Posters

Abstract

As wind energy development accelerates worldwide, many more campaigns are using profiling and scanning lidar, and more campaigns are occurring in locations with low aerosol density (“clean air” or “clear sky”). Lidar backscatter signal quality depends on the presence of aerosols advecting in the wind: clean air sites reduce lidar range and data availability, and can lead to increased project uncertainties. Large scale deployment of wind lidar requires (1) simulators that can reliably predict lidar data availability (2) novel techniques to improve range and availability to reduce energy yield assessment uncertainties. In this presentation we present a new range data availability simulator, capable of estimating WindCube and WindCube Scan data availability worldwide based on a physical model of the WindCube and ERA5 reanalysis data. Next, we present a technique to optimize profiling lidar data availability by dynamically switching between traditional wind field reconstruction algorithms (WFR): 4-beam scalar, vector, and hybrid averaging. For each of these traditional WFR techniques, data filtering limits are derived from estimates of 10-minute wind speed uncertainty as a function of carrier-to-noise ratio (CNR) and Data Availability. Line-of-Sight (LOS) uncertainties are measured in the laboratory with a novel fiber optic tool. Precise filtering and dynamic WFR switching allow for 10-15% improvement of data availability in Clean Air conditions. Finally, we present Moustiquaire or MosquitoNet, a new lidar range boosting technique based on a convolultional neural network that recovers low CNR data via 2D images of LOS wind speed versus time. For the profiling WindCube, Moustiquaire boosts Data Availability in clean air conditions: for WindCube Scan, range is improved by 650m on average, and 60m for WindCube profilers, roughly 30% improvements in range across all devices.