[Mitgcm-support] CAUTION !!!

[mitgcm-support at dev.mitgcm.org]

>> in the MIT/SIO KPP classic code, the internal wave diffusivity is set to
>> zero since we pass in the value we normally use as a vertically
>> const. background viscosity. It represents the same thing and should be put
>> in only once. This all comes down to the choice of the vertical background
>> viscusity/diffusivity and that has NOTHING to do with KPP itself but the
>> interior ocean.

In testing above strategy I discovered that numerical results are different
when background diffusivities are set through KPP internal wave diffusivity
parameter or through model vertical background diffusivity.  The reason is
that S/R blmix relies on background diffusivity terms to be included in terms
computed by S/R ri_iwmix, especially in regions where bathymetry outcrops
mixed layer.  It's not a serious problem, but it's one more reason why Classic
and UV KPP codes give different numerical results.

For UV KPP, I am taking the approach of eliminating dif?iw variables but
adding model background diffusivities inside S/R ri_iwmix rather than after
the KPP code as is currently done in the Classic implementation.

Dimitris