A Popular New Migration Tool Could Save Birds from Deadly Building Collisions

BirdCast can accurately forecast the biggest migration movements days in advance. This is good news for birds and birders alike.
Migration forecast map for the United States. Warmer colors indicate that the model predicts a greater number of migrating birds. Map: Benjamin Van Doren

In the Spring of 2017, after 400 birds crashed to their deaths in a single night by flying into a brightly lit high-rise in Galveston, Texas, Houston Audubon and local partners quickly worked with the building owners to help shut off the lights for the remainder of migration season. By doing so, they surely spared the lives of many other migratory birds that are attracted to or confused by lights as they make their annual journeys north or south. 

While this was a huge win, Houston Audubon realized it could make an even bigger difference, says Richard Gibbons, conservation director of Houston Audubon. This past spring, the chapter started a program called Lights Out for Birds to let homeowners and businesses know which nights would be most active for migrating birds. These alerts were based on a recently released tool from the Cornell Lab of Ornithology called BirdCast, which uses observation data and doppler weather radar to predict and track bird migrations. “This is not just me prognosticating, looking at tea leaves or something,” Gibbons says. “This is based on science.” 

Upon its release, BirdCast's popularity with birders itching to see migrants was immediate, but as Houston Audubon showed, it also has huge potential value as a conservation tool. This week, new research published in Science looking at the modeling used to predict migration patterns this spring, further supports this idea. Not only are the forecasts highly accurate, the researchers found, but BirdCast is able to reliably predict migration timing up to three days in advance.

The images from BirdCast should look familiar to anyone who’s watched a TV weather report. Fed by radar precipitation data, generated every few minutes from 143 towers country-wide, the instruments also pick up moisture-laden animals passing through the sky. “Birds are like large raindrops in the atmosphere,” says Benjamin Van Doren, an ornithologist from the University of Oxford who works on the forecasts. These traces of wildlife are a nuisance to meteorologists, who have to edit them out of standard weather data, but a boon to anyone searching for birds

Using new machine-learning techniques, Van Doren and Kyle Horton, who studies bird migrations at Cornell, trained an algorithm to forecast migration activity using 23 years of radar data and bird observations. The findings show that migration largely happens in a few big bursts: Half of all migrants move on roughly eight nights of the entire season, Horton says. For spring migrations, their model closely predicts the intensity of migration actually observed, explaining about 70 percent of bird activity three days in advance. (The percentage jumps to 80 percent only a day out.) 

In spring, a warmer temperature is the biggest predictor of which nights will see this flurry of activity. Different factors play into birds’ decisions to move in the fall, but the majority still take to the sky on fewer than 10 nights. Fall migration could be more deadly because there are more young birds in the mix; those inexperienced travelers might be more thrown off by bright lights than a weathered springtime migrator. “There may be some learning here,” Horton says. “Young birds may be skewed in terms of their attraction to light.”

The forecasts predicting big migration nights are published for free online, so hypothetically anyone—including owners of large lit-up buildings or stadiums that attract and kill migrating birds—could use the tool to prevent bird deaths. However, "there’s a big step sometimes between developing a model and actually putting into practice,” says Judy Shamoun-Baranes, an ecologist at the University of Amsterdam. For over a decade, she’s worked with the European Space Agency to develop forecast models for the Dutch and Belgian Air Force that use local radar data to predict hourly bird activity three days in advance—not for the birds' benefit, but for people's. To avoid dangerous collisions with planes, pilots will wait to hold training exercises if large numbers of birds are expected in the area.

Ideally, building owners would follow a similar tack and fully integrate bird conservation into building management (or, even better, just keep their lights off all the time). But it’s more likely that they’ll need a push from outsiders like Houston Audubon. “The more groups, chapters, bird clubs that can help build a groundswell of awareness, the more likely we are to have collective success," Gibbons says. 

To its credit, this past year the city of Galveston followed Audubon's lead and encouraged anyone who lives or works in a building that’s at least three stories tall to turn off their lights on certain nights—though not necessarily the heaviest nights as predicted by BirdCast. But if the city had done so, it could have potentially had an even bigger impact. And if other cities and towns also took up the cause, millions of birds could be saved each year. After all, thanks to BirdCast, the information is out there. Everyone just needs to tune in. 


Audubon is a nonprofit dedicated to saving birds and the places they need. You can support our conservation work by making a donation today.