A spatial machine learning approach to exploring the impacts of coal mining and ecological restoration on regional ecosystem health

Abstract

Ecosystem health is an important approach to measuring urban and regional sustainability. In previous studies, the spatiotemporal changes of ecosystem health have been addressed using comprehensive assessment index system. However, the quantitative contribution of human activities and climate change to ecosystem health was less examined. In this study, Shuozhou City, a coal resource-based city, was chosen to explore the response of ecosystem health to human activities using the Geographically Weighted Artificial Neural Network (GWANN) model. The results showed a distinct improvement of ecosystem health in Shuozhou City from 1990 to 2020. The contribution of human activities increased during the study period, while the contribution of climate change decreased as a consequence of coal mining expanding. By 2020, human activities contributed 76% to ecosystem health compared with 24% of climate change. The direct impact of coal mining on ecosystem health occurred mainly in the surrounding areas within a radius of 6 km and 17 km under low and high mining intensity respectively. Ecosystem health will further decline by 2030 based on the scenario in which current coal mining is continued. However, only stopping mining activities in small coal mining areas for ecological restoration but keeping large coal mining areas in production, will realize 92.6% of restoration effects on ecosystem health as compared to ceasing all mining activities. This study examines the effects of coal mining on ecosystem health in resource-based cities, and underscores the importance of large coal mining sites in ecological restoration.