Evaluation of Turbulent Transport and Dissipation Closures in Second-Order Modeling

We show that the turbulence statistics from our (96)3 large-eddy-simulation (LES) studies of a convective boundary layer are in excellent agreement with those from the Deardorff–Willis laboratory convection tank. Using these LES data, we evaluate contemporary parameterizations for turbulent transport and dissipation in second-order closure models of the convective boundary layer. The gradient-diffusion parameterization for turbulent transport fares poorly, due in large part to the direct influence of buoyancy. This leads to poor predictions of the vertical profiles of some turbulence statistics. We also find that the characteristic length scales for the mechanical and thermal dissipation rates typically used in second-order closure models are a factor of 2–3 too small; this leads to underpredictions of turbulence kinetic energy levels. Finally, we find that the flux and variance budgets for conservative scalars are substantially different in top-down and bottom-up diffusion. In order to reproduce these differences accurately, it seems necessary to model the turbulent transport, pressure covariance, and molecular destruction terms differently in top-down and bottom-up diffusion.