Search

The satellite data obtained from various airborne as well as space-borne Hyperspectral sensors, often termed as imaging spectrometers, have great potential to map the mineral abundant regions. Narrow contiguous bands with high spectral resolution of imaging spectrometers provide continuous reflectance spectra for different Earth surface materials. Detailed analysis of resultant reflectance spectra, derived through processing of hyperspectral data, helps in identification of minerals on the basis of their reflectance characteristics. EO-1 Hyperion sensor contains 196 unique channels out of 242 bands (L1R product) covering 0.4-2.5 μm range has also been proved significant in the field of spaceborne mineral potential mapping. Present study involves the processing of EO-1 Hyperion image to extract the mineral end members for a part of a gold prospect region. Mineral map has been generated using spectral angle mapper (SAM) method of image classification while spectral matching has been done using spectral analyst tool in ENVI. Resultant end members found in this study belong to the group of minerals constituting the rocks serving as host for the gold mineralisation in the study area. [ABSTRACT FROM AUTHOR]

The article focuses on utilizing machine learning (ML) techniques to classify minerals based on spectral signatures from satellite-based hyperspectral sensors for understanding the geology of a region. Topics include the application of machine learning to reduce human efforts in remote sensing; the classification of lithological units using ML algorithms; and the accuracy assessment of ML algorithms in distinguishing various lithounits in the Nidar ophiolite complex in Ladakh, India.