Restructure GFL submodule for MPAS, WRF, and CCPP

[haiqinli]

The GFL submodule is restructured for MPAS, WRF and CCPP. The Fortran subroutines were reformatted from *F to *F90. It is tested with WRF and global MPAS at 15km horizontal resolution and L60 vertical layers.

Priority Reviewers

• Please list the developers/collaborators you'd like to prioritize for review
  • @clark-evans
  • @AndersJensen-NOAA
  • @joeolson42

[clark-evans]

In MPAS/module_cu_gfl_mpas.F90, old line 167 and new line 172, the PBL check for ishallow = 0 is changed from bl_mynn to bl_mynnedmf. Should it instead be bl_mynn .or. bl_mynnedmf? I don't know Laura's implementation of MYNN-EDMF well enough to know if the GFL shallow cu scheme should also be inactive in that scenario.

@joeolson42

(See code snippet here - though you have to expand the large diff first.)

[joeolson42]

It looks like it's now set to work with my mynnedmf and will no longer work with Laura's, which passes my code evaluation.

The only other thing I see, that I didn't flag before, is the physical constants seem to be defined in the scheme instead of being brought in from a common source file to maintain consistency with all other schemes in a modeling framework. The impact probably isn't noticeable for short/medium range forecasts, but may be problematic for longer range. You might consider bringing them in from a model-specific reference file.

[haiqinli]

There is no PBL checking in Laura's implementation. We added the checking of "bl_mynn" before the MYNN EDMF submodule, and updated it to "by_mynnedmf" to be consistent with our GSL physics suite.

Yes, I added a placeholder of model-specific reference file as module_cu_gfl_common.F90, and plan to add physical constants next time.

[clark-evans]

There is no PBL checking in Laura's implementation. We added the checking of "bl_mynn" before the MYNN EDMF submodule, and updated it to "by_mynnedmf" to be consistent with our GSL physics suite.

Yes, I added a placeholder of model-specific reference file as module_cu_gfl_common.F90, and plan to add physical constants next time.

Understood re: Laura's implementation of GF, but to clarify, I was wondering if our implementation of GFL should check for both Laura's MYNN-EDMF implementation (bl_mynn) and our MYNN-EDMF implementation (bl_mynnedmf) and not just the latter. Joe says no, so I'm OK with moving forward as is, but I just wanted to clarify for the potential unfortunate instances where someone could use our GFL with Laura's MYNN-EDMF.