Abstract
Curtailing the operation of wind turbines when wind speeds are low is a common and effective method used to reduce bat mortality. However, bats are not always present or killed during periods of lower wind, and additional research is needed to understand the factors that influence bat migration activity and mortality, which could be used to develop smarter curtailment regimes that maximize energy production while reducing bat mortality. Pilot Hill Wind, LLC is operating the Pilot Hill Wind Farm (PHWF) and Kelly Creek Wind, LLC is operating the adjacent Kelly Creek Wind Farm (KCWF; Projects) in Kankakee, Ford, and Iroquois counties, Illinois. Pilot Hill Wind, LLC and Western EcoSystems Technology, Inc. were awarded funding from the American Wind and Wildlife Institute to enhance ongoing research at both facilities in 2018 to quantify potential relationships between bat activity, bat fatalities, and weather patterns that could be used to implement cost-effective strategies for reducing bat mortality. Previous studies at the PHWF documented relationships between precipitation occurring on-site and bat fatalities recorded 48- hours later, and we hypothesized the lag in fatalities following precipitation could be because bat migration initiated outside the PHWF was being triggered by broader weather fronts in the region.
The objectives of this study were to 1) determine if weather data collected at off-site locations would be a better predictor of bat activity and fatalities than weather data collected on-site, 2) identify if weather variables could be used to optimize curtailment by increasing energy production when bats were not present, and 3) determine if bat activity and fatalities were concentrated in portions of the night.
Fifteen turbines were searched within 80 meters (m; 263 feet [ft]) of the turbine base at the PHWF; 10 of these were searched daily and five were searched twice a week. Anabat SD2 detectors were installed on eight turbines at the PHWF and four at the adjacent KCWF to record bat activity. Only bat carcasses estimated to have died the previous night from the PHWF were used to quantify relationships between bat mortality and weather. Acoustic data from PHWF and KCWF were used to quantify relationships between bat activity and weather. NRG Systems’ Bat and Avian Mortality Monitoring System (BAMM), a camera based carcass detection system, was installed on 10 turbines at PHWF in an effort to determine the time of night when a fatality occurred.
The capabilities of logistic regression to predict migratory tree bat activity and fatalities were examined. Fatalities and activity data were divided into exploratory and validation data sets to test model performance. Barometric pressure and wind direction from 12 different airports located within 161 kilometers (km; 100 miles [mi]) of both Projects were used as measures of potential high and low pressure systems that may pass through the region and trigger bat activity and migration. Twelve weather variables collected from on-site meteorological towers and turbine sensors were included as potential covariates, including day of study, wind speed, air temperature, wind direction, precipitation, relative humidity, and barometric pressure. Estimated moonlight and peak moonlight measured at the Kankakee Airport, located 9.5−41.2 km (5.9−25.6 mi) from turbines, were included as variables, and assumed representative of conditions at both Projects. Wind direction recorded 48.0−121.0 km (29.8−75.2 mi) north of both Projects at airports along the Kankakee River were better predictors of hoary and silver-haired bat fatalities than weather variables recorded on-site. Eastern red bat activity was best predicted by combinations of changing barometric pressure and wind direction recorded on-site. Both the PHWF and KCWF lack significant forested areas within 10.0−19 km (6.2−1.8 mi) of both Projects. The lack of forested areas near the Project, the lack of bat calls recorded within 120 minutes (min) of sunset, and the selection of weather variables from airports located 48-121 km north of the Projects (as the best predictors of fatalities) suggest hoary and silver-haired bat migration and subsequent occurrence in both Projects were influenced by larger scale weather fronts away from both Projects. This was the first study to quantify data from broad scales, and suggests smart curtailment approaches for hoary and silverhaired bats could be improved by utilizing a combination of on-site variables such as wind speed and temperature, with off-site measurement of broader weather fronts to develop smarter curtailment approaches.
The BAMM system identified the time of death of four hoary bat carcasses and one eastern red bat carcass; this was the first research study to assign a time of death to a bat carcass. The time of death for the five fatalities occurred between 2300 H and 0230 H. The wind speeds were variable on the nights when all five bat carcasses were found, ranging from 0 to approximately 12 meters/second (m/s; 26.8 miles per hour [mph]); all five fatalities occurred when wind speeds were between 2.8 and 3.9 meters per second. One of the five bats was killed when wind speeds were below cut-in speed (3.0 m/s [6.7 mph]). The five fatalities occurred during periods of the night when wind speeds were declining relative to periods just prior to the time of death. All five carcasses detected by BAMM occurred within 15.0 m (49.2 ft) of the base of wind turbines.
No bat activity was recorded until 90 min after sunset, and few calls were recorded until 120 min after sunset; activity occurred until sunrise. Both Projects lack significant forest cover within 10.0−19.0 km (6.2−11.8 mi) of turbines, which suggests bats have to fly significant distances before reaching turbines at both Projects. Our results suggest curtailment could be delayed until 90−120 min after sunset with little impact on bat fatalities at both Projects, and potentially other projects located long distances from significant amounts of forest habitat and roost sites could similarly delay curtailment.