Unsupervised Detection of Active, New, and Closed Coal Mines with Reclamation Activity from Landsat 8 OLI/TIRS Images

Abstract

Monitoring and classification of surface coal mine regions have several research aspects, as they have huge impacts on the eco-environment of a region. Mining regions change over time immensely. New potential mines get opened. Whereas, few mines stay active and many of them get closed. Closed mine regions are reclaimed to environment through plantation. In the past, semi-supervised and supervised techniques have been used to detect mine classes and assess the changes of land use and land cover classes. In this work, mine regions are detected in an adaptive manner from satellite images unlike the techniques in the literature. Further, a change detection technique is used to detect active, new, and closed surface coal mines in a region. Detected closed mines are further analysed to evaluate reclamation of that region. Average precision and recall of active, new, and closed mining regions of the proposed technique are found to be [$$84.7\%,62.8\%$$], [$$74.2\%,64.5\%$$], and [$$70.1\%,58.2\%$$], respectively.