Lithological mapping enhancement by integrating Sentinel 2 and gamma-ray data utilizing support vector machine: A case study from Egypt

Abstract

Hybrid data fusion mostly gives a better diagnosis to lithological units compared to single-source mapping techniques. Rock unit discrimination depends mainly on variations in the concentrations of chemical elements. Remote sensing datasets reflect these variations as different spectral reflectances, while gamma-ray spectrometric measurements enable recording the varied concentrations of K, Th, and U in these rock units. Accordingly, in this study, we use Support-Vector Machine (SVM) learning algorithm to classify combined high spectral resolution Sentinel 2 data with K, Th, and U content of the rocks to better differentiate a lithologically complex area in Egypt. SVM classifier has been trained and tested on a reference map (built from FCCs, principal and independent component analysis of remote sensing images, as well as previous geological maps) to allocate 13 lithological targets. K, Th, U, and total count maps are interpolated using the inverse distance weighted (IDW) method, cubically resampled, and fused with Sentinel 2 data. We concluded that incorporating any single chemical concentration in the allocation gives better results than using remote sensing data solely and raised the Overall Accuracy by 4.14%, 5.11%, and 6.83% by adding U, K, and Th, respectively. Moreover, blending the total count band (K + Th + U) with Sentinel 2 data outstandingly boosts the classification accuracy by 7.77 %. We performed field reconnaissance to verify the classification results. The study demonstrates the effectiveness of integrating Sentinel 2 data with airborne geophysical spectrometric data, and the proposed approach may prove a more precise and sophisticated lithological map.