Mapping mineral chemistry of a lateritic outcrop in new Caledonia through generalized regression using Sentinel-2 and field reflectance spectra

Mining is fundamental for human development, yet it currently requires innovative spatial techniques as it faces diverse environmental and social pressures. With the free Sentinel-2 data of the Copernicus programme, new opportunities arise for studies related to nickel laterite, especially with its reported potential in mapping iron-oxide. This work utilizes samples from drill-holes extracted from Tiebaghi, New Caledonia. The chemical composition and the hyperspectral reflectance of each sample are obtained. The reflectance spectra are resampled to Sentinel-2's characteristics, and generalized linear regression was used to accurately predict Fe2O3, MgO, SiO2, Al2O3, and nickel content where three regression approaches were compared: Ridge, Elastic Net, and the Least Absolute Shrinkage and Selection Operator (LASSO). With the resulting regression models, mineral chemistry of an outcrop in the vicinity of the drill-holes is mapped by a scene of Sentinel-2. The work shows the great potential of free satellite imagery in mapping chemical characteristics of minerals and rocks. It opens up great opportunities for monitoring outcrops and for achieving more efficient mineral exploration.