Towards RTOFS in ufs-community

[NickSzapiro-NOAA]

Continuing discussion here @sanAkel @raphaeldussin @DeniseWorthen

Out of curiosity, I did want to ask if there was interest for RTOFS and HAFS to share some workflow/scripts. Like
https://github.com/hafs-community/HAFS/blob/develop/scripts/exhafs_ocn_prep.sh
seems to initialize from RTOFS and make gfs_forcings from atmosphere output.

I started a ufs-weather-model branch at:
https://github.com/NickSzapiro-NOAA/ufs-weather-model/tree/marine_datm
with two main features: 1) some example test configurations and 2) a particular CDEPS branch.

The CDEPS branch has a "generic" datamode to directly input any atmospheric variables (prepared offline).
Review at ESCOMP is expected to take a while, as they're busy preparing for CMIP7. But, it may be slower in CDEPS on Ursa (??).
For this, "workflow scripts" would prepare datm input files from ERA5 (or CFORCE update), GFS,...

One goal is a useful test suite (in tests/parm/my_rt.conf) at global 1/4 degree ocean-ice.

A working idea is to base around coupled "GFS"or "SFS" tests that track those configurations. Can get MOM6, CICE6, CMEPS restarts (among other things) from those too.

Then make flavors like

• cpld_woa_sfs initializes MOM6 via the World Ocean Atlas override
• cpld_pcice_sfs prescribes CICE6 to an input ice concentration from different year with slabs of ice (2m thick in Arctic and 1m thick in Antarctic)
• cpld_insert_sic_sfs: TODO Exercise adjust_aice feature already in CICE
  or planning ones for
• datm_era5_s2s TODO : will have data atmosphere - MOM6 - CICE6 instead. Inputs are the ERA5-based data atmosphere forcing files, processed with TODO script
• datm_cplhist_s2s TODO: with data atmosphere too, but the atmospheric forcing is from CMEPS history (of GFS/SFS test)
• datm_gfs_s2s TODO : with data atmosphere too, but the atmospheric forcing is from UFSATM history (of GFS/SFS test)

Hopefully can make some configuration switches and couplings to help build & understand model behavior

[sanAkel]

Continuing discussion here @sanAkel @raphaeldussin @DeniseWorthen

Out of curiosity, I did want to ask if there was interest for RTOFS and HAFS to share some workflow/scripts. Like https://github.com/hafs-community/HAFS/blob/develop/scripts/exhafs_ocn_prep.sh seems to initialize from RTOFS and make gfs_forcings from atmosphere output.

Yes @NickSzapiro-NOAA. Definitely there is not only an interest, but [work](see https://github.com/sanAkel/ufs-ocean/blob/main/scripts/grib_to_nc_mapping.md) ✔️ towards this goal.

Unfortunately that work has not (yet) been pulled into this repo.

[sanAkel]

@NickSzapiro-NOAA Can you please close this issue in favor of a discussion: #146 ?

Thanks!

[NickSzapiro-NOAA]

Oh, nice. I don't see an option to convert to discussion ... please do if you can @sanAkel

[sanAkel]

@NickSzapiro-NOAA ✅

[NickSzapiro-NOAA]

I also just heard about nos-utils https://nos-utils.readthedocs.io/en/latest/index.html for preparing atmospheric forcing from output on wcoss2

[sanAkel]

I also just heard about nos-utils https://nos-utils.readthedocs.io/en/latest/index.html for preparing atmospheric forcing from output on wcoss2

@NickSzapiro-NOAA Thanks. I looked up the scripts, the gfs forcing is no good for the UFS DATM (active ocean + sea ice) for anything like the RTOFS. It seems to miss the essential elements of radiative fluxes.

[NickSzapiro-NOAA]

Thanks for pointing out the GFS forcing script!

It's not an exact match but it's in the ballpark, no?
DEFAULT_VARIABLES = ["uwind", "vwind", "prmsl", "stmp", "spfh", "dlwrf", "dswrf", "prate"]
in nos-utils is close to what CDEPS GEFS data mode needs:
https://github.com/ESCOMP/CDEPS/blob/main/datm/datm_datamode_gefs_mod.F90#L21-L39

As is, would have to partition shortwave and rain/snow somewhere along the way

[NickSzapiro-NOAA]

I see some comments in there about "UFS-Coastal" options to blend GFS and HRRR

Interesting if, say, could add RRFS to the blend too

[DeniseWorthen]

I'm a little confused by the title; we currently have analogs for RTOFS, albeit at lower resolution (the basic DATM-OCN-ICE tests). We don't have a weekly configuration (like gfsv17_c1152) that runs at scale for easy sandboxing. That would be worthwhile adding, IMHO.

But it seems here you're talking about workflow development; are you taking on the task of developing a workflow for RTOFS?

[NickSzapiro-NOAA]

Thanks Denise. We do already have low-resolution example versions.

I guess I'm trying to support developments (particularly in data atmosphere and UFS/CICE initialization and configuration) to help make the operational application work as well as I can.

I'm not trying to make an RTOFS workflow, but I'm also not eager to make separate/temporary/one-off ways to do things.

For data atmosphere forcing, I can get some files from HPSS. But maybe there is already a sound, vetted (NOS) workflow to make better forcing ... as gulf stream, ice extent, etc. are maybe better prescribed in RRFS. So it makes sense to me to collaborate and co-produce on that

[sanAkel]

@NickSzapiro-NOAA Can you please itemize what exactly do you want to discuss here?

[NickSzapiro-NOAA]

For a full resolution weekly test, we need MOM6 and CICE6 restarts with DA increments/adjustment files as well as datm forcing at target resolution. What files should we use for a full RTOFS test?

Ideally, those would come from application workflow (like the GFS, GEFS, SFS RTs do).

[sanAkel]

Thanks @NickSzapiro-NOAA. I will address/answer your items in a few PRs in next few weeks.