A Bayesian Network-based risk dynamic simulation model for accidental water pollution discharge of mine tailings ponds at watershed-scale

Abstract

Mine tailings ponds that contain heavy metals are sources of potential risk to human security and ecosystem health. China particularly faces challenge of accidental water pollution risk from more than 8869 mine tailings ponds in serve by 2015, some of which are close to residential areas and other important infrastructures within 1 km downstream. To address watershed-scale risk assessment of accidental water pollution from mine tailings ponds, a Bayesian Network-based Risk Dynamic Simulation (BN-RDS) model was proposed to simulate "sources/stressors-receptors-endpoints" risk routes. An accidental water pollution convection-diffusion simulation was coupled to Bayesian Networks to perform the risk dynamic simulation and risk evolution quantification at watershed-scale. This method was applied to the risk assessment of 23 tailings dams in 12 sub-watersheds covering the Guanting Reservoir basin (the major backup drinking water source for Beijing) in Zhangjiakou City, China. The result indicated that ecosystem health and property security were the endpoints at the highest risk in the overall watershed. Spatially, the combined risk distribution map showed the risk was higher in the downstream of the Guanting Reservoir Watershed and in its two tributary basins (the Qingshui River and the Longyang River). This research highlighted a probabilistic approach to accidental water pollution risk assessment of tailings ponds with verifiable and tangible results for risk managers and stakeholders.

Keywords: Accidental water pollution; Acute lead/cadmium exposure; Bayesian networks; Mine tailings ponds; Risk assessment.