Clm50BgcCropG.derecho_intel.clm-default

The ERI test points to 5.4 finidat and 5.3 fsurdat; update the fsurdat?
From meeting with @ekluzek: Yes, make a temporary fsurdat and update namelist_defaults.

• This did NOT fix the need for use_init_interp. Looking at git blame in namelist_defaults_ctsm.xml current line 1515 suggests that this line has included use_init_interp historically. I removed it last month with 71f2db2 and may have not run ctsm_sci on it. So I'm adding back use_init_interp to this line and retesting.

The SMS test points to 5.3 finidat and 5.4 fsurdat; update the finidat?
From meeting with @ekluzek: No, update namelist_defaults to use_init_interp = TRUE.

• Retest is in the queue
• Confirm that tests failing in ISSP tests need to point to CMIP6 ctsm5.3 fsurdat/landuse files to PASS #3250 passed and remove from expected fails
• After resolving the last failure, I plan on submitting aux_clm and ctsm_sci again, while I work on build-namelist_test.pl

[slevis-lmwg]

Regarding most recent commit 14f647d:
For the record, this MODEL_BUILD passes also in my b4b-dev testing (ctsm5.3.077-30-gefe185a89).

On a different note, as per today's software meeting (2025/10/2):
Add cmip6 to IHist namelist_defaults.

[slevis-lmwg]

@ekluzek now that I've added back the cmip6 defaults for IHist cases, it seems silly to omit them for I2000 cases, so I will add them for consistency.

[slevis-lmwg]

UPDATE FROM MEETING WITH @ekluzek

Skip this list below. Instead, consider the "lai streams" test an expected failure and merge this PR. Resolve the expected failure in the later PR where I add Peter's new raw lai dataset and new raw noanthro dataset.

• Expect new raw lai dataset from Peter L.
• Pointing now to that, repeat make all to ensure it still works
• Point to the new fsurdat/landuse files in namelist_defaults
• Repeat this PR's complete testing:
  • ./build-namelist_test.pl
  • ctsm_sci
  • aux_clm
  • python

[slevis-lmwg]

@ekluzek this PR is ready for me to merge and needs your approval first.

[ekluzek]

This is great, We had already gone through this. I had one comment from our discussion that I neglected to push go on earlier.

I also have some comments/questions on the XML variable description.

Approving, this is ready to go from my perspective, you can decide on how to handle my suggestions. It would be good to get the other PR to ctsm5.4 alpha branch, and then for this one run baselines and bring it in right away.

[slevis-lmwg]

My testing is up-to-date:

• aux_clm in tests_1001-161326de and tests_1001-163228iz (baselines: alpha-ctsm5.4.CMIP7.12.ctsm5.3.075)
• ctsm_sci in tests_0930-102805de (baseline: ctsm_sci-alpha-ctsm5.4.CMIP7.12.ctsm5.3.075)
• ./build-namelist_test.pl in /glade/derecho/scratch/slevis/temp_work/ctsm5.4/mk_ctsm54_datasets/bld/unit_testers/test.out