how to implement passive OBC in ROMS?

[leonjld]

I know there are Free Surface OBC,M2 OBC,M3 OBC and Tracer OBC, but I do not know what to specify. So do I basically choose one for every type of OBC (FS,M2,M3 and T?). I don't have any information at OBC, but many OBC need an input file ocean_bry.nc, how can I circumvent this? Is it possible to specify the FOA sheme by Palma and Matano (2000)? The case I am running is a simple wind-driven upwelling situation with one offshore boundary and two cross-shore boundaries. I have successfully run the model using cyclic cross-shore OBC and closed offshore boundary, but now I want to replace the cyclic OBC for other choices. The closed offshore boundary also seems to give me some reflections, although I have already specified "SPONGE" in the cppdef.h. Is there better way to do this? Thanks a lot!

[kate]

When you specify SPONGE, do you also set it up the way you want in ana_hmixcoef.h? Did you try the radiation boundary conditions? I'm not familiar with Palma and Matano, but feel free to try it out yourself. Just be careful to comply with the timestepping scheme used in ROMS. Also, check out Sasha's boundary condition talk from the ROMS meeting last fall (http://www.myroms.org/index.php?page=ROMS_2007_agenda).

So do I basically choose one for every type of OBC (FS,M2,M3 and T?).

Yes.

Note, the combination of radiation, sponge, and nudging to some exterior value that most of us use was arrived at through a lot of trial and error. What are your goals that you want to get away from the periodic boundary condition?

[leonjld]

thanks, kate. I didn't know I need to change the ana.hmixcoef.h when specifying SPONGE. For Sasha's boundary condition talk, are you referring to:

Alexander F. Shchepetkin, IGPP, UCLA (30 min)
10 Years of the ROMS Project: An Overview with Emphasis on what is Overlooked, Overdue, or Missing

My main reason for not to specify cyclic boundary condition is because I want to specify alongshore pressure gradient which is missing in cyclic OBC. Another reason is for cyclic OBC energy won't radiate out of the model domain which I believe results in unreasonble accumulation of energy in the domain.

[kate]

Yes, that is the talk. A few pages of it are about a boundary condition. It's confusing because his code uses a different timestepping than mine, but we all want his results.

I just wanted you to be sure you had a good reason for abandoning periodic BC's. We all wish we could use something so well-behaved.

[nobuhitomori]

I had similar questions for OBC but there was no documant about it. I did simple numerical test with analytical tidal forcing and here is the result:
http://www.oceanwave.jp/research/ROMS/roms_obc.pdf

NM

[leonjld]

Thanks Nobuhito Kun, The memorandum is very helpful to understand the some of the OBC of ROMS. I hope ROMS group can give a more complete memorandum of the OBC in the similar way, and how to set OBC using netcdf file or analytical expressions.

[kate]

Say, Nobuhito, would you like to put that on the wiki? Please send an email to Dave Robertson for access. Or send me the LaTeX source file and I can do it.

[nobuhitomori]

Kate,

I've put LaTeX and figures on
http://www.oceanwave.jp/tmp/roms_obc.tgz
I'm wondering you could modify it to wiki.
Sorry for very late reply.