Check useless lbc_lnk in icb code

[mathiot]

Context

After a grep on lbc_lnk (by curiousity), I found some lbc_lnk I don't really understood why they are needed.
After checking which interpolated variable at icb location are used, I think some are useless (8/16 for icbthd.F90 and 2/16 for icbdyn.F90).

Proposal

• Build an idealised test case to test the iceberg code (GYRE like with a lot of calving point)
• Check if these lbc_lnk / interpolation are useful (block below are simply as reminder of where these calls exactly are):

1 : I am no convinced these 2 call of lbc_lnk. I don't understand why stored_ice need to be different to 0 over the haloes.

src/OCE/ICB/icbclv.F90

@@ -172 +172 @@
       END DO
       !
       DO jn = 1, nclasses
-         CALL lbc_lnk( 'icbclv', berg_grid%stored_ice(:,:,jn), 'T', 1._wp )
+         CALL lbc_lnk( 'icbclv', berg_grid%stored_ice(:,:,jn), 'T', 1._wp )    ! useless PM ?
       END DO
-      CALL lbc_lnk( 'icbclv', berg_grid%stored_heat, 'T', 1._wp )
+      CALL lbc_lnk( 'icbclv', berg_grid%stored_heat, 'T', 1._wp )              ! useless PM ?
       !
       IF( nn_verbose_level > 0 .AND. icntmax > 1 )   WRITE(numicb,*) 'icb_clv: icnt=', icnt,' on', narea
       !

2 : zsst and zcn are computed but not used => need to find an easy to use solution in case more variables are used in the future (grounding scheme, new sea-ice/iceberg drag formulation ...)

src/OCE/ICB/icbdyn.F90

@@ -269 +269 @@
 
       ! Interpolate gridded fields to berg
       nknberg = berg%number(1)
-      CALL icb_utl_interp( pxi, pe1, zuo, zui, zua, zssh_x,                     &
-         &                 pyj, pe2, zvo, zvi, zva, zssh_y, zsst, zcn, zhi, zff )
+      CALL icb_utl_interp( pxi, pe1, zuo, zui, zua, zssh_x,                     & ! zsst, zcn useless PM ?
+         &                 pyj, pe2, zvo, zvi, zva, zssh_y, zsst, zcn, zhi, zff ) 
 
       zM = berg%current_point%mass
       zT = berg%current_point%thickness               ! total thickness

3 : pt%e1, pt%e2, pt%ssh_x, pt%ssh_y, zff, pt%ui, pt%vi, pt%hi are computed but not used => need to find an easy to use solution in case more variables are used in the future (new melting formulations).

src/OCE/ICB/icbthm.F90

@@ -82 +82 @@
          !
          pt => this%current_point
          nknberg = this%number(1)
-         CALL icb_utl_interp( pt%xi, pt%e1, pt%uo, pt%ui, pt%ua, pt%ssh_x,   &
+         CALL icb_utl_interp( pt%xi, pt%e1, pt%uo, pt%ui, pt%ua, pt%ssh_x,   &   ! pt%e1, pt%e2, pt%ssh_x, pt%ssh_y, zff, pt%ui, pt%vi, pt%hi useless PM ?
             &                 pt%yj, pt%e2, pt%vo, pt%vi, pt%va, pt%ssh_y,   &
             &                 pt%sst, pt%cn, pt%hi, zff )
          !

4 : fr_e, tt_e and hi_e seems to be used only for the thermodynamic (1d with no icb over the haloes). So probably these 3 lbc_lnk call are useless.

src/OCE/ICB/icbutl.F90

@@ -82 +82 @@
       CALL lbc_lnk_icb( 'icbutl', ff_e, 'F', +1._wp, 1, 1 )
       CALL lbc_lnk_icb( 'icbutl', ua_e, 'U', -1._wp, 1, 1 )
       CALL lbc_lnk_icb( 'icbutl', va_e, 'V', -1._wp, 1, 1 )
-      CALL lbc_lnk_icb( 'icbutl', fr_e, 'T', +1._wp, 1, 1 )
-      CALL lbc_lnk_icb( 'icbutl', tt_e, 'T', +1._wp, 1, 1 )
+      CALL lbc_lnk_icb( 'icbutl', fr_e, 'T', +1._wp, 1, 1 ) ! useless PM ?
+      CALL lbc_lnk_icb( 'icbutl', tt_e, 'T', +1._wp, 1, 1 ) ! useless PM ?
 #if defined key_si3
       hi_e(1:jpi, 1:jpj) = hm_i (:,:)  
       ui_e(1:jpi, 1:jpj) = u_ice(:,:)
@@ -93 +93 @@
       zssh_lead_m(:,:) = ice_var_sshdyn(ssh_m, snwice_mass, snwice_mass_b)
       ssh_e(1:jpi, 1:jpj) = zssh_lead_m(:,:) * tmask(:,:,1)
       !
-      CALL lbc_lnk_icb( 'icbutl', hi_e , 'T', +1._wp, 1, 1 )
+      CALL lbc_lnk_icb( 'icbutl', hi_e , 'T', +1._wp, 1, 1 ) ! useless PM ?
       CALL lbc_lnk_icb( 'icbutl', ui_e , 'U', -1._wp, 1, 1 )
       CALL lbc_lnk_icb( 'icbutl', vi_e , 'V', -1._wp, 1, 1 )
 #else

[mathiot]

• To keep in mind, every nn_sample time step icb_trj_write, icb%* need to be fully defined.

[mathiot]

In 13243:

Error: Failed to load processor CommitTicketReference

No macro or processor named 'CommitTicketReference' found

[mathiot]

In 13265:

Error: Failed to load processor CommitTicketReference

No macro or processor named 'CommitTicketReference' found

[mathiot]

In 13273:

Error: Failed to load processor CommitTicketReference

No macro or processor named 'CommitTicketReference' found

[mathiot]

In 13276:

Error: Failed to load processor CommitTicketReference

No macro or processor named 'CommitTicketReference' found

[mathiot]

Done as part of commit 14030 (merge end 2020)