[MITgcm-support] Zonal velocity bands

[Jason Goodman]

On Mar 26, 2010, at 1:57 PM, Neil Swart wrote:

> I'm not sure I understand your comment on periodicity for the forcing.

> tauMax=0.1;
> x=((1:nx)-0.5)/(nx-1); % nx-1 accounts for a solid wall
> y=((1:ny)-0.5)/(ny-1); % ny-1 accounts for a solid wall
> [X,Y]=ndgrid(x,y);
>
> tau=tauMax*sin(1.2*pi*Y);

With a nonzero zonal velocity at the southern boundary, you'll get a  
northward Ekman flow at the south wall: by continuity, this can only  
be balanced by a ferocious vertical upwelling.  I'm guessing you're  
getting some sort of numerical instability as a result.

Try checking the vertical advective CFL number in the monitor output  
(advcfl_wvel_max).  If it's a large fraction of 1, you're in trouble.