Abstract
In recent years, research into the occurrence of bats at the Dutch North Sea has shown that there is regular seasonal migration over sea. However, so far, little is known about their migration ecology, the fatality risks at offshore wind turbines, and the number of individuals migrating over sea. Since the Dutch government wants to boost the further development of wind energy production in the southern North Sea, the Ministry of Economic Affairs commissioned to Rijkswaterstaat the elaboration of an integrated Wind at Sea Ecological Programme (in Dutch: Wozep). This study, as part of the Wozep-project Behaviour and Collision Risk of Bats (Bats_2), investigates how bat behaviour can be studied near offshore wind turbines. To find out whether it is wise to continue and further develop behavioural research at wind turbines in the context of the Wozep programme, we first conducted a feasibility study on land. A stereoscopic setup consisting of two thermal cameras was devised and used to collect footage of bats in August and September 2016 at the Wind Turbine test Site in the Wieringermeer. The cameras were positioned in such a way that the overlapping field of view in both cameras could be used to determine 3D bat paths at a distance of 80 m around a single wind turbine. In addition, acoustic bat activity was measured with a 12 channel bat detector at three different heights with microphones in each wind direction. Tailor made 3D analysis tools were developed to synchronize both thermal cameras, calibrate the stereoscopic setup, determine 2D tracks in the left and the right view and eventually to reconstruct the x, y, z coordinates of bat positions in time.
The conclusion of this feasibility study is that computer vision components can be used to measure bat flight trajectories in 3D to study the effects of wind turbines on the bat mortality and bat flight behaviour. The bat detector configuration is able monitor bat activity simultaneously at multiple heights in different directions, which can be used for species identification and to assess the accuracy of the estimated bat flight trajectories. During nights with much bat activity several thousand recordings can be obtained and therefore it is necessary to identify these automatically. Before a system can be built for wind turbines at sea, further improvements are needed to mature the current system and meet the goal to automatically detect bat fight paths and fatalities. Before application at sea we recommend to build a prototype on land with multiple stereo setups to monitor the complete area of the rotors and improve and automatize the current 3D analysis tools.