Towards Improved Wind Turbine Clutter Detection Using Real-Time Turbine Operational Data

Abstract. Weather radar is essential for precipitation monitoring, nowcasting and severe-weather warning. However the increasing deployment of wind turbines near radar sites increasingly compromises operational data quality. Rotating turbine blades enhance Doppler spectral noise floor, which can be used for automated clutter detection, but the detectability of wind turbine clutter (WTC) depends on turbine operation and radar-turbine viewing geometry. This study investigates the drivers of WTC detection using one year operational radar observations and two months spotlight-mode scans of an individual turbine from the German Meteorological Service (DWD, Deutscher Wetterdienst) weather radar at Boostedt, together with operating data from twelve turbines at Gönnebek at 30 s resolution. The available turbine operating data correlate well with the observed WTC at a temporal resolution of 5 min. This sampling interval is sufficient to capture the relevant temporal variability of WT operation data, whereas aggregation to 10 min or longer increasingly reduces temporal information. The detection algorithm is largely insensitive to precipitation compared to clear sky condition. Rotor speed is the dominant control on detectability while radar-relative nacelle orientation provides an additional, site-dependent contribution. Front- and back-view geometries yield broader Doppler signatures and more reliable detection, whereas side-view echoes often occur as short-lived spectral flashes that routine scans and therefore the WTC detection algorithm may miss. Blade pitch angle has little independent explanatory power once rotor speed is considered. Future assessments should account for turbine operation and radar-beam geometry in addition to distance from the radar. Extending detection to turbine-affected dual-polarimetric variables is needed because their impact may extend beyond the area identified by the current WTC-Flag.