A landslide dynamic deformation and fracture monitoring method and system

Abstract

This invention discloses a landslide dynamic deformation and rupture monitoring method integrating InSAR and UAV, belonging to the field of geological disaster monitoring technology. It aims to address the problems of highly concealed bedding landslides in high mountain and canyon areas, insufficient spatiotemporal resolution of traditional monitoring methods, and the difficulty of using a single technology to simultaneously monitor macroscopic deformation and microscopic rupture. The core of its technical solution is as follows: first, establish a monitoring spatial benchmark and preprocess multi-source SAR data; then, calculate the global deformation field and optimize zoning through a combination of PS-InSAR and SBAS-InSAR; finally, collect rupture features using UAV and construct a three-dimensional model; and establish a deformation-rupture correlation through spatiotemporal fusion of multi-source data, combined with a dynamic early warning mechanism, to achieve full-scale dynamic monitoring and graded early warning of landslides, providing support for early identification and risk management of landslides.

Description

Technical Field

[0001] This invention relates to landslide ground deformation and surface fracturing monitoring technology, and more particularly to a method and system for monitoring dynamic deformation and fracturing of landslides. Background Technology

[0002] Bedding-parallel landslides are a typical type of landslide in high mountain and canyon areas. Due to their thin rock strata, fractured rock mass, and well-developed weak interlayers, their failure modes, under inducing factors, are characterized by concealment, suddenness, complexity, rapidity, and high destructiveness, seriously threatening critical infrastructure and the safety of residents' lives and property in mountainous areas. As a common geological hazard, bedding-parallel landslides are often accompanied by significant surface deformation; therefore, monitoring their dynamic deformation and fracturing processes is crucial.

[0003] Traditional methods for monitoring bedding landslides and understanding landslide evolution mainly rely on field surveys, model tests, and numerical simulations. Field surveys are easily limited by terrain conditions, making manual investigation difficult; model tests can only reveal qualitative information, not quantitative details; traditional finite element and discrete element methods are more focused on analyzing the stable state of landslides than on presenting their dynamic evolution process.

[0004] In recent years, Interferometric Synthetic Aperture Radar (InSAR) technology has been widely used in landslide monitoring and early identification due to its ability to penetrate clouds and vegetation, providing large-scale, high-precision monitoring of surface deformation. InSAR technology can detect millimeter-level surface deformation by analyzing phase changes in radar signals, making it suitable for the early identification and monitoring of geological hazards such as landslides and earthquakes. However, the application of InSAR monitoring technology in landslides still faces several challenges. InSAR technology is limited by baseline and phase unwrapping techniques when monitoring localized, minute deformations, and its accuracy and reliability may be affected in areas with complex terrain. Furthermore, InSAR acquires vertical deformation along the elevation direction, while landslide deformation is a vector deformation along the sliding surface; this difference in deformation direction makes direct use of InSAR monitoring results for early landslide identification unreliable. UAV (Unmanned Aerial Vehicle) low-altitude photogrammetry, as an emerging monitoring method, has shown great potential in the field of geological hazard monitoring due to its flexibility, low cost, and high resolution. UAVs can monitor crack changes at close range, and aerial triangulation can perform joint adjustment calculations to construct high-precision digital surface models (DSMs), realizing three-dimensional spatial coordinate data of landslides. However, the UAV measurement process is easily affected by sudden extreme weather, wind direction, and flight altitude. Bedding landslides are located in high mountain and canyon areas with uneven vegetation cover, requiring the integration of different InSAR technologies for ground deformation monitoring to improve monitoring accuracy. Furthermore, the digital extraction and intelligent identification of multi-source information from the massive point cloud and image data acquired by UAV photography is a pressing technical problem that needs to be solved.

[0005] Interferometric Synthetic Aperture Radar (InSAR) technology has significant advantages in observing areas of slow surface deformation, enabling regional, comprehensive, and large-scale surveys of potential major geological hazards. Currently, InSAR technology is widely used in landslide geological hazard monitoring and identification both domestically and internationally. Based on InSAR technology, differential interferometry (DInSAR) and temporal InSAR, namely permanent scatterer interferometry (PSInSAR) and small baseline set interferometry (SBAS-InSAR), have been gradually developed, capable of detecting surface deformation at the millimeter level. However, the accuracy and reliability of InSAR are severely affected by spatial incoherence, making it difficult to further meet the needs of landslide monitoring research. UAV photogrammetry technology is widely used for observing and identifying landslide hazards in high-risk areas, areas with concentrated hazards, or hazard points, including topography, surface fracturing signs, and even rock mass structures. High-resolution images acquired by UAV orthophoto / oblique photogrammetry drones at different time points can generate different digital surface models (DSMs). Their sub-centimeter error range can be used to calculate surface offsets and compare changes in landforms before and after the event. LiDAR can not only acquire high-precision DSMs from laser point cloud data, but also obtain high-precision digital elevation models (DEMs) by removing vegetation. Combined with DEM visualization methods such as red stereo maps and sky view factors, it is possible to clearly display landslide micro-topography such as large accumulations, landslide walls, landslide boundaries, cracks, and landslide embankments. Using a single solution method can lead to large local errors in the monitoring area due to complex ground reflection zones, making it difficult to identify the ground fracturing state of potential landslide sites. Therefore, effectively integrating and complementing InSAR monitoring and UAV measurements for early landslide identification can effectively overcome these challenges.

[0006] To address the limitations and challenges of existing technologies, this invention proposes a method for monitoring the dynamic deformation and rupture of landslides. This method identifies macroscopic deformation of landslides using InSAR; delineates surface micro-deformation and ground fissures using UAV photogrammetry and airborne LiDAR; and conducts ground surveys to examine deep displacement within the slope, groundwater levels, and fissure development. The aim is to combine the macroscopic monitoring capabilities of InSAR with the microscopic monitoring advantages of UAV, and through data fusion technology, achieve comprehensive and high-precision monitoring of bedding landslide areas, effectively improving the efficiency and accuracy of geological disaster monitoring. Summary of the Invention

[0007] The purpose of this invention is to provide a method for monitoring the dynamic deformation and rupture of landslides by integrating InSAR and UAV, which can identify the deformation characteristics and signs of landslides in the survey area, determine the degree of landslide development, delineate the boundary range of landslides, and identify slope deformation cracks. This provides a more effective method for early identification of landslides, solves the shortcomings of traditional single measurement methods, and identifies landslides before they slide, thus providing support for the monitoring, early warning, and risk management of bedding landslides.

[0008] To achieve the above objectives, this application adopts the following technical solution: In a first aspect, the present invention provides a method for monitoring the dynamic deformation and rupture of landslides, comprising: By integrating field geological survey data, historical engineering geological data, and multi-temporal high-resolution optical remote sensing images, and then performing correction and preprocessing, the accurate range and geographical location of the target landslide are determined. Based on the accurate range and geographical location of the target landslide, SAR image data that meets the constraints of time and space baselines are selected, and a multi-source SAR database is obtained through preprocessing. Based on a multi-source SAR database, an interferometric pair network was constructed using the PS-InSAR method. A digital elevation model of the monitoring area was synchronously integrated, and a differential interferogram was generated using the two-track differential interferometry method. The amplitude deviation index method was used to screen permanent scatterers in the differential interferogram, and a spatiotemporal linear deformation model was constructed. The landslide surface deformation rate was obtained by solving the spatiotemporal linear deformation model. A short baseline interferometric pair network is constructed based on multi-source SAR data. Interferograms are generated by conjugate multiplication of master and slave image pixels. SBAS-InSAR technology is used to complete image interferometry and phase unwrapping to interpret surface deformation information and deformation rate. The interferogram is masked by combining digital elevation model. Based on the unwrapped phase, the deformation rate is calculated by SVD method. After integration, a deformation time series is generated, and the time-series deformation field is reconstructed to provide information on the change of surface deformation over time. Based on spatial statistical methods and kernel density estimation, cluster analysis and partitioning of landslide surface deformation rate are performed, and SBAS-InSAR deformation rate raster data are converted into vector surface data to generate a spatial distribution map of landslide deformation risk level. Based on the spatial distribution map of landslide deformation risk level, a spatial heterogeneous fusion model is constructed to differentiate the data of dense and sparse deformation areas and obtain the monitoring results of ground deformation across the entire region. Data was acquired through UAV oblique photography and airborne LiDAR, and 3D point clouds were generated using the SfM algorithm. Deep learning algorithms were combined to automatically identify surface fracture features, and a constrained triangulation algorithm was used to construct a 3D real-world model. The results of the global ground deformation monitoring are georeferenced and scale-matched with the 3D reality model. Based on the spatiotemporal coupling coefficient and dynamically optimized early warning threshold, a closed-loop monitoring and hierarchical early warning mechanism is formed.

[0009] As a further improvement of the present invention, based on the accurate range and geographical location of the target landslide, SAR image data that meets the constraints of time and spatial baselines is selected, and through preprocessing, a multi-source SAR database is obtained, including: Based on the accurate range and geographical location of the target landslide, satellite ascent and descent data are selected, the time baseline covers at least one complete hydrological year, and at least N>25 synthetic aperture radar (SAR) image data are acquired to form a SAR image data set, so that the spatial baseline is constrained within the set range. For SAR image datasets, terrain phase contribution errors are eliminated; then, orbital drift is corrected using precise orbital data, and phase unwrapping is performed using a minimum cost flow algorithm, so that the residual phase error is controlled within a set range.

[0010] As a further improvement to the present invention, the spatiotemporal linear deformation model is constructed as follows:

[0011] in, For deformation phase, The phase of the residual error in the DEM. For atmospheric delayed phase, Noise phase; As a further improvement to this invention, the method of clustering and partitioning the landslide surface deformation rate based on spatial statistical methods and kernel density estimation, converting SBAS-InSAR deformation rate raster data into vector surface data, and generating a spatial distribution map of landslide deformation risk levels includes: Spatial statistical methods from hotspot analysis are used to perform significant cluster analysis on PS-InSAR deformation points, calculate the local spatial autocorrelation index and statistical significance of each deformation point, and identify deformation hotspots and coldspots. Kernel density estimation is combined to quantify the spatial distribution density of PS points and define the threshold boundaries between dense and sparse regions. An irregular triangular mesh is constructed using a triangulation algorithm, with the spatial coordinates of PS points as nodes to generate the initial mesh. Significantly clustered spatial regions are selected, and the boundary of the triangular mesh is corrected using inverse distance weighted interpolation to extract the topological boundary between dense and sparse regions. A raster-to-vector conversion algorithm is used to convert the deformation rate raster data obtained by SBAS-InSAR into vector surface data with attribute topology, and the PS point density zoning results are fused to generate a spatial distribution map of landslide deformation risk level.

[0012] As a further improvement of the present invention, the step of using spatial statistical methods in hotspot analysis to perform significant cluster analysis on PS-InSAR deformation points, calculating the local spatial autocorrelation index and statistical significance of each deformation point, and identifying deformation hotspots and coldspot regions includes: Collect InSAR data, including multi-temporal radar imagery; InSAR processing involves generating interferograms, calculating phase differences, and generating deformation maps. Identify deformation signals in the deformation diagram and eliminate non-deformation factors; Perform hotspot analysis on the deformation map to identify deformation hotspot regions; Spatial autocorrelation theory statistics are used to quantify the spatial heterogeneity of deformation, extract high-confidence deformation hot zones, and optimize the geometric accuracy of hot zone boundaries through triangulation to obtain the results of hotspot analysis; Based on the results of hotspot analysis, identify points or regions with significant deformation; Extract the deformation rate and deformation amount of deformation points or regions; perform spatial clustering of the deformation rate and deformation amount of deformation points or regions to identify deformation areas; perform time series analysis based on the data to obtain the changing trend of deformation points over time.

[0013] As a further improvement of the present invention, the step of constructing an irregular triangular mesh using a triangulation algorithm, generating an initial mesh using the spatial coordinates of PS points as nodes, filtering out spatially clustered regions, correcting the triangular mesh boundary using inverse distance weighted interpolation, and extracting the topological boundary lines between dense and sparse regions includes: DEM is generated by constructing a grid using boundary points. The ground model with a spatial triangular mesh structure is connected by discrete points at the boundary of the landslide monitoring area using a corresponding algorithm to generate the corresponding DEM image. The basic process of establishing an irregular triangular network is to first connect the three nearest discrete points to form a triangle, and then connect the nearest discrete points outward based on each side of the triangle to form another three triangles, and so on, until all discrete points are connected. UAVs were used to acquire image data of different resolutions, GPS RTK was used to measure field control points, and high-precision DEM data was generated using 3D reconstruction and aerial triangulation. Then, an irregular grid was constructed using boundary points to generate reference surface DEM data.

[0014] As a further improvement of the present invention, the step of georeferencing and scale matching the global ground deformation monitoring results with the three-dimensional real-scene model, and forming a closed-loop monitoring and hierarchical early warning mechanism based on the spatiotemporal coupling coefficient and dynamically optimized early warning threshold, includes: Perform geographic coordinate registration to determine a common geographic coordinate system, and perform geometric correction to ensure that InSAR and UAV images are aligned in the same coordinate system; The image coordinates are transformed to a unified projected coordinate system, and the image is resampled to match the resolution; the boundary lines of InSAR and UAV images are identified, and image processing techniques are used to smooth the boundary lines; Based on pixel and feature overlay, InSAR deformed images and UAV deformed images are overlaid, the overlaid image is analyzed, the registration error and overlay quality are evaluated, the overlaid image is interpreted, deformation hotspots are identified, deformation patterns are analyzed, and the overlay results are displayed graphically.

[0015] As a further improvement of the present invention, the early warning threshold based on the spatiotemporal coupling coefficient and dynamic optimization includes: The dynamic early warning threshold is adaptively optimized by introducing a Kalman filter update mechanism and fusing real-time monitoring data to correct the early warning model parameters.

[0016] in, The creep coefficient of the soil and rock mass. for t Real-time measured deformation-fracture increment The observation matrix; The threshold dynamic adjustment function is obtained by training with a historical disaster case database, and the early warning threshold is obtained.

[0017] As a further improvement of the present invention, the pixel- and feature-based overlay method involves overlaying InSAR deformed images and UAV deformed images, analyzing the overlaid image, evaluating registration error and overlay quality, interpreting the overlaid image, identifying deformation hotspot regions, analyzing deformation patterns, and displaying the overlay result graphically, including: Construct a spatiotemporal kriging interpolation model that integrates the InSAR deformation rate field and the UAV rupture displacement vector:

[0018] In the formula, , Weighting coefficients ; Granger causality test was used to analyze the spatiotemporal lag relationship between deformation-driven and fracture response; Calculate the Landslide Activity Index (LAI):

[0019] in, , Deformation rate With crack length The contribution weights are determined through principal component analysis; An interactive monitoring platform was developed based on the WebGL engine, integrating time-series deformation cloud maps, fracture propagation animations, and LAI heatmaps. The reliability of the fused data was cross-verified using ground-based GNSS monitoring stations; Output a dynamic monitoring report of landslides, including a deformation-fracture coupling matrix, a risk level zoning map, and an early warning decision tree.

[0020] Secondly, the present invention provides a landslide dynamic deformation and rupture monitoring system, comprising: The location determination module is used to integrate field geological survey data, historical engineering geological data and multi-temporal high-resolution optical remote sensing images, and then determine the accurate range and geographical location of the target landslide after correction and preprocessing. The database processing module is used to select SAR image data that meets the constraints of time and space baselines based on the accurate range and geographical location of the target landslide, and obtain a multi-source SAR database through preprocessing. The deformation model construction module is used to construct an interferometric pair network based on a multi-source SAR database using the PS-InSAR method, synchronously integrate the digital elevation model of the monitoring area, generate differential interferograms using the two-track differential interferometry method, and use the amplitude deviation index method to screen permanent scatterers in the differential interferograms to construct a spatiotemporal linear deformation model; the landslide surface deformation rate is obtained by solving the spatiotemporal linear deformation model. The deformation model construction module is used to construct a short baseline interferometric pair network based on multi-source SAR data, and generate interferograms by multiplying master and slave image pixels by conjugate. It uses SBAS-InSAR technology to complete image interferometry and phase unwrapping, interpreting surface deformation information and deformation rate. Combined with digital elevation model, the interferogram mask is processed, and the deformation rate is calculated using the SVD method based on the unwrapped phase. After integration, the deformation time series is generated, the temporal deformation field is reconstructed, and information on the change of surface deformation over time is provided. The distribution map generation module is used to perform cluster analysis and partitioning of landslide surface deformation rate based on spatial statistical methods and kernel density estimation, convert SBAS-InSAR deformation rate raster data into vector surface data, and generate a spatial distribution map of landslide deformation risk level. The monitoring results acquisition module is used to construct a spatial heterogeneous fusion model based on the spatial distribution map of landslide deformation risk level, differentiate the data of dense and sparse deformation areas, and obtain the monitoring results of ground deformation across the entire area. The overlay and fusion module is used to acquire data through UAV oblique photography and airborne LiDAR, generate 3D point clouds using the SfM algorithm, automatically identify surface fracture features using deep learning algorithms, and construct 3D reality models using constrained triangulation algorithms. The matching and early warning module is used to perform geographic registration and scale matching between the monitoring results of ground deformation across the entire region and the three-dimensional real scene model. Based on the spatiotemporal coupling coefficient and dynamically optimized early warning threshold, a closed-loop monitoring and hierarchical early warning mechanism is formed.

[0021] Thirdly, the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the landslide dynamic deformation and rupture monitoring method.

[0022] Fourthly, the present invention provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program, which, when executed by a processor, implements the landslide dynamic deformation and rupture monitoring method.

[0023] Fifthly, the present invention provides a computer program product, the computer program product including computer instructions, characterized in that the computer instructions instruct a computer to execute the landslide dynamic deformation and rupture monitoring method.

[0024] The beneficial effects of this invention compared to the prior art are as follows: This invention considers the fusion and complementarity of InSAR and UAV technologies in monitoring landslide deformation and fracturing. InSAR technology can detect surface deformation at the millimeter level, while UAV technology can capture local cracks, faulting, and rupture characteristics of landslides. The combination of the two technologies significantly improves monitoring accuracy and broadens the monitoring range of landslides. Utilizing UAV technology for close-range monitoring and real-time data acquisition and transmission allows for rapid response and capture of the immediate dynamics of landslide areas, while combining it with InSAR technology provides periodic macroscopic deformation data. This invention enables continuous dynamic monitoring of landslides, making early warning and rapid response to landslide disasters possible. InSAR technology is not limited by weather and lighting conditions, while UAV technology, through flexible flight path planning, can monitor under various environmental conditions, greatly improving the environmental adaptability of the monitoring method. Through advanced data fusion algorithms, this invention can effectively integrate data acquired by InSAR and UAV, improving data consistency and reliability, and providing more comprehensive information and reliable early identification for landslide disaster analysis. Attached Figure Description

[0025] Figure 1 This invention provides a flowchart for monitoring the dynamic deformation and rupture of landslides. Figure 2 This is a schematic diagram of a landslide dynamic deformation and rupture monitoring principle provided by the present invention; Figure 3 This is a slope deformation rate diagram of the study area provided by the present invention; Figure 4 This is a flowchart of the UAV monitoring and landslide image acquisition process provided by the present invention; Figure 5 This is a rendering of the three-dimensional real-scene model of the research area provided by the present invention. Detailed Implementation

[0026] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0027] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0028] This invention provides a method for monitoring the dynamic deformation and rupture of landslides by integrating synthetic aperture radar interferometry (InSAR) and unmanned aerial vehicle (UAV) technologies. It addresses the problems of the strong concealment of bedding landslides in high mountain and canyon areas and the insufficient spatiotemporal resolution of traditional monitoring methods. By collaboratively interpreting multi-source remote sensing data, it achieves full-scale dynamic tracking of landslides from macroscopic deformation to microscopic rupture. The core technologies include: (1) Multi-time series InSAR deformation calculation: Based on PS-InSAR (Permanent Scatter Interferometry) and SBAS-InSAR (Short Baseline Set Interferometry), the millimeter-level deformation rate field of the landslide area is obtained. PS-InSAR extracts high-precision deformation for stable scatterers, and SBAS-InSAR suppresses surface decoherence through short baseline combination; (2) UAV high-resolution 3D modeling: Sub-meter-level digital elevation model (DEM) and surface model (DSM) are constructed using UAV oblique photography and LiDAR point cloud data to extract geometric parameters of rupture features such as ground fissures and faults; (3) Multi-scale data fusion and risk zoning: The Getis-OrdGi* hotspot analysis method is used to perform significant clustering of InSAR deformation points to identify deformation hotspot areas; the potential sliding surface spatial distribution is delineated by combining slope and aspect analysis in ArcGIS; and the InSAR deformation rate field and UAV rupture features are integrated through georegistration and multi-resolution image fusion to generate a landslide dynamic evolution spatiotemporal correlation map. This method can effectively capture the co-evolutionary pattern of landslide rupture expansion and deformation accumulation, providing high-precision technical support for early identification, boundary delineation and early warning decision-making of hidden landslides. It is applicable to geological disaster prevention and control in vegetated areas and areas with complex terrain.

[0029] like Figure 1 The diagram shown is a flowchart of the method of the present invention. Further description of specific embodiments of the present invention is provided below. The specific steps are as follows: Step 1: Landslide Location Information Extraction and Spatial Benchmark Establishment: Based on the geological conditions obtained from the on-site reconnaissance and the engineering data of bedding landslides, high-resolution satellite images of the landslide are combined with field geological survey data (including the distribution of soft interlayers and surface rupture signs) and historical engineering geological data of bedding landslides (covering the burial depth of the sliding surface, sliding direction, and historical monitoring records). Multi-temporal high-resolution optical remote sensing images (such as WorldView-3, spatial resolution ≤0.5 m) are combined with correction and preprocessing to obtain the accurate range and geographical location of the target landslide.

[0030] Step 2: Multi-source SAR data optimization and preprocessing: Sentinel-1 satellite ascent and descent data (C-band, IW mode) were selected, with a time baseline covering two complete hydrological years. At least N>25 synthetic aperture radar (SAR) image data were acquired to form a SAR image dataset T, with a spatial baseline constraint within 200m. Radiometric calibration, multi-look processing (azimuth × range: 5×1), and DEM registration (SRTM 30m DEM) were performed using the ESA SNAP platform to eliminate terrain phase contribution errors. Orbit drift was corrected using precise orbit data (POD), and phase unwrapping was performed using the minimum cost flow (MCF) algorithm, with residual phase residuals controlled within ±0.5 rad.

[0031] In steps 2 and 3, PS-InSAR and SBAS are used respectively. InSAR obtains InSAR ground deformation images of the monitored area. Specifically, it combines... Figure 2 To explain, Figure 2 This is a schematic diagram of a landslide dynamic deformation and rupture monitoring principle provided by the present invention.

[0032] InSAR deformable image acquisition and interpretation: Obtain N SAR image data to form a SAR image dataset, where N is greater than or equal to 25 and the time span is at least 2-3 years; the image data consists of SAR imaging data of the same landslide observation scene and meets the requirements of interferometry; the obtained SAR image dataset is processed using permanent scatterer and short baseline set InSAR, i.e., PS-InSAR and SBAS. InSAR processing yields InSAR landslide deformation images. Radiometric and geometric corrections are applied to the images, while the influence of atmospheric delay on the radar signal is eliminated. Interferometry is performed on the registered radar images to generate interferograms, and a phase unwrapping algorithm is used to address phase entanglement issues within the interferograms. Landslide topography and deformation signals are separated using external elevation data (DEM), and the deformation rate is calculated based on the time intervals between the acquisition of N radar images. Anomalies caused by decoherence or noise are identified and eliminated, and deformation features, such as deformation amplitude, direction, and distribution, are extracted from the interferograms. Subsequent monitoring cycles and data acquisition plans are planned based on monitoring requirements and the dynamic characteristics of the target area.

[0033] PS-InSAR solution process: ① Permanent scatterer screening: Based on the amplitude deviation index (ADI) threshold (ADI<0.25) and phase stability criterion (temporal coherence>0.75), high signal-to-noise ratio permanent scatterers (PS points) are extracted from the interferogram. ②Separate deformation phase and atmospheric delay through temporal phase modeling:

[0034] In the formula, For linear deformation rate, For time intervals, This represents DEM error.

[0035] ③ Model Solving and Optimization: Singular Value Decomposition (SVD) is used to iteratively solve the deformation rate field. ) and elevation error ( The convergence condition is that the root mean square residual (RMS) ≤ 1.5 mm / yr; ④ Atmospheric phase correction: The atmospheric delayed phase is separated based on spatiotemporal filtering (STUN algorithm), and coupled with the GACOS (Generic Atmospheric Correction Online Service) model to eliminate regional tropospheric effects; ⑥ Result verification: Through cross-verification of remote sensing image results, the spatiotemporal development process and variation law of landslide ground deformation can be obtained from long-term deformation images. Finally, a millimeter-level deformation rate field (accuracy ±2 mm / yr) is output, providing a quantitative basis for landslide dynamic evolution analysis.

[0036] The solution process for SBAS-InSAR: ① Differential Interferometry and Phase Unwrapping: The interferometric pair is processed by two-track differential processing to generate a differential interferogram. The SNAPHU algorithm (Statistical-Cost Network Flow Method) is used for phase unwrapping, and the unwrapping threshold is set to the standard deviation of the phase gradient.

[0037] ② Deformation Model Construction: Based on the assumption of time-series deformation accumulation, a joint model of linear deformation rate and elevation residual error is established.

[0038] In the formula, For the first Deformation rate over time period For time intervals, Divide the time period into totals; ③ Singular Value Decomposition and Regularized Solution: Singular value decomposition (SVD) is performed on the underdetermined system of equations, and Tikhonov regularization (smoothing factor) is introduced. The inversion of the time-series deformation rate field is converged under the condition that the root mean square residual (RMS) ≤ 2 mm / yr. ④ Atmospheric phase correction: Spatiotemporal high-pass filtering (cutoff frequency: 30 days in time domain, 1 km in spatial domain) is used to separate the atmospheric delayed phase, and MERIS atmospheric data is fused to correct the regional water vapor effect; ⑤ Result Verification and Accuracy Evaluation: Perform spatial consistency analysis with PS-InSAR solution results (RMSE≤3.5mm / yr), output temporal deformation rate field (accuracy ±2.5 mm / yr) and cumulative deformation (spatial resolution ≤50 m) (e.g. Figure 3 This provides quantitative support for the analysis of the multi-stage evolution mechanism of landslides.

[0039] Step 3: High-precision calculation of PS-InSAR landslide deformation rate field: Based on the N SAR image dataset (N≥25) obtained in step (2), a spatiotemporal baseline joint optimization strategy is adopted to select the master image (Master) with low vertical baseline (B⊥<200 m) and short time baseline (Δt<50 days) to construct a Delaunay triangulation interferometric pair network; a high-precision digital elevation model (DEM, data source is TanDEM-X 12m or ALOS World 3D-30m) of the monitoring area is integrated synchronously, and a differential interferogram is generated by the two-track differential interferometry method, and precise orbit data (POD) is introduced to correct the orbit phase error. The amplitude deviation index method (ADI<0.25) is used to screen permanent scatterers (PS points) and construct a spatiotemporal linear deformation model:

[0040] in, For deformation phase, The phase of the residual error in the DEM. For atmospheric delayed phase, Using the noise phase, the landslide surface deformation rate and corresponding monitoring results are calculated. Step 4: Landslide Temporal Deformation Field Reconstruction and Rate Interpretation Based on SBAS-InSAR: For the N SAR image dataset (N≥25) obtained in step (2), a short baseline interferometric pair network is constructed (vertical baseline threshold B⊥≤250 m, temporal baseline threshold Δt≤120 days). The Delaunay triangulation optimization strategy is used to generate the interferometric pair set. By multiplying the corresponding pixels in the registered master and slave images by conjugate, an interferogram is obtained. The SAR image data is interferometrically and unwrapped using short baseline ensemble aperture radar interferometry SBAS-InSAR to interpret the landslide ground deformation monitoring results and deformation rate. The interferogram is processed by multi-view filtering, removal of flat phase and terrain phase errors, adaptive filtering, etc., and a digital elevation model (DEM) is used for masking to eliminate overlapping and shadow areas. Then, based on the unwrapped phase, the deformation rate is calculated using the SVD method. The obtained temporal deformation rate is integrated to generate a deformation time series, thereby reconstructing the landslide temporal deformation field and providing information on the change of surface deformation over time.

[0041] Step 5: PS-InSAR Deformation Field Partitioning and Vector Transformation Based on Spatial Hotspot Analysis: The Getis-Ord Gi* spatial statistical method in hotspot analysis is used to perform significant clustering analysis on PS-InSAR deformation points, calculate the local spatial autocorrelation index (Z-score) and statistical significance (p<0.05) of each deformation point, and identify deformation hotspots (Z-score>2.58) and coldspot regions (Z-score<-2.58); combine kernel density estimation (KDE, bandwidth=100 m) to quantify the spatial distribution density of PS points, and define the threshold boundaries of dense regions (density ≥150 points / km²) and sparse regions (density <50 points / km²). An irregular triangular network (TIN) was constructed using the Delaunay triangulation algorithm, with the spatial coordinates of PS points as nodes to generate the initial grid. Significantly clustered areas were screened based on the local Moran's I index (I>0.3, p<0.01), and the TIN boundary was corrected using inverse distance weighted interpolation (power parameter=2) to extract the topological boundary between dense and sparse areas. A raster-to-vector conversion algorithm was used to convert the deformation rate raster data obtained from SBAS-InSAR in step (4) into vector surface data with attribute topology, and the PS point density zoning results were fused to generate a spatial distribution map of landslide deformation risk levels.

[0042] Step 5 employs hotspot analysis to partition the InSAR image and obtain relevant parameters of deformation points. The implementation steps are as follows: 1) Data preparation: Collect InSAR data, including multi-temporal radar images; 2) Data preprocessing: Perform radiometric correction, geometric correction, etc., to ensure data quality; 3) InSAR processing: Generate interferograms, calculate phase differences, and generate deformation maps; 4) Deformation map analysis: Identify deformation signals in the deformation map and remove non-deformation factors such as atmospheric effects and orbital errors; 5) Application of hotspot analysis: Determine the parameters for hotspot analysis, such as significance level (p-value) and hotspot intensity (e.g., Z-score). Use hotspot analysis software or tools to analyze the deformation map and identify deformation hotspot regions. The calculation of the hotspot intensity Z-score is as follows:

[0043] Where: X is the observed value, such as the number of events in a hotspot area. It is the average number of events expected under the null hypothesis. Standard deviation represents the degree of variation in the number of events.

[0044] The Getis-Ord Gi* statistic from spatial autocorrelation theory is used to quantify spatial heterogeneity of deformation:

[0045] in, , Spatial weight matrix The inverse distance squared function was used. The significance level was set to α=0.01 (Z≥2.58) to extract high-confidence deformation hot zones. The geometric accuracy of the hot zone boundary was optimized by Delaunay triangulation (node ​​spacing ≤50m).

[0046] 6) Deformation point identification: Based on the results of hotspot analysis, identify points or regions with significant deformation; 7) Deformation parameter extraction: Extract parameters such as deformation rate and deformation amount of the deformation points; 8) Spatial clustering and time series analysis: Based on the results of hotspot analysis, perform spatial clustering of the deformation points to identify deformation regions. Perform time series analysis based on the data to study the changing trend of deformation points over time.

[0047] In step 5, hotspot analysis is a semi-automatic method for extracting ground motion. Using hotspot analysis, a clustering method, to partition InSAR images helps identify high-risk areas for landslide deformation. Hotspot analysis can capture the clustering of deformation data within a certain range; the calculation formula is as follows:

[0048] In the formula, d For distance, e ij ( d () is the spatial weight matrix, representing the elements. i and j The topological relationship between them, the molecule represents the distance from the topology. i point d All within the scope x j The sum of the attribute values ​​of the elements, with the denominator being all x j The sum of the attribute values ​​of the features does not include... x j itself.

[0049] The formula for calculating the straight-line distance between two points is: ; in( x G , y G ), ( x Gn , y Gn () represents two coordinates.

[0050] After hotspot analysis, three parameters can be obtained for each deformation point: Z-score, P-value, and Gi_bin.

[0051] The Z-score represents a multiple of the standard deviation.P The value represents the probability, and Gi_bin represents the confidence interval. Among them, the Z score, ... P All values ​​can be used to measure statistical significance. It can determine whether the clustering of observed high or low values ​​is more significant than our expectation in a random distribution of the same values, and then obtain the corresponding image partitions accordingly.

[0052] Step 6: Spatial heterogeneity fusion of multi-source InSAR deformation field and full-domain monitoring results: Based on the PS-InSAR deformation hotspot partitioning results obtained in Step 5, a spatial heterogeneity fusion model is constructed, and the deformation data of dense and sparse areas are processed differently. The PS-InSAR deformation in the sparse area is converted into SBAS-InSAR vector data to obtain the final ground deformation monitoring results of the entire study area. In step 6, PS-InSAR and SBAS InSAR deformation field data fusion steps: ① Dense Area Data Optimization: For dense PS point areas (density ≥ 150 points / km²), Kriging interpolation (semi-variogram model: Gaussian type, nugget value = 0.5 mm², range = 500 m) is used to convert the deformation rate of discrete PS points into 30 m raster data, preserving local deformation details (interpolation residual standard deviation ≤ 1.0 mm / yr). ② Data compensation in sparse regions: For sparse regions of PS points (density < 50 points / km²), based on the SBAS-InSAR temporal deformation field (spatial resolution 50 m), the data is resampled to 30 m resolution using bilinear interpolation, and a spatial weight matrix is ​​constructed:

[0053] In the formula, The distance between PS points and SBAS pixels (unit: m) is used to weight and fuse PS-InSAR discrete point data to suppress data noise in low-density areas. ③ Global deformation field integration: An adaptive weighted fusion algorithm is adopted, and the weight function is defined as follows:

[0054] in, , To determine the local PS point density (points / km²), ensuring that dense areas are dominated by PS data. Sparse regions depend on SBAS data .

[0055] ④ Topology consistency verification: The spatial continuity of the fused data is verified by global Moran's I index, and the accuracy is evaluated by leave-one-out cross-validation (RMSE ≤ 2.5 mm / yr). ⑤ Standardized output of results: Generate a 30 m resolution global deformation rate vector dataset (GeoPackage format), with attribute fields including deformation rate (mm / yr), density weight (α) and 95% confidence interval, supporting multi-scale deformation mechanism analysis and dynamic risk classification of landslides.

[0056] Step 7: High-resolution surface fracturing feature acquisition by UAV: ​​Near-field, high-resolution monitoring of bedding landslide areas is conducted using multi-rotor unmanned aerial vehicle (UAV) oblique photography and airborne LiDAR sensor technology; high-density 3D point clouds (density ≥ 500 points / m²) are generated based on the Structure from Motion (SfM) algorithm and multi-view stereo matching (MVS) technology, and aerial triangulation accuracy is optimized using ground control points (GCPs); detailed surface fracturing features are automatically identified from the point cloud and orthophotos using a deep learning algorithm (U-Net architecture): ground fissures, bulges, troughs, faults, and landslide steep walls; the spatiotemporal expansion rate of surface fracturing is quantified through differential DEM and deformation vector field analysis, and a crack development index (CDI) is constructed to assess fracturing activity.

[0057] In step 7, UAV photogrammetry is used to obtain information on surface rupture of landslides and virtual reality models in the monitored area.

[0058] UAV Deformation Image Acquisition and Interpretation: UAV photography can acquire multiple images of a corresponding area, which are then stitched and fused to generate seamless high-resolution images. Digital Elevation Models (DEMs) and Digital Surface Models (DSMs) are generated from image data and point cloud data to construct a comprehensive virtual reality 3D model of the landslide (e.g., Figure 5 As shown in the figure, detailed features of landslide surface fracturing (ground fissures, mounds, troughs, faults, landslide steep walls, etc.) are extracted. A detailed surface failure analysis of the landslide is conducted to identify the spatial and temporal patterns of landslide deformation and to obtain the dynamic spatiotemporal evolution law of landslide surface fracturing in the study area.

[0059] Step 8: High-precision 3D real-scene modeling of landslides based on multi-source UAV data: Collect image data acquired by UAVs, fit the geometric features of the main sliding surface of the landslide using the Random Sample Consensus (RANSAC) algorithm, extract the landslide topographic boundary using the Alpha Shape algorithm (boundary fitting error ≤ 0.5 m), construct an initial triangulation network using the constrained Delaunay triangulation algorithm to avoid topographic distortion, dynamically densify the triangulation network density in key areas based on topographic curvature analysis, and generate a high-precision TIN model, integrate LiDAR point cloud and multispectral data to construct a multi-attribute 3D model of the landslide, with attribute fields including slope, aspect, surface rupture type, and activity level, and realize interactive visualization of the model through the WebGL platform, supporting profile analysis, volume calculation (error ≤ 3%), and deformation vector field overlay display, finally outputting a high-precision 3D real-scene model of the landslide.

[0060] Furthermore, in step 8, a high-precision TIN model (triangle side length ≤ 0.3m) is constructed, and a 1cm resolution DEM is generated using Kriging interpolation. Micro-topographic anomalies are identified through curvature field analysis. Planar curvature (Profile Curvature): Detects the orientation of tensile cracks; Plan Curvature: Identifies the direction of shear fracture propagation; The Terrain Distortion Index (TDI) is constructed by integrating slope, aspect, and surface roughness.

[0061] In the formula, For elevation gradient, This represents the standard deviation of local roughness.

[0062] In step 8, the TIN method is used to construct the DEM image. Since the elevation change before the landslide was relatively small, a grid construction method using boundary points can be used to generate the DEM. Discrete points at the boundary of the landslide monitoring area are connected using an appropriate algorithm to form a spatial triangular mesh ground model, thus generating the corresponding DEM image. The basic process of establishing an irregular triangular mesh is to first connect the three nearest discrete points to form a triangle, then connect adjacent discrete points outwards based on each side of this triangle to form another three triangles, and so on, until all discrete points are connected. UAVs are used to acquire image data of different resolutions, GPS RTK is used for field control point measurement, and high-precision DEM data is generated using 3D reconstruction and aerial triangulation methods. Finally, an irregular grid is constructed using boundary points to generate the reference surface DEM data.

[0063] In step 8, slope and aspect analysis is performed on the obtained DEM using ArcGIS. The calculation formula is:

[0064] Its Slope x for x Slope y for y The slope and aspect are defined by the direction of the slope. Slope and aspect analysis, combined with the preceding hotspot analysis, divides the study area into visible, low-coherence, and overlapping shadow areas.

[0065] In step 8, the obtained InSAR deformed image and UAV deformed image are overlaid, including the geographic coordinate registration of the landslide image and the fusion of image boundary lines.

[0066] InSAR images use the WGS84 coordinate system, which is a geocentric coordinate system; UAV images use the Beijing 54 coordinate system, which is a geocentric coordinate system. Geographic coordinate registration involves placing both types of images into the same coordinate system. This involves converting the coordinate projection of the UAV image to the coordinate system of the InSAR data. This step can be achieved using image registration tools in ArcGIS through coordinate system transformation or feature point selection. After registration, boundary lines in the images are eliminated using image fusion techniques, making the landslide image appear as a continuous and unified whole.

[0067] Step 9: Spatiotemporal coupling of multi-source remote sensing data and collaborative analysis of landslide dynamic evolution: Based on the global InSAR deformation field generated in Step 6 and the UAV three-dimensional real scene model constructed in Step 8, a multi-scale fusion framework is adopted to realize the collaborative monitoring of the entire process of landslide deformation-rupture, and a dynamic early warning mechanism is implemented for the entire process of landslide evolution, forming a closed-loop monitoring chain of data acquisition → feature interpretation → model coupling → dynamic early warning.

[0068] Further, in step 9, the InSAR deformed image and the UAV deformed image are overlaid, including geographic coordinate registration of the landslide image and fusion of image boundary lines. Geographic coordinate registration is performed using ArcGIS software or remote sensing processing software to determine a common geographic coordinate system (such as WGS84), and geometric correction is performed to ensure that the InSAR and UAV images are aligned in the same coordinate system. The image coordinates are transformed to a unified projected coordinate system, and the images are resampled to match the resolution. The boundary lines of the InSAR and UAV images are identified; these boundary lines may be terrain features or man-made boundaries. Image processing techniques, such as edge detection and image fusion, are used to smooth the boundary lines and reduce discontinuities during image overlay. Based on pixel and feature overlay, methods such as Laplacian pyramids and multi-resolution analysis are used to overlay the InSAR deformed image and the UAV deformed image. The overlaid image is analyzed to evaluate registration errors and overlay quality. The overlaid image is interpreted, deformation hotspots are identified, deformation patterns are analyzed, and the overlay results are displayed graphically, using different color schemes to enhance the visual effect.

[0069] Furthermore, in step 9, the InSAR and UAV data are converted to the same geographic coordinate system for precise georegistration. Scale matching is performed based on the spatial resolution of the InSAR and UAV data, which may require resampling of the UAV data. Macroscopic ground deformation analysis is conducted using InSAR data to identify deformation rates and patterns, while microscopic surface fracturing analysis is performed using UAV data to identify the distribution, length, and width of cracks. The macroscopic deformation and microscopic fracturing data are then fused using various fusion techniques, such as pixel-level fusion, feature-level fusion, or decision-level fusion. Deformation features, such as deformation rate, cumulative deformation, and fracturing development, are extracted from the fused data for spatial analysis to identify the spatial distribution characteristics of deformation and fracturing. If the data span is sufficient, time-series analysis is performed to study the dynamic changes of deformation and fracturing over time.

[0070] Furthermore, in step 9, the deformation-fracture coupling quantification model defines the spatiotemporal coupling index (SCI):

[0071] In the formula, For the deformation rate field, This represents the crack length field. Landslide stages are determined based on the SCI value range: SCI < 0.3: Initial creep stage (blue alert) 0.3≤SCI<0.6: Constant velocity deformation stage (yellow warning) SCI≥0.6: Accelerated destruction phase (red alert).

[0072] Furthermore, the dynamic early warning threshold is adaptively optimized by introducing a Kalman filter update mechanism and integrating real-time monitoring data to correct the early warning model parameters.

[0073] in, This represents the creep coefficient of the soil and rock mass. for t Real-time measured deformation-fracture increment The observation matrix is ​​used. A threshold dynamic adjustment function is obtained by training through a historical disaster case database (100+ landslide samples) to achieve region-specific early warning.

[0074] Step 9 involves using a multi-scale fusion framework to achieve coordinated monitoring of landslide deformation and rupture: ① Unified spatiotemporal data benchmarks: By using affine transformation and least squares matching algorithm, the InSAR deformation field is aligned with the UAV model coordinate system to the national geodetic coordinate system, with a registration residual (RMS) ≤ 0.3 pixels; Wavelet transform multiresolution analysis (decomposition level = 5, mother wavelet = Daubechies 4) was used to achieve spatial scale matching and eliminate the resolution difference between InSAR and UAV data (30 m). 0.1 m); ② Deformation-fracture correlation modeling: Constructing a spatiotemporal kriging interpolation model and fusing the InSAR deformation rate field ( V InSAR ) and UAV rupture displacement vector (D UAV ):

[0075] In the formula, , Weighting coefficients The result was determined through cross-validation using the mutability function. Granger causality test (lag order = 3, significance level p < 0.05) was used to analyze the spatiotemporal lag relationship between deformation driving and fracture response; ③ Dynamic risk assessment and visualization: Define the Landslide Activity Index (LAI):

[0076] in, , Deformation rate ( ) and crack length ( The contribution weights of ) were determined by principal component analysis (PCA); An interactive monitoring platform was developed based on the WebGL engine, integrating temporal deformation cloud maps, fracture extension animations, and LAI heatmaps.

[0077] ④ Accuracy verification and engineering output: The reliability of the fused data was cross-validated using ground-based GNSS monitoring stations (horizontal accuracy ±3 mm, elevation accuracy ±5 mm) (RMSE ≤ 4 mm / yr). Outputs a landslide dynamic monitoring report (PDF / A format), including a deformation-fracture coupling matrix, a risk level zoning map (GeoTIFF), and an early warning decision tree (XML Schema), which can be directly connected to the geological disaster management system.

[0078] The second objective of this application is to provide a landslide dynamic deformation and rupture monitoring system, comprising the following modules: The location determination module is used to integrate field geological survey data, historical engineering geological data and multi-temporal high-resolution optical remote sensing images, and then determine the accurate range and geographical location of the target landslide after correction and preprocessing. The database processing module is used to select SAR image data that meets the constraints of time and space baselines based on the accurate range and geographical location of the target landslide, and obtain a multi-source SAR database through preprocessing. The deformation model construction module is used to construct an interferometric pair network based on a multi-source SAR database using the PS-InSAR method, synchronously integrate the digital elevation model of the monitoring area, generate differential interferograms using the two-track differential interferometry method, and use the amplitude deviation index method to screen permanent scatterers in the differential interferograms to construct a spatiotemporal linear deformation model; the landslide surface deformation rate is obtained by solving the spatiotemporal linear deformation model. The deformation model construction module is used to construct a short baseline interferometric pair network based on multi-source SAR data, and generate interferograms by multiplying master and slave image pixels by conjugate. It uses SBAS-InSAR technology to complete image interferometry and phase unwrapping, interpreting surface deformation information and deformation rate. Combined with digital elevation model, the interferogram mask is processed, and the deformation rate is calculated using the SVD method based on the unwrapped phase. After integration, the deformation time series is generated, the temporal deformation field is reconstructed, and information on the change of surface deformation over time is provided. The distribution map generation module is used to perform cluster analysis and partitioning of landslide surface deformation rate based on spatial statistical methods and kernel density estimation, convert SBAS-InSAR deformation rate raster data into vector surface data, and generate a spatial distribution map of landslide deformation risk level. The monitoring results acquisition module is used to construct a spatial heterogeneous fusion model based on the spatial distribution map of landslide deformation risk level, differentiate the data of dense and sparse deformation areas, and obtain the monitoring results of ground deformation across the entire area. The overlay and fusion module is used to acquire data through UAV oblique photography and airborne LiDAR, generate 3D point clouds using the SfM algorithm, automatically identify surface fracture features using deep learning algorithms, and construct 3D reality models using constrained triangulation algorithms. The matching and early warning module is used to perform geographic registration and scale matching between the monitoring results of ground deformation across the entire region and the three-dimensional real scene model. Based on the spatiotemporal coupling coefficient and dynamically optimized early warning threshold, a closed-loop monitoring and hierarchical early warning mechanism is formed.

[0079] A third objective of this application is to provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the landslide dynamic deformation and rupture monitoring method.

[0080] A fourth objective of this application is to provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the landslide dynamic deformation and rupture monitoring method.

[0081] A fifth objective of this application is to provide a computer program product comprising computer instructions that instruct a computer to execute the landslide dynamic deformation and rupture monitoring method.

[0082] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0083] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0084] This application may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application may take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, readable storage media, optical storage, etc.) containing computer-usable program code.

[0085] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0086] Obviously, the described embodiments are only some, not all, of the embodiments in this application. All other embodiments obtained by those skilled in the art based on the embodiments in this application without inventive effort should fall within the scope of protection of this application.

[0087] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application and not to limit them. Although this application has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation methods of this application. Any modifications or equivalent substitutions that do not depart from the spirit and scope of this application should be covered within the protection scope of this application.

Claims

1. A method for monitoring dynamic deformation and rupture of landslides, characterized in that, include: By integrating field geological survey data, historical engineering geological data, and multi-temporal high-resolution optical remote sensing images, and then performing correction and preprocessing, the accurate range and geographical location of the target landslide are determined. Based on the accurate range and geographical location of the target landslide, SAR image data that meets the constraints of time and space baselines are selected and preprocessed to obtain a multi-source SAR database. Based on a multi-source SAR database, an interferometric pair network was constructed using the PS-InSAR method. A digital elevation model of the monitoring area was synchronously integrated, and a differential interferogram was generated using the two-track differential interferometry method. The amplitude deviation index method was used to screen permanent scatterers in the differential interferogram, and a spatiotemporal linear deformation model was constructed. The landslide surface deformation rate was obtained by solving the spatiotemporal linear deformation model. A short baseline interferometric pair network is constructed based on multi-source SAR data. Interferograms are generated by conjugate multiplication of master and slave image pixels. SBAS-InSAR technology is used to complete image interferometry and phase unwrapping to interpret surface deformation information and deformation rate. The interferogram is masked by combining digital elevation model. Based on the unwrapped phase, the deformation rate is calculated by SVD method. After integration, a deformation time series is generated, and the time-series deformation field is reconstructed to provide information on the change of surface deformation over time. Based on spatial statistical methods and kernel density estimation, cluster analysis and partitioning of landslide surface deformation rate are performed, and SBAS-InSAR deformation rate raster data are converted into vector surface data to generate a spatial distribution map of landslide deformation risk level. Based on the spatial distribution map of landslide deformation risk level, a spatial heterogeneous fusion model is constructed to differentiate the data of dense and sparse deformation areas and obtain the monitoring results of ground deformation across the entire region. Data was acquired through UAV oblique photography and airborne LiDAR, and 3D point clouds were generated using the SfM algorithm. Deep learning algorithms were combined to automatically identify surface fracture features, and a constrained triangulation algorithm was used to construct a 3D real-world model. The results of the global ground deformation monitoring are georeferenced and scale-matched with the 3D reality model. Based on the spatiotemporal coupling coefficient and dynamically optimized early warning threshold, a closed-loop monitoring and hierarchical early warning mechanism is formed.

2. The method for monitoring dynamic deformation and rupture of landslides according to claim 1, characterized in that, Based on the accurate extent and geographical location of the target landslide, SAR image data that meets the constraints of temporal and spatial baselines is selected and preprocessed to obtain a multi-source SAR database, including: Based on the accurate range and geographical location of the target landslide, satellite ascent and descent data are selected, the time baseline covers at least one complete hydrological year, and at least N>25 synthetic aperture radar (SAR) image data are acquired to form a SAR image data set, so that the spatial baseline is constrained within the set range. For SAR image datasets, terrain phase contribution errors are eliminated; then, orbital drift is corrected using precise orbital data, and phase unwrapping is performed using a minimum cost flow algorithm, so that the residual phase error is controlled within a set range.

3. The method for monitoring dynamic deformation and rupture of landslides according to claim 1, characterized in that, The spatiotemporal linear deformation model is constructed as follows: in, For deformation phase, The phase of the residual error in the DEM. For atmospheric delayed phase, This is the noise phase.

4. The method for monitoring dynamic deformation and rupture of landslides according to claim 1, characterized in that, The method, based on spatial statistical methods and kernel density estimation, performs cluster analysis and partitioning of landslide surface deformation rates, converts SBAS-InSAR deformation rate raster data into vector surface data, and generates a spatial distribution map of landslide deformation risk levels, including: Spatial statistical methods from hotspot analysis are used to perform significant cluster analysis on PS-InSAR deformation points, calculate the local spatial autocorrelation index and statistical significance of each deformation point, and identify deformation hotspots and coldspots. Kernel density estimation is combined to quantify the spatial distribution density of PS points and define the threshold boundaries between dense and sparse regions. An irregular triangular mesh is constructed using a triangulation algorithm, with the spatial coordinates of PS points as nodes to generate the initial mesh. Significantly clustered spatial regions are selected, and the boundary of the triangular mesh is corrected using inverse distance weighted interpolation to extract the topological boundary between dense and sparse regions. A raster-to-vector conversion algorithm is used to convert the deformation rate raster data obtained by SBAS-InSAR into vector surface data with attribute topology, and the PS point density zoning results are fused to generate a spatial distribution map of landslide deformation risk level.

5. The method for monitoring dynamic deformation and rupture of landslides according to claim 4, characterized in that, The method employs spatial statistical methods from hotspot analysis to perform significant clustering analysis on PS-InSAR deformation points, calculates the local spatial autocorrelation index and statistical significance of each deformation point, and identifies deformation hotspots and coldspots, including: Collect InSAR data, including multi-temporal radar imagery; InSAR processing involves generating interferograms, calculating phase differences, and generating deformation maps. Identify deformation signals in the deformation diagram and eliminate non-deformation factors; Perform hotspot analysis on the deformation map to identify deformation hotspot regions; Spatial autocorrelation theory statistics are used to quantify the spatial heterogeneity of deformation, extract high-confidence deformation hot zones, and optimize the geometric accuracy of hot zone boundaries through triangulation to obtain the results of hotspot analysis; Based on the results of hotspot analysis, identify points or regions with significant deformation; Extract the deformation rate and deformation amount of deformation points or regions; perform spatial clustering of the deformation rate and deformation amount of deformation points or regions to identify deformation areas; perform time series analysis based on the data to obtain the changing trend of deformation points over time.

6. The method for monitoring dynamic deformation and rupture of landslides according to claim 4, characterized in that, The irregular triangular mesh is constructed by the triangulation algorithm, and the initial mesh is generated using the spatial coordinates of PS points as nodes. Select areas with significant spatial clustering, correct the triangular mesh boundaries using inverse distance weighted interpolation, and extract the topological boundaries between dense and sparse regions, including: DEM is generated by constructing a grid using boundary points. The ground model with a spatial triangular mesh structure is connected by discrete points at the boundary of the landslide monitoring area using a corresponding algorithm to generate the corresponding DEM image. The basic process of establishing an irregular triangular network is to first connect the three nearest discrete points to form a triangle, and then connect the nearest discrete points outward based on each side of the triangle to form another three triangles, and so on, until all discrete points are connected. UAVs were used to acquire image data of different resolutions, GPS RTK was used to measure field control points, and high-precision DEM data was generated using 3D reconstruction and aerial triangulation. Then, an irregular grid was constructed using boundary points to generate reference surface DEM data.

7. A method for monitoring dynamic deformation and rupture of landslides according to claim 1, characterized in that, The process involves georeferencing and scale matching the results of global ground deformation monitoring with a 3D reality model, and forming a closed-loop monitoring and tiered early warning mechanism based on the spatiotemporal coupling coefficient and dynamically optimized early warning thresholds. This includes: Perform geographic coordinate registration to determine a common geographic coordinate system, and perform geometric correction to ensure that InSAR and UAV images are aligned in the same coordinate system; The image coordinates are transformed to a unified projected coordinate system, and the image is resampled to match the resolution; the boundary lines of InSAR and UAV images are identified, and image processing techniques are used to smooth the boundary lines; Based on pixel and feature overlay, InSAR deformed images and UAV deformed images are overlaid, the overlaid image is analyzed, the registration error and overlay quality are evaluated, the overlaid image is interpreted, deformation hotspots are identified, deformation patterns are analyzed, and the overlay results are displayed graphically.

8. The method for monitoring dynamic deformation and rupture of landslides according to claim 1, characterized in that, The early warning threshold based on the spatiotemporal coupling coefficient and dynamic optimization includes: The dynamic early warning threshold is adaptively optimized by introducing a Kalman filter update mechanism and fusing real-time monitoring data to correct the early warning model parameters. in, The creep coefficient of the soil and rock mass. for t Real-time measured deformation-fracture increment The observation matrix; The threshold dynamic adjustment function is obtained by training with a historical disaster case database, and the early warning threshold is obtained.

9. The method for monitoring dynamic deformation and rupture of landslides according to claim 1, characterized in that, The pixel- and feature-based overlay method superimposes InSAR deformed images and UAV deformed images, analyzes the superimposed image, evaluates registration error and overlay quality, interprets the superimposed image, identifies deformation hotspots, analyzes deformation patterns, and displays the overlay results graphically, including: Construct a spatiotemporal kriging interpolation model that integrates the InSAR deformation rate field and the UAV rupture displacement vector: In the formula, , Weighting coefficients ; Granger causality test was used to analyze the spatiotemporal lag relationship between deformation-driven and fracture response; Calculate the Landslide Activity Index (LAI): in, , Deformation rate With crack length The contribution weights are determined through principal component analysis; An interactive monitoring platform was developed based on the WebGL engine, integrating time-series deformation cloud maps, fracture propagation animations, and LAI heatmaps. The reliability of the fused data was cross-verified using ground-based GNSS monitoring stations; Output a dynamic monitoring report of landslides, including a deformation-fracture coupling matrix, a risk level zoning map, and an early warning decision tree.

10. A landslide dynamic deformation and rupture monitoring system, characterized in that, include: The location determination module is used to integrate field geological survey data, historical engineering geological data and multi-temporal high-resolution optical remote sensing images, and then determine the accurate range and geographical location of the target landslide after correction and preprocessing. The database processing module is used to select SAR image data that meets the constraints of time and space baselines based on the accurate range and geographical location of the target landslide, and obtain a multi-source SAR database through preprocessing. The deformation model construction module is used to construct an interferometric pair network based on a multi-source SAR database using the PS-InSAR method, synchronously integrate the digital elevation model of the monitoring area, generate differential interferograms using the two-track differential interferometry method, and use the amplitude deviation index method to screen permanent scatterers in the differential interferograms to construct a spatiotemporal linear deformation model; the landslide surface deformation rate is obtained by solving the spatiotemporal linear deformation model. The deformation model construction module is used to construct a short baseline interferometric pair network based on multi-source SAR data, and generate interferograms by multiplying master and slave image pixels by conjugate. It uses SBAS-InSAR technology to complete image interferometry and phase unwrapping, interpreting surface deformation information and deformation rate. Combined with digital elevation model, the interferogram mask is processed, and the deformation rate is calculated using the SVD method based on the unwrapped phase. After integration, the deformation time series is generated, the temporal deformation field is reconstructed, and information on the change of surface deformation over time is provided. The distribution map generation module is used to perform cluster analysis and partitioning of landslide surface deformation rate based on spatial statistical methods and kernel density estimation, convert SBAS-InSAR deformation rate raster data into vector surface data, and generate a spatial distribution map of landslide deformation risk level. The monitoring results acquisition module is used to construct a spatial heterogeneous fusion model based on the spatial distribution map of landslide deformation risk level, differentiate the data of dense and sparse deformation areas, and obtain the monitoring results of ground deformation across the entire area. The overlay and fusion module is used to acquire data through UAV oblique photography and airborne LiDAR, generate 3D point clouds using the SfM algorithm, automatically identify surface fracture features using deep learning algorithms, and construct 3D reality models using constrained triangulation algorithms. The matching and early warning module is used to perform geographic registration and scale matching between the monitoring results of ground deformation across the entire region and the three-dimensional real scene model. Based on the spatiotemporal coupling coefficient and dynamically optimized early warning threshold, a closed-loop monitoring and hierarchical early warning mechanism is formed.

Images