Using machine learning methods, scientists from China, together with colleagues from Iceland and Italy, discovered almost 110 thousand previously unknown craters in the low and middle latitudes of the moon, and also determined the age of 18 thousand of them. The research results will form the basis for a new database, and the proposed methods can be used to study other bodies in the solar system. The article was published in the journal Nature Communications.
A team of researchers from Jilin, Trent, and Icelandic Universities used the transfer learning method, which is one of the machine approaches, to detect and count impact craters on the Moon. In this way, scientists using neural networks were able to identify 109,956 new craters in the low and middle latitudes of the Moon.
Most of the surface of the Earth’s satellite is covered with craters, but the results of manual and automatic counting of their number do not correspond to each other. Difficulties arise for a variety of reasons, for example, it is often difficult to detect irregular craters or collapsed craters with automated methods.
The authors trained a deep neural network using data from 7895 previously identified and 1411 dated craters. Using the information collected by the automatic interplanetary stations Chang’e-1 and Chang’e-2, the algorithm was able to find tens of times more craters in the middle and low latitudes of the Moon than all previous methods had detected.
In addition, the neural network estimated the age of 18,996 craters over eight kilometers in diameter. The results of the work will form the basis of a new database of lunar craters. Scientists also believe that their approach can be adapted for other bodies in the solar system.
