Bug in sbc_cpl_init when running with no ice model

[charris]

This line of code

CASE( 'conservative'  )   ;   srcv( (/jpr_rain, jpr_snow, jpr_ievp, jpr_tevp/) )%laction = .TRUE.

is wrong if running without an ice model (jpr_ievp should not be included in that case). If no-one objects I will put a check based on k_ice in tomorrow.

[smasson]

but if you have no ice model there is no point to do use the conservative option (which is used to conserves the total flux aver cells that ar partially covered by sea ice). You should simply use the "oce only" option.

[charris]

I agree it is not really required, but the 'conservative' option is currently available in the no ice case:

      !                                                      ! ========================= !
      IF( k_ice <= 1 ) THEN                                  !  heat & freshwater fluxes ! (Ocean only case)
         !                                                   ! ========================= !
         !
         !                                                       ! total freshwater fluxes over the ocean (emp)
         SELECT CASE( TRIM( sn_rcv_emp%cldes ) )                                    ! evaporation - precipitation
         CASE( 'conservative' )
            emp(:,:) = frcv(jpr_tevp)%z3(:,:,1) - ( frcv(jpr_rain)%z3(:,:,1) + frcv(jpr_snow)%z3(:,:,1) )
         CASE( 'oce only', 'oce and ice' )
            emp(:,:) = frcv(jpr_oemp)%z3(:,:,1)
         CASE default
            CALL ctl_stop( 'sbc_cpl_rcv: wrong definition of sn_rcv_emp%cldes' )
         END SELECT

For the purposes of minimising coupling and atmosphere changes when switching the ice model on and off it would be quite useful to be able to make use of this.

[smasson]

ok
you could also redefine sn_rcv_emp%cldes to "oce only" if nn_ice == 0

[smasson]

I was just suggesting to overwrite the definition of cldes if you have no ice model, for example:
instead of

      REWIND( numnam )                    ! ... read namlist namsbc_cpl
      READ  ( numnam, namsbc_cpl )

you do

      REWIND( numnam )                    ! ... read namlist namsbc_cpl
      READ  ( numnam, namsbc_cpl )

      IF( nn_ice == 0 ) THEN
         sn_rcv_emp%cldes  = 'oce only'
         sn_rcv_qns%cldes  = 'oce only'
         sn_rcv_qsr%cldes  = 'oce only'
         sn_snd_crt%cldes  = 'oce only'
         sn_snd_temp%cldes = 'oce only'
      ENDIF

[charris]

But that doesn't help people (like me) who still want the option to make use of the code in sbc_cpl_rcv which allows you to receive evaporation and precip separately even in the no ice case.

[charris]

Following up on this again, we still want to be able to receive rain and snow separately when running without ice (to avoid the need to modify other aspects of the coupling etc when turning the ice model on / off). And given the code in sbc_cpl_rcv is set up to do this surely the code in sbc_cpl_init should be consistent. So I'm still proposing just altering

      SELECT CASE( TRIM( sn_rcv_emp%cldes ) )
      CASE( 'oce only'      )   ;   srcv(                                 jpr_oemp   )%laction = .TRUE. 
      CASE( 'conservative'  )   ;   srcv( (/jpr_rain, jpr_snow, jpr_ievp, jpr_tevp/) )%laction = .TRUE.
      CASE( 'oce and ice'   )   ;   srcv( (/jpr_ievp, jpr_sbpr, jpr_semp, jpr_oemp/) )%laction = .TRUE.
      CASE default              ;   CALL ctl_stop( 'sbc_cpl_init: wrong definition of sn_rcv_emp%cldes' )
      END SELECT

to

      SELECT CASE( TRIM( sn_rcv_emp%cldes ) )
      CASE( 'oce only'      )   ;   srcv(                                 jpr_oemp   )%laction = .TRUE. 
      CASE( 'conservative'  )
         srcv( (/jpr_rain, jpr_snow, jpr_ievp, jpr_tevp/) )%laction = .TRUE.
         IF ( k_ice < 1 )  srcv(jpr_ivep)%laction = .FALSE.
      CASE( 'oce and ice'   )   ;   srcv( (/jpr_ievp, jpr_sbpr, jpr_semp, jpr_oemp/) )%laction = .TRUE.
      CASE default              ;   CALL ctl_stop( 'sbc_cpl_init: wrong definition of sn_rcv_emp%cldes' )
      END SELECT

unless there are objections in the next week.

[smasson]

ok for me. please proceed.
seb

[charris]

Change as agreed now on trunk in [4068] and 3.4_stable in [4069].