Passing additional surface variables into the atmospheric model #28 #29
Conversation
zhihong-tan
requested review from
laurenchilutti and
thomas-robinson
as code owners
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.
For compatibility with AM4, a ifdef-statement with a conditional compilation flag "do_am5phys" is added in the call to physics_driver_down to prevent the additional surface variables from being passed into the old AM4 physics package. This flag should be defined if AM5 is compiled with NCEP TKE-based EDMF.
zhihong-tan
force-pushed
the
znt_f3b1r1
branch
from
e3256b8
to
9072c35
Compare
|
Does the solo driver also need to be updated? |
|
@menzel-gfdl Would you have the time to review this today? |
|
Hi All,
Sorry for the complication. I would need to do another commit for
atmos_driver to remove the conditional compilation and submit a PR for the
atmos_phys repository for compatibility. I will finish these changes by
tomorrow.
Thanks,
Zhihong
…On Thu, Nov 9, 2023 at 12:28 PM Lauren Chilutti ***@***.***> wrote:
***@***.**** approved this pull request.
—
Reply to this email directly, view it on GitHub
<#29 (review)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AC3AZH2GJMXRRQRJRCUMHRLYDUHFDAVCNFSM6AAAAAA6CYEDAWVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMYTOMRTGEZTQOJXGY>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
The added variables are now passed as optional arguments into physics_driver_down. This requires consistent changes in the atmos_phys package.
Two if-associated statements are added before the checksums of bnd_type%shflx and bnd_type%lhflx to pass the null model build.
laurenchilutti
requested review from
menzel-gfdl,
laurenchilutti and
uramirez8707
|
@zhihong-tan I see you made a few commits - are these changes ready to review? And has the atmos_phys PR been created? |
|
@laurenchilutti I have completed and committed my code changes, but I would need to go through again a couple of verification tests before submitting the merge request (I will do this today). And I have created my own forks of atmos_phys on both Github and Gitlab. |
|
@laurenchilutti I have created the PRs and merge requests on Gitlab on atmos_solo_land and atmos_phys repositories. Especially, the update on atmos_phys is needed along with the updates on FMScoupler and atmos_drivers for the standard AM4 to run successfully. Could you please go over these PR/MRs for me? Also, could you please suggest who I should assign the MRs to (on Gitlab)?Thank you! |
|
Additionally, I have also created my own fork of the atmos_phys repository on Github and made the same changes as those on Gitlab on my branch. However, because Github's atmos_phys has not been updated since 2021.02, I have not been successful in creating a working version of AM4 with the up-to-date repositories of other components (e.g., FMS and FMScoupler) on Github. Should I still submit the PR on Github? |
I took care of the atmos_phys PR and tagged it as 2023.04-beta1 (It will be tagged as 2023.04 once we run regression testing on it)
The github version of atmos_phys is only updated as needed for publications, so it does not need to be updated in this case. Thanks |
|
@thomas-robinson and @menzel-gfdl when you get a moment could you please review this PR this is associated with changes in atmos_phys which you can look at Gitlab for. |
| write(outunit,100) 'lnd_ice_atm_bnd_type%wind ',mpp_chksum(bnd_type%wind ) | ||
| write(outunit,100) 'lnd_ice_atm_bnd_type%thv_atm ',mpp_chksum(bnd_type%thv_atm ) | ||
| write(outunit,100) 'lnd_ice_atm_bnd_type%thv_surf ',mpp_chksum(bnd_type%thv_surf ) |
There was a problem hiding this comment.
Just double checking these entities will be defined for all non-AM5 simulations as well e.g. AM3/AM4.
The full name for the NCEP TKE-based EDMF replaces the informal short name (HanEDMF) in the comments only. No other code changes.
|
@thomas-robinson @menzel-gfdl would you be able to review this today? |
The term 'surface' in the comments for the added fields is clarified as the lowest atmos level as Sergey suggested.
|
@thomas-robinson can you give this a final review and merge if approved? |
Description
This PR is linked to issue #28 and it depends on issue #92 (#93) of the FMScoupler repository.
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 atmos_driver (coupled/atmos_model.F90) to pass these variables into the physics driver. (A "ifdef" statement is added to maintain backward compatibility with AM4's physics driver).
How Has This Been Tested?
The AM4 8-day 'basic' run with my updated FMScoupler and atmos_drivers can bitwise reproduce the results with those run with the default code. This test is done with the AM4-2023.02 xml obtained with the command below:
git clone -b AM4_2023.02 http://gitlab.gfdl.noaa.gov/mdt/xml.git xml_2023.02
And the model codes are compiled and run on the ncrc5.intel22-classic-prod-openmp platform.
Note that both changes in FMScoupler and atmos_drivers need to be incorporated at the same time for the model to compile and run correctly.
Checklist: