Each spring, billions of birds fly vast distances to spend the summer in North America, most of them traveling at night. It's a trip fraught with peril: Many slam into wind turbines or brightly lit buildings. Now, a new forecasting system for bird migration could help put an end to millions of those deaths by warning wind farm operators and building managers of incoming migrations 3 days in advance.
Although hawks and other large species migrate during the day, most small birds migrate at night to avoid predators and enjoy better flying conditions. The daily legs of these migrations depend heavily on the weather. If conditions are too cold or rainy, migrants hang out in trees until the skies clear. And birds are more likely to continue their journeys when warm air signals an incoming, southerly tailwind. Since 2012, the Cornell Lab of Ornithology has made predictions about these migrations by using observer sightings and regional weather reports on its BirdCast website.
To scale up and automate these forecasts Benjamin Van Doren, a Ph.D. student at the University of Oxford in the United Kingdom, and Kyle Horton, a postdoc at the Cornell lab, built a computer model of weather and bird migration. They began with weather radar, the only effective way to monitor night-time migrations. Individual birds can't be detected, but radar can reveal the density of birds in the airspace: 60 to 70 birds per cubic kilometer in a light migration, and as many as 1700 in a heavy one.
Van Doren and Horton calculated the number of migratory birds on a massive scale. They analyzed radar images from 143 sites across the continental United States, examining the same half-hour period each night between March and May, the migration season. By looking at records from 1995 to 2017, they also were able to estimate for the first time the average number of birds migrating on any given night across the entire United States: about 200 million, they report on the preprint server bioRxiv. The mass movement peaks at 520 million birds, typically in early May.
Next, Van Doren and Horton compared all these movements to the local weather at the time when birds began a particular leg of their journey. They found that air temperature was the most powerful predictor of how many birds would be flying that night. By combining forecasts of air temperature and other factors such as air pressure and wind, they constructed a model that could reliably predict bird movements up to 3 days in the future, and they suspect that longer forecasts are possible. "It's impressive how precise the prediction is, and how accurate," says Wouter Vansteelant, a freelance bird migration researcher in Bennekom, the Netherlands, who was not involved in the research.
BirdCast is now posting automated forecasts for the entire United States. "This is the most significant update since we first began using radar to study bird movements," Horton said in a statement. The Cornell team hopes that in addition to telling birdwatchers where and when to spot migratory species, the forecast will help owners of tall buildings know when to turn off unnecessary lights. Moreover, wind farms—some of which employ monitors to watch for hawks and other daytime migrants—might shut down turbines when waves of nighttime migratory birds are about to pass through.