[MITgcm-support] High vertical velocity/noise next to land

gael forget gforget at mit.edu
Tue Feb 23 20:23:44 EST 2010


Hi David and Martin,

I wanted to mention an option that I recently added to MITgcm,
and which may be of some relevance to your discussion.

I think I add a similar issue when setting up ECCO v4 (using the lat- 
lon-cap grid @ 1 degree res.).
I wanted something less "aggressive" than the CD scheme that would  
still damp the velocity grid point noise.
The best compromise I found consists in adding bi-harmonic viscosity  
just near land (i.e. the ocean bottom).

To this end I added the ALLOW_3D_VISCA4 option in MITgcm. It allows  
you to provide fields of
biharmonic viscosity coefficients. The so-prescribed viscosity will  
be added on top of what you already use.

Cheers,
Gael

On Jan 28, 2010, at 5:41 AM, Martin Losch wrote:

> Hi David,
>
> good question.
> I think ISOTROPIC_COS_SCALING should be defined  
> ISOTROPIC_COS_SCALING, but make shure that is only applies to the  
> gad_*.F files (-> GAD_OPTIONS.h), and not the mom_*.F files.
> In an effort to reduce the horizontal diffusion as much as  
> possible, I usually set diffKh/4T/S so zero and use an  
> unconditionally stable advection scheme (such as 33).
>
> Martin
>
> On Jan 28, 2010, at 1:10 PM, David Munday wrote:
>
>> 'm experimenting with the viscAhgrid and viscA4grid parameters at  
>> the moment, although it looks like I might need values as small as  
>> 0.05 to give comparable viscosity/diffusivity to the constants  
>> that I was prescribing before. I suspect that friction/ 
>> dissipation, and time step, might still need some tuning. If I'm  
>> using diffK4T & diffK4S to set biharmonic coefficient for  
>> temperature and salinity, do I then need to define  
>> ISOTROPIC_COS_SCALING in GAD_OPTIONS.h to ensure that it's scaled  
>> properly?
>
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support




More information about the MITgcm-support mailing list