Groundwater extracted from alluvial aquifers close to rivers is vulnerable tocontamination by infiltrating river water. Infiltration is often increased during highdischarge events, when the levels of waterborne pathogens are also increased. Watersuppliers with low-level treatment thus relyon alternative measures derived from information on system state to manage the resource and maintain drinking-water quality. Inthis study, a combination of Self-Organizing Maps and Sammon’s Mapping (SOM-SM)was used as a proxy analysis of a multivariate time-series to detect critical system stateswhereby contamination of the drinking water extraction wells is imminent. Groundwaterhead, temperature and electrical conductivity time-series from groundwater observationwells were analysed using the SOM-SM method. Independent measurements (spectralabsorption coefficient, turbidity, particle density and river stage) were used. This approach can identify critical system states and can be integrated into an adaptive, online,automated groundwater-management process.