Hi

Bug in sbcrnf.F90. Too many rivers (to cross) if nn_fsbc > 1 .....

A simple fix is to put a

IF( MOD( kt-1, nn_fsbc ) == 0 ) THEN <<CODE>> ENDIF

block around the following code:

100	         ! C a u t i o n : runoff is negative and in kg/m2/s
101	
102	         emp (:,:) = emp (:,:) - ABS( sf_rnf(1)%fnow(:,:) )
103	         emps(:,:) = emps(:,:) - ABS( sf_rnf(1)%fnow(:,:) )

cheers Stephane

In the vvl case, and without a flat bottom, the vertical velocity computation in wzvmod is wrong. Check wn at the surface, you will notice that it does not match the emp flux as it should. This is because the mut array, the inverse depth at T-points, is not masked in the following lines:

         DO jk = jpkm1, 1, -1
            wn(:,:,jk) = wn(:,:,jk+1) - fse3t(:,:,jk) * hdivn(:,:,jk) &
              &        - ( zssha(:,:) - sshb(:,:) ) * fsve3t(:,:,jk) * mut(:,:,jk) / z2dt
         END DO

To correct this, multiply mut by tmask ! Jerome

Testing interpolation on the fly (IOF) for some special configurations (ROSS sea, including the south pole)... I found a bug in the allocation of fly_dta array in fldread.F90. This bug causes a crash later on in the run. It is also the case for a PERIANT8 configuration.

So, I suggest to change the following lines in fldread.F90:

         ALLOCATE( ref_wgts(nxt_wgt)%fly_dta(ref_wgts(nxt_wgt)%jpiwgt+2, ref_wgts(nxt_wgt)%jpjwgt+2) )
         IF( ref_wgts(nxt_wgt)%cyclic ) ALLOCATE( ref_wgts(nxt_wgt)%col2(2,ref_wgts(nxt_wgt)%jpjwgt+2) )

by

         ALLOCATE( ref_wgts(nxt_wgt)%fly_dta(ref_wgts(nxt_wgt)%jpiwgt+2, ref_wgts(nxt_wgt)%jpjwgt+3) )
         IF( ref_wgts(nxt_wgt)%cyclic ) ALLOCATE( ref_wgts(nxt_wgt)%col2(2,ref_wgts(nxt_wgt)%jpjwgt+3) )

The difference is in the +3 instead of +2 in the second dimension.

Jean-Marc