Application of multisensor remote sensing and Machine Learning for mapping alteration minerals in the Mountain Pass District, California, USA

Mapping alteration minerals in carbonatite-hosted mineral systems is challenging because diagnostic spectral responses are distributed across multiple lithologic units, structurally controlled zones, and hydrothermal overprints. In the Mountain Pass District of California, this challenge is important because alteration minerals and carbonate-rich host rocks are spatially associated with a well-known rare earth element (REE) system. This study develops an integrated multisensor remote sensing and machine-learning workflow to improve alteration mineral mapping in this geological setting using Landsat-9, ASTER, and Sentinel-2 imagery. Band ratio analysis, principal component analysis (PCA), fuzzy logic modeling, and supervised machine-learning classification were combined to identify and map iron oxides, hydroxyl-bearing minerals, and carbonate minerals that reflect hydrothermal alteration and lithologic variability across the district. The integration of multispectral datasets improved spectral discrimination and delineation of alteration zones, while Random Forest and Support Vector Machine classification achieved predictive performance, with overall accuracies greater than 90% and Kappa coefficients exceeding 0.85 based on validation with U.S. Geological Survey field data. The resulting maps highlight alteration-rich and carbonate-associated zones near known mineralized areas and provide an interpretable framework for regional targeting and follow-up investigation. Because the multispectral workflow does not directly detect REE-bearing mineral phases or quantify REE distribution, the outputs should be interpreted as proxy-based exploration-guidance layers rather than direct REE prospectivity maps. Nevertheless, the study demonstrates that integrating Landsat-9, ASTER, and Sentinel-2 with statistical and machine-learning methods offers an effective approach for alteration mineral mapping and for guiding follow-up geochemical, mineralogical, and hyperspectral investigations in carbonatite-related terrains.