[MITgcm-devel] [MITgcm-cvs] MITgcm/pkg/seaice CVS Commit

Fri Nov 5 02:59:30 EDT 2010

Jean-Michel, Oliver has already contacted me about the exch2 issue.
In addition, the mod that I checked in is not thread-safe.  So it definitely
needs to be changed.

The issue I am trying to address is following.  Sea ice open boundary
conditions are problematic because there can be a mismatch between
the ice velocity that is imposed at the boundaries and that which is
computed by the sea ice dynamic solver.  This can happen, for example, 
when specifying monthly mean sea ice boundary conditions because of
higher-frequency sea ice variability or when interpolating sea ice OBCS
spatially from a lower-resoluton integration.

When there is convergence, ice piles up and cause model to crash.
This is what flag OBCS_SEAICE_AVOID_CONVERGENCE
takes care of by allowing ice to exit the domain of integration
when there is convergence at the edges.

Divergence at the edges can also be problematic.  For example,
Michael is integrating a hi-res Pine Island Glacier domain with OBCS
from a CS510 integration.  Ice divergence at the edge causes
ice formation -> salt rejection -> convection -> warmer water to rise to
surface.  It looks ugly and eventually causes integration to blow up.

My attempted solution was to (i) put the open boundaries 4 grid point away
from the edge as you suggested in your October 28 email and
(ii) "not" to specify uice or vice at the edges but rather to
use that which is computed by ice dynamic solver which extends a few
point outside the OBCS.  This did not work as expected.  There are
some sea ice artifacts in the region between the OBCS and the edge
of the integration domain, which cause integration to crash.

So I am thinking that I should revert back to original OBCS mask in
seaice_init_varia.F and to implement instead a relaxation region for
sea ice tracer quantities similar to rbcs.

Suggestions?

Dimitris

On Nov 4, 2010, at 7:41 PM, Jean-Michel Campin wrote:

> Hi Dimitris,
> 
> I was going to make my changes to prevent "links" of
> tracers (T, S, pTraver or seaice thermodynamics variables)
> when using high order Adv.Scheme with OBCS (and when no
> opposite land barrier prevents EXCH to transfert tracers
> values). 
> 
> I started to look at seaice_init_varia.F (because it's related to exch)
> but I don't undestand your latest changes (I know very little
> about seaice dynsolver too, so may be it's the reason). In particular, 
> 1) It does not seems compatible with Oliver's recent changes to make 
> OBCS works with exch2; from tag-index, on Oct 13:
>> o pkg/obcs: add support for exch2
>>  - the position of the boundary and prescribed values are specified using
>>    a global domain with exch2 facets stacked
>>    - in x for N,S boundaries (like W2_mapIO=-1)
>>    - in y for E,W boundaries (so E,W boundaries do not overlap)
> 2) without exch2, I have some concern when the OB are further inside the 
> domain (OB_Iw > 1, e.g., OB_Iw =4 ): what will be the effect of setting
> those masks to zero just for i=1 ?
> 
> Thanks,
> Jean-Michel
> 
> On Tue, Nov 02, 2010 at 05:23:50PM -0400, Dimitris Menemenlis wrote:
>> Update of /u/gcmpack/MITgcm/pkg/seaice
>> In directory forge:/tmp/cvs-serv22627/pkg/seaice
>> 
>> Modified Files:
>> 	seaice_init_varia.F 
>> Log Message:
>> For OBCS, close southern and western boundaries for sea ice solver
>> based on myYGlobalLo and myXGlobalLo instead of OB_Js and OB_Iw.
>> 
>> 
>> _______________________________________________
>> MITgcm-cvs mailing list
>> MITgcm-cvs at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-cvs
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel