Automated Spectral Deconvolution for Quantitative Dye Tracing

Abstract Dye tracing using fluorescent dyes is an essential tool for investigating flow path distribution, conduit networks, and connectivity in complex hydrogeologic systems, such as karst and fractured aquifers. The breakthrough curve, derived from the dye concentration measured at a downstream location with near‐continuous time resolution, provides quantitative information on water travel time distribution. Accurate measurement of dye concentration, however, is often challenged by the presence of natural organic materials (NOMs), including humic acids, fulvic acids, and chlorophylls, which contribute significant background fluorescence and interfere with dye signals. This study introduces a robust, automated spectral deconvolution framework aimed at precisely quantifying multiple fluorescent dye concentrations in hydrologic dye tracing. The framework employs an automated multi‐peak fitting algorithm, equipped with fluorescence peak information for both common NOMs and dye compounds, to effectively separate dye‐specific fluorescence from NOM background signals. The methodology is capable of processing large data sets generated from the high‐frequency water sampling required for breakthrough curve estimation. Three dye tracing field campaigns were conducted to compare the proposed approach with common practice of using in situ fluorometers. The results from these campaigns demonstrate significant improvements in both accuracy and efficiency when using the automated approach. A Python‐based code, along with the water tracing data from the three campaigns, is provided as open source in the public domain, inviting the community to utilize the proposed dye tracing platform for accurate and efficient dye tracing in natural water systems.