Presented paper

IMWA2019 Students work

Developing A Risk Assessment For Local Groundwater Degradation By Mine Waste Storage Facilities

Zielke-Olivier, Josepha SDR; Vermeulen, P Danie
University of the Free State, South Africa

A linear and nonlinear statistic approach was chosen to develop a risk assessment predicting the groundwater pollution potential and SO4 concentrations considering 12 environmental and spatial variables at an industrial and coal mining complex. Linear regression models and nonlinear classification and regression trees indicated that the explanatory variables lnWLDepth and vadose zone were most significant in predicting the potential pollution risk and SO4 values. Furthermore, the simple linear and robust regression model showed that the study area around a particular discard dump had generally higher SO4 concentration than predicted by the control area with a factor of 95.5 and 88.3%, respectively. This could be caused by additional environmental factors not being considered for the study site such as a backfilled area, wetlands and groundwater flow paths possibly related to underground mining activities. On the other hand, tree models were able to identify additional relationships between SO4 concentrationsand environmental factors including distance to pollution, fault and stream as these models recognize nonlinear relationships. Tree models were found to be useful visual tools to develop a site-specific risk assessment and predict SO4 concentrations in the local groundwater.