Spy Satellites, Weather Radar, Drones, and Machine Learning Uncover a Meteorite-Strewn Field

Combining open datasets from the US DOD and the Australian Bureau of Meteorology, and drone smarts find the biggest cache since 1969.

A team led by Andy Tomkins, professor at Australia's Monash University's School of Earth, Atmosphere, and Environment, has come across one of the most densely-packed collection of meteorites in the country — using a combination of data from spy satellites, weather radar, and drones to track the meteorites down and machine learning to map them.

The team's mission began in July 2013 when the US Department of Defense (DOD) announced an atmospheric explosion captured by its satellite network and believed to be caused by a six-tonne asteroid, around five feet wide, entering the atmosphere — exploding with a force equivalent to 220 tonnes of TNT.

Hadrien Devillepoix, PhD, at Curtin University’s Space Science and Technology Center (SSTC), immediately seized the opportunity to find fresh meteorites that would have made it to the ground — and used the Austrian Bureau of Meteorology's recently-opened weather radar data to figure out where they might have landed. "We were very excited because a few meteorites have been found using this method using the USA’s radar network," Devillepoix explains, "and if we were to find any from this event, it would be the first time that has been achieved elsewhere."

The weather radar data pointed to a near-four mile wide fall zone where meteorites were likely to be found. With so much area to cover, Tomkins' team turned to more technology to figure out where best to look: a machine learning system developed by Seamus Anderson, which can analyze drone imagery to identify meteorites — and to distinguish them from terrestrial rocks and other debris.

"This is an amazing opportunity to refine our approach to meteorite recovery," Anderson claims. "It typically takes a search team hundreds of hours to fully search such a large area – a drone can do it in less than a day. This is a world first: Using artificial intelligence and machine learning to map a meteorite strewn field."

The approach proved more than successful: the first meteorite was found within ten minutes, and over several days a further 43 were recovered for a total of around nine pounds of extraterrestrial matter. "This discovery is very exciting," says Tomkins, "because it’s the first meteorite strewn field resulting from a new fall event to be defined since the famous Murchison meteorite fall in 1969."

Several of the meteorites recovered from the site are on display at the South Australian Museum's Six Extensions exhibit, while Devillepoix credits their discovery to "the value of open datasets."

Gareth Halfacree
Freelance journalist, technical author, hacker, tinkerer, erstwhile sysadmin. For hire: freelance@halfacree.co.uk.
Latest articles
Sponsored articles
Related articles
Latest articles
Read more
Related articles