Bird Collision Avoidance: empirical evidence and impact assessments.

Description

Driven by ambitious renewable energy targets, the offshore wind industry has experienced a significant growth over the last 20 years, particularly in Europe, where most of installed capacity can be found. The consenting of offshore wind projects requires the identification, prediction and evaluation of environmental effects of project proposals. Seabirds are known to interact with offshore wind farms (OWF), mostly with negative consequences. Birds may collide with turbines, be displaced from their feeding grounds or wind farms may act as barriers to seabird movement (Drewitt & Langston, 2006; Johnston et al., 2014; Krijgsveld et al., 2011; Vanermen et al., 2013). JNCC, Scottish Natural Heritage, Natural England, Natural Resources Wales and the Department of Agriculture, Environment and Rural Affairs (Northern Ireland) as statutory nature conservation bodies (SNCBs), provide advice to Government regulators on whether a proposed OWF development is likely to have an adverse impact on seabird populations. In order to assess the number of birds which might collide with structures, collision risk modelling is usually undertaken as part of the impact assessment process. Various collision risk methodologies and models are available (Masden and Cook 2016). In the UK the most widely used model is the Band (2012) collision risk model. Band (2012) relies on input parameters describing the turbine design, bird morphology, and bird ecology and behaviour. Studies have shown that Band (2012) is particularly sensitive to the avoidance rate input parameter which describes an aspect of bird behaviour (Masden 2015). The avoidance rate parameter is applied towards the end of the Band (2012) modelling process in order to adjust the 'otherwise' predicted number of collisions for actions taken by birds in order to avoid turbines. Such avoidance can occur at several scales. Avoidance rates for OWFs have usually been estimated based on comparisons of observed collisions with those that would be expected in the absence of avoidance behaviour and as such in practice include other uncertainties and variabilities associated with the modelling process. Avoidance rates have been summarised and combined in Cook et al (2014).