An assessment of turbulence closure schemes based on ocean observations and large eddy simulations

Various turbulence closure schemes have been developed and used to model upper-ocean processes. Most of such schemes are based on the Reynolds-averaged Navier-Stokes (RANS) equations and tuned according to ocean observations or laboratory experiments. For RANS methods, the effect of wave-driven Langmuir turbulence (LT) is usually not explicitly included in regional and global ocean models, although recently modifications with explicit LT effects have been proposed. We assess the performance of several turbulence closure schemes, based on two well-studied observed oceanic test cases. The first test case is based on a wind event, during which the mixed layer rapidly deepened. The second test case is based on a heating event with weak winds and swell waves, during which the upper ocean did not restratify. Both cases have been previously investigated in detail using large eddy simulations (LES) with and without LT effects. Those LES results are only consistent with observations if the LT effect is included. For the wind event, RANS models that have been tuned to ocean observations agree better with observations and LES with LT. For the heating event, none of the RANS models without explicit LT effects reproduce the inhibited restratification. Our results highlight the importance of properly representing sea-state dependent LT dynamics in upper ocean RANS models.