Supervised Classification of Landsat-8 Band Ratio Images for Geological Interpretation of Sokoto, Nigeria

Abstract

Landsat-8 OLI imagery of Sokoto, Nigeria, was processed to emphasize the geology features and mineral potential of the area. Band ratios   were assigned to RGB. Band ratio    highlights ferric ion minerals,  emphasizes ferrous minerals, and  distinguishes iron oxide minerals from carbonate minerals. In a second technique, band ratio  was replaced with  in order to accentuate clay minerals with high reflectance within band 7. The last technique evaluated in this study used spectral information from minimum noise fraction image to map surface geology. Supervised classification training sites were selected using five classes (clay, ironstone, alteration zone, water and vegetation). The band ratio classification using maximum likelihood classification was fairly accurate and matched the geologic map of the area, also showing an alteration zone that coincided with the migmatite-quartz/mica schist contact. The classified image was finally passed through a filtering effect for generalization of the data. This filtering effect was helpful in discriminating the pixels of ironstone and those of the alteration zone on the classified map.