Plate Nº 94 · recorded October 10, 2026
Earth & ClimateReported finding
Thermal cameras log 22 possible bird collisions at offshore turbine
Thermal cameras and AI logged 22 possible bird collisions at a Dutch offshore wind turbine over 11 months, including two confirmed strikes and one rated highly probable—marking an early empirical dataset at sea.
By Elena Vasquez3 min read570 words
In brief
- Researchers logged 22 possible bird collisions during an 11-month pilot at the Hollandse Kust Zuid offshore wind farm.
- Two of the 22 events were confirmed and one was rated highly probable after human review.
- Wildlife Imaging Systems installed 16 thermal cameras on a single turbine, operating around the clock through fog, rain and darkness.
- The system was commissioned by energy company Vattenfall and the data was analyzed by Wageningen University & Research.
- Past offshore collision estimates relied largely on theoretical models, since struck birds typically fall into open water with no remains to recover.
Thermal cameras and artificial intelligence logged 22 possible bird collisions at a single offshore wind turbine during an 11-month Dutch pilot, generating empirical data on songbird and bat strikes that until now has been largely out of reach at sea, researchers report.
The setup ran at the Hollandse Kust Zuid offshore wind farm, where the wildlife technology firm Wildlife Imaging Systems (WIS) installed 16 thermal cameras on one turbine. The cameras covered every angle and operated continuously for nearly a year, including through fog, rain and darkness.
What did the system actually record?
Wageningen University & Research (WUR) scientists analyzed the resulting footage. Across a spring and a fall migration season, the AI flagged 22 "possible" collision events. Two were confirmed on review and one rated highly probable. The remaining 19 sit in lower-confidence categories the team is still refining.
The detections span small migratory songbirds at night, larger birds during daylight hours, and bat encounters. The data matter because seabirds and songbirds that strike offshore blades usually drop into open water, leaving few remains. Past offshore collision estimates rely on theoretical models, not direct counts.
"Until now, estimates of offshore bird collisions were mainly based on theoretical models," says Jesper Kyed Larsen, a bioscience expert at Vattenfall, the energy company that initiated the study. "At sea, it is extremely difficult to verify fatalities because there are no remains to recover. This study demonstrates how we can get the empirical data needed to improve collision estimates and secure efficient mitigation measures."
Why thermal cameras, and why now?
Millions of songbirds cross the North Sea twice a year, mostly under cover of darkness. The cameras register the heat signatures of birds and bats in flight, and machine-learning software classifies the resulting tracks.
The pilot did not aim to settle whether offshore wind power is dangerous for birds. The team stresses that one turbine at one site cannot speak for the wider North Sea fleet.
"Safeguarding biodiversity is an important part of the energy transition," says Karen Krijgsveld, a bird ecology researcher at Wageningen University & Research. "On land, we already know quite a lot about when and why birds collide with turbines, and which species are most at risk. As offshore wind energy continues to expand, we need to develop that same understanding at sea."
Krijgsveld describes the WIS system as a meaningful step toward measuring offshore collisions directly—an ability that, until now, has existed mainly in computer models.
What are the limits of the result?
The findings rest on a single turbine over roughly 11 months. Weather, bird species, turbine layout and rotor speed all vary across the North Sea, so the observed collision rate cannot yet be extrapolated to other farms. The authors flag this explicitly.
Detection accuracy also needs further work. With only three events out of 22 firmly classified, the AI still flags many ambiguous near-misses or false positives that a human reviewer must sort out.
The team is now planning a follow-up phase. Researchers will spread monitoring to more turbines and locations, refine the algorithm and gather collision data across a wider range of weather and seasonal conditions. The eventual goal: link collisions to specific circumstances—species, weather, time of night—so operators can design more targeted mitigation, from adjusted blade painting to curtailment during heavy migration nights.
The report, authored by Krijgsveld and colleagues, is available with DOI 10.18174/721562.
via Phys.org Biology (Source)
More from Elena Vasquez
Nearby plates
- Floating Scarecrow Cuts Seabird Deaths in Fishing Nets by Two-Thirds
- Marine Heatwaves Hide Months of Extra Ocean Warming, Study Finds
- Back-to-Back Arctic Storms Can Double Sea Ice Loss, Study Finds
- Raptors banded in Europe are dying in West Africa, study finds
- Tracking 4,581 Animals Reveals a Hidden Human Effect on Wildlife