Mining is an important activity of the primary sector with strong economic and environmental impacts. All over the world, governments have made efforts to regulate mine restoration by monitoring and assessing the evolution of mined sites. Our work aims to synthesize various remote sensing applications into a single workflow in order to obtain cartographic products using Unmanned Aerial Systems (UAS), not only for mine restoration management, but also as a way of monitoring mining activity as a whole. The workflow performs image processing and terrain analysis calculations, which conduct a supervised classification of the land cover. The resulting mapping products include orthoimagery, Digital Surface Models (DSM), land cover maps, volume variation calculations, dust deposition, detection of erosion problems, and drainage network evaluation maps. The data obtained from red-green-blue (RGB) sensors has a spatial resolution of 4-10 cm, providing information that allows the characterization of land covers with an overall accuracy of 91%. In comparison, if using multispectral sensors with the same flight conditions than RGB, image spatial resolution diminishes and land cover characterization accuracy drops to 81%. The resulting digital maps can be fully integrated into Geographic Information Systems (GIS), allowing the quantification of environmental features and spatial changes. Our study provides the basis for creating a large-scale, replicable and ready-to-use workflow suited for monitoring the exploitation of minerals and mine restoration using RGB imagery obtained through drones.
Keywords: DSM; Drones; Open-pit mines restoration; Photogrammetry; Soil erosion; kNN classifier.
Copyright © 2020 The Authors. Published by Elsevier Ltd.. All rights reserved.