[Mitgcm-support] checkpoint36

[mitgcm-support at dev.mitgcm.org]

Does it help if we do the shapiro as a tendency
i.e. 

 du/dt_filter = shapN(u)/time_scale

 Gu = Gu + du/dt_filter
   :
 This is what most atmos. models do.

 Doesn't help with FFT, but they are history when cube-sphere
 flys....
 
Chris
Jean-Michel Campin wrote:
> 
> Hello,
> 
> > Jean-Michel Campin wrote:
> > > Concerning Borotropic Crank-Nickelson time stepping:
> > > I had to apply the filter (FFT or Shapiro) to
> > > uVel and vVel when the 2Dflow_Divergence is not
> >
> > Yes, I noticed, but is this a problem with the way it is now?
> >
> > A.
> Like it is now, it seems to work. But the small adjustment test
> I run with shapiro show some differences (near the poles ?) compared
> to non-filtering run.
> 
> The (only ?) alternative would be to apply the filters at the end
> of the time step, after solve pressure. if the 2D solver is able
> to work with non filtered gUnm1 as input, ...
> I am not sure it works like this. and then eta will not be filtered ?
> 
> I propose to discussed this later ?
> 
> See you,
> 
> Jean-Michel