In this study, radiometric datasets which are best interpreted in conjunction with multispectral remote sensing data were applied in discriminating and classifying the lithologies of the Kebbi–Sokoto basement terrain for effective lithological mapping of the region. K/Th/U radiometric and Landsat-8 remote sensing data were enhanced through several processing techniques to display the characteristics of surface responses needed as input parameters in the classification of the surface rocks. Potassium radioelement concentration, Th/K channel ratio, normalization and ternary maps were produced using radiometric data while band combination and band ratio images were obtained from the Landsat-8 data. Several band ratios highlighted the major rock–forming minerals in the area such as feldspar, quartz, hematite, goethite, illite, kaolinite, biotite, amphibolite and dolomite. The major lithological expressions deduced from the combined data analysis namely; sandstone, felsic gneiss, clay, migmatite, granitoid and carbonates were used as representative classes to train the classification algorithm for the lithological interpretation. Supervised classification algorithm employing maximum likelihood classifier was used to obtain the lithological classification for basement rock units in the region. The classification results which were tested on band ratio b6/b2, b6/b7, b6/b5*b4/b5 image, b4/b2, b5/b6, b6/b7 image and b4/b2, b6/b7, b6/b5 image using a contingency table produced overall classification accuracy values of 79%, 92%, and 88% respectively. The classified images displayed quite similar geological feature discrimination with the radiometric ternary map and geological map of the area. The study emphasized the benefit of using geophysical and multispectral datasets to produce a more complete classification of the surficial Precambrian rocks exposed in the region.
