Custom Query (964 matches)

The T and S advection schemes are read in read_phypar.F and ThreeGhostPoints is then set accordingly. It isn't until much later in inp_par.F that the biology, etc. advection schemes are read. If I have the old standard U3 for T and S and something fancier for biology, ThreeGhostPoints is not being set.

The logical switch Lmpdata is located in the wrong place in ad_step3d_t.F, tl_step3d_t.F, and rp_step3d_t.F:

!
!-----------------------------------------------------------------------
!  Time-step horizontal advection term.
!-----------------------------------------------------------------------
!
      Lmpdata=ANY(tl_Hadvection(:,ng)%MPDATA).and.                      &
     &        ANY(tl_Vadvection(:,ng)%MPDATA)

It needs to be located outside of the AD_SUPPORTED, TL_SUPPORTED, and RP_SUPPORTED directives, repectively.

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

Fixed couple of issues in the weak constraint 4D-Var algorithm:

• Removed some legacy code in routines forcing.F and tl_forcing.F when processing the impulse forcing for the weak constraint 4D-Var formulation. The legacy code that included the forcing for ubar and vbar in 3D solutions were removed from forcing.F when SOLVE3D is activated. Similarly, the forcing for tl_ubar and tl_vbar was removed from tl_forcing.F.

• Added logic to def_impulse.F, get_data.F, set_data.F, tl_get_data.F, tl_set_data.F, and checkvars.F to avoid processing the weak constraint impulse forcing for ubar (tl_ubar) and vbar (tl_vbar) when SOLVE3D is activated.

Many thanks to Andy Moore for his help in tracking this bug.