Fix mass non-conservation in VDIFF PBL mixing

[nicholasbalasus]

Name and Institution (Required)

Name: Nick Balasus
Institution: Harvard University

Describe the update

As pointed out in #2409 by @ohsangik98, the methane simulation does not conserve mass. This is the result of the mass conservation adjustment in vdiff_mod.F90 relying on a mass tendency (dqbot) with units of kg species/kg total air that is then converted to mass using kg dry air instead of kg total air. This is the main fix here.

Other minor changes:

• Do not convert State_Chm concentration values from kg/kg dry to mol/mol dry before PBL mixing (as this is just undone right after in VDIFFDR). According to inline comments from @lizziel, the call to AIRQNT should still work with these units.
• Delete code from the (from what I can tell) unused VDIFFAR.
• Convert pint to Pa from hPa for the top edge as is done with the rest of the array.
• Update the hard-coded molecular weights in global_ch4_mod.F90.

Expected changes

Mass balance can be achieved for the methane simulation. I tested this with a year long simulation for methane.

In general, concentrations will be higher in wet places. This change will impact all species using the VDIFF PBL mixing scheme (GEOS-Chem default), not just methane.

Reference(s)

Thanks to Jintai Lin and Chris Holmes (@cdholmes) for advice.

Related Github Issue

#2409

[yantosca]

Thanks @nicholasbalasus! I'll start testing this to bring it into 14.5.0.

[yantosca]

All GEOS-Chem Classic integration tests passed, except ModelE2.1:

==============================================================================
GEOS-Chem Classic: Execution Test Results

CodeDir   : edaa239 GEOS-Chem & HEMCO updates: Merge PR #2318, PR #284, PR #285
GEOS-Chem : 443a91a57 mass conservation fixes
HEMCO     : d0d5fca PR #285 post-merge fix: Update CHANGELOG.md
Cloud-J   : d20050f Update version to 7.7.3
HETP      : 2a99b24 Merge pull request #2 from geoschem/bugfix/initialize_local_variables

Using 24 OpenMP threads
Number of execution tests: 30

Submitted as SLURM job: 44461598
==============================================================================
...
gc_2x25_ModelE2.1_fullchem..........................Execute Simulation....FAIL
...
Summary of test results:
------------------------------------------------------------------------------
Execution tests passed: 29
Execution tests failed: 1
Execution tests not yet completed: 0

The ModelE2.1 failure is due to a missing file as described in #2427.

[yantosca]

All GCHP integration tests also passed:

==============================================================================
GCHP: Execution Test Results

CodeDir       : 86bcfa9 GEOS-Chem & HEMCO updates: Merge PR #2318, PR #284, PR #285
MAPL          : 231d53c Merge pull request #36 from geoschem/feature/improve_hflux_regridding
GMAO_Shared   : 4ddb3ec Merge pull request #2 from geoschem/feature/mapl-upgrade
ESMA_cmake    : ad5deba Added ecbuild as a submodule of ESMA_cmake
gFTL-shared   : 4b82492 Merge branch 'upstream_v1.5.0' into feature/v1.5.0
FMS           : 259759d Merge pull request #3 from geoschem/feature/update_gmao_libs
FVdycoreCubed : af42462 Merge PR #8 (Add PLEadv diagnostic for offline advection in GCHP)
geos-chem     : 443a91a57 mass conservation fixes
HEMCO         : d0d5fca PR #285 post-merge fix: Update CHANGELOG.md
yaFyaml       : 19afe50 Merge branch 'upstream_v1.0.4' into feature/v1.0.4
pFlogger      : 2c4b724 Merge branch 'upstream_v1.9.1' into feature/v1.9.1
Cloud-J       : d20050f Update version to 7.7.3
HETP          : 2a99b24 Merge pull request #2 from geoschem/bugfix/initialize_local_variables

Number of execution tests: 12

Submitted as SLURM job: 44455702
==============================================================================

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%  All execution tests passed!  %%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%