Presentations

Abstract

Lidar are widely deployed as part of wind applications worldwide. Lidar measurement depends on the presence of sufficient aerosols in the air which serve as the backscatter target for the lidar beam. Poor aerosol density can lead to reduced lidar data availability, thus increasing uncertainty of wind resource assessment (WRA), increasing the duration of power performance tests, or limiting the range of offshore scanning lidar met campaigns. These can ultimately contribute to increased project financing costs in extreme cases. Innovations to boost lidar data availability are essential for universal use of lidar for wind energy applications. In this study, deep learning quality control (DLQC) algorithms are applied to several years of lidar data and evaluated against IEC-compliant met masts. Another class of novel range boosting algorithms using dynamic selection of traditional wind field reconstruction algorithms (D3LOS) are demonstrated in parallel. For both novel algorithms, uncertainties of the recaptured data are precisely quantified against a 200m IEC-compliant lidar validation met mast in Germany, and at a 140m met mast in South Africa during low-aerosol density events. The analysis demonstrates substantial lidar data availability improvements across both sites, with superior availability gains coming from DLQC. The associated uncertainties in wind speeds generated with recaptured, lower-quality data are well-behaved, and in line with theoretical quantifications of lidar uncertainty. These new lidar algorithms are a significant improvement over traditional filters and can be applied to any WindCube lidar, profiler, Nacelle, and Scanning. These innovations improve the reliability and flexibility of lidar as a replacement for met masts, and help accelerating the transition to renewable energy.