Abstract
Large-eddy simulations of the atmospheric boundary layer have been performed over a range of stabilities between neutral and free convective conditions. The variation of various non-dimensionalized turbulence statistics over this stability range is presented and the results are compared with observations where possible. The robustness of the model results is also assessed by comparing those from high and low resolution simulations, and by reference to a number of additional sensitivity tests. The simulation results for the variation with stability of the mean wind and temperature profiles and various similarity coefficients are presented. The large-eddy model datasets are then used to evaluate critically the performance of a number of simple closure schemes suitable for use in boundary layer parametrizations in large-scale weather forecasting and climate prediction models. The potential significance of the shortcomings of the simplest mixing length schemes is discussed, and an assessment is made of the types of closure most likely to give a significant improvement in performance without an excessive computational overhead. Results are also presented from large-eddy simulations of the baroclinic boundary layer. The effects of the shear in the geostrophic wind on scaled turbulence statistics and the mean wind profiles are discussed. It is shown that this shear does not lead to significant degradation of the performance of two simple closure models, in either neutral or convective conditions. Finally simulation results for the entrainment flux at the top of the boundary layer are presented. A parametrization of this flux is developed, based on the boundary layer root mean square vertical velocity.