Identification of regolith-hosted Nb-REE deposits in alkaline rocks using multisource spectroscopic system: A case study of the Puxiong deposit, Yunnan Province, China

Abstract

Regolith-hosted rare earth element (REE) deposits are of global strategic importance, with China supplying over 90% of the world’s medium and heavy REEs. However, exploration in subtropical regions is hindered by dense vegetation, variable weathering intensities, and complex lithological settings. This study addresses these challenges by focusing on the Puxiong Nb-REE deposit in Yunnan Province, China, and by employing a cost-effective methodology that integrates ASTER and Sentinel-2 multispectral data with ALOS PALSAR DEM and GF-3 SAR imagery. As a complementary to hyperspectral and airborne approaches, our method demonstrates the feasibility of extracting REE-related information from multisource remote sensing data. Techniques including band ratio analysis, principal component analysis (PCA), constrained energy minimization (CEM), directional filtering, and K-means clustering were combined with geological and geomorphological data to delineate potential REE zones. A key innovation of this study is the use of Nd³⁺ abundance as a direct tracer for Nb-REE localization, enhanced by a regression model that incorporates indicator minerals (e.g. kaolinite, montmorillonite, hematite, chlorite, calcite, illite, and halloysite) to refine predictions. The integration of ALOS PALSAR DEM enabled detailed geomorphological analysis, while GF-3 SAR imagery improved the detection of structural features, further refining the delineation of potential zones. Results reveal a strong spatial correlation between Nd³⁺-rich areas and favorable lithological units, such as alkaline rocks and residual slope deposits, as well as regions with moderate-to-high indicator mineral assemblages abundance. Six high-potential zones were identified, including the Changlinggang-Baiyunshan and Jiasha areas, with validation by comparison against known geochemical datasets confirming the accuracy of these predictions. The proposed multi-factor constrained model highlights the potential for scalable and efficient exploration of regolith-hosted REE deposits, particularly in subtropical regions, by integrating multisource remote sensing and geological insights. This study provides a robust framework for enhancing exploration strategies for regolith-hosted Nb-REE deposits across diverse metallogenic environments.