Potential effects of using machine vision monitoring to estimate eagle fatality risk at wind facilities
Kimberly Peters
DNV GL Energy
Dr. Kimberly Peters is a Senior Project Biologist at DNV GL. She received her MSc in Fisheries and Wildlife Science from North Carolina State University, and PhD in Zoology from Clemson University. Over the last 20 years, Dr. Peters has led research and conservation programs on migratory shorebirds, grassland birds, wind energy bird and bat fatalities, and bird-aircraft strike-risk. Prior to joining DNV GL, she served as Director of Bird Monitoring at New Jersey Audubon and as Chief Scientist and Director of Bird Conservation at Mass Audubon, with a focus on migratory bird conservation. Dr. Peters provides environmental and wildlife support to wind developers in both the US and Canada, has published research on the analysis of bird and bat fatality estimates, has co-authored reports on radar-based passage-rates for migratory birds and bats at operational and proposed wind-development sites, and is the lead author of the Canadian Wind Energy Association’s Wind Energy and Bat Conservation Toolkit.
Abstract
Much emphasis has been placed on obtaining accurate and reliable estimates of eagle activity at prospective and operational wind facilities because of concerns about potential strike-risk from wind turbines. Projected... [ view full abstract ]
Much emphasis has been placed on obtaining accurate and reliable estimates of eagle activity at prospective and operational wind facilities because of concerns about potential strike-risk from wind turbines. Projected collision risk for a specific wind project is typically estimated using the U.S. Fish & Wildlife (USFWS) Bayesian Model, which incorporates prior probability distributions based on external projects to relate eagle activity to risk. The model is then parameterized by site-specific eagle counts; however, the model is highly sensitive to survey effort, which can be limited when using standard point-count surveys by biologists. Recent developments in machine vision technology offer an opportunity for substantially increasing survey effort and could therefore influence eagle fatality estimates as a function of effort alone. We conducted a series of simulation studies based on the monitoring capabilities of a newly-developed machine vision system, IdentiFlight (Renewable Energy Systems), under varied project and eagle use conditions to determine how automated monitoring could affect strike-risk estimates. Results showed that risk estimates were markedly lowered as compared to using standard point-count methods, particularly under conditions in which eagles were rarely observed on site or were temporarily using the site (e.g. migratory periods, subadult wandering). The Bayesian eagle risk model assumed that expected eagle observations increased linearly at a 1:1 (or lower) ratio as observation time increased, and our application of the model further assumed that machine vision for a given time window is comparable to human observers; these assumption have yet to be verified with field testing. The potential regulatory and economic impacts of using machine vision monitoring at planned and operational wind facilities are discussed in the context of the USFWS Eagle Conservation Plan Guidance recommendations, including expected impacts if model assumptions are violated.
Authors
-
Kimberly Peters
(DNV GL Energy)
-
Tom Hiester
(Renewable Energy Systems)
Topic Areas
Evaluating novel approaches (e.g., conceptual, methodological, technological) to avoiding, , Eagles , U.S. - No Specific Region , Technology - detection or deterrent , Land-based
Session
00 » Posters (12:30 - Friday, 2nd December, Centennial Ballroom)
Presentation Files
The presenter has not uploaded any presentation files.
