[MITgcm-devel] seaice adjoint and EVP

[Martin Losch]

Hi,

I have corrected the bug in seaice_ocean_stress (I am wondering if  
this will fix the problem that Dimitris has with evp. I don't think  
so, but there is always hope o:), and updated lab_sea but not the  
adjoint of that (because I can't).

also I have reduced the number of exchanges in the lsr solver to 2  
exch_uv_3d_rl calls per iteration. (the evp solver has 1 per iteration).
While doing that it occured to me that the lsr solver may actually  
have serious problems along the face edges because it first computes  
u (which is not updated along face edges if u=v), and then v which is  
again not properly updated (if v=-u), so I should probably not use  
that for llc.

As a side note: It is probably time to figure out which exchanges are  
not needed in seaice_model.F. My impression is that none of these

#ifdef SEAICE_EXTERNAL_FORCING
C--   Atmospheric state and runoff are from
C     pkg/exf, which does not update edges.
       CALL EXCH_UV_AGRID_3D_RL( uwind, vwind, .TRUE., 1, myThid )
#ifdef ALLOW_ATM_TEMP
       _EXCH_XY_R8( atemp,  myThid )
       _EXCH_XY_R8( aqh,    myThid )
       _EXCH_XY_R8( precip, myThid )
       _EXCH_XY_R8( evap,   myThid )
#else /* ALLOW_ATM_TEMP */
       STOP 'ABNORMAL END: S/R SEAICE_MODEL: ATM_TEMP undef'
#endif /* ALLOW_ATM_TEMP */
#ifdef ALLOW_DOWNWARD_RADIATION
       _EXCH_XY_R8( lwdown, myThid )
       _EXCH_XY_R8( swdown, mythid )
#else /* ALLOW_DOWNWARD_RADIATION */
       STOP 'ABNORMAL END: S/R SEAICE_MODEL: DOWNWARD_RADIATION undef'
#endif /* ALLOW_DOWNWARD_RADIATION */
       _EXCH_XY_R8( runoff, myThid )
#else /* SEAICE_EXTERNAL_FORCING */

are necessary (as they are exchanged in exf) and could be removed.

Martin

On 15 May 2007, at 16:04, Jean-Michel Campin wrote:

> Hi Martin,
>
>> Jean-Michel, can I do these changes now or would like me to wait a
>> little longer?
>
> Sorry, I forgot to tell that my changes were all done.
> Go for it, please.
>
> Jean-Michel
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel