Hyperspectral proximal sensing for carbonate rocks characterization in the SWIR (Short-Wave Infrared)

In this work proximal hyperspectral images are used to make a compositional map of one sample of ornamental carbonate rock (formed by variable content of dolomite and calcite) in terms of mineralogical composition. The hyperspectral dataset consists of a total number of 278 bands corresponding to the short-wave infrared (SWIR) wavelengths. After visual analysis and interpretation, 8 points or pixels were selected as reference spectra for image classification through the Spectral Angle Mapper (SAM) algorithm. The results show the spatial distribution of calcite and dolomite based on their characteristic and diagnostic absorption features (at 2335 and 2315 nm respectively), and areas with different proportions of calcite and dolomite mixture, and the presence of carbonates and clay mineral mixtures. The applied technique demonstrates the potential of hyperspectral proximal sensing procedures for mineral characterization of samples in the laboratory, expanding the application for the analysis and interpretation in field outcrops.