Fix single_column logic

[billsacks]

Description of changes

There is no ATM_DOMAIN_FILE; this should be PTS_DOMAINFILE (as noted in ESCOMP/CMEPS#164)

This will result in the following variables being set correctly in single_column mode, for datm, dice and docn:

• model_meshfile (null; I think this should have no impact)
• model_maskfile (null; I think this should have no impact)
• nx_global (this impacts cpl hist and rest files, but I think not in a critical way)
• ny_global (this impacts cpl hist and rest files, but I think not in a critical way)

Resolves #419

Specific notes

Contributors other than yourself, if any: @mvertens

CDEPS Issues Fixed (include github issue #):

• Resolves single_column logic in buildnml is incorrect #419

Are there dependencies on other component PRs (if so list): no

Are changes expected to change answers (bfb, different to roundoff, more substantial): bfb, but change in dimension sizes for some cpl hist fields in single-column tests will lead to baseline failures for these single-column tests

Any User Interface Changes (namelist or namelist defaults changes): For datm, dice and docn, in single_column mode, model_meshfile and model_maskfile are set to null, and nx_global and ny_global are set to 1

Testing performed (e.g. aux_cdeps, CESM prealpha, etc):

Ran testing on these changes along with CMEPS changes to remove single-column logic (ESCOMP/CMEPS#679). The following test results describe the results from testing these two changes together.

Ran the following, all with baseline comparisons:

(1) aux_cdeps

(2) aux_cime_baselines

(3) The two tests from the test_scam test list (but moved to derecho):

SCT_D_Ln7.ne3_ne3_mg37.QPC5.derecho_intel.cam-scm_prep
SCT_D_Ln7.ne3_ne3_mg37.QPC6.derecho_intel.cam-scm_prep_c6

(4) A G compset test (since G compsets weren't covered in the other testing):

SMS_D.TL319_t232.G_JRA_RYF.derecho_intel

(5) A selection of single-point tests from aux_clm - the derecho tests suggested in ESCOMP/CMEPS#676 (comment) (not the izumi tests from there, though):

ERS_D_Ld5_Mmpi-serial.1x1_mexicocityMEX.I1PtClm60SpRs.derecho_gnu.clm-CLM1PTStartDate
ERS_D_Mmpi-serial_Ld5.1x1_brazil.I2000Clm50FatesRs.derecho_gnu.clm-FatesCold
ERS_Ly3_Mmpi-serial.1x1_smallvilleIA.IHistClm50BgcCropQianRs.derecho_gnu.clm-cropMonthOutput
ERS_Ly5_Mmpi-serial.1x1_smallvilleIA.I1850Clm50BgcCrop.derecho_gnu.clm-ciso_monthly
ERS_Ly6_Mmpi-serial.1x1_smallvilleIA.IHistClm50BgcCropQianRs.derecho_intel.clm-cropMonthOutput
SMS_D_Ld1_Mmpi-serial.ne3_ne3_mg37.I2000Clm50SpRs.derecho_gnu.clm-ptsRLA
SMS_D_Ld1_Mmpi-serial.ne3_ne3_mg37.I2000Clm50SpRs.derecho_gnu.clm-ptsROA
SMS_D_Ld1_Mmpi-serial.ne3_ne3_mg37.I2000Clm50SpRs.derecho_intel.clm-ptsRLA
SMS_D_Ly6_Mmpi-serial.1x1_smallvilleIA.IHistClm45BgcCropQianRs.derecho_intel.clm-cropMonthOutput
SMS_Ld10_D_Mmpi-serial.CLM_USRDAT.I1PtClm60Bgc.derecho_gnu.clm-default--clm-NEON-HARV
SMS_Ld10_D_Mmpi-serial.CLM_USRDAT.I1PtClm60Fates.derecho_intel.clm-FatesFireLightningPopDens--clm-NEON-FATES-NIWO
SMS_Ld10_D_Mmpi-serial.CLM_USRDAT.I1PtClm60SpRs.derecho_intel.clm-default--clm-NEON-TOOL--clm-nofireemis
SMS_Lm3_D_Mmpi-serial.1x1_brazil.I2000Clm50FatesCruRsGs.derecho_intel.clm-FatesColdHydro
SMS_Ly3_Mmpi-serial.1x1_numaIA.I2000Clm50BgcDvCropQianRs.derecho_gnu.clm-ignor_warn_cropMonthOutputColdStart
SSPMATRIXCN_Ly5_Mmpi-serial.1x1_numaIA.I2000Clm60BgcCropQianRs.derecho_intel.clm-ciso_monthly
SUBSETDATAPOINT_Ld5_D_Mmpi-serial.CLM_USRDAT.I2000Clm60BgcCropCrujra.derecho_intel.clm-default
SUBSETDATAREGION_Ld5_D_Mmpi-serial.CLM_USRDAT.I2000Clm60BgcCropCrujra.derecho_intel.clm-default

All tests passed (other than expected failures) and were bit-for-bit. SMS_Ln9_P1.T42_T42.2000_DATM%QIA_SLND_SICE_DOCN%DOM_SROF_SGLC_SWAV.derecho_intel.datm-scam has different dimension sizes in the cpl hist file (as expected), but when I used ncks to extract just point 1 from the baseline cpl hist file, it's bit-for-bit.

For the two SCP tests, in the baseline, I subset the ocn dimensions to take just the first point (keeping the atm dimensions as is because there is apparently still an issue in nx/ny on the atm side when running with CAM), and then compared the atm & cpl files from case2 against baseline (since these aren't normally compared, and these are the important single-column files); these were bit-for-bit.

Differences in NLCOMP (all expected):

• I compset 1x1 tests now have mesh_mask UNSET instead of null
• I1Pt CLM_USRDAT tests now have mesh_mask UNSET instead of null
• I compset pts tests now have mesh_mask set to a file instead of null; datm model_maskfile and model_meshfile null instead of set, and nx_global and ny_global 1 instead of 0
• I compset SUBSETDATAPOINT CLM_USRDAT has mesh_mask UNSET instead of null (no substantive diffs in SUBSETDATAREGION namelists)
• SMS_Ld5_P1.1x1_mexicocityMEX.2000_DATM%1PT_SLND_SICE_SOCN_SROF_SGLC_SWAV_SESP.derecho_intel.datm-1PT has mesh_mask UNSET instead of null
• SMS_Ln9_P1.T42_T42.2000_DATM%QIA_SLND_SICE_DOCN%DOM_SROF_SGLC_SWAV.derecho_intel.datm-scam has mesh_mask set to a file rather than null; and for both docn_nml and datm_nml, model_maskfile and model_meshfile are now null rather than set, and nx_global and ny_global are 1 rather than 128/64
• SCT_D_Ln7.ne3_ne3_mg37.QPC6.derecho_intel.cam-scm_prep_c6 and SCT_D_Ln7.ne3_ne3_mg37.QPC5.derecho_intel.cam-scm_prep: case2 has diffs: mesh_mask now set to a file instead of null; docn_in has model_meshfile null instnead of a file, and nx_global 1 instead of 488

Hashes used for testing: ctsm5.4.047 for the selected tests from aux_clm; cesm3_0_alpha09d for others

[billsacks]

@ekluzek and @jtruesdal - I have run pretty extensive testing on this (see testing description in the top-level comment) so feel pretty good about it, but let me know if you want to review this or do some additional testing before it's merged. In particular, the only single-column tests I could find with CAM were:

SCT_D_Ln7.ne3_ne3_mg37.QPC5.derecho_intel.cam-scm_prep
SCT_D_Ln7.ne3_ne3_mg37.QPC6.derecho_intel.cam-scm_prep_c6

@jtruesdal - should any other testing be run on CAM single-column?

[jtruesdal]

@billsacks Thanks for your efforts here. The test you identified will make a bit for bit comparison with a column in a global run. I will be testing the other IOP files this afternoon just to make sure all our default cases run. I'll reply here when I'm done.

[ekluzek]

Looks good to me. And the testing gives me confidence that it's correct. Thanks for working on this.

[mvertens]

@billsacks - Thanks for your extensive testing on this!