Studying the hydrogeochemical characteristics of hot springs provides essential geochemical information for monitoring earthquake precursors and understanding the relationship between fluids, fractures, and earthquakes. This paper investigates the hydrogeochemical characteristics of hot springs along the Tingri–Nyima Rift (TNR) in southern Tibet, a seismically active zone at the collision front of the Indian and Asian-European plates. The major elements, hydrogen, and oxygen isotopes of seven thermal springs were analyzed from July 2019 to September 2021. The findings indicate that Mount Everest’s meteoric water, which has a recharge elevation of roughly 7.5–8.4 km, is the main source of recharge for the hot springs. The water samples have two main hydrochemical types: HCO₃-Na and Cl-Na. The temperature of the geothermal reservoir is between 46.5 and 225.4 °C, while the circulation depth is between 1.2 and 5.0 km based on silica-enthalpy mixing models and traditional geothermometers. Furthermore, continuous measurements of major anions and cations at the Yundong Spring (T06) near Mount Everest reveal short-term (8 days) seismic precursor anomalies of hydrochemical compositions before an M_L4.7 earthquake 64.36 km away from T06. Our study suggests that seismicity in the northern section of the TNR is controlled by both hydrothermal activity and tectonic activity, while seismicity in the southern section is mainly influenced by tectonic activity. In addition to magnitude and distance from the epicenter, geological forces from deep, large fissures also affect how hot springs react to seismic occurrences. A fluid circulation model is established in order to explain the process of groundwater circulation migration. The continuous hydrochemical monitoring of hot springs near Everest is critical for studying the coupling between hot springs, fractures, and earthquakes, as well as monitoring information on earthquake precursory anomalies near Everest.

Deep groundwater dynamics, influenced by multiple factors, exhibit multi-scale periodicity, nonlinearity, and non-stationarity, posing significant challenges for reliable groundwater level prediction and anomaly detection. To address this issue, this study develops a hybrid deep learning model for simulating and predicting normal groundwater dynamics, which integrates multi-source observations—including barometric pressure, solid tide, and historical groundwater levels—with convolutional neural networks (CNNs), Long Short-Term Memory (LSTM), and Attention mechanisms. An Exponentially Weighted Moving Average (EWMA) control chart is then applied to the prediction residuals for statistical anomaly detection. The results show that CNN kernels capture multi-period local features induced by barometric and tidal effects; LSTM gating mechanisms learn long-term temporal dependencies; and the Attention mechanism dynamically assigns weights to both temporal steps and input features. Compared with single-source inputs or standalone LSTM/CNN models, the hybrid model enhances prediction accuracy (MAE < 0.02, RMSE < 0.03, R² > 0.99), while integration with the EWMA control chart helps to identify the possible anomalies. Collectively, these results provide a reliable means for predicting deep groundwater levels and detecting hydrological anomalies under complex multi-scale periodic forcings, holding significant practical value in groundwater resource management and disaster warning.