Enhancing minute-scale lidar-based power forecasts of offshore wind farms towards an operational use

Abstract. Minute-scale power forecasts are gaining importance to support the integration of volatile wind power in particular for offshore wind farms with geographical concentration of generation capacity. Lidar-based approaches have proven useful as an alternative to statistical methods, however, their forecast horizon needs to be extended from currently 5 minutes to at least 15 minutes to be useful for end-users. In this work, we utilize data from an extensive offshore measurement campaign and adapt a lidar-based forecasting approach, e.g. to include wind profile and lidar inclination measurements for improved tilt correction and vertical extrapolation of wind speed, to forecast wind speed and power with horizons of up to 30 minutes. We evaluate individual turbine wind speed and power forecasts and compare them against the benchmark persistence. Further, the impact of forecast characteristics on the forecast skill are analysed. Our results revealed the lidar-based forecast's ability to outperform persistence up to a 16 minute forecast horizon during unstable conditions. An increased wind vector age and propagation duration was found to reduce the forecast skill. Wind farm power forecasts are analysed neglecting large propagation durations, which increased the forecast skill and forecast horizon for which persistence was outperformed. We discuss the applied wind speed extrapolation approaches, the impact of lidar trajectories and wind vector propagation on forecast skill and the value of lidar-based minute-scale power forecast for end-users. In conclusion, the skill of the lidar-based approach highly depends on atmospheric conditions and the forecast characteristics. When considering this for its operational use, the lidar-based forecast has the potential to improve wind farm power forecasts for relevant forecast horizons.