Huge asteroids have crashed into Earth before – the RIP dinosaurs – and if we don’t keep an eye out for all these errant space rocks, they could crash into our world again, with devastating consequences. That’s why Ed Lu and Danica Remy of the Asteroid Institute have started a new project to track as many as possible.
Lu, a former NASA astronaut and executive director of the institute, led a team that developed a new algorithm called THOR, which harnesses massive computing power to compare points of light seen in different images of the night sky, then associates them to reconstruct a trajectory of an asteroid in the solar system. They have already discovered 104 asteroids with the system, according to an announcement they released on Tuesday.
While NASA, the European Space Agency and other organizations have their own asteroid searches underway, they all face the challenge of analyzing telescope images containing thousands or even 100,000 asteroids. Some of these telescopes do not or cannot take multiple images of the same region on the same night, making it difficult to tell if the same asteroid appears in multiple photos taken at different times. But THOR can make the connection between them.
“The magic thing about THOR is that he realizes that out of all these asteroids, this one in a certain frame, and this one in another frame four nights later, and this one seven nights later are all the same object and can be put together like the trajectory of a real asteroid,” says Lu. in the direction of Earth. Such a formidable task would not have been possible with older, slower computers, he adds. “It shows the importance of computation in advancing astronomy. What drives this , is that computing is becoming so powerful, so cheap, and so ubiquitous.
Astronomers typically spy on asteroids with what’s called a “tracklet,” a vector measured from multiple images, usually taken within an hour. These often involve an observation model with six or more images, which researchers can use to reconstruct the asteroid’s course. But if the data is incomplete, for example because a cloudy night obstructs the telescope’s view, then that asteroid will remain unconfirmed, or at least not found. But that’s where THOR, which stands for Tracklet-less Heliocentric Orbit Recovery, comes in, helping to determine the trajectory of an asteroid that would otherwise have been missed.
While NASA benefits from telescopes and surveys dedicated to detecting potentially dangerous asteroids, other datasets abound. And THOR can use almost any of them. “THOR turns any astronomical dataset into a dataset where you can search for asteroids. This is one of the most interesting aspects of the algorithm,” says Joachim Moeyens, co-creator of THOR, fellow at the Asteroid Institute and graduate student at the University of Washington. For this first demonstration, Moeyens, Lu and their colleagues searched billions of images taken between 2012 and 2019 from telescopes managed by the National Observatory of Optical Astronomy, many of them by a sensitive camera mounted on the Blanco telescope of 4 meters in the Chilean Andes.