[Mitgcm-support] RE: Conservation z/z*

[mitgcm-support at dev.mitgcm.org]

The exchange (point 3) shouldn't be needed so somewhere a loop range has
been changed. That it didn't show up in any tests is bad; do we need a
xz-slice test?

A.


> -----Original Message-----
> From: Jean-Michel Campin [mailto:jmc at gulf.mit.edu]
> Sent: Tuesday, March 25, 2003 1:13 PM
> To: adcroft at mit.edu
> Cc: jmc at gulf.mit.edu
> Subject: Re: Conservation z/z*
> 
> 
> Alistair,
> 
> 1) I don't like phiMin=0 & cyclic BC in Y direction,
> because at the end I don't know what dxG is for j=2 
> (lat=1.deg) and not sure that it is equal to dxG(j=1) 
> (lat=0). But if I change it to phiMin=-.5, I get the same 
> drift in eta.
> 
> 2) same drift when using linear FS (good or me, its not my
> recent coding that seems wrong). 
> 
> 3) I add an exchange of gUnm1,gVnm1 in forwardstep just
> before calling solve for P and it give a perfect
> conservation of volume and temperature.
> 
> 4) why is this exchange(gUnm1,gVnm1) needed ?
> Normaly, it should be OK without.
> I continue my investigation.
> 
> See you,
> 
> Jean-Michel
>