Evaluation of SWIR and TIR hyperspectral remote sensing applied to quantitative mineralogy of hydrothermally altered rocks in Las Bambas, Peru

Abstract

This study applies hyperspectral shortwave infrared (SWIR) ( = 1-3 m) and thermal infrared (TIR) ( = 5-15 m) spectroscopy to hydrothermally altered rocks from the Las Bambas Porphyry Cu skarn cluster, southeastern Peru. The goal is to understand how spectrally derived mineral abundances relate to actual mineral abundances determined through traditional techniques, especially the pixel-counting method and x-ray diffraction (XRD). A total of 46 bulk rock samples including various types of skarns, intrusive rocks, and metasedimentary rocks were scanned by the HySpex SWIR sensor to produce mineral maps. Thin sections of the corresponding samples were scanned by Fourier-transform infrared (FTIR) micro-spectroscopy in the TIR spectral range. The produced images were processed using principal component analysis (PCA). Spectral Angle Mapper classification (SAM) was applied to categorize the minerals from the images. The mineral abundances were generated by pixel-counting, analogous to traditional point counting. The calculated mineral abundances were compared with those derived from the powder X-ray diffraction (XRD) analyses. SWIR results correlate well with XRD results for carbonates (R2 = 0.99), though this correlation is influenced by the fact that our rocks typically have either very little or very abundant carbonate minerals. For mineral abundance estimates calculated by thermal infrared analysis, the technique is most effective in estimating the abundance of feldspar, garnet, pyroxene, and carbonate minerals, with the R2 values between 0.75 – 0.99. Errors are largely due to differences in spatial sampling of each technique with regard to the rock texture, which can include coarse to very finely crystalline minerals. The abundance of pyroxene is more accurately estimated in the intrusive rocks than in the skarns. This is because pyroxene grains are distributed evenly as large and euhedral crystals in the intrusive samples. Mineral mapping at fine scales to regional scales is a critical element of exploration for economic mineral deposits, efficient development of mine sites, and monitoring and minimization of environmental impact from resource extraction and processing. This study demonstrates the potential of SWIR and TIR spectroscopy for quantifying mineral abundances in a controlled setting and illuminates some of the problems related to mineral detectability even in pixels with little or no spectral mixing. This result helps shed light on the geology of the Las Bambas ore deposits. Perhaps more importantly, the results inform the application of spectroscopic methods to mining and mineral exploration deployable at laboratory, field or remote sensing scales.