[MITgcm-devel] conservation Problem with KPP viscosity

[Jean-Michel Campin]

Hi Dimitris,

I have a temporary fix for this non-conservation problem,
which consists in exchanging KPPviscAz after kpp_calc.

I think it would be good to sort out why this exch is needed, 
and to fix it properly (in kpp_calc, something to do with the
computational domain), but for now, I am thinking of checking-in 
my fix (adding the exch call). OK like this ?

Jean-Michel

On Mon, Apr 30, 2007 at 04:58:14PM -0700, Dimitris Menemenlis wrote:
> Jean-Michel, any more ideas about this KPP conservation problem.  From your 
> description below it sounds like problem is not caused by KPP itself but by 
> some inaccuracy with time-stepping scheme in the presence of large, 
> spatially varying vertical viscosities ???
> 
> >Seems that the problem a conservation of T & S in the donuts
> >is related to KPP viscosity:
> >When I comment out CALL KPP_CALC_VISC in calc_viscosity.F,
> >I get a good conservation of T (~1.e-5 J/m2) and S (~1.e-8 g/m2),
> >over 1 year, which is what I expect given the machine precision.
> >On the contrary, with KPP-viscosity on, the drift in T & S are
> >still small ( ~1 J/m2 & ~1.e-5 g/m2) over the same period,
> >but much larger than what they should be.
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel