Improvement of the Ocean Mixed Layer Model by Using Large Eddy Simulation and Inverse Estimation

Examination of the pattern of bias in the ocean mixed layer model (OMLM) (Noh model) prediction under different conditions from the comparison with large eddy simulation (LES) data reveals that mixed layer depth (MLD) under convection can be over and underestimated depending on the wind stress. It is found to be associated with the bias in the relative magnitude of turbulent kinetic energy (TKE) budget terms in the OMLM, which in turn depend on various empirical parameters in the model. Sensitivity tests are carried out for the coefficients in the TKE flux, the length scale, and the Prandtl number on the OMLM in order to investigate their effects on the TKE budget terms. The effect of parameterization of convection, such as convective adjustment and non-local mixing, is also examined. The inverse estimation using model Green's functions with constraints of LES data is then applied to obtain the simultaneous optimization of various empirical constants in the OMLM. The performance of the optimized OMLM is evaluated in comparison with the original Noh model and other OMLMs.