The objective of this project is to refine and improve the methods of Large Eddy Simulation (LES) to enable the incorporation of turbulence and land-surface effects in mesoscale computer models of atmospheric flow. The work includes representation of the influence of the land surface on atmospheric processes, definition and representation of local and regional forcing of atmospheric motion, resolution of key energy-transporting phenomena, and evaluation of the accuracy of the simulations. The major task is to link (small-scale) turbulence models to (larger-scale) mesoscale models. For this, two mesoscale computer codes are employed: the Advanced Regional Prediction System (ARPS) and the Weather Research and Forecasting Model (WRF). Linking these to turbulence models requires accounting for surface boundary conditions and for velocity fluctuations smaller than the computational grid. Accurate representation of the land surface is achieved through use of a highly resolved land-surface module that characterizes many combinations of vegetation and soil type. Turbulence models are extended to the larger domain using recently proposed algebraic stress models to account for the effects of motions smaller than the grid size. Two-way interactive nesting grids enable improved simulations of the interaction of airflow with the terrain. The ultimate aim is to develop a detailed methodology for improving local and regional weather predictions by accurately simulating the effects of small-scale phenomena within the larger-scale models. Other broad impacts of the research include improved ability to predict the dispersion of pollutants, as well as the capacity to simulate flow in a given setting to provide guidance for the design of micrometeorological field experiments.