[Mitgcm-support] 5lAIM 3d asymmetry

[mitgcm-support at dev.mitgcm.org]

Alistair,

 Think we may have missed something a while back when we 
switched to interpolating for 3d AIM velocity terms. 
For example we map vAim = baryV2P(vVel) and then map back
gv=baryP2V(vt_Pbl), then exch(gV). However, because
vt_Pbl is only defined for tile interior I don't see how
the baryP2V can be correct for multi-tile runs?

e.g

      DO J=1,sNy
       DO I=1,sNx
        JN = J
        IF ( JN .EQ. sNy ) JN = sNy-1
        I3 = sNx*(JN )+I
         Katm = _KD2KA( Klev )
         gV(i,j,kLev,bi,bj) =
     &   gV(i,j,kLev,bi,bj) + 0.5*(VT_PBL(I2,katm)+VT_PBL(I3,katm))
       ENDDO
      ENDDO

      _EXCH_XYZ_R8( gV , myThid)


QUESTION:
 Could this explain the asymmetry of those long 8-node atmos. 
 tests, or were the asymmetries also there with hs94 stuff?

ALSO:
 Am I using the right mailing list?

Chris