[Mitgcm-support] MITgcmUV checkpoint33

[mitgcm-support at dev.mitgcm.org]

A new checkpoint has been made of models/MITgcmUV tagged "checkpoint32".

Tar-balls are available at http://mitgcm.lcs.mit.edu/download/

I'm making this checkpoint since Patrick left for Germany before we
could finish these changes. The corresponding ECCO changes/checkpoint
will have to wait until he returns.

Changes since last checkpoint (c32):

1. Bug fix for implicit diffusion  (thanks to Chris Edwards)
   The lower boundary conditions over finite height topography
   have been incorrect since very early on. The fix involves
   just one routine; impldiff.F

2. Units of the primary forcing arrays have been restored to their
   original units. The arrays have the following units:

    Fu     N/m^2   - zonal wind stress
    Fv     N/m^2   - meridional wind stress
    Qnet   W/m^2   - upward surface heat flux
    Qsw    W/m^2   - upward surface heat flux in short wave
    EmPmR  m/s     - Evaporation - Precipitation - Runoff

   - Note the signs are ocean-centric.
   - Forcing data (input files) was always in these units and still is.
   - Changing the units of Fu and Fv to  m/s^2  broke the ability
     to force a variable thickness layer (non-linear free-surface)
   - Changing the units of EmPmR to  psu/s  broke the ability to
     add fluid to the free-surface (natural boundary conditions)
   - the KPP package has been altered accordingly and tested
      (verification/natl_box)

3. Format of dumps has been restored to the former
   - those of you who couldn't read your output after c32 now can
   - those of you who do want the new format introduced in c32 can
     read the comments in write_state.F to enable it

Known problems in c33:
1. The output in verification/natl_box is out of date. The output
    changed due to the fix in impldiff.F