Question about radiation boundary condition

[xupeng66]

Hello,

I have a question about the radiation boundary condition that is currently implemented in ROMS. For the example shown in Marchesiello 2001, the authors mention that only the difference between model and climatology is radiated out at the open boundaries. I looked into the latest source code of ROMS and saw that the radiation boundary conditions appear to be applied to the total fields instead of their differences from climatology. If this is correct, I am wondering if radiating out only the differences is a special treatment that does not generally work well in cases other than the one described in that paper.

Thanks!
Guangyu

[kate]

I guess it depends on how well you trust your climatology vs your boundary fields. Things have come a long way since that paper was written in terms of the global model reanalysis products that are available as boundary conditions.

[xupeng66]

Thanks, Kate. I am trying to build a regional simulation using the global reanalysis products from NEMO as open boundary conditions/forcing (i.e., one-way nesting). In this case, should I use the radiation+nudging open boundary conditions and radiate out only the difference between the interior solutions and NEMO results?

[kate]

How will you know until you try it?

[wilkin]

...the authors mention that only the difference between model and climatology is radiated out at the open boundaries.

Where do you see this in Marchesiello et al (2001)?

My reading of it says the radiation calculation is based on the prognostic variables u,v,T,S separately, and the active/passive distinction controls the strength of the nudging to phi^ext. But the exterior/boundary values do not enter the phase speed calculated for radiation.

For the barotropic velocity the FLather condition (Marchesiello's eqn 12) does consider the difference between the interior and exterior solution, but only for this term. And in Flather the radiation speed is set by physics - sqrt(gh).

[xupeng66]

Kate, I agree that the proof of the pudding is in the eating so I will have to try and find out how those open boundary treatments work.

John, thanks for your reply. On page 17, lines 5-7, the authors state that "Perkins et al. (1997) suggest that continuity at the open boundary is improved by radiating out model-data differences rather than absolute model values." The paper does not include any details of how that is implemented numerically but my guess is that 'radiating out model-data differences' is done by substituting those differences for the absolute variables in the oblique OBC formulas. Please let me know if this makes sense to you.

[wilkin]

Yes, Perkins et al. (1997) choose to radiate "Q_inner minus Q_outer" in their nested model application, but this is not what Marchesiello et al. (2001) do and it is not implemented in ROMS.

Radiating the difference between model and climatology - as opposed to the difference in nested grids as in Perkins - would not amount to the same correction step that Perkins advocates.

The Perkins approach emphasizes nested grids, and uses information from the "outer" coarse grid at points inside the nested grid perimeter.

To do this in ROMS you could load the information you have from NEMO reanalysis into "climatology" arrays CLIMA(ng)%tclm - then that would be available to the boundary condition routines (t3dbc_im.F etc.) and you could modify the radiation speed calculation accordingly.

Let us know how it works out.

[xupeng66]

Thanks for explaining, John. I will set up a test run and keep you posted on how it works.