Document Type : Original Research Paper

Authors

• Ahmed Mahmoud Abdelhameed ¹
• Maher Abdelateef El Amawy ²
• Ayman Mahrous ³
• Mohamed El-Khouly ⁴
• Adel Fathy ⁵

¹ Department of Space Environment, Institute of Basic and Applied Sciences, Egypt-Japan University of Science and Technology (E-JUST), New Borg El-Arab City, Alexandria 21934, Egypt. B) Geology Department, Faculty of Science, Sohag University, Sohag 82524, Egypt

² Department of Geology, Faculty of Science, Benha University, Benha 13518, Egypt

³ Department of Space Environment, Institute of Basic and Applied Sciences, Egypt-Japan University of Science and Technology (E-JUST), New Borg El-Arab City, Alexandria 21934, Egypt. B) Department of Physics, Faculty of Science, Helwan University, Helwan, Cairo 11795, Egypt

⁴ Nanoscience Program, Institute of Basic and Applied Science, Egypt-Japan University of Science and Technology (E-JUST), New Borg El-Arab City, Alexandria 21934, Egypt

⁵ Department of Space Environment, Institute of Basic and Applied Sciences, Egypt-Japan University of Science and Technology (E-JUST), New Borg El-Arab City, Alexandria 21934, Egypt. B) Physics Department, Faculty of Science, Fayoum University, Fayoum 63514, Egypt

Abstract

Hyperspectral imaging (HSI), combined with advanced machine learning algorithms (MLAs), has unlocked novel research opportunities and revolutionized geological mapping by enabling precise lithological classification. Accurately detailed geological mapping is one of the most essential requirements for targeting mineralization. However, achieving comprehensive lithological mapping remains a challenge, hindering systematic mineral exploration. This work explores the use of PRISMA hyperspectral data and the Support Vector Machine (SVM) and Artificial Neural Network (ANN) algorithms to objectively map the Precambrian rock assemblages at the El Ineigi area in the Central Eastern Desert (CED) of Egypt. For this purpose, PRISMA data in HDF5 format were first pre-processed and subsequently transformed through principal component analysis (PCA). The processed spectral data were then combined with extensive fieldwork and previously existing geological maps and classified using SVM and ANN to achieve enhanced discrimination of the exposed rock units in the study area. Our results conclusively demonstrate the exceptional capability of PRISMA data for detailed lithological mapping. The SVM and ANN classification achieved remarkably high overall accuracy, successfully generating a robust geological map that clearly discriminates between various Neoproterozoic basement rock units in the El Ineigi area. Through the integration of diagnostic spectral signatures with field verification, we confidently identified all major mappable units, including metavolcanics, metagabbro-diorite complexes, younger granites, and Wadi deposits. The proposed integrated approach demonstrates superior performance compared to traditional mapping techniques, offering enhanced discrimination precision and operational efficiency. These findings strongly support the combined use of PRISMA hyperspectral data and MLAs for lithological mapping applications.

Keywords

• PRISMA hyperspectral data
• Support vector machine (SVM)
• Artificial Neural Network (ANN)
• Lithological mapping
• El Ineigi area

Main Subjects

• Mineralogy, Petrology & Gemology