Hidden insights into the fall of meteors, long unrecognized in the observational data, are finally being revealed thanks to AI in new research out of Flagstaff, Arizona’s Lowell Observatory.
Detailed in a new paper published in Icarus, researchers used data from the Lowell Observatory Cameras for All-Sky Meteor Surveillance (LO-CAMS) network, part of the Global Meteor Network, which characterized over 28,000 meteor events.
Their work goes a long way toward expanding the parameters used to classify meteors, providing a much deeper understanding of what makes these space rocks so unique.
Meteors Explained
Meteors are objects that burn up in the sky in a brilliant streak across the sky, commonly called shooting or falling stars. Meteorites are those space rocks that manage to hold together and land on the Earth’s surface.
The terminology used to describe these objects can be confusing, as the distinction between a meteoroid, a meteor, and a meteorite may not always be readily apparent. When still in space, we call these rocks meteoroids, but as soon as they enter the Earth’s atmosphere, the differences are noted.
While meteors are a common sight in the night sky and have fascinated humans for millennia, there is still a great deal to learn about them, according to researchers at Lowell Observatory.
“Meteors have been observed for centuries, but only recently have we had datasets large and detailed enough to apply modern machine-learning methods,” said lead author Sam Hemmelgarn. “This allows us to extract physical information that was previously hidden in the data.”
Advancing Meteor Observations
Traditionally, only a few parameters were used to characterize meteors; the new work expanded this to 13, including speed, brightness, duration, height, and atmospheric density.
“Our goal was to move beyond traditional classification schemes,” said co-author Nick Moskovitz. “Modern meteor networks capture a wealth of observational information, and we wanted a framework that could fully take advantage of that.”
The team combined multiple machine learning algorithms to identify natural groupings in the data, which mirrored existing physical meteoroid models. Three key factors that dictate a meteor’s behavior upon atmospheric entry emerged from this analysis. These were its size and shape, how easily it heats up, also known as “activation,” and its path of travel.
“One of the most exciting results was how clearly the ‘activation’ behavior separated asteroidal material from cometary material,” Hemmelgarn explained. “That tells us we’re capturing something fundamentally physical, not just statistical patterns.”
A New Classification Scheme
As a result of their work, the researchers developed Hclass, a new classification system to identify a meteor’s hardness. On the hardest end of the new scale is dense material with a high iron content, generally associated with asteroids, while the distant end contains fragile, porous material likely to come from cometary debris.
The scheme in this new classification system is multi-layered, allowing for more general or more granular classifications depending on the researcher’s need. Additionally, it works with a broad range of datasets, from single digits to millions of observations.
“Hclass gives us a more nuanced view of meteoroid composition,” Hemmelgarn said. “It bridges the gap between classical meteor theory and the realities of modern, large-scale observations.”
The team tested their new scale by fitting data from known meteor showers to it and then examining how those matters behaved in real-world observations. Their validation was successful, with the meteors behaving as expected based on their classifications.
“This work shows that machine learning isn’t just about handling big data,” Moskovitz said. “It’s about turning those data into physical understanding of where this material comes from and how our solar system works.”
The paper, “A Machine Learning Approach to Meteor Classification,” appeared in Icarus on April 27, 2026.
Ryan Whalen covers science and technology for The Debrief. He holds an MA in History and a Master of Library and Information Science with a certificate in Data Science. He can be contacted at ryan@thedebrief.org, and follow him on Twitter @mdntwvlf.