Identification of the disturbance and trajectory types in mining areas using multitemporal remote sensing images

Abstract

Surface coal mining disturbances affect the local ecology, human populations and environmental quality. Thus, much public attention has been focused on mining issues and the need for monitoring of environmental disturbances in mining areas. An automated method for identifying mining disturbances, and for characterizing recovery of vegetative cover on disturbed areas using multitemporal Landsat imagery is described. The method analyzes normalized difference vegetation index (NDVI) data to identify sample points with multitemporal spectral characteristics (“trajectories”) that indicate the presence of environmental disturbances caused by mining. A typical disturbance template of mining areas is created by analyzing NDVI trajectories of disturbed points and used to describe NDVI multitemporal patterns before, during, and following disturbances. The multitemporal sequences of disturbed sample points are dynamically matched with the typical disturbance template to obtain information including the disturbance year, trajectory type, and the nature of vegetation recovery. The method requires manual analysis of randomly selected sample points from within the study area to calculate several thresholds; once those thresholds are determined, the method's application can be automated. We applied the method to a stack of 26 Landsat images over a 32-year period, 1984 to 2015, for mining areas of Martin County KY and Logan County WV in eastern USA. When compared with the samples determined by direct interpretation, the method identified mining disturbances with 97% accuracy, the disturbance year with 90% accuracy, and disturbance-recovery trajectory type with 90% accuracy.