Update Prototype Test suites P7->P8

[DeniseWorthen]

Description

The existing P7 test suite for coupled and uncoupled needs to be updated to the initial P8 test suite by enabling RRTMGP.

Solution

Make the required changes in default_vars and the standalone P7 tests, rename the tests and make the required changes in rt.conf

Alternatives

Related to

[DeniseWorthen]

A strategy is needed for naming the SDFs that are used for the Prototype configuration. The relevant current SDFs in FV3/ccpp/suites are:

P7 SDFs:

coupled: FV3_GFS_v16_coupled_nsstNoahmpUGWPv1, FV3_GFS_v16_coupled_p7_rrtmgp 
standalone: FV3_GFS_v16_nsstNoahmpUGWPv1, FV3_GFS_v16_p7_rrtmgp

The initial P8 SDFs could be named:

coupled: FV3_GFS_v16_coupled_p8
standalone: FV3_GFS_v16_p8

These P8 SDFs would initially be simple copies of the SDFs currently being used (e.g. nsstNoahmpUGWPv1) except for the suite_name in the xml file. The P8 SDFs would then be updated as required by any new features added from physics.

The issue arises with the rrtmgp tests and SDFs.

1) there is a conflicting SDF file currently in FV3/ccpp/suites: suite_FV3_GFS_v16_p8.xml. This file is currently un-used by any test. It is identical to FV3_GFS_v16_p7_rrtmgp except for the suite_name at the top of the xml file.

2) it is not clear whether the p7_rrtmgp tests should also be updated to P8 (this would require _rrtmgp version SDFs to also be created). If p8_rrtmgp tests are desired, then the existing suite_FV3_GFS_v16_p8.xml should be renamed suite_FV3_GFS_v16_p8_rrtmgp.xml.

@junwang-noaa @dustinswales

[dustinswales]

These P8 SDFs would initially be simple copies of the SDFs currently being used (e.g. nsstNoahmpUGWPv1) except for the suite_name in the xml file. The P8 SDFs would then be updated as required by any new features added from physics.

The issue arises with the rrtmgp tests and SDFs.

1. there is a conflicting SDF file currently in FV3/ccpp/suites: suite_FV3_GFS_v16_p8.xml. This file is currently un-used by any test. It is identical to FV3_GFS_v16_p7_rrtmgp except for the suite_name at the top of the xml file.
2. it is not clear whether the p7_rrtmgp tests should also be updated to P8 (this would require _rrtmgp version SDFs to also be created). If p8_rrtmgp tests are desired, then the existing suite_FV3_GFS_v16_p8.xml should be renamed suite_FV3_GFS_v16_p8_rrtmgp.xml.

@DeniseWorthen @junwang-noaa The existing P8 RRTMGP can be removed. This is a relic from the last PR. Recall we were going to name the new GP p7 SDFs as p8, but backed off of this.

I don't have a strong opinion on what the SDFs should be named, but a few comments/questions.

If I recall correctly, there are mini-p8 prototypes with different "new" physics turned on (e.g. one with Thompson MP and another with RRTMGP)? So it may be that there are p8+GP p8+Thompson SDFs. Or maybe GP is p8 and there is p8 and p8+Thompson (or the opposite)?

[yangfanglin]

I'd suggest to use FV3_GFS_v16_coupled_p8a, FV3_GFS_v16_p8a, then updated to 8b, 8c etc along with the testing of each mimi prototype, and eventually to FV3_GFS_v16_coupled_p8 FV3_GFS_v16_p8 with the final P8 configuration. At that time, SDF files for mini prototypes can be removed. For schemes that are still in development and have not been included in mini prototypes, developers should take the responsibility of keeping them updated to follow the develop branch.

[DeniseWorthen]

Thanks @dustinswales . For your existing p7_rrtmgp tests, do you want to keep those up-to-date with whatever the current P8 configuration is? Or do you want to maintain them at P7?

[dustinswales]

@DeniseWorthen My preference would be to freeze the p7 RRTMGP tests as they are and build onto the baseline P8 (P8a) prototype.

RRTMGP is to be included as mimi P8 at some point, so as long as the coupled/un-coupled P8a's begin with the P7+RRTMGP SDFs there shouldn't be anything to maintain on the GP side. This is somewhat complicated by the need to add the coupling between two potential mimi P8's (RRTMGP and the refined Thompson MP).

[yangfanglin]

@dustinswales I'd prefer not to include RRTMGP in p8a before it can be claimed to be ready to be included in any of the upcoming mini prototypes. Thompson MP will likely be included in p8b. As you know more development is needed to run RRTMGP along with Thompson MP. There are also other scientific and computational questions related to RRTMGP that need to be addressed (as discussed offline in separate emails), and we are still waiting for the evaluation of RRTMGP that Qingfu is conducting. Could it be possible for you to set up a branch of your own for RRTMGP development and keep it updated ?

[DeniseWorthen]

@yangfanglin My concern with what you suggest is that we don't have the resources to maintain multiple possible P8 configurations in the RTs. That would require carrying a p8X(X=a,b,c etc) test suite for both coupled and standalone in the RTs.

From my perspective, while multiple mini-Ps may be required or performed, the RTs need to update only once a feature is known to be included in the P8 feature set.

[yangfanglin]

@DeniseWorthen Understood. I'd suggest still keeping the SDF files used for all mini prototypes for a while, and only run RT for the latest mini prototype. If this still does not work, it is fine to maintain only one SDF file and keep updating it.

[dustinswales]

@yangfanglin Sorry, I'm not included in any of the emails regarding the mini prototype's for P8. Last I heard GP was to be included.

I'm a bit confused on why I would keep a parallel GP development branch up-to-date with the P8 development? My preference would be to wait until you are ready for GP, then I can update the code base.

[yangfanglin]

@dustinswales let's take the discussion of GP issues offline. For the code managers, my recommendation is to move forward without carrying RRTMGP. The physics team will announce when it is ready to be used for P8.

[DeniseWorthen]

Sounds good. I will

1) delete the existing, un-used suite_FV3_GFS_v16_p8_rrtmgp.xml 2) create new p8 standalone and coupled P8 suites which are copies of nsstNoahmpUGWPv1. The initial RT commit for P8 will be to turn on CICE albedos, which requires no change to SDF. 3) not update the p7_rrtmgp tests. They will continue to use the existing P7 configuration.

[junwang-noaa]

@yangfanglin Do you have a physics namelist with P8 physics schemes available?

[yangfanglin]

@junwang-noaa please see https://docs.google.com/spreadsheets/d/1NmfS-2q6xrQNxG8uatfNNqu0Z1WjLE0xLT59BRCrXSw/edit#gid=0 for the physics updates planned for P8. Column D provides the readiness status of each update.

[lisa-bengtsson]

@yangfanglin just to follow up on our offline discussion in this thread; there seem to be a challenge in that we have to make a scientific decision wether RRTMGP should be included in P8, but it cannot be included if it is not a feature that is turned on in the P8 control test. Since we seem to have reached consensus on inclusion of Thompson MP, perhaps a way forward is to first include Thompson MP in the "P8 ORT test cases". Then on top of this configuration create a PR which adds the RRTMGP feature to the "P8 ORT test cases", including potential source code updates that goes with the incoming Sundqvist cloud fraction scheme and/or changes in Xu-Randall cloud fraction scheme. What do you think of doing it in this order?

[yangfanglin]

@lisa-bengtsson I believe the code managers (Denise etc) plan to do exactly what you described here. After the P8B configuration is announced they will add Thompson MP to P8 ORT test cases. It is then up to the developers to add other features on top of the P8B version.

[yangfanglin]

@RuiyuSun Ruiyu, please work with @DeniseWorthen and @junwang-noaa to add Thompson MP to the P8 ORT tests. I believe all "Plan-A" updates related to Thompson MP we discussed have been committed to the develop branch.

[lisa-bengtsson]

@RuiyuSun @yangfanglin in the "Plan-A" configuration, were there updates to the Xu-Randall cloud fraction code?

[yangfanglin]

@lisa-bengtsson yes.

[lisa-bengtsson]

@yangfanglin is it only this alpha parameter: https://github.com/NCAR/ccpp-physics/blob/a5dcdceda45f95d1e74880f67a0929e9d2464068/physics/GFS_rrtmgp_thompsonmp_pre.F90#L225

[yangfanglin]

@lisa-bengtsson I think the answer is yes based on Ruiyu's presentation he gave before the new year. @RuiyuSun Ruiyu please confirm, and share the ppt with Lisa

[lisa-bengtsson]

@yangfanglin I spoke with @dustinswales just now and we agreed that once the tests for P8B are generated (including Thompson MP), Dustin will create a PR updating the ORT tests to turn on the RRTMGP feature. In this PR he will also update the alpha parameter in the Xu-Randall scheme on the RRTMGP side. After this we can discuss how to approach the Sundqvist cloud fraction to be seen by RRTMGP if this solution will be part of any prototype. @junwang-noaa @DeniseWorthen can you keep us informed when the timing is right for such a PR to update the tests to include RRTMGP?

[RuiyuSun]

Lisa,

It is shown 'changes in cloud cover' section in slide #4 of this file. https://docs.google.com/presentation/d/1SXsYIyVTHws4YaqzGpUFvfbEb7Uh01Ju/edit#slide=id.p4

Ruiyu

On Wed, Jan 5, 2022 at 4:24 PM Fanglin Yang @.***> wrote:

@lisa-bengtsson https://github.com/lisa-bengtsson I think the answer is yes based on Ruiyu's presentation he gave before the new year. @RuiyuSun https://github.com/RuiyuSun Ruiyu please confirm, and share the ppt with Lisa

— Reply to this email directly, view it on GitHub https://github.com/ufs-community/ufs-weather-model/issues/940#issuecomment-1006088566, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKFOCLDGYYQIJCG3VCFEGJLUUSZHNANCNFSM5JHHLZVQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you were mentioned.Message ID: @.***>

--

Ruiyu Sun, PhD

IMSG at NOAA/NWS/NCEP/EMC

5830 University Research Ct., Rm. 2097

College Park, MD 20740

@. @.>

301-683-3787

[lisa-bengtsson]

Thank you @RuiyuSun, when Thompson MP is turned on @dustinswales has to make the same updates in the Xu-Randall code on the RRTMGP side (to reflect the code change when RRTMGP is turned on), so perhaps you could share with us the PR that has these changes? Thanks.

[RuiyuSun]

Lisa,

This is the PR https://github.com/NCAR/ccpp-physics/pull/789.

Ruiyu

On Wed, Jan 5, 2022 at 5:10 PM lisa-bengtsson @.***> wrote:

Thank you @RuiyuSun https://github.com/RuiyuSun, when Thompson MP is turned on @dustinswales https://github.com/dustinswales has to make the same updates in the Xu-Randall code on the RRTMGP side (to reflect the code change when RRTMGP is turned on), so perhaps you could share with us the PR that has these changes? Thanks.

— Reply to this email directly, view it on GitHub https://github.com/ufs-community/ufs-weather-model/issues/940#issuecomment-1006117830, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKFOCLCULE5RBBUJBHUMW23UUS6WNANCNFSM5JHHLZVQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you were mentioned.Message ID: @.***>

--

Ruiyu Sun, PhD

IMSG at NOAA/NWS/NCEP/EMC

5830 University Research Ct., Rm. 2097

College Park, MD 20740

@. @.>

301-683-3787

[dustinswales]

@lisa-bengtsson Looks like there are only a few minor changes to make on the GP side. Thanks @RuiyuSun