Issue with radiation fluxes in DATM+SCHISM configuration

[uturuncoglu]

Initial Mail:

Hi Ufuk,

I hope you are doing okay.

I am reaching out to hear your thoughts on UFS-Coastal/spack stack/ESMF. As you know, on Stampede3, currently an older version of UFS Coastal is available, with containerized Spack Stack (v1.6.0), and ESMF (v8.6). Both Saeed and Kyungmin were successful with installing/testing them. This version of UFS-Coastal has some remapping/interpolation issues.

We are hoping that Spack Stack v1.9.1 and ESMF 8.8.0 will be installed so that we can use the latest version of UFS-Coastal. Kyungmin actually tried to install them but encountered issues related to the Intel MPI runtime. It's unclear why this error happened. Kyungmin thinks this might be due to downloading or using the wrong Spack-Stack tag or image variant. I also wonder if this is related to the EULA (end user license agreement) issue, but not sure.

Two questions:

1. Do you think, if we request TACC to install Spack Stack v1.9.1 and ESMF 8.8.0 (rather than individual users installing them), we could bypass the license issue?
2. Do you know if ESMF8.8 would address the issue of solar radiation remapping and interpolation with CDEPS?

Any thoughts will be appreciated. Happy to try to set up a short meeting if it is easier to talk.

Thank you!
Ayumi

--
Ayumi Fujisaki-Manome (she/her/hers)
Associate Research Scientist
Cooperative Institute for Great Lakes Research (CIGLR), School for Environment and Sustainability (SEAS)
Climate & Space Sciences and Engineering (CLASP)
University of Michigan

Follow-up from (saeed.memari@mines.edu):

Hi All,
Thank you both for the follow-up. I wanted to briefly summarize the issue. The problem occurred when shortwave and longwave radiation variables were passed through the mediator to SCHISM. While other atmospheric fields (wind, pressure, air temperature, humidity) remapped correctly, the solar fluxes showed unrealistic values after interpolation, either extremely high or near zero. The data itself was verified to be correct and worked fine in standalone SCHISM runs, so the issue appeared to originate from the mediator’s remapping of solar radiation (and/or CDEPS?).
I’ve attached a short PDF summarizing my earlier discussion with Ufuk on this topic, which may help as a refresher. Since then, I believe the more recent UFS and ESMF updates (as Joseph mentioned earlier) may have addressed this issue—but could you please confirm if that is indeed the case? If not, we can still use the previous ESMF (v8.6) version available in the EPIC Spack-Stack (v1.6) when compiling UFS.
Best regards,
Saeed

[uturuncoglu]

The problem was in the data file. See following information for more,

---

Hi Saeed,

It took a little bit of time, but I think I found the issue with the radiation. As I expected, it is not related to CDEPS or the coupling interface but to the data file. The CDEPS expects double data type for all the variables, but in your input file, radiation components and also precipitation are in float type. Anyway, you could use the following NCO command to make them double.

ncap2 -s 'dswrf=double(dswrf);dlwrf=double(dlwrf);prate=double(prate)' HRRR_UFS_NYH_Allvars_2021.nc HRRR_UFS_NYH_Allvars_2021_fixed.nc

Then, you could run your case with the fixed file. On my end, it works fine.

I think you could run your original configuration first but you might want to create another stream just for shortwave using following datm.stream,

stream_info: atmmesh.01 atmmesh.02

taxmode01: limit
mapalgo01: redist
tInterpAlgo01: linear
readMode01: single
dtlimit01: 1.5
stream_offset01: 0
yearFirst01: 2021
yearLast01: 2021
yearAlign01: 2021
stream_vectors01: null
stream_mesh_file01: "INPUT/saeed_ESMFmesh.nc"
stream_lev_dimname01: null
stream_data_files01: "INPUT/HRRR_UFS_NYH_Allvars_2021_fixed.nc"
stream_data_variables01: "uwind Sa_u10m" "vwind Sa_v10m" "prmsl Sa_pslv" "stmp Sa_t2m" "dlwrf Faxa_lwdn" "spfh Sa_q2m"

taxmode02: limit
mapalgo02: redist
tInterpAlgo02: coszen
readMode02: single
dtlimit02: 1.5
stream_offset02: 0
yearFirst02: 2021
yearLast02: 2021
yearAlign02: 2021
stream_vectors02: null
stream_mesh_file02: "INPUT/saeed_ESMFmesh.nc"
stream_lev_dimname02: null
stream_data_files02: "INPUT/HRRR_UFS_NYH_Allvars_2021_fixed.nc"
stream_data_variables02: "dswrf Faxa_swdn”

Anyway, let me know if you have any issues.

PS: I’ll create an issue on the UFS Coastal side to keep tracking and put our previous exchange to it. Once you have a working case. We could close it.

Best,

—ufuk