ccpp-physics
ccpp-physics copied to clipboard
NCAR
Move NoahMP to its own submodule
@DomHeinzeller @ligiabernardet @barlage @ncarchen This is a placeholder issue for setting up the NoahMP LSM as its own submodule within the ccpp-physics repository.
According to @barlage via the AMS 2021 meeting, @ncarchen and @barlage have already begun a GitHub repository to contain the unified (between WRF and the NWM) NoahMP LSM: https://github.com/NCAR/noahmp
This should be what ends up as the ccpp-physics submodule. Depending on what the owners of the unified NoahMP repository want to do in terms of the internal NoahMP code, it may be necessary to follow the route of RRTMGP and have only the CCPP interface to NoahMP exist in the ccpp-physics repository, OR, if all users/developers of the code within the unified NoahMP repository will be using CCPP-compliant hosts, it would be possible to avoid the extra software layer and make the code within the unified NoahMP repository CCPP-compliant.
In the meantime, @barlage and the land team at EMC have plans to update the NoahMP version in the ccpp-physics repository to match a snapshot of the code in the unified NoahMP repository, since what exists in the ccpp-physics today "is equivalent to WRF v3.6 so probably 5-6 years old likely" (from personal communication with @barlage).
Will the update of Noah-MP in the ccpp-physics repository be used for the CCPP v5 release and the associated UFS SRW App v1.0 release (code freeze January 22)? This is getting very close to the deadline to allow thorough testing.
On Mon, Jan 11, 2021 at 11:54 AM grantfirl [email protected] wrote:
@DomHeinzeller https://github.com/DomHeinzeller @ligiabernardet https://github.com/ligiabernardet @barlage https://github.com/barlage @ncarchen https://github.com/ncarchen This is a placeholder issue for setting up the NoahMP LSM as its own submodule within the ccpp-physics repository.
According to @barlage https://github.com/barlage via the AMS 2021 meeting, @ncarchen https://github.com/ncarchen and @barlage https://github.com/barlage have already begun a GitHub repository to contain the unified (between WRF and the NWM) NoahMP LSM: https://github.com/NCAR/noahmp
This should be what ends up as the ccpp-physics submodule. Depending on what the owners of the unified NoahMP repository want to do in terms of the internal NoahMP code, it may be necessary to follow the route of RRTMGP and have only the CCPP interface to NoahMP exist in the ccpp-physics repository, OR, if all users/developers of the code within the unified NoahMP repository will be using CCPP-compliant hosts, it would be possible to avoid the extra software layer and make the code within the unified NoahMP repository CCPP-compliant.
In the meantime, @barlage https://github.com/barlage and the land team at EMC have plans to update the NoahMP version in the ccpp-physics repository to match a snapshot of the code in the unified NoahMP repository, since what exists in the ccpp-physics today "is equivalent to WRF v3.6 so probably 5-6 years old likely" (from personal communication with @barlage https://github.com/barlage).
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/NCAR/ccpp-physics/issues/551, or unsubscribe https://github.com/notifications/unsubscribe-auth/AE7WQAV2JUUQF4LSN2XGEGLSZNCNBANCNFSM4V57WSWQ .
@ligiabernardet I don't have any details on the timeline. Mike Barlage mentioned that there was probably not time to do this for the upcoming release. I put an issue here just so that we don't forget about it and can reference that the issue has been closed when completed.
Sounds good. In talking with SRW App release code lead (Jeff Beck) he said this is not needed for the SRW App v1.0 release. So we are all on the page that this change will not be used for the CCPP v5 release.
On Tue, Jan 12, 2021 at 9:15 AM grantfirl [email protected] wrote:
@ligiabernardet https://github.com/ligiabernardet I don't have any details on the timeline. Mike Barlage mentioned that there was probably not time to do this for the upcoming release. I put an issue here just so that we don't forget about it and can reference that the issue has been closed when completed.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/NCAR/ccpp-physics/issues/551#issuecomment-758765597, or unsubscribe https://github.com/notifications/unsubscribe-auth/AE7WQAQNZAOEY6YKYJPP3TDSZRYQHANCNFSM4V57WSWQ .
I am going to talk with Mike tomorrow. I think we should include those Noah MP bug fixes to CCPP v5 release. The code is under my fork and has been tested. My branch is several commits behind and I will synchronize with the master today. So they will be ready to be merged if the decision is made.
It is always good to have bug fixes. However, note that we have limited time for further testing of CCPP v5 and note that this will impact the code distributed with the UFS Short-Range App (attn @Jeff Beck - NOAA Affiliate [email protected]). They both have code freezes this week. Can you please tell us the nature of these bug fixes?
On Mon, Jan 18, 2021 at 8:33 AM HelinWei-NOAA [email protected] wrote:
I am going to talk with Mike tomorrow. I think we should include those Noah MP bug fixes to CCPP v5 release. The code is under my fork and has been tested. My branch is several commits behind and I will synchronize with the master today. So they will be ready to be merged if the decision is made.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/NCAR/ccpp-physics/issues/551#issuecomment-762322518, or unsubscribe https://github.com/notifications/unsubscribe-auth/AE7WQAU5MRPQIKM33BVZO23S2RIFVANCNFSM4V57WSWQ .
Is the UFS Short-Range App configured to use NOAH MP as the default in CCPP v5 release ? If not, I think the bug fix can go in at a later date with other NOAH MP updates -- Fanglin
On Mon, Jan 18, 2021 at 6:16 PM ligiabernardet [email protected] wrote:
It is always good to have bug fixes. However, note that we have limited time for further testing of CCPP v5 and note that this will impact the code distributed with the UFS Short-Range App (attn @Jeff Beck - NOAA Affiliate [email protected]). They both have code freezes this week. Can you please tell us the nature of these bug fixes?
On Mon, Jan 18, 2021 at 8:33 AM HelinWei-NOAA [email protected] wrote:
I am going to talk with Mike tomorrow. I think we should include those Noah MP bug fixes to CCPP v5 release. The code is under my fork and has been tested. My branch is several commits behind and I will synchronize with the master today. So they will be ready to be merged if the decision is made.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <https://github.com/NCAR/ccpp-physics/issues/551#issuecomment-762322518 , or unsubscribe < https://github.com/notifications/unsubscribe-auth/AE7WQAU5MRPQIKM33BVZO23S2RIFVANCNFSM4V57WSWQ
.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/NCAR/ccpp-physics/issues/551#issuecomment-762509372, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKY5N2OHARX6YRG6NITUJSDS2S6K5ANCNFSM4V57WSWQ .
-- Fanglin Yang, Ph.D. Chief, Model Physics Group Modeling and Data Assimilation Branch
NOAA/NWS/NCEP Environmental Modeling Center
https://www.emc.ncep.noaa.gov/gmb/wx24fy/fyang/ https://www.emc.ncep.noaa.gov/gmb/wx24fy/fyang/
The UFS SRW App will be distributed with two CCPP suites: GFSv15p2 (operational GFS) and RRFS_v1alpha (which uses NoahMP). Suite RRFS_v1alpha, with NoahMP, is very important for the UFS SRW App release. Therefore, if there are known bugs in NoahMP, we need to either fix the bugs or publish the problem/fix as a "known issue".
The bug fixes includes:
(1) Some updates to prevent the model from crashing. The Noah MP is running in single precision currently and it has caused a significant number of crashing cases from our tests. Almost all crashing cases were gone after we used double precision for Noah MP like other physics. (2) Noah MP uses its own albedo and emissivity, which are different from those used in the surface radiation calculations outside Noah MP. (3) The wrong downward longwave forcing to Noah MP. Noah MP expects the downward longwave before emission, however the current driver passed the one after emission. Inside Noah MP it is multiplied by surface emissivity again. So generally Noah MP is getting less downward longwave.
For (1) and (3), they are straightforward and we only need to modify the code under CCPP physics. For (2), we need to make some modifications in the other parts of the model. I am waiting for them to remove IPD completely then commit my change.
Helin
On Mon, Jan 18, 2021 at 6:16 PM ligiabernardet [email protected] wrote:
It is always good to have bug fixes. However, note that we have limited time for further testing of CCPP v5 and note that this will impact the code distributed with the UFS Short-Range App (attn @Jeff Beck - NOAA Affiliate [email protected]). They both have code freezes this week. Can you please tell us the nature of these bug fixes?
On Mon, Jan 18, 2021 at 8:33 AM HelinWei-NOAA [email protected] wrote:
I am going to talk with Mike tomorrow. I think we should include those Noah MP bug fixes to CCPP v5 release. The code is under my fork and has been tested. My branch is several commits behind and I will synchronize with the master today. So they will be ready to be merged if the decision is made.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <https://github.com/NCAR/ccpp-physics/issues/551#issuecomment-762322518 , or unsubscribe < https://github.com/notifications/unsubscribe-auth/AE7WQAU5MRPQIKM33BVZO23S2RIFVANCNFSM4V57WSWQ
.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/NCAR/ccpp-physics/issues/551#issuecomment-762509372, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALPHKYEMV5UJWBHTHTES7QLS2S6K3ANCNFSM4V57WSWQ .
@HelinWei-NOAA Thanks for providing information about the nature of the fixes. I am looking forward to seeing a PR. I am concerned about the timing of the fixes because the release freeze date is this Friday 01/22. To my knowledge, there will not be PRs related to IPD removal before Friday. I am not aware of the timeline for when the IPD removal commits are scheduled. Would it be possible for you to submit a ccpp-physics PR at least for (1) and (3) so we can start reviewing? For expediency, I suggest directing the PR to the release/public-v5 branch. Later we can bring it to master. We also need to make sure that the fixes are sufficiently tested for the release. @JeffBeck-NOAA Would your team be running some SRW App tests for the release after these fixes are submitted? The CCPP team can run some single-column model tests.
@ligiabernardet, yes, we will definitely need to run a few tests with the RRFS_v1alpha SDF within the SRW App (all three pre-defined domains with varying ICs/LBCs) to ensure these changes don't cause any unexpected results.
I feel like maybe I missed some discussion along the way since this took a turn to bug fixing Noah-MP rather than the issue of moving Noah-MP to a submodule. Should we have a new issue that can be referenced by Helin's bug-fix PR?
Yes, pls start a new issue
On Tue, Jan 19, 2021 at 7:34 PM Michael Barlage [email protected] wrote:
I feel like maybe I missed some discussion along the way since this took a turn to bug fixing Noah-MP rather than the issue of moving Noah-MP to a submodule. Should we have a new issue that can be referenced by the PR?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/NCAR/ccpp-physics/issues/551#issuecomment-763286507, or unsubscribe https://github.com/notifications/unsubscribe-auth/AE7WQAVHTD5ZZI4IUABWFIDS2Y6LTANCNFSM4V57WSWQ .