[MITgcm-support] Questions about LES / turbulent viscosities

[David Hebert]

Thanks Martin for your input regarding the sub-grid scale parameters. I 
am looking through the code and see Mellor Yamada 82 package. It looks 
like this computes vertical viscosities and diffusivities based on 
Richardson number. I am trying to model flow over submerged alps, and 
perhaps this might suffice for a variable viscosity/diffusivity package. 
Perhaps this can be used in conjunction with Smagorinski for horizontal 
parameters? I sill plan on running in non-hydrostatic mode, but would 
like some turbulent mixing effects downstream of the alps.

Thanks,

David

Martin Losch wrote:
> David,
> the LES capabilities of the MITgcm are not fully developed.
> As you noticed, flow dependent viscosities (Smagorinski, Leith) are 
> only implemented for the horizontal viscosities in the momentum 
> equations (for both horizontal and vertical velocities). Vertical 
> viscosities are constant (if you don't use KPP or other mixed layer 
> models).
> There is no equivalent for tracers. I generally use flux limiting 
> advection schemes to take care of the "sub-grid turbulence". This is 
> probably not the best sgs-model, but often sufficient.
>
> For non-hydrostatic modelling, I usually do not use KPP, not sure how 
> these two features "cooperate". Also, KPP is supposed to model shallow 
> and deep convection. In LES I suspect, that you want to resolve these 
> processes explicitly.
>
> Implicit viscosity/diffusion needs to be turned on with KPP (and other 
> mixed layer schemes), because large vertical 
> viscosities/diffusivities, as computed by KPP, can make the explicit 
> scheme unstable. Otherwise it is usually not required (but you can 
> always turn it on).
>
> Implementing 3D smagorinski viscosities is on someone's list, but I 
> don't know how far down.
>
> Here are some more earlier postings of this list that deal with 
> Smagorinski, etc.
> http://forge.csail.mit.edu/pipermail/mitgcm-support/2006-June/004086.html
>
> Martin
>
> On 16 Sep 2008, at 19:44, David Hebert wrote:
>
>> I have been looking at the manual regarding LES viscosities (section 
>> 2.21) and am hoping someone can answer the following
>>
>> I see viscC2Smag appears in the pkg/mom_vecinv routines. Does this 
>> imply that vector invariant form must be used with viscC2Smag?
>> It appears that this will only affect horizontal viscosities. If I am 
>> running in nonhydristatic mode, do I need to use KPP (section 6.4.2 
>> of the manual) for enhanced vertical viscosity based on shear, etc.
>> Is viscC2Smag only used for momentum equations? Do I need to use KPP 
>> for turbulent mixing of tracers?
>> Along those lines, in the KPP section I see there are flags for 
>> implicitViscosity and implicitDiffusion. I am having trouble finding 
>> what these flags do outside of KPP. Are they required for viscC2Smag? 
>> Where can I find more info about the implicit flags?
>> Thanks for your help.
>>
>> David
>>
>> _______________________________________________
>> MITgcm-support mailing list
>> MITgcm-support at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-support
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support