Assimilation of Real-Time Observations for Short-Term Wind Power Forecasting and Renewable Systems Management

Short-term forecasts of winds provided by state-of-the-art numerical weather prediction (NWP) systems are commonly used for wind energy applications and for efficient management of renewable resources. These forecasting tools represent atmospheric conditions at high spatial and temporal resolution, and are based on advanced physical models and a variety of observations. In this study, the WRF model and WeatherBug surface data assimilation are used to provide short-term predictions of winds for lead times between 0 and 24 hours. Real-time observations of wind speed, direction, temperature, humidity and pressure from the WeatherBug surface network, as well as NWS data from local airports, are used in this wind power forecasting system. Observations are passed through rigorous quality control procedures prior to assimilation. The study is focused on a mostly rural area complex topography in the Northern U.S. where there are more than 100 surface sites that are in close proximity to several operating wind farms. Typical wind turbines with hub at 80m height are deployed there. Assimilation of dense real-time surface observations enables more realistic modeling of micro-scale variations in wind power capacity in the area as it becomes important for dynamic resource management. The results of this study show that the coupling of a weather forecasting system with real-time surface observations improves the accuracy of short-term wind power forecasts. Spatially varying impact of data assimilation on forecast accuracy over the selected area is also discussed.