We select CB and MB superconducting gravimeter (SG) stations with longer data length (2004-2020) and better observation quality, and use the oceanic Ni1o index (ONI) as the El Ni1o-southern oscillation (ENSO) index to assess the possible impact on SG observations at the interannual timescale (primarily 1-5 a). First, we remove the influences of instrument response, atmospheric tides, tidal effects, hydrological factors, and polar motion from the CB and MB observation series. Then, we apply the Morlet wavelet transform and cross-wavelet analysis to the data. The results show there is a significant correlation between SG observations and ONI, primarily at periods of approximately 1.5 a and 3 a. Specifically, gravity variations at CB station exhibit a negative correlation with ONI, while at MB station, the correlation is positive, with a time delay in the gravity observation relative to ONI. Furthermore, the ENSO effect on CB station is more pronounced than that on MB station, especially at about 3 a period, where ENSO shows a negative correlation with gravity variations at CB station. We hypothesize that this discrepancy is largely due to the different geographical locations of stations. However, more detailed quantitative studies are needed to further explore this relationship.

This paper uses historical landslide and hydrogeological data from Anhui province as the original dataset, employing five typical machine learning models and selects nine environmental disaster-inducing factors as initial input variables. To explore the impact of different machine learning input variable selection combinations on the performance of various models, multiple factor selection methods and several evaluation metrics are used to examine the prediction accuracy of each algorithm model under different selection combinations for landslide susceptibility. The evaluation results show that the combination of the full selection set and the lightweight gradient boosting machine (LightGBM). model yields the best performance. Therefore, this combination is chosen for the landslide susceptibility mapping in the study area. Moreover, the study suggests that factor selection for machine learning inputs does not necessarily improve model performance in landslide susceptibility studies of the research area. Additionally, ensemble models outperform individual estimator models, and LightGBM, using a leaf-node-based decision tree algorithm, shows improved performance compared to other ensemble algorithms.

Based on the seismic monitoring data from the Ganzhou Earthquake Monitoring Center Station of the Jiangxi Seismological Agency, real-time monitoring of blasting vibrations from different locations and explosive quantities at the Ganzhou Tieshanlong Tungsten Mine was conducted from multiple directions and long distances. The microseismic signals generated by blasting were obtained and analyzed to identify the propagation patterns of blasting vibration signals in medium and far blasting zones. Considering the influence of complex environmental conditions on blasting vibration signals, a Butterworth filter was used to eliminate trend components, and a combination of wavelet denoising and HHT denoising was applied to analyze the time-frequency characteristics, frequency band energy characteristics, and the impact of terrestrial environment on the spectral energy propagation of microseismic signals under different monitoring base stations and different blasting events. The study shows that the Sadovsky empirical formula is still applicable to the propagation patterns of blasting vibration signals in the medium and far blasting zones (32-120 km), where: 1) in the medium and far blasting zones, signal energy is highly concentrated in the low-frequency band and shifts downward with increasing distance from the blast center (R); 2) when the distance from the blast center is fixed, signal energy increases significantly with the increase in explosive quantity (Q), and the concentrated frequency band fluctuates within a normal range; 3) the attenuation of blasting vibration signals is also affected by the surface environment along the propagation path.

A post-earthquake affected population prediction model based on support vector machines(SVM) optimized by particle swarm optimization(PSO) is established, and the SST(safety stock theory) earthquake emergency supply demand prediction model is constructed. Nine indicator parameters, including seismic hazard and damage severity, are selected and processed through dimensionality reduction and redundancy removal as input variables for the PSO-optimized SVM model to predict the affected population. Based on the relationship between the affected population and emergency supplies in disaster areas, the SST model is applied to indirectly estimate the quantities of typical supplies required in the immediate aftermath of the Jiuzhaigou earthquake. The experimental results are as follows: By employing an error comparison analysis method to validate the model's effectiveness, the PSO-SVM model demonstrates a 14.27% reduction in prediction error compared to the SVM model, with a significant improvement in prediction accuracy. The estimated demand for typical supplies in the aftermath of the Jiuzhaigou earthquake provides a certain degree of reference, indicating that the PSO-SVM-SST prediction model possesses both theoretical and practical rationality and utility.

This paper reviews and summarizes traditional estimation methods, highlighting their reliance solelyon observed seismic activity data. If the return period of the maximum earthquake exceeds the data coverage period or seismic activity rates fluctuate over longer timescales, these methods become ineffective. Furthermore, in the absence of physical constraints, traditional approaches may predict unbounded growth in the maximum event magnitude as the temporal scope expands.To address these limitations, this study introduces a new model based on seismic moment conservation. This model integrates instrumental records, fault slip rates, fault geometry, and seismic activity patterns, assuming that seismicity follows the Gutenberg-Richter(G-R) law. By simulating missing large earthquakes and their aftershocks, it constructs a long-term seismic catalog and imposes physical constraints on the maximum magnitude using the seismic moment accumulation rate, effectively mitigating the reliance of traditional methods on observational periods.The model is applied to predict the M_max and return period of the Aerjin fault zone, the Kunlun fault zone, and the Longmenshan fault zone. The results show high consistency with historical earthquake records, paleoearthquake data, and observed slip rates, validating the model's efficacy. This research offers a new perspective on maximum magnitude estimation and provides a valuable tool for seismic hazard assessment in regions with incomplete historical records.

Based on the statistical analysis, correlation fitting and preprocessing of high-precision continuous observation data of four trace hydrogen observation points in fault zones of northwestern Yunnan over the past five years, we evaluate the seismic prediction efficiency of observational data, and analyze the characteristics and causes of trace hydrogen concentration anomalies related to earthquakes combining with previous studies. The results show that: 1) The concentration variations of hydrogen escaping from hot springs in northwestern Yunnan are correlated with temperature and air pressure, showing distinct annual variation features. In contrast, hydrogen escaping from soil is essentially uncorrelated with meteorological elements. Comparison with previous studies, there is a significant spatial heterogeneity in the distribution of trace hydrogen concentration in fault zones, which is not controlled by a single factor. 2) All of four trace hydrogen observation points in northwestern Yunnan are effective seismic prediction indicators with statistical significance. The short-term prediction effect of hydrogen escaping from soil is better but with relatively lower credibility. The prediction efficiency of hydrogen escaping from hot springs is mainly in medium to long term, with relatively higher credibility. 3) There is a close correlation between hydrogen anomalies in fault zones and seismic activities in northwestern Yunnan. Seismic precursory anomalies typically manifest as sudden increases or sawtooth fluctuations. Due to differences in hydrogen sources, escaping processes, and geothermal storage temperatures in different regions, the geochemical background values of hydrogen escaping from hot springs and soil also exhibit significant spatial heterogeneity.

Based on the observational data of two types of borehole strainmeters at Xuzhou seismic station, we use the R-value test method to calculate and score the forecasting efficiency of trend-breakinganomalies, annual breaking anomalies and rate anomalies recorded by the two instruments. A systematic comparative analysis of the forecasting efficiency of the two borehole strain instruments at the same site is conducted to clarify their complementarity in earthquake precursor monitoring. The results show that the borehole body strainmeter performs well in the forecasting of trend-breaking anomalies and annual breaking anomalies, and is suitable for capturing the earthquake precursor anomaly signals related to long-term geological tectonic activities. The borehole component strainmeter shows a strong ability in the forecasting of trend-breaking anomalies and rate anomalies, and is more adept at capturing the trend-breaking precursor anomaly signals caused by stress state changes of geological body, and has high sensitivity for the monitoring of rapid changes in regional stress and strain. These findings reveal the efficiency differences of borehole strainmeters in earthquake precursor monitoring and provide the application focus and optimization direction of the two types of borehole strainmeters in earthquake forecasting.

Based on ground-based single-antenna GNSS receivers, the global navigation satellite system-interferometric reflectometry(GNSS-IR) technique can utilize the signal-to-noise ratio(SNR) to invert the vertical distance between the reflective surface and the antenna, thereby estimating water level heights. To reduce equipment costs and enhance the accessibility of GNSS-IR technology, while fully leveraging existing deformation monitoring GNSS stations, this study proposes the use of universal BDS/GNSS receivers for monitoring the structural deformation of large bridges. By integrating quality control, error correction models, and result fusion methods, effective monitoring of water level changes beneath bridges is achieved. Processing data from two universal BDS/GNSS receivers installed on the main cable of the Guojiatuo Changjiang bridge in Chongqing, which are used for monitoring bridge structural deformation, the study results indicate that the GNSS-IR water level time series, after quality control, error correction, and fusion processing, are highly consistent with the data trends from traditional hydrological stations, with a correlation coefficient reaching 0.99. This outcome not only validates the feasibility of this method in monitoring water level changes beneath large bridges but also demonstrates the broad application prospects of universal BDS/GNSS receivers in river water level monitoring, providing robust data support for flood early warning and water resource assessment.

We quantify surface water storage anomalies using GRACE satellite Mascon data and hydrological model data combining with water level and area data, estimate groundwater storage anomalies in Dongting Lake basin based on water balance equation, analyze the spatiotemporal variation characteristics and influencing factors of groundwater storage anomalies. The results show that after incorporating changes in surface water storage, the correlation between the inverted groundwater storage anomalies at the lake scale and the observed groundwater data improved from 0.40 to 0.65, reaching 0.75 in areas with dense groundwater wells. The overall groundwater storage anomalies in Dongting Lake basin show a downward trend, with a loss rate of-5.05 mm/a. Spatially, there is a significant difference between the northern and southern areas. Precipitation influences groundwater storage anomalies through its impact on surface water, with some lag due to hydrological processes, while human activities have a limited effect. This study reduces uncertainties generated in separating surface water storage, providing data support for groundwater dynamic research and water resource management in Dongting Lake basin.

In order to improve the accuracy of ZTD model, a fine-scale stratification ZTD (ZTDFS) model in China is proposed. The new model is finely stratified below 10 km and takes into account the seasonal and daily variation characteristics of each layer. At the same time, the model also fully considers the spatiotemporal variation of the vertical lapse rate of third-order exponential function. The radiosonde data shows that the mean RMSE of ZTDFS model is 31.43 mm, which is 36.6% smaller than that of GPT3 model. The GNSS ZTD data shows that above 500 m, the mean RMSE of ZTDFS model is 28.64 mm, which is improved by 9% compared to GPT3. With the increase in altitude, the performance advantage of ZTDFS model gradually emerges.