Hydrochemistry-Based Statistical Model for Sourcing Groundwater Inrush into Underground Mining Works: A Case Study in Eastern Serbia

Discriminant analysis was applied to a baseline groundwater quality data set to develop a model for sourcing groundwater inrush into underground mining works. Such inrush events compromise the safety of the working environment and the feasibility of mining; thus, a comprehensive characterization of the hydrogeological system is required. In this study, a statistical model for groundwater sourcing was developed and validated based on a dataset of 179 groundwater chemical analyses related to the "Cukaru Peki" Cu-Au underground mine in eastern Serbia. The studied groundwaters are primarily of low TDS (235-715 mg/L), with a neutral to slightly alkaline pH, while the mine waters are alkaline (pH > 9). The evolution of groundwater chemical composition with depth was observed; Ca-HCO3 water prevails in the shallow zones, transforming to Na-SO4 type at depth. Two discriminant functions were delineated by statistical analysis, explaining 65.3% and 31.2% between-group variability, with SO42-, Na, Al, pH, Sr, Li, and SiO2 identified as grouping factors. The resulting discriminant functions were applied to 11 mine water samples to examine their relationship with the main hydrogeological units. Among the examined groundwater inflow samples, three were classified as having a different origin than the lithological unit at the mine from which the water was sampled, bringing valuable information for planning mitigation measures. The demonstrated approach highlights the importance of comprehensive baseline hydrogeological investigations at mine sites to obtain relevant data for controlling groundwater inrushes during development of the declines and shafts.