Connecting Meteorite Spectra to Lunar Surface Composition Using Hyperspectral Imaging and Machine Learning

We present an innovative, cost-effective framework integrating laboratory Hyperspectral Imaging (HSI) of the BECHAR 010 lunar meteorite with ground-based lunar HSI and supervised Machine Learning (ML) to generate high-fidelity mineralogical maps. A \SI{3}{\milli\metre} thin section of BECHAR 010 was imaged under a microscope with a \SI{30}{\milli\metre} focal length lens at \SI{150}{\milli\metre} working distance, using 6x binning to increase the signal-to-noise ratio, producing a data cube (X $\times$ Y $\times$ $\lambda$ = $791 \times 1024 \times 224$, \SI{0.24}{\milli\metre} $\times$ \SI{0.2}{\milli\metre} resolution) across \SIrange{400}{1000}{\nano\metre} (224 bands, \SI{2.7}{\nano\metre} spectral sampling, \SI{5.5}{\nano\metre} FWHM spectral resolution) using a Specim FX10 camera. Ground-based lunar HSI was captured with a Celestron 8SE telescope (\SI{3}{\kilo\metre}/pixel), yielded a data cube ($371 \times 1024 \times 224$). Solar calibration was performed using a Spectralon reference (\SI{99}{\percent} reflectance \SI{< 2}{\percent} error) ensured accurate reflectance spectra. A Support Vector Machine (SVM) with a radial basis function kernel, trained on expert-labeled spectra, achieved \SI{93.7}{\percent} classification accuracy (5-fold cross-validation) for olivine (\SI{92}{\percent} precision, \SI{90}{\percent} recall) and pyroxene (\SI{88}{\percent} precision, \SI{86}{\percent} recall) in BECHAR 010. Local Interpretable Model-agnostic Explanations (LIME) identified key wavelengths (e.g., \SI{485}{\nano\metre}, \SI{22.4}{\percent} for M3; \SI{715}{\nano\metre}, \SI{20.6}{\percent} for M6) across 10 pre selected regions (M1 to M10), indicating olivine-rich (Highland-like) and pyroxene-rich (Mare-like) compositions. Spectral Angle Mapper (SAM) analysis revealed angles from \SI{0.26}{\radian} to \SI{0.66}{\radian}, linking M3 and M9 to Highlands and M6 and M10 to Mares. K-means clustering of lunar data identified 10 mineralogical clusters (\SI{88}{\percent} accuracy), validated against Chandrayaan-1 Moon mineralogy Mapper ($\rm M^3$) data (\SI{140}{\metre}/pixel, \SI{10}{\nano\metre} spectral resolution). A novel push-broom HSI approach with telescope, achieves 0.8 arcsec resolution for lunar spectroscopy, inspiring full-sky multi-object spectral mapping.