Predicting Bird Migration Using AI and Weather Radar

Artificial intelligence (AI) is being used more and more these days to predict bird migration using old weather radar imagery.

Here’s how it works:

Mass migrations occur twice a year, in spring and autumn. More than 4 billion birds are estimated to fly over North America.

Experts say more than 500 million people could move overnight during peak migration season. The journey is difficult as winged warriors must travel hundreds to thousands of miles and contend with extreme weather, food shortages and light pollution.

Bright, steady light can confuse birds, alter their normal migration paths, and in some cases cause death or injury. This is one of the reasons the FAA mandated flashing lights on buildings and towers in his 2016.

And this allure of light is one reason New York City’s Central Park is one of the best birding spots on the East Coast. Attracted by the bright lights, various migratory birds gather and use the park’s vast forests as a resting place.

Another example of the effect of light on bird migration is the annual 9/11 memorial light. During migration season in September each year, thousands of floodlights are installed near the World Trade Center grounds.

On a clear night, its light can be seen from over 80 miles away. Environmentalists soon noticed that many dead and injured birds were appearing near the monument. The birds were drawn to the light and flew around it until they were exhausted.

Volunteers from the Audubon Society of New York City are now stationed at the floodlight each year. They work two-hour shifts, counting birds. When the number of birds exceeds 1,000, the lights are turned off for 20 minutes to allow the birds to escape the crowds of the exhibit.

Once upon a time, ornithologists, scientists who studied birds, tracked their movements through markers. They attached colored plastic or metal bands to the legs of the birds, and scientists from other parts of the world observed the birds and shared with the bands which species they observed.

As you can imagine, the numbers of migrating birds are so large that the markers capture only a small portion of the population movements each spring and fall. Additionally, we were unable to accurately predict which nights saw more migration activity, so we can warn communities to reduce the use of outdoor lighting.

But in 2018, the science of bird migration made great strides. Birdcast was born.

This is a project developed and operated by Colorado State University, Cornell Institute or Ornithology, and the University of Massachusetts.

BirdCast uses artificial intelligence, weather and historical weather radar data to predict the intensity of bird migrations up to three days in advance.

Since the 1940s, weather radars have been able to detect objects such as insects, birds, rain and snow. As weather radar technology improved, data was stored in digital form.

Roosting birds display a unique signature on the radar when they take off. It looks like a circle on the radar display.

Machines have been taught to remove this confusion when chasing storms. This same filter is now being used to scrutinize past radar records to create decades-long migration records.

The computer was told what the birds looked like on radar compared to precipitation. The computer program was then tasked with examining decades of data, clearing rain and revealing bird migration data. They found that the number of birds flying each night depends mainly on wind, temperature, pressure and precipitation.

Now that ornithologists know the key factors that influence migration, they can use this information to predict future migrations.

The next thing experts want to work on is to find out what kinds of birds migrate. The Cornell Institute for Ornithology has developed software that can identify bird species by listening to their calls.

They hope to integrate this technology into BirdCast to provide a more complete picture of one of the world’s largest migrations.

Our team of meteorologists digs deep into meteorological science and analyzes timely weather data and information. For more weather and climate articles, check out our Weather Blog section.