An empirical study supported by the Wind Wildlife Research Fund testing the thresholds used in automated curtailment and successfully predicting eagle entry into rotor-swept zones at multiple wind energy facilities.
Curtailment is the slowing or stopping of wind turbines in order to minimize avian collision fatalities, but results in reduced power generation. When paired with information on an approaching bird’s risk of collision, informed curtailment can help reduce risks for birds while also reducing power loss. IdentiFlight, an automated monitoring systems used to trigger curtailment at wind energy facilities, while better at detecting eagles compared to human observers, can frequently produce false positives where curtailment occurs when no eagle is present. Additionally, with current thresholds for triggering curtailment, about 70% of the curtailments initiated do not result in the eagle entering the rotor-swept zone, thus apparently making curtailment unnecessary. Prior to this study, thresholds used to initiate this process (i.e., flight altitude, distance, relative flight bearing) had not been verified at wind facilities. This study used IdentiFlight eagle data from a single wind energy facility in Wyoming to model eagle collision risk and identify the covariates most important in predicting an eagle’s risk of collision after being detected by the system.. The best fit model, which incorporated flight altitude, distance, relative flight bearing, and flight speed, was further tested on a subset of data at the WY facility and on eagle data from a facility in CA with different weather patterns and topography. The model successfully predicted the likelihood of an eagle entering the rotor-swept zone with better accuracy than current automated systems. This model has the potential to be used across many wind energy facilities in order to more accurately minimize protected species mortality while maximizing the amount of wind power generation.