Bug domcfg with isf

[mathiot]

Context

Under some condition domisf leads to negative vertical scale factor: case where isolated grid point in bathymetry where ice shelf almost grounded.

Analysis

This comes from the position of the modification of the bathy because of isolated grid point in the bathymetry.
It has been included before the computation of misfdep (zgr_isf_kspace) but it should be done at the very beginning in zgr_isf_zspace when compatibility between isf draft and bathy is done.

Fix

in domisf.F90, move the bloc:

      ! 0.0 fill isolated grid point in the bathymetry
      ! will be done again later on in zgr_bat_ctl, but need to be done here to adjust risfdep/misfdep respectively
      icompt = 0 
      DO jj = 2, jpjm1
         DO ji = 2, jpim1
            ibtest = MAX(  mbathy(ji-1,jj), mbathy(ji+1,jj),   &   
               &           mbathy(ji,jj-1), mbathy(ji,jj+1)  )
               IF( ibtest < mbathy(ji,jj) ) THEN
                  mbathy(ji,jj) = ibtest
                  bathy(ji,jj)  = gdepw_1d(ibtest+1)
                  icompt = icompt + 1 
               END IF
         END DO
      END DO
      !   
      ! ensure halo correct 
      zdummy(:,:) = FLOAT( mbathy(:,:) ) ; CALL lbc_lnk('domisf', zdummy, 'T', 1._wp ) ; mbathy(:,:) = INT( zdummy(:,:) )
      !   
      IF( lk_mpp )   CALL mpp_sum('domisf', icompt )
      IF( icompt == 0 ) THEN
         IF(lwp) WRITE(numout,*)'     no isolated ocean grid points'
      ELSE
         IF(lwp) WRITE(numout,*)'    ',icompt,' ocean grid points suppressed'
      ENDIF

out of zgr_isf_kspace in zgr_isf_zspace at the top as 0.0 to be sure mbathy and bathy not modify before starting to compute the isfdraft.

And in domzgr, move the call to zgr_isf_zspace after the computation of mbathy and before the call to zgr_isf_kspace:

domzgr.F90

@@ -894 +894 @@
       IF(lwp) WRITE(numout,*) '    ~~~~~~~ '
       IF(lwp) WRITE(numout,*) '              mbathy is recomputed : bathy_level file is NOT used'
 
-      ! compute position of the ice shelf grounding line
-      ! set bathy and isfdraft to 0 where grounded
-      IF ( ln_isfcav ) CALL zgr_isf_zspace
-
       ! bathymetry in level (from bathy_meter)
       ! ===================
       zmax = gdepw_1d(jpk) + e3t_1d(jpk)        ! maximum depth (i.e. the last ocean level thickness <= 2*e3t_1d(jpkm1) )
@@ -915 +911 @@
          WHERE( 0._wp < bathy(:,:) .AND. bathy(:,:) <= zdepth )   mbathy(:,:) = jk-1
       END DO
 
-      ! Check compatibility between bathy and iceshelf draft
-      ! insure at least 2 wet level on the vertical under an ice shelf
-      ! compute misfdep and adjust isf draft if needed
-      IF ( ln_isfcav ) CALL zgr_isf_kspace
+      ! if under ice shelf cavities, compute misfdep for ocean point (i.e. the top level)
+      IF ( ln_isfcav ) THEN
+         ! compute position of the ice shelf grounding line
+         CALL zgr_isf_zspace
+         ! compute misfdep and adjust isf draft if needed
+         CALL zgr_isf_kspace
+      END IF

[mathiot]

In 12869:

Error: Failed to load processor CommitTicketReference

No macro or processor named 'CommitTicketReference' found

[mathiot]

In 12871:

Error: Failed to load processor CommitTicketReference

No macro or processor named 'CommitTicketReference' found

[mathiot]

As I was not able to reproduced the bug described by Clement B. (Mercator), I will not push this. I send an email to Clement to send me the information needed to reproduce the bug. I am waiting feedback.