[MITgcm-support] Open boudary conditions in MITgcm

[Martin Losch]

Hi Vigdis,

I do realize that the open boundary package is probably not the most  
mature part of the code, so that I am not surprised that there seem  
to be so many problems with it (browse the support archives and you  
will find a lot of questions about obcs).

I use obcs with a simple box with four open boundaries and a linear  
free surface (and no topography). I prescribe U,V,T,S along the  
boundaries and yes, Eta=SSH is zero along the boundaries.  
Nevertheless, the velocity field imposes a jet entering the domain  
and I have no problems representing that jet in the SSH (as if the  
Eta-values on the boundary do not matter). I suspect that there are  
other thing that go wrong in your case.

Prescribe the velocity field and make sure that it is balanced:  
integral of divU=0. (where U is the vertically integrated transport)
Make sure that you do not have any topography gradients across the  
boundary. I am not quite sure whether this means that the  
topographies along the south and the north have to be the same (see  
Gianmaria's emails yesterday and the day before yesterday). It may be  
possible that for obcs we will have to extend the topography beyond  
the computational domain. You can try that by increasing your  
computational domain by the width of the overlap region (Oly, it can  
also be any number larger than Oly, just to make shure that you don't  
run into trouble with the overlaps) and specifying the topography in  
this new part of the domain, so that:
j=1 is the first point on your grid
j=1+Oly is the open boundary point
and depth(j=1 to 1+Oly) = const.
and the same for j=Ny, j=Ny-Oly, so that your actual domain ranges  
from 1+Oly to Ny-Oly.
If that works, we may have to change the code to account for that  
problem.

Hope that helps,

Martin
On Mar 1, 2006, at 12:00 PM, Vigdis Tverberg wrote:

>
> I am trying to run the MITgcm for a domain with a barotropic slope  
> current
> entering in the south and exiting in the north. But no matter what  
> open
> boundary condition I use, the routine solve_for_pressure.F sets the  
> open
> boundary value for surface pressure equal to zero. I interpret that  
> the
> variable  cg2d_x is the surface pressure (or surface elevation). A  
> barotropic
> current have to maintained by a sloping surface elevation ,though,  
> so the
> model run crashes quite fast.
>
> Do you have a suggestion to how to solve this, or guide me to  
> someone who is
> working with similar problems?
>
> Best regards,
> Vigdis Tverberg
> Norwegian Polar Institute
> Polarmiljøsenteret
> N-9296 TROMSO
> Norway
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support