Characterizing Wind Power Forecast Uncertainty with Numerical Weather Prediction Spatial Fields

Abstract

Successful integration of wind power into power systems can be facilitated through better understanding of future uncertainty in wind power generation. This paper explores a new approach to characterizing this uncertainty using measures of the variability in the wind speeds predicted at multiple grid points in a Numerical Weather Prediction (NWP) system. This approach is compared to the traditional approach of using the spread from an NWP ensemble by using two measures of uncertainty; forecast errors in single time-series forecasts and observed temporal variability. Results show that the multiple grid point approach has a comparable skill level to NWP ensembles for predicting these uncertainty measures and in particular, demonstrates very good skill in predicting large forecast errors. These results also provide a positive evaluation of a terrain standardization method described in a companion paper. A possible extension of this work is to combine the multiple grid point approach with NWP ensembles to improve uncertainty characterization.