Enhancing Wind Power Forecasts via Bias Correction Technologies for Numerical Weather Prediction Model

This study aims to improve wind power forecasts using the bias correction technologies for NWP wind speeds, which proposed a suitable post processing strategy to correct NWP outputs. In this study, three NWP models (Radar Weather Research and Forecasting (RWRF), Deterministic Weather Research and Forecasting (WRFD), and a 20-member WRF ensemble prediction system (WEPS)) provided by Taiwan Central Weather Bureau (CWB) were systematically bias corrected using decaying average (DA) and probability matching mean (PMM) to improve wind-speed forecasts. With the bias correction of NWP products, this study performed hourly-ahead wind power forecasts using machine learning (Long short-term memory (LSTM), Gated Recurrent Unit (GRU), and eXtreme Gradient Boosting (XGBoost)) and deep learning (Transformer) models. The results demonstrate that the forecasting accuracy after the bias correction of NWP wind speeds was improved. Compared with the experiments without bias correction, the RMSE for wind speed in RWRF, WRFD and WEPS during the validation months are decreased by 46.13%, 5.84% and 3.82%, respectively, and the accuracy of wind power prediction is improved as well. In this regard, the necessity of NWP correction in a typical forecasting process is verified.