Updates to TEMPO in ccpp-physics need updates at ufsatm level

[benwgreen]

I have been trying to run TEMPO regression tests after updating ccpp-physics to ensure all CCPP hooks for the existing TEMPO-based suite definition file (SDF) have been implemented correctly, see ccpp-physics Issue 330.

@AndersJensen-NOAA suggested a series of changes within Issue 330, which I have synthesized into a ccpp-physics branch in my repository.

The problem is, when the tracers get updated to add two extra for hail (lthailaware = .true.) -- cf. GFS_PBL_generic_common.F90 (scroll down to Line 47, where the imp_physics_tempo flag gets added), if both ltaerosol and lthailaware are true, then the number of tracers increases to 14.

Unfortunately, the current logic on Line 1206 of ccpp/data/CCPP_typedefs.F90 is unable to handle the case of TEMPO with lthailaware (in fact, Model%lthailaware does not exist in this file).

Thus, if I make the modifications to CCPP as described above and then run the TEMPO regression tests with the ufsatm code as-is, I will get runs that do not crash for:
1.) control_p8_ugwpv1_tempo_intel
2.) control_p8_ugwpv1_tempo_aerosol_intel

BUT, I will get a crash for:
3.) control_p8_ugwpv1_tempo_aerosol_hail_intel

Because in that case, the value of Interstitial%nvdiff should be equal to 14, but it is not set correctly on Line 1206 of ccpp/data/CCPP_typedefs.F90.

My proposed fix is to separate TEMPO out into its own elseif block such that Line 1206 is only for Thompson, and then (at Line 1215), the TEMPO elseif block can set up the value of Interstitial%nvdiff correctly.

Specifically, my proposed fix in ccpp/data/CCPP_typedefs.F90 would be:
elseif ( Model%imp_physics == Model%imp_physics_tempo) then
Interstitial%nvdiff = 9
if (Model%ltaerosol) then
Interstitial%nvdiff = Interstitial%nvdiff + 3
endif
if (Model%lthailaware) then
Interstitial%nvdiff = Interstitial%nvdiff + 2
endif
if (Model%satmedmf) Interstitial%nvdiff = Interstitial%nvdiff + 1

This would have to be done in tandem with Issue 330 in ccpp-physics, and a new baseline for the TEMPO tests (at least, the control_p8_ugwpv1_tempo_aerosol_hail_intel test) would be made.

I would appreciate feedback from @AndersJensen-NOAA and @dustinswales. My fear is that there are other TEMPO-specific issues like this lurking around in ufsatm.

[AndersJensen-NOAA]

Thanks @benwgreen for finding another microphysics-specific area of code:

As you suggested, this commit should fix things:
4e56b4f

[benwgreen]

Hey @AndersJensen-NOAA would you please double-check the following files to make sure that arrays are being initialized correctly for TEMPO species? I pulled this list based on looking for Thompson and TEMPO flags (imp_physics_thompson and imp_physics_tempo):
1.) ccpp/data/CCPP_typedefs.F90:

Thanks @benwgreen for finding another microphysics-specific area of code:

As you suggested, this commit should fix things: 4e56b4f

Thanks for catching that second block (around Line 1300) that I missed before!

[benwgreen]

@AndersJensen-NOAA In poking around the code, there is seemingly inconsistent treatment of ltaerosol and mraerosol when it comes to TEMPO.
In some of your changes (e.g., Line 47 of GFS_PBL_generic_common.F90 at the CCPP level), you only consider ltaerosol (no consideration of mraerosol, but in other places (e.g., Line 1212 of ccpp/data/CCPP_typedefs.F90), the logic allows for the possibility of mraerosol to be used with TEMPO. Furthermore, at Lines 4976, 4978, and 4979 in data/ccpp/GFS_typedefs.F90, the code clearly suggests that either ltaerosol or mraerosol is acceptable in TEMPO.

If mraerosol is allowable in TEMPO, then we need to make some changes to the CCPP level (specifically, your commits in Issue 330).
If mraerosol is not allowable in TEMPO, then we should make changes at the ufsatm level to spit out an error message when imp_physics_tempo and mraerosol are both true.

[AndersJensen-NOAA]

@AndersJensen-NOAA In poking around the code, there is seemingly inconsistent treatment of ltaerosol and mraerosol when it comes to TEMPO. In some of your changes (e.g., Line 47 of GFS_PBL_generic_common.F90 at the CCPP level), you only consider ltaerosol (no consideration of mraerosol, but in other places (e.g., Line 1212 of ccpp/data/CCPP_typedefs.F90), the logic allows for the possibility of mraerosol to be used with TEMPO. Furthermore, at Lines 4976, 4978, and 4979 in data/ccpp/GFS_typedefs.F90, the code clearly suggests that either ltaerosol or mraerosol is acceptable in TEMPO.

If mraerosol is allowable in TEMPO, then we need to make some changes to the CCPP level (specifically, your commits in Issue 330). If mraerosol is not allowable in TEMPO, then we should make changes at the ufsatm level to spit out an error message when imp_physics_tempo and mraerosol are both true.

@benwgreen When we originally connected tempo we leveraged the Thompson infrastructure, and Thompson works with both ltaerosol and mraerosol. For tempo v3, I'm trying to clean things up and start simple, and so mraerosol will not work with Tempo. I'll fix these inconsistencies. Thanks for pointing them out.

[AndersJensen-NOAA]

This commit removes tempo -mraerosol options
9bfa57b