Deciphering craters on the moon is quick and easy

The surface of the moon tells the story of the inner solar system. Each impacting meteorite leaves its mark, and together those craters hold a record of events on and around the Moon over the past 4 billion years.

But the record can be difficult to read. The age and spatial density of craters are critical measurements for decoding the Moon’s impact history, but analyzing these properties takes time and sometimes requires bringing samples back to Earth.

JH Fairweather and colleagues show in an article published in Earth and space science, machine learning is a quick and easy way to improve our understanding of lunar craters. By training an algorithm on more than 50,000 images, the researchers were able to estimate the age and density of many of the moon’s features.

First, the machine learning algorithm’s estimates differed significantly from those derived by hand by other researchers. But with a bit of manual curation, Fairweather and his colleagues were able to bring their automated estimates of the crater’s age and density into line with previous estimates.

Lighting conditions presented a problem. If the craters were partially shadowed by rocks or located on slopes with uneven illumination, the algorithm had trouble analyzing them accurately. Improved accuracy by eliminating such craters. The presence of rocks or buried craters led the algorithm to overestimate crater ages by 10%–45%, but if rocks, buried craters, and other unwanted material were removed it could determine very accurate ages of young lunar surfaces and impact craters. pictures

Although machine learning can provide a wealth of information about the moon’s surface, the researchers caution that the algorithms still require careful oversight.