Remote-sensing-based approaches have shown potential for minute-scale forecasting and need to be further developed towards an operational use. In this work we extend a lidar-based forecast to an observer-based probabilistic power forecast by combining it with a SCADA-based method. We further aggregate individual turbine power using a copula approach. We found that the observer-based forecast benefits from combining lidar and SCADA data and can outperform persistence for unstable stratification.

Very short-term wind power forecasts are gaining increasing importance with the rising share of renewables in today's energy system. In this work, we developed a methodology to forecast wind power of offshore wind turbines on minute scales utilising long-range single-Doppler lidar measurements. The model was able to outperform persistence during unstable stratification in terms of deterministic and probabilistic scores, while it showed large shortcomings for stable atmospheric conditions.

This paper focuses on the use of scanning lidars for very short-term forecasting of wind speeds in a near-coastal area. An extensive data set of offshore lidar measurements up to 6 km has been used for this purpose. Using dual-doppler measurements, the topographic characteristics of the area have been modelled. Assuming Taylor's frozen turbulence and applying the topographic corrections, we demonstrate that we can forecast wind speeds with more accuracy than the benchmarks persistence or ARIMA.