Refinement of lunar TiO2 analysis with multispectral features of Chang’E-1 IIM data

Abstract

We develop a method based on the samples from Apollo and Luna landing sites to determine lunar TiO2 content with Chang’E-1 interference imaging spectrometer (IIM) imagery. By analyzing the nonlinear relationship between the optical and compositional parameters of lunar soil samples, the method employs two Support Vector Machines (SVMs) to estimate the titanium abundance of the lunar surface. Developed with the soil compositions of the Apollo and Luna sample-return stations, the RMS (root mean square) error of our method is 0.24 wt% TiO2, and the correlation coefficient of the TiO2 values and our predicted ones is 99.72 %. Compared with the other 3 models, the method proposed in this paper exhibits a good performance for determining the chemical composition of the lunar surface. TiO2 maps of Sinus Iridum, part of the Marius Hills plateau, and part of Mare Smythii are produced using our method, which could be useful for future lunar missions.