Completing the FATES-CLM nitrogen coupling

[slevis-lmwg]

I think of this as the parent issue: NGEET/fates#1270

Currently we can only send synthetic uptake boundary conditions to FATES, and send nothing back to CLM.

This issue will address coupling to CLM's:

• Uptake competition code
• Nitrogen fixation code

In meeting with @rgknox and @adrifoster we discussed that I should focus on the
else !----------NITRIF_DENITRIF-------------!
section of SoilBiogeochemCompetitionMod.F90 (<-- for guidance see Allocation in E3SM-Project/E3SM#4325)
and on new bc_in lines in utils/clmfates_interfaceMod.F90.
@rgknox suggested that for now FATES need not worry about .not. use_nitrif_denitrif.

Testing starting from ctsm5.3.065 (fates: sci.1.84.0_api.40.0.0):
A motivation behind this update is to perform single-point FATES-MIMICS simulations, so I modified this existing test as below
ERS_D_Ld30.f45_f45_mg37.I2000Clm50FatesCruRsGs.izumi_nag.clm-FatesColdPRT2
Side notes about PRT2 tests:

• This one fails on izumi_nag as per Runtime fail due to timeout in fates PRT2 test on izumi NGEET/fates#983
• Currently PRT2 tests set fates_cnp_prescribed_nuptake = 1 and fates_cnp_prescribed_puptake =1 as per changing prescribed N and P uptake to 1 in the FATES PRT2 test #2624

Single-point tests
PASS ERS_D_Ld30.1x1_brazil.I2000Clm60FatesCrujraRs.derecho_intel.clm-FatesColdPRT2
Turning on mimics shows that we will need to troubleshoot the FATES-MIMICS coupling:
FAIL ERS_D_Ld30.1x1_brazil.I2000Clm60FatesCrujraRs.derecho_intel.clm-FatesColdPRT2--clm-mimicsFatesCold
Fates and fireemis cannot both be true, so...
FAIL ERS_D_Ld30.1x1_brazil.I2000Clm60FatesCrujraRs.derecho_intel.clm-FatesColdPRT2--clm-mimicsFatesCold--clm-nofireemis DUMPS CORE and states "floating overflow" in line 1180 SoilBiogeochemDecompCascadeMIMICSMod.F90

So I tried our preexisting FATES-MIMICS test with nofireemis appended
FAIL ERP_P256x2_Ld30.f45_f45_mg37.I2000Clm60FatesRs.derecho_intel.clm-mimicsFatesCold--clm-nofireemis
CH4 Conservation Error in CH4Mod during diffusion, nstep, c, errch4 (mol /m^2.timestep) 1 2 NaN
Same test in debug mode
ERP_D_P256x2_Ld30.f45_f45_mg37.I2000Clm60FatesRs.derecho_intel.clm-mimicsFatesCold--clm-nofireemis
traceback

floating overflow
...
floating invalid
...
process killed (SIGTERM)
197  BandDiagonalMod.F90
415  SoilTemperatureMod.F90
883  clm_driver.F90

@rgknox suggests switching from intel to gnu for better uninitialized variable catching.

• Made EXPECTED FAILURE: ERP_P72x2_Ld30.f45_f45_mg37.I2000Clm51FatesRs.cheyenne_intel.clm-mimicsFatesCold #2261 a sub-issue and updated it with the above FATES-MIMICS test-results.

[slevis-lmwg]

From meeting with @rgknox

• First thing to try... Uncomment sections of subroutine UpdateNLitterFluxes in utils/clmfates_interfaceMod.F90, remove the hardwiring to 0, and PASS ERS_D_Ld30.1x1_brazil.I2000Clm60FatesCrujraRs.derecho_intel.clm-FatesColdPRT2.
• Try non-PRT2 test in case it fails: PASS ERS_D_Ld30.1x1_brazil.I2000Clm60FatesCrujraRs.derecho_intel.clm-FatesCold
• In ELM's AllocationMod focus on RD (rel. demand) modifications, note that many of the mods are due to refactoring, and corresponding mods in CLM should go in SoilBiogeochemCompetitionMod: see 2025/8/8 post.
• ch4Mod appears already done.
• Need to address harvest (ask Charlie if need help): commented out hrv_deadstem_to_prod* lines in clmfates_interfaceMod for fates_parteh_mode == 2.
• When ready to turn on MIMICS, address EXPECTED FAILURE: ERP_P72x2_Ld30.f45_f45_mg37.I2000Clm51FatesRs.cheyenne_intel.clm-mimicsFatesCold #2261 and may need "if mimics" section near line 158 in src/biogeochem/CNNStateUpdate1Mod.F90 (lines 168-177 may need to be copied to the interface code above).
• This point repeats below: Think about how CLM can tell FATES to not run the phosphorus code. Ryan wrote:

It would take a complete re-write of the FATES side code to not cycle and allocate P along with N. So I think going forward, we just use synthetic boundary conditions for P and just make sure that it is not limiting