Ed Lu wants to save Earth from killer asteroids.
Or at least, if there is a big space rock streaking our way, Lu, a former Nasa astronaut with a doctorate in applied physics, wants to find it before it hits us — hopefully with years of advance warning and a chance for humanity to deflect it.
On Tuesday, the B612 Foundation, a nonprofit group that Lu helped found, announced the discovery of more than 100 asteroids. (The foundation's name is a nod to Antoine de Saint-Exupéry's children's book "The Little Prince": B612 is the home asteroid of the main character.)
That by itself is unremarkable. New asteroids are reported all the time by sky-watchers around the world. That includes amateurs with backyard telescopes and robotic surveys systematically scanning the night skies.
What is remarkable is that B612 did not build a new telescope or even make new observations with existing telescopes. Instead, researchers financed by B612 applied cutting-edge computational might to years-old images — 412,000 of them in the digital archives at the National Optical-Infrared Astronomy Research Laboratory, or NOIRLab — to sift asteroids out of the 68 billion dots of cosmic light captured in the images.
"This is the modern way of doing astronomy," Lu said.
The research adds to the "planetary defence" efforts undertaken by Nasa and other organisations around the world.
Today, of the estimated 25,000 near-Earth asteroids at least 140m in diameter, only about 40 per cent of them have been found. The other 60 per cent — about 15,000 space rocks, each with the potential of unleashing the energy equivalent to hundreds of million of tons of TNT in a collision with Earth — remain undetected.
B612 collaborated with Joachim Moeyens, a graduate student at the University of Washington, and his doctoral adviser, Mario Juric, a professor of astronomy. They and colleagues at the university's Institute for Data Intensive Research in Astrophysics and Cosmology developed an algorithm that is able to examine astronomical imagery not only to identify those points of light that might be asteroids, but also to figure out which dots of light in images taken on different nights are actually the same asteroid.
In essence, the researchers developed a way to discover what has already been seen but not noticed.
Typically, asteroids are discovered when the same part of the sky is photographed multiple times during the course of one night. A swath of the night sky contains a multitude of points of light. Distant stars and galaxies remain in the same arrangement. But objects that are much closer, within the solar system, move quickly, and their positions shift over the course of the night.
Astronomers call a series of observations of a single moving object during a single night a "tracklet". A tracklet provides an indication of the object's motion, pointing astronomers to where they might look for it on another night. They can also search older images for the same object.
Many astronomical observations that are not part of systematic asteroid searches inevitably record asteroids, but only at a single time and place, not the multiple observations needed to put together tracklets.
The NOIRLab images, for example, were mainly taken by the Victor M. Blanco 4 Metre Telescope in Chile as part of a survey of almost one-eighth of the night sky to map the distribution of galaxies in the universe.
The additional specks of light were ignored because they were not what the astronomers were studying. "They're just random data in just random images of the sky," Lu said.
But for Moeyens and Juric, a single point of light that is not a star or a galaxy is a starting point for their algorithm, which they named Tracklet-less Heliocentric Orbit Recovery, or THOR.
The motion of an asteroid is precisely dictated by the law of gravity. THOR constructs a test orbit that corresponds to the observed point of light, assuming a certain distance and velocity. It then calculates where the asteroid would be on subsequent and previous nights. If a point of light shows up there in the data, that could be the same asteroid. If the algorithm can link together five or six observations across a few weeks, that is a promising candidate for an asteroid discovery.
In principle, there are an infinite number of possible test orbits to examine, but that would require an impractical eternity to calculate. In practice, because asteroids are clustered around certain orbits, the algorithm needs to consider only a few thousand carefully chosen possibilities.
Still, calculating thousands of test orbits for thousands of potential asteroids is a humongous number-crunching task. But the advent of cloud computing — vast computational power and data storage distributed across the internet — makes that feasible. Google contributed time on its Google Cloud platform to the effort.
"It's one of the coolest applications I've seen," said Scott Penberthy, director of applied artificial intelligence at Google.
So far, the scientists have sifted through about one-eighth of the data of a single month, September 2013, from the NOIRLab archives. THOR churned out 1354 possible asteroids. Many of them were already in the catalogue of asteroids maintained by the International Astronomical Union's Minor Planet Centre. Some of them had been previously observed, but only during one night and the tracklet was not enough to confidently determine an orbit.
The Minor Planet Centre has confirmed 104 objects as new discoveries so far. The NOIRLab archive contains seven years of data, suggesting that there are tens of thousands of asteroids waiting to be found.
"I think it's awesome," said Matthew Payne, director of the Minor Planet Centre, who was not involved with developing THOR. "I think it's hugely interesting and it also allows us to make good use of the archival data that already exists."
The algorithm is currently configured to find only main-belt asteroids, those with orbits between Mars and Jupiter, and not near-Earth asteroids, the ones that could collide with our planet. Identifying near-Earth asteroids is more difficult because they move faster. Different observations of the same asteroid can be separated farther in time and distance, and the algorithm needs to perform more number crunching to make the connections.
"It'll definitely work," Moeyens said. "There's no reason why it can't. I just really haven't had a chance to try it."
THOR has the ability not only to discover new asteroids in old data but also to transform future observations as well. Take, for example, the Vera C Rubin Observatory, formerly known as the Large Synoptic Survey Telescope, currently under construction in Chile.
Financed by the National Science Foundation, the Rubin observatory is an 8.4m telescope that will repeatedly scan the night sky to track what changes over time.
Part of the observatory's mission is to study the large-scale structure of the universe and spot distant exploding stars, also known as supernovas. Closer to home, it will also spot a multitude of smaller-than-a-planet bodies whizzing around the solar system.
Several years ago, some scientists suggested that the Rubin telescope's observing patterns could be adjusted so that it could identify more asteroid tracklets and thus locate more of the dangerous, as-yet-undiscovered asteroids more quickly. But that change would have slowed down other astronomical research.
If the THOR algorithm proves to work well with the Rubin data, then the telescope would not need to scan the same part of the sky twice a night, allowing it to cover twice as much area instead.
"That in principle could be revolutionary, or at least very important," said Zeljko Ivezic, the telescope's director and an author on a scientific paper that described THOR and tested it against observations.
Returning to the same spot in the sky every two nights instead of every four could benefit other research, including the search for supernovas.
"That would be another impact of the algorithm that doesn't even have to do with asteroids," Ivezic said. "This is showing nicely how the landscape is changing. The ecosystem of science is changing because software now can do things that 20, 30 years ago you would not even dream about, you would not even think about."
For Lu, THOR offers a different way to accomplish the same goals he had a decade ago.
Back then, B612 had its sights on an ambitious and far more expensive project. The nonprofit was going to build, launch and operate its own space telescope, called Sentinel.
At the time, Lu and the other leaders of B612 were frustrated by the slow pace of the search for dangerous space rocks. In 2005, Congress passed a mandate for Nasa to locate and track 90 per cent of near-Earth asteroids with diameters of 140m or more by the end of 2020. But lawmakers never provided the money Nasa needed to accomplish the task, and the deadline passed with less than half of those asteroids found.
Raising US$450 million ($689m) from private donors to underwrite Sentinel was difficult for B612, especially because Nasa was considering an asteroid-finding space telescope of its own.
When the National Science Foundation gave the go-ahead to construct the Rubin observatory, B612 re-evaluated its plans. "We could quickly pivot and say, 'What's a different approach to solve the problem that we exist to solve?'" Lu said.
The Rubin observatory is to make its first test observations in about a year and become operational in about two years. Ten years of Rubin observations, together with other asteroid searches could finally meet the US Congress' 90 per cent goal, Ivezic said.
Nasa is accelerating its planetary defense efforts. Its asteroid telescope, NEO Surveyor, is in the preliminary design stage, aiming for launch in 2026.
And later this year, its Double Asteroid Redirection Test mission will slam a projectile into a small asteroid and measure how much that changes the asteroid's trajectory. China's national space agency is working on a similar mission.
For B612, instead of wrangling a telescope project costing almost a half-billion dollars, it can contribute with less expensive research endeavours such as THOR. Last week, it announced that it had received US$1.3m of gifts to finance further work on cloud-based computational tools for asteroid science. It also received a grant from Tito's Handmade Vodka that will match up to US$1m from other donors.
B612 and Lu are now not just trying to save the world. "We're the answer to a trivia question of how vodka is related to asteroids," he said.
This article originally appeared in The New York Times.
Written by: Kenneth Chang
Images by: B612 Asteroid Institute/University of Washington DiRAC Institute/OpenSpace Project via The New York Times
© 2022 THE NEW YORK TIMES