Important Corrections

[arango]

This update contains several minor corrections:

• check_multifile.F: Added the source/sink files (SSF structure) for multifile management. Splitting the SSF input data into multiple files is now possible. Also, corrected the forcing data time range checking in the adjoint model. We need to have instead:

          IF (Lcheck) THEN
            IF (model.eq.iADM) THEN
              Tmin=Tscale*FRC(i,ng)%time_min(1)
              ...
            END IF
          END IF
  

  Many thanks to Julia Levin for bringing this problem to my attention.

• qet_cycle.F: Replaced the multifile LowerBound logic with an ELSE statement to also allow computing the LOWER time-snapshot in the time cycling case:

  !
  !  If not processing a multi-file field, find LOWER time-snapshot
  !  and its associated time record (Tindex).
  !
        ELSE
          IF (job.gt.0) THEN     ! forward:   Tval(i) =< mday =< Tval(i+1)
            IF (Linside) THEN
              tstart=Tmin
              DO i=2,ntime
                IF ((tstart.le.mday).and.(mday.le.Tval(i))) THEN
                  Tstr=tstart
                  Tindex=i-1     ! one is added when processing
                  EXIT
                END IF
                tstart=Tval(i)
              END DO
            ELSE
              Tstr=Tmax          ! LowerBound for next multifile or
              Tindex=ntime       ! time cycling
            END IF
          ELSE
            ...
          ENDIF
        ENDIF
  

  Many thanks to Dezhou Yang for ​reporting this issue.

• read_asspar.F: Corrected loading of 4D-Var parameters HdecayB and VdecayB. The order of this input values in input script s4dvar.in is (4,1:NT,Ngrids):

             CASE ('HdecayB(isTvar)')
               Npts=load_r(Nval, Rval, 4*MT*Ngrids, Rboundary)
               DO ng=1,Ngrids
                 DO itrc=1,NT(ng)
                   DO ib=1,4
                     HdecayB(isTvar(itrc),ib,ng)=Rboundary(ib,itrc,ng)
                   END DO
                 END DO
               END DO
               ...
             CASE ('VdecayB(isTvar)')
               Npts=load_r(Nval, Rval, 4*MT*Ngrids, Rboundary)
               DO ng=1,Ngrids
                 DO itrc=1,NT(ng)
                   DO ib=1,4
                     VdecayB(isTvar(itrc),ib,ng)=Rboundary(ib,itrc,ng)
                   END DO
                 END DO
               END DO
  

  The documentation in s4dvar.in indicates that the tracer boundary decorrelation scales need to be specified in the (1:4,1:NT,1:Ngrids) to facilitate having different values for each tracer. For example for Ngrids=1 and NT=2, we can have:

  ! Boundary conditions error covariance: horizontal, isotropic decorrelation
  ! scales (m). A value is expected for each boundary edge in the following
  ! order:
  !                  1: west  2: south  3: east  4: north
  
  HdecayB(isFsur) == 40.0d+3  40.0d+3   40.0d+3  40.0d+3  ! free-surface
  HdecayB(isUbar) == 40.0d+3  40.0d+3   40.0d+3  40.0d+3  ! 2D U-momentum
  HdecayB(isVbar) == 40.0d+3  40.0d+3   40.0d+3  40.0d+3  ! 2D V-momentum
  HdecayB(isUvel) == 40.0d+3  40.0d+3   40.0d+3  40.0d+3  ! 3D U-momentum
  HdecayB(isVvel) == 40.0d+3  40.0d+3   40.0d+3  40.0d+3  ! 3D V-momentum
  HdecayB(isTvar) == 4*40.0d+3  4*15.0d+3                 ! (4,1:NT) tracers
  
  ! Boundary conditions error covariance: vertical, isotropic decorrelation
  ! scales (m). A value is expected for each boundary edge in the following
  ! order:
  !                 1: west  2: south  3: east  4: north
  
  VdecayB(isUvel) == 10.0d0  10.0d0  10.0d0  10.0d0       ! 3D U-momentum
  VdecayB(isVvel) == 10.0d0  10.0d0  10.0d0  10.0d0       ! 3D V-momentum
  VdecayB(isTvar) == 4*10.d0  4*20.d0                     ! (4,1:NT) tracers
  

  It allows to use the compact notation to specify different values for temperature and salinity.

• Added new makefile configurations files Linux-ftn-cray.mk and Linux-ftn-intel.mk. Many thnaks to Mark Hadfield for providing us these two new compiling files.