A meteorite-hunting AI will scout for space rocks buried in polar ice


Antarctic meteorite expeditions can be grueling, but scientists designed a new AI that could help locate hotspots.

Meteorites are the records of our solar system.Examples By studying the remains of planets that have formed and evolved for literally eons, scientists can learn important cosmological insights.

But before they can do that, they must locate these elusive cosmic gems. So far, about 62 percent of meteors recovered on Earth have been found in Antarctica. Previous meteorite expedi­tions to the world’ s coldest and iciest continents have relied on pure luck, and the past experiences of experienced space rock hunters.

Striking meteorite finds are also hard won. Most trips are usually hindered by the sheer amount labor it takes to unearths a find. Scientists have developed an alternative, low-cost method for finding meteorite stranding sites, the areas where space rocks crash into Earth.

A machine learning algorithm was used to develop an AI that could predict the probability that a meteorite would hit any point in Antarctica. According to the study conducted by the researchers, the system was able to identify over 80% of previously known MSZs in that particular region.

Researchers are confident that it will be able to conquer unexplored territories. “It has never been done before on a continental scale and on a local scale,” says Veronica Tollenaar from the University of Brussels. This type of data driven analysis has never been done before.

While searching in the scientific literature, there are good reasons why scientists still rely on their intuition, which is usually informed by important research and previous experience. But, she adds, their tool can be used to validate claims. The AI is able not only to verify or refute the ‘gut feeling’ many experts follow to track meteorites, but also curb expensive meteorite recovery efforts at the same time.

Tollenaar and his team trained a machine learning algo­rithm to recognize 2,254 places where researchers found meteorites, and 2,1 million more where they didn’t. They then created an index that ranks MSPs according to the ease of a potential field visit and the success of a potential field trip.

Some of these unexplored zones were even found to be located near a number of Antarctica’s 70 permanent research stations.

To accurately map the entire continent and create predictions, a technique uses satellite-based data on Antarctic surface temperature, surface velocity (speed), surface slope, and radar reflectivity (backscatter)—when electromagnetic energy is reflected back from the surface of the earth to its source. The team needed to take into account these factors because when meteors fall to the ground, most often they become trapped in Antarctica‘s thick, snow-capped ice sheet, which covers almost all of the continent.

These meteorites are carried under the region by ice flows, and many of them end up at the bottom of the ocean. But some particularly lucky individuals rise back to the surface from beneath the ice in what are known as “blue ice” areas, places where snow melts faster than its accumulation. These areas are characterized by a distinctive blue hue caused by snow melt.

Some of these old space rocks come from between Earth and Mars, but those of a more rarefied pedigree are lunar meteorites: ones that some scientists say hail from the dark side (the far side) of the moon. Regardless of their celestial origins, all of these rocks can really tell us a lot regarding how the Earth looks from its interior.

According to Zekolari, who has visited Antartica once before, exploratory expeditions can be difficult, however the scientific value of recovered meteors is too valuable to pass up. “A lot of the things we know about how old our solar systems are, we really know them because of meteorites.” says Zekolari.

Of the 45,00 meteorites scientists have already found, Tollenaars team expects about 300,00 more still lie somewhere beneath Antartica’s dense ice sheet. Their study also suggests that since the first Antarctic Meteorite was found in 1912, less than fifteen percent of all those at its surface have ever been recovered!

Zekollari explains that some of the discrepancy between fieldwork studies and lab experiments stems from how fieldwork studies are usually conducted. “So, so far being in the field has been just driving around, typically in a snow scooter, and we’ve been just looking for meteorites,” he says. “But we might be missing quite a lot.”

Zekollari‘s dream is to use drones in a similar fashion as their artificial intelligence. This new tool could not only help expeditions make their own individual searches more efficient, but could also improve meteorite exploration missions worldwide.

Zekollari‘s dream is to use drones in a similar fashion as their AI. This new tool could not only help expeditions make their individual searches more efficient, but could also improve meteorite exploration missions across the globe.

As for what‘s next for the project: the study authors are already preparing for a potential mission to the Antarctic sometime in the next year. They hope to test the new AI algorithm there.

“We want as many meteors as possible, and we hope everyone finds them. With this study, we try our best to point people in the right direction,” Zekolari says.


Please enter your comment!
Please enter your name here