Imposing sea level time series at boundaries

[mdottori]

The question is simple, but I cannot find where this is explained. I need to input a time series of sea level at the southern boundary and don't know how to do so. Anyone?

[kate]

Sea level is not one of the primary variables in the model but is rather updated from horizontal divergences. It is therefore really tricky (impossible?) to simply impose it as a boundary condition. For tides, people use a combination of Chapman/Flather boundary conditions, with reduced physics if you don't have the barotropic flow through that boundary. I personally use a tidal model which comes with barotropic flows for my boundary condition.

[julian.kuhlmann]

So does that mean the Chapman/Flather condition only makes sense for tides? Why? And what would be better for non-tidal sea level and barotropic momentum?

[mdottori]

I just saw a file with a variable called zeta_south, and similar ones for the other boundaries. I wonder if that's not the variable to input a changing sea level condition at the boundary. Besides, why is it possible to input tides but not other changes in sea level at the boundaries?

Sea level is not one of the primary variables in the model but is rather updated from horizontal divergences. It is therefore really tricky (impossible?) to simply impose it as a boundary condition. For tides, people use a combination of Chapman/Flather boundary conditions, with reduced physics if you don't have the barotropic flow through that boundary. I personally use a tidal model which comes with barotropic flows for my boundary condition.

[julian.kuhlmann]

Does anybody have further suggestions on what kind of zeta/ubar/vbar boundary conditions to use in non-tidal applications?

[kate]

We use Chapman/Flather for all frequencies of boundary conditions. This requires both the outside zeta and the ubar/vbar through the boundary.

[tony1230]

Hi kate,

We use Chapman/Flather for all frequencies of boundary conditions. This requires both the outside zeta and the ubar/vbar through the boundary.

Then, what would be the suggestion if we have only zeta but without ubar/vbar?


-shou

[julian.kuhlmann]

Thank you, Kate! Then I just misread your answer. I thought when you said "it is tricky/impossible to simply impose zeta" you meant we shouldn't use it at all, but now I gather that simply imposing would mean the option Clamped, but we should use Chapman and Flather instead.

[kate]

Yes, it's clamped zeta that's not going to be telling the model anything about what to do for ubar/vbar at the boundary. Some people get by with the reduced physics option, estimating the velocity from other fields, so you might as well try that if you don't have the ubar/vbar.

[aurelie.rivier]

Hello,
I'm wondering how velocity is estimated from other fields with the reduced physics option.
Thanks you very much.

[kate]

I was going to suggest that you see for yourself by searching the code for FSOBC_REDUCED, but I find that it has been in part replaced by the "reduced" option for LBC in code such as ROMS/Nonlinear/u2dbc_im.F. Terms there are pressure gradient, Coriolis, and top and bottom stresses.