Self-organizing maps for the identification of groundwater salinity sources based on hydrochemical data

摘要

Growing inorganic and expansive hydrochemical datasets and large differences in the measured concentrations require methods that are capable of compressing data without the loss of critical information and subsequently displaying it in a condensed and comprehensive way. Here we train an artificial neural network, Kohonen's self-organizing map (SOM), to model inorganic hydrochemical clusters and associate the salinity source with the distribution of the ionic concentration spatial variation at a former potash mining site. Kohonen's self-organizing maps are applied to project the data onto a two-dimensional grid and the geometric relationship of the projected vectors is subsequently used to perform a hierarchical cluster analysis. The SOM clustering approach succeeded in assigning the groundwater samples automatically according to their inorganic chemical composition. Five different clusters, three geogenic and two anthropogenic, were identified. The final outcome is displayed and compared with the classification from Piper plotting of the same dataset. In order to see the SOM clustering results in the large scale hydrogeological context, the distribution of the clusters is displayed spatially. This approach is a tool for the hydrogeologist to quickly analyze large datasets and present them in a clear and concise format.