[MITgcm-devel] KPP and smoothing

[Dimitris Menemenlis]

David, my KPP is a bit rusty but what I have been using for cubed sphere 
configuration is the default KPP_OPTIONS.h file, i.e.,
#define KPP_SMOOTH_SHSQ
#define KPP_SMOOTH_DBLOC

When I initially included these filters in KPP, they were tested on a 
quasi-global lat/lon configuration with 1 degree horizontal grid spacing except 
meridionally near the equator where it was 1/3 degree.  The two options above 
were found to be the minimum requirement for removing the grid scale noise, 
which mostly occurred at the Equator in regions of local shear mixing.

The comments "smooth zonal shear meridionally" and "smooth meridional shear 
zonally" are misleading and I have just replaced them.  The flags apply 
smoothing in both directions either for dVsq or for dbloc:

c     dVsq   (nx,ny,Nr)    - velocity shear re surface squared
c                            at grid levels for bldepth             (m^2/s^2)
c     dbloc  (nx,ny,Nr)    - local delta buoyancy at interfaces
c                            for ri_iwmix and bldepth                 (m/s^2)

The reason for not smoothing dVsq, in addition that it does not seem to be 
needed to remove the grid scale noise, is because it is used in mixed layer and 
surface convection computations, which would be smeared if dVsq was smoothed.

D.