Abstract
As part of its license conditions for the Horns Rev 2 offshore wind farm (HR2) DONG Energy A/S is obliged to undertake post-construction monitoring of bird migration. This report contains the results of the monitoring on bird migration carried out during September 2010 to May 2012. The post-construction monitoring followed up on the baseline program during 2008. The purpose of the post-construction monitoring was to monitor possible impacts of the operation of Horns Rev 2 on bird migration, including collision risk, barrier effects and cumulative effects as a consequence of Horns Rev 1(HR1) and Horns Rev 2 being located in the same marine region. The methods applied in the post-construction monitoring have as far as possible been based on the same field methods as used during the baseline, which again were based on the methods developed and applied during the PSO 1 monitoring program for the HR1 wind farm and the investigations carried out as part of the EIA for HR2.
This implied using horizontal radar at HR1 and HR2 with the same technology as during the baseline, while the species-specific monitoring has been based on radar- and rangefinder-based tracking methods capable of collecting species-specific data on behavioural reactions of migrating birds to HR2 and HR1 for the calculation of avoidance and collision rates. In order to assess the relative importance of migration intensities offshore visual observations were also undertaken at Blåvandshuk during the same observation periods as on HR1 and HR2.
In total 1,785 species-specific tracks were recorded, of which 1,047 were 3D tracks made by laser rangefinder and 738 were 2D tracks made by radar. The majority of the tracks recorded by both devices were of Common Scoter (55.1 %). With these data, the data collected during the baseline in 2008 and the investigations related to HR1 the knowledge of spatio-temporal and directional trends in the movements of birds along Horns Rev has now reached a level which enables generalisations concerning bird migration to be made for the whole region. These generalisations include differences and similarities in the composition of species and functional groups along Horns Rev.
In an attempt to generalise the flight patterns and behavioural reactions to the wind farms of different bird species altitude and flight regression models were developed using a generalised additive mixed model design. The models successfully incorporated weather-induced variation in flight patterns and species-specific responses to the wind farms, and made it possible to assess and predict likely effects in terms of collision risks and barrier effects, including cumulative effects, for the main species of birds occurring in the region. The models and the predictions show general flight patterns in relation to weather, wind and distance to the turbines which can be extrapolated beyond the sites of data collection. However, it is important to note that due to the sensitivity of the radar devices to wind induced sea clutter the data was collected during calm weather conditions, and is therefore biased towards these conditions. The response to e.g. wind speeds might therefore change at higher wind speeds.
The flight directions (both specific-specific tracks and the automated radar recordings) corroborated earlier findings indicating the dominance of ‘local’2 seabirds at the two wind farms, and in comparison to the situation on the coast of Blåvandshuk movements of birds on long-distance migration constituted a small proportion of the total number of movements. Movements of one species, Common Scoter, outnumbered those of other species, which was reflected in the 55% of all recorded tracks, and the large number of radar signals recorded between November and March. Due to the higher abundance of Common Scoter at HR2 as compared to HR1 the highest densities of species-specific tracks and radar signals were recorded at HR2. Compared to the baseline the post-construction monitoring documented the presence of events of migrating passerines on Horns Rev, especially Meadow Pipits during autumn. Yet, these events were also noted on Blåvandshuk at an even larger scale. Thus, the occurrence of large numbers of passerines offshore on Horns Rev seems to be related to mass migration during autumn rather than specific offshore corridors. The visual recordings of terns showed that some movements of terns are noted on HR1 but not at the coast, indicating the presence of an offshore corridor.
Prominent barrier effects and a reduced risk of colliding with the turbines of HR1 and HR2 could be determined for most key species. Gannets were seen in the HR2 wind farm despite the fact that the species has not previously been recorded in the HR1 wind farm. Accordingly, for all main species the barrier effect could be judged as partial as no species completely abandoned the wind farms. An avoidance corridor where densities of flying birds peaked was determined for Common Scoter to be 1,500-2,500 meter distance from the wind farm, and at HR1 at 1,000-2,000 m distance. The probability of the Common Scoter to fly towards a wind turbine decreased more steeply at a distance of 1.5 km, with a tendency for a delayed response during spring as compared to autumn, possibly as a result of habituation during the staging period. These results are in line with the findings during the PSO program, and it can be safely concluded that due to the limited spatial scale of the barrier effect of local seabirds at HR1 and HR2 no cumulative barrier effect exists between the two wind farms.
The altitude profiles of the majority of species showed a preference for low altitude flights, particularly for the seabird species which all predominantly were recorded flying below the rotors when they approached the two wind farms. Only large gull species (Herring Gull, Lesser Black-backed Gull, Great Black-backed Gull), Gannets, raptors, pigeons and passerines were often recorded flying at rotor height close to the wind farms. The flight models indicated that most species flew higher in tail and side winds in comparison to head winds although the variable was not always significant. Common Scoters and large gull species flew, according to the models, at higher altitudes closer to the turbines. Gannets and Common Scoters increased altitude with increasing wind speeds while large gulls, passerines and pigeons increased flight height with decreasing wind speed. According to the models the species flying closer to the rotor swept area were Gannets and large gulls, whereas Common Scoters and terns flew well below the rotor swept area in all-weather situations. The risk for the Gannets and large gull species of potentially colliding with the rotors increased when the birds were flying in tail or side winds and for the Gannets at intermediate wind speeds and for large gull species at low wind speeds. The risk was higher at Horns Rev 2 than at Horns Rev I as the rotor swept area reaches 10 m closer to the sea surface at Horns Rev 2. The modelled average flight altitudes for passerines and pigeons were based on a small sample size and included a range of different species which consequently increased the uncertainties of the results.
A higher collision rate was seen across all assessed species, except small gulls, for HR2 than for HR1. Thus, the height of the lower tip of the rotor over the sea surface (10 m higher for HR1 than HR2) seems to constitute an important design parameter in relation to mitigating collision risks to seabirds. The specific macro avoidance rates were lower than those recorded at offshore wind farms in the Netherlands and Belgium, and those recorded earlier during the monitoring at HR1. This fact together with higher tendency to fly at rotor height at HR2 compared to these other wind farms increased the risk of collision of seabirds at HR2. These differences may be caused by movements on Horns Rev being dominated by local feeding or resting seabirds which may have a higher tendency to habituate to the wind farm than birds on long-distance migration. The worst case estimated total mortality of seabirds per ‘winter’ season was 553 at HR2 and 415 at HR1, of which Common Scoter and large gulls comprised the vast majority of potential victims. The large number of estimated collisions by Common Scoter was mainly driven by the large abundance of birds at HR2. Although only a minority entered the wind farm perimeter at rotor height the flux of birds was sufficiently large to cause regular collisions with the rotor blades. The high estimated collision rates for large gulls were mainly driven by a combination of low avoidance rates and high proportions flying at rotor height. Despite the relatively high number of estimated collisions the toll of the species concerned was estimated to lead to insignificant impacts at the population level. Thus, the estimated level of in-combination collisions by seabirds for HR2 and HR1 should be seen as less problematic than those reported for raptors at the Nysted and Rødsand 2 offshore wind farms.
The collision models have provided detailed estimates of collisions using a deterministic approach, and a methodology for which most parameters can be safely set using the field data at hand from radar and rangefinder tracks. However, for two of the model parameters very few data are available, and as they represent the within wind farm behaviour of birds and interactions with the rotor blades the estimated collisions should only be regarded as approximations; proportion trying to cross the swept area without showing avoidance and the probability of being hit by the rotor-blades. To establish realistic micro-avoidance rates it is recommended exploring the possibility for applying surveillance networks comprising a combination of tracking in the periphery and inside the wind farm.