The Navy's Coupled Ocean/Atmospheric Mesoscale Prediction System (COAMPS) rou-tinely uses Louis's surface flux parameterization with added "gustiness" effects driven by large turbulent eddies. A unique feature of this parameterization is that the flux does not need to be solved iteratively which is usually necessary for a full Monin-Obkhov similarity formulation. In recent years, many parameterization schemes have been proposed to appropriately account for the gustiness factor. This study first evaluates the modified Louis's scheme and compares it with TOGA COARE parameterization described by Fairall et al. and then attempts to address the is-sue of scale dependence of the gustiness factor as applied in a mesoscale model.

It is found that the current "gustiness" factor in the COAMPS's surface parameteriza-tion is significantly larger than that in COARE scheme and what is implied by many observa-tions. As a result, the stability effect is significantly suppressed at low wind regimes. This feature leads to a relatively weak sensitivity of the scheme to the atmospheric stability at the surface. To correct this unrealistically large "gustiness" effects and preserve the simplicity of Louis's scheme at the same time, a different coefficient is introduced to the gustiness formulation. The results simulated by the COAMPS with the modified scheme are encouraging.

Because the "gustiness" represents the effects of subgrid wind variability driven by large turbulent eddies in a numerical model, it may strongly depend on grid resolution of the model. A mesoscale model is sometimes run at very fine resolution (e.g., 1km) and thus can partly resolve the turbulent eddies that produce the "gustiness". Clearly, for these types of simulations, the pa-rameterized gustiness effects should be reduced. This issue is investigated using data from a nu-merically simulated low-wind convective boundary layer by a large-eddy simulation model. The results suggest that the coefficient for the "gustiness" should be dependent on the grid scale and the scale of large-eddy size. Consequently, a simple formulation is developed and tested in the COAMPS.