Adding MYNN-SFC submodule to MPAS

[joeolson42]

This PR removes the current traditionally implemented MYNN surface layer scheme and replaces it with the MYNN-SFC submodule, which has alrady been implemented into WRF (develop branch) for the upcoming v4.8 release. The submodule version uses the same config_sfclayer_scheme = "sf_mynnsfclay" option to be activated and uses the same set of variables except for a few (qgh, qcg, regime, pi3d, and qc3d) that were no longer used within the regular hrrrv5 suite. This submodule contains the scheme from the CCPP, which was used as the initial form, but it has decomposed into 3 different modules for land, water, and ice. There was a bug found in the SPP implementation for zq over water (now fixed). SPP was also added over snow and to another bulk flux option over land. New MYNN-specific options were added (sf_mynnsfc_sfcflux_land and sf_mynnsfc_sfcflux_water), which will be carried to all other model frameworks; the old options (iz0tlnd and isftcflx - hard-wired in MPAS) carried different meaning for different surface layer schemes, causing confusion. The new options are:

config_mynn_sfcflux_land = 0:constant Czil (default), 1:variable Czil, 2:Yang

config_mynn_sfcflux_water = 0:COARE3.0 (default), 1:COARE3.5, 2:Davis/COARE3.5, 3:Davis/Garratt, 4:Taylor-Yelland

Test results show very little differences relative to the non-submodular version.

Priority Reviewers

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

[clark-evans]

We haven't specifically discussed MYNN surface-layer with Bill and Michael, but they've indicated they'd like to keep their version + our version of other parameterizations for at least one major release after our version becomes available in their codebase. Thus, I'm concerned that they'll want you to re-do this to keep both or they might not take it in. In the latter case, we'd be left with more painful merges with the upstream code until the issue is rectified.

Thus, while I'm all in favor of this PR's approach in a vacuum, I'm concerned about upstream implications.

[joeolson42]

I didn't remove their version, just my old version. Laura's version should still work, although I haven't tested it.

[clark-evans]

I didn't remove their version, just my old version. Laura's version should still work, although I haven't tested it.

Ah, that's right - my mistake. I saw files removed in the physics_wrf subdirectory and forgot that your initial MYNN surface-layer implementation went in there.

The convection_permitting CI tests all passed, so Laura's version still works. The ufs-community versions of those tests return no differences, whereas the MPAS-Dev versions of those tests return small differences relative to their baselines. For the latter, it's a combination of what is documented in #166 and the RRTMG changes with #201, so nothing to do with MYNN surface layer.