Investigating critical element concentrations in topsoils from tropical to subtropical climates using proximal sensor data and auxiliary variables
Soil spectroscopy; Environmental monitoring; Mining prospecting; Machine learning.
Critical elements such as lithium (Li) and rare earth elements (REEs) are essential for modern technology and the global transition to green energy, yet they are increasingly recognized as emerging contaminants. Understanding their natural background levels in soils is therefore crucial for both exploration and environmental assessment. Proximal sensing technologies offer rapid and cost-effective tools to investigate these elements across large and heterogeneous landscapes. This project evaluates the potential of portable proximal sensors to assess Li and REE concentrations in highly weathered tropical to subtropical soils from Minas Gerais (MG), Brazil, an environmentally diverse region that also hosts known deposits of these critical elements. The work is organized into four studies. The first developed a new matrix-matched soil calibration for pLIBS capable of detecting low Li concentrations. The second and third studies used pXRF, Vis-NIR, and auxiliary soil information with Random Forest models to predict Li and REE concentrations, respectively. The fourth applies the Geo-Detector Method to identify the main environmental factors controlling the spatial distribution of these elements in MG soils. We hypothesize that proximal sensors can generate powerful and reliable information for quantifying Li and REEs in soils and for improving the understanding of their behavior in tropical environments. The findings aim to support both mineral exploration and environmental monitoring efforts.