Abstract
Collision of birds with wind turbines is an important negative effect of wind energy generation. Assessments of the potential numbers of bird collisions are required prior to the construction of wind farms. Collision rate models (CRMs) are used as a tool to estimate numbers of collision victims for wind farm initiatives.
In the past couple of decades various CRMs have been developed. These models are all based on the theoretical calculation of collision probabilities (theoretical models).
In this paper we introduce an empirical model, the Flux Collision Model (FCM), in which actual knowledge of species (group)-specific collision probabilities collected in existing wind farms on land is used to calculate collision rates for planned wind farms.
An important quality of the FCM is that it provides a means to use empirical information to assess collision rates in Environmental Impact Assessments (ETAS) for wind farm initiatives. In addition, no detailed information on bird behaviour close to the rotor is needed, as this information is already incorporated in the empirical collision probability.
In two case studies, one offshore and one on land, we compare and discuss the use and performance of the empirical FCM and the theoretical SOSS Band model for predicting collision rates of birds at wind farm initiatives. To date, no actual collision rates are known for the offshore situation. Accordingly, in the FCM, collision probabilities derived from wind farms on land were used. Nevertheless, in the offshore case study, the results of the FCM were comparable with those of the SOSS Band model.
Basic sensitivity analyses for both the FCM and the SOSS Band model showed that purely theoretically both models are equally sensitive to changes in avoidance rates. However, because lower values for avoidance are applied in the FCM (wind farm avoidance) than in the SOSS Band model (overall avoidance), in practice the effect of realistic variation in avoidance rates on the resulting collision rates is much smaller for the FCM than for the SOSS Band model.
Our results show that the FCM provides a valuable addition to the existing suite of (theoretical) CRMs. The predictive value of the theoretical SOSS Band model is constrained by the limited availability of knowledge on species (group)-specific (wind turbine) avoidance rates, which is not the case for the FCM. By contrast, the reliability of the empirical FCM is determined only by variation in the availability and quality of information on species (group)-specific collision probabilities.
The choice of which CRM to use (theoretical or empirical) seems not to depend on the location of the wind farm initiative as being offshore or on land, but on the availability and reliability of species (group)-specific information in existing wind farms. The availability of a reliable collision probability supports the use of the FCM, while the availability of information on (overall) avoidance rates in the absence of a species (group) specific collision probability supports the use of a theoretical CRM like the SOSS Band model.
Synthesis and applications. The predictive power of collision rate models relies in the first place on the quality of the input information, and second on the theoretical details of the model calculations. Although the FCM is less dependent on measurements of avoidance rates, an urgent need remains to obtain information on actual collision rates and corresponding collision probabilities as well as avoidance rates in existing wind farms both offshore and on land in order to accurately determine the impact of wind energy on bird populations.