Surface coal mine area monitoring using multi-temporal high-resolution satellite imagery

Surface mining activities, exploitation of ore and stripping and dumping overburden, cause changes on the land cover and land use of the mine area. Sustainable mining requires continuous monitoring of these changes to identify the long-term impacts of mining on environment and land cover to provide essential safety measures. In this sense, digital image classification provides a powerful tool to obtain a rigorous data and hence diminishes the essence of time-consuming and costly field measurements. There are various image classification techniques, serving different features for different purposes, and the Support Vector Machine (SVM) classification method based on statistical machine learning theory stands out to be an effective and accurate image classification technique among them.This research study investigates the use of SVM classification methods for identifying, quantifying, and analyzing the spatial response of landscape due to surface mining activities in Goynuk open cast mine, Turkey, from year 2004 to 2008. The research algorithm essentially entails (i) acquiring data, (ii) pre-processing the data, (iii) performing image classification, (iv) accuracy assessment and change detection, and (v) analysis of results. The results showed that SVM classification method can effectively be utilized for high spatial resolution multispectral satellite images for identifying the changes in surface coal mine area.

Research Highlights
► We investigated SVM classification use for change detection due to surface mining.
► Classification accuracy for Ikonos image was higher than that for Quickbird image.
► We have found out that mine and dump area decreased by 192.5 ha from 2004 to 2008.
► Forest area increased by 6.73% from 854 ha to 911 ha from 2004 to 2008.
► SVM classification agorithm was effectively utilized for the classification.