HAFS-MOM6 IC coupled using CDEPS

[JohnSteffen-NOAA]

Hello all,

I'm working on the standalone HAFS-MOM6 IC coupled configuration which supplies MOM6 SST to FV3, one-way coupled. My working directory on Orion is here,

/work/noaa/hwrf/save/jsteff/hafs_couplehycom_cdeps/

I followed the README instructions in this directory, which I modified from Ling Liu. However, after install/compile of the HAFS system, the generated executable in ./HAFS/sorc/hafs_forecast.fd/tests/fv3_32bit.exe produces an error after 30s of runtime,

" 879: FATAL from PE 879: atmosphere_dynamics: error in ccpp_physics_init for group fast_physics: Invalid suite FV3_HAFS_v0_gfdlmp_tedmf_nonsst"

Both Ufuk and Bin suggested that there may be an issue with the HAFS model version or build. Any suggestions on how to check this and correct the error?

*I will note, this configuration runs successfully when using a previously compiled fv3_org.exe from Ling. A copy of the original directory is here,

/work/noaa/hwrf/save/jsteff/HAFS_couplling_orig

And successfully generated output files are moved to here,

/work/noaa/hwrf/save/jsteff/HAFS_couplling_orig/cpld_regional_natl_c192_docn_mom6/output_fv3org

Thank you for your advice!

Best,
John

[evankalina]

Hi @JohnSteffen-NOAA! Thanks for posting this question. I'm not able to see your directories (permission denied), but this type of error would mean that the FV3_HAFS_v0_gfdlmp_tedmf_nonsst wasn't included in the fv3 executable at compile time. In the HAFS application-level feature/hafs_couplehycom_cdeps branch, the sorc/build_forecast.sh script includes this physics suite in the call to compile.sh. Does your build_forecast.sh script look like the one in the branch? Where did you get your code from?

[lliu1220]

@JohnSteffen-NOAA
Try the following to build:

git clone -b feature/hafs_couplehycom_cdeps --recursive https://github.com/hafs-community/HAFS
git clone -b develop --recursive https://github.com/hafs-community/HAFS
cd HAFS/sorc/hafs_forecast.fd
git checkout feature/hafs_couplehycom_cdeps
git submodule update --init —recursive

and then you could compile the model by amodifying build_forecast.sh as following,

./compile.sh "$target" "SUITES=HAFS_v0_gfdlmp_nonsst 32BIT=Y CDEPS=Y CDEPS_DOCN=Y CMEPS=Y" 32bit YES NO

then you could link those files to your run directory,

cd RUN_test_fv3_docn
ln -sf $YOUR_PATH/HAFS/sorc/hafs_forecast.fd/NEMS/src/conf/module-setup.sh.inc module-setup.sh
ln -sf $YOUR_PATH/HAFS/sorc/hafs_forecast.fd/modulefiles/orion.intel/fv3 modules.fcst
ln -sf $YOUR_PATH/HAFS/sorc/hafs_forecast.fd/tests/fv3_32bit.exe .

[JohnSteffen-NOAA]

Hi @lliu1220 , with the recent updates made to feature/hafs_couplehycom_cdeps some of these steps are outdated and no longer work. Maybe @uturuncoglu could provide some insight. Thank you all for the help!

[evankalina]

@JohnSteffen-NOAA It appears that your input data file doesn't exist:

Orion-login-1[39] ekalina$ pwd
/work/noaa/hwrf/save/jsteff/hafs_couplehycom_cdeps/cpld_regional_natl_c192_docn_mom6/INPUT_DOCN
Orion-login-1[40] ekalina$ ls -ltr hycom_201908_surf_nolev.nc
lrwxrwxrwx 1 jsteff hwrf 80 Aug 2 13:07 hycom_201908_surf_nolev.nc -> ../../old_RUN_test_fv3_docn/DOCN_MOM6_input_data/hycom_201908_surf_nolev_mom6.nc
Orion-login-1[41] ekalina$ ls -ltr ../../old_RUN_test_fv3_docn/DOCN_MOM6_input_data/hycom_201908_surf_nolev_mom6.nc
ls: cannot access ../../old_RUN_test_fv3_docn/DOCN_MOM6_input_data/hycom_201908_surf_nolev_mom6.nc: No such file or directory

[JohnSteffen-NOAA]

Hi @evankalina , that is an oversight by me. It's a broken link from when I initially copied Ling's directory. I have it set up now where the input files from HYCOM and MOM6 are located in my working directory and linked from there. I'll add this step to the README file and this should provide all the necessary steps to test this configuration.

I ran the fv3_32bit.exe and it works. There is a NetCDF start+count error towards the end of the run, but that is a separate issue I have already discussed with Ling.

Thanks for all your help! I appreciate it.

Best,
John

[evankalina]

@JohnSteffen-NOAA Great to hear that! Please let me know if you run into any other issues.

[YongzuoLi-NOAA]

Hi @JohnSteffen-NOAA I am reading this. My working directory on Orion is here, /work/noaa/hwrf/save/jsteff/hafs_couplehycom_cdeps/ I followed the README instructions in this directory, which I modified from Ling Liu.
But I can not find README. README.md there has no Ling Liu modification.

[JohnSteffen-NOAA]

Hi @YongzuoLi , try this directory here, /work/noaa/hwrf/save/jsteff/hafs_couplehycom_cdeps_mom6 . This work was performed before a HAFS-CDEPS approved PR, so further modifications may be needed. There's an updated HAFS-CDEPS configuration which can import/use DOCN in hycom formatting. What are you trying to do?

[YongzuoLi-NOAA]

Hi @JohnSteffen-NOAA Thank you for the information. I now see /work/noaa/hwrf/save/jsteff/hafs_couplehycom_cdeps_mom6/README and will follow it. I now work on regional MOM6-CDEPS-HAFS. Are your HAFS and MOM6 (SIS2 GFDL repo ?) here for regional ?

[JohnSteffen-NOAA]

@YongzuoLi Yes, these configurations are setup for regional HAT10 domain. I would suggest following up with Cameron Book. He is also currently working on HAFS-CDEPS runs using MOM6 IC.

[YongzuoLi-NOAA]

@JohnSteffen-NOAA Yes, I am working Cameron Book now.