[MITgcm-devel] obcs in seaice dynamics

[Menemenlis, Dimitris (3248)]

Jean-Michel, it might be good to add a couple of checks in obcs_check
for folks that are using some of the hacks, e.g., COMPUTE_UVICE and
AVOID_CONVERGENCE.  All that is needed is a warning to define
OBCS_UVICE_OLD if it is not already defined?  That way folks that
use the experimental options and that have custom OBCS_OPTIONS.h,
which doesn't include "#define OBCS_UVICE_OLD" will know why
their code stops working.

I can take care of adding these checks if you like.  Let me know

Dimitris Menemenlis

On Oct 21, 2011, at 12:15 PM, Jean-Michel Campin wrote:

> Hi,
> 
> I have added a new implementation of OBCS in seaice-dyanmics.
> and kept the old one with #define OBCS_UVICE_OLD.
> Will switch seaice_obcs test exp to use the new version soon.
> 
> I did not implement any of the OBCS_SEAICE_* "experimental"
> options in the new version:
> - some may not be really used (SMOOTH_UVICE_PERP, SMOOTH_UVICE_PAR)
> - a COMPUTE_UVICE equivalent would need to be really different 
>  (+would be better in a separated obcs_seaice_calc.F S/R) and I 
>  don't know what it should be.
> - can certainly put back AVOID_CONVERGENCE in obcs_adjust_uvice.F,
>  but one thing I noticed in my simple test was that in the old version,
>  the zero seaiceMaskU,V @ Southern & Western boundaries was
>  slowing down the ice velocity near the boundaries and generating
>  some divergence/convergence. So may be we don't need anymore
>  this AVOID_CONVERGENCE piece of code.
> 
> I also had to put some masking in the water-drag coeff in order
> to get a valid wind-stress at the open-boundary. The wind stress
> at the open-boundary is not used for ocean dynamics but is used
> in vertical mixing scheme (KPP, GGL90, ...), so better to be right.
> 
> And regarding seaice_obcs test exp., it's still not "clean"
> (meaning some value @ Western Boundary can influence the solution
> at the Eastern boundary, and results will change if OBCS_uvApplyFac
> is set to zero), but this is due to KPP and not to seaice.
> 
> Cheers,
> Jean-Michel
> 
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel