Abstract
This report presents data on infrared monitoring investigations by use of Thermal Animal Detection System (TADS) on autumn migrating waterbirds at the Nysted offshore wind farm, Denmark in 2004.
The aims of the report were twofold:
- to collect data on the number of waterbird collisions and on the near rotor evasive behaviour using TADS, and
- to gather information (data collected by TADS and radar) to develop a deterministic predictive collision model in order to estimate the number of Common Eiders Somateria mollissima which collide with the sweeping rotor blades of the 72 wind turbines.
The results from the collision monitoring study confirm the findings from the same site in spring 2004, when a relatively low migration volume around the near vicinity of the turbines was also documented. During autumn operation, the TADS recorded 1,944 thermal video sequences automatically at one turbine, of which five were triggered by birds passing the field of view. No birds were recorded as passing the sweep area of the rotor-blades nor colliding with any part of the turbine during the 28,571 minutes (equivalent to 476 hours) of monitoring.
A single passerine was observed approaching the rotor-blades, and ceased its onward flight hovering on its wings before it returned in the direction it came from. The remaining five sequences showed three flocks of passerines and two flocks of waterbirds passing within the near vicinity of the turbine but beyond the reach of the rotor blades.
Hence, out of six events four were passerines passing the field of view of the TADS, and this despite the fact, that the present monitoring scheme was designed for measuring waterbird collisions. This demonstrates that the TADS can evenly well be used for monitoring passerines as waterbirds, especially if a larger telephoto lens is applied.
The values, which were imputed in a collision model, were obtained partly from the conclusions of the present study and from the literature. The model estimated that on average 68 Common Eiders would collide with the turbines in one autumn season, with a range of 3 to 484 individuals. The estimated average number of collisions of 68 individuals lie within range of the published estimates from the literature.
The model in its present form, as a deterministic model, must be characterized as a preliminary solution. Before the preferred stochastic approach can be applied, enabling the variance of the data of the input parameters to be incorporated in the final collision estimate, the last radar data collected in 2005 will have to be included.