Skip to main content

Weather Radar Predicts Nighttime Bird Migrations

Thumbnail
Painted buntings fly under the cover of darkness to reach breeding grounds in the southern U.S. | Kyle G. Horton

In North America, when air temperatures warm in early May, more than 500 million migratory songbirds take flight each night, according to a study that used a new method to predict continental-scale bird migrations using weather forecasting radar systems.

Combining radar data with climate models, which revealed the important atmospheric conditions that influence the timing of migration events, the researchers created a forecasting model able to accurately predict the timing, intensity and location of nocturnal migration events as far as a week in advance, as well as estimate the numbers of birds making the journey each night.

The findings , published in the September 14 issue of Science, could be used to reduce bird deaths resulting from human activities and infrastructure, according to University of Oxford's Benjamin Van Doren, the study's lead author, and colleagues.

Each year billions of birds navigate the skies on biannual migrations, traveling great distances back and forth between breeding and wintering sites that sometimes lay a continent apart. Like ancient humans crossing vast seas, some birds plot their courses using the stars. Others follow the ebb and flow of Earth's electromagnetic field — using it not just as a compass, but as a kind of visual heads-up display to identify their position on the globe.

Not all of these birds arrive at their destinations. According to the researchers, migratory bird populations in North America are in decline, mostly due to habitat loss. Human-caused threats, such as collisions with buildings, communication towers or vehicles, result in the estimated deaths of over 630 million birds each year in the U.S. alone. While collisions with aircraft account for only a small proportion of this total, bird strikes worldwide are responsible for at least $1.2 billion of aircraft damage annually and endanger the lives of those aboard. For example, 2009's "miracle on the Hudson," where the pilot of US Airways flight 1549 was forced to make an emergency landing on the Hudson River with two disabled engines, occurred following a collision with a large flock of geese.

Widespread light pollution also can lure migrating birds and send them off-course in urban areas, causing some to waste precious migration time and energy and others to crash into the sides of nearly invisible high-rise buildings.

Successful migration requires coordination of a variety of physiological, behavioral and ecological systems. "Birds must navigate accurately using multiple cues, time their journeys precisely, deal with winds that may blow them off course, find food in transient and unfamiliar environments, and often re-engineer their bodies to store energy and save weight," said Van Doren.

Furthermore, environmental drivers, such as temperature, atmospheric pressure and precipitation play important roles in determining when large groups of birds take flight. With so many intermingling factors, the ecology of avian migration is exceedingly complicated — in short, it's hard to know exactly when birds of a feather will flock together.

"Studying the movements of one species, let alone the hundreds, is a fundamental challenge," said Cornell University co-author Kyle Horton. "Because the drivers of avian migration are so complex, we need a system that captures that complexity."

To build that system, Van Doren and Horton used data from the meteorological Next-Generation Radar (NEXRAD) network and powerful machine learning techniques to create a forecasting model which demonstrates an exceptional ability to predict the movement of birds with high spatial accuracy and at a continental scale.

Though the NEXRAD weather radar network is designed to detect weather systems, which aid in forecasting the next bout of rain, they also pick up a surprising number of flying creatures like groups of migratory songbirds flying under the cover of night.

"Ornithologists have been using the U.S. network of radar to study bird migration since it first came online decades ago, but this initially required a great deal of manual labor," said Van Doren.

"That's where applications in machine learning come into play, allowing us to accurately predict when and where migrants are likely to take to flight," added Horton.

On radar, the signatures of groups of migrating birds are subtle, ghostly low- to mid-energy echoes, which can be difficult or time-consuming to identify within each radar sweep. However, using machine learning to automate the identification of these signals amid the sometimes-chaotic background of other signals, such as those from precipitation or airborne swarms of insects, the researchers were able to track the springtime flights of millions of migrating birds nationwide for the last 23 years.

The study shows the value of "big data" ecology research, particularly in understanding the complex nature of bird migration. "We demonstrate that machine learning methods can capture this complexity and yield a tool with direct applications for conservation," said Van Doren. "To finally be able to harness the full potential of this continental and multi-decadal dataset is incredibly exciting."

According to Van Doren and Horton, most birds in North America — particularly songbirds like warblers, sparrows and thrushes — migrate at night, making use of a cooler, more stable atmosphere and allowing daylight hours to be spent searching for food at stopover sites to fuel up for the next leg.

"It's amazing to think that billions of birds enter and leave North America each year, all while we're asleep," said Van Doren.

"Knowing when and where these birds will be flying has the potential to reduce mortality, for instance shutting down wind turbines during big flights or turning off lights, which escalate the threat of collisions," said Horton.

Van Doren and Horton also suggest that the predictions could help global health workers in tracking the spread of avian-borne diseases.

Van Doren hopes that bird migration forecasts will become a useful tool for decreasing these negative impacts on birds while engaging people with the "amazing phenomenon that is bird migration."

"Birds are visible and charismatic representatives of our natural world. People have connected with birds for millennia, and birds' migrations connect people continents apart," said Van Doren.

"By counting their numbers and tracking their migrations, birds can inform us about the state of the natural world and teach us how animals are responding to change."

Author

Walter Beckwith