How to correct HYCOM SSH before using it as OBCs in ROMS

[isoares]

Dear forum and ROMS users,

I am working on the implementation of Ensemble Optimal Interpolation (ENOI) in a regional domain of ROMS in the North Pacific Ocean (see figure) .

I am using 6-hourly Hycom GLBy 0.08 T/S/U/V/SSH in the open boundaries of my ROMS regional grid. I am using the Chapman condition (Cha) on SSH and the Flather condition (Fla) on the external velocities. I also use the RadNud condition on the internal velocities and on T/S. In the surface I use the BULK fluxes, which are computed from CFSR v2.1 3-hourly variables. I am using air temp at 2 m, wind U/V at 10 m, specific humidity at 2m , air pressure at MSL, precipitation rate, and net long/short radiations.

This is working well. After 8 years of integration the results are looking good and the model is stable. However, the long-term mean of SSH shows higher values near the open boundaries, and smaller values in the middle of the domain. This is causing a negative impact in the DATA Assimilation.

When I run ROMS with ENOI, the results show higher errors near the open boundaries (see the figure).
I suspect that I need to correct the HYCOM SSH, so it will have the same MSL as the satellite along-track SLA that I am assimilating, before I impose hycom SSH in the boundaries.

The question is: Does anybody in the forum has a similar problem ?

and second question: Does anybody know where I will get the Global Hycom long-term mean sea level ?

I already tried the hycom forum, but I didn't get any answer from them.

If this is a topic which have already been addressed in the forum, please let me know. If not, I hope this will create a discussion that will be useful to other user too.

Thank You .

Ivan

npo_roms_grids_new.png (219.03 KiB) Viewed 4069 times

mean_innov_2019.png (112.98 KiB) Viewed 4069 times

[Dan_chan]

Dear Ivan,
I am new in ROMS and Data assimilation, so I am sorry for I couldn't solve your problems. Besides, I'd like to ask you some question on your result. As shown in npo_roms_grids_new.png, what did 2 rectangle mean? Do they relate to Nested Grids? How could you realize nested grids in OI. If it comes to your privacy on work, I am very very sorry.
Thank you for your help.
Dan

[ptimko]

Monthly means for HYCOM should be available here:

https://data.hycom.org/datasets/GLBb0.0 ... a/meanstd/

The files are in native HYCOM format but there maybe a utility for converting the files to netcdf format in the source code under utilities or tools, looks like there may be some python code for reading the files at least. I should have source code and matlab scripts that will read HYCOM output in its native format but I don't have a routine for output into netcdf if that's what you need. It might take me awhile to locate the source code and matlab scripts I wrote as it was about 10 years ago that I wrote them and haven't used them recently.

Patrick

[isoares]

Dear Patrick, thank you for replying.

I actually got an answer from Michael McDonald in the Hycom forum. He point me to the link https://www.hycom.org/dataserver/gofs-3pt1/reanalysis where I was able to download the long term mean SSH in NetCDF format.

Best Regard,

Ivan

[isoares]

Dear Dan,

Yes, the boxes are the model domains, and yes, the smaller domain is nested on the the larger domain. The Data Assimilation is working only in the larger domain. If you have any questions please feel free to contact me directly in my email ivan@atlantech.com.br

Best,