[Mitgcm-support] Re: gm/redi clipping that works

[mitgcm-support at dev.mitgcm.org]

Mesdames et Mesjeurs,

ca y est. Nous sommes dans le bizness.

I've gone through all slope_limit cases (not for slope_psi yet).
Made some slight modif's to the calculation.
The changes will affect numbers since exact zero 
cases are now treated differently.
Still needs to be tested for physical consistency 
of the forward results.

Patrick

Patrick Heimbach wrote:
> 
> The old code seems to do what I was pointing out w.r.t.
> the gradmod == 0 values:
> Instead of trying to set those artificially to small_number
> it does not bother about those, but instead sets
> K13 = K23 = K33 = 0.
> That makes the scheme much less sensitive to small_number.
> 
> P.