A comparison of optical remote sensing data for automatic fracture network mapping in Lokoja region, Central Nigeria

Abstract

This study evaluates the efficiency of some open source optical remote sensing data (Landsat OLI, Sentinel-2 A, ASTER and DEM) in automatic lineaments extraction over Lokoja region of Central Nigeria. Various image processing techniques involving principal component analysis (PCA), directional filtering and shaded relief were employed followed by a robust lineament extraction technique and novel false lineaments filtration method. The result indicated significant variation in the number, length and accuracy of extracted lineaments across the datasets. Comparison of results by way of accuracy assessment was achieved after superimposition of extracted lineaments on geological map and shading map of the study area as well as by comparing their orientations with field obtained fracture data. Landsat OLI and Sentinel-2 A demonstrated better performance by extracting longer and more numerous lineaments many of which correlated with lithological boundaries and boundaries between shaded and unshaded areas. Orientations of the extracted lineaments indicated a majorly ENE-WSW and NE-SW trend while the field obtained fracture orientations showed a majorly NNW-SSE and NE-SW directions. All the images failed to identify lineaments along the NW-SE directions. The findings of this study underscored the importance of selecting appropriate datasets for regional geological investigations.