This study takes the agricultural irrigation district of Changji, Xinjiang as the research object. By integrating time-series InSAR, geographically weighted regression (GWR), and wavelet analysis methods, and based on Sentinel-1A data from 2017 to 2020, the surface deformation field was obtained. Multi-source data were fused to quantify the spatial heterogeneity of driving factors, and periodic characteristics of the deformation time series were extracted. The results show that: 1)The deformation in the study area exhibits a spatial pattern of stable in the south, subsiding in the center,and uplifting in the north. The maximum subsidence rate in the central subsidence funnel reached -90.1 mm/a, while the uplift rate in the northern area reached 21.7 mm/a. Temporally, the deformation showed continuous accumulation superimposed with seasonal fluctuations, with the maximum cumulative subsidence reaching -407.3 mm in August 2020. 2)The GWR model revealed significant spatial heterogeneity in the driving mechanisms: stability in the south is controlled by abundant rainfall and coarse-grained aquifers; subsidence in the central area is closely related to groundwater over-extraction and fine-grained, easily compressible strata; uplift in the north is mainly driven by human-regulated water level recovery. 3)Wavelet analysis identified two main periodicities of 372 days and 348 days in the subsidence areas, whose spatial distribution is highly consistent with the intensity of agricultural irrigation. This indicates that agricultural extraction activities are the primary factor controlling the spatial differentiation of these periodic characteristics. This study provides a basis for differentiated prevention and control of land subsidence and adaptive water resource management in arid regions.

This study proposes an improved DeepLabV3 model for the identification of retrogressive thaw slumps in frozen ground regions. The method integrates the Sobel operator with a boundary-enhanced dual attention (BEDA) module to accurately extract retrogressive thaw slump boundaries; a deformable context pyramid (DCP) module is employed to adapt to retrogressive thaw slumps of varying scales and morphological characteristics. Weighted residual connections are introduced to preserve the original semantic information and enhance detail representation. The dataset of retrogressive thaw slumps was constructed through visual interpretation using SBAS-InSAR surface deformation results, mountain shadow information, and Google Earth imagery. Experimental results show that the overall outputs of the model are closer to the ground truth labels, and the Recall and MIoU of the improved model are increased by 1.59% and 2.09%, respectively, compared with the baseline model.

Based on Sentinel-1A ascending and descending SAR data, the co-seismic InSAR deformation field and range-direction pixel offset of the January 7, 2025, M_W7.1 Dingri earthquake were obtained. Combined with elastic half-space dislocation theory, the particle swarm optimization (PSO) algorithm and the Akaike-Bayesian information criterion (ABIC) were used to invert single- and double-fault slip distribution models. Static Coulomb stress changes at depths of 5 km, 10 km, and 15 km were also calculated. The results show that although the double-fault model, constrained by surface rupture traces identified via the pixel offset tracking (POT) method, exhibits higher roughness, it more accurately reproduces the surface rupture trajectory at fault strike variations. The double-fault model achieved fitting accuracies of 97.2% and 96.5% for ascending and descending InSAR data, respectively, outperforming the single-fault model (96.8%/95.9%). The inverted seismogenic fault has a strike of 186° and a dip angle of 55°, with co-seismic slip mainly concentrated in the northern section and a secondary slip core between the two fault strands. The total seismic moment released was 6.12×10¹⁹ Nm (M_W7.1). Coulomb stress perturbations generally decreased on both sides of the fault, while stress increased at a depth of 15 km in the southern segment, showing strong spatial correlation with aftershock distribution, indicating that stress in this region is prone to release through aftershocks. Local stress increases were also observed at different depths near the northern and southern ends of the fault, which should be key areas for future research.

Aiming at the problem that the static relative positioning technology has a long calculation cycle and cannot timely reflect abrupt deformations when using Beidou/GNSS technology for infrastructure deformation monitoring currently, this paper proposes a real-time fusion method for monitoring results of static and RTK with different time scales and positioning accuracies, based on wavelet decomposition and dynamic weighted fusion of sliding time windows. Through experimental tests in the ultra-short baseline field, it is verified that this method can provide millimeter-level monitoring accuracy results when the monitored object is stable or deforms slowly, and can timely output the actual displacement when abrupt deformation occurs, with accuracy superior to that of RTK monitoring results. The method can be implemented based on general-purpose GNSS data processing software available on the market, and has a good prospect of engineering application.

To address the limited positioning accuracy of global geomagnetic field models in localized regions, this paper proposes a geomagnetic positioning method based on a genetic algorithm-optimized extreme learning machine (GGA-ELM). By introducing an elite backpropagation strategy, the initial weights and thresholds of the ELM network are optimized, thereby improving training efficiency and reducing the risk of getting stuck in local optima. Based on actual aeromagnetic data from the Yuyao and Weishui regions in China, and in combination with the EMM2017, WMM2020, and IGRF13 geomagnetic field models, this study analyzes the differences in positioning accuracy between model data and observed data. Experiments demonstrate that the EMM2017 model achieves the best overall positioning accuracy at both sites (Yuyao: 6.29 m in longitude and 5.92 m in latitude; Weishui: 6.81 m in longitude and 5.89 m in latitude), significantly outperforming the WMM2020 and IGRF13 models. In the measured aeromagnetic data, the error in the latitudinal direction was slightly greater than that in the longitudinal direction (with a difference in root mean square error of approximately 0.5 m), whereas the geomagnetic field model data exhibited a characteristic where the longitudinal error was greater than the latitudinal error. The GGA-ELM method performed consistently at both locations, with positioning accuracy differences of less than 5%, verifying the robustness of the GGA-ELM method. This study provides a theoretical basis for model selection and algorithm optimization in geomagnetic navigation.

Addressing the challenges of weak scattering and low coherence in airport environments, we establishe a "macro-to-fine" hierarchical deformation monitoring framework based on the Chinese domestic "Fucheng-1" high-resolution SAR satellite, integrating SBAS-InSAR and an improved IPTA technique. A dynamic noise suppression baseline is constructed to optimize the amplitude dispersion index model, and an enhanced IPTA algorithm incorporating adaptive adjustment parameters is proposed, effectively mitigating the difficulty of coherent point selection in low-scattering areas. The results demonstrate that: 1) The improved algorithm increases the density of effective monitoring points within the airport area by 31.7% compared to conventional methods, with an internal accuracy better than 4.1 mm/a. 2) By incorporating multi-source environmental and meteorological data, the physical mechanism of minor differential settlement along the 1.5 km to 2.0 km runway segment is revealed to be driven by seasonal freeze-thaw cycles. The findings confirm the superior applicability and robustness of domestic high-resolution commercial SAR satellites combined with advanced time-series InSAR algorithms for monitoring subtle deformation in transportation infrastructure.

Based on Sentinel-1 data from 2015 to 2024, technical methods such as SBAS-InSAR deformation inversion, independent component analysis (ICA), K-means clustering analysis, and wavelet transform analysis are used to study the surface deformation characteristics of Xi'an city over approximately 10 years. The results show that the average surface deformation rate in Xi'an ranges from －40 to 11 mm/a, with a spatiotemporal heterogeneity in deformation distribution. There are three modes of surface deformation in Xi'an: stable fluctuation, continuous subsidence, and subsidence followed by uplift. K-means clustering analysis of surface deformation within a 1 km buffer zone along metro lines reveals that the stable deformation areas account for the largest proportion (54%), the subsidence areas (16.2%) distribute in the southern and western urban areas, and uplift areas (29.8%) concentrate in the central urban area. The four metro lines with the largest surface deformation are line 2, 3, 4, and 5, with most stations showing a trend of subsidence followed by stability or even uplift. Further analysis indicates that groundwater recharge in Xi'an in 2018 was the main reason for the change in deformation trend of metro lines. The rainfall is an important factor inducing seasonal fluctuations in metro line deformation. The research results provide an important basis for the safe use of urban underground space and the prevention of geological disasters.

Conventional InSAR monitoring can only obtain one-dimensional deformation along the line-of-sight (LOS) direction, making it difficult to directly invert the three-dimensional movement mechanism of landslides. Moreover, the quantitative relationship between deformation time series and triggering factors such as rainfall remains unclear. To address this, we take the Nuole landslide as an example. By introducing the surface parallel flow(SPF) model, which is suitable for landslides whose movement direction is controlled by topography, as a constraint and integrating ascending and descending time-series InSAR observations, we successfully derived a high-precision three-dimensional deformation field. The results reveal that the landslide exhibits composite movement characteristics in the westward, northward, and downward directions, with westward motion being dominant at a maximum deformation rate of -140 mm/a. Furthermore, variational mode decomposition (VMD) was applied to accurately extract periodic signals from the deformation time series, and wavelet analysis was employed to quantify the triggering effect of rainfall on deformation. The study identified a significant common annual oscillation period between deformation and rainfall, with periodic deformation lagging behind rainfall by approximately one-quarter of the period (about 3 months). This lag period reflects the complete physical process of rainfall infiltration, slip-zone softening, and deformation response. The study provides an effective approach for quantitative analysis of the three-dimensional movement mechanism of landslides and rainfall-induced triggering mechanism.

Taking the Urumqi region characterized by complex tectonic activity and variable environment as a case study, we systematically analyze the noise characteristics and environmental loading impact of GNSS coordinate time series based on multi-year continuous GNSS observations. The results indicate that the regional noise time series are predominantly characterized by a combination model of white noise plus flicker noise or white noise plus power-law noise. The noise amplitude of vertical component is significantly higher than that of the horizontal component, and the spatial distribution exhibits notable heterogeneity. Influenced by seasonal hydrological mass migration and atmospheric loading, the annual signals in the vertical series are largely attributable to environmental loading (> 70%). Moreover, long-term variations in groundwater levels can introduce deviations comparable to tectonic trends. At URU2 station, the cumulative effect over 14.6 years without correction can reach up to ±36.5 mm. The study further reveals the heightened noise complexity and greater disturbance at urban and alluvial stations (URU2 station), highlighting the critical impact of local geological conditions and human activities on station stability.

Taking the Yingxiu section of the Longmenshan fault zone as the study area, a 2D vertical-profile viscoelastic finite element model is constructed. The model fully considers the vertical stratification and lateral heterogeneity of the medium in the study area. Based on the slip-weakening friction law, numerical simulations are carried out for the entire periodic spontaneous rupture process of thrust earthquakes, covering three key stages: interseismic, coseismic, and postseismic stress adjustments.The results show that, on the premise that the simulated interseismic velocity field is in good agreement with the pre-earthquake GPS observations of the Wenchuan earthquake, the difference between the static and dynamic friction coefficients exerts a significant control on the coseismic surface displacement. The optimally fitted static and dynamic friction coefficients are 0.35 and 0.23, respectively. The simulated maximum coseismic slip on the fault plane during a single earthquake is approximately 7 m. Since the postseismic deformation simulation in this study only considers the viscoelastic relaxation effect of the middle and lower crust without incorporating postseismic afterslip, the simulated near-field postseismic deformation is lower than the observed values, while the far-field simulation results agree well with the observations. Under the computational parameters and assumptions adopted in this study, the recurrence interval of the Wenchuan earthquake is simulated to be approximately 700 to 770 years, which is generally consistent with estimates based on the deep detachment layer locking model for the Yingxiu section of the Longmenshan fault zone.