EERE Success Story—ThermalTracker: Imaging for Coexistence

When studying the potential effects of offshore wind turbines to wildlife, like birds and bats, the traditional way of surveying is to have a person in a boat or aerial survey during daylight hours or in fair weather conditions.

But even in perfect weather conditions, surveying can be challenging—and especially problematic for observing bats, which thrive at night. Now, with a new tool known as ThermalTracker, these challenges and limitations are removed. The innovative technology for processing thermal video can also process information five times faster than a human.

Supported by the Wind Energy Technologies Office (WETO), ThermalTracker software was developed by Pacific Northwest National Laboratory’s (PNNL) Marine Sciences Laboratory. The open source, plug-and-play software can be used with any standard desktop computer, thermal camera, and statistical software to quantify the presence and behaviors of animals in offshore locations with near real-time, continuous monitoring.

This technology provides a tool for researchers and the wind industry to meet monitoring requirements for offshore locations to collect more data about bird and bat activity at proposed wind farm sites. Currently, no other software automatically extracts the flight tracks of birds and bats from thermal video.

With the groundbreaking ThermalTracker, a unique algorithm takes a segment of video, for example, 10 seconds of recording at 30 frames per second, which is 300 video frames, and compresses it into one single image that can show the entire flight track of a bird or bat through the field of view. So in one image, viewers can see the entire flight track.

However, when it comes to identifying the species of the birds there are some challenges with thermal video. Although thermal cameras have the advantage of day and night and low visibility operation, coupled with the fact that the animals stand out against the background making them easy to detect, the resolution of thermal cameras is not as a high as one might have with a normal digital camera. As a result, the images don’t have as detailed shape of the animal or color information —critical to identifying birds and bats species.

To address that issue, researchers used the actual shape of the flight track to distinguish between different types of birds and bats and found they could successfully discern the difference between a seagull, a tern, swallow, or bat. In addition, wing flap patterns help provide a clue as to the species.

“This is really impressive for initial prototype software development,” said Jocelyn Brown-Saracino, WETO’s Acting Program Manager for Market Acceleration and Deployment for Wind Energy at DOE.

Currently, Version 1 of the ThermalTracker software is on GitHub, a website for sharing open-source software. This year, the team at PNNL has been developing Version 2 —a real-time version to work within the camera stream. Version 2 does the processing directly from the live camera with the addition of stereo vision, which means users will see a 3D flight track feed.

This is important because it helps researchers and regulators know where the animals are flying relative to the rotor swept zone (the area that the wind turbine’s turning blades pass through), and distinguishes how far the bird is from the camera based on its size, which will help in identifying the species.

In tests with video recorded at the Marine Sciences Lab, the Version 1 software has been able to detect 81% of the animals it recorded and correctly classify to the taxonomic family 82% based on the flight tracks.

The new version will also help identify avoidance behavior such as birds and bats avoiding turbine areas. This behavior would be included in risk models that regulators use when permitting wind farm sites.

The ThermalTracker software will not only output the positions of the bird flight track, but also the direction it is traveling, how fast, whether it’s in a straight line, or changes direction. While Version 1 does this already, Version 2 will do this in 3D. Version 2 is already on the horizon with testing underway that incorporates these special 3D capabilities. The goal is to develop it as a tool for environmental consultants and biologists.

Next year, the team hopes to do field testing around a wind farm to collect relevant data to see how birds and bats fly around the wind turbines.