Hongyan Dong

Establishing the groundwater geochemical background (GGBC) is fundamental for distinguishing seismic precursors from non-tectonic environmental influences. This study characterizes the spatiotemporal evolution of the GGBC in Beijing using 60 groundwater samples collected from 17 seismic monitoring wells between 2018 and 2025. Major ion concentrations, stable water isotopes (δ²H and δ¹⁸O), Self-Organizing Map (SOM) clustering, Principal Component Analysis (PCA), ionic ratios, and mineral saturation indices were integrated to analyze the hydrochemical framework for establishing the GGBC baseline. Piper diagram results reveal three main hydrochemical types: Na·HCO₃ type, Ca·Mg·HCO₃ type, and sulfate type (subdivided into Na·SO₄ and Ca·Mg·SO₄/HCO₃ subtypes). Hydrogen and oxygen isotope data from 2024 indicate that recharge sources range from deep bedrock fissure water to shallow mixtures of meteoric and surface water. Gibbs diagrams, SOM, and PCA consistently demonstrate that water–rock interaction (including carbonate, silicate, and evaporite dissolution, as well as cation exchange) dominates the GGBC. Ionic ratios and saturation indices further confirm that silicate weathering and forward cation exchange are persistently widespread, while carbonate and evaporite dissolution are locally important. Notably, 12 wells of the 17 monitoring wells (71%) retained the same hydrochemical type across all campaigns in which they were sampled. This stability shows no detectable temporal correspondence with either the V-shaped pattern of seismic activity (strong–weak–strong, 2018–2025) or the regional groundwater level rise caused by water diversion projects at the available annual-to-multi-year sampling resolution. Instead, the high stability of the GGBC reflects long-term water–rock equilibrium that is resilient to minor, transient perturbations. This study demonstrates that major ions are reliable indicators of the GGBC for assessing the sensitivity of the Beijing Plain to short-term seismic or hydrological disturbances. However, the limited temporal resolution of the sampling and incomplete datasets preclude definitive conclusions regarding earthquake-related geochemical signals. To strengthen regional earthquake monitoring and forecasting capabilities, we recommend establishing a continuous, long-term, high-frequency, multi-parameter monitoring network—including stable isotopes, trace elements, and dissolved and exsolved/free gases—with high data completeness in the Beijing area.

In comparison with conventional hydrological parameters such as water levels and temperatures, geochemical changes induced by earthquakes have become increasingly important. It should be noted that hydrogen (δ²H) and oxygen isotopes (δ¹⁸O) offer the greatest potential as precursor proxies of earthquakes. Here, we conducted high-resolution sampling (weekly, 59 samples), measuring consecutive δ²H and δ¹⁸O levels at the two sites of the WLY well and SS spring in the Yan-Huai Basin of Beijing from June 2021 to June 2022. During the period of this sampling, several small earthquakes of ML > 1.6 occurred in Beijing. We used statistical methods (analysis of variance) to test the significant differences, used Self-Organizing Maps (SOMs) for data clustering, and then used Bayesian Mixing Models (MixSIAR) to calculate the proportions of the source contributions. We found significant four-stage patterns of change processes in δ²H and δ¹⁸O at both sites. The WLY well exhibited a distinct four-stage variation pattern: initial stable development (WT1) followed by a rapid rise (WT2) and sudden fall (WT3) before the small earthquakes, and finally gradual stabilization after earthquakes (WT4). In contrast, the SS spring displayed an inverse pattern, beginning with stable development (ST1), then undergoing a rapid falling (ST2) and sudden rising (ST3) before the small earthquakes, and finally stabilizing through stepwise reduction after the earthquakes (ST4). The most likely mechanisms were differences in the time of rupture between the carbonate in WLY and granite in SS under sustained stress. The stress induced source mixing of fluid from the surface or deeper groundwater-source reservoirs. The hypothesis was supported by the MixSIAR model, calculating the variational proportion of source contributions in the four stages. This work permitted the use of high-resolution isotopic data for statistical confirmation of concomitant shifts during the earthquakes, provided the mechanisms behind them, and highlighted the potential for the consecutive monitoring of hydrogen and oxygen isotopes indicators in earthquake-prediction studies.