[MITgcm-devel] Re: [MITgcm-support] obcs_apply_ptracer

[Martin Losch]

It would be easiest to start from an unproblematic topography in the  
first place, so yes, providing an offline matlab routine would be  
simple.
Initial conditions need not change, as this will not require  
"digging" but only "filling in". It should go like this:
if "depth on boundary > depth one point inside bounday" then depth on  
boundary = depth one point inside boundary
and that's it. In fortan I would have coded it like this (R_low < 0  
for ocean in z-coordinates):
>          DO K=1,Nr
> #ifdef ALLOW_OBCS_NORTH
>           DO I=1,sNx
>            J=OB_Jn(I,bi,bj)
>            IF ( J .NE. 0 .AND.
>      &          ( R_low(I,J,bi,bj) .LT. R_low(I,J-1,bi,bj) ) ) THEN
>             ichanged = ichanged + 1
>             R_low(I,J,bi,bj) = R_low(I,J-1,K,bi,bj)
>            ENDIF
>           ENDDO
> #endif
>       ENDDO

How and where would you implement this matlab routine?

Martin

On 14 Apr 2008, at 15:32, Dimitris Menemenlis wrote:
> Martin, thank you for catching the staggertimestep problem.  I will  
> fix.
> This will require modifying:
> MITgcm/verification/lab_sea/results/output.salt_plume.txt
> MITgcm/verification/seaice_obcs/results/output.txt
>
>> PS. it turns out, that it is not trivial to correct the topography  
>> along the
>> boundaries, because the obcs information is not yet available at  
>> the time
>> r_low is read. This is a problem as it is probably dangerous to  
>> change hfacc
>> after this grid initialisation phase, isn't it? I would need to call
>> obcs_init_fixed  (or at least the first part where OB_Jn/s, and  
>> OB_Ie/w are
>> set) before ini_depth (or from within ini_depth).
>
> Would it be easier to provide a matlab routine that takes care of  
> this digging offline?  Also, will this digging require sensible  
> initial T/S/ETAN values, or are they not needed because they are  
> outside the open boundaries?
>
> D.
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel