Horizontal Mixing & spatially varying metric factors m and n

[aryansafaie]

Dear ROMS Friends,

In setting values for the harmonic and biharmonic horizontal diffusion and viscosity coefficients (TNU2, VISC2, etc), I'm taking guidance from the following page (the one Kate refers people to) : https://www.myroms.org/wiki/Horizontal_Mixing

I have a couple questions based on the information from there:
1) I cannot find a copy of the Wajsowicz 1993 paper for whatever reason, and I do not understand what is meant by the spatially-varying metric factors 1/m and 1/n... can someone point me in the right direction to understand m and n?

2) Running a quick calculation as described by the bottom section of that webpage (Guidelines for Coefficient Values), for my setup, VISC2 < 2.026 m^2/s for ∆x = 10 m and ∆t = 5 s, and TNU2 = VISC2/10. In a few previous runs, I mistakenly set VISC2 ~ 10^-4, and my simulation would blow up after a few hours of simulation time. Is the following reasoning appropriate then: too little viscosity will have the fluid flow traveling across cells too fast to be resolved by my time step (5 s), so a CFL violation and blow up will ensue?

Any help or suggestions for reading material are greatly appreciated.
Best,
-aryan

[kate]

In the ROMS literature, m=1/dx and n=1/dy. https://www.myroms.org/wiki/Curvilinear ... sformation

The numerical schemes will often generate 2dx noise at some rate. You want your damping time on the 2dx stuff to be short enough to combat the growth. Too little viscosity will allow this stuff to grow and dominate the solution. You can plot your results to see what's going on for sure.

[aryansafaie]

Thank you very much Kate.
That helps!