[MITgcm-support] turning off gent-mcwilliams

[Jean-Michel Campin]

Hi Martin,

The things you are doing are not really "standard":
I agree that being able to use GMRedi for T & S but not
for all the passive tracers could be interesting. But the opposite
looks like a particular case that we may not want to support.

Anyway, apart for the vertical diffusivity (sightly more complicated,
I am trying to figure out how it's currently done), 
you would need to 
1) keep "CALL GMREDI_CALC_TENSOR" at the beginning of thermodynamics
2) call gmredi_x,y,r_transport in gad_calc_rhs only for some
 ptracers but not for T & S

If you are using the advective form of GM and not the skew flux,
it become more complicated (since the same transport flow is used 
for T,S and tracers, in gad_calc_rhs). In this case, I recommend 
a) if you are using a linear advection scheme, 
  to use GM_AdvSeparate = TRUE (it's not the default), so that you
  avoid this problem.
b) if you are using a nonlinear one (77,33), might not be a problem
 since GAD_ADVECTION compute all the transports and is called for 
 each tracers.

I think it's also the reason why PTRACERS_useGMRedi has not been
implemented earlier, since it's not so easy to do it for all the
different cases, without recomputation and/or significant changes.

I am in the process of making changes in KappaRT & KappaRS, 
so we may run into some merging conflict when checking-in 
those modifications. Let's try, and we will see how many broken things
we get.

Have fun !

See you,

Jean-Michel