[MITgcm-devel] Re: arctic with thsice

[Martin Losch]

Dimitris,
I am moving this to the devel-list:
if the stripes persist, (in spite of A. updating the code and B.  
using SEAICE_maskRHS=.true.), please try uncommenting the lines with  
seaice_sigma12 in the following part of seaice_evp.F
>           DO J=1,sNy
>            DO I=1,sNx
>             stressDivergenceX(I,J,bi,bj) =
>      &           ( sig11(I  ,J  ) * _dyF(I  ,J,bi,bj)
>      &           - sig11(I-1,J  ) * _dyF(I-1,J,bi,bj)
>      &           + seaice_sigma12(I,J+1,bi,bj) * _dxV(I,J+1,bi,bj)
>      &           - seaice_sigma12(I,J  ,bi,bj) * _dxV(I,J  ,bi,bj)
>      &           ) * recip_rAw(I,J,bi,bj)
>      &           -
>      &           ( seaice_sigma12(I,J  ,bi,bj)
>      &           + seaice_sigma12(I,J+1,bi,bj) )
>      &           * _tanPhiAtU(I,J,bi,bj) * recip_rSphere
>      &           +
>      &           ( sig22(I,J) + sig22(I-1,J) ) * 0.5 _d 0
>      &           * _tanPhiAtU(I,J,bi,bj) * recip_rSphere
> C     one metric term  missing for general curvilinear coordinates
>             stressDivergenceY(I,J,bi,bj) =
>      &           ( sig22(I,J  ) * _dxF(I,J  ,bi,bj)
>      &           - sig22(I,J-1) * _dxF(I,J-1,bi,bj)
>      &           + seaice_sigma12(I+1,J,bi,bj) * _dyU(I+1,J,bi,bj)
>      &           - seaice_sigma12(I  ,J,bi,bj) * _dyU(I  ,J,bi,bj)
>      &           ) * recip_rAs(I,J,bi,bj)
>      &           -
>      &           ( sig22(I,J) + sig22(I,J-1) ) * 0.5 _d 0
>      &           * _tanPhiAtV(I,J,bi,bj) * recip_rSphere
> C     two metric terms missing for general curvilinear coordinates
>            ENDDO
>           ENDDO
and/or the HEFFM in
C     first step stress equations
           DO j=0,sNy+1
            DO i=0,sNx+1
C     sigma1 and sigma2 are computed on C points
             seaice_sigma1 (I,J,bi,bj) = ( seaice_sigma1 (I,J,bi,bj)
      &           + SEAICE_deltaTevp *  0.5 * recip_evp_tau
      &           * PRESS(I,J,bi,bj) * ( seaice_div(I,J,bi,bj)-1. _d 0 )
      &           )/(1. + 0.5*SEAICE_deltaTevp*recip_evp_tau)
      &           *hEffM(I,J,bi,bj)
             seaice_sigma2 (I,J,bi,bj) = ( seaice_sigma2 (I,J,bi,bj)
      &           + SEAICE_deltaTevp * 0.5 * recip_evp_tau
      &           * PRESS(I,J,bi,bj) * seaice_tension(I,J,bi,bj)
      &           )/(1. + 0.5*SEAICE_deltaTevp*recip_evp_tau*ecc2)
      &         *hEffM(I,J,bi,bj)
C     sigma12 is computed on Z points
             seaice_sigma12(I,J,bi,bj) = ( seaice_sigma12(I,J,bi,bj)
      &           + SEAICE_deltaTevp * 0.25 * recip_evp_tau *
      &           0.25*( PRESS(I,J  ,bi,bj) + PRESS(I-1,J  ,bi,bj)
      &                + PRESS(I,J-1,bi,bj) + PRESS(I-1,J-1,bi,bj) )
      &           * seaice_shear(I,J,bi,bj)
      &           )/(1. + 0.5*SEAICE_deltaTevp*recip_evp_tau*ecc2)
      &           *hEffM(I,J  ,bi,bj)*hEffM(I-1,J  ,bi,bj)
      &           *hEffM(I,J-1,bi,bj)*hEffM(I-1,J-1,bi,bj)
C     recover sigma11 and sigma22
             sig11(I,J) = 0.5 *
      &           ( seaice_sigma1(I,J,bi,bj)+seaice_sigma2(I,J,bi,bj) )
             sig22(I,J) = 0.5 *
      &           ( seaice_sigma1(I,J,bi,bj)-seaice_sigma2(I,J,bi,bj) )
            ENDDO
           ENDDO

or maybe even save seaice_sigma1/2/12 (they are part of the dumpFreq  
output). That would help with the debugging.

Martin

On 25 Apr 2007, at 19:35, Dimitris Menemenlis wrote:

> Here is comparison of evp and lsr sea ice after four year  
> integration of Arctic domain.
>
> There are stripes in evp ice velocity over open water.
>
> I noticed you checked in a bug fix yesterday.  Was that for evp?   
> Should I rerun?
>
> D.
>
> -- 
> Dimitris Menemenlis <menemenlis at jpl.nasa.gov>
> Jet Propulsion Lab, California Institute of Technology
> MS 300-323, 4800 Oak Grove Dr, Pasadena CA 91109-8099
> tel: 818-354-1656; cell: 818-625-6498; fax: 818-393-6720
> <evp_vs_lsr.ps.gz>