Passing additional surface variables into the atmospheric model #92 #93
Conversation
The shflx, lhflx, wind, thv_atm, thv_surf variables are passed from the surface coupler to the atmospheric model as required by the NCEP TKE-EDMF scheme. And some useful diagnostics are added to output the surface fluxes before the implicit surface coupling.
Changing the names of fms subroutines (register and send diagnostic variables, and get data from xgrid).
The added diagnostics for surface fluxes before the implcit surface coupling have been removed to minimize the changes from the original codes.
|
@zhihong-tan have you tested this with the doubly periodic, aquaplanet, or any other model that uses the simple coupler? Those models use the surface flux file but not the full coupler. It's possible that this code doesn't affect those models, or it may require changes to their respective coupler implementations. |
|
@zhihong-tan Please remove any comments with your name and the date. Git will track that information, so we don't need to clutter the code with comments. |
|
There are also some compile errors in the CI that need to be addressed. |
|
Looks like the compile errors are coming from the atmosphere null model we're using for the CI. To fix it would require a change in https://github.com/NOAA-GFDL/atmos_null. The fields added to the |
|
@thomas-robinson Thanks for your suggestions! I have not run the doubly-periodic or aquaplanet models with simple coupler, and I think they may not work with my modified surface_flux.F90. I will change the two added output variables from surface_flux subroutine to 'optional' and test whether this is sufficient for compatibility with the simple coupler. |
|
@zhihong-tan Do you have an update on this? |
The same code modifications as the full coupler are now incorporated into the simple coupler, which is now compatible with the updated surface_flux subroutine.
@thomas-robinson I have modified the simple coupler and removed my name tags in the comments as suggested. The updates have been pushed to my branches of the FMScoupler and atmos_drivers repositories. I have verified that the new codes work with both the default AM4 and the simple version under the doubly_periodic repository. The 8-day regression run results are exactly reproduced for the "c96L33_am4p0" experiment in the default AM4 and for the "aqua_c96L33_am4p0_control" experiment in the doubly_periodic version. |
|
Should I also clone the atmos_null repository, modify the codes, and submit a merge request? Or could you do that for me? (I don't know how to do the CI when multiple repositories are changed). |
|
@zhihong-tan the fastest approach would be for you to update atmos_null and any other repositories that would require updating. If you want MSD to do it, the time frame would be a few months. We should also set up a meeting to go over these changes and discuss why they are necessary at the coupler level. I'll reach out via email for that. |
Thanks for your suggestion! I have forked and updated the atmos_null to be compatible with the updated coupler, for which I have submitted a merge request. I will explain my proposed updates for these three repositories (FMScoupler, atmos_drivers, and atmos_null) in our meeting tomorrow. |
|
I'm not sure if you will also need to update atmos_solo_land on the gfdl gitlab |
|
@zhihong-tan @thomas-robinson - I'd really like to see all of the related PRs in different repos, including the internal gitlab ones, be linked in to the PR description here so that we can view the changes as a whole. As I've looked at the various PRs individually, I believe I've seen changes to interfaces in atmos-drivers that will require changes in the AMx physics driver interfaces. My concern right now is exposing these new variables will be good for AM5 development, but not for AM4 and AM3 models without macro-controlled conditional compilation. |
This PR is linked to issue #92.
A new boundary layer scheme (NCEP TKE-based eddy-diffusivity mass-flux scheme) is being implemented in the prototype-AM5 model, and it requires five additional surface variables (shflx, lhflx, wind, thv_atm, and thv_surf) as inputs. Thus, I would like to add a few lines in the surface coupler (full/atm_land_ice_flux_exchange.F90 and shared/surface_flux.F90) to pass these variables into the atmospheric model. My branch contains the proposed changes, and the model results are unchanged with these changes.