Historical socio-environmental assessment of resource development footprints using remote sensing

Abstract

Characterising the spatial and temporal dimensions of resource development projects is critical for understanding their present and future risk and opportunity profiles. For mining projects, pre-existing land uses, such as agriculture or human settlements, create spatial constraints that drive trade-offs between the development and operation of a mine, and other alternative land uses. Spatial analysis using historical remote sensing data can be used to “go back in time” to assess changes in mining land cover and better understand the physical determinants underlying present day complexities. This article describes a remote sensing workflow for characterising socio-environmental land cover change in mining projects. The workflow is applied in two mine sites in Papua New Guinea and the Lao People's Democratic Republic. An idealised hierarchical land cover classification scheme from which a selection of land covers were derived using Landsat is first described. Using a combination of supervised classification with manual input, and automated classification using CLASlite, 8 scenes are mapped at each mine site. Overall accuracies were greater than 90% for all years at both mine sites. The mapped spatial patterns over time demonstrate the dynamic nature of both mining landscapes with rapidly increasing mine waste footprints (i.e. area on a surface). As undisturbed land becomes scarce, the data shows mining and community activities are in closer proximity to each other. The results represent one of the most detailed cross-disciplinary remote sensing studies in mining in terms of high thematic, spatial and temporal resolution conducted to date. The data demonstrates how methods and perspectives from various social and environmental disciplines can combine to provide a more holistic perspective of a mining footprint.