Résumé
Karst aquifers play an essential role in drinking water supply, but their vulnerability to anthropogenic pollution, particularly during rainfall events, requires adapted monitoring methods. In this study, we developed an approach combining high-frequency in situ measurements and laboratory analyses to characterise the intensity and occurrence of organic matter fluorescence peaks, in particular Tryptophan-Like Fluorescence (TLF) peaks, associated with faecal contamination. We adapted a background noise correction method based on multiple regression applied to multispectral signals from field fluorometers. A specific calibration in TLF equivalents of the signals from an optical sensor measuring fluorescence at 285 nm is used to quantify contamination peaks. This method was tested on a dataset including 95 samples analysed in the laboratory and high-frequency data spanning 2015–2025 at 15-minute intervals from the Lez karst spring (France), showing its effectiveness in correcting optical interferences and detecting low pollution episodes. The analysis of time series underlined the importance of high-frequency monitoring to capture rapid water flow dynamics, while confirming the indispensable complementarity of laboratory analyses to better discriminate the fluorophores controlling the intensity of high-frequency in situ signals. The results highlight critical contamination periods, particularly in autumn and spring, and confirms the suitability of TLF for Early Warning System (EWS) applications. This approach broadens the methods used for water quality management in karst environments and proposes a methodological framework transferable to other sites.