Landslide deformation mode identification method based on InSAR multi-dimensional observation

Abstract

The invention discloses a landslide deformation mode recognition method based on InSAR multi-dimensional observation, and belongs to the technical field of geological disaster dynamic recognition and monitoring, in the method, a two-dimensional deformation decomposition formula can convert InSAR observation data of different incident angles into actual deformation rates of the earth surface in the horizontal direction and the vertical direction. By introducing topographic information and incident angle parameters, sight line displacement is converted into displacement in the horizontal direction and the vertical direction, so that the real deformation characteristics of the landslide are reflected more accurately, the thickness of a soil landslide mass is inverted by using a mass conservation equation, and the landslide form is obtained by combining the topographic information. The shape and distribution characteristics of the sliding surface can be further analyzed, and more comprehensive information is provided for landslide mode recognition. The landslide deformation mode can be judged more accurately by integrating the landslide shape and the two-dimensional displacement distribution characteristic inversion landslide mode, so that a more reliable basis is provided for landslide early warning and disaster prevention and control, and the landslide disaster risk is effectively reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of dynamic identification and monitoring of geological disasters, and particularly relates to a landslide deformation pattern recognition method based on InSAR multi-dimensional observation. Background Technique

[0002] Landslide deformation pattern recognition refers to a technical method that analyzes the monitoring data of the landslide body and its surrounding environment to identify the deformation characteristics and laws before the occurrence of the landslide, so as to provide a scientific basis for landslide early warning and prevention. Landslide deformation patterns usually include multiple aspects such as the geometric shape, movement characteristics, deformation rate, and deformation stage of the landslide. The main methods for identifying landslide deformation patterns are: field investigation, remote sensing technology, ground survey, GPS positioning, InSAR technology, acoustic detection, etc. Accurately identifying the movement characteristics and deformation patterns of active landslides is of great significance for geological disaster prevention and control work. For landslide deformation pattern recognition, the existing main technical means include field geological investigation, GNSS monitoring, geophysical exploration, InSAR deformation calculation, etc. GNSS monitoring uses satellite positioning means to continuously observe the surface or deep deformation of a single point, and the observation data is relatively accurate.

[0003] However, the monitoring data of a single point cannot obtain the overall surface deformation information of the landslide. Increasing the monitoring points is less economical. Field geological investigation identifies macroscopic deformation characteristics by manually traversing the landslide body, and then identifies the landslide deformation pattern. In the implementation process of this method, the geological experience of the participants is required to be relatively high. In addition, it cannot obtain deep deformation characteristics and can only be used as an auxiliary verification means for landslide pattern recognition. Summary of the Invention

[0004] The purpose of the present invention is to provide a landslide deformation pattern recognition method based on InSAR multi-dimensional observation in order to solve the problems mentioned above.

[0005] The technical solution adopted by the present invention is as follows: A landslide deformation pattern recognition method based on InSAR multi-dimensional observation, which includes the following steps:

[0006] S1: Prepare data, collect radar remote sensing data covering the research area, including radar data with different incident angles. Obtain high-precision aerial flight data for calculating landslide terrain information such as slope and aspect;

[0007] S2: Process remote sensing data, preprocess the radar remote sensing data, including intensity map generation, image registration, interference combination construction, differential interference phase calculation, phase unwrapping, etc. Calculate the annual average deformation amount obtained from radar data with different incident angles;

[0008] S3: Perform data registration. Pixel-level registration is carried out on the annual average deformation raster files of different radar beam incident angles of multi-source orbits, and the deformation amounts obtained from radar data with different incident angles of any grid in the study area are extracted.

[0009] S4: Calculate landslide topographic information, including slope and aspect, using high-precision aerial flight data. The maximum slope direction of different geological units is statistically analyzed using spatial analysis tools in Arcgis.

[0010] S5: Perform two-dimensional displacement decomposition. The annual average deformation information at different incident angles is converted into the actual two-dimensional deformation field on the ground surface using the two-dimensional deformation decomposition formula to obtain the deformation rates in the horizontal and vertical directions.

[0011] S6: Based on the mass conservation equation, perform the inversion of the slip surface morphology and invert the thickness of the soil landslide mass. Using the landslide topographic data and elevation information, the current elevation is subtracted from the thickness of the landslide mass to obtain the landslide morphology. According to the maximum thickness of the landslide and the distribution characteristics of the thickness of the landslide mass on both sides, the slip surface is divided into concave and planar types.

[0012] S7: Conduct landslide zoning. Use the high-precision surface elevation model to draw the shaded relief map of the landslide mass. According to the geomorphic distribution characteristics, the landslide body is divided into three parts: the rear edge of the landslide, the front edge of the landslide, and the landslide mass.

[0013] S8: Identify the landslide deformation mode. Statistically analyze the vertical and horizontal average deformation rates of the three parts of the landslide. Combine the multi-dimensional displacement difference distribution characteristics of different landslide zones and the shape of the slip surface to judge the landslide deformation mode.

[0014] S9: Result verification. Use methods such as field geological surveys to verify the identification results to ensure accuracy, and then the entire landslide deformation mode identification process based on InSAR multi-dimensional observations can be ended.

[0015] In a preferred embodiment, in step S1, the radar remote sensing data includes: collecting radar remote sensing data covering the study area, including radar data with different incident angles. These data can be obtained from SAR satellite data providers or relevant databases, such as Sentinel-1, ERS-1 / 2, and ENVISAT.

[0016] In a preferred embodiment, in step S2, the specific processing methods include:

[0017] ①. Set the terrain factors in the Range direction and the Azimuth direction to 5 and 1 respectively to generate an intensity map reflecting the actual terrain ratio of the study area.

[0018] ②. Select the radar intensity image obtained in autumn and winter as the main image, and register the intensities of the remaining images to this main image, with the requirement that the registration error is less than one-quarter of a pixel. Using the polynomial coefficients obtained from the intensity map registration iteration as the standard, register the original single-look complex raw data to obtain the resampled single-look complex data (rslc). Calculate the resampled intensity map (rmli) based on the resampled single-look complex data;

[0019] ③. Set up the time baseline and spatial baseline to construct the interferometric combination. To best meet the interferometric conditions, it is necessary to shorten the time and spatial baseline intervals of the interferometry. Therefore, set the maximum time baseline to 24 days and the spatial baseline to 100 m;

[0020] ④. Perform pixel registration on the rslc that constitutes the interferometric combination, then substitute the pixel values for conjugate multiplication to obtain the interferometric calculation result. Further substitute the terrain parameters of the study area in the radar coordinate format to obtain the differential interferometric phase;

[0021] ⑤. Use the minimum cost flow method to unwrap the differential interferometric data, and stack multiple differential interferometric phases to calculate the phase within the stacking time period. The stacking calculation formula is: Finally, calculate the phase information as deformation information to obtain the annual average deformation amount of the study area;

[0022] ⑥. Collect radar data with different incident angles and calculate according to the above method in turn to obtain the deformation values in different line-of-sight directions covering the same time period.

[0023] In a preferred embodiment, in step S3, it specifically includes: resample the grid files (including annual average deformation information) calculated with different radar incident angles to make the resolutions of different grid data consistent; then use the georeferencing tool in Arcgis to perform georegistration on the grid data with consistent resolutions, and further use the extract by mask tool for double registration verification to ensure that multiple grid files coincide precisely.

[0024] In a preferred embodiment, in step S4, the spatial analysis tools in Arcgis specifically refer to the aspect and slope tools to solve the aspect and slope information of the same geological unit in the study area, and based on the above results, statistically analyze the maximum slope drop direction of different geological units.

[0025] Resample the aspect and slope grid files with the sampling benchmark being the size of the grid file containing the annual average deformation amount information in step A2 to make the resolutions of different grid data consistent

[0026] Using the georeferencing tool in Arcgis, georeference the terrain raster data and the deformation raster data with the same resolution, and further use the extract by mask tool for double registration verification to ensure that multiple raster files are precisely coincident.

[0027] Define the maximum slope direction as the horizontal direction, and the direction perpendicular to the maximum slope direction as the vertical direction to establish a coordinate system.

[0028] In a preferred embodiment, in step S5, the annual deformation information at different incident angles obtained by calculation is denoted as dLos1 and dlos2 respectively. Using the raster calculator tool in Arcgis, substitute the dLos1, dlos2, Aspect, and slope raster data into the following formula to calculate the actual two-dimensional deformation field of the surface.

[0029]

[0030] Where d H and d V are the surface deformation rates in the horizontal and vertical directions respectively, α represents the angle between the satellite flight direction and the true north direction, β is the angle between the maximum slope direction of the slope and the true north direction, and θ represents the satellite incident angle.

[0031] In a preferred embodiment, in step S6, the thickness of the sliding mass of the landslide tapers off at the two side boundaries, and the sliding mass of the soil landslide shows uniform rheological characteristics in space. The landslide depth and the vertical deformation obtained by InSAR can be solved by the following formula

[0032]

[0033] Where h is the landslide thickness, t is the time, f is the rheological coefficient, Vsurf is the surface deformation rate. Substitute the decomposed two-dimensional displacement field information, and the information on the thickness of the landslide slope surface can be obtained using the least squares method.

[0034] In a preferred embodiment, in step S6, introduce the landslide terrain data obtained by high-precision aerial flight, subtract the landslide thickness obtained by back-inverting the sliding surface from the current elevation, and the landslide morphology can be obtained. According to the maximum thickness of the landslide and the distribution characteristics of the landslide thickness on both sides of the maximum thickness, the landslide sliding surface is divided into a concave type and a planar type. Among them, within the surface range with the position of the maximum thickness as the center and twice the maximum thickness as the radius, if the landslide thickness decreases by no more than one-fourth of the thickness, the sliding surface is of the planar type; otherwise, it is of the concave type.

[0035] In a preferred embodiment, in step S7, during the drawing process of the mountain shadow map, a high-precision surface elevation model is used to draw the mountain shadow map of the landslide body. The mountain shadow map can intuitively reflect the topographic undulations and geomorphic features of the landslide body, which helps to better understand the spatial distribution and deformation characteristics of the landslide body.

[0036] During the landslide zoning process, the rear edge of the landslide is usually located at the highest point of the landslide body, the front edge of the landslide is located at the lowest point of the landslide body, and the landslide body is located between the rear edge and the front edge of the landslide.

[0037] Analyze the vertical and horizontal deformation rates of the three parts of the landslide to understand the deformation characteristics and differences in different regions. This helps to more accurately judge the deformation mode and potential risks of the landslide.

[0038] In a preferred embodiment, in step S8, statistically analyze the vertical and horizontal average deformation rates of the three parts of the landslide, and calculate statistical indicators such as their mean, variance, maximum value, and minimum value. These indicators can reflect the intensity, range, and change trend of the landslide deformation.

[0039] In step S9, use field geological drilling or trench survey methods to conduct on-site investigations on the structure, material composition, deformation characteristics, etc. of the landslide body to verify the accuracy of the InSAR identification results. If GNSS monitoring data exists in the study area, compare the InSAR identification results with the GNSS monitoring data to analyze whether the monitoring results of the two methods are consistent, and further verify the reliability of the InSAR identification results. If there is a landslide deformation prediction model in the study area, the InSAR identification results can be compared with the model prediction results to analyze whether the prediction results of the two methods are consistent, and further verify the accuracy of the InSAR identification results.

[0040] In summary, due to the adoption of the above technical solutions, the beneficial effects of the present invention are:

[0041] 1. In the present invention, the two-dimensional deformation decomposition formula is the core of the entire recognition process, which can convert InSAR observation data with different incident angles into the actual deformation rates in the horizontal and vertical directions of the ground surface. By introducing terrain information and incident angle parameters, the line-of-sight displacement is converted into displacements in the horizontal and vertical directions, thereby more accurately reflecting the true deformation characteristics of landslides. At the same time, the two-dimensional deformation decomposition formula can provide more comprehensive deformation information, which helps to more accurately identify the deformation patterns of landslides. For example, the horizontal deformation and vertical deformation of rotational landslides show a linear relationship, while the horizontal deformation of translational and pushing landslides is relatively stable, and the vertical deformation shows linear growth or decrease. By analyzing the two-dimensional deformation distribution characteristics, different types of landslides can be more effectively distinguished. The two-dimensional deformation decomposition formula provides an important data basis for subsequent inversion of the slip surface morphology and landslide pattern recognition. By analyzing the deformation rates in the horizontal and vertical directions, the slip surface morphology and deformation pattern of landslides can be further inferred, providing a scientific basis for landslide warning and disaster prevention.

[0042] 2. In the present invention, the mass conservation equation is used to invert the thickness of the soil mass landslide body, and combined with terrain information to obtain the landslide morphology, which can significantly reduce costs and improve work efficiency compared with traditional on-site investigation methods such as drilling and trenching. This is of great significance for large-scale landslide monitoring and identification. At the same time, the mass conservation equation can consider the material composition and deformation characteristics of the landslide body, so as to more accurately invert the thickness of the slip surface. This is of great significance for understanding the deformation mechanism of landslides and predicting the development trend of landslides. By inverting the thickness of the slip surface, the shape and distribution characteristics of the slip surface can be further analyzed, providing more comprehensive information for landslide pattern recognition.

[0043] 3. In the present invention, inverting the landslide pattern by comprehensively considering the slip surface shape and two-dimensional displacement distribution characteristics can more accurately judge the deformation pattern of landslides, thereby providing a more reliable basis for landslide warning and disaster prevention. For rotational landslides, the deformation of the rear edge of the landslide needs to be focused on; for translational and pushing landslides, the deformation of the front edge of the landslide needs to be focused on. By taking targeted prevention and control measures, the risk of landslide disasters can be effectively reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a schematic diagram of two-dimensional displacement decomposition of the present invention;

[0045] Figure 2 is a multi-dimensional deformation distribution diagram of InSAR for rotational landslides in the present invention;

[0046] Figure 3 is a schematic diagram of the deformation pattern and field verification of rotational landslides in the present invention;

[0047] Figure 4 is a multi-dimensional deformation distribution diagram of InSAR for translational landslides in the present invention. Detailed implementation manners

[0048] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0049] Refer to Figures 1-4 ,

[0050] Embodiment:

[0051] A landslide deformation mode recognition method based on InSAR multi-dimensional observation, characterized in that it includes the following steps:

[0052] S1: Perform data preparation, collect radar remote sensing data covering the study area, including radar data with different incident angles. Obtain high-precision aerial flight data for calculating landslide terrain information, such as slope and aspect;

[0053] S2: Perform remote sensing data processing, preprocess the radar remote sensing data, including intensity map generation, image registration, interference combination construction, differential interference phase calculation, phase unwrapping, etc. Calculate the annual average deformation amount obtained from radar data with different incident angles;

[0054] S3: Perform data registration, perform pixel-level registration on the annual average deformation amount raster files of different radar beam incident angles with multi-source orbits, and extract the deformation amounts obtained from radar data with different incident angles of any grid in the study area;

[0055] S4: Calculate landslide terrain information, including slope and aspect, using high-precision aerial flight data. Use the spatial analysis tool in Arcgis to statistically analyze the maximum slope drop direction of different geological units;

[0056] S5: Perform two-dimensional displacement decomposition, use the two-dimensional deformation decomposition formula to convert the annual average deformation information of different incident angles into the actual two-dimensional deformation field on the ground surface, and obtain the deformation rates in the horizontal and vertical directions;

[0057] S6: Based on the mass conservation equation, perform the inversion of the slip surface morphology and invert the thickness of the soil landslide sliding mass. Use the landslide terrain data and elevation information, subtract the sliding mass thickness from the current elevation to obtain the landslide morphology. According to the maximum thickness of the landslide and the distribution characteristics of the sliding mass thickness on both sides, the slip surface is divided into a concave type and a planar type;

[0058] S7: Perform landslide zoning, use the high-precision surface elevation model to draw the shaded relief map of the sliding mass mountain body. According to the geomorphic distribution characteristics, divide the landslide body into three parts: the landslide rear edge, the landslide front edge, and the landslide body;

[0059] S8: Landslide deformation mode recognition, statistically calculate the vertical and horizontal average deformation rates of the three parts of the landslide. Combine the multi-dimensional displacement difference distribution characteristics of different landslide zones and the shape of the slip surface to judge the landslide deformation mode;

[0060] S9: Result verification, use field geological surveys and other methods to verify the recognition results to ensure accuracy, and then the entire landslide deformation mode recognition process based on InSAR multi-dimensional observations can be ended.

[0061] In step S1, the radar remote sensing data includes: collect radar remote sensing data covering the study area, including radar data with different incident angles. These data can be obtained from SAR satellite data providers or relevant databases, such as Sentinel-1, ERS-1 / 2, ENVISAT, etc.

[0062] The aerial flight data includes: obtain high-precision aerial flight data for calculating landslide terrain information. The aerial flight data can be obtained through aerial photogrammetry or lidar scanning, etc.

[0063] Data quality requirements:

[0064] Radar remote sensing data: The data quality is required to be good without noise interference to ensure the accuracy of subsequent data processing and deformation analysis results.

[0065] Aerial flight data: The data is required to have high resolution and complete coverage to ensure the accuracy and integrity of the terrain information.

[0066] Data time span:

[0067] Radar remote sensing data: Collect radar remote sensing data covering multiple time periods in the study area for time series analysis to capture the dynamic deformation characteristics of the landslide.

[0068] Aerial flight data: Select aerial flight data with a time close to that of the radar remote sensing data acquisition to ensure the timeliness of the terrain information.

[0069] Data preprocessing:

[0070] Radar remote sensing data: Preprocess the radar remote sensing data, including denoising, radiometric correction, geometric correction, etc., to ensure the accuracy and consistency of the data.

[0071] Aerial flight data: Preprocess the aerial flight data, including denoising, mosaicking, coordinate transformation, etc., to ensure the accuracy and integrity of the data..

[0072] In step S2, the specific processing methods include:

[0073] ①. Set the terrain factors in the Range direction and Azimuth direction to 5 and 1 respectively, and generate an intensity map reflecting the actual terrain ratio of the study area;

[0074] ②. Select the radar intensity image obtained in autumn and winter as the main image, and register the intensities of the other images to this main image, requiring the registration error to be less than one-quarter of a pixel. Using the polynomial coefficients obtained from the intensity map registration iteration as the standard, register the original single-look complex raw data to obtain the resampled single-look complex data (rslc). Calculate the resampled intensity map (rmli) based on the resampled single-look complex data;

[0075] ③. Set the time baseline and spatial baseline to construct an interference combination. To best meet the interference conditions, it is necessary to shorten the time and spatial baseline intervals of the interference. Therefore, set the maximum time baseline to 24 days and the spatial baseline to 100 m;

[0076] ④. Perform pixel registration on the rslc that constitutes the interference combination, and then substitute the pixel values for conjugate multiplication to obtain the interference calculation result. Further substitute the terrain parameters of the study area in the radar coordinate format to obtain the differential interference phase;

[0077] ⑤. Use the minimum cost flow method to unwrap the differential interference data, and stack multiple differential interference phases to calculate the phase within the stacking time period. The stacking calculation formula is: Finally, calculate the phase information as deformation information to obtain the annual average deformation amount of the study area;

[0078] ⑥. Collect radar data with different incident angles and calculate according to the above method in turn to obtain the deformation values in different line-of-sight directions covering the same time period.

[0079] In step S3, it specifically includes: resample the raster files (including annual average deformation information) calculated from different radar incident angles to make the resolutions of different raster data consistent; then use the georeferencing tool in Arcgis to georeference the raster data with consistent resolutions, and further use the extract by mask tool for double registration verification to ensure that multiple raster files coincide precisely..

[0080] In step S4, the spatial analysis tools in Arcgis specifically refer to the aspect and slpoe tools to solve the aspect and slope information of the same geological units in the study area, and based on the above results, statistically analyze the maximum slope drop direction of different geological units.

[0081] Resample the aspect and slope raster files with the raster file size containing the annual average deformation amount information in step A2 as the sampling reference to make the resolutions of different raster data consistent

[0082] Using the georeferencing tool in Arcgis, georegister the terrain raster data and the deformed raster data with the same resolution, and further use the extract by mask tool for double registration verification to ensure that multiple raster files coincide precisely.

[0083] Define the maximum slope direction as the horizontal direction, and the direction perpendicular to the maximum slope direction as the vertical direction to establish a coordinate system.

[0084] In step S5, the annual deformation information obtained at different incident angles is denoted as dLos1 and dlos2 respectively. Using the raster calculator tool in Arcgis, substitute the dLos1, dlos2, Aspect, and slope raster data into the following formula to calculate the actual two-dimensional deformation field of the surface.

[0085]

[0086] where d H and d V are the surface deformation rates in the horizontal and vertical directions respectively, α represents the angle between the satellite flight direction and the true north direction, β is the angle between the maximum slope direction of the slope and the true north direction, and θ represents the satellite incident angle.

[0087] In step S6, the thickness of the sliding mass of the landslide deposit tapers off at both side boundaries, and the sliding mass of the soil landslide shows uniform rheological characteristics in space. The landslide depth and the vertical deformation obtained by InSAR can be solved by the following formula

[0088]

[0089] where h is the landslide thickness, t is the time, f is the rheological coefficient, and Vsurf is the surface deformation rate. Substitute the decomposed two-dimensional displacement field information, and the information on the thickness of the landslide slope can be obtained using the least squares method.

[0090] In step S6, introduce the landslide terrain data obtained by high-precision aerial flight. Subtract the landslide thickness obtained by back-inverting the sliding surface from the current elevation to obtain the landslide morphology. According to the maximum thickness of the landslide and the distribution characteristics of the landslide thickness on both sides of the maximum thickness, the landslide sliding surface is divided into concave and planar types. Among them, within the surface range with the position of the maximum thickness as the center and twice the maximum thickness as the radius, if the landslide thickness reduction does not exceed one-fourth of the thickness, then the sliding surface is of the planar type; otherwise, it is of the concave type.

[0091] In step S7, during the process of drawing the mountain shadow map, use the high-precision surface elevation model to draw the mountain shadow map of the landslide body. The mountain shadow map can intuitively reflect the terrain undulation and geomorphic characteristics of the landslide body, which helps to better understand the spatial distribution and deformation characteristics of the landslide body.

[0092] During the landslide zoning process, the trailing edge of the landslide is usually located at the highest point of the landslide body, the leading edge of the landslide is located at the lowest point of the landslide body, and the landslide body is located between the trailing edge and the leading edge of the landslide.

[0093] Analyze the vertical and horizontal deformation rates of the three parts of the landslide to understand the deformation characteristics and differences in different areas. This helps to more accurately judge the deformation pattern and potential risks of the landslide.

[0094] In step S8, statistically analyze the vertical and horizontal average deformation rates of the three parts of the landslide, and calculate statistical indicators such as their mean, variance, maximum value, and minimum value. These indicators can reflect the intensity, range, and change trend of the landslide deformation.

[0095] In step S9, use field geological drilling or trench survey methods to conduct on-site investigations on the structure, material composition, deformation characteristics, etc. of the landslide body to verify the accuracy of the InSAR identification results. If GNSS monitoring data exists in the study area, compare the InSAR identification results with the GNSS monitoring data to analyze whether the monitoring results of the two methods are consistent, and further verify the reliability of the InSAR identification results. If there is a landslide deformation prediction model in the study area, the InSAR identification results can be compared with the model prediction results to analyze whether the prediction results of the two methods are consistent, and further verify the accuracy of the InSAR identification results.

[0096] It can be known from the above that:

[0097] In the present invention, the two-dimensional deformation decomposition formula is the core of the entire identification process, which can convert InSAR observation data with different incident angles into the actual horizontal and vertical deformation rates of the ground surface. By introducing terrain information and incident angle parameters, the line-of-sight displacement is converted into horizontal and vertical displacements, so as to more accurately reflect the true deformation characteristics of the landslide. At the same time, the two-dimensional deformation decomposition formula can provide more comprehensive deformation information, which helps to more accurately identify the deformation pattern of the landslide. For example, the horizontal deformation and vertical deformation of a rotational landslide show a linear relationship, while the horizontal deformation of a pulling type and a pushing type landslide is relatively stable, and the vertical deformation shows a linear increase or decrease. By analyzing the two-dimensional deformation distribution characteristics, different types of landslides can be more effectively distinguished. The two-dimensional deformation decomposition formula provides an important data basis for subsequent slip surface morphology inversion and landslide pattern recognition. By analyzing the deformation rates in the horizontal and vertical directions, the slip surface morphology and deformation pattern of the landslide can be further inferred, providing a scientific basis for landslide warning and disaster prevention.

[0098] In the present invention, the thickness of the landslide mass of a soil landslide is inverted using the mass conservation equation, and the landslide morphology is obtained in combination with topographic information. Compared with traditional on-site investigation methods such as drilling and trenching, the cost can be significantly reduced and the work efficiency can be improved. This is of great significance for large-scale landslide monitoring and identification. At the same time, the mass conservation equation can take into account the material composition and deformation characteristics of the landslide mass, so as to more accurately invert the thickness of the slip surface. This is of great significance for understanding the deformation mechanism of the landslide and predicting the development trend of the landslide. By inverting the thickness of the slip surface, the shape and distribution characteristics of the slip surface can be further analyzed, providing more comprehensive information for landslide pattern recognition.

[0099] In the present invention, inverting the landslide pattern by comprehensively considering the shape of the slip surface and the two-dimensional displacement distribution characteristics can more accurately judge the deformation pattern of the landslide, thus providing a more reliable basis for landslide early warning and disaster prevention and control. Different types of landslides have different slip surface shapes and two-dimensional displacement distribution characteristics. For example, the slip surface of a rotational landslide is concave, and the slip surfaces of translational and pushing landslides are planar. By analyzing the slip surface shape and two-dimensional displacement distribution characteristics, different types of landslides can be more effectively distinguished. The results of landslide pattern recognition can provide an important basis for landslide early warning and disaster prevention and control. For example, for rotational landslides, the deformation of the rear edge of the landslide needs to be focused on; for translational and pushing landslides, the deformation of the front edge of the landslide needs to be focused on. By taking targeted prevention and control measures, the risk of landslide disasters can be effectively reduced.

[0100] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the element.

[0101] The above description enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A landslide deformation pattern recognition method based on InSAR multi-dimensional observation, characterized by: The steps include: S1: Perform data preparation and collect radar remote sensing data covering the study area, including radar data at different incidence angles; Obtain high-precision aerial data to calculate landslide terrain information, including slope and aspect; S2: Perform remote sensing data processing and pre-process radar remote sensing data, including intensity map generation, image registration, interference combination construction, differential interference phase calculation, and phase unwrapping; calculate the average annual deformation obtained from radar data at different incident angles; S3: Perform data registration, perform pixel-level registration on the annual average deformation raster files of different radar beam incidence angles of multi-source orbits, and extract the deformation obtained by radar data of different incidence angles of any grid in the study area; S4: Use high-precision aerial data to calculate landslide terrain information, including slope and slope direction; use ArcGIS spatial analysis tools to calculate the maximum slope direction of different geological units; S5: Perform two-dimensional displacement decomposition, and use the two-dimensional deformation decomposition formula to convert the annual average deformation information of different incident angles into the actual two-dimensional deformation field of the surface, and obtain the deformation rate in the horizontal and vertical directions; S6: Based on the mass conservation equation, the sliding surface morphology is inverted to invert the thickness of the sliding body of the soil landslide; using the landslide terrain data and elevation information, the current elevation is subtracted from the sliding body thickness to obtain the landslide morphology; according to the maximum thickness of the landslide and the thickness distribution characteristics of the sliding body on both sides, the sliding surface is divided into a concave type and a flat type; S7: Perform landslide zoning and draw a landslide mountain shadow map using a high-precision surface elevation model; divide the landslide into three parts: the rear edge of the landslide, the front edge of the landslide, and the landslide body according to the landform distribution characteristics; S8: Identification of landslide deformation patterns, statistics of vertical and horizontal average deformation rates of three parts of the landslide; The deformation mode of the landslide is determined by combining the multi-dimensional displacement difference distribution characteristics and sliding surface shape of different landslide zones; S9: Result verification: Use field geological survey methods to verify the identification results to ensure accuracy, and then the entire landslide deformation pattern identification process based on InSAR multi-dimensional observation can be completed.

2. The landslide deformation pattern recognition method based on InSAR multi-dimensional observation according to claim 1, characterized in that: In step S1, the radar remote sensing data includes: collecting radar remote sensing data covering the study area, including radar data at different incident angles; these data are obtained from SAR satellite data providers or related databases, including Sentinel-1, ERS-1 / 2, and ENVISAT.

3. The landslide deformation pattern recognition method based on InSAR multi-dimensional observation as claimed in claim 1, characterized in that: In step S2, the specific method of processing includes: ①. Set the terrain factors in the Range and Azimuth directions to 5 and 1 respectively, and generate an intensity map that reflects the actual terrain proportions of the study area; ②. Select the radar intensity image acquired in autumn and winter as the main image, and register the remaining image intensities to the main image, requiring the registration error to be less than a quarter of a pixel; use the polynomial coefficients obtained by iterative intensity map registration as the standard to register the original single-view complex raw data to obtain the resampled single-view complex data (rslc); calculate the resampled intensity map (rmli) based on the resampled single-view complex data; ③. Set the time baseline and space baseline to construct the interference combination; the interference conditions are not met to the maximum extent, and the time and space baseline intervals of the interference need to be shortened. Therefore, the maximum time baseline is set to 24d and the space baseline is set to 100m; ④. Perform pixel registration on the RSLC constituting the interference combination, and then bring in the pixel value for conjugate multiplication to obtain the interference calculation result, and further bring in the terrain parameters of the study area in radar coordinate format to obtain the differential interference phase; ⑤. Use the minimum cost flow method to untangle the differential interference data, and superimpose multiple differential interference phases to calculate the phase within the superposition time period. The superposition calculation formula is: Finally, the phase information is calculated as deformation information to obtain the average annual deformation of the study area; ⑥ Collect radar data at different incident angles, and calculate them according to the above method in turn to obtain deformation values ​​in different line of sight covering the same time period.

4. The method for landslide deformation pattern recognition based on InSAR multi-dimensional observation according to claim 1, characterized in that: The step S3 specifically includes: resampling the raster files calculated at different radar incident angles; making the resolutions of different raster data consistent; then using the georeferencing tool in ArcGIS to georeference the raster data with consistent resolutions, and further using the extract by mask tool to perform double registration and verification to ensure that multiple raster files accurately overlap.

5. The landslide deformation pattern recognition method based on InSAR multi-dimensional observation according to claim 1, characterized in that: In step S4, the spatial analysis tools in ArcGIS, specifically aspect and slpoe, are used to solve the slope direction and slope information of the same geological unit in the study area, and based on the above results, the maximum slope direction of different geological units is counted; Resample the aspect and slope raster files, with the sampling basis being the size of the raster file containing the annual average deformation information in step A2; make the resolution of different raster data consistent; use the georeferencing tool in ArcGIS to georeference the terrain raster data and the deformation raster data with consistent resolution, and further use the extract by mask tool to perform double registration and verification to ensure that multiple raster files are accurately overlapped; Define the maximum slope direction as the horizontal direction, and the direction perpendicular to the maximum slope direction as the vertical direction, and establish a coordinate system.

6. The method for landslide deformation pattern recognition based on InSAR multi-dimensional observation according to claim 1, characterized in that: In step S5, the calculated annual average deformation information of different incident angles is recorded as dLos1 and dlos2 respectively, and the dLos1, dlos2, aspect and slope raster data are substituted into the following formula using the raster calculator tool in ArcGIS to calculate the actual two-dimensional deformation field of the surface; where d H and d V are the lateral and vertical surface deformation rates respectively, α represents the angle between the satellite flight direction and the true north, β is the angle between the maximum slope drop direction and the true north, and θ represents the satellite incidence angle.

7. The method for landslide deformation pattern recognition based on InSAR multi-dimensional observation according to claim 1, characterized in that: In step S6, the thickness of the accumulation landslide body pinches out at the boundaries on both sides, and the soil landslide body presents uniform rheological characteristics in space. The landslide depth and the vertical deformation obtained by InSAR are solved using the following formula: Among them, h is the thickness of the landslide, t is the time, f is the rheological coefficient, Vsurf is the surface deformation rate, and the decomposed two-dimensional displacement field information is brought in, and the thickness information of the landslide slope can be obtained using the least squares method.

8. The method for landslide deformation pattern recognition based on InSAR multi-dimensional observation according to claim 1, characterized in that: In step S6, the landslide topographic data obtained by high-precision aerial flight is introduced, and the sliding body thickness obtained by sliding surface inversion is subtracted from the current elevation to obtain the landslide morphology; the landslide sliding surface is divided into concave type and plane type according to the maximum thickness of the landslide and the distribution characteristics of the sliding body thickness on both sides of the maximum thickness; wherein the maximum thickness occurs at the center of the circle, within the surface range with twice the maximum thickness as the radius, if the landslide thickness decreases by no more than one-fourth of the thickness, the sliding surface is plane type, otherwise it is concave type.

9. The method for landslide deformation pattern recognition based on InSAR multi-dimensional observation according to claim 1, characterized in that: In the step S7, a high-precision surface elevation model is used to draw a mountain shadow map of the landslide body during the mountain shadow map drawing process; the mountain shadow map can intuitively reflect the terrain undulations and geomorphic features of the landslide body, which helps to better understand the spatial distribution and deformation characteristics of the landslide body; In the landslide zoning process, the rear edge of the landslide is located at the highest point of the landslide body, the front edge of the landslide is located at the lowest point of the landslide body, and the landslide body is located between the rear edge and the front edge of the landslide; The vertical and horizontal deformation rates of the three parts of the landslide were analyzed to understand the deformation characteristics and differences in different areas; This helps to more accurately determine the deformation pattern and potential risk of landslides.

10. The method for landslide deformation pattern recognition based on InSAR multi-dimensional observation according to claim 1, characterized in that: In the step S8, the vertical and horizontal average deformation rates of the three parts of the landslide are statistically analyzed to calculate the mean, variance, maximum and minimum statistical indicators; These indicators can reflect the intensity, scope and changing trend of landslide deformation; In step S9, a field investigation is conducted on the structure, material composition and deformation characteristics of the landslide body by using field geological drilling or trenching methods to verify the accuracy of the InSAR identification results.

Images