Modify diffusivity and viscosity, and EOS

[taimoorsohail]

From Thompson + Stewart, 2014:

We apply a horizontal Laplacian viscosity of 12 m2s−1 and a vertical Laplacian viscosity of 3×10−4m2s−1. We also employ a horizontal biharmonic viscosity with Courant-Friedrichs-Lewy number 0.1, and biharmonic Leith and modified-Leith viscosities [Fox-Kemper and Menemenlis, 2008] with coefficients both equal to 1. The only explicit mixing of potential temperature and salinity is via a vertical diffusivity of 5×10−6m2s−1. This mixing is supplemented by the K Profile Parametrization (KPP) [Large et al., 1994], which enforces a surface mixed layer of at least 50 m depth. The ocean’s equation of state is computed using the numerically efficient algorithm of McDougall et al.[2003] (25 term EOS).

We should ry to match these parameters as closely as possible

[navidcy]

What do they mean by "horizontal biharmonic viscosity with Courant-Friedrichs-Lewy number 0.1"?

[glwagner]

That's a lot of parameterizations!

[navidcy]

It is what it is.

[glwagner]

I suggest using WENO vector invariant in place of the lateral parameterizations because the effective resolution will probably be higher. Replacing KPP is more difficult though. I think to start you can use RiBasedVerticalDiffusivity with the calibrated default parameters.

[navidcy]

We can just start with convective adjustment for a mixed-layer parametrization, no?

[navidcy]

So you are suggesting WENO vector invariant and then drop the biharmonic viscosities?

[glwagner]

Convective adjustment works as well! RiBasedVerticalDiffusivity is calibrated which I regard as an improvement (and it will mix in the presence of wind stress too which is helpful). On lateral parameterizations, we find that with WENO vector invariant there's no need for biharmonic viscosities. These are typically employed to stabilize noisy second-order momentum advection schemes. We can also compare results with and without a Leith parameterization.

[taimoorsohail]

This is great, thanks @glwagner !