Multiscale (microscopic to remote sensing) preliminary exploration of auriferous-uraniferous marbles: A case study from the Egyptian Nubian Shield

Abstract

Since their recent first record within the Egyptian Nubian Shield, auriferous and uraniferous marbles (Au = 0.98–2.76 g/t; U = 133–640 g/t) have rarely been addressed, despite not only their probable economic importance but also the fact that it is a new genetic style of gold and uranium mineralization in the Nubian Shield rocks. This is mainly attributed to the inadequate localization of these marbles within harsh terrains, as well as the cost and time spent with conventional fieldwork for their identification compared to the main lithological components of the Nubian Shield. On the contrary, remote sensing and machine learning techniques save time and effort while introducing reliable feature identification with reasonable accuracy. Consequently, the current research is an attempt to apply the well-known machine learning algorithm (Support vector Machine—SVM) over Sentinel 2 remote sensing data (with a spatial resolution of up to 10 m) to delineate the distribution of auriferous-uraniferous marbles in the Barramiya-Daghbagh district (Eastern Desert of Egypt), as a case study from the Nubian Shield. Towards better results, marbles were accurately distinguished utilizing ALOS PRISM (2.5 m) pan-sharpened Sentinel 2 data and well-known exposures during fieldwork. With an overall accuracy of more than 90%, a thematic map for auriferous-uraniferous marbles and the major rock units in the Barramiya-Daghbagh district was produced. Marbles are spatially related to ophiolitic serpentinite rocks, as consistent with their genesis within the Neoproterozoic oceanic lithosphere. Field and petrographic investigations have confirmed the newly detected Au and U-bearing zones (impure calcitic to impure dolomitic marbles in Wadi Al Barramiya and Wadi Daghbagh areas and impure calcitic marble in Gebel El-Rukham area). Additionally, X-ray diffraction (XRD), back-scattered electron images (BSEIs), and Energy Dispersive X-ray spectroscopy (EDX) results were integrated to verify our remote sensing results and petrographic investigations. Different times of mineralization are indicated, ranging from syn-metamorphism (gold in Wadi Al Barramiya and Gebel El-Rukham) to post-metamorphism (gold in Wadi Daghbagh and uranium in all locations). Based on the application of geological, mineralogical, machine learning and remote sensing results for the construction of a preliminary exploration model of the auriferous-uraniferous marble in the Egyptian Nubian Shield, we recommend a detailed exploration of Au and U-bearing zones in Barramiya-Dghbagh district and applying the adopted approach to other districts of similar geological environments.

A review on advancements in lithological mapping utilizing machine learning algorithms and remote sensing data

“… In cases where a higher number of classes need to be distinguished, the use of Sentinel 2 is recommended [ 88 ] SVM PALSAR/Sentinel-2 The combination of PALSAR DEM data and Sentinel 2 multispectral data through the SVM algorithm enabled improved differentiation of rock units based on their topographic variations, resulting in a more precise lithological classification. [ 89 ] SVM Sentinel-2 By employing pan-sharpened Sentinel 2 data and SVM, the researchers attained an overall accuracy of over 90% in generating the thematic map. …”

Impact of DEMs for Improvement Sentinel 2 Lithological Mapping Utilizing Support Vector Machine: A Case Study of Mineralized Fe-Ti-Rich Gabbroic Rocks from the South Eastern Desert of Egypt

“…As a supervised technique introduced by Cortes and Vapnik [52], SVM is considered an efficient classifier in setting apart lithological classes through building a decision boundary hyperplane [37,53,54]. Finding a hyperplane that divides data points into distinct classes is the basic purpose of an SVM.…”

“…Toward impartial lithological classifications, several Machine Learning Algorithms (MLAs) have been widely implemented [9,14,[21][22][23][24][25][26][27][28][29][30][31][32]. One of the most powerful algorithms that is mostly utilized and recommended by several studies is the support vector machine [5,9,14,29,31,[33][34][35][36][37][38].…”