dev/gfdl main candidate 2021-03-26

[marshallward]

This PR is a minor update to the main branch, unrelated to the FMS2 update. This is a joint update from GFDL and EMC.

Review request:

• ✔️ COAPS / FSU: @abozec / @awallcraft
• ✔️ EMC: @jiandewang
• ✔️ ESMG: @kshedstrom
• ✔️ NASA-GMAO: @sanAkel
• ✔️ NCAR: @alperaltuntas / @gustavo-marques

The main purpose of this PR is to restore the history between main, dev/gfdl, and dev/emc. This was caused (by myself) when main was accidentally updated with content from dev/gfdl during the main -> dev/gfdl update.

Although we reversed the accidental update to main, the timing was such that the new content ended up in dev/emc before the reversal.

To resolve this, we locally merged dev/emc into dev/gfdl. The dev/emc -> dev/gfdl merge contained both our own content as well as new content from dev/emc.

Although the number of changes is less than usual, there is at least one answer-changing modification to the NUOPC coupler (flux sign bugfix), as well as another potentially detected by NCAR.

---

GFDL summary:

• Testing: Recurse target submodules, LDFLAGS hook #1355: Testing: Recurse target submodules, LDFLAGS hook (@marshallward)
• Change units of slope returned from calc_isoneutral_slopes() to "Z L-1" #1351: Change units of slope returned from calc_isoneutral_slopes() to "Z L-1" (@adcroft)
• Explicitly set (1x1) io_domain as a default #1350: Explicitly set (1x1) io_domain as a default (@Hallberg-NOAA)
• Merge from main #1349: Merge from main (@marshallward)
• Bugfix: empty restart_input_dir for *_solo.res #1348: Bugfix: empty restart_input_dir for *_solo.res (@herrwang0)
• Bugfix: FGNV streamfunction vertical bounds #1346: Bugfix: FGNV streamfunction vertical bounds (@marshallward)
• Typo correction #1345: Typo correction (@hmkhatri)
• Update CVMix submodule #1344: Update CVMix submodule (@adcroft)
• Depth-integrated momentum budget diagnostics #1343: Depth-integrated momentum budget diagnostics (@hmkhatri)
• Fix a missing line regarding vel_underflow #1340: Fix a missing line regarding vel_underflow (@herrwang0)
• dynamic ice shelf #1338: dynamic ice shelf (@OlgaSergienko)
• Fix the bug that Rayleigh_[uv] was not initialized in MOM_barotropic #1329: Fix the bug that Rayleigh_[uv] was not initialized in MOM_barotropic (@herrwang0)

Contributors:

• @marshallward: 3
• @herrwang0: 3
• @adcroft: 2
• @hmkhatri: 2
• @Hallberg-NOAA: 1
• @OlgaSergienko: 1

---

EMC summary:

• fixed missing halo in two halo updates for taux and tauy in mom_surface_forcing_nuopc.F90 in nuopc_cap
• allow MOM6 to use a mesh optionally in nuopc_cap
• fixed sign error on fprec passed to ocean in mom_surface_forcing_nuopc.F90 in nuopc_cap (from NCAR)
• allow ocean->land topo edits in MOM_shared_initialization.F90

[kshedstrom]

I approve this PR.

[sanAkel]

@marshallward Thanks for requesting to review.

Thanks to @mathomp4, we are on about a bi-weekly sync schedule with dev/gfdl. So comparing with what we use in GEOS-ESM the only relevant change is a parameter addition of: ALLOW_LANDMASK_CHANGES, which is okay!
(Please note that we use FMS_cap driver.)

[jiandewang]

approval

[abozec]

FSU/COAPS approves too

[alperaltuntas]

@jiandewang, @DeniseWorthen, The newly added code block in config_src/drivers/nuopc_cap/mom_cap.F90 impacts CESM's nuopc configuration. Specifically, this PR changes our geomtype from ESMF_GEOMTYPE_MESH to ESMF_GEOMTYPE_GRID. Did you intend to change geomptype for CESM configurations as well? If so, could you remind us the rationale? Also mentioning @mvertens.

[DeniseWorthen]

@alperaltuntas This was not intended to change CESM configuration. The only intent was to allow UFS to optionally use the mesh instead of the current grid. I realized only later that by setting the config variable false by default this would possibly impact CESM.

[alperaltuntas]

Thanks, @DeniseWorthen! We'll wait for you to address this and we will then test the PR again. (No rush on our side.)

[DeniseWorthen]

@alperaltuntas Does switching the default value for use_mommesh to true work for CESM? I've tested it in ufs and it will work on our end. We will eventually switch our default to using the mesh for the mom_cap.

diff --git a/config_src/drivers/nuopc_cap/mom_cap.F90 b/config_src/drivers/nuopc_cap/mom_cap.F90
index c96c98cdd..7dc95e13f 100644
--- a/config_src/drivers/nuopc_cap/mom_cap.F90
+++ b/config_src/drivers/nuopc_cap/mom_cap.F90
@@ -136,7 +136,7 @@ integer              :: logunit  !< stdout logging unit number
 logical              :: profile_memory = .true.
 logical              :: grid_attach_area = .false.
 logical              :: use_coldstart = .true.
-logical              :: use_mommesh = .false.
+logical              :: use_mommesh = .true.
 character(len=128)   :: scalar_field_name = ''
 integer              :: scalar_field_count = 0
 integer              :: scalar_field_idx_grid_nx = 0
@@ -349,7 +349,7 @@ subroutine InitializeP0(gcomp, importState, exportState, clock, rc)
   write(logmsg,*) use_coldstart
   call ESMF_LogWrite('MOM_cap:use_coldstart = '//trim(logmsg), ESMF_LOGMSG_INFO)

-  use_mommesh = .false.
+  use_mommesh = .true.
   call NUOPC_CompAttributeGet(gcomp, name="use_mommesh", value=value, &
        isPresent=isPresent, isSet=isSet, rc=rc)
   if (ChkErr(rc,__LINE__,u_FILE_u)) return

[alperaltuntas]

@DeniseWorthen, yes, setting use_mommesh to true by default works for CESM.

[DeniseWorthen]

Thanks for testing @alperaltuntas. I will have Jiande commit the change when he is back in the office on 4/1.

[marshallward]

Thank you @DeniseWorthen we'll also merge the work into this PR when it's ready.

[jiandewang]

@marshallward I just merged dev-gfdl-main-candidate-2021-03-26 branch to the original branch I provided to you (https://github.com/jiandewang/MOM6/tree/feature/for-GFDL-hot-fixing-20210322), and committed the changing of default value as True for use_mommesh in nuopc_cap/mom_cap.F90.
thanks for the fixing @DeniseWorthen @alperaltuntas

[marshallward]

Thanks @jiandewang this branch has now been updated.

Since this change is very minor and isolated to the NUOPC cap, I'll merge this after NCAR approval.

[alperaltuntas]

I approve.

[marshallward]

Thanks all, merging this now. Histories between EMC and GFDL (and hopefully everyone else) should now be in sync.