Landslide susceptibility evaluation method, device and equipment based on InSAR and medium

Abstract

The invention provides a landslide susceptibility evaluation method, device, equipment and medium based on InSAR, and the method comprises the steps: obtaining landslide basic data, extracting evaluation index data, analyzing the multicollinearity among the evaluation index data, and constructing a landslide susceptibility evaluation index system; selecting all vector points in a disaster body range of the landslide record library as landslide positive samples, selecting non-landslide negative samples, and constructing training set and test set samples according to the landslide positive samples and the non-landslide negative samples; based on the random forest model and the RF-InSAR model of the sample after the SBAS-InSAR optimization, constructing a landslide susceptibility evaluation model; and training a landslide susceptibility evaluation model by using the training set and the test set samples, and inputting the to-be-analyzed data into the landslide susceptibility evaluation model to evaluate the landslide susceptibility, so as to solve the problem of accurately and reliably evaluating the landslide susceptibility under the condition of comprehensively considering various influence factors.

Description

Technical Field

[0001] The present invention relates to the technical field of geological disaster risk assessment, and in particular to a landslide susceptibility assessment method, device, equipment and medium based on InSAR. Background Art

[0002] With the continuous expansion of human activities and the continuous changes in the natural environment, the problem of landslide susceptibility has become increasingly prominent and has become an important topic of concern. Landslide is a common geological disaster, usually caused by the combined action of multiple factors. With global climate change, extreme rainfall events occur frequently, and a large amount of rainwater seeps into the ground, increasing the water content of the rock and soil. On the one hand, it leads to an increase in the weight of the rock and soil, and on the other hand, it reduces the shear strength of the rock and soil, thereby greatly increasing the susceptibility of landslides. Therefore, in-depth research on the susceptibility of landslides and accurate assessment of the possibility of landslides in different regions are of vital importance for formulating scientific and reasonable disaster prevention and mitigation measures, protecting people’s lives and property, and promoting social stability and development.

[0003] Traditional landslide susceptibility assessment methods mainly rely on geological surveys, remote sensing image interpretation and statistical analysis, which have the problems of strong subjectivity, discontinuous data, and susceptibility to human factors. These methods often fail to fully consider the combined effects of multiple factors such as geology, topography, and climate, resulting in limited accuracy and reliability of landslide susceptibility assessment results.

[0004] Synthetic Aperture Radar (InSAR) has the characteristics of all-weather, all-time, and high-precision, and can monitor small surface deformations. In landslide monitoring, InSAR technology can provide high-resolution surface deformation data, help identify potential landslide bodies and monitor landslide activities, and provide a new data source and technical support for landslide susceptibility assessment. By combining InSAR technology to optimize sample selection, the potential risk of landslides can be more accurately assessed, and early warning and response capabilities can be improved.

[0005] Therefore, there is an urgent need to propose a landslide susceptibility evaluation method, device, equipment and medium based on InSAR to solve the technical problem of accurately and reliably evaluating the susceptibility of landslides while comprehensively considering various influencing factors. Summary of the invention

[0006] In order to overcome the problems existing in the related art, the present disclosure provides a landslide susceptibility evaluation method, device, equipment and medium based on InSAR to solve the technical problem of accurately and reliably evaluating the susceptibility of landslides while comprehensively considering various influencing factors in the related art.

[0007] One or more embodiments of this specification provide a landslide susceptibility assessment method based on InSAR, comprising the following steps:

[0008] Acquiring basic landslide data, extracting evaluation index data from the basic landslide data, analyzing multicollinearity among the evaluation index data, and constructing a landslide susceptibility evaluation index system;

[0009] All vector points within the hazard body range of the landslide catalog library are selected as landslide positive samples, and vector points within a preset range outside the landslide are selected as negative samples, and the negative samples are optimized by the SBAS-InSAR method based on ground deformation information to obtain non-landslide negative samples, and training set and test set samples are constructed according to the landslide positive samples and the non-landslide negative samples;

[0010] Based on the random forest model and the RF-InSAR model of the sample optimized by SBAS-InSAR, the landslide basic data and the landslide susceptibility evaluation index system are combined to construct a landslide susceptibility evaluation model;

[0011] The landslide susceptibility evaluation model is trained using the training set and the test set samples, and the data to be analyzed is input into the landslide susceptibility evaluation model for landslide susceptibility evaluation.

[0012] Preferably, the basic landslide data include landslide catalog data, various evaluation index data and Sentinel-1B sentinel data.

[0013] Preferably, the analysis of the multicollinearity among the evaluation index data specifically comprises the following steps:

[0014] The multicollinearity between the evaluation index data is analyzed by calculating VIF and TOL, the multicollinearity between the evaluation index data is ranked, and the evaluation index data with VIF less than a first threshold or TOL greater than a second threshold is used to construct a landslide susceptibility evaluation index system.

[0015] Preferably, the selecting of data output by the SBAS-InSAR method based on ground deformation information as non-landslide negative samples specifically comprises the following steps:

[0016] SBAS-InSAR obtains surface deformation information through image registration, differential interferogram generation, phase unwrapping, orbit refinement and re-leveling, elevation reconstruction and geocoding processing;

[0017] The data in the surface deformation information is converted into WGS-84-UTM-zone-48N projection coordinate system data, the raster layer of the evaluation factor is extracted and obtained, and the landslide catalog data is superimposed to obtain a non-landslide negative sample.

[0018] Preferably, the method further comprises cross-validating each combination of hyperparameter values ​​in the grid of the landslide susceptibility evaluation model using a grid search-based method, calculating the evaluation index, and selecting the group with the highest accuracy as the hyperparameter of the optimization model.

[0019] One or more embodiments of this specification provide a landslide susceptibility assessment device based on InSAR, including an indicator system construction module, a sample construction module, an assessment model construction module and an assessment module;

[0020] The indicator system construction module is used to obtain basic landslide data, extract evaluation indicator data from the basic landslide data, analyze the multicollinearity between the evaluation indicator data, and construct a landslide susceptibility evaluation indicator system;

[0021] The sample construction module is used to select all vector points within the hazard body range of the landslide catalog library as landslide positive samples, select vector points within a preset range outside the landslide as negative samples, and optimize the negative samples based on the SBAS-InSAR method of ground deformation information to obtain non-landslide negative samples, and construct training set and test set samples according to the landslide positive samples and the non-landslide negative samples;

[0022] The evaluation model building module is used to build a landslide susceptibility evaluation model based on the random forest model and the RF-InSAR model of the sample optimized by SBAS-InSAR, combined with the landslide basic data and the landslide susceptibility evaluation index system;

[0023] The evaluation module is used to train the landslide susceptibility evaluation model using the training set and the test set samples, and input the data to be analyzed into the landslide susceptibility evaluation model to perform landslide susceptibility evaluation.

[0024] Preferably, the basic landslide data include landslide catalog data, various evaluation index data and Sentinel-1B sentinel data.

[0025] Preferably, the indicator system construction module is further configured as follows:

[0026] The multicollinearity between the evaluation index data is analyzed by calculating VIF and TOL, the multicollinearity between the evaluation index data is ranked, and the evaluation index data with VIF less than a first threshold or TOL greater than a second threshold is used to construct a landslide susceptibility evaluation index system.

[0027] One or more embodiments of the present specification provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned InSAR-based landslide susceptibility assessment method when executing the computer program.

[0028] One or more embodiments of the present specification provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-mentioned landslide susceptibility assessment method based on InSAR.

[0029] The present invention provides a landslide susceptibility evaluation method, device, equipment and medium based on InSAR, which has the advantages of providing reliable data support for the evaluation work by obtaining basic landslide data as basic data for landslide susceptibility evaluation, extracting evaluation index data from the basic landslide data, analyzing the multicollinearity between the evaluation index data, and constructing a landslide susceptibility evaluation index system, which can ensure that each evaluation index data has a certain relative independence, and avoid the possible high correlation that leads to a decrease in model accuracy; selecting all vector points within the disaster body range of the landslide catalog library as landslide positive samples, selecting vector points within a preset range outside the landslide as negative samples, and optimizing the negative samples based on the SBAS-InSAR method of ground deformation information to obtain non-landslide negative samples, and according to the landslide The positive samples and the non-landslide negative samples are used to construct training set and test set samples, and the interference of geological disaster potential points is eliminated by selecting the landslide positive samples and the non-landslide negative samples; the RF-InSAR model of the samples optimized by the random forest model and SBAS-InSAR is combined with the basic landslide data and the landslide susceptibility evaluation index system to construct a landslide susceptibility evaluation model, and the reconstruction of the digital elevation model is completed by using INSAR to make the RF model susceptibility evaluation more accurate; the landslide susceptibility evaluation model is trained by using the training set and test set samples, and the data to be analyzed is input into the landslide susceptibility evaluation model for landslide susceptibility evaluation, and the accuracy and reliability of landslide susceptibility evaluation are improved by combining InSAR technology with machine learning methods, and the landslide susceptibility evaluation can be widely used in landslide monitoring and early warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate one or more embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0031] Figure 1 A schematic flow chart of a landslide susceptibility assessment method based on InSAR provided for one or more embodiments of this specification;

[0032] Figure 2 The impact factor provided for one or more embodiments of this specification, Figure 2 (a) is the elevation, Figure 2 (b) is the slope direction, Figure 2 (c) is the slope, Figure 2 (d) is the plane curvature, Figure 2 (e) is NDVI, Figure 2 (f) is the lithology, Figure 2 (g) is the average rainfall over many years, Figure 2 (h) is the distance from the river, Figure 2 (i) is the distance from the road, Figure 2 (j) is the distance from the fault;

[0033] Figure 3 A schematic diagram of the structure of a landslide susceptibility assessment device based on InSAR provided for one or more embodiments of this specification;

[0034] Figure 4 A schematic diagram of the structure of a computer device provided for one or more embodiments of this specification. DETAILED DESCRIPTION

[0035] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below in conjunction with the drawings in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the protection scope of the present invention.

[0036] The present invention is described in detail below in conjunction with specific implementation methods and the accompanying drawings.

[0037] Method Embodiment

[0038] According to an embodiment of the present invention, a landslide susceptibility assessment method based on InSAR (Interferometric Synthetic Aperture Radar) is provided. Figure 1 FIG. 2 is a flow chart of a landslide susceptibility assessment method based on InSAR provided in this embodiment. The landslide susceptibility assessment method based on InSAR according to an embodiment of the present invention comprises the following steps:

[0039] S110, monitor the landslide and obtain basic landslide data, including landslide catalog data, various evaluation index data and Sentinel-1B sentinel data. Specifically, the landslide catalog data is mainly established through visual interpretation based on Google Earth images. Various evaluation index data are derived from the 30m resolution DEM (Digital Elevation Model) of ASTERGDEM (Advanced Spaceborne Thermal Emission and Reflection Radiometer Global Digital Elevation Model) provided by the Geospatial Data Cloud, and four types of evaluation index data of slope, aspect, curvature and elevation are generated in ArcGIS10.8. Evaluation index data are extracted from the basic landslide data, the multicollinearity between the evaluation index data is analyzed, and a landslide susceptibility evaluation index system is constructed. The evaluation index data include evaluation indicators of topography, geological conditions, meteorology and hydrology, vegetation and human activities, among which topography includes: slope, slope aspect, plane curvature and elevation; geological conditions include distance from faults and lithology; meteorology and hydrology include distance from rivers and average rainfall over many years; vegetation and human activities include NDVI (Normalized Difference Vegetation Index) and distance from roads.

[0040] The lithology and fault data of geological conditions were obtained from the 1:200,000 geological map provided by the Geological Cloud of the China Geological Survey, and the meteorological and hydrological data were obtained by interpolating the rainfall data provided by the National Meteorological Information Center to obtain the rainfall distribution map of the study area.

[0041] Road data and river data were obtained with the help of the 91 Satellite Map Assistant platform, Landsat8OLI-TIRS satellite remote sensing data were obtained from the geospatial data cloud, and the normalized vegetation index was calculated in ENVI5.3 software.

[0042] S120, selecting all vector points within the hazard body range of the landslide catalog library as landslide positive samples, selecting vector points within a preset range outside the landslide as negative samples, and optimizing the negative samples based on the SBAS-InSAR method of ground deformation information to obtain non-landslide negative samples, and constructing training set and test set samples according to the landslide positive samples and the non-landslide negative samples.

[0043] S130. A landslide susceptibility evaluation model is constructed based on a random forest model and an RF-InSAR (Random Forest-InSAR) model of samples optimized by a SBAS-InSAR (Small Baseline Subset Interferometric Synthetic Aperture Radar) model, in combination with the basic landslide data and the landslide susceptibility evaluation index system. The parameters used are the Gini Impurity (GI) index, and the accuracy of the model is verified by using a ROC curve (Receiver Operating Characteristic curve) and an AUC (Area Under the Curve) value. In the parameter optimization stage, a grid search-based method is used to cross-validate each combination of hyperparameter values ​​in the grid, calculate the evaluation index, and select the combination with the highest accuracy as the hyperparameter of the optimization model.

[0044] S140, using the training set and the test set samples to train the landslide susceptibility assessment model, and inputting the data to be analyzed into the landslide susceptibility assessment model to perform landslide susceptibility assessment, and drawing a landslide disaster susceptibility zoning map.

[0045] It also includes S150, using a grid search-based method to cross-validate each combination of hyperparameter values ​​in the grid of the landslide susceptibility evaluation model, calculating the evaluation index, and selecting the combination with the highest accuracy as the hyperparameter of the optimization model.

[0046] The method provided in this embodiment provides reliable data support for the evaluation work by acquiring basic landslide data as basic data for landslide susceptibility evaluation, extracting evaluation index data from the basic landslide data, analyzing the multicollinearity between the evaluation index data, and constructing a landslide susceptibility evaluation index system, which can ensure that each evaluation index data has a certain relative independence, and avoid the possible high correlation that leads to a decrease in model accuracy; select all vector points within the disaster body range of the landslide catalog library as landslide positive samples, select vector points within a preset range outside the landslide as negative samples, and optimize the negative samples based on the SBAS-InSAR method of ground deformation information to obtain non-landslide negative samples, and construct a training set based on the landslide positive samples and the non-landslide negative samples. The training set and the test set samples are used to eliminate the interference of geological disaster potential points by selecting the landslide positive samples and the non-landslide negative samples; the landslide susceptibility evaluation model is constructed based on the RF-InSAR model of the samples optimized by the random forest model and SBAS-InSAR, combined with the basic landslide data and the landslide susceptibility evaluation index system, and the reconstruction of the digital elevation model is completed by INSAR to make the RF model susceptibility evaluation more accurate; the landslide susceptibility evaluation model is trained by using the training set and the test set samples, and the data to be analyzed is input into the landslide susceptibility evaluation model for landslide susceptibility evaluation. By combining the InSAR technology with the machine learning method, the accuracy and reliability of the landslide susceptibility evaluation are improved, and the landslide susceptibility evaluation can be widely used in landslide monitoring and early warning.

[0047] In one embodiment, the analyzing the multicollinearity among the evaluation index data specifically comprises the following steps:

[0048] The multicollinearity between the evaluation index data is analyzed by calculating VIF (Variance Inflation Factor) and TOL (Tolerance). The evaluation index data include five types of evaluation indicators: topography, geological conditions, meteorology and hydrology, vegetation and human activities. The multicollinearity between the evaluation index data is ranked, and the evaluation index data with VIF less than the first threshold or TOL greater than the second threshold is used to construct a landslide susceptibility evaluation index system, wherein the multicollinearity of the evaluation index data is related to the correlation. A low degree of multicollinearity indicates that the correlation between the factors is weak, and vice versa. VIF and TOL are inversely proportional to each other. When VIF<10 or TOL>0.1, it indicates that the factor does not have multicollinearity and can be used for modeling. The formula is:

[0049]

[0050] Where: is the sample determination coefficient, and its size determines the explanatory power of the explanatory variable on the dependent variable, that is, the model's ability to fit the sample. SSR stands for the regression sum of squares, which is used to measure the impact of changes in the independent variable on the dependent variable; SSE stands for the residual sum of squares, which represents the impact of random errors on the dependent variable; and the total sum of squares of deviations SST stands for the sum of the impacts caused by SSR and SSE. If the correlation between factor Xi and other factors is higher, then The closer it is to 1, the larger the corresponding VIF value is and the smaller the TOL value is.

[0051] Table 1 Multicollinearity analysis of evaluation index data

[0052]

[0053] The method provided in this embodiment can ensure a certain degree of relative independence between various evaluation index data by constructing a sample set and calculating the multicollinearity between evaluation index data, thereby avoiding the possible high correlation that may lead to a reduction in model accuracy.

[0054] In one embodiment, S120, selecting vector points within a preset range outside the landslide as negative samples, and optimizing the negative samples based on the SBAS-InSAR method of ground deformation information to obtain non-landslide negative samples, specifically includes the following steps:

[0055] The basic principle of S1201 and SBAS-InSAR is to use the phase information of radar waves to measure the small displacement changes between two observations. Since the time baseline and space baseline between SAR image pairs are small, the coherence of the interference results is high, and the surface deformation information can be extracted more accurately. The surface deformation information is obtained through the processing flow of image registration, differential interferogram generation, phase unwrapping, orbit refinement and re-leveling, elevation reconstruction and geocoding.

[0056] SAR imaging systems mainly include airborne SAR, ground-based SAR and satellite-based SAR. The interferometric measurement mode of the satellite-based SAR system above 600 kilometers above the surface is repeated orbit interferometric measurement, which can simultaneously record the phase information related to the distance between the ground and the radar and the backscattering intensity information of ground objects. Figure 2 As shown, it is an impact factor diagram provided by this embodiment, Figure 2 (a) is the elevation, Figure 2 (b) is the slope direction, Figure 2 (c) is the slope, Figure 2 (d) is the plane curvature, Figure 2 (e) is NDVI, Figure 2 (f) is the lithology, Figure 2 (g) is the average rainfall over many years, Figure 2 (h) is the distance from the river, Figure 2(i) is the distance from the road, Figure 2 (j) is the distance from the fault. The interval between the two image acquisition orbits is several meters, which is called the spatial baseline. Projecting the spatial baseline, the component parallel to the radar line of sight direction is called the parallel baseline, and the component perpendicular to the line of sight direction is called the vertical baseline. The time interval between the two imaging is called the temporal baseline.

[0057] Assume that the radar sensor passes through the ground observation point P twice, and S1 and S2 respectively represent the positions of the radar at the observation point P at different times. B is the spatial baseline, representing the spatial distance between S1 and S2, B⊥ is the vertical baseline, and B∥ is the parallel baseline.

[0058] α is the angle between the spatial baseline and the horizontal line, and θ is the incident angle of the radar sensor. H is the vertical height from the ground when the sensor first images, and h is the elevation of point P relative to the ground. The distances of the radar from the ground observation point P at positions S1 and S2 are r and r + Δr respectively. The phase values in the echo information collected by the sensor at S1 and S2 are respectively:

[0059]

[0060] In the formula, λ is the wavelength of the radar. The phase difference obtained from the two observations of point P is:

[0061]

[0062] That is

[0063]

[0064] Among them, because B << r, so B∥ can be used instead, that is:

[0065] Δr≈B∥ = B×sin(θ - α);

[0066] Therefore, it can be obtained that:

[0067]

[0068] h = H - r×cosθ;

[0069] Calculate the elevation h of the ground.

[0070] Using the SBAS-InSAR method, select the interferometric image pairs with smaller temporal and spatial baselines for differential interferometric multi-look processing, select the high-coherence points based on the coherence threshold, perform phase unwrapping on these points, calculate the cumulative deformation value at the moment of image acquisition using singular value decomposition and the least squares method, and then remove the atmospheric influence through filtering processing in the spatial and temporal domains to obtain the final deformation result.

[0071] In synthetic aperture radar (InSAR) data processing, the purpose of image registration is to ensure the consistency of the relative positions of the same ground objects in the SAR image. This process involves the determination of control points, the establishment of geometric deformation models, image resampling and the output of registered images, including the following steps:

[0072] Image registration includes the determination of control points, the establishment of geometric deformation models, image resampling and the output of registered images.

[0073] The image pairs are interferometrically processed to generate an interferogram, and then flattened, filtered, and phase-unwrapped differential interferograms are generated.

[0074] Phase unwrapping is used to further reduce the influence of noise and calculate the pixel coherence coefficient, and the true phase relationship between pixels is restored from the entangled phase.

[0075] The linear interference phase caused by no elevation change on the surface is removed by orbit refinement and re-leveling, which simplifies the complexity of phase unwrapping and improves the inversion accuracy of DEM.

[0076] Elevation reconstruction and geocoding.

[0077] After image registration, the image pairs are interferometrically processed to generate an interferogram, and then flattened, filtered, and phase unwrapped to prepare for subsequent processing steps. Phase unwrapping is a key step in restoring the true phase of the interferogram from the principal value or phase difference. This process further reduces the influence of noise and calculates the pixel coherence coefficient, and restores the true phase relationship between pixels from the entangled phase. After phase unwrapping is completed, a digital elevation model (DEM) can be reconstructed based on InSAR principles, imaging geometry, and orbital parameters.

[0078] S1202, converting the data in the surface deformation information into WGS-84-UTM-zone-48N projection coordinate system data, extracting and obtaining the grid layer of the evaluation factor, superimposing the landslide catalog data, and obtaining a non-landslide negative sample, specifically, through geocoding, that is, converting the slant distance coordinate system to a reference mapping coordinate system (such as WGS84), completing the spatial positioning of the elevation information, and facilitating analysis in GIS. The purpose of removing the flattening effect is to remove the image of the terrain phase by setting up GCP points, remove the linear change interference phase generated on the interference pattern due to the lack of elevation change on the surface, simplify the complexity of phase unwrapping, and thus improve the inversion accuracy of the DEM.

[0079] By calculating the root mean square error of the annual average deformation rate, the coherent points in the range of -5 to 5 mm / yr are set as stable areas. The number of coherent points in this range is 490,975, accounting for 55.8% of the total number of coherent points in the study area. Considering that most deformation areas do not have landslide morphological characteristics and cannot be collected by the landslide catalog library, and such areas are not suitable as negative sample sampling areas, the stable area range is used as the sampling area of ​​negative samples in subsequent analysis to exclude the interference of geological disaster risk points.

[0080] A RF traditional machine learning model was constructed based on random sampling of the entire area. At the same time, a corresponding coupling model was constructed based on the InSAR optimized samples. A landslide susceptibility zoning study was completed in the study area, and the results were divided into five prone areas: very low, low, medium, high, and very high using the natural breakpoint method for RF and RF-InSAR susceptibility mapping.

[0081] Table 2 Landslide susceptibility zoning results of traditional machine learning model and InSAR-based optimization model

[0082]

[0083]

[0084] Based on the optimized sample construction of InSAR results, it can be seen that after the sample is optimized by the ground deformation rate results, the prediction performance of the model has been significantly improved, and the AUC value is 0.8021, indicating that it has good prediction performance.

[0085] Device Embodiment

[0086] According to an embodiment of the present invention, a landslide susceptibility assessment device based on InSAR is provided. Figure 3 As shown, it is a schematic diagram of the structure of the landslide susceptibility assessment device based on InSAR provided in this embodiment. According to the landslide susceptibility assessment device based on InSAR in this embodiment of the present invention, it includes an indicator system construction module 31, a sample construction module 32, an evaluation model construction module 33 and an evaluation module 34.

[0087] The indicator system construction module 31 is used to obtain basic landslide data, extract evaluation indicator data from the basic landslide data, analyze the multicollinearity between the evaluation indicator data, and construct a landslide susceptibility evaluation indicator system. The basic landslide data includes landslide cataloging data, various evaluation indicator data and Sentinel-1B sentinel data.

[0088] The sample construction module 32 is used to select all vector points within the disaster body range of the landslide catalog library as landslide positive samples, select vector points within a preset range outside the landslide as negative samples, and optimize the negative samples based on the SBAS-InSAR method of ground deformation information to obtain non-landslide negative samples, and construct training set and test set samples according to the landslide positive samples and the non-landslide negative samples.

[0089] The evaluation model building module 33 is used to build a landslide susceptibility evaluation model based on the random forest model and the RF-InSAR model of the sample optimized by SBAS-InSAR, combined with the basic landslide data and the landslide susceptibility evaluation index system.

[0090] The evaluation module 34 is used to train the landslide susceptibility evaluation model using the training set and the test set samples, and input the data to be analyzed into the landslide susceptibility evaluation model to perform landslide susceptibility evaluation.

[0091] The device provided in this embodiment, the index system construction module 31 obtains basic landslide data as basic data for landslide susceptibility evaluation, provides reliable data support for the evaluation work, extracts evaluation index data from the basic landslide data, analyzes the multicollinearity between the evaluation index data, and constructs a landslide susceptibility evaluation index system, which can ensure that each evaluation index data has a certain relative independence, and avoids the possible high correlation that leads to a decrease in model accuracy; the sample construction module 32 selects all vector points within the disaster body range of the landslide catalog library as landslide positive samples, selects vector points within a preset range outside the landslide as negative samples, and optimizes the negative samples based on the SBAS-InSAR method of ground deformation information to obtain non-landslide negative samples, and constructs a landslide positive sample and a non-landslide negative sample according to the landslide positive sample and the non-landslide negative sample. The training set and the test set samples eliminate the interference of geological disaster risk points by selecting the landslide positive samples and the non-landslide negative samples; the evaluation model construction module 33 builds a landslide susceptibility evaluation model based on the random forest model and the RF-InSAR model of the samples optimized by SBAS-InSAR, combined with the basic landslide data and the landslide susceptibility evaluation index system, and uses INSAR to complete the reconstruction of the digital elevation model to make the RF model susceptibility evaluation more accurate; the evaluation module 34 uses the training set and the test set samples to train the landslide susceptibility evaluation model, and inputs the data to be analyzed into the landslide susceptibility evaluation model for landslide susceptibility evaluation. By combining InSAR technology with machine learning methods, the accuracy and reliability of landslide susceptibility evaluation are improved, and it can be widely used in landslide monitoring and early warning.

[0092] In one embodiment, the indicator system construction module 31 is further configured as follows:

[0093] The multicollinearity between the evaluation index data is analyzed by calculating VIF and TOL, the multicollinearity between the evaluation index data is ranked, and the evaluation index data with VIF less than a first threshold or TOL greater than a second threshold is used to construct a landslide susceptibility evaluation index system.

[0094] The device provided in this embodiment can ensure that each evaluation index data has a certain degree of relative independence by constructing a sample set and calculating the multicollinearity between the evaluation index data, thereby avoiding the possible high correlation that may lead to a reduction in model accuracy.

[0095] The embodiment of the present invention is an apparatus embodiment corresponding to the above-mentioned method embodiment. The specific operations of the processing steps of each module can be understood by referring to the description of the method embodiment, which will not be repeated here.

[0096] like Figure 4 As shown, the present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the landslide susceptibility assessment method in the above embodiment is implemented, or when the computer program is executed by a processor, the landslide susceptibility assessment method based on InSAR in the above embodiment is implemented.

[0097] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0098] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0099] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and the contents not described in detail in the specification of the present invention belong to the common knowledge of those skilled in the art.

Claims

1. A landslide susceptibility assessment method based on InSAR, characterized in that: The following steps are involved: Acquiring basic landslide data, extracting evaluation index data from the basic landslide data, analyzing multicollinearity among the evaluation index data, and constructing a landslide susceptibility evaluation index system; All vector points within the hazard body range of the landslide catalog library are selected as landslide positive samples, and vector points within a preset range outside the landslide are selected as negative samples, and the negative samples are optimized by the SBAS-InSAR method based on ground deformation information to obtain non-landslide negative samples, and training set and test set samples are constructed according to the landslide positive samples and the non-landslide negative samples; Based on the random forest model and the RF-InSAR model of the sample optimized by SBAS-InSAR, the landslide basic data and the landslide susceptibility evaluation index system are combined to construct a landslide susceptibility evaluation model; The landslide susceptibility evaluation model is trained using the training set and the test set samples, and the data to be analyzed is input into the landslide susceptibility evaluation model for landslide susceptibility evaluation.

2. The landslide susceptibility assessment method according to claim 1, characterized in that: The basic landslide data include landslide catalog data, various evaluation index data and Sentinel-1B data.

3. The landslide susceptibility assessment method according to claim 1, characterized in that: The analysis of the multicollinearity between the evaluation index data specifically includes the following steps: The multicollinearity between the evaluation index data is analyzed by calculating VIF and TOL, the multicollinearity between the evaluation index data is ranked, and the evaluation index data with VIF less than a first threshold or TOL greater than a second threshold is used to construct a landslide susceptibility evaluation index system.

4. The landslide susceptibility assessment method according to claim 1 or 2, characterized in that: The method of selecting vector points within a preset range outside the landslide as negative samples and optimizing the negative samples based on the SBAS-InSAR method of ground deformation information to obtain non-landslide negative samples specifically includes the following steps: SBAS-InSAR obtains surface deformation information through image registration, differential interferogram generation, phase unwrapping, orbit refinement and re-leveling, elevation reconstruction and geocoding processing; The data in the surface deformation information is converted into WGS-84-UTM-zone-48N projection coordinate system data, the raster layer of the evaluation factor is extracted and obtained, and the landslide catalog data is superimposed to obtain a non-landslide negative sample.

5. The landslide susceptibility assessment method according to claim 1, characterized in that: It also includes the use of a grid search-based method to cross-validate each combination of hyperparameter values ​​in the grid of the landslide susceptibility assessment model, calculate the evaluation index, and select the group with the highest accuracy and use it as the hyperparameter of the optimized model.

6. A landslide susceptibility assessment device based on InSAR, characterized in that: It includes indicator system construction module, sample construction module, evaluation model construction module and evaluation module; The indicator system construction module is used to obtain basic landslide data, extract evaluation indicator data from the basic landslide data, analyze the multicollinearity between the evaluation indicator data, and construct a landslide susceptibility evaluation indicator system; The sample construction module is used to select all vector points within the hazard body range of the landslide catalog library as landslide positive samples, select vector points within a preset range outside the landslide as negative samples, and optimize the negative samples based on the SBAS-InSAR method of ground deformation information to obtain non-landslide negative samples, and construct training set and test set samples according to the landslide positive samples and the non-landslide negative samples; The evaluation model building module is used to build a landslide susceptibility evaluation model based on the random forest model and the RF-InSAR model of the sample optimized by SBAS-InSAR, combined with the landslide basic data and the landslide susceptibility evaluation index system; The evaluation module is used to train the landslide susceptibility evaluation model using the training set and the test set samples, and input the data to be analyzed into the landslide susceptibility evaluation model to perform landslide susceptibility evaluation.

7. The landslide susceptibility assessment device according to claim 5, characterized in that: The basic landslide data include landslide catalog data, various evaluation index data and Sentinel-1B data.

8. The landslide susceptibility assessment device according to claim 5, characterized in that: The indicator system construction module is also configured as: The multicollinearity between the evaluation index data is analyzed by calculating VIF and TOL, the multicollinearity between the evaluation index data is ranked, and the evaluation index data with VIF less than a first threshold or TOL greater than a second threshold is used to construct a landslide susceptibility evaluation index system.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the landslide susceptibility assessment method based on InSAR as described in any one of claims 1 to 5 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the landslide susceptibility assessment method based on InSAR as claimed in any one of claims 1 to 5 are implemented.

Images