Computing subgrid-scale TKE

[phyo-wai-thaw]

Hi,

I’m running LES with the Dynamic Smagorinsky model and I would like to perform a grid-resolution check.

Could you suggest how subgrid-scale TKE can be computed diagnostically?

Thanks in advance.

[glwagner]

I'm not sure this is possible, although a scaling argument suggests that there is a velocity scale $u_\Delta^2 = \nu_e \Delta$, where $\Delta$ is the grid spacing and $\nu_e$ is the eddy viscosity. Another diagnostic is the effective Reynolds number: $Re_{eff} = Re_{actual} \nu / \nu_e$.

cc @tomchor

[tomchor]

There are some ways to estimate the SGS TKE in these models but they're not trivial and usually rely on some "Kolmogorovian" assumptions (so you gotta check if you expect them to be valid). I think this is a relatively simple approach and probably a good place to get started: https://link.springer.com/article/10.1007/s10546-011-9618-0

That said, I'd argue you probably don't need to estimate SGS TKE if all you're trying to do is a grid-resolution check. @glwagner already mentioned a couple of approaches, you can also compare your grid resolution with the Ozmodov scale and run your model at different resolutions. You can check Appendix B of this paper to see what I did as an example

[jagoosw]

I think a common way to check the resolution is sufficient is to plot the energy spectrum to check the inertial range is represented like I did here: #4367 (comment) (there is a link to code in that comment that includes working out the spectrum)

[omni-front]

not sure about this one, but maybe try calculating it using the eddy viscosity and grid spacing like @glwagner mentioned? that velocity scale $u_\Delta^2 = \nu_e \Delta$ could give you some insight.