In ancient Rome, high priests divined events by the flights, types and songs of birds. Augurs, as the Roman officials were called, advised emperors. They planned cities. They sanctioned war.
Birdcast, a technological application in its first year of development at the Cornell Lab of Ornithology, is inaugurating a scientific version of that dated art. The project aims to provide real-time forecasts of bird migration, much like a weather forecast substituting ‘migrations’ for ‘storms’ and ‘birds’ for ‘rain.’
Andrew Farnsworth, research associate in Information Science for Migration Studies and Acoustic Monitoring at the Cornell Lab of Ornithology, announced the Birdcast project to an audience at a press-only event held at the ILR Conference Center in midtown Manhattan last Thursday.
Birdcast developers ultimately plan to have the project predict the complex, dynamic phenomena of bird migrations using an integrated suite of radar, acoustic and ground observation data, he said. The project is a collaborative effort funded by the National Science Foundation and the Leon Levy Foundation, with partners at Microsoft, the National Oceanic and Atmospheric Administration, the University of Massachusetts at Amherst, and the Cornell Lab of Ornithology.
In mid-August, tens of thousands of Blackpoll Warblers, each weighing about as much as a pencil, embark on a journey, Farnsworth said as he described his favorite feat of bird migration. Many eat their body weight, literally doubling their mass, in preparation. The warblers descend from the northern latitudes of North America on a multi-thousand kilometer trip in which they soar above the Atlantic Ocean for several days, he said. Spending dozens of hours aloft, not stopping or feeding en route, the birds eventually touch down in northern South America. Exodus completed.
In advance of predictive computer models, Farnsworth and his colleagues calculate and forecast odysseys like those of the Blackpoll Warbler weekly on a website they just launched: www.Birdcast.info.
“Right now, we’re talking in human terms but in a couple of years, probably beginning next year or the year after, we’re going to start automating the process so that we can use models to tell us what’s going to happen,” Farnsworth said.
Once developed, Farnsworth and his team will assess the computer models against their own forecasts.
“We’re probably going to have some opportunities where I go against the machine and say, ‘Alright, I think the model is wrong in these places. Here’s what I think is going to happen,’” he said.
Mapping Birds With The Doppler Effect
According to Farnsworth, Birdcast developers are taking advantage of the network of about 140 radar stations spanning the continental U.S. that monitors atmospheric phenomena such as weather.
“As it turns out, radar is also extremely good at detecting other things in the atmosphere — notably, birds,” he said.
By subtracting aerial occurrences like rain, insects and bats from radar imagery, the researchers can obtain patterns of bird migration containing information on density, direction and velocity. Birds tend to be incredibly directed, powerful flyers at night when they migrate — usually moving across, and about five to six meters per second above, wind speed. The researchers can distinguish the birds by comparing their speeds with wind speeds that are recorded by balloons. Insects, which mostly float on the wind, and bats, which tend to be less directed fliers, are separable components, too.
Though Farnsworth and his colleagues currently perform such analyses manually, they plan to replace their current methods with simple algorithms. The only problem, though, is that the radar’s pixel-data provides no information about the species that are flying. But Birdcast plans to solve this issue: using acoustics data.
Eavesdropping on Flight Calls
Flight calls are species specific, commonly faint, high-pitched, single note vocalizations that tend to last less than a quarter of a second long, Farnsworth said.
“Most birds use them in social migratory contexts when they are migrating at night to stay in touch with one another,” he said. “In a sense they are trying to say, ‘Where are you?’ ‘What direction are you going?’, ‘Are you going down to land?’”
Benjamin Van Doren ’16 works with Farnsworth on the Birdcast project. One of Van Doren’s tasks is to run “primitive” detectors on hundreds of hours of raw audio recordings, that use software to pick out possible calls; he then filters through the calls and assigns tentative identifications that are, in turn, confirmed or revised by team members such as Farnsworth, he said.
“Once you have those exemplars, you can use them to train more automated systems so someone doesn’t have to go through and spend dozens of hours going through and identifying the calls by hand,” Van Doren said. But one doesn’t have to be an official Birdcast team member to get involved with the project, there are opportunities for regular bird enthusiasts, too.
“For a project like this we’re really interested in the amateur recordists that get inspired to build a microphone, put it on top of their house and send us the information it collects,” Farnsworth said, citing the website www.oldbird.org, which teaches users how to build a microphone for $20 to $50.
“Some of it is very much a ‘DIY’ approach,” he said.
Another project called eBird, contains a database of over 100 million observations from birders over the past 10 years. It provides a platform on which to gather data from a distance, Farnsworth said.
Birdcast researchers can use eBird data, uploaded by citizen scientists to the eBird database, to verify the accuracy of future computer-generated forecasts.
Biomarkers with Wings
Predicting when and where heavy bird migrations occur will help people know the best times to shut off or dim city lights so as not to interfere with a migration, Farnsworth said. And it may also yield insights about what habitats are most important to preserve.
In addition to bird conservation efforts, forecasts may provide valuable information about our global, connected ecosystem, he said. Van Doren wrote in an email, “not only can studying migratory birds help shed light on the birds themselves, but the health of entire ecosystems consisting of many, many other species.
“Birds, because they are so visible, are great bioindicators: if a bird species is in decline, it may mean something is wrong with the ecosystem or ecosystems in which it lives, but that may be difficult to detect otherwise.”
Farnsworth noted that bird migration is, at a basic level, a response to climate change, adding that this year saw patterns where birds were arriving two to three weeks ahead of their typical times.
“Birds are able to predict what’s coming in the future in a way that we can’t yet because we’re not tuned into whatever they’re sensing,” Farnsworth said.
“It certainly seems like they can do it both at the very minute level scale of, say, freaking out before an earthquake, and also months in advance, anticipating when a drought might occur.”
Farnsworth mentioned that Dickcissel, a blackbird relative, may have predicted something that most market analysts got wrong about this past summer.
“Corn futures suggested that this year was going to be a huge corn production year,” Farnsworth said. “But it turned out it was a complete disaster because there was a dramatic drought.”
Dickcissel, he explained, tend to disperse and move around a lot in years when there will be drought.
“It looked like these birds were doing that long before the drought actually occurred," he said. Perhaps Augurs were on to something.