Power production from wind turbines can vary significantly from manufacturer's ratings due to atmospheric stability and wind shear. In this study, remotely sensed and in-situ data from a wind farm in the High Plains of Central North America were examined to quantify the effects of atmospheric conditions in the boundary layer on power generation. Several approaches for segregating time periods by atmospheric conditions were applied to this dataset, including methods based on the time-of-day, the power law exponent α, the bulk Richardson number R_B, and diurnal cycles in wind and temperature. These classifications were used to generate stability-dependent power curves. For this site, all classification metrics indicated underperformance during stable/night regimes and overperformance during convective/day regimes at moderate wind speeds (7-12m/s). A simple attempt at forecasting power production values proved both the feasibility and the utility of applying meteorological classifications for forecasting applications. The success in diagnosis and forecasting of power production using boundary layer data demonstrate that power output is strongly influenced by boundary layer stability, but further research is required that involves measurements taken across the rotor-disk; remote sensing of such profiles is recommended.