Modeling study of turbulent mixing over the continental shelf: Comparison of turbulent closure schemes

[1] The sensitivity of model-produced time-dependent wind-driven circulation on the continental shelf to the turbulent closure scheme employed is studied with a two-dimensional approximation (variations across-shelf and in depth) using the Princeton Ocean Model. The level 2.5 Mellor-Yamada closure (MY), k-epsilon closure, and K-Profile Parameterization schemes are used to evaluate the mesoscale fields and the spatial and temporal variability of mixing. All three submodels produce similar features in the mesoscale circulation. They produce qualitatively similar eddy diffusivities and eddy viscosities, although the turbulent structure and the mixing intensities can differ quantitatively. The k-epsilon length scale follows the buoyancy length scale when stratification is important. In contrast, the length scale produced by the q(2)l equation in the MY scheme deviates substantially from the buoyancy scale unless a stratification-dependent limitation is imposed. During upwelling-favorable winds, the majority of turbulent mixing occurs in the top and the bottom boundary layers and in the vicinity of the vertically and horizontally sheared coastal jet. Turbulent mixing in the coastal jet is primarily driven by shear-production. The near-surface flow on the inner shelf becomes convectively unstable as wind stress forces the upwelled water to flow offshore in the surface layer. During downwelling-favorable winds, the strongest mixing occurs in the vicinity of the downwelling front. The largest turbulent kinetic energy and dissipation are found near the bottom of the front. Turbulence in the bottom boundary layer offshore of the front is concentrated between recirculation cells which are generated as a result of symmetric instabilities in the boundary layer flow.