RRFS_v1beta not Recognizing Physics Suite Changes

[rgsmith27]

I'm attempting to replicate an experiment in Gallus Jr. & Harrold where they ran the SRW using the RRFS_v1beta physics package to simulate the August 10 2020 derecho. They performed two simulations, one with convective parameterization off and one with convective parameterization on, and obtained wildly different results.

I've attempted this experiment myself but I am getting identical results with and without convective parameterization enabled. The results are incredibly similar to the results obtained in the study when convective parameterization is turned off, which makes me think that it is being disabled somewhere. The run_fcst logs seem to indicate that the change in physics schemes has been recognized, so I'm unsure where the issue is occuring. Below I've linked logs, as well as images from the two runs to illustrate the issue.

run_fcst logs without gf
run_fcst logs with gf

[jdkublnick]

Hello @rgsmith27,

Your question has been forwarded to an internal subject matter expert (SME), and you will get a response soon.

Thank you,

Josh

[MichaelLueken]

Hi @rgsmith27,

The issue is that the FV3_RRFS_v1beta physics suite doesn't use the Grell-Freitas convection scheme. This documentation shows the parameterizations used by the FV3_RRFS_v1beta physics suite.

In August 2020, the experimental RRFS would have been using the FV3_RAP physics suite - https://github.com/NOAA-GSL/regional_workflow/blob/feature/RRFS_dev1/ush/config.sh.RRFS_NA_13km. Please look for the CCPP_PHYS_SUITE entry in the above file. The documentation for this physics suite also notes that it is set up to use Grell-Freitas convection scheme. Please try rerunning your experiment using FV3_RAP and see if the results change between turning on or off the GF convection scheme.

[gspetro-NOAA]

Yes, likely all of the components and physics in the study have been evolving since the study was conducted. It might be worth trying to contact the authors for more precise details (repo used, hash) if you are trying to reproduce their work. Also, physics schemes can be manually added/removed from physics suites, although it's not encouraged, and there may be limitations to what is possible, as @MichaelLueken mentioned. See FAQ for more.