Ensuring water security and enabling timely responses to contamination events in water distribution systems (WDSs) rely heavily on the accurate and timely localization of contamination sources. Despite advances in water quality monitoring technologies, such as continuous sensing and grab-sampling, the coverage of monitoring remains sparse in most WDSs, making it difficult to accurately pinpoint the source of contamination. This paper introduces a novel source localization methodology designed to overcome these challenges by integrating sparse continuous sensing with targeted manual grab-sampling. The proposed approach iteratively narrows down the set of probable contamination sources by applying heuristics that account for the timing and signals from sensor measurements. To further address the uncertainty inherent in source localization, the methodology generates a probabilistic distribution over potential source locations. This distribution highlights areas requiring closer attention and guides where subsequent samples should be collected, effectively reducing uncertainty in the localization process. The methodology's performance is validated through extensive analysis, demonstrating that combining fixed sensors with adaptive sampling significantly improves precision, accuracy, and localization speed, particularly in sparse sensor networks. The proposed approach advances the use of water quality sensing technology for source localization, with further research needed to optimize its effectiveness in improving WDS security and maximizing public health protection.