"gls" in the output file, and calculation of TKE dissipation

[tongbo]

Hi all,

I wondered what the "gls" is in the output file. It is "tke*l" ("l" is the generic length) according to the glossary I found in the .in file. So then I could calculate TKE dissipation using "epsilon = c_mu0^3*tke^1.5/l = c_mu0^3*tke^1.5/(gls/tke)" - I am using the GLS k-epsilon closure. But the calculated "epsilon" seems to be an order of magnitude larger than the shear production.

Another possibility is that "gls" is just "epsilon" for the k-epsilon closure, but in this case, "epsilon=gls" is still around an order of magnitude larger than the shear production.

I wondered if anyone has tried to calculate TKE dissipation and close the TKE budget. I guess TKE dissipation happens locally and there is not a lot of advection of TKE, so the dissipation should generally balance with shear production everywhere. Thank you.

Tong

[jcwarner]

here is a good paper with lots of info:
Warner, J.C., Sherwood, C., Arango, H., and Signell, R. (2005). “Performance of Four Turbulence Closure Models Implemented Using a Generic Length Scale Method.” Ocean Modelling, v. 8/1-2, p. 81-113.

the GLS scheme is a method that transports tke and 'gls' = generic length scale. Depending on the parameter settings, gls can be dissipation or turb freq omega, etc. The values are set in the ocean.in. Please review that paper and see if that helps with the understanding.
-j

[tongbo]

Hi John,

Thanks a lot for this reply. I am well aware of this paper and the GLS method. I am just a bit confused by what ROMS writes in the output file. I think gls = epsilon for GLS k-epsilon closure, but according to the glossary in ocean.in, the output gls is always tke*l even if I am not using the k-kl closure.

Thanks,
Tong

[jcwarner]

oh ok. let me look at that glossary, it might not be worded clearly.
can you paste that part of the glossary here?
if you choose k-e coeffs, then gls will be e (dissipation).

[ggerbi]

Tong,

Look at equations 32 and 34 in the documentation here:
https://www.myroms.org/wiki/Vertical_Mi ... erizations

The meaning of the gls variable that gets computed and saved is (I believe) determined by the p, m, and n coefficients that you choose.

You can recover the dissipation from the expression in (32) using the gls_p gls_m, and gls_n variables in the output.

Does that answer your question?

Greg

[tongbo]

Thank John and Greg for helping with this. Yes I think "gls" = "epsilon" for my case, so I can directly use the "gls" variable from the output file as dissipation. And I also compared the vertical distribution of gls (tke dissipation) and it is generally consistent with shear production.

Below is the glossary for gls in ocean.in:

Hout(idMtls) == T ! gls turbulent length scale
----- glossary
! Hout(idMtls) Write out turbulent kinetic energy times length scale.

And here is the annotation in my output .nc file, and the unit is also for k-kl:

Variable "gls" In file "..."
float gls(ocean_time=...);
:long_name = "turbulent generic length scale";
:units = "meter3 second-2";

I think it's just the glossary has not been updated, or maybe it was copied from a k-kl closure before I used it. Thank you so much!
Tong

[wilkin]

Variable "gls" takes on different meanings, and units, depending on whether the parameters set in roms.in are for k-kl, k-epsilon, k-omega or "GLS".

Since roms.in is processed at run time, not compile time, the logic becomes rather tortured for setting the output netcdf long_name and units attributes for variable gls.

Perhaps we should document, more explicitly, inn the netcdf file that output variable "gls" is variable "psi" in the equations (32) and (34), and the Table, at the WikiROMS link John Warner directed you to https://www.myroms.org/wiki/Vertical_Mi ... erizations

[tongbo]

It makes sense to me now. Thank you John for this very clear explanation.