The scientists, using data on 7,895 previously identified craters and 1,411 dated craters, were able to apply machine learning to train a deep neural network. With information from China’s first and second lunar orbiters — Chang’e 1 and Chang’e 2 — the network identified 109,956 new craters. The two unmanned spacecraft launched in 2007 and 2010, respectively.
“Impact craters (are) the most diagnostic features of the lunar surface. That is in great contrast to the surface of the Earth. It is very difficult to trace the Earth’s history of being impacted by asteroids and comets over the past 4 billion years,” said study author Chen Yang, of the College of Earth Sciences at Jilin University and the Key Laboratory of Lunar and Deep Space Exploration at the Chinese Academy of Sciences.
“Earth and the moon have been struck by the same impactor population over time, but large lunar craters have experienced limited degradation over billions of years. Therefore, lunar impact craters can trace the evolution of the Earth,” she said via email.
The craters on the moon lack water, an atmosphere and tectonic plate activity — three forces that erode the Earth’s surface, meaning that all but the most recent meteor impacts aren’t visible.
This latest study isn’t the first to deploy machine learning to detect lunar craters, said Mohamad Ali-Dib at the Institute for Research on Exoplanets at the University of Montreal.
“Machine learning can be used to detect craters on the moon,” he said via email. Craters are “a window into the dynamical history of the solar system.