[MITgcm-support] viscosities

Martin Losch Martin.Losch at awi.de
Wed Jan 10 11:27:12 EST 2007


Hi Lina,

before you go on and use Leith or Smagorinsky schemes for  
viscosities, I recommend this for constant viscosity and variable  
grid spacing: viscAhGrid and viscA4Grid (for bi-harmonic viscosity).  
They are scaled like this to give their full viscosity contribution:
viscAh = 0.25*L**2*viscAhGrid/deltaT
viscA4 = 0.25*0.125*L**4*viscA4Grid/deltaT

Also with your high resolution, try using bi-harmonic viscosity  
(maybe together with harmonic viscosity).

I found that the Leith/Smagorinsky schemes work very well, once the  
circulation has pickup up. Please have a look at the documentation,  
start here:
http://mitgcm.org/r2_web_testing/latest/online_documents/node85.html
and/or read Baylor's excellent recipe for using these schemes:
http://forge.csail.mit.edu/pipermail/mitgcm-support/2006-June/ 
004086.html
I've had long discussion with Baylor on this subject, starting here:
http://forge.csail.mit.edu/pipermail/mitgcm-support/2006-November/ 
004459.html
which helped me a lot, maybe it will help you too.

Also in pkg/mom_common/mom_calc_visc.F you'll find a summary of the  
different viscosity parameters along with some recommended values.

Good luck!

Martin

On 10 Jan 2007, at 16:57, lina sitz wrote:

> Hi,
> I'm Lina Sitz and I'm new using MITgcm. I'm studying the vortex- 
> pair formation in a basin, product of a tidal-jet arriving from a  
> channel.
> Maximum velocities of the jet are 1 m/seg approximately and the  
> tidal period is 12 hs. The eddies have a radio of 2.5 km  
> approximately. The horizontal dimension of the region is 15x30 km  
> and the depth is 25 m.
> Because of the size of the channel and the dimension of the problem  
> I would like to use a variable resolution (between 50 and 400 m),  
> but I would like to know what kind of viscosity I must to set, when  
> running MITgcm. I tried with different values of constant  
> viscosities, taking into acount the stabilities constraint, but the  
> solution have noise, even using a grid with constant resolution (50  
> m). I don't know if in this kind of problem is more adequate to use  
> Smagorinsky or Leith viscosities or if the values that I'm using  
> are not correct.
>
> I apologize because of my English and I would appreciate any  
> suggestion.
>
> Thanks in advance!
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support




More information about the MITgcm-support mailing list