[MITgcm-support] obcs/exf interactions
Patrick Heimbach
heimbach at MIT.EDU
Wed Oct 6 17:42:12 EDT 2004
Hi Alistair
several things:
the old ecco_... very likely had missing exchanges
in case you'd prescribe obcs.
As you may remember we dropped adding xz-/yz-exchanges
since there were becoming too many of them.
Instead, full fields should be exchanged once obcs are applied.
That's particularly true for the initialisation part of obcs.
These exchanges are not incorporated in any ecco-branch checkpoint.
They're in the latest version of the MAIN branch though.
Another issue with the old version was incorrect mdsreadfield
for xz-/yz- fields so that you needed to pre-tile them.
That's fixed too in the latest.
Nevertheless, the pattern still looks weird to me,
I think, obcs/exf related bugs tended to look different.
How about dropping your merged code and take a fresh
copy of the HEAD of the main branch?
-Patrick
On Wednesday 06 October 2004 16:20, chris hill wrote:
> I haven't seen anything like this before, but I would guess that it is
> connected with OBCS. Would it be easy to do a simple perturbation
> experiment starting from a stable stratification with no exf?
>
> Chris
>
> On Wed, 2004-10-06 at 14:27, Alistair Adcroft wrote:
> > Resending msg:
> >
> > I've mostly sorted out this upgrade of Tom's experiment from
> > ecco_c50_e29 to HEAD. The last problem has me puzzled - the attached
> > figure is of (U,V) and the difference between the two versions of the
> > model after just two steps using the same forcing/parameters.
> >
> > The straight line is hard to see in the original fields and I can't
> > decide whether it comes from Tom's older code or is in the new model.
> > The line does not line up with tile boundaries (black grid in last
> > panel). The best I can determine is that the difference is something to
> > do with exf prescribing the OBC values. Various permutations of code
> > led me to this conclusion but I do not know which is right; I can only
> > see the line when I do a difference of outputs.
> >
> > Anyone else seen a problem like this?
> >
> > A.
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://dev.mitgcm.org/mailman/listinfo/mitgcm-support
--
----------------------------------------------------------
Patrick Heimbach Massachusetts Institute of Technology
FON: +1/617/253-5259 EAPS, Room 54-1518
FAX: +1/617/253-4464 77 Massachusetts Avenue
mailto:heimbach at mit.edu Cambridge MA 02139
http://www.mit.edu/~heimbach/ USA
More information about the MITgcm-support
mailing list