[MITgcm-support] help-noise with the internal wave simulation

[Jody Klymak]

Hi,


> On Dec 26, 2019, at  2:13 AM, JIATONG <jiatong at radi.ac.cn> wrote:
> 
>     And here is my question. I am trying to model nonliear internal waves with two open boundaries without free surface (rigid lid) in 2D 
> in nonHydrostatic. In the model setup, I input the barotropic tide at west and east boundaries over the bottom topography, and want the 
> barotropic tide can evolve into nonlinear internal wave(internal solitary wave). dx is set to 50m and dz is set to 5m. dt is 5s. However, 
> the T of the model output is filled with noises and I can not identify the wave signal(its lenght scale is about several hundred meters). 
> I've tried to set various dx, dz and dt steps, but  without positive result. And I also tried the Laplacian viscosity coefficient changing 
> from 1e-5 to 1e-1, some very small noise have been removed, but there are still some large noises. 


What do you mean by “large noises”?  The model results you are plotting look reasonable to me if your forcing is moderately strong.  You would expect a lot of turbulence near the generation site.  

If you are worried about grid-scale noise, you can often lessen that by changing to a higher-order advection scheme, at the risk of introducing more numerical dissipation and diffusivity: https://mitgcm.readthedocs.io/en/latest/algorithm/algorithm.html#advection-schemes  But the real advnatage of that is increasing the time step, not particularly to make the results look better.  

Cheers,   Jody


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.mitgcm.org/pipermail/mitgcm-support/attachments/20191226/ed9875ca/attachment.html>