A review on hyperspectral imagery application for lithological mapping and mineral prospecting: Machine learning techniques and future prospects

Hyperspectral remotely sensed imagery is a pertinent instrument in lithological and mineral alterations mapping for a vast range of ore mineralization. This imagery typically provides an ideal characterization and exploitation of the Earth's outcrop, using wide-ranging spectral/spatial data for the reconnaissance stages during ore mineral exploration. The application of hyperspectral remote sensing datasets derived from satellite and airborne platforms has proven to be instrumental in surmounting prevalent challenges encountered in mineral exploration endeavors. Because of the exponential surge in hyperspectral remote sensing data acquisition from disparate platforms, the scientific community has been incited to develop sophisticated and resilient data processing approaches using artificial intelligence (AI) techniques. Additionally, recent studies have witnessed the integration of machine learning (ML) algorithms with conventional image processing techniques and geological surveys, featuring the upward significance of hyperspectral remote sensing lithological mapping and mineral prospecting (HLM-MP). Although there are previous reviews that broached the use of HLM-MP, there is still a lack of an updated comprehensive review on the subject. This review article unequivocally demonstrates the potential inherent in harnessing hyperspectral imaging datasets and ML algorithms, facilitating precise mapping of crucial geological features and enabling the production of significantly enhanced mineral prospectivity mapping. Furthermore, this review identifies promising prospects for the utilization of deep learning (DL), multisource data integration, and cloud computing when processing hyperspectral remote sensing data, thereby further refining HLM-MP investigations.