Socio-environmental land cover time-series analysis of mining landscapes using Google Earth Engine and web-based mapping

Abstract

Mining contributes to land cover changes, directly and indirectly affecting the natural landscape and local communities. The need for robust analyses of social and environmental data at the mine site and mine region scale is essential for regulators and mining companies to effectively identify, monitor, sustainably mitigate and manage mining impacts. This study characterized and evaluated land cover changes and their concurrent impacts on socio-environmental land uses in a Philippines mining landscape. Using composites of multispectral Landsat images, vegetation indices and a Digital Elevation Model (DEM), classified land use and land cover time-series maps were created. High-level land covers with a coarse thematic resolution that could be successfully characterized using Landsat historical imagery were first mapped using supervised Random Forest classification in Google Earth Engine (GEE). Subsequently, web-based mapping by local experts was used to characterise key fine thematic resolution land use categories within selected zones of importance which were not possible to characterise using Landsat alone. The time-series provided an accurate estimate of change, and revealed significant temporal trends at the regional scale between 1994 and 2018. Trends included a significant decrease in primary vegetation and an increase in built-up areas, mining and irrigated agriculture. Some notable fine-scale land use changes revealed by the analysis were the increase in social development projects post 2005 and conversions between citrus and paddy agriculture; these were initially balanced but leaned prominently towards paddy cultivation post 2010. An assessment of land use and land cover transitions provided key insights to several socio-environmental indicators, including environmental quality, habitat loss, population distribution and livelihood, essential to characterise and support the management of the mine's socio-ecological system. The paper concludes by reflecting on the methods developed, evaluating their limitations and presenting potential ways to improve the workflow to better support social change assessments.