The objective of the present study was to produce a robust, quantitative prediction of fatality rates of “rufa” Red Knots (Calidris canutus rufa) resulting from collisions with the physical structures of a yet-to-be-constructed offshore wind energy facility (the “Facility”) that has been approved for construction in federal waters of Nantucket Sound, Massachusetts. To accomplish this objective, we assembled a technical team consisting of field leading experts in offshore wind bird collision risk assessment, collision risk modeling, and Red Knot biology to synthesize existing technical information on this subject, and to use this synthesis as the basis for developing an original quantitative collision risk modeling effort.
A comprehensive review of technical literature related to bird collision risk modeling at offshore wind energy facilities was performed as an initial step in this process. On the basis of this review, we adopted a modeling approach that included the Band (2012) model to represent a subset of collision dynamics, with various additional elements incorporated to represent the most important biological and meteorological dynamics of the system of interest, as hypothesized and conceived by the Project’s technical team on the basis of best available scientific information. We developed an original simulation model to represent this system, and conducted a series of simulations with the model in order to produce quantitative predictions of Red Knot fatality rates resulting from collisions with the structures of the approved Facility. These simulations included variation in several of the model’s inputs in order to characterize the sensitivity of the model’s fatality rate predictions to changes in these inputs.
The overall average collision fatality rate for rufa Red Knots at the approved Facility predicted by our model was 0.16 Red Knots per year equivalent to one fatality every 6.25 years, composed of 0.10 predicted fatalities per fall migration season and 0.060 predicted fatalities per spring migration season, under baseline, or default model inputs. Predicted fatalities scaled linearly with population size, such that under the assumption of a “recovered” population three times the size of the current population, the predicted fatality rates were exactly three times higher.
Collision fatality rates were largely driven by collisions of Red Knots with stationary structures, and particularly turbine towers, in our modeled results, with turbine towers accounting for roughly 90% of all collision fatalities in most simulations. The influence of collisions with turbine towers was also evidenced by the result that for all model iterations in which at least one collision occurred, the average number of collisions was approximately eight, corresponding to the number of Red Knot wingspan lengths in the tower’s diameter. Red Knots were represented flying in chevron-shaped flocks, aligned wingtip to wingtip in our model, hence when the path of a flock intercepted a turbine tower, seven or eight collisions typically resulted, depending on the elevation at which the flock encountered the tapered, cylindrical towers.
Biologically realistic representations of the influence of variations in wind speed and direction, precipitation, and visibility were incorporated into the model in the form of behavioral switches between high and low elevation migratory flight altitude distributions (effective headwind speed, heavy precipitation), fall migratory flight departure delay decisions (effective headwind speed), and avoidance rate parameters (heavy precipitation, visibility, and effective headwind speed). Simulation results indicated that of these, only effective headwind speed exerted a strong influence on the fatality rates. The relative insensitivity of fatality rates to variation in precipitation was explained by the relative rarity of heavy precipitation events during the migratory seasons of interest, as characterized by an empirical summary of hourly precipitation data from regional meteorological data. Low visibility events occurred more frequently than heavy precipitation events within the Facility region during the seasons of interest. Nonetheless, low visibility conditions were still far less common than high visibility conditions, resulting in an overall weak influence of variation in visibility on modeled fatality rates. By contrast to heavy precipitation and low visibility , the effective headwind conditions that were hypothesized to trigger behavioral switches in migrating Red Knots occurred more commonly, and consequently exerted a more important influence on modeled fatality rates. Specifically, fatality rates were significantly higher when the effective headwind speed threshold for fall migratory flight departure was below the effective headwind speed threshold for switching from the high-altitude to the low-altitude flight distribution. Under such conditions, fall migrating birds would commonly decide to fly under conditions that would cause them to fly at a lower altitude. In all other conditions, fall migrating birds would either tend to fly at a higher altitude, or not fly at all.
Relaxing the Band model’s assumption of perpendicular approach angle had a marginally significant influence on fatality rates for approach angles close to perpendicular to the rotor (head on) and a negligible influence for approach angles close to parallel to the rotor. Altering the mean flight altitudes had a slight influence on modeled fatality rates for the high-altitude flight distribution, which was elected by migrating birds in our model under favorable weather conditions, and a moderate influence on fatality rates for the low-altitude flight distribution, which was elected by migrating birds in our model under poor weather conditions.
Avoidance behaviors were incorporated in our model at two distinct spatial scales. Macro-avoidance was defined as the avoidance of the entire modeled wind farm by a flock of birds that sees the Facility from a distance. Our model included the assumption that macro-avoidance ability was diminished to near zero (0.01) under low visibility conditions or heavy precipitation. Micro-avoidance was defined as the avoidance of a wind energy Facility structure by a flock that was present within the Facility, and approaching the structure. In our model, micro-avoidance capacity was reduced by the presence of heavy precipitation, strong headwinds, and low visibility, and separate micro-avoidance rates were included for stationary versus (vs.) moving structures. Modeled fatality results generally exerted a moderate level of sensitivity to variation in macro-avoidance rates and a high level of sensitivity to variation in micro-avoidance rates. These patterns are consistent with a general trend reported in previous studies of bird collision risk with offshore wind energy facility structures, and highlight the importance of characterizing birds’ behavioral responses to wind energy facilities for understanding collision risk.