An autonomous GPS geofence alert system to curtail avian fatalities at wind farms

Wind energy developments continue to proliferate globally as nations seek clean and renewable alternative energy sources to fossil fuels. However, wind farms do not come without environmental costs [1]. A growing literature is documenting the serious impacts that wind farms can have on resident and migratory avifauna populations through mortalities from direct collisions with turbines [2, 3]. Recent estimates indicate that wind farms in North America are responsible for up to 368,000 bird fatalities annually [4]. Consequently, the development and implementation of effective measures to reduce wind energy impacts on wildlife is recognized as a top priority by biologists, conservation organizations, regulators and the private sector (see reviews by [5–7]).

Recent rapid advances in the miniaturization, accuracy and utility of biotelemetry devices [8] offer technological solutions to the challenge of reducing bird fatalities at wind farms. Our aim was to successfully develop and demonstrate a biotelemetry system that would provide automated alerts to select recipients if a telemetered bird flew within proximity of a wind farm so that appropriate measures could be taken in sufficient time to minimize the risk of collision injury. We built in and programmed real-time virtual fence (geofence) technology into prototype avian GPS-tags to provide this early-warning capability for large bird species such as eagles, vultures and cranes that have been documented suffering collision injuries from wind turbines [3, 9]. Golden eagle populations in North America appear to be declining [10], and Pagel et al. [11] found a minimum of 67 golden eagle mortalities at 32 wind energy facilities in 10 states between 1997 and 2012. Global populations of vultures are declining precipitously from anthropogenic threats [12], and recent studies indicate that mortalities from wind turbine collisions are an increasing impediment to their recovery [13]. The geofence system offers a unique technological solution to mitigate wind turbine collisions with these threatened species.

A geofence is a virtual boundary delineated around an area of interest that triggers: (1) A cue to the telemetered animal (e.g. electric shock); (2) a change in the location fix rate attempted by the unit, or; (3) an alert to managers whenever the animal crosses the boundary edge. Geofences are increasingly recognized as an effective platform to enhance the spatiotemporal flexibility of wildlife management. For example, geofences have been successfully integrated into the management of mammalian populations that come into conflict with or are disturbed by human activities, such as elephants and wolves (see review by [14]). However, until now geofence alert technology has been prohibitively too large or too complex to incorporate into avian biotelemetry. We report on the first autonomous GPS geofence alert system custom developed specifically for avian applications with a view to minimizing the risk of bird collisions with wind turbines.

The geofence tags provided highly accurate location data and transmitted SMS alert messages within a few minutes of crossing a virtual geofence boundary. The planar 2D accuracy of the two test GSM-GPS tags deviated from the DGPS flight path by a mean of 4.4 m (±0.5 SE) and 2.1 m (±0.6 SE), respectively. The vertical 3D accuracy of the test unit locations deviated by 27.2 m (±0.1 SE) and 42.6 m (±0.6 SE). Because the GPS fix rate had to first change from the normal 15-min interval to 30 s once the tags crossed the first outer virtual boundary, it took longer to receive an SMS alert after the tags first crossed the outer geofence compared to when they crossed the inner second geofence. Hence, the timing of the initial SMS alert could range from 1 to 15 min depending on where the tag was during its fix cycle. The mean time that we received an SMS alert after we flew two the tags across the outer geofence was 9:14 min (±1 min SE). The mean time that we received an SMS alert after we flew two the tags across the inner geofence was 1:17 min (±16 s SE) (Fig. 4). Consequently, we recommend that users install a minimum of two geofences around a wind farm, with an outer zone acting as a buffer to trigger the higher location fix rate function and corresponding reduction in alert message delays. We also received SMS alerts from the units on average a minute faster when we flew across the geofence boundaries compared to when we drove the units across the geofences. The mean time that we received an SMS alert after we drove two the tags across the inner geofence was 2:18 min (±37 s SE) compared to the mean alert time of 1:17 min (±16 s SE) when we flew. Higher altitudes may quicken response times by providing better access to the GPS satellite constellation and cellular communication networks.

A condor flying at the 15 km/h mean flight speed recorded from the tracked Baja population and flying directly towards the ESJWP site would take around 2 h to reach the closest wind turbine after it crossed the outer boundary of the 40 km geofence zone and triggered an alert. The condor that flew across the ESJWP site in 2007 would have taken between 3 and 4 h to reach the closest turbines after it flew across the 40 km geofence boundary. If this bird had flown directly towards the ESJWP at its maximum recorded flight speed of 47 km/h for its entire flight path it would have taken 43–46 min to reach the turbines after it had triggered the first alert. In the unlikely event that a telemetered condor flew directly towards the ESJWP site at a sustained speed of 100 km/h it would still not reach the closest wind turbine until 20–22 min had passed since it triggered the first geofence SMS alert.