Laplacian Viscosity Damping Time

[aryansafaie]

Dear all,

This question is in regards to motivating the damping time of the Laplacian viscosity used in horizontal mixing. I know that we want to have that
time scale be resolved by our model time step but large enough to resolve features of interest.
For some context, my application is concerned with the nearshore ~1 km of a linearly sloped bottom with grid spacing
dx = dy = 10 m, 30 vertical levels, TS_MPDATA and UV_C4ADVECTION for tracer and momentum advection, respectively, and we're
interested in features like fronts that have time scales of ~4-6 hours. I've typically used VISC2 and TNU2 values corresponding to a damping time of 60 seconds
(i.e. VISC2 for me = 0.1689 m2/s).


I'm aware of the guidance for Horizontal Mixing found at https://www.myroms.org/wiki/Horizontal_Mixing, and I've consulted some
very nice resources such as:
-Griffies, Stephen. Fundamentals of ocean climate models. Princeton university press, 2018.
-Lee, Mei-Man, Andrew C. Coward, and AJ George Nurser. "Spurious diapycnal mixing of the deep waters in an eddy-permitting global ocean model." Journal of physical oceanography 32.5 (2002): 1522-1535.
-Griffies, Stephen M., and Robert W. Hallberg. "Biharmonic friction with a Smagorinsky-like viscosity for use in large-scale eddy-permitting ocean models." Monthly Weather Review 128.8 (2000): 2935-2946.

So my question is: ignoring for now a thorough and systematic sensitivity analysis involving many simulations to understand the influence of VISC2 and TNU2, what has been your motivation for the choice of these values? Have you seen that as long as you fulfill the damping time criteria, you are left with a pretty arbitrary range of values to choose from for VISC2 and TNU2 (and therefore t_damp?)
I guess generally I'm just asking around for your experience in choosing VISC2 and TNU2, because I'm trying to understand what motivates that choice for spin-down time.

Thanks in advance.
-aryan