The proposed talk will report on ongoing development of a new turbulence parameterization aimed at use within mesoscale models. The basic structure is that of Mellor and Yamada's Level 2.5 model, but with the turbulence length scale ("l") computed via a prognostic equation for the dissipation rate ("epsilon") of turbulent kinetic energy. This, in principle, is an improvement over typically used algebraic formulations for "l", which are almost always designed in reference to equilibrium one-dimensional atmospheric boundary layers (ABLs). The epsilon-equation, on the other hand, since it properly recognizes the effects on "l" of memory, transport, and local sources and sinks, provides a more general parameterization that, for example, does not depend on "a priori" 1-D boundary layer assumptions. This makes the equation attractive for use in computing in complex mesoscale flows. The epsilon-equation employed is modified by us from its standard engineering form by making one of its closure parameters, c_{epsilon 1}, a function of gradient Richardson number. The form of this functional dependence is derived by matching the equation to Monin-Obukhov theory for the unstably and stably stratified atmospheric surface layer.

During the talk, we will show results from a series of one-dimensional computations of the neutral and stable atmospheric boundary layers (NBL and SBL, respectively) employing the new parameterization. These computations represent initial testing of the parameterization to assure that its predictions relax to classical 1-D ABL structure with appropriate external forcing. This relaxation is not achieved when the standard engineering form of the epsilon-equation is retained, which seems to have been the biggest reason for the lack of common use of the equation in mesoscale models to date. Prelimanary model predictions using our modified epsilon-equation show encouragingly good agreement with field data as well as with direct numerical and large-eddy simulation results of the NBL and SBL. Current work is being carried out on similar 1-D computations of the convective boundary layer, for which we also hope to present results at the upcoming meeting.