Hydrogeochemical characterization and suitability assessment of groundwater in a typical coal mining subsidence area in China using self-organizing feature map

In the Taiping coal mining area in Zoucheng City, where there is a focus on agricultural production, a unique collapsed pond has been formed due to the dense population, a high phreatic water level and coal mining subsidence. A one-year field study was undertaken to investigate the concentrations of cations and anions in the pore groundwater, as well as the collapsed water and surface water, to clarify the hydrogeochemical controls, the seasonal cycle characteristics and the intended uses for the groundwater. The results, obtained from a self-organizing feature map, the K-means clustering algorithm and the Durov diagrams, revealed that the hydrochemical dataset could be classified into five clusters, corresponding to a SO4-Na type (Clusters 1 and 2), a mixed type (Cluster 3), a HCO3-Ca type (Cluster 4) and a SO4-Na∙Ca type (Cluster 5), respectively, with clear seasonal changes in the five pore groundwater samples. Based on the Gibbs, Gaillardet and chloro-alkaline index (CAI) diagrams, rock weathering, cation exchange and evaporative crystallization, especially the erosion of silicate rock, were the primary processes controlling the hydrogeochemistry. Meanwhile, the suitability of the groundwater evaluation methods of random forest (RF), genetic algorithm-support vector machine (GA-SVM) and back-propagation (BP) neural network were found to be superior to the traditional Quality Standard for Groundwater of China (SGQC), the Fisher and the F analysis methods. Among them RF has the optimal simulation accuracy and effect. As a result of quality assessment of the groundwater, the quality of the shallow groundwater was generally poor and was only fit for purpose after appropriate treatment. Moreover, it is speculated that the main factors affecting the groundwater quality were the unique mode of collapse of the pond formed as a result of the high phreatic water level, the natural conditions such as rainwater recharge and groundwater runoff, the dense population, mining and agricultural development, and chemical pollution. This innovative study describes an optimization method for assessment of groundwater suitability and highlights the importance of minimizing excessive groundwater extraction, developing continuous water quality monitoring plans, and managing and preventing potential hazards.