Merge HAFSv0.2 configuration into develop and sync HAFS submodules as of 202107 #75
Conversation
…es not mask out EPAC ocean (From Hyun-Sook).
…rations/experiments. *The radiation scheme calling time steps (fhswr and fhlwr) now become config options. *Calculate blocksize according npy/layouty for input.nml.
…) options to control inline post and turning on/off history output capabilities. The inline post capability is not fully tested in the HAFS workflow yet.
…gth and dissipation mixing length in the GFS scale-aware TKE EDMF PBL scheme) with the value of 100.0, when the GFS scale-aware TKE EDMF PBL scheme is used. These changes were made by ChunxiZhang-NOAA and BinLiu-NOAA based on the rlmx/elmx value of 100.0 used by AndrewHazelton-NOAA.
*Increase KMP_STACKSIZE to 2048m in ush/hafs_runcmd.sh.inc
currently overwrites g2/3.4.1 with g2/3.4.2, but hafs_gfs2ofs2.x needs g2/3.4.1. Once this is fixed from the hpc-stack side, this can be removed. *Load upp/10.0.6 in modulefile.hafs.*
… is now needed for Jet as well. So, make this temporary change for all platforms by reloading g2/3.4.1 after loading upp.
running the obs-preproc step to deal with tempdrop drifting, an obs_proc task is added in HAFS workflow to deal with tempdrop drifting.
…ta in the monitoring mode, since the tempdrop drifting corrected observations are now being assimilated. (From Jason.Sippel).
*Update submodule hafs_gsi.fd Note: These modifications are from Xu Lu and Xuguang Wang (OU) to enable TDR and NEXRAD radial wind OBS being properly ingested/assimilated in the EnKF analysis. POC: [email protected].
Porting hafs_change_prepbufr from hwrf_prepbufr under hafs_tools.fd by BijuThomas-NOAA and BinLiu-NOAA.
Note: Xu and Xuguang (OU) and Jon (UMD) reported that the mem001 file instead of the ensemble mean is used in the enkf_update step. After confirming with Ting (EMC), the exhafs_enkf.sh script is changed to use the real ensemble mean file generated from enkf_mean in the enkf_udate step.
…using t(tv) as a control vector variable.
…coss_cray and wcoss_dell_p3 to io1x96, so that the inline_post can run properly on these platforms.
near-surface mixing length in GFS sa-TKE EDMF PBL scheme.
…tion of near-surface mixing length in GFS sa-TKE EDMF PBL scheme
…S sa-TKE EDMF PBL scheme.
*Add a hafsv0p2a_regional_da_AL.conf configuration with a smaller domain and L81 levels
*Use 40x30 PEs for FV3ATM compute for the forecast job *Use 20x12 PEs for FV3aTM compute for ensemble forecast jobs
…le the fix of boundary-crawler issue working properly. Note: Andrew Hazelton reported that in their initial HAFSV0.2B test that 02L (Bill) was being tracked incorrectly along the boundary of the nest even though the tracker source code change made it into the HAFS repository already. And Tim Marchok confirmed that because this boundary-crawler issue only occurred under specific circumstances, the fix checks specifically for them. So, in order for the checking algorithms to even get called, trkrinfo%gridtype must = "regional" and inp%nesttyp must = "fixed".
…file for channel selections above 10 hPa into the monitoring mode. (From Li Bi)
BinLiu-NOAA
requested review from
ChunxiZhang-NOAA,
evankalina,
JunghoonShin-NOAA,
LinZhu-NOAA,
mrinalbiswas,
panll and
ZhanZhang-NOAA
…cales as HAFS application/workflow config options. Also, the following default values were given as follows: gsi.s_ens_h=150, gsi.s_ens_v=-0.5, enkf.corrlength=500, enkf.lnsigcutoff=1.3 based on the HAFS DA community discussion. Note: *These default localization scales are currently set based on the HWRF D03 DA settings and the HAFS DA community discussions. Further tests will be needed to determine and fine-tune these localization scale settings for HAFS DA system in the future. *Beside, we plan to also explore using the hybens_info file to specify the vertical profile of horizontal and vertical localization length scales for HAFS EnVar and EnKF analysis.
… feature/hafs_sync_202107
BinLiu-NOAA
changed the title
|
The regression tests on ufs-weather-model level were finished successfully. |
|
The regression tests with the feature/hafs_sync_202107 branch were successful on WCOSS-CRAY |
|
The regression tests with the feature/hafs_sync_202107 branch were successful on WCOSS-DELL |
|
The regression tests were successful on Hera with the feature/hafs_sync_202107 branch. |
|
Note: There is a known issue from the ufs-weather-model side: NOAA-EMC/fv3atm/issues/344, which changed the Time dimension and contents of the sfc_data.nc restart files. This broke the HAFS DA capability since DA/GSI need to read these sfc_data restart files. Fixing is ongoing from the ufs-weather-model side. And this HAFS DA capability will be restored next time when we sync ufs-weather-model develop branch after the fix. Currently, if you want to run the HAFS DA configurations/experiments, please keep using the feature/hafs_ensda branch instead. |
|
@mrinalbiswas, not sure if your application level regression tests go through on Orion successfully. If so, we are ready to merge this PR. Thanks! |
|
@BinLiu-NOAA The Orion tests are going very slow. The forecasts were on the queue for past three days. Now forecasts jobs are done. Post and products are running at different stages. I am hoping that they will be done by today. However, since the forecast jobs are done I am little confident that the tests will pass. So, you can go ahead with the merge, if needed. I will report any issues or report success when the tests are done on my end. |
|
@BinLiu-NOAA RT on Orion passed. |
|
Thanks @mrinalbiswas! |
This PR tries to:
And this PR addresses issues #72, #73, #64.
Notes:
Testing:
Again, appreciate it if you could help to run the regression tests as you helped previously!
@evankalina, @panll, @mrinalbiswas (Hera and Orion)
@JunghoonShin-NOAA (Jet, we probably will skip Jet since it is occupied by the HFIP real-time parallel experiments under reservation)
@LinZhu-NOAA (wcoss_dell_p3)
@BijuThomas-NOAA (wcoss_cray)
@ChunxiZhang-NOAA (ufs-weather-model level regression tests)
To conduct the HAFS application level regression tests, you can:
A. Clone and checkout the branch, e.g.,
git clone -b feature/hafs_sync_202107 --recursive https://github.com/hafs-community/HAFS.git ./hafs_sync_202107
B. Build and install HAFS
cd hafs_sync_202107/sorc
./install_hafs.sh # Afterward, please check or edit the ../parm/system.conf if needed.
C. Edit and run the cronjob driver as a cron job
cd ../rocoto
vi cronjob_hafs_rt.sh # make sure to use the correct platform and directory
./cronjob_hafs_rt.sh
To run the ufs-weather-model level regression tests:
cd sorc/hafs_forecast.fd/tests
vi detect_machine.sh and update this line into export ACCNR=${ACCNR:-hurricane} or your own compute project name.
Launch the ufs-weather-model level regression tests
nohup ./rt.sh -e > rt.log 2>&1 &