With new, more advanced telescopes coming online almost annually, the amount of data astronomers are able to collect is astounding – the term petabyte has become more common in my usage these days than ever before. And with all of this new data comes the same old struggle to process it all. This issue is one of the reasons community science projects became popular in the last two decades. We simply needed more eyeballs to find objects like exoplanets and supernovae.

And while community science projects remain popular, one of the hottest topics in research these days seems to be machine learning and artificial intelligence. That’s not to say that human researchers are becoming obsolete, but machine learning can free up those researchers for less trivial tasks than staring at thousands of light curves or images of rocks. Now, instead, they can take the best options from the machine learning results and essentially proofread them.

Not that we want you all to stop counting rocks. That’s important work as well, and not every science project can be solved with machine learning.

But sometimes, really wild science objectives can be solved with machine learning, and in a new paper published in Nature Communications, a team of researchers has used an algorithm to determine the origin of a martian meteorite. When I say ‘origin’, I mean the actual crater from whence the meteorite came from Mars. Wild, right?

The meteorite is cataloged as NWA 7034 and known informally as ‘Black Beauty.’ It’s a brecciated Martian rock, which means it contains sharp, angular fragments of different rock types all cemented together. For those into geology, that definition makes this meteorite a sedimentary rock.

That sedimentary nature is what makes Black Beauty unusual and special – it’s the only brecciated Martian rock available for us to study here on Earth. As lead author Anthony Lagain notes: For the first time, we know the geological context of the only brecciated Martian sample available on Earth, 10 years before NASA’s Mars Sample Return mission is set to send back samples collected by the Perseverance rover currently exploring the Jezero crater.

To understand Black Beauty’s origin, Lagain and his team developed a machine learning algorithm that could analyze a large amount of high-res planetary images from Mars to find impact craters. The algorithm used multiple layers of data collected on Mars with a variety of missions to determine where this particular type of rock could be found and eventually identified the exact crater, now informally named Karratha. Lagain explains the importance of the discovery, stating: Finding the region where the ‘Black Beauty’ meteorite originates is critical because it contains the oldest Martian fragments ever found, aged at 4.48 billion years old, and it shows similarities between Mars’ very old crust, aged about 4.53 billion years old, and today’s Earth continents. The region we identify as being the source of this unique Martian meteorite sample constitutes a true window into the earliest environment of the planets, including the Earth, which our planet lost because of plate tectonics and erosion.

Going forward, the team is also adapting this algorithm to be used to find impact craters on the Moon and Mercury. Co-author Gretchen Benedix notes: This will help to unravel their geological history and answer burning questions that will help future investigations of the Solar System such as the Artemis program to send humans on the Moon by the end of the decade or the BepiColombo mission, in orbit around Mercury in 2025.

And we definitely look forward to seeing just what this latest machine learning algorithm finds. Although, as a reminder, we will have more rocks and boulders and craters for you all to identify when we finish setting up new community science projects during the hiatus, so stay tuned for those.

More Information

Curtin University press release

NAU press release

“Early crustal processes revealed by the ejection site of the oldest martian meteorite,” A. Lagain et al., 2022 July 12, Nature Communications