Groundwater radon concentrations can reflect the changes of crustal stress and strain. Scholars and scientific institutions have also recorded groundwater radon precursor anomalies before earthquakes. Therefore, groundwater radon monitoring is an effective means of predicting seismic activities. However, the variation of radon concentrations within groundwater is not only affected by structural factors, but also by environmental factors, such as air pressure, temperature, and rainfall. This causes difficulty in identifying the possible precursor anomalies. Therefore, the EMD-LSTM model is proposed to identify the radon anomalies. This study investigated the time series data of groundwater radon from well #32 located in Sichuan province. Three models (including the LSTM (Long Short-Term Memory) model with auxiliary data, the EMD-LSTM (Empirical Mode Decomposition Long Short-Term Memory) model with auxiliary data, and the EMD-LSTM model without auxiliary data) were developed in order to predict groundwater radon variations. The results indicated that the prediction accuracy of the EMD-LSTM model was much higher than that of the LSTM model, and the EMD-LSTM model without auxiliary data also can obtain an ideal prediction result. Furthermore, the different durations of seismic activities T (T = ±10, ±30, ±50, and ±100) were also investigated by comparing the identification results. The identification rate of the precursor anomalies was the highest when T = ±30. The EMD-LSTM model identified five possible radon anomalies among the seven selected earthquakes. Taking well #32 as an example, we provided a promising method, that was the EMD-LSTM model, to detect the groundwater radon anomalies. It also suggested that the EMD-LSTM model can be used to identify the possible precursor anomalies within future studies.

Continuous observation data from a period of 26 years show that water radon concentrations in the Panjin observation well document pre-seismic anomalies prior to earthquakes of 4.8 ≤ M ≤ 7 within a radius of 300 km. Among these earthquakes, two distinct groups with different water radon concentration anomalies and anomaly mechanisms are apparent. The abnormal characteristics of water radon concentrations clearly reflect the processes of stress change, while Cl⁻ concentration, Ca²⁺ concentration, Mg²⁺ concentration, and escaping gas flow only document part of these processes. According to Global Positioning System main strain rate fields and area strain rate fields, the change in anomalous behavior coincides with the 2011 Great Tohoku M 9.1 earthquake. This event caused the stress state of the study area, located in eastern China, to change from a relative compressive stress state to a tensile state, and may be the main reason for the change in the precursory characteristics of water radon concentrations (from increasing to decreasing prior to earthquakes). Regardless, water radon concentration in the well remains a good pre-seismic indicator for earthquakes of M ≥ 4.8. In the near future (∼50–100 years), water radon anomalies in the Panjin observation well prior to earthquakes of M ≥ 4.8 will most likely manifest as a V-shaped concentration change. Helium and neon isotopic compositions of gas samples from the Panjin observation well show that the present relatively high levels of water radon concentrations are normal and not an earthquake precursor.

Due to frequent large earthquakes in the Lanping-Simao fault basin—located in China’s Yunnan Province—the Simao observation well has observed groundwater discharge, as well as Ca²⁺, Mg²⁺, and HCO₃⁻ concentrations every day between 2001–2018. Over 18 years of observations, M ≥ 5.6 earthquakes within a radius of 380 km from the well were seen to cause hydrochemical variations. In this study, we investigated CO₂ release and groundwater mixing as possible causes of regional earthquake precursors, which were caused by the characteristics of the regional structure, lithology, water-rock reactions, and a GPS velocity field. Precursory signals due to CO₂ injection are normally short-term changes that take two months. However, groundwater mixing linked to earthquakes was found to take, at the earliest, 15 months. The proportion of shallow water that contributes to mixing was found to significantly increase gradually with the stronger regional strain. These finding delineate the two mechanisms of earthquake-induced hydrochemical variations in an observation well, and would contribute to a better understanding of chemical changes before events in the Simao basin.

Springs offer insight into the sources and mechanisms of groundwater recharge and can be used to characterize fluid migration during earthquakes. However, few reports provide sufficient annual hydrochemical and isotopic data to compare the variation characteristics and mechanisms with both atmospheric temperature and seismic effects. In this study, we used continuous δ²H, δ¹⁸O, and major ion data from four springs over 1 year to understand the groundwater origin, recharge sources, circulation characteristics, and coupling relationships with atmospheric temperature and earthquakes. We found that (1) atmospheric temperatures above and below 0 °C can cause significant changes in ion concentrations and water circulation depth, resulting in the mixing of fresh and old water in the aquifer, but it cannot cause changes in δ²H and δ¹⁸O. (2) Earthquakes of magnitude ≥ 4.8 within a 66 km epicentral distance can alter fault zone characteristics (e.g., permeability) and aggravate water–rock reactions, resulting in significant changes in δ²H, δ¹⁸O, and hydrochemical ion concentrations. (3) Hydrogen and oxygen isotopes are the most sensitive precursory seismic indicators. The results of this study offer a reference for the establishment of long-term hydrochemical and isotopic monitoring, with the potential for use in earthquake forecasting.

Groundwater level, permeability and chemical components can be affected by earthquakes, however there are few comprehensive investigations on the combination of long-term continuous monitoring data and multiple strong earthquakes. In this study, continuous two-year dataset of groundwater levels and chemical compositions of groundwater (Ca2+, Mg2+ and HCO3-) in well #32 were collected to analyze the groundwater dynamic changes induced by earthquakes in the aquifer-aquitard system. The groundwater level appeared co-seismic rise change induced by Yangbi MW 6.1 earthquake and Luding MW 6.6 earthquake. The vertical permeability, estimated by the tidal response model, exhibited decrease changes during the period of Yangbi MW 6.1 earthquake and Luding MW 6.6 earthquake. Meanwhile, the continuous two-year chemical compositions showed that Ca2+ and HCO3- concentrations decreased, and Mg2+ concentrations increased during the two earthquakes period. The correlation between the vertical permeability and chemical compositions showed that there was a significant negative correlation between the vertical permeability and Mg2+, and a significant positive between the vertical permeability and Ca2+, HCO3-. A possible mechanism for observed fluctuations in some chemical compositions during earthquakes periods was that the reduction of mixing effect of different groundwater caused by permeability decreased. The flow of groundwater richened in Ca2+ and HCO3- from the overlying aquifer to the observation aquifer has been reduced. Meanwhile, due to the weakening of dilution effect, the Mg2+ concentration of the observation aquifer increased. This study can enhance understanding of the groundwater dynamic changes induced by earthquakes.