Update UFS model with physics updates for GFSv17 #3840
Conversation
|
Gaea tests can be found here: /gpfs/f6/ira-sti/world-shared/Jessica.Meixner/updatemodel20250625/RUNTESTS/EXPDIR |
|
There has been a minor update to the physics settings to appropriately match what is here: https://docs.google.com/document/d/1LOJ0Mip2ZXOHKQxIA_Pkv6GlTXUnOHQbFQl2Psuf2TQ/edit?tab=t.0 which was originally incorrectly updated for lower resolutions. Thanks to @yangfanglin for spotting this. @XiaqiongZhou-NOAA - this PR is ready for your review. |
|
2 more UFS PRs have made their way through the PR queue. |
Recent GEFSv13 reforecast evaluation suggest we might need this update (rf_cutoff =300) too. @XiaqiongZhou-NOAA might provide more input shortly. |
|
@XiaqiongZhou-NOAA @bingfu-NOAA we probably should set rf_cutoff=300.0 for all resolutions |
Okay - so you want rt_cutoff=300 for all resolutions and also set for GEFS? What about SFS? |
|
SFS can also use rf_cutoff=300 |
JessicaMeixner-NOAA
requested review from
yangfanglin,
bingfu-NOAA and
XiaqiongZhou-NOAA
|
I've updated rt_cuttoff and requested reviews from you all to confirm that I have updated as it should be. |
|
NOTE: this PR since we went to top of ufs now needs and update to gdas app see: NOAA-EMC/GDASApp#1778 |
adding gocart changes for head of develop
|
Experiment C96C48mx500_S2SW_cyc_gfs FAILED on Hercules in Build# 1 in |
emcbot
added
CI-Hercules-Failed
|
All the errors on hercules that I've looked at so far have been srun errors. Installing a generate_workflows to test out from my account so I can try re-running/debugging if necessary. |
|
It looks like Hera may need a module file update. All of the forecasts failed at launch time with GLIBCXX errors. For C48_ATM |
DavidHuber-NOAA
added
the
CI-Hera-Failed
|
I'm wondering if hercules needs a module and/or environment update or if srun was just having errors this weekend? My re-run had the exact same problems the CI tests had. Module update seems odd because don't we use the ufs-weather-model modules in g-w? And those have been running with what's there. I wonder if there are hera environment variables that need updated versus modules? I'll check out the ufs commit history. |
|
On WCOSS2, the UPP has changed its module names. upp.sh should be updated with the following patch: diff --git a/dev/jobs/upp.sh b/dev/jobs/upp.sh
index 4d070e9c..50b2825e 100755
--- a/dev/jobs/upp.sh
+++ b/dev/jobs/upp.sh
@@ -18,7 +18,7 @@ if [[ "${MACHINE_ID}" == "wcoss2" ]]; then
set +x
source "${HOMEgfs}/ush/module-setup.sh"
module use "${HOMEgfs}/sorc/ufs_model.fd/FV3/upp/modulefiles"
- module load "${MACHINE_ID}"
+ module load "${MACHINE_ID}_intel"
module load prod_util
module load cray-pals
module load cfp |
No, we moved to loading the workflow modules for the R&D machines when we updated to spack-stack 1.5.1, but it looks like we need to temporarily switch back to using the UFS modules since the model has upgraded to 1.9.2 and the workflow is still using 1.6.0. |
|
How close is your spack-stack PR? Do we just combine this w/that so you can not have that work around? |
JessicaMeixner-NOAA
dismissed stale reviews from aerorahul and DavidHuber-NOAA
via
6dc014e
|
@JessicaMeixner-NOAA I am having issues with the 2nd full cycle gdas_fcst and enkfgdas_efcs jobs for all tests on Hera and MSU. I would be happy to combine PRs and tackle these issues together, which would also allow me to remove a kludge in Thoughts @aerorahul? |
|
We should at minimum see if combining helps resolve issues for both of us maybe? The sooner we can get part/all of these updates in the better. I'll see if on hercules if i load ufs-modules if any of the forecasts will go through. |
|
@JessicaMeixner-NOAA Sounds good to me. I'll give that a try on Hera. |
|
Okay, I updated the dev/jobs/fcst.sh job to load the ufs modules, I get this module load error and then I get this error: This makes me lean towards combining the PRs as the path forward. That being said - we have run on both Gaea and WCOSS2 with the updated model (not perhaps this exact version but similar). |
|
Given that the spack-stack Pr is stuck... I'm going to work on updating this even though many changes will have to be un-done w/the spack-stack PR. |
|
@DavidHuber-NOAA - did the WCOSS2 regtests pass? I'm assuming hera is not going to work until #3798 is through? |
|
@JessicaMeixner-NOAA apologies for not posting the results. Unfortunately, the working space on stmp has been scrubbed, but based on emails from cron, it looks like the UPP jobs failed (both the atmanlupp and goes_upp for the extended tests). The runtimes were all quite short, so my guess is this was caused by module loads and was probably fixed by 6dc014e. |
|
Let me know when you are ready and I will kick off another suite of tests on Cactus. |
|
No worries @DavidHuber-NOAA - I'll do a few tests on hercules and wcoss2 and let you know when this is ready to be re-tested. In the meantime - I'll continue to engage on the spack-stack Pr as well. |
DavidHuber-NOAA
removed
the
CI-Wcoss2-Ready
|
On hercules, loading the UWM modules does not work because then we get an error: |
|
On WCOSS2 - I did get have almost all tests pass: The SFS test failure is here /lfs/h2ouple/noscrub/jessica.meixner/updateModelDev/upt03/COMROOT/C96mx100_S2S_upt03/logs/1994050100/sfs_fcst_mem001_seg1.log.0 And I'm getting a "permission denied" This might be an issue for you too @DavidHuber-NOAA Also tagging @jiandewang since it's ocean and @XiaqiongZhou-NOAA @NeilBarton-NOAA since it's an SFS test. |
|
I believe this was fully wrapped into #3798. Can you confirm @JessicaMeixner-NOAA? |
|
Confirm - This was merged into #3798. |
Description
Updates model with the last of the known physics updates (at this time) for the model. Changes in the model can be seen here:
ufs-community/ufs-weather-model@9193010...1f0ab43
This includes ufs-community/ufs-weather-model#2730 which requires a change of cutoff level
It also includes updates to use the new spack-stack and porting to URSA. When spack-stack was updated, GOCART was updated and it includes the following updates (thanks to @bbakernoaa) and are described here:
This pull request introduces several updates to configuration files for the UFS GOCART model, focusing on adding new parameters, improving wet removal schemes, and updating external data sources. Below is a summary of the most important changes, grouped by theme.
Wet Removal Scheme Enhancements:
wet_removal_scheme: ufsalong with rainout efficiency parameters (fwet_ice,fwet_snow,fwet_rain) to multiple configuration files, includingCA2G_instance_CA.bc.rc,CA2G_instance_CA.br.rc,CA2G_instance_CA.oc.rc,DU2G_instance_DU.rc, andSS2G_instance_SS.rc. These parameters define the efficiency of wet removal processes for different precipitation types. [1] [2] [3] [4] [5]New Parameters for Physical and Chemical Processes:
time_days_hydrophobic_to_hydrophilic(conversion rate of hydrophobic to hydrophilic aerosols) andtime_days_chemical_destruction(rate of chemical destruction of carbon species) inCA2G_instance_CA.bc.rc,CA2G_instance_CA.br.rc, andCA2G_instance_CA.oc.rc. These parameters enhance the representation of aerosol lifecycle processes. [1] [2] [3]soil_clay_factorinDU2G_instance_DU.rcto scale clay fraction in dust emissions.Updates to External Data Sources:
ExtData.otherfor variables such asDU_RDRAG,DU_GVF, andDU_LAI, pointing to new datasets for albedo drag, green vegetation fraction, and leaf area index.volcano_srcfilen_explosiveandvolcano_srcfilen_degassinginSU2G_instance_SU.rc, improving the handling of volcanic SO₂ sources.Dust Emission Scheme Adjustments:
alphaparameter in the Fengsha dust emission scheme from0.1to0.16and addeddrag_partition_option: 2inDU2G_instance_DU.rc, refining the dust emission calculations.<!--Type of change
Change characteristics
Is this a breaking change (a change in existing functionality)? NO
Does this change require a documentation update? NO
Does this change require an update to any of the following submodules? YES
We will need an update to GDAS App due to MOM6 updates. Not sure of the GDAS App PR yet, but SOCA PR is Update MOM6 to f9e6e2e JCSDA-internal/soca#1177
Fix file update: New GOCART input for updated ufs wm #3864
How has this been tested?
Full CI on Gaea C6 (will post log in comment)
Checklist