A landfill pile stability dynamic risk early warning method based on multi-temporal remote sensing

Abstract

The application provides a landfill heap stability dynamic risk early warning method based on multi-temporal remote sensing, and belongs to the technical field of waste landfill surface deformation monitoring, geological disaster risk early warning and engineering management. The method uses multi-period high-precision digital elevation models, adopts an improved iterative nearest point algorithm with rigid reference constraint to realize point cloud registration, and constructs a three-dimensional displacement vector field; the uneven settlement and hidden crack risk area are automatically identified through settlement gradient and displacement divergence. The elevation change, displacement anomaly and external environmental factors are integrated, a weighted superposition model is used to quantitatively evaluate the slope instability risk index, and a hierarchical early warning spatial distribution map is generated. The method realizes high-precision, automatic and dynamic spatial monitoring and risk early warning of the deformation process of the landfill heap, and significantly improves the safety management efficiency of the landfill.

Description

Technical Field

[0001] This invention belongs to the field of landfill surface deformation monitoring, geological disaster risk early warning and engineering management technology, and more specifically relates to a dynamic risk early warning method for landfill stability based on multi-temporal remote sensing. Background Technology

[0002] As a crucial terminal facility for urban solid waste treatment, landfills are subject to continuous and dynamic changes in their topography, internal stress, and surrounding hydrogeological environment. Uneven settlement of the landfill can lead to significant safety and environmental risks such as slope instability and rupture of the anti-seepage system. Traditional landfill monitoring relies primarily on manual inspections, single-point settlement observations, and periodic topographic surveys, which suffer from low efficiency, limited coverage, and delayed early warning. In recent years, UAV remote sensing technology (including oblique photography and lidar) has been applied to landfill topographic mapping, enabling efficient acquisition of high-precision surface models for purposes such as capacity calculation and slope analysis. However, existing technological solutions mostly focus on static 3D reconstruction and visualization at a single time point, essentially providing a "snapshot of the current state."

[0003] Current technologies have significant limitations, such as the inability to quantitatively extract the three-dimensional deformation field (including vertical settlement and horizontal displacement) of the pile surface from multiple data periods, and the inability to conduct stability trend analysis and early warning based on deformation patterns. There is a lack of direct and intelligent links between existing monitoring results (such as DOM and DSM) and management decisions (when to reinforce slopes and where to focus on seepage prevention) driven by quantitative risk models.

[0004] Therefore, developing an integrated method that can fuse multi-temporal remote sensing data to achieve quantitative early warning of stability changes from landfill surface monitoring has become an urgent technical requirement for the intelligent and refined management of landfills. Summary of the Invention

[0005] This invention aims to solve the technical problems of existing landfill stability monitoring and early warning methods, such as insensitivity to dynamic deformation processes, difficulty in automatically identifying potential danger zones, insufficient spatiotemporal resolution, reliance on manual interpretation for risk warnings, and lack of spatial visualization of warning information. It aims to achieve high-precision, automated, and dynamic spatial identification and graded early warning of landfill settlement, deformation, and slope instability risks, thereby effectively improving the safety assurance capability of landfill operation.

[0006] To achieve the above objectives, the present invention employs the following technical solution: the method comprises: Obtain high-precision digital elevation models of the target landfill at at least two different time points; Differential calculations are performed on the high-precision digital elevation models of two adjacent time periods to generate an elevation change field on the surface of the landfill. Rigid infrastructure around the landfill is extracted as a reference feature. An improved iterative nearest point (ICP) algorithm with rigid feature constraints is used for point cloud registration to construct a three-dimensional displacement vector field. Extract the settlement gradient of the elevation change field and the displacement divergence of the three-dimensional displacement vector field, and combine them with a preset settlement rate threshold to automatically identify and mark potential uneven settlement areas or hidden crack risk areas. By integrating the elevation change field, the three-dimensional displacement vector field, and external induced environmental data, a weighted superposition analysis model is used to calculate the slope instability risk index of each region of the waste pile, and a graded early warning spatial distribution map is generated based on the index range.

[0007] In one scheme, the digital elevation model includes: solving the lidar echo signal, using multiple echo signals to distinguish the land surface and vegetation, and combining waveform decomposition algorithm to optimize land feature identification; performing voxel-based downsampling, using statistical outlier detection to remove abnormal elevation points to achieve noise removal, and then converting the three-dimensional point cloud into a digital land surface model and a digital elevation model. Spatial interpolation is used to generate a rasterized elevation model from point clouds, thus completing the numerical expression of the elevation function over the entire region. Radiometric and geometric corrections are performed on the images, and full waveform analysis is used to improve the accuracy of ground point classification. Multi-sensor data fusion and beam triangulation are used to improve the stereo positioning accuracy of the digital elevation model. Data consistency checks and benchmark elevation stability tests are performed on the digital elevation models at all time points to ensure the spatiotemporal integrity and high accuracy of the digital elevation models.

[0008] In one scheme, the construction of the three-dimensional displacement vector field includes: performing differential calculations on two adjacent high-precision digital elevation models in time series to generate an elevation change field on the surface of the landfill, and extracting the rigid infrastructure around the landfill as reference features. The rigid infrastructure includes surrounding roads, retaining walls, and office buildings. The reference features are extracted from the point cloud using spatial analysis, shape index method, geometric template matching, and maximum flat support domain analysis.

[0009] In one approach, the uneven settlement area or hidden crack risk zone is identified by extracting the settlement gradient from the elevation change field, using the central difference to calculate the spatial first derivative of the elevation change to obtain the gradient expression, and combining edge detection technology to identify settlement fracture and slope change zone areas. Based on the three-dimensional displacement vector field, the displacement divergence at each spatial location is calculated. The spatial derivative extraction method based on the Sobel operator or the Prewitt operator is used to distinguish the local aggregation, cracking tendency and collapse trend of the landfill.

[0010] In one scheme, the calculation of the slope instability risk index of each area of ​​the landfill includes: using a weighted linear superposition analysis model, setting the weight parameters of multiple factors, constructing a comprehensive risk index, and quantitatively calculating the slope instability risk index of each area based on the normalized values ​​and spatial distribution of each factor, so as to achieve a spatiotemporal resolution assessment of the instability risk of the landfill. The obtained risk index is classified according to a preset threshold, and a corresponding graded early warning spatial distribution map is generated. The risk range can be divided into grades using the zone method to form a color-banded visualization layer covering the entire area, and linked to the engineering management platform through the geographic information system interface.

[0011] In one scheme, the construction of the three-dimensional displacement vector field includes: using an improved iterative nearest point algorithm that introduces rigid feature constraints, and achieving high-precision spatial alignment of the two time point clusters through the registration constraints of rigid reference features; in this registration process, rigid infrastructure acts as anchor points and participates in error minimization optimization with high weight, thereby improving registration accuracy and stability. After registration, the point clouds from the two phases are subjected to point-by-point vector difference in three-dimensional space to construct a three-dimensional displacement vector field of the landfill surface. Vertical and horizontal components are obtained for each spatial location, which truly reflects the dynamic evolution process of the landfill. In one approach, preset settlement rate and divergence thresholds are set to automatically screen out areas where settlement rate, gradient amplitude, or divergence abnormally exceeds limits. Abnormal or high-risk points are automatically marked and partitioned. For the detected high-risk areas, connectivity analysis, morphological spatial filtering, and other methods are used to remove isolated abnormal points, thereby enhancing the continuity and engineering applicability of the final area identification.

[0012] Beneficial effects of this invention: (1) It has leaped from the traditional “single-phase static mapping” to “multi-phase dynamic analysis”. In the monitoring of landfills, a three-dimensional displacement vector field and quantitative risk index model have been systematically constructed and applied, which enables stability monitoring to have predictive and early warning capabilities.

[0013] (2) The risk warning map and warning information generated by this method are intuitive, quantitative and spatial management tools that can be applied to the daily management of landfills, significantly improving the scientific and intelligent level of landfill safety and environmental management. Attached Figure Description

[0014] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0015] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Typical embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0016] Unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. To facilitate understanding, the invention will now be described more fully with reference to the accompanying drawings. Typical embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to make the disclosure of the invention more thorough and complete.

[0017] like Figure 1 As shown, a dynamic risk early warning method for landfill stability based on multi-temporal remote sensing is proposed. It includes the following steps: S1. Obtain a high-precision digital elevation model of the target landfill at at least two different time points.

[0018] To meet the demand for high precision, mainstream remote sensing data sources include airborne or ground-based LiDAR and high-resolution UAV photogrammetry. LiDAR, with its high spatial resolution and vertical accuracy, is widely used in landfill topography reconstruction. After selecting a suitable remote sensing platform, flight paths must be planned to ensure coverage of the landfill area, and repeated flights or observations at multiple time points must be conducted to ensure that the time intervals reflect the landfill's evolution characteristics. After data acquisition, for the LiDAR point cloud, the echo signals are first processed, using multiple echo signals to distinguish between the land surface and vegetation, and then waveform decomposition algorithms are used to optimize feature identification.

[0019] In the data processing stage, to eliminate noise and outliers, voxel grid-based downsampling needs to be introduced, and statistical outlier detection methods are used for abnormal elevation points. Points with outliers exceeding a threshold are removed. The noise-removed 3D point cloud can then be converted into a Digital Surface Model (DSM) or a Digital Elevation Model (DEM). To generate a rasterized elevation model from the point cloud, spatial interpolation methods (Kriging interpolation, inverse distance-weighted IDW) are used. In this process, the discrete point cloud is mapped to a regular raster through spatial interpolation, yielding the elevation function. Numerical representation over the entire region. The interpolation model can be expressed as:

[0020] in, The elevation of the target location is... Given the elevation value of point i, By assigning weights to each known point relative to the target location, this method is better able to reveal terrain details in areas with dense spatial data.

[0021] The quality of a high-precision DEM (Digital Image Model) is determined not only by the accuracy of the sensor itself but also by external interferences such as illumination, cloud cover, and changes in ground features. Therefore, imagery requires radiometric and geometric correction. For the raw waveform returned by lidar, echo data can be analyzed into ground points, low-lying vegetation points, and tall-lying vegetation points. Full waveform analysis methods can further improve the accuracy of ground point classification. For example, a Gaussian mixture model can be used to fit the waveform energy distribution, clustering different ground feature categories, with elevation separation measured by the change from peak to trough.

[0022] After data generation, the DEMs of all time phases undergo data consistency checks, and the elevation stability of the benchmark points is examined to ensure that the monitoring of changes across the entire stack is meaningful. In some cases, multi-sensor data fusion is introduced, such as co-processing airborne LiDAR and multi-view aerial imagery, using collinearity equations to improve the stereo positioning accuracy of the DEM. The mathematical principle is based on the projection equation:

[0023] in, For the actual three-dimensional coordinates of the ground features, For pixel coordinates, focal length, These are the coordinates of the spatial control points. Rigorous geometric constraints from multi-temporal, multi-source remote sensing data ensure the spatiotemporal integrity and high accuracy of the DEM data.

[0024] All elevation models represent the three-dimensional surface conditions of the landfill at different points in time. Through the above system process, each elevation model has sufficient spatial resolution and vertical accuracy to support subsequent analysis of landfill settlement, deformation and dynamic risk.

[0025] S2. Perform differential calculation on the high-precision digital elevation models of two adjacent time periods to generate the elevation change field of the landfill surface. Extract the rigid infrastructure around the landfill (such as surrounding roads, retaining walls, office buildings, etc.) as reference features. Use the improved iterative nearest point (ICP) algorithm with rigid feature constraints to perform point cloud registration and construct a three-dimensional displacement vector field.

[0026] Differential calculations are performed on two consecutive high-precision digital elevation models to reveal the dynamic changes in the surface of the landfill. Each acquired elevation model is represented as a raster point cloud or a regular grid, and can be used with discrete functions. and These represent the elevations of a location during the two measurement periods. The landfill topography has undergone irreversible changes between the two observations; therefore, pixel-level difference calculations are performed directly on the DEM. This yields the overall surface elevation change field, which can be used to assess the spatial distribution of settlement, uplift, or subsidence. However, monitoring is inevitably affected by external disturbances, observation errors, or equipment drift. Simple difference analysis often fails to accurately reflect the true elevation changes, necessitating higher requirements for spatiotemporal registration.

[0027] Before differencing, the quality of registration directly determines the reliability of the difference values. To eliminate spatial mismatches caused by non-deformation factors in the landfill as a whole, region, or locality, rigid infrastructure must be introduced as baseline features. These rigid structures, such as surrounding roads, retaining walls, and office buildings, should have constant or highly stable coordinate positions in the elevation models of both periods, and are therefore identified as anchor points for registration. Through spatial analysis, these target areas are identified and segmented. Using methods such as shape index method, geometric template matching, and maximum flat support domain analysis, rigid features in the point cloud are extracted from the dynamic regions of the landfill. The projections in data from different periods can serve as the basis for registration.

[0028] Point cloud registration after introducing rigid reference features requires high-precision alignment of point cloud data from two different periods. Here, the Iterative Closest Point (ICP) algorithm is employed, with the addition of rigid feature constraints for improvement. The improved ICP aims to minimize the registration error between rigid intervals between point cloud sets from different periods (referred to as S and T). Its mathematical model is as follows:

[0029] in, For rotation matrix, It is a translation vector. Register corresponding points for ordinary point clouds. , Here are the coordinates of the extracted rigid datum features in the two-phase point clouds. Weighting factors are used to strengthen rigid feature registration constraints. For regular point pairs, rotation and translation are updated in each iteration to minimize the overall point pair error; while for rigid feature point pairs, they participate in ICP optimization with higher weights to ensure that the reference anchor points strictly coincide, thereby minimizing the overall rigid body drift caused by observation system errors and environmental changes.

[0030] Each iteration of the ICP algorithm includes nearest neighbor matching, registration transformation parameter estimation, error accumulation calculation, and termination / convergence determination. Nearest neighbor matching is accelerated using a kd-tree data structure. After the matching point set is determined, the rotation is solved using the Singular Value Decomposition (SVD) method. Peaceful relocation For the error constraint term of the rigid datum feature, this is equivalent to increasing the KLambda norm, strengthening the role of the rigid anchor point in minimizing the overall error. Iterative updates continue until the error converges.

[0031]

[0032] After improved ICP registration, the point cloud pairs exhibit strict overlap in the rigid foundation area, while the elevation changes in dynamic regions accurately reflect the positional shifts caused by the settlement, collapse, and uplift of the waste pile. Following registration, point-by-point vector difference is performed between the two point clouds in three-dimensional space to construct a three-dimensional displacement vector field. For each corresponding point in the survey area, three-dimensional displacement components can be obtained. ,in Describe vertical changes, Describes horizontal displacement. The vector field of the entire point cloud. The microscopic process of dynamic evolution of landfill bodies is depicted.

[0033] The elevation change field (i.e., the change in the vertical component) and the three-dimensional displacement field (including both horizontal and vertical components) together reveal where significant settlement, lateral movement, or latent slippage occurred within the landfill. Visualization techniques such as contour rendering and interpolated heatmaps were used to map the differential and registered petrified results into a spatial distribution map, revealing the dynamic information of each pixel and point. This significantly improved the sensitivity for identifying early anomalies, accelerated settlement, or localized structural damage within the landfill.

[0034] S3. Extract the settlement gradient of the elevation change field and the displacement divergence of the three-dimensional displacement vector field, and combine them with a preset settlement rate threshold to automatically identify and mark potential uneven settlement areas or hidden crack risk areas.

[0035] The elevation change field records the local undulations on the overlying surface, while the three-dimensional displacement vector field reveals the full spatial picture of this local motion. To accurately identify abnormal settlement and potential cracking hazards, it is essential to uncover the physical mechanisms underlying the elevation and displacement data. Using mathematical field theory methods, the elevation change field... It can not only show the increase or decrease of amplitude, but also reflect slopes, deformation transition zones, and potential fault locations using the spatial first derivative (gradient). For the raster expression, the settlement gradient can be approximated by the directional derivative as follows:

[0036] The numerical implementation uses central difference:

[0037]

[0038] The magnitude of the gradient The degree of abruptness of surface changes is given. Gradient extreme zones often coincide with important geological structures such as subsidence faults and slope change zones; therefore, gradient field analysis and edge detection of the elevation difference matrix help delineate risk areas. Simultaneously, high-gradient regions also require the integration of a three-dimensional displacement field to capture complex deformation signals such as lateral shear and contraction.

[0039] The three-dimensional displacement vector field is obtained through point-to-point difference, and at each spatial location there is To quantify regional volumetric change trends—especially to identify precursory effects of localized accumulation, cracking, or bulging in landfills—the divergence of the displacement field can be calculated:

[0040] In large-scale applications on the Earth's surface, the principal contribution takes the form of two-dimensional divergence (dz term omitted):

[0041] Two-dimensional meshes can use the Sobel or Prewitt operator, for The spatial derivatives of each field are calculated and summed. Positive divergence indicates localized tension and soil sparsity, while negative divergence suggests unit volume aggregation or collapse. Compared to simple vertical settlement deformation, three-dimensional divergence is more sensitive to latent shear and multi-directional strain, which is crucial for the complex mechanical response and early warning spatial zoning of landfills. High divergence values ​​in the displacement field often coincide with precursors of cracks, surface tension, and landfill fragmentation.

[0042] In the section on marking and automatically identifying risk areas, a preset settlement rate threshold is used. and divergence threshold This serves as a threshold for screening and grading. For each elevation grid cell and colored location, if the settlement rate is monitored continuously for a given period...

[0043] Points exceeding empirically or statistically defined critical values ​​(such as those corresponding to structural failure or environmental tolerance limits) are considered settlement anomalies. Boundary thresholds are simultaneously established for both the gradient modulus and the divergence field.

[0044] Locations that meet any of the above conditions are automatically marked as high-risk sub-regions.

[0045] In practice, uneven settlement in landfills often presents as "band-like" or "cluster-like" distributions, and isolated high values ​​are often considered as anomalies and eliminated. Therefore, using methods such as 8-connectivity region screening or morphological dilation-erosion algorithms, spatial filtering can be performed on high-risk points in the initial screening, making the final identification results more meaningful for engineering applications.

[0046] S4. Integrate the elevation change field, the three-dimensional displacement vector field, and external induced environmental data, calculate the slope instability risk index of each area of ​​the waste pile through a weighted superposition analysis model, and generate a graded early warning spatial distribution map based on the index range.

[0047] Spatial resolution assessment of landfill slope instability risk relies on the organic integration of elevation change fields, three-dimensional displacement vector fields, and externally induced environmental data. By constructing a weighted and superimposed comprehensive risk index, quantitative and hierarchical dynamic early warning can be achieved. Elevation change field Reflecting the subsidence and uplift trends of different blocks, the three-dimensional displacement vector field The system provides the directionality and magnitude of surface movements, while externally induced environmental data encompasses driving variables such as rainfall, groundwater dynamics, landfill gas, leachate pressure, and landfill operation intensity. These data, in raster or vector form, are spatially matched with the landfill elevation model to ensure all factors involved in risk calculations share the same spatial reference standard. Each data field undergoes normalization and dimensional adjustment to ensure its impact on overall risk can be directly superimposed without distortion due to inconsistencies in dimensions.

[0048] To reflect the different contributions of each participating factor, a weighted linear combination (WLC) model is used to establish the risk index of the spatial cell. The risk assessment model can be set as follows:

[0049] in, For the weights of different factors, Each item For the normalized input variables, This represents all environmental factors (such as rainfall, landfill operations, etc.). Elevation changes. Directly related to the vertical deformation rate, reflecting the effects of rapid settlement, accumulation, or uplift; three-dimensional displacement magnitude. Reflects the rate of local surface movement, indicating sedimentation, slippage, or rifting phenomena; divergence This highlights the nature of volumetric changes, such as criteria for in-plane tensile cracking or localized aggregation. External inducing factors are mainly based on monitored quantities, and are reflected by expert experience, discriminant analysis, or machine learning weight back calculation to determine the driving force of slope instability under optimal configuration.

[0050] Weight The determination of the factors can be achieved using methods such as Analytic Hierarchy Process (AHP), fuzzy comprehensive evaluation, entropy weighting, or reverse training based on historical instability incidents, aiming to quantitatively reflect the relative contribution of each factor to instability. Weighted results The index represents the slope instability potential of each spatial unit, physically representing the probability intensity of an instability event occurring in that area. The spatial distribution of the index is directly implemented on the GIS platform, using regular raster assignment or triangular mesh interpolation to achieve spatial mapping of the complete field. For high-risk areas, the index... It will be significantly higher than the surrounding area, highlighting it as a region of abrupt changes in the coupling of subsidence, slippage, and loosening.

[0051] The generation of the tiered early warning chart is based on the common probabilistic failure or decision threshold zoning method, which divides the index range into risk levels, such as...

[0052] in, Risk levels can be determined through accident statistical analysis, empirical comparison, or extreme value theory. The graded risk distribution map is immediately converted into a color-coded spatiotemporal visualization layer, covering the entire landfill area. Through a GIS interface, it is linked to the engineering management system to issue real-time warnings or scheduling tasks for different risk levels.

[0053] Example: I. Implementation Steps 1. Data Acquisition In April and July 2024, two phases of UAV-borne LiDAR aerial surveys were conducted on the target landfill to obtain three-dimensional point cloud data covering the entire area.

[0054] 2. Digital Elevation Model (DEM) Construction Using full-waveform point cloud data, two high-precision DEMs of different time phases were generated through waveform calculation, multi-echo separation, voxel lattice noise reduction, and spatial interpolation. The data accuracy was verified by actual measurements at ground benchmarks, with a horizontal RMSE better than 0.15 m and an elevation RMSE better than 0.08 m.

[0055] 3. Point cloud registration Approximately 20 rigid structures, such as roads and retaining walls around the landfill, were extracted. An improved ICP algorithm, which incorporates rigid references, was used to achieve high-precision spatial alignment of the point clouds from the two phases.

[0056] 4. Calculation of Elevation Variation Field and Three-Dimensional Displacement Field The registered DEM is subjected to pixel-by-pixel subtraction to obtain the elevation change field; at the same time, a three-dimensional displacement vector field is constructed to extract the vertical and horizontal components of each point in space.

[0057] 5. Automatic detection of abnormal areas Set a settlement rate threshold (e.g., 10 cm / 3 months) to identify and mark areas of abnormal settlement; calculate displacement divergence to screen for high-risk areas of displacement accumulation and tensile cracking.

[0058] 6. Risk Index Assessment and Early Warning Chart Generation By integrating environmental factors such as elevation changes, displacement anomalies, and rainfall (average daily rainfall of 62 mm during the same period), a slope instability risk classification map is generated through comprehensive weighted analysis.

[0059] II. Key Data Example Table

[0060] This method detected a continuous settlement zone on the south slope of the landfill, with a settlement rate of -23.7 cm / 3 months. The local settlement gradient was significant, and the risk index exceeded 0.8, classifying it as a Level IV high-risk area. Combined with continuous heavy rainfall, a spatial risk distribution map was generated, providing a scientific basis for landfill engineering inspection and reinforcement.

[0061] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0062] It should be understood that the above detailed description of the technical solutions of the present invention with reference to preferred embodiments is illustrative and not restrictive. Those skilled in the art can modify the technical solutions described in the embodiments or make equivalent substitutions for some of the technical features based on reading this specification; however, these modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for dynamic risk early warning of landfill stability based on multi-temporal remote sensing, characterized in that: The method includes: Obtain high-precision digital elevation models of the target landfill at at least two different time points; Differential calculations are performed on the high-precision digital elevation models of two adjacent time periods to generate an elevation change field on the surface of the landfill. Rigid infrastructure around the landfill is extracted as a reference feature. An improved iterative nearest point (ICP) algorithm with rigid feature constraints is used for point cloud registration to construct a three-dimensional displacement vector field. Extract the settlement gradient of the elevation change field and the displacement divergence of the three-dimensional displacement vector field, and combine them with a preset settlement rate threshold to automatically identify and mark potential uneven settlement areas or hidden crack risk areas. By integrating the elevation change field, the three-dimensional displacement vector field, and external induced environmental data, a weighted superposition analysis model is used to calculate the slope instability risk index of each region of the waste pile, and a graded early warning spatial distribution map is generated based on the index range.

2. The method for dynamic risk early warning of landfill stability based on multi-temporal remote sensing according to claim 1, characterized in that: The digital elevation model includes: solving the lidar echo signal, using multiple echo signals to distinguish the land surface and vegetation, and combining waveform decomposition algorithm to optimize land feature identification; performing voxel-based downsampling, using statistical outlier detection to remove abnormal elevation points to achieve noise removal, and then converting the three-dimensional point cloud into a digital land surface model and a digital elevation model. Spatial interpolation is used to generate a rasterized elevation model from point clouds, thus completing the numerical expression of the elevation function over the entire region. Radiometric and geometric corrections are performed on the images, and full waveform analysis is used to improve the accuracy of ground point classification. Multi-sensor data fusion and beam triangulation are used to improve the stereo positioning accuracy of the digital elevation model. Data consistency checks and benchmark elevation stability tests are performed on the digital elevation models at all time points to ensure the spatiotemporal integrity and high accuracy of the digital elevation models.

3. The method for dynamic risk early warning of landfill stability based on multi-temporal remote sensing according to claim 1, characterized in that: The construction of the three-dimensional displacement vector field includes: performing differential calculations on two adjacent high-precision digital elevation models in time series to generate an elevation change field on the surface of the landfill, and extracting the rigid infrastructure around the landfill as reference features. The rigid infrastructure includes surrounding roads, retaining walls, and office buildings. The reference features are extracted from the point cloud using spatial analysis, shape index method, geometric template matching, and maximum flat support domain analysis.

4. The method for dynamic risk early warning of landfill stability based on multi-temporal remote sensing according to claim 1, characterized in that: The aforementioned uneven settlement area or hidden crack risk area is identified by extracting the settlement gradient from the elevation change field, calculating the spatial first derivative of the elevation change using the central difference, and combining edge detection technology to identify settlement fracture and slope change zone areas. Based on the three-dimensional displacement vector field, the displacement divergence at each spatial location is calculated. The spatial derivative extraction method based on the Sobel operator or the Prewitt operator is used to distinguish the local aggregation, cracking tendency and collapse trend of the landfill.

5. The method for dynamic risk early warning of landfill stability based on multi-temporal remote sensing according to claim 1, characterized in that: The calculation of the slope instability risk index of each region of the landfill includes: using a weighted linear superposition analysis model, setting the weight parameters of multiple factors, constructing a comprehensive risk index, and quantitatively calculating the slope instability risk index of each region based on the normalized values ​​and spatial distribution of each factor, so as to achieve a spatiotemporal resolution assessment of the instability risk of the landfill. The obtained risk index is classified according to a preset threshold, and a corresponding graded early warning spatial distribution map is generated. The risk range can be divided into grades using the zone method to form a color-banded visualization layer covering the entire area, and linked to the engineering management platform through the geographic information system interface.

6. The method for dynamic risk early warning of landfill stability based on multi-temporal remote sensing according to claim 3, characterized in that: The construction of the three-dimensional displacement vector field includes: using an improved iterative nearest point algorithm that introduces rigid feature constraints, and achieving high-precision spatial alignment of the two time point clusters through the registration constraints of rigid reference features; in this registration process, rigid infrastructure acts as anchor points and participates in error minimization optimization with high weight, thereby improving registration accuracy and stability; After registration, the point clouds of the two phases are vector-differentiated point by point in three-dimensional space to construct a three-dimensional displacement vector field of the landfill surface. Vertical and horizontal components are obtained at each spatial location, which truly reflects the dynamic evolution process of the landfill.

7. The method for dynamic risk early warning of landfill stability based on multi-temporal remote sensing according to claim 4, characterized in that: The system sets preset settlement rate and divergence thresholds to automatically screen out areas where settlement rate, gradient amplitude, or divergence abnormally exceeds limits. It then automatically marks and identifies abnormal or high-risk points by region. Furthermore, it employs connectivity analysis and morphological spatial filtering methods to remove isolated abnormal points from the detected high-risk areas, thereby enhancing the continuity and engineering applicability of the final region identification.

Images