Classification of Phosphate Sedimentary Facies and Estimation of Carbonate-Fluorapatite Abundance Using Hyperspectral Infrared Imaging

Over recent years, the use of hyperspectral infrared imaging has significantly increased in the mining sector, offering numerous applications from geological exploration and mining to sorting and the rehabilitation. However, this technology remains underutilized in the phosphate mining industry, particularly in countries like Morocco, where phosphates represent over 70% of the world's reserves. In this study, the objective is to investigate the use of hyperspectral infrared imagery as a tool to identify and characterize sedimentary phosphate facies for automated facies core logging applications as well as to identify the spectral signature of Carbonate-Fluorapatite (CFA), the primary phosphate mineral phase in sedimentary phosphates, in order to estimate its abundance.To achieve this, six samples have been carefully selected from the Benguerir phosphate sequence to represent the commonly encountered indurated facies. The samples were scanned using a core scanner equipped with three hyperspectral sensors: a Visible Near-Infrared (VNIR) camera, a Short-Wavelength Infrared (SWIR) camera, and a Medium-Wavelength Infrared (MWIR) camera. The covered wavelength interval ranges from 0.4 µm to 5.3 µm, with spatial resolutions varying from 0.117 mm/pixel to 0.228 mm/pixel. Eight facies were identified in the studied samples and characterized through petrography and XRF geochemical analysis of the whole rock. Subsequently, a spectral library was established for each of these facies. Moreover, a sample area rich in CFA was selected and characterized by automated SEM using Tescan Integrated Mineral Analyzer (TIMA). The results indicate that all the facies exhibit distinguishable signatures in the various VNIR, SWIR, and MWIR intervals. However, the SWIR and MWIR intervals proves to be the most effective sensors for distinguishing these facies. The results indicate also that the Spectral Angle Mapper (SAM) is the most efficient method, achieving an overall accuracy of 98,75% in distinguishing the studied facies in the MWIR wavelength range. Additionally, several statistical methods were also tested to estimate the abundance of CFA using the spectral signature derived from the comparison between the SEM mineral maps and corresponding hyperspectral images. Band rationing (B(3.4µm)/B(4.7µm)) * (B(3.4µm)/B(3.9µm)) has demonstrated effective in identifying and estimating the abundance of CFA demonstrating the potential of hyperspectral imaging as a rapid and cost-effective method for the characterization of phosphates in terms of their apatite content.