Creation of a WRF Ensemble for Improved Wind Forecasts at Turbine Height

The inherent variability of wind requires accurate forecasts to optimize wind power generation. The weather research and forecasting (WRF) model with 10-km horizontal resolution was used to explore improvements in wind speed forecasts at hub height (80m). Results were validated using wind speed measurements at 80 m from a meteorological tower at the Pomeroy wind farm in northwestern Iowa. An ensemble consisting of different planetary boundary layer (PBL) schemes showed little spread between the individual PBL members. A second configuration using three random perturbations of the Global Forecast System (GFS) model produced more spread in the wind speed forecasts, but less model skill. A third ensemble with members having different initialization times showed model spread comparable to that from the perturbation results, but model skill was not compromised. In addition, we examined post-processing techniques such as bias correction of the diurnal cycle, training of the model for the day 2 forecast based on day 1 results, and bias correction based on observed wind direction. Early evaluation suggests that the ensemble mean of the first and third ensembles provides a more skillful wind forecast than any particular member, and further improvements occur when a bias correction of the diurnal cycle is applied to these forecasts.