Mapping of potential rare earth deposits in the Schiel alkaline complex using sentinel-2B multispectral sensor

Abstract

In this study, the images of Sentinel-2B multispectral sensor are used to map minerals associated with rare earth elements (REE) in an attempt to discover new potential targets for REE deposits in the Schiel Alkaline Complex of South Africa. Spectral Information Divergence (SID) classification was applied to Sentinel-2B images to map areas containing potential monazite and hematite. Generally, the accuracy of SID classification is warranted by selecting appropriate threshold values, which are normally user-defined. In this study, a simple thresholding approach is proposed; this approach requires at least one area, with a confirmed existence of a mineral being sought, to be used as a reference point for selecting optimum threshold value. The aim of this approach is to determine the lowest threshold value at which mineral classification in a reference point starts. The smaller the threshold value, the more likely the classified pixels represent minerals in question in the ground. Following this approach, a value of 0.02 and 0.03 was used as the maximum acceptable threshold value for mapping REE-bearing mineral (monazite) and iron-rich mineral (hematite), respectively. The proposed approach was validated through field sampling and geochemical analysis of samples collected at potential REE deposit targets, including one rock sample (S3) collected in the area void of minerals of interest, based on SID classification. The analytical results show that the samples collected at anomalous mineral clusters have relatively high REE contents that are well beyond the average crustal abundance and S3, thus suggesting the viability of the proposed approach.