15 October 2026, Volume 46 Issue 10
-
Select all|
-
Journal of Geodesy and Geodynamics. 2026, 46(10): 1171-1181. https://doi.org/10.14075/j.jgg.2025.12.450
To improve the estimation accuracy of the three-dimensional velocity field derived from GNSS coordinate time series, this paper proposes a data processing strategy combining global environmental load model correction with variational Bayesian principal component analysis (VBPCA) filtering. Firstly, based on the atmospheric, hydrological, and non-tidal ocean load models provided by the German Research Centre for Geosciences (GFZ), environmental load correction is performed on the original coordinate time series of 17 GNSS stations in the Hong Kong region. The results show that environmental load correction can effectively reduce the root mean square (RMS) values of the vertical (U) time series with an average reduction of approximately 20% and weaken the spatial correlation between stations. However, such correction leads to an increase in the RMS values and spatial correlation of the horizontal north (N) and east (E) time series. Secondly, the VBPCA algorithm which does not require missing data interpolation is adopted to filter and eliminate the common-mode errors of the corrected coordinate time series. After VBPCA filtering, the RMS values in the N, E, and U directions are reduced by approximately 20%, 13%, and 18%, respectively. After the joint processing of environmental load correction and VBPCA filtering, the velocity uncertainties in the N, E, and U directions are averagely reduced by about 63%, 47%, and 68%, respectively. The calculated results based on the ITRF2020 international terrestrial reference frame indicate that the average horizontal crustal movement rate of the GNSS stations in the study area is 34.6±0.2 mm/a with a dominant movement direction of N109.5°E. The vertical movement presents a slow uplift trend, while local subsidence occurs in partial areas. This study provides an effective approach for the construction of high-precision GNSS velocity fields in regions affected by significant environmental load effects and data missing.
-
Journal of Geodesy and Geodynamics. 2026, 46(10): 1182-1190. https://doi.org/10.14075/j.jgg.2026.02.034
Based on the time-variable gravity field data from GRACE/GRACE-FO, a GRACE-DSI drought index with trend and seasonal components removed was constructed. Under a three-dimensional identification framework, drought events in the Yangtze River Basin from 2003 to 2023 were identified and analyzed, and the variations in TWSA components and their interactions were examined through contribution rates and exchange intensity. The results show that: 1) Compared with the run theory, the three-dimensional identification framework detected 19 drought events during the study period, demonstrating clear advantages in accurate drought identification. 2) Drought frequency and intensity exhibited pronounced spatial heterogeneity across the basin. The prolonged drought from 2022 to 2023 was the most severe event, with the highest intensity and an affected area exceeding 90%. 3) Variations in TWSA were primarily dominated by GWSA and SMSA, with strong interactions among components in the western upstream and the northern parts of the middle and lower reaches.
-
Journal of Geodesy and Geodynamics. 2026, 46(10): 1191-1204. https://doi.org/10.14075/j.jgg.2026.02.033
To address the limitations of traditional landslide susceptibility assessments, namely the underutilization of dynamic surface deformation information and insufficient model interpretability, this study proposes an interpretable landslide susceptibility evaluation framework that integrates Stacking-InSAR technology with machine learning, using Guizhou Province, China as the study area. Annual mean surface deformation velocities were derived from Sentinel-1A imagery and combined with fourteen static conditioning factors encompassing topography, land cover, climate, and human activities for modeling. Comparative experiments across multiple machine learning models revealed that without InSAR-derived deformation data, the AUC values ranged from 0.896 to 0.921. Upon integrating deformation information, the overall model performance improved by 3.5%-5.6%, with the InSAR-TabPFN model achieving the optimal performance (AUC = 0.951). Susceptibility mapping results indicated that the areal proportion of high and relatively high susceptibility zones decreased from 22.1% to 13.11%, significantly reducing the misclassification rate of stable areas. SHAP interpretability analysis identified distance to rivers, surface deformation velocity, land use type, slope gradient, and distance to roads as the dominant controlling factors of landslide occurrence, with a notable interaction effect observed between surface deformation and distance to rivers. The findings demonstrate that the TabPFN model incorporating InSAR dynamic deformation data offers significant advantages in predictive accuracy. Furthermore, the integration of SHAP-based interpretability analysis enables the quantitative expression of individual factor contributions and their interaction mechanisms, effectively bridging the gap between predictive accuracy and model interpretability that has long constrained conventional landslide susceptibility assessments.
-
Journal of Geodesy and Geodynamics. 2026, 46(10): 1205-1211. https://doi.org/10.14075/j.jgg.2026.02.037
This study systematically reviews three mainstream methods for tidal harmonic analysis of well water levels: the spectrum analysis method based on fast Fourier transform (FFT), the time domain least squares method, and the Bayesian statistical method (Baytap-G). The mathematical models and computational procedures of each method are elaborated. To address the issues of operational complexity and high entry barriers associated with existing specialized tools, unified and user-friendly MATLAB implementations are provided. Using the well water level observation data from Nanhua county bureau well in Yunnan province as an example, the calculation results of the three methods are compared and analyzed. The results indicate that the three methods yield generally consistent outcomes for the principal tidal constituent (M2), yet each has its own applicable scenarios: the spectrum analysis method is suitable for rapid screening, the time-domain least squares method offers a good balance between computational efficiency and parameter estimation stability, and the Bayesian method demonstrates advantages under complex noise conditions.
-
Journal of Geodesy and Geodynamics. 2026, 46(10): 1212-1219. https://doi.org/10.14075/j.jgg.2025.12.454
To investigate the relationship between the spatial distribution characteristics of soil gas Rn and CO2 concentrations and tectonic activity along the Zaoshi fault and the Changjiangbu fault, three survey lines were deployed across the Zaoshi fault to measure soil gas Rn and CO2 concentrations. The threshold method was adopted to extract the information and intensity of anomalous measurement sites, and an integrated evaluation system was established using the entropy weight method. The results show that: 1)The Pearson correlation coefficient of Rn and CO2 concentrations along the three survey lines is 0.524 4, indicating synchronous variations and the presence of anomalous high-value points.2)Gas emission is relatively strong at the intersection of the two faults and weak at both ends, reflecting the spatial heterogeneity of fault activity.3)Rn exhibits higher entropy value and score contribution than CO2, serving as a more sensitive indicator for reflecting subsurface structural changes in the study area.This study provides a useful reference for the detection of buried faults, analysis of fault activity, and comprehensive evaluation of multi-index soil gas.
-
Journal of Geodesy and Geodynamics. 2026, 46(10): 1220-1227. https://doi.org/10.14075/j.jgg.2026.01.005
Through shallow seismic exploration and controlled source audio-frequency magnetotelluric method, typical geophysical profiles of the western boundary fault zone in Yangqu rift basin are given, and fine fault structural profile characteristics are obtained. The results show that the fault zone can be divided into the northern and southern sections, with Wujiazui-Matou village as the boundary. The north section is Panlong- Gangshang fault zone, with NW strike and length of about 13 km. The south section is the east side fault zone of Qizi mountain, with SN strike and length of about 20 km. Four main branch faults are developed in the south and north sections. Controlled by branch faults, multiple secondary horsts and secondary depressions are formed in the rift basin along the east-west direction. The results of joint drilling profile and stratigraphic chronology test show that the latest activity age of the southern section of the fault zone is late Pleistocene, and the minimum buried depth of the upper breakpoint is 21.1 m, the maximum fault distance is 3.9 m, and the maximum sliding rate is 0.158 mm/a. The results can be used as an important reference for the evolution of Taiyuan basin, the source exploration of large earthquakes, and the seismic risk analysis.
-
Journal of Geodesy and Geodynamics. 2026, 46(10): 1228-1235. https://doi.org/10.14075/j.jgg.2025.12.443
Using continuous waveform data collected by 66 nodal seismometers deployed in the Zaoyang sag area of the Nanxiang basin from March to April 2025, cross-correlation calculations were performed to extract 998 Rayleigh wave phase velocity dispersion curves. A direct surface wave inversion method was applied to obtain a high-resolution shear wave velocity structure down to 3 km depth in the region. The H/V spectral ratio method was also used to preliminarily estimate the sediment thickness. A sudden change in the shear wave velocity interface was observed at 15 km along the survey line: the 2.5 km/s contour rises from 2 km depth to 0.5 km depth from north to south, with a variation amplitude of 1.5 km, consistent with the location of the Xiangfan-Guangji fault. Within the 0-15 km range of the survey line, the shear wave velocity characteristics correlate well with the variation in Cretaceous residual thickness in the Zaoyang sag of the Nanxiang basin, indicating that the Zaoyang sag was a basin controlled by the western segment of the Xiangfan-Guangji fault during the Cretaceous. Based on the morphology of the basin boundary, it is inferred that the western segment of the Xiangfan-Guangji fault behaved as a normal fault dominated by extension, with strike-slip characteristics, during the Cretaceous. The fault blocks within the low-velocity zone and the low-velocity continuity near the surface reflect the basin sedimentation process from differential block uplift at the early stage of basin formation to stable Neogene sedimentation.
-
Journal of Geodesy and Geodynamics. 2026, 46(10): 1236-1244. https://doi.org/10.14075/j.jgg.2026.03.069
Based on the one-dimensional crustal velocity models of three tectonic subregions, western, central and eastern Inner Mongolia, this study adopted the PTD method based on the Pn-Pg travel-time difference to invert the focal depths of earthquake events in Inner Mongolia from 2019 to 2025, with the results verified by the CAP waveform inversion method. Through simulated data tests, the impacts of velocity model errors and phase picking noise on the inversion results were evaluated. The results show that earthquakes in Inner Mongolia are dominated by shallow-focus earthquakes with an average depth of 13.6±7.2 km, and the dominant depths are concentrated in the range of 8.5 to 21.0 km. The spatial distribution exhibits distinct zonal characteristics: the focal depths in the Alxa block of western Inner Mongolia are generally deeper with a mean value of 16.2 km; those in the Hetao graben of central Inner Mongolia are significantly shallower with a mean value of 9.7 km; and those in the Greater Khingan Mountains tectonic belt of eastern Inner Mongolia fall in between with a mean value of 12.7 km. This distribution pattern of shallower in the middle and deeper on both sides is highly correlated with the variations in regional crustal thickness and differences in tectonic stress fields. This study provides depth data support for understanding the regional seismogenic environment and the mechanism of strong earthquakes.
-
Journal of Geodesy and Geodynamics. 2026, 46(10): 1245-1253. https://doi.org/10.14075/j.jgg.2026.02.027
To address the limitations of traditional single denoising method when processing nonlinear and non-stationary seismic signals, we propose a joint denoising method based on variational mode decomposition (VMD) and an improved wavelet threshold decomposition (IWTD). First, a standard VMD decomposition is performed on the seismic signal based on the pre-set number of mode decomposition layers and penalty factors. Second, the intrinsic modal functions (IMFs) obtained from the VMD decomposition are filtered using cross-correlation coefficients to precisely separate signal-dominated components and noise-dominated components. Next, a continuously differentiable improved wavelet threshold function is constructed to perform secondary refined denoising of the selected noise-dominated IMFs, thereby addressing the inherent shortcomings of traditional hard threshold function (discontinuity) and soft threshold function (constant bias). Finally, a weighted comprehensive evaluation scheme for signal denoising is designed to comprehensively assess multiple critical indicators, including effective bandwidth envelope consistency, noise suppression rate, power spectrum consistency, signal kurtosis, and arrival time detection accuracy. The experimental results of both synthetic signals and actual seismic signals demonstrate that, this method achieves comprehensive scores of 76.299 8 and 76.227 2, respectively, exceeding other models by up to 29.782 5 and 27.177 1 compared to empirical mode decomposition (EMD), Wiener filtering, and wavelet threshold filtering. It better preserves the amplitude-frequency characteristics and effective energy of seismic signals, providing a more reliable data foundation for subsequent high-precision phase picking and seismic sequence analysis.
-
Journal of Geodesy and Geodynamics. 2026, 46(10): 1254-1262. https://doi.org/10.14075/j.jgg.2025.11.369
This paper presents a gravity-gradient inversion method for fault-model parameters that integrates the covariance matrix adaptation evolution strategy (CMA-ES) with simulated annealing (SA). To mitigate the boundary-effect problem of CMA-ES in bounded parameter spaces, SA is introduced to perform a coarse global search, thereby broadening the search space and improving convergence behavior. Tests on single-fault and complex three-fault models under noise-free conditions and various Gaussian noise levels (10% and 20%) demonstrate that the hybrid strategy combines the global search capability of SA with the adaptive local-search advantage of CMA-ES, exhibiting excellent convergence and stability while significantly enhancing inversion efficiency. Application of this method to the inversion of EGM2008 differential-curvature gravity-gradient data in the Thrace basin, Turkey, yields geometric parameters such as fault dip angles that agree well with seismic data and previous research findings, with computational efficiency being substantially improved. The results indicate that the proposed method enables qualitative to semi-quantitative interpretation of buried-fault parameters, offering a new technical approach for investigating regional fault-structural characteristics.
-
Journal of Geodesy and Geodynamics. 2026, 46(10): 1263-1272. https://doi.org/10.14075/j.jgg.2025.12.458
High-precision correction of zenith tropospheric delay (ZTD) can effectively enhance the positioning accuracy of global navigation satellite systems (GNSS). This study proposes a window-enhanced, real-time, high-precision gridded ZTD model for Chinese mainland, integrating machine learning algorithms (XGBoost and BP neural networks) with grid partitioning approach and utilizing the fifth-generation reanalysis data (ERA5) from the European Centre for Medium-Range Weather Forecasts (ECMWF). We developed two models, namely the GGZTD-XG and GGZTD-BP models, for real-time high-precision ZTD estimation. Their performance is compared with commonly used GPT3 and GGZTD-P models. The results show that, using ERA5 ZTD data as the reference, the GGZTD-XG model achieved higher accuracy than the other models, with the root mean square error (RMSE) reduced by 0.75 cm (22.32%) and 1.53 cm (36.96%) respectively and the GGZTD-BP model also demonstrated notable improvement. Furthermore, validation using GNSS ZTD and MERRA-2 data confirmed the superior performance of the proposed models. Therefore, the new models exhibit significant potential for applications in high-precision GNSS positioning.
-
Journal of Geodesy and Geodynamics. 2026, 46(10): 1273-1284. https://doi.org/10.14075/j.jgg.2025.12.460
The latest version of the international reference ionosphere (IRI) model, IRI2020, faces significant challenges in predicting the mid-latitude summer nighttime anomaly (MSNA), an ionospheric phenomenon characterized by anomalous nighttime enhancement of total electron content (TEC). Using GNSS TEC and CODE TEC observational data, this study systematically evaluates the predictive performance of IRI2020 under varying solar activity conditions in MSNA regions. The main conclusions are as follows: 1) Comparison with GNSS TEC reveals that the applicability of IRI2020 exhibits pronounced solar activity dependence and hemispheric asymmetry. During the ascending phase and solar maximum years, the model shows a systematic underestimation of nighttime TEC by approximately 10 TECu and fails to accurately capture the typical characteristics of MSNA; the prediction accuracy during the descending phase and solar minimum years is markedly superior to that during the ascending phase and solar maximum, with the northern hemisphere generally outperforming the southern hemisphere. 2) monthly-averaged MSNA index (MSI) heatmaps based on CODE TEC data indicate that the model underestimates MSNA to varying degrees during summer throughout the entire solar cycle. The spatial extent of the anomaly in the southern hemisphere is poorly predicted, while the northern hemisphere shows acceptable performance only during solar minimum years, suggesting limited overall quantitative capability. 3) Regional residual analysis further reveals that IRI2020 exhibits significant underestimation during solar maximum years (particularly pronounced in the southern hemisphere, with deviations of approximately 11 TECu), accompanied by large error dispersion; prediction accuracy improves substantially during solar minimum years, with error distributions becoming more convergent. Overall, the predictive capability of IRI2020 is notably limited during summer periods in MSNA regions, whereas it demonstrates higher adaptability during most non-summer periods.
-
Journal of Geodesy and Geodynamics. 2026, 46(10): 1285-1293. https://doi.org/10.14075/j.jgg.2026.01.004
Addressing the issue of traditional methods typically estimate the wide-lane (WL) uncalibrated phase delay (UPD) as a daily constant, ignoring its time-varying characteristics during flex power periods, which may affect the accuracy of UPD estimation. We propose a piecewise constant model for UPD estimation that accounts for the effects of flex power. In this model, UPD is estimated in segments according to flex power transitions, and the specific impact of flex power on satellite WL UPD is analyzed. The results show that flex power can impact the between-satellite single-difference WL UPD by up to 0.912 cycles for certain satellites. Furthermore, precise point positioning ambiguity resolution (PPP-AR) experiments were conducted to validate the performance of UPD products generated by the proposed method. Compared with the traditional daily constant estimation strategy, the piecewise constant strategy estimates UPD more accurately, increasing the proportion of WL ambiguity residuals within ±0.25 cycles by 2 percentage points. Consequently, positioning accuracy in E, N, and U directions improved by 18.2% (0.6 cm), 8.7% (0.2 cm), and 4.0% (0.2 cm), respectively. These results indicate that adopting a piecewise constant model during flex power periods can improve the precision of UPD estimation, thereby enhancing the accuracy of fixed solutions.
-
Journal of Geodesy and Geodynamics. 2026, 46(10): 1294-1304. https://doi.org/10.14075/j.jgg.2026.01.019
To improve the accuracy and realism of subglacial terrain reconstruction in Antarctica, this paper proposes a reconstruction method based on prior information from a global reference terrain database. First, a global reference terrain library is constructed and a multi-dimensional progressive matching strategy is designed to introduce terrain data from regions with similar features, thereby enriching the information available for interpolation. Second, within the existing terrain interpolation framework, a reference terrain-guided branch based on a deep⁃shallow feature fusion mechanism is developed to enable efficient utilization of prior information. Experiments conducted in the Gamburtsev Mountains and the Weddell Sea sector demonstrate that, compared to Kriging interpolation and the method without a reference mechanism, the proposed method effectively improves both the detail diversity and roughness fidelity of the reconstructed terrain.
