[MITgcm-support] Smagorinsky viscosity

Malte Jansen mfj at uchicago.edu
Fri Mar 13 17:10:45 EDT 2015


Thanks, Jody!
-Malte

> On Mar 13, 2015, at 9:15 PM, Jody Klymak <jklymak at uvic.ca> wrote:
> 
> 
> Hi Malte,
> 
> I think its true that you can make an anisotropic non-hydrostatics simulation that better simulates non-linear internal waves, if that is what you are interested in.  But delta x still needs to be pretty small.  See Vitousek and Fringer 2011 (Oc. Modelling).  If getting the proper shape of non-linear internal waves is important, then non-hydrostatic is worth the extra computational expense.  From an energetic point of view, its not as clear that it helps. As I said it depends on what you are trying to simulate.
> 
> Cheers,   Jody
> 
> 
> 
>> On 13 Mar 2015, at  11:45 AM, Malte Jansen <mfj at uchicago.edu> wrote:
>> 
>> Jody,
>> 
>> Are you arguing that the combination of non-hydrostatic with anisotropic grid is never desirable? It is my understanding that in the presence of stratification, turbulence (and waves) generally remain anisotropic, even for scales/processes where it might be preferable not to use the hydrostatic approximation. E.g. I thought that a hydrostatic model cannot reproduce the exact dispersion relation for internal waves, nor would it be adequate to properly represent symmetric instability - yet the aspect ratios here are still commonly much larger than one. There seem to be a range of publications which make use of non-hydrostatic models with anisotropic grids, so I don’t seem to be the only one who thinks this is reasonable, but I’m certainly not an expert on this smaller-scale stuff, so please do corrected me if I’m wrong.
>> 
>> Thanks,
>> Malte
>> 
>> 
>>> On Mar 13, 2015, at 6:27 PM, Jody Klymak <jklymak at uvic.ca> wrote:
>>> 
>>> 
>>> 
>>>> Do I understand it correctly though that Smag3D uses the same viscosity coefficient for the horizontal and vertical? I’m using a 1:5 grid aspect ratio, so I don’t think I would want that. 
>>> 
>>> Yes, Smagorinsky uses just one viscosity for all three dimensions.  
>>> 
>>> I'm not sure why you would want 3-D Smagoronsky if your grid is 1:5.  Non-hydrostatic is a bit questionable as well.  What are you trying to simulate?
>>> 
>>> Cheers,   Jody
>>> 
>>> 
>>> 
>>>> Thanks,
>>>> Malte
>>>> 
>>>>> On Mar 12, 2015, at 3:37 PM, Jody Klymak <jklymak at uvic.ca> wrote:
>>>>> 
>>>>> Hi Malte,
>>>>> 
>>>>> I just looked into this with Ruth Musgrave's help as well.  3-D Smagorinsky works pretty well. Your grid can be anisotropic, but probably not too anisotropic.  If you are using non-hydrostatic you probably don't want a grid that is too anisotropic anyways.   As written, it does not do any enhanced diffusivity, though I have a straight forward mod to enhance the diffusivities as well.  I guess I dont' know about any issues with the vertical advection scheme.  I'd have to take a look at it.  
>>>>> 
>>>>> I believe it just uses Dearborn's scheme.  So the Smagorinsky constant is just a constant.  I found to get energy to balance you had to tune this constant, so it would be great if someone moved towards a Germano type scheme.
>>>>> 
>>>>> I dont' have access to my files right now, but will in a few hours, but its pretty straightforward to try.  A little less straightforward if you want your own diagnostics into the dissipation terms and matching diffusivity.
>>>>> 
>>>>> I don't know anything about 2-D Smagorinsky.  
>>>>> 
>>>>> Cheers,   Jody
>>>>> 
>>>>>> On Mar 12, 2015, at  4:03 AM, Malte Jansen <mfj at uchicago.edu> wrote:
>>>>>> 
>>>>>> Hi all,
>>>>>> 
>>>>>> Can anybody give me (or point me to) a brief update on the Smagorinsky viscosity implementation? The documentation seems to say that the vertical component of the Smagorinsky viscosity has not yet been implemented. Is that still true? 
>>>>>> 
>>>>>> I also found an old post stating that Smag does not yet work at all for the non-hydrostatic version. Is that still true?
>>>>>> 
>>>>>> There now also is a Smag3D. Is there any documentation for that? Do I see it correctly that Smag3D is fully isotropic (and thus probably not recommended if the model grid is anisotropic)?
>>>>>> 
>>>>>> I am running a non-hydrostatic model with non-isotropic grid. So my question is basically if there is an existing implementation of the Smagorinsky viscosity that should be adequate?
>>>>>> 
>>>>>> Thanks,
>>>>>> Malte
>>>>>> _______________________________________________
>>>>>> 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
>>>> 
>>>> _______________________________________________
>>>> MITgcm-support mailing list
>>>> MITgcm-support at mitgcm.org
>>>> http://mitgcm.org/mailman/listinfo/mitgcm-support
>>> 
>>> --
>>> Jody Klymak    
>>> http://web.uvic.ca/~jklymak/
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> _______________________________________________
>>> 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
> 
> --
> Jody Klymak    
> http://web.uvic.ca/~jklymak/
> 
> 
> 
> 
> 
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support



More information about the MITgcm-support mailing list