MYNN-EDMF update

[joeolson42]

MYNN-EDMF update including MYNN-related modules

1. Updates to the MYNN-EDMF from RRFS_dev, which include updates made for RRFS "Base E" (pdf here)+ some addition mods below:
2. Including snow mixing. This required passing snow into the MYNN, modifying the negative mixing ratio checks, and adding a new tendency array for output.
3. Includes a patch for ensuring cloud fractions for all grid cells where qc > 1e-6 and qi > 1e-9. See pdf.
4. Misc code cleaning: removing some test code, removing from if-check from within loops, cleaning up indentations and interfaces, etc.
5. Updates to the sgscloud_radpre module for use with MYNN. This removes a bug for supplying grid mean mixing ratios and also adda a section for use with SAS convection scheme and imports the Ferrier-Aligo flag for future coupling tests.
6. Included many changes to control the specified model precision.

Companion PRs:
NOAA-EMC/fv3atm#630
ufs-community/ufs-weather-model#1636

[yangfanglin]

Joe, could you please add a link to the ppt about these updates you presented at the RRFS meeting ? Thx

[ericaligo-NOAA]

Joe, nice talk the other day. You said there was no noticeable differences with a winter case. I'm wondering if one would see a difference in the onset of snow at the surface when the snow is included in the mixing. Some years ago Brad Ferrier and I thought about mixing snow in the MYJ PBL. The NAM tended to have a delay in the onset of snow reaching the ground. Vertical cross sections showed it was dry in the lowest few hundred mbs, but RH was near 100% above that. We wondered if mixing snow would increase the RH sooner in the lowest levels.

[joeolson42]

Fanglin,

I have update the description to include links to some pdfs.

Eric,

I should clarify that I didn't see any differences in the overall cloud cover (this wasn't the ultimate fix for the "missing cloud fraction" problem) but there may be some other small impacts, like you mention. I'll definitely take a closer look. We will also be running more winter retros soon. Summer retros have been getting most of the CPU hrs.

-joe

[ericaligo-NOAA]

Thanks, Joe! It will be interesting to see if there is a difference in snow onset for the winter cases.

[dustinswales]

I don't have any major concerns, just the mismatch of the working precision in the scheme, which should get resolved. I know it's an invasive step, so I'm fine with it being part of a follow up PR.

[grantfirl]

Only minor comments from me. You definitely want to check if the #ifdef SINGLE_PREC is still needed in order to not break the recent single precision changes by NRL and Sam.

[grantfirl]

@joeolson42 I didn't catch this earlier, but please update all of your PR branches to the top of their respective develop branches: ufs/dev for ccpp/physics, develop for fv3atm, and develop for ufs-weather-model.