Abstract

Wind energy prediction requires regional weather modeling to cope with intra-hour, multi-hour and day-ahead decision making. In addition, microscale modeling is needed to support wind turbine sitting decisions and turbine operations. In the last 10 years, the U.S. National Center for Atmospheric ... Show moreWind energy prediction requires regional weather modeling to cope with intra-hour, multi-hour and day-ahead decision making. In addition, microscale modeling is needed to support wind turbine sitting decisions and turbine operations. In the last 10 years, the U.S. National Center for Atmospheric Research has developed a Real-Time Four-Dimensional Data Assimilation (RTFDDA) and forecasting system to support diverse weather-critical applications such as wind energy forecasting. The RTFDDA, built upon the Weather Research and Forecasting (WRF) model, is a rapidly cycling forecasting system that effectively combines all available weather observations with the full-physics WRF model to produce high-accuracy multi-scale 4D weather information from the synoptic scale to the microscale (< 2 km). Recently, an LES version of RTFDDA has been employed, which allows successive downscaling from synoptic-scale numerical weather predictions (based on global models), to regional weather predictions (mesoscale weather processes), and to smaller microscale weather modeling with LES models. The WRF RTFDDA LES system, with six nested domains having grid sizes of 30, 10, 3.333, 1.111, 0.369 and 0.123 km, is being used to model intra-farm wind variations for a large wind farm in northern Colorado, with 274 wind turbines. A multiple-day simulation is presented to illustrate 1) the model’s ability to simulate the rich intra-farm wind features, as verified against the high spatial and temporal turbine-hub-height wind speed observations, 2) the LES model’s potential use in the future for operational real-time prediction, as computing power grows, 3) the importance of a two-way nesting approach for the mesoscale and LES grids, 4) the limitations of mesoscale models that make use of parameterized boundary-layer mixing schemes, and 5) the need for further understanding the transition from full parameterization of large eddies in ~1 km grid-increment models to the explicit modeling of the large eddies in domains with grid increments of 10s of meters. Show less