Multivariate Statistical Analysis of Hydrochemical and Microbiological Natural Tracers as a Tool for Understanding Karst Hydrodynamics (The Unica Springs, SW Slovenia)

Various multivariate statistical techniques (MST) can provide valuable insights into water quality variability. Despite numerous studies in which these methods have been used, their potential has not been fully exploited. This paper presents an improved approach to better understand the hydrodynamics of karst systems. The integrated application of hierarchical cluster and principal component analysis in combination with factor analysis allowed the construction of an advanced multivariate chemograph. The analytical procedure was applied in a binary karst aquifer known for its complex hydrodynamics and mixing of water with similar hydrochemical composition. In addition, the study area provides access to an integral groundwater flow system (ponor‐cave‐spring) and offers extensive prior hydrogeological knowledge. The approach allowed reduction and discrimination of the main parameters affecting water quality characteristics. Their identification enabled recognition of three predominant recharge components: (a) stored water impact with Cl and electrical conductivity, (b) sinking stream impact with turbidity and bacteria composition and (c) karst aquifer impact with Ca/Mg ratio as principal parameters. The results supported innovative characterization of the dominant processes and isolation of temporal hydrodynamic phases of individual monitoring points within the aquifer system. On this basis, a spatio‐temporal conceptual model was developed and the hydrodynamic behavior of the main springs was revealed. The applied methodology demonstrated to be useful in ascertaining functioning of a complex karst system under flood event conditions. Plain Language Summary Karst aquifer systems contain important water resources. The quality of karst springs can deteriorate significantly after rain events, but it is difficult to distinguish how water flows and mixes in the subsurface, especially in large and complex systems. Statistical methods are powerful tools for studying these issues, but most common approaches are inadequate in some cases to reveal the origin of the water and its fate. In this paper, we present an approach in which we combined different statistical methods to explain the dynamics of water flow based on the physicochemical and microbiological properties of water. The application of these methods led to the discrimination of parameters most useful for a reliable interpretation of statistical results, such as turbidity, bacteria, Cl, EC, and Ca/Mg, and to t