Analysis of hydrochemical evolution in main discharge aquifers under mining disturbance and water source identification

Abstract

To discuss the hydrochemical evolution characteristics of the mining process of Peigou Coal Mine, based on the test results of 43 water samples collected at different times from three main discharge aquifers, namely, Carboniferous Taiyuan Formation limestone water (L7-8 + L5-6 water), Ordovician limestone water (including Taiyuan Formation L1-4), and Permian main mining coal seam roof and floor sandstone water (roof and floor water), a hydrochemical evolution model of the mining disturbances since 2003 has been established. The carbonate and sulphate dissolution and pyrite oxidation in Ordovician limestone water significantly decreased and then increased in 2006, and silicate weathering was weak. The carbonate and sulphate dissolution, silicate weathering and pyrite oxidation of roof and floor sandstone water increased. At the same time, a water source identification model suitable for the Peigou Coal Mine was developed by comparing the Fisher discriminant and the BP (back propagation) neural network discriminant. The accuracy rates of Fisher discriminant and BP neural network discriminant are 81.40% and 83.72% respectively, which indicates that BP neural network is more accurate. Finally, the evolution of hydraulic connection between aquifers is analysed. We speculate that there is a fracture development channel between Ordovician limestone water and roof and floor water aquifers that is affected in 2005 by the mining disturbance. This study has significance for examining the hydrochemical evolution of groundwater in mines and acting as a guideline to prevent and control water inrushes.