Enable runs that use RRTMGP with 127 vertical levels #300
Labels
enhancement
New feature or request
Comments
|
The code is committed in PR #378, close issue. |
pjpegion
pushed a commit
to NOAA-PSL/ufs-weather-model.p7b
that referenced
this issue
Jul 20, 2021
…pp-physics (ufs-community#300) - renames the ccpp-framework and ccpp-physics branches from master to main in .gitmodules - updates the submodule pointer for ccpp-physics for several small changes described in NCAR/ccpp-physics#658 - updates the submodule pointer for ccpp-framework for a small update described in NCAR/ccpp-framework#372 - add missing Intel DEBUG flags specific to CCPP
epic-cicd-jenkins
pushed a commit
that referenced
this issue
Apr 17, 2023
This PR must be merged after PR #[26](ufs-community/ufs-srweather-app#26) into ufs-community/ufs-srweather-app. ## DESCRIPTION OF CHANGES: ### Modifications to workflow -- these are needed mostly to make it more convenient to formulate/run the WE2E tests: * Introduce the new workflow variable USE_USER_STAGED_EXTRN_FILES to indicate whether or not to look in user-specified directories for external files. Previously, this was determined by checking whether the variables EXTRN_MDL_SOURCE_DIR_ICS and EXTRN_MDL_SOURCE_DIR_LBCS are empty or not. Now, there is an explicit flag to set. * Make the number of attempts for the rocoto tasks (maxtries) user-specifiable workflow variables (instead of setting them to a default of 1). * Allow the workflow variable EXPT_BASEDIR to be set to a relative path. In this case, the relative path gets created under the usual default location of EXPT_BASEDIR (which is ${SR_WX_APP_TOP_DIR}/../expt_dirs). * Bug fix - update the configuration file name for UPP from postxconfig-NT-fv3sar.txt to postxconfig-NT-fv3lam.txt to match the name in the EMC_post repository (release branch). ### Modifications to script that runs the WE2E tests (run_experiments.sh): * Add new arguments expt_basedir, testset_name, and verbose. expt_basedir and verbose can be used to override the default values of EXPT_BASEDIR and VERBOSE in the workflow. testset_name can be used to specify a subdirectory under expt_basedir in which to place all the experiment directories. This is convenient to isolate sets of tests in different directories. * Bug fix - Source the default workflow configuration file for each WE2E test, not just once at the beginning of the script. * Rewrite the way EXPT_SUBDIR is set so that it is set properly whether or not expt_basedir and testset_name are specified as input arguments. * Instead of using "cat" with heredocs to create the workflow configuration file for each test, store the contents of the configuration file in a variable (str) and write it out to a file after it is complete. * If running the WE2E tests on hera or cheyenne, increase maxtries for various tasks that will likely need more than one attempt (because non-reproducible bugs in codes cause the tasks to fail). * Remove unused code. ### Modifications to WE2E test configuration files: * Remove lines that set workflow variables (e.g. RUN_TASK_MAKE_GRID) to their default values in config_defaults.sh because they effectively do nothing; better to keep the test configuration files shorter. * Add 4 new WE2E tests that test fetching of files from various external models from NOAA HPSS. These use the new workflow variable USE_USER_STAGED_EXTRN_FILES described above. * Remove lines in some WE2E configuration files that now get generated in run_experiments.sh (e.g. comments included in those configuration files that run in NCO mode). * Rename WE2E test configuration files to clarify their purpose, i.e. what workflow feature/capability they test. * Update the list of WE2E test names in baselines_list.txt. ## TESTS CONDUCTED: The WE2E tests were run on both hera and cheyenne. The experiment directories can be found here: hera: /scratch2/BMC/det/Gerard.Ketefian/UFS_CAM/PR_feature_WE2E_testing/expt_dirs/try13 cheyenne: /glade/scratch/ketefian/PR_feature_WE2E_testing/expt_dirs/try14 Note that: * On both machines: * Test suite_FV3_CPT_v0 fails due to a problem with the namelist (likely a namelist variable name is incorrect). This is a preexisting issue, i.e. it also occurs in the develop branch. * Test suite_FV3_RRFS_v1beta fails in the make_orog task because orography statistics files required by the gravity wave drag (GWD) parameterization in the RRFS_v1beta scheme aren't available. This will be temporarily remedied by using a different GWD parameterization in this suite. That change will happen in a future PR. * On hera: * Test grid_GSD_RAP13km fails because run_post_f003 fails (even after 20 tries) while run_post_f005 hangs. Others run_post_f### tasks that succeed take many attempts. This did not happen with the previous version of EMC_post being used in the workflow. Thus, it is either due to this change in EMC_post version or the change in the grid generation code (from regional_grid to regional_esg_grid). Needs further investigation. * All tests other than grid_GSD_RAP13km, suite_FV3_CPT_v0, and suite_FV3_RRFS_v1beta succeed. * On cheyenne: * Test grid_GSD_RAP13km fails because run_post_f005 hangs. All other run_post_f### tasks succeed, although some take many attempts. As on hera, this did not happen with the previous version of EMC_post being used in the workflow, so it is either due to the change in EMC_post version or the change in the grid generation code (from regional_grid to regional_esg_grid). Needs further investigation. * The four new tests get_extrn_files_from_hpss_... that try to fetch the external model files from NOAA HPSS fail. This is expected because cheyenne does not provide access to NOAA HPSS. Thus, these tests do not need to be run on cheyenne. * **All tests that use the HRRRX and/or RAPX as external model files for ICs/LBCs fail in both the make_ics and make_lbcs tasks.** * These tests are: grid_GSD_HRRR_AK_50km, nco_GSD_HRRR25km_HRRRX_RAPX, nco_GSD_HRRR3km_HRRRX_RAPX, nco_GSD_SUBCONUS3km_HRRRX_RAPX, suite_FV3_GSD_SAR, suite_FV3_GSD_v0, suite_FV3_RRFS_v1beta. (Recall that suite_FV3_RRFS_v1beta fails in the make_orog task (before the make_ics and make_lbcs tasks) because the orography statistics files needed by the FV3_RRFS_v1beta suite are not available.) * These failures do not happen on hera and thus are likely due to a bug in chgres_cube. This needs further investigation. * All tests other than the ones mentioned above succeed. The failures listed above are due missing input files or to possible bugs in the chgres_cube and EMC_post codes (or possibly due to the changeover to the new regional_esg_grid code, which will not be reverted), or they may indicate inconsistencies in the software versions loaded in the modulefiles on cheyenne. They do not indicate any problems with this PR.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Description
The UFS Weather Model cannot run a 127-level configuration with RRTMGP. It is expected to error out with message
array plev has values outside range. This configuration is needed for the test toward GFSv17/GEFSv13.Solution
A one-line change is needed in submodule https://github.com/earth-system-radiation/rte-rrtmgp/, in branch dtc/ccpp. Note that rte-rrtmgp is a submodule of ccpp-physics.
The change is described in earth-system-radiation/rte-rrtmgp@be10425
Alternatives
Apply the one-line change manually until the problem is resolved.
Related to
An issue will be created in the ccpp-physics repository
The text was updated successfully, but these errors were encountered: