[MITgcm-support] noise near the bottom

Jody Klymak jklymak at uvic.ca
Tue Jul 30 15:49:14 EDT 2019

Dear Yangxin He,

Thats undoubtably because the flow has gone quite non-linear here, and the numerical scheme is producing noise.  To get rid of this, you can expensively turn on the nonhydrostatic code, in which case you’ll still get wiggles, but they won’t necessarily have grid-scale oscillations, or you can add more lateral diffusion and viscosity (at the expense of freee-running waves elsewhere).

Personally, I just put up with this grid-scale “speckling”, its basically numerical dissipation, and it usually occurs at shocks where you’d expect dissipation.  But if your application requires this to be more precise, I’d increase the resolution and turn on non-hydrostatic code.  But for breaking wave problems I’ve not found this to substantively affect the large-scale energetics.  

I’d suggest you turn down your lateral diffusivity.  Advection scheme 33 will allow running w/ this set to zero.  But you may consider turning on lateral viscosity.  

Cheers,   Jody

More information about the MITgcm-support mailing list