[MITgcm-devel] highOrderVorticity & CS-grid

[Jean-Michel Campin]

Hi,

Regarding highOrderVorticity & CS-grid (to the devel-list
so that we keep it somewhere):

I just look to the code, and was surprised to see
(in config_check.F) this:
      IF ( highOrderVorticity .AND. useCubedSphereExchange ) THEN
        WRITE(msgBuf,'(2A)')
     &   'CONFIG_CHECK: highOrderVorticity Scheme does not work',
     &   ' on CubedSphere grid'
        CALL PRINT_ERROR( msgBuf , myThid)
        STOP 'ABNORMAL END: S/R CONFIG_CHECK'
      ENDIF

I put those lines 2 months ago, and trying to remember what the
problem was ...

Well, it does not seem obvious (because each value that is used
is computed correctly, velocity & vorticity), but depending
on which face is considered, the vorticity in the halo region
(near a corner) is not the same:

with cs_32x32, gv(32,33) on face.1 should be equal to gu(1,1) face.3
gv(32,33) face.1 uses Z(34,33)_f1 = Z(2,33)_f2
gu( 1, 1) face.3 uses Z( 1, 0)_f3 = Z(1,32)_f2
and clearly Z(2,33)_f2 is not equal to Z(1,32)_f2

At the time I put the stop, I was checking that the solution
at the edge of 1 face (near a corner) was correct and mach the
value on the next face.

A fix could be to consider the average of the 2,
Z(2,33)_f2 + Z(1,32)_f2 / 2 for both ?

Jean-Michel