Operational wind power forecasting, turbine micrositing, and turbine design require high-resolution simulations of atmospheric flow over complex terrain. The use of both Reynolds-Averaged Navier Stokes (RANS) and large-eddy (LES) simulations is explored for wind energy applications using the Weather Research and Forecasting (WRF) model. To adequately resolve terrain and turbulence in the atmospheric boundary layer, grid nesting is used to refine the grid from mesoscale to finer scales. This paper examines the performance of the grid nesting configuration, turbulence closures, and resolution (up to as fine as 100 m horizontal spacing) for simulations of synoptically and locally driven wind ramping events at a West Coast North American wind farm. Interestingly, little improvement is found when using higher resolution simulations or better resolved turbulence closures in comparison to observation data available for this particular site. This is true for week-long simulations as well, where finer resolution runs show only small changes in the distribution of wind speeds or turbulence intensities. It appears that the relatively simple topography of this site is adequately resolved by all model grids (even as coarse as 2.7 km) so that all resolutions are able to model the physics at similar accuracy. The accuracy of the results is shown in this paper to be more dependent on the parameterization of the land-surface characteristics such as soil moisture rather than on grid resolution.