This dataset presents a comprehensive approximately two-year continuous monitoring effort of subsurface mine water dynamics in the Reiche Zeche mine, a decommissioned silver mining site in the Ore Mountains of Central Europe. This dataset integrates high-resolution geochemical sampling, hydrological in situ sensing, and UV-Vis spectrometry complemented by machine learning modeling to characterize hotspots and hot moments of metal mobilization across stratified and mixed hydrological regimes. Within 42 sampling campaigns (February 2022 - May 2024), water samples were collected from 26 water flowing and dripping sites, with detailed hydrological and automated spectroscopic monitoring at four distinct locations (sites 1, 2, 3A, and 3B). Dissolved metal(loid) and organic carbon concentrations, and stable water isotopes (δ2H and δ18O) were recorded alongside hourly flow rates and optical absorbance spectra. A cubist-based modeling framework was applied to trace metal concentrations from spectral data, with models validated using autosampler measurements. This curated dataset supports the study of contaminant transport under variable flow conditions in legacy mining environments and is intended to facilitate data reuse for hydrogeochemical modeling, machine learning applications, and environmental monitoring research.