Design-based inference and data integration allow the efficient estimation and mapping of onshore wind turbines presence across large spatial scales

Wind energy development is a major driver of change for terrestrial ecosystems. As wind turbines are seldom mapped on a regular basis, conservationists increasingly use aerial/satellite images to address this gap. Nevertheless, the manual identification of turbines is labour intensive, preventing conservationists from mapping wind energy development across large spatial scales. In this study we adopted a design-based sampling approach to render sustainable this effort for estimating the total number of onshore wind turbines in Sardinia (Italy) and to map their spatial distribution from high-resolution aerial images (1:500). We also adopted data integration to incorporate previous knowledge on wind turbines presence, achieved from an opportunistic survey, to improve the quality of our estimate and map. We finally estimate the precision of the total estimate and map from a pseudopopulation bootstrap procedure that exploits the estimated map as a pseudopopulation. We estimated a total of 1168 turbines with a relative standard error of 0.7%, and a bootstrap 0.95 confidence interval of 1155–1181 turbines. We also provided a map that estimated a very scarce presence of turbines outside the area covered by the opportunistic survey. The resulting map will be crucial to identify overlaps between wind turbines and biodiversity hotspots, as well as to study spatiotemporal patterns of wind energy development.