History bugfixes

[peverwhee]

Summary

1. Add optional sampled_on_subcycle argument to addfld. If that's true for a field, we don't override the averaging flag to "I" if nhtfrq = 1 (every timestep). This reenables subcycling averages.
2. Check for incompatible duplicates
   • compatible duplicates (flags match) are ok - just take the first one (to keep existing CAM behavior)
   • incompatible duplicates (flags differ) - endrun
3. Remove time_bounds field from instantaneous history files
4. Update default filename for monthly instantaneous files to include full timestamp
5. Update git-fleximod to 0.9.2

Mods

src/control/cam_history.F90

• Remove nhtfrq=1 override
  • Add sampled_on_subcycle logic to keep override for subcycle accumulated fields
• modify list_index to optionally check for duplicates/compatibility
• endrun with message including all duplicates found (if necessary)
  • Sample error message: ERROR: FLDLST: Found duplicate field(s) with different averaging flags. Place in separate fincl lists: "Q:A", "Q:I" (fincl3). "T:A", "T:I" (fincl3). "U:A", "U:I" (fincl4).
• Remove time_bounds from instantaneous history files
• Override instantaneous file name for monthly files to include full timestep

src/control/cam_history_support.F90

• Add sampled_on_subcycle to field_info object

src/physics/*

• update subcycle addfld calls to include sampled_on_subcycle=.true.

bld/namelist_files/use_cases/sd_waccm_sulfur.xml

• remove bad XML comment line

.lib/git-fleximod/*

• update to 0.9.2

closes #1149
closes #1150
closes #1166
closes #1167

[brian-eaton]

I'll just point out that the reason for setting the averaging flag to I when nhtfrq=1 is to get the correct timestamps. If the flag is A then the timestamps are set to midpoint values which is not correct for a single timestep.

[peverwhee]

@adamrher could you confirm that the addfld calls I've amended represent the full set of fields that are sampled on the subcycle?

[adamrher]

Hi @peverwhee, I found some missing addfld calls.

micro_pumas_cam.F90, marcop_driver.F90, microp_aero.F90 all look good. There are a 5 addflds you missed, in stats_init_clubb in clubb_intr.F90. I couldn't figure out how to add a review comment that deep into clubb_intr, but they start at line 5116.

microp_aero.F90 has some dependencies that have addfld calls. These are: ndrop.F90, ndrop_bam.F90, nucleate_ice_cam.F90, hetfrz_classnuc_cam.F90. All addflds in those modules need the sample_on_subcycle flag.

Lastly, the drivers in subcol_SILHS.F90 are called inside the macmic loop if use_subcol_microp=.true. Support for subcolumns is a grey area but it shouldn't be too much work (?) to add sample_on_subcycle to all the addflds in subcol_SILHS.F90.

[peverwhee]

@adamrher thanks for taking the time to check these! I've added/removed the sampled_on_subcycle arguments as you described.

• stats_init_clubb in clubb_intr.F90 (5 fields)
• ndrop.F90 (all addflds)
• ndrop_bam.F90 (all addflds)
• nucleate_ice_cam.F90 (all addflds)
• hetfrz_classnuc_cam.F90 (all addflds)
• subcol_SILHS.F90 (all addflds)
• REMOVE FROM check_energy.F90 (all addflds)