High Flyer

By Kasey Rahn

Boone and Crockett University Programs Fellow Marco Salvo uses machine learning to help prevent bird-aircraft collisions

Collisions between avians and aircraft, called bird strikes, are costly, kill birds, and are dangerous to people. Costs related to strikes are in the billions, including aircraft repairs and flight delays. From 1988 to October 2024, there were 499 human fatalities and 361 aircraft destroyed that were attributed to wildlife strikes globally for military and civil aviation. Boone and Crockett Fellow Marco Salvo, a master’s student at the University of Montana (UM), hopes to help change that.

For his master’s project, Salvo is exploring ways to use bird-dedicated radar systems to help inform and reduce bird strikes. The project is funded by the U.S. Department of Defense and conducted in collaboration with the U.S. Air Force.

“There’s a great interest in the Air Force and across the aviation industry as a whole to reduce the risk of bird strikes to planes,” Salvo says. “One tool that’s been increasingly used to help reduce that risk is bird-dedicated radar systems. The same way that radars can detect planes flying through the sky, there are specific radar systems that can detect birds.”

Salvo says radar systems record birds’ flight patterns, and the data is displayed as tracks, essentially mapping bird routes. The catch is that the radar can’t differentiate between bird species or numbers.

“It could be a small sparrow or a flock of 200 Canada geese,” Salvo says. “It gives a really good idea of intensity of bird use on the base…but it doesn’t give you an idea of risk.”

Salvo and his team are using the tracks and other data collected by the radar to see if they can build machine-learning models capable of identifying the number, size, and species of birds from radar signatures. The technology could be used to inform wildlife management on and around military bases and to help inform and ultimately prevent bird strikes.

Salvo is currently focused on predicting specific categories of birds, including waterfowl, songbirds, herons, and raptors. The work is ongoing, but initial results show that the scientists can predict songbirds and great blue herons at a bare minimum. Their next goal is to narrow in on raptors and waterfowl. The results will be supplied to the Air Force and shared publicly in a scientific journal.

A second part of Salvo’s master’s work will also examine how weather, the time of day, and seasonal changes influence bird activity patterns.

“When I started this, I had no idea that radar could detect birds. That application is really cool,” Salvo says. “I think this project is a really good example of how we’re using modern technology, big data, and advances in machine learning and artificial intelligence to assist in wildlife management.”

Digging Into Data

Salvo spent two years collecting bird data at Air Force bases in Nebraska and South Dakota. The effort required long days in the field, clutching a tablet in one hand and a pair of binoculars in the other. The tablet showed the radar tracks in real-time, and Salvo would find and identify the correlating bird, explains Mike Clawson, Ph.D., owner of Clawson Statistical Services, a consulting firm. Clawson is the project’s analyst and is guiding Salvo’s work.

Salvo identified several thousand bird tracks by hand, using the points to create a database of known tracks—baseline information about which bird species coordinated to which radar signatures.

The radar also collects dozens of other metrics, including speed, distance to radar, radar reflectivity, and more. The machine learning models essentially compare these metrics to information in the database to identify the unique characteristics of different birds. The algorithms then use that information to predict bird identifications for other radar tracks.

The work requires an incredible amount of data, Salvo says. A month’s worth of data from one radar is easily 20 gigabytes, and two years of data in South Dakota recorded 120 million unique tracks.

Clawson says the undertaking is a novel approach—and a level of work other researchers might not tackle until a doctoral program.

Applied Management: From Elk to Avians

The Univerity Programs sponsors Salvo's graduate fellowship, covering expenses with living stipends so he can focus on his research.

Salvo’s master’s research advisor is Joshua Millspaugh, Professor of Wildlife Biology and Director of the Boone and Crockett Program at UM and Camp Fire Conservation Fund Professor at the State University of New York. But Salvo’s been a part of UM’s Boone and Crockett Program since well before he started his graduate education.

Originally from New York City, Salvo headed west to study wildlife biology at UM as an undergraduate student. During his freshman year, he took a weekend workshop about hunting for sustainability taught by Millspaugh at the Club’s Theodore Roosevelt Memorial Ranch.

A few years later, Salvo officially joined the lab when he started working as a field technician on a doctoral study about elk. He conducted plant surveys and set up camera traps, and he eventually built his senior thesis around the project. When the project’s lead left the program following the COVID-19 pandemic, Salvo stepped in to wrap up the project, working for the lab as a research associate.

Salvo says the project aimed to assess the effects of low-intensity mechanical thinning treatments in forests on elk forage over long periods. The researchers found clear benefits to elk forage quality and quantity in the short term. As you get further out in time, 10-20 years or so, the forest returns to a more natural state like what’s found in an untreated forest.

When the chance came up to turn the radar research into a master’s project, Salvo took it. He says he was drawn to the project’s tangible outcomes—data that managers can use to make decisions and, in this case, ultimately save lives. “That,” he says, “is work worth doing.”

Salvo will graduate in December 2025. Afterward, he wants to pursue a career in academia and applied management. It’s the thread that connects all his work.

Salvo says that applied management looks at conservation problems important to natural resource managers and answers questions they can act on. Large-scale anthropogenic changes on the landscape have made it so wildlife and humans impact one another constantly, so finding those answers is incredibly important.

Clawson describes Salvo as smart, ambitious, and hardworking. But it’s Salvo’s curiosity—especially his keen interest in the details—that sets him apart.

“That’s one of the most important things you can have as a scientist,” Clawson says, “and Marco has it in spades.”