Beijing An international team of researchers has identified 109,000 effective craters on the moon using machine learning methods that have not been previously identified. The study, led by Jilin University researchers, was published in the journal Nature Communications.
Effective craters are the most prominent feature of the lunar surface and occupy most of the lunar surface.
Craters with irregular and severely degraded effects are usually difficult to detect with traditional automatic identification methods.
According to the report, in order to identify the craters and estimate their age, the researchers applied a transfer learning method and trained deep neural networks with data from previously identified craters.
By combining data collected by China’s Chang-1 and Chang-2 lunar probes, researchers identified 109,956 new effective craters. He also estimated the age of 18,996 newly discovered craters, with a diameter of over 8 kilometers.
Yang Chen of Jilin University, one of the researchers, said that the moon crater database is of great importance for scientific research on the moon.
“The strategy adopted can be applied to aid in crater study, which gives reliable suggestions for planetary research,” Yang said.
Yang stated that this research model has been applied to identify small impact pits at the Chang-5 probe site.
Leave a Reply