[MITgcm-support] viscAh & viscAz!

Baylor Fox-Kemper baylor at MIT.EDU
Tue Jul 19 09:49:43 EDT 2005


HI Goran,
   The viscosities produced by Leith and Smagorinsky just get passed on 
to the rest of the model, where boundary conditions are set just as if 
there were a constant viscosity.  Thus, unless there is a bug in the 
corner values of the Leith and Smag viscosities, a similar effect 
should occur with constant viscosity.  Neither the Leith nor Smag 
should go to zero at the walls.  In fact, that would make the 
mathematical formulation of the model ill-posed...
   Can you try a short run with constant viscosity to see what happens?  
If the problem is there as well, that will be a good clue as to what's 
going on.
   Also what boundary conditions are you using?  There are (I believe) 
some numerical problems that can occur with no-slip and biharmonic 
viscosity (whether or not C4leith is used)...
   Also, if you are forcing with buoyancy and wind, I found in a similar 
calculation intense sinking in that region, but I thought it was 
physical...
    Cheers,
     -Baylor


On Jul 18, 2005, at 3:47 PM, Göran Broström wrote:

> Hi again
>
> I have a question about lateral boundary conditions and the new 
> schemes. What are the boundary conditions??
>
> The reason I ask is the when I used biharmonic friction, 
> eddy-permitting resolution (0.3 degree), and vertical walls we got a 
> very intense sinking in the northeastern corner (using a simple box 
> 0-60N, 0-60E as the ocean). Joseph LaCasce thinks it can be something 
> weird with the boundary conditions (he sent a question to the mitgcm 
> list but did not receive any answers) although we have not yet 
> completed a detailed analysis of the dynamics of the eastern boundary. 
> (the intense downwelling seems to disappear with smooth sloping walls 
> but that is another story). What will happen at the walls with the 
> Leith scheme, will horizontal mixing go towards zero at the walls or 
> will the boundary conditions come into play and possibly be important?
>
>
> Sincerely
> Göran
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support





More information about the MITgcm-support mailing list