A combined modelling system for short-term wind power forecasting based on mesoscale Numerical Weather Prediction

The study presents the preliminary results of a short term (from 6 hours to 3 days) forecasting system of the hourly production from a wind power source with assigned technical and dimensional characteristics, starting from the hourly forecasts of wind speed and direction. The wind forecast is performed by the Weather Research and Forecast (WRF) mesoscale model on four 10-day periods in 2018, providing estimates for the next 72 hours every 6 hours. Weather forecasts are used for estimating wind power from wind data using a sigmoid wind turbine power curve. The system is validated considering wind observations and wind energy production data at an operating wind farm, leading to normalized average absolute daily errors of about 5%.