Kantha_Clayson mixed layer method and wave model

[luohaogit]

Hi all,

I have two questions about ROMS.

(1) Mixed layer depth: Has anybody ever used the Kantha_Clayson (second-moment closure of turbulence) mixed layer method (in my25_corstep.F [Mellor-Yamada] or gls_corstep.F [Generic length-scale]) in the real domain? Did this method has advantage over LMD [Large/McWilliams/Doney] mixed layer? We are trying to run some experiment in the domain of Gulf of Mexico. Any recommendation of method to compute the mixed layer depth? If using the GLS method, which set of values are suggested?

! Suggested values for various parameterizations:
!
! MY2.5 K-epsilon K-omega K-omega K-tao
!
! GLS_P = 0.d0 3.0d0 -1.0d0 -1.0d0 -3.0d0
! GLS_M = 1.d0 1.5d0 0.5d0 0.5d0 0.5d0
! GLS_N = 1.d0 -1.0d0 -1.0d0 -1.0d0 1.0d0
! GLS_Kmin = 5.0d-6 7.6d-6 7.6d-6 7.6d-6 7.6d-6
! GLS_Pmin = 5.0d-6 1.0d-12 1.0d-12 1.0d-12 1.0d-12
!
! GLS_CMU0 = 0.5544d0 0.5477d0 0.5477d0 0.5477d0 0.5477d0
! GLS_C1 = 0.9d0 1.44d0 0.555d0 0.52d0 0.173d0
! GLS_C2 = 0.5d0 1.92d0 0.833d0 0.8d0 0.225d0
! GLS_C3M = 0.9d0 -0.4d0 -0.6d0 -0.6d0 0.0d0
! GLS_C3P = 0.9d0 1.0d0 1.0d0 1.0d0 0.0d0
! GLS_SIGK = 1.96d0 1.0d0 2.0d0 2.0d0 1.46d0
! GLS_SIGP = 1.96d0 1.30d0 2.0d0 2.0d0 10.8d0
!

(2) Wave-induced mixing: We knew that ROMS has been coupled with SWAN wave model, also in the gls_corstep, there is an option of CRAIG_BANNER (wave-induced turbulence). Did these wave models have anything to do with the mixed layer depth?

Any answer and recommendations are really appreciated.

Thanks

Hao

[ggerbi]

I would guess that most of us who use GLS use the Kantha and Clayson stability functions.

The two-equation models in GLS require two choices:
1) choose a length-scale model (the numbers in ocean.in).
2) choose a stability function (choose in cppdefs).

John Warner's 2005 paper (Ocean Modelling, 8, 81-113) on turbulence closure covers a lot of the ground in how the different choices affect model output. I also recommend Umlauf and Burchard, 2005 (Cont. Shelf Res., 25, 795-827). At the end of the day, there is little good comparison (because it is difficult) that shows which closure model produces the most "realistic" results. In truth they all have similar output. I use mostly k-epsilon and k-omega. Others use Mellor-Yamada or other schemes.

The two-equation models (GLS) are very different from the KPP model (LMD). The GLS models parameterize the evolution of turbulence dynamics. The KPP model computes turbulent diffusivities from bulk properties of the fluid. The KPP model is constructed to parameterize the surface boundary layer (mixed layer), and has bits in it that account for processes specific to the surface boundary layer. Depending on your needs, KPP may be well-suited for open-ocean applications.

GLS models do not account for specific surface boundary layer processes in such a unique way as KPP, but they do a much better job if there is spatial variability in your forcing, topography, hydrography, or turbulence conditions. If you are simulating coastal regions, then GLS is probably the better choice.

If the boundary layer is shallow enough, wave breaking will thicken the surface boundary layer. Wave breaking is probably most important at depths shallower than 10 times the significant wave height, so if you already have a deep boundary layer then wave breaking may not be important. There is plenty of discussion suggesting that Langmuir turbulence is important in thickening the boundary layer, but there is no parameterization for that in ROMS at present. As far as I know, there is no ROMS parameterization for wave breaking in KPP.

[luohaogit]

Thank you so much for your reply. Your explanation is very helpful.

Thanks again,

Hao

[jcwarner]

a follow up paper that describes the surface flux of tke (CRAIG_BANNER) is
Carniel, S., Warner, J.C., Chiggiato, J., and Sclavo, Mauro. (2009). “Investigating the impact of surface wave breaking on modeling the trajectories of drifters in the Northern Adriatic Sea during a wind-storm event” Ocean Modelling, 30, 225-239.