[MITgcm-support] viscAh & viscAz?

[Baylor Fox-Kemper]

Hi Van Thinh,
   The way that things stand now, the Smagorinsky and Leith  
viscosities would be applied only in the U and V momentum equations  
(not the W equation).  The W equation only sees the constant viscAhW,  
which is usually set to be equal to viscAh.  This is an inconsistent  
application of the lateral viscosites, even if it 'improves' the  
noise, it is not really set up correctly.
   I am looking as to how to change things so this won't be the case.
   Cheers,
     -Baylor

On Jun 14, 2006, at 11:14 AM, Van Thinh Nguyen wrote:

> Hi Baylor,
>
> When I run the non-hydrostatic mode with the option of Smagorinsky  
> and Leith, the program had no warning or error, it just crashed  
> after 1/2 tidal period (22320 s), it happened the same when i  
> played only with viscAh & viscAz (increased to 1.). And with the  
> option of Smagorinsky & Leith, it shows a great reduce the noises  
> with the low values of viscAh &
> viscAz (10-5). So I'm not sure that Smagorinsky & Leith wouldn't  
> work with non-hydrostatic?
>
> In case to run non-hydrostatic mode with Smagorinsky and/or Leith,  
> some works have to be implemented, could you please give me some  
> hints which subroutines have to be concerned?
>
> Thanks a lot!
>
> Van Thinh
>
> -------------------------------------------
>
> On Fri, 9 Jun 2006, Baylor Fox-Kemper wrote:
>
>> Hi Van Thinh,
>> Are you using non-hydrostatic mode?  If so, Leith and Smagorinsky  
>> are not yet implemented, and you can only use viscAh or viscA4.
>> If you are using hydrostatic mode, and you are at O(10m)  
>> resolution, it is probably most sensible to be using the  
>> Smagorinsky scheme.  You should be able to run it with no other  
>> viscosities.  If you are having trouble with noise, you might try  
>> adding some biharmonic viscosity.
>> Try this to get started:
>>
>> &PARM01
>> viscAr=5.E -5,
>> # This is harmonic Smagorinsky Coefficient
>> # 0.2-0.9=value from 3-d turbulence theory; Smagorinsky (1993)
>> # 2.2-4=stable oceanic values (Griffies and Hallberg, 2000)
>> # 1.0 smag is approximately the same as 1.85 leith+leithd
>> # for this grid spacing
>> viscC2Smag=1.0,
>> viscC4Smag=1.0,
>> # This is the max Viscosity used, be it Smag or Leith, and it scales
>> # with grid size and timestep
>> viscAhGridMax=0.5,
>> viscA4GridMax=0.5,
>>
>> implicitDiffusion=.TRUE.,
>> implicitViscosity=.TRUE.,
>>
>> # These two might be particularly useful, depending on your problem
>> usejamartwetpoints=.TRUE.,
>> staggerTimeStep=.TRUE.,
>>
>> # You should stick with the 33, it reduces noise.
>> tempAdvScheme=33,
>> saltAdvScheme=33,
>> nonHydrostatic=.FALSE.,
>> &
>>
>> You can play with the values of the Smagorinsky coefficients to  
>> see what effect they have.
>>
>> If you are using nonhydrostatic mode, let me know (I am overdue on  
>> setting up the variable viscosity for nonhydrostatic mode anyway!)
>> Cheers,
>>  -Baylor
>>
>> On Jun 8, 2006, at 3:37 PM, Van Thinh Nguyen wrote:
>>
>>> Dear all,
>>> In oder to reduce the noises, I am playing with parameters viscAh  
>>> & viscAz, so I have some questions, could someone please help me?
>>> 1. From the documentation, we can have two ways to set viscAh:
>>>   a. By estimating velocity & grid scale
>>>   b. By using viscAhReMax and/or viscC2Smag (or maybe combine  
>>> with Leith
>>>      viscosity)
>>>  So, I understand that when I use option b) then I don't need to set
>>>  viscAh or viscAz in &PARM01, is that correct?
>>> 2. Here are some options in data file:
>>> &PARM01
>>> #
>>> viscAz=1.E-5,
>>> viscAh=1.E-5,
>>> tempAdvScheme=33,
>>> viscA4=0.E12,
>>> diffK4T=0.E4,
>>> diffKhT=1.E-5,
>>> diffKzT=1.E-5,
>>> diffK4S=0.E4,
>>> diffKhS=1.E-5,
>>> diffKzS=1.E-5,
>>> implicitDiffusion=.TRUE.,
>>> implicitViscosity=.TRUE.,
>>> viscC2leith=1.,
>>> viscC2leithd=1.,
>>> ##
>>> &
>>> &PARM04
>>> viscC2Gmag=4.,
>>> viscAhGridMax=0.5,
>>> &
>>> I even try to increase viscAh & viscAz (to 1.E-1), as well as  
>>> increase viscC2leith=viscC2leithd=1.
>>> but I still get noises.
>>> I know that you should know my simulation (problem &domain length  
>>> scale) related to grid space (dX=10m) & time step (1s). However,  
>>> I just wanna know
>>> that my above options are correct or anything in conflict.
>>> My problem has to be dealed with the range of viscAh=1.E-6 &  
>>> viscAz=1.E-6, but I got stuck in this simulation due to the noises.
>>> Thanks so much for your helps.
>>> Van Thinh
>>> -----------------------------------------------------
>>> _______________________________________________
>>> 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