Machine learning-based mapping of pyroxene on global lunar surface

The advancement of remote sensing technology has created a gargantuan dataset with high spatial and spectral resolution enabling the use of Machine Learning (ML) methods for analysis. The Moon Mineralogy Mapper (M3) imaging spectrometer onboard the Chandrayaan-1 mission captured hyperspectral data of the lunar surface within the wavelength of 430-3000 nm in 85 contiguous hyperspectral bands. In this study, the distribution of pyroxene on the global lunar surface is mapped with M3 hyperspectral data using the support vector machine classifier (SVM) and reflectance spectra as a feature, and the result was compared with the previous mapping result. This study demonstrates that the use of machine learning is an efficient way for mapping of minerals using hyperspectral data on the lunar surface, and this technique can be used to map minerals of other planetary bodies as well in the future.