mix of fixes for catchCN CLM 5.1

[biljanaorescanin]

This PR is an attempt to fix several issues on the CatchCN PR branch associated with #1129.

Related PRs:

• Fix for endrun so code exits cleanly.
• Base PR: Updates to add new Catchment-CN version: Catchment-CN5.1 #1129

Fixes implemented in this PR:

1. Fixed indices:

   • CanopyStateType.F90: read vegwp from cnpft(...,76..79) (not 78+).
   • CNVegCarbonFluxType.F90: read litterfall accumulators from cnpft(...,80/81) (not 82/83).
   • PhotosynthesisMod.F90: do not read 76/77 for alpha (those are vegwp now)

2. Fixed calculation of multi-day averages in GEOS_CatchCNCLM51GridComp.F90:

• T2MMIN5D
• SNDZM5D

3. Fixed inconsistent application of CN_zone_weight when translating between CNCOL restart array and CTSM structures:

   • CNProductsMod.F90: for prod100/prod10 (C: slots 7/8; N: slots 21/22).
   • CNVegCarbonStateType.F90:seedc_grc.
   • CNVegNitrogenStateType.F90: seedn_grc.

4. In CNCLM_Driver.F90, subroutine CN_exit(), fixed assignment: "CNCOL(nc,nz,[8,22]) = prod10_grc".

5. Surface albedo: Added missing if condition to fix mapping of CNPFT(nc,nv,nz,[73,74]) to [LAI,SAI].

6. Consistent use of FVEG_MIN=1.e-4 parameter for the minimum vegetation area fraction that is needed for a patch to become active.

Open issues:

• PhotosynthesisMod.F90: set alphapsn* = 0._r8 at init -- not sure if this is correct, or if these variables should be in the restart file.
• In CatchmentCNRst.F90, verify vegetation class variables for consistency (incl. iclass, map_pft). List of CTSM veg classes is in GEOS_CatchCNCLM51GridComp.F90:

  GEOSgcm_GridComp/GEOSagcm_GridComp/GEOSphysics_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOScatchCN_GridComp/GEOScatchCNCLM51_GridComp/GEOS_CatchCNCLM51GridComp.F90

  Line 136 in 3717bd2

  ! map catchment type into PFT

• Identify which CNCOL and CNPFT restart variables can be re-set to non-negative and implement consistently.
• Review treatment of CTSM no-data-values (spval=1.e26) and NaNs; translate between CTSM no-data-value (spval) and MAPL_UNDEF when filling/reading out CNCOL and CNPFT restart arrays.
• Determine if restart variables should be initialized to spval or NaN immediately after allocation (see also 348beeb).
• Determine if totn_col is needed in restart; should be calculated as sum of component pools (CNVegNitrogenStateType.F90). Keeping totn_col in the restart should not be an error as such.
• Understand why restart variables have crazy or impossible values (e.g., ANNAVG_T2M_COL should be a temperature in units of K). See excel sheet of all restart variables, including names, units (where available) and min/mean/max values:
  Restart file for Y2000/M01/D01 from Eunjee's "cycle3" global spinup simulation
  GEOSldas_CN51_280_global_c3.catchcnclm51_internal_rst_stats.xlsx

History of PR:
Disabled after after bc6d321 (because resetting restart variables to 0. does not make them scientifically meaningful):

• Fix restart N fields
  • Targeted restart clamp for CNCOL 24/36/37 is the real fix we need for the ERROR seen on the associated branch.
• Sanitize restarts to prevent corrupted values from detonating totn_col and NBAL
  • At ingest (SoilBiogeochemNitrogenStateType.Init): clamp mineral N (24/36/37) and the soil pool components that build totals (17, 18–20, 25–28).

Code testing:

• After bc6d321:
  • Code runs with bad rst values from - spinup C3 with RESTART:1
  • Code runs from remapped spinup C3 to normal ranges with option RESTART:2

On 12 September 2025, @gmao-rreichle verified the consistency of CNCOL and CNPFT variable indices between filling and reading out data from/to CTSM structures.

@gmao-rreichle @weiyuan-jiang

[github-actions]

This PR is being prevented from merging because you have not added one of our required labels: 0 diff, 0 diff trivial, Non 0-diff, 0 diff structural, 0-diff trivial, Not 0-diff, 0-diff, automatic, 0-diff uncoupled, github_actions. Please add one so that the PR can be merged.