Changeset 9482 for branches/2017/dev_merge_2017/NEMOGCM/NEMO/NST_SRC/agrif_lim3_interp.F90

Timestamp:

2018-04-13T19:03:31+02:00 (6 years ago)

Author:

clem

Message:

The configuration Agrif_Nordic is now restartable for any nn_fsbc(Parent) and nn_fsbc(Child). However we do not authorized nn_fsbc(Child) to be larger than 1 because otherwise it breaks the CFL too easily. The simulation would still run but the results would not be satisfactory. I still have to verify this point with the last version of the code

File:

• branches/2017/dev_merge_2017/NEMOGCM/NEMO/NST_SRC/agrif_lim3_interp.F90 (modified) (4 diffs)

branches/2017/dev_merge_2017/NEMOGCM/NEMO/NST_SRC/agrif_lim3_interp.F90

@@ -53 +53 @@
       !!-----------------------------------------------------------------------
       !
-      IF( Agrif_Root() .OR. nn_ice==0 )  RETURN   ! do not interpolate if inside Parent domain or if child domain does not have ice
+      IF( Agrif_Root() .OR. nn_ice==0 )  RETURN   ! do not interpolate if inside Parent Grid or if child domain does not have ice
       !
       SELECT CASE( cd_type )
       CASE('U','V')
-         IF( PRESENT( kiter ) ) THEN  ! interpolation at the child sub-time step (only for ice rheology)
+         IF( PRESENT( kiter ) ) THEN  ! interpolation at the child ice sub-time step (only for ice rheology)
             zbeta = ( REAL(lim_nbstep) - REAL(kitermax - kiter) / REAL(kitermax) ) /  &
                &    ( Agrif_Rhot() * REAL(Agrif_Parent(nn_fsbc)) / REAL(nn_fsbc) )
-         ELSE                         ! interpolation at the child time step
+         ELSE                         ! interpolation at the child ice time step
             zbeta = REAL(lim_nbstep) / ( Agrif_Rhot() * REAL(Agrif_Parent(nn_fsbc)) / REAL(nn_fsbc) )
          ENDIF
@@ -204 +204 @@
                !
                DO jk = 1, nlay_s
-                  e_s(i1:i2,j1:j2,jk,jl) = ptab(:,:,jm) * tmask(i1:i2,j1:j2,1)
+                  e_s(i1:i2,j1:j2,jk,jl) = ptab(i1:i2,j1:j2,jm) * tmask(i1:i2,j1:j2,1)
                   jm = jm + 1
                END DO
                !
                DO jk = 1, nlay_i
-                  e_i(i1:i2,j1:j2,jk,jl) = ptab(:,:,jm) * tmask(i1:i2,j1:j2,1)
+                  e_i(i1:i2,j1:j2,jk,jl) = ptab(i1:i2,j1:j2,jm) * tmask(i1:i2,j1:j2,1)
                   jm = jm + 1
                END DO
@@ -216 +216 @@
             !
 !!==> clem: this interpolation does not work because it creates negative values, due
-!           to negative coefficients when mixing points (for ex. z7)
+!!          to negative coefficients when mixing points (for ex. z7)
 !!
 !         ELSE                          ! ==> complex interpolation (only one ghost cell available)
 !            !! Use a more complex interpolation since we mix solutions over a couple of grid points
 !            !! it is advised to use it for fields modified by high order schemes (e.g. advection UM5...)
-!            !! clem: for some reason (I don't know why), the following lines do not work 
-!            !        I think there is an issue with Agrif_SpecialValue here (not taken into account properly)
 !            ! record ztab
 !            jm = 1
@@ -358 +356 @@
 !         ENDIF  ! nbghostcells=1
          
-         ! integrated values
-         vt_i (i1:i2,j1:j2) = SUM(      v_i(i1:i2,j1:j2,:),            dim=3 )
-         vt_s (i1:i2,j1:j2) = SUM(      v_s(i1:i2,j1:j2,:),            dim=3 )
-         at_i (i1:i2,j1:j2) = SUM(      a_i(i1:i2,j1:j2,:),            dim=3 )
-         et_s(i1:i2,j1:j2)  = SUM( SUM( e_s(i1:i2,j1:j2,:,:), dim=4 ), dim=3 )
-         et_i(i1:i2,j1:j2)  = SUM( SUM( e_i(i1:i2,j1:j2,:,:), dim=4 ), dim=3 )
-
-         at_ip(i1:i2,j1:j2) = SUM( a_ip(i1:i2,j1:j2,:), dim=3 ) ! melt ponds
-         vt_ip(i1:i2,j1:j2) = SUM( v_ip(i1:i2,j1:j2,:), dim=3 )
-         !
-         ato_i(i1:i2,j1:j2) = 1._wp - at_i(i1:i2,j1:j2)         ! open water fraction  
-
          DO jl = 1, jpl
             WHERE( tmask(i1:i2,j1:j2,1) == 0._wp )   t_su(i1:i2,j1:j2,jl) = rt0   ! to avoid a division by 0 in sbcblk.F90