CCPP-compliant Grell-Freitas convection scheme

[climbfuji]

This PR is based mostly on the work by @haiqinli and supersedes PR #139. For further details, see below.

[climbfuji]

Associated PRs:

https://github.com/NCAR/NEMSfv3gfs/pull/60
https://github.com/NCAR/FV3/pull/87
#166
NCAR/ccpp-framework#149

[climbfuji]

@haiqinli Can you please provide a paragraph of information (for example like the description in #165) about this PR, which version of the code, what has been left out, what needs to be done etc?

[climbfuji]

These PRs were tested successfully on MacOSX with clang+gfortran to give bit-for-bit identical results for the fv3_control run both with and without CCPP compared to the baseline. Attached is also a screenshot of average precipitation rate for a 1h forecast on a C96 mesh, left fv3_control and right fv3_gf (identical to fv3_control, but with GF deep/shallow instead of SAS).

The code is up to date with the current reference branches, including the latest Thompson MP PRs.

Todo:

• 120h forecast fv3_control versus fv3_gf on Theia for C96 mesh
• 24h forecast on a C384 mesh and/or a 6h forecast on a C768 mesh
• usual standard regression tests on Theia/Intel

[climbfuji]

Standard regression test logs on Theia/Intel, all tests passed.

rt_ccpp_hybrid.log
rt_ccpp_ref.log
rt_ccpp_standalone.log
rt_ccpp_static.log
rt_full.log

[climbfuji]

Test of GF versus SAS: 120h integration on C96 mesh, plots of total precipitation [mm/day] for SAS, GF, and SAS-GF.

[climbfuji]

This PR is ready to go from my side. If changes need to be made, for example to the standard_name choices, we can either redo the CCPP tests or create a follow-up PR.

[haiqinli]

This PR and associated PRs listed below add the Grell-Freitas scale-aware and aerosol-aware convective scheme to CCPP and enable its use in NEMSfv3gfs.

This GF scheme is based on the version in FV3GFS version 1.

[haiqinli]

This looks good to me. It is designed to be independent of global models.

[climbfuji]

Most of the code was written by Haiqin, I only implemented corrections on top of the original PRs and brought the code up to date. My corrections were reviewed by Haiqin. Merging it now.