The Moon has a lot more craters than we thought

More than 109,000 new craters have recently been identified in the low and mid-latitude regions of the Moon. All were spotted using artificial intelligence powered by data collected by Chinese orbiters.

Lunar impact craters, formed during meteor strikes, can be thought of as "fossils" recording the history of our solar system. These marks vary in size, shape, and can also overlap and erode over time. This is why their identification remains very complicated and time-consuming.

To overcome these difficulties, Chinese astronomers recently used machine learning, training a deep neural network (a computer using layers of mathematical calculations that feed the inside each other) with data from thousands of previously identified craters. Once these notions "learned", the algorithm would then take care of discovering new ones.

The face of the Moon is more "marked" than previously thought

As part of this work, the researchers "fed" this AI with data collected by the Chang'e-1 and Chang'e-2 lunar orbiters, revealing

strong>109,956 additional craters on the surface of the Moon. The number of craters recorded on the surface of the Moon is now more than a dozen times greater than before.

"This is the largest lunar crater database with automatic extraction for the mid and low latitude regions of the Moon “, said the study’s lead author Chen Yang, of Jilin University in China.

The vast majority of these brands are "small" to "medium" in size (from 1 to 100 kilometers in diameter), reads the review NatureCommunication . A few of these craters, on the other hand, are much larger, irregularly shaped and very eroded, measuring up to 550 km in diameter .

Thanks to this algorithm, Chinese researchers were also able to date nearly 19,000 of these craters. These covered the five geological periods of the Moon , some dating back around four billion years.

These same researchers now plan to continue this work by feeding their algorithm with data collected by the Chang'e 5 lander, which recently brought back to Earth the first lunar samples in more than forty years. They also aim to adapt and apply their machine learning approach to other bodies in the solar system, such as the planet Mars.