[MITgcm-support] Internal waves. Noise near bottom.

Klymak Jody jklymak at uvic.ca
Mon May 2 10:21:56 EDT 2011


Hi Serge,

I can't tell too much from what you sent that one model is behaving "better" than the other.  Some plots of both models on the same plot at the same time steps might help us understand what features you are not liking in the MITgcm. 

Then, it woudl be nice to know what diffusivities and viscosities you have used in the sigma-coordinate model, and, if you are happy with your Reynolds number, use the same in the MITgcm runs. 

Then, you should check out the cell shaving code and make sure you have set everything correctly.  

Finally, if you want to fundamentally reduce the stepiness, use more vertical resolution.

Cheers,   Jody


On 2011-05-02, at 12:25 AM, fancer fancer wrote:

> Jody,
> 
> I worry about noise low frequency, what come from bottom steps. For comparison I attached the screenshots of pictures from other model. These pictures look more clearly, perhaps for reason what this model uses sigma-coordinate. And I trying to create the MITgcm project with similar output. Could you tell me Is this possible? 
> 
> We are investigating an internal waves in closed basin, generated as a result of deviation of pycnocline. These deviations may be caused for various reason, for example, wind-surge phenomena and etc. I trying to make a serial of experiments for comparison our model and MITgcm.
> 
> Thanks for response and interest.
> 
> Sincerely,
> Serge Semin
> 
> On Sun, May 1, 2011 at 9:19 PM, Klymak Jody <jklymak at uvic.ca> wrote:
> The new runs look very different from the old run; all the very strong vertical excursions have been suppressed.   What "noise" are you worried about in the new runs?    
> 
> I can't quite tell what you are trying to do in your model, but note that shocks or solitons will not be properly represented unless you turn non-hydrostatic to be "on".  In hydrostatic mode, most shocks will try to dissipate energy by shedding grid-scale oscillations in their wake.  Perhaps some of the "steppiness" you are seeing is related to this phenomena.   
> 
> Finally, I would look at the cell-shaving code again, or perhaps someone who understands it better can suggest proper settings for it.  I don't think you are going to get rid of some internal waves being generated at the steps.  However, they should really have little energy and affect your solution very little.  
> 
> Cheers,   Jody
> 
> 
> 
> On 2011-04-30, at 11:37 PM, fancer fancer wrote:
> 
>> Jody,
>> 
>> Thanks for response.
>> Now I've put diffusion, but still with no positive results (pictures with U,W and T in attach). I've set:
>> diffKhT=1.E-2,
>> diffKzT=1.E-3,
>> diffK4T=0.E-3,
>> diffShT=0.E-2,
>> diffSzT=0.E-3,
>> diffS4T=0.E-3,
>> momViscosity=.TRUE.,
>> viscAh=1.E-2,
>> viscAz=1.E-3,
>> But noise above bottom steps still present.
>> Is it right value?
>> Is there something else, what can help to restrain noises?
>> 
>> Sincerely, 
>> Serge Semin
>> 
>> On Fri, Apr 29, 2011 at 10:12 PM, Jody Klymak <jklymak at uvic.ca> wrote:
>> Serge,
>> 
>> According to your "data" file you have no explicit viscosity or diffusion. So any internal waves generated at the steps rattle around with only numerical diffusivity to dissipate them.  I'd put some diffusivity and viscosity in, even if it is molecular.
>> 
>> Cheers,  Jody
>> 
>> 
>> On Apr 28, 2011, at  12:23 PM, fancer fancer wrote:
>> 
>> Hi Dear Modelers,
>> 
>> I am trying to model ideal case of internal waves propagation in closed basin without free surface (rigid lid) in 2D area with two layer stratification (linear tilting) and parabolic bottom.
>> Now I faced with problem of partial discretization. I can not get a simulation without noises near the largest steps of the bottom.
>> I've tried to set various dx, dz and dt steps, but without positive result. I've tried to use  various tempAdvScheme (33, 77), unfortunately with same result. And finally I've tried to set various hFacMin, now it is in 0.00005 value (best results), but it didn't helped me too.
>> We have an results of modelling through other system with sigma-coordinate discretization, where we set 1000x50 points on 10000x10 m grid. There's no noise.
>> I put the pictures with T, U and W of modeling in result directory of attached project.
>> Could you tell me how should I modify my project (in attach) for clearing the model? or how should I calculate dx and dz steps for model stabilization? (CFL in norm)
>> 
>> And one more question. My area is 2D (z and x coordinates), but models stability depends of dy step. Is this normal or something wrong with my project? If second is true, could you tell me that can be wrong?
>> 
>> Thanks for help in advance.
>> 
>> Sincerely,
>> Serge Semin
>> <01_tilting_stratification_parabola.tar.gz>_______________________________________________
>> 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
>> 
>> <tilting_stratification_parabola.zip>_______________________________________________
>> 
>> 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
> 
> 
> <comparison.zip>_______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20110502/e24a2776/attachment.htm>


More information about the MITgcm-support mailing list