Assessing avoidance rates of seabirds at offshore wind farms from aerial digital images
Mark Rehfisch, Stuart Clough
APEM, Stockport, UK
The avoidance rate is the key input variable in collision risk models (CRM) requiring a three-dimensional empirical analysis of the spatiotemporal distribution of birds relative to individual wind turbines (Chamberlain et al. 2006). However, the number of potential seabird collisions resulting from CRM has frequently been based on precautionary, possibly unrealistically low, avoidance rate estimates. Collision rates are therefore likely to be overestimated in current environmental impact studies. These inflated mortality estimates are often accepted by decision makers, making wind farm consent more difficult. Here, we present a novel approach for estimating seabird macro- and micro-avoidance of offshore windfarms from high-resolution digital images from aerial survey.
We present results from a case study based on four post-construction aerial surveys of the Greater Gabbard offshore windfarm (GGOWF) carried out in 2014 during a period of high Northern gannet (Sula bassana) autumn passage off the East Anglian coast in the southern North Sea. Due to the widespread use of potentially precautionary avoidance rates gannet has become a major consenting risk for offshore windfarms in the North Sea and elsewhere.
This approach calculates macro- and micro-avoidance through the statistical analysis of the spatial distribution of birds in flight relative to the turbine positions. The method makes it possible to gather information on bird distribution and individual approach distances to offshore wind turbines over a wide spatial scale and in three dimensions.
In total 336 gannets were recorded during the four surveys of which eight individuals were recorded within and 328 individuals outside the wind farm footprint. A zero-inflated negative binomial model was used to describe the relationship between the distance to the nearest turbine and gannet counts outside of the wind farm footprint. Gannet numbers increased significantly from zero close to the turbines to background density levels 2 km away from the nearest turbine. A macro-avoidance value of 95.02% was calculated for gannets as the percentage change from their background outside of the wind farm area compared to their density within the wind farm footprint. Observing no birds closer than 359 m to a turbine suggested 100% micro-avoidance and an overall avoidance value of 100%.
In conclusion, it is not unreasonable to suggest that an avoidance rate of 99.5%, at least for gannets on autumn passage, may be appropriately precautionary for use in CRM.
Our digital aerial survey approach can capture the avoidance behaviour of seabirds efficiently and rapidly over a wide spatial scale. Results arising from the use of this method will contribute to reducing uncertainty in predicting impacts from offshore wind farms on seabird populations at both a local and national level.