Landslide movement analysis method based on unmanned aerial vehicle remote sensing and three-dimensional geological modeling

Abstract

The application is a landslide movement analysis method based on unmanned aerial vehicle remote sensing aerial survey and three-dimensional geological modeling, high-resolution topographic data of a landslide hidden danger area is obtained by using a laser scanner technology carried by an unmanned aerial vehicle, a finite element Geostudio software is used to calculate a landslide instability damage range and a stability coefficient under different scenarios, the stability coefficient is used as a reduction coefficient of a rock-soil body strength parameter, then rock-soil body strength parameters under different scenarios are determined in discrete element PFC software, and a discrete element is used to realize three-dimensional damage movement process analysis of a landslide under different scenarios. The application can compare a landslide scale under different rainfall and earthquake scenarios and movement characteristics after instability of different scale landslides by using PFC software, can effectively simulate large deformation under fluid-structure coupling conditions, determine rock-soil body strength parameters under different rainfall (or earthquake) scenarios, and carry out three-dimensional movement process of a landslide under different scenarios for a landslide hidden danger point.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of engineering geology, in particular to a landslide movement analysis method based on unmanned aerial vehicle remote sensing and three-dimensional geological modeling. BACKGROUND

[0002] The formation of landslides is related to many factors, including geological structure, topography, soil properties, rainfall, groundwater level changes, seismic activity, human engineering activities, etc. Among them, rainfall is the most common factor that triggers landslides, especially after continuous or heavy rainfall, soil moisture increases, and soil shear strength decreases, thereby increasing the probability of landslide occurrence. This disaster often has the characteristics of suddenness, strong destructiveness and difficulty in prediction. Therefore, it is necessary to accurately and timely predict landslides triggered by rainfall.

[0003] However, the current evaluation of landslide movement process and its risk mainly focuses on the retrospective analysis of the landslides that have occurred, that is, the calculation results of numerical simulation methods are compared with the known actual results. These landslide cases usually have detailed field investigation data and survey data. However, for landslide hazard points that have not yet occurred, the prediction of the movement process under specific rainfall conditions and different seismic conditions still has no effective method. Especially, there is a lack of landslide disaster prediction framework combining remote sensing, numerical simulation and other technical means, which makes it difficult to analyze the landslide movement under different rainfall scenarios and different seismic scenarios. SUMMARY

[0004] In view of the deficiencies of the prior art, the technical problem to be solved by the present application is to provide a landslide movement analysis method based on unmanned aerial vehicle remote sensing and three-dimensional geological modeling, aiming to solve the problem that it is difficult to accurately predict the three-dimensional movement of landslides under different scenarios (such as different intensity rainfall, earthquake). The present application has important scientific reference value for formulating effective landslide prevention and control strategies.

[0005] In order to solve the above technical problems, the technical scheme of the present application is as follows:

[0006] A landslide movement analysis method based on unmanned aerial vehicle remote sensing and three-dimensional geological modeling, the process of the method is:

[0007] S1, using an unmanned aerial vehicle to obtain high-resolution remote sensing images of the landslide hazard investigation area, generating a digital elevation model DEM, an orthographic image and an inclined model of the investigation area, obtaining the topographic conditions and high-resolution topographic data of the landslide hazard point;

[0008] S2, the digital elevation model obtained in step S1 is converted and imported into Geostudio software, a Geostudio landslide model is generated, a seepage field of the slope under different rainfall scenarios is calculated in the SEEP / W module of the Geostudio software, and stability analysis is performed by using the SLOPE / W module to obtain the position of the landslide sliding surface and the stability coefficient of the landslide hazard point under different rainfall scenarios, and then the damage range of the landslide is determined; in the QUAKE / W stress analysis module of the Geostudio software, the initial stress and initial pore water pressure of the slope are set, and the transverse seismic acceleration data is input as a superior directory, and then the slope stability under different earthquake scenarios is analyzed by combining the Newmark deformation module to obtain the position of the sliding surface and the stability coefficient of the landslide hazard point under the scenario, and then the damage range of the landslide under different earthquake scenarios is determined;

[0009] S3, using the high-resolution terrain data of the landslide hazard point obtained in step S1, a three-dimensional geological model of the landslide is established in PFC3D software by using ball-wall contact, the sliding surface in the three-dimensional geological model of the landslide is determined by S2, and the sliding bed in the three-dimensional geological model of the landslide is determined based on the damage range of the landslide obtained by S2;

[0010] After obtaining the stability coefficients of the landslide hazard point under different rainfall scenarios and different earthquake scenarios in step S2, the stability coefficients are used as strength reduction factors of the rock and soil strength parameters of the landslide, and the strength parameter values under different rainfall scenarios and different earthquake scenarios are obtained by multiplying the rock and soil strength parameter values under natural conditions by the corresponding strength reduction factors; the rock and soil strength parameter values under natural conditions are the data without rainfall and earthquake determined by experience or experiment;

[0011] The landslide is discretized into particles, and the discrete element particles in PFC are given according to the strength parameter values of the corresponding scenarios to obtain the three-dimensional geological model of the landslide under different scenarios;

[0012] S4, based on the three-dimensional geological model of the landslide under different scenarios obtained in step S3, the initial conditions and boundary conditions of the landslide are set in PFC3D software, and the landslide is numerically simulated, and the velocity, displacement, and three-dimensional motion range diagram under different scenarios are output, and the three-dimensional motion characteristics of the landslide under different scenarios are analyzed.

[0013] Further, the analysis of the three-dimensional motion characteristics of the landslide under different scenarios includes the analysis of the motion speed, displacement, motion range and accumulation depth of the landslide particles, and the comparison of the possible motion range and maximum motion speed of the landslide under different scenarios, so as to compare the landslide danger under different scenarios.

[0014] Further, the rock-soil mass strength parameter value includes a damping coefficient, a friction coefficient, cohesion, and an internal friction angle.

[0015] Further, the unmanned aerial vehicle is a DJI Mavic3E unmanned aerial vehicle equipped with an RTK module, and two routes with a certain overlap degree are adopted to realize oblique photography when the unmanned aerial vehicle plans a route, and the overlap degree is 50-80%.

[0016] Compared with the prior art, the beneficial effects of the present application are:

[0017] (1) Previous researches are mainly back analysis or inverse analysis of the motion process of the occurred landslide, and the present application mainly aims at landslide hidden danger points, and high-resolution topographic data of the landslide hidden danger area can be quickly and accurately obtained by using the laser scanner technology carried by the unmanned aerial vehicle. The landslide hidden danger points are usually not subjected to detailed field investigation, and there is no accurate survey data, therefore, based on the characteristics that the finite element software can perform fluid-structure coupling analysis, the Geostudio software is used to analyze the instability range of the landslide hidden danger point under different rainfall and earthquake scenarios, and the landslide damage range and sliding surface under different scenarios can be accurately determined.

[0018] (2) The finite element software cannot effectively simulate large deformation and landslide motion, and the discrete element cannot consider different rainfall infiltration conditions. The present application combines Geostudio and PFC for coupling analysis, uses Geostudio to analyze the landslide stability under different rainfall and earthquake scenarios, uses the stability coefficient of the landslide hidden danger point as the strength reduction coefficient, and then uses the coefficient to reduce the strength of the rock-soil parameters in the discrete element PFC software under the corresponding scenario, so as to obtain the rock-soil mass strength parameters under different scenarios, which are used for calculating the motion process analysis after the landslide instability.

[0019] (3) The high-resolution image data obtained by the unmanned aerial vehicle remote sensing is used to establish a three-dimensional geological model of the landslide in PFC, and in reality, only the analysis in three-dimensional scale can accurately reflect the motion process after the landslide instability, so as to accurately predict the motion characteristics and danger of the landslide. Based on the aforementioned determined rock-soil mass strength parameters under different scenarios, the three-dimensional motion process analysis of the landslide under different rainfall and earthquake scenarios is carried out. The two indexes of landslide motion range and motion speed are mainly analyzed, so that the danger of the landslide under different scenarios can be accurately predicted and compared.

[0020] The present application obtains high-resolution terrain data of landslide hidden danger area by using the laser scanner technology carried by the unmanned aerial vehicle, calculates the landslide instability damage range and stability coefficient under different scenarios (different rainfall, earthquake) by using finite element Geostudio software, takes the stability coefficient as the reduction coefficient of rock-soil strength parameter, and then determines the rock-soil strength parameter under different scenarios in the discrete element PFC software, and realizes the three-dimensional damage motion process analysis of landslide under different scenarios by using the discrete element. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 Unmanned aerial vehicle route (a), (b) and obtained DEM data (c) and landslide hidden danger point inclination model (d).

[0022] Figure 2 The figure of landslide hidden danger point damage range and stability coefficient under 50-year return period rainfall.

[0023] Figure 3 The three-dimensional motion diagram of landslide under 50-year return period rainfall (wherein T refers to time).

[0024] Figure 4 The three-dimensional motion diagram of landslide under earthquake scenario (wherein T refers to time).

[0025] Figure 5 The velocity-time relationship diagram of monitoring particles under different scenarios.

[0026] Figure 6 The displacement-time change diagram of monitoring particles under different scenarios. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical scheme and advantages of the present application more clear and obvious, the present application is further described in detail below in combination with the drawings and examples. Of course, the specific examples described here are only used to explain the present application, and are not used to limit the present application.

[0028] The landslide motion analysis method based on unmanned aerial vehicle remote sensing aerial survey and three-dimensional geological modeling of the present application comprises the following steps:

[0029] S1, DJI Mavic3E unmanned aerial vehicle with RTK module acquires high-resolution remote sensing of landslide hazard investigation area, generates digital elevation model DEM, orthographic image and tilt model of landslide hazard point investigation area, obtains topographic conditions and high-resolution topographic data of landslide hazard point.

[0030] The unmanned aerial vehicle technology operation process is as follows:

[0031] (1) Determine the working area on Google Earth, save it as a KMZ file, and import it into the unmanned aerial vehicle.

[0032] (2) Set up control points and unmanned aerial vehicle routes evenly distributed in the area. In order to ensure the smoothness of data processing, two routes with a certain degree of overlap are used to realize oblique photography when planning the route, and the overlap can be set between 50%-80%.

[0033] (3) Use Pix4Dmapper software to process and obtain dense point cloud data, generate digital elevation model DEM, orthographic image and oblique model of the investigation area.

[0034] S2, convert the digital elevation model of step S1 and import it into Geostudio software, divide the grid in Geostudio software, establish the Geostudio landslide model, and analyze the sliding surface and damage range of the landslide hazard point:

[0035] Calculate the seepage field of the slope under different rainfall scenarios in the SEEP / W module of the Geostudio software, and inherit the results of this step, and use the SLOPE / W module to analyze the stability, obtain the sliding surface position and stability coefficient of the landslide hazard point under different rainfall scenarios, and the sliding surface position above is the instability damage range of the landslide, and then determine the damage range (depth) of the landslide; In the QUAKE / W stress analysis module of the Geostudio software, set the initial stress of the slope, the initial pore water pressure and input the transverse seismic acceleration data as the upper directory, and then combine the Newmark deformation module to analyze the stability of the slope under different earthquake scenarios, obtain the sliding surface position and stability coefficient of the landslide hazard point under different earthquake scenarios, and then determine the damage range (depth) of the landslide;

[0036] The specific steps for analyzing the sliding surface and damage range of the landslide hazard point in the finite element Geostudio software are as follows:

[0037] (1) Open the Geostudio software, select the SEEP / W module, set the unit time, duration and step number;

[0038] (2) Using ArcGIS to process the high-resolution terrain data obtained in S1, converting the digital elevation model into projection coordinate data that can be recognized by AutoCAD. In AutoCAD, use the boundary command to generate a dwg file, then import it into the SEEP / W module of the Geostudio software to generate a Geostudio landslide model, and then perform meshing;

[0039] (3) Set the material model, volumetric water content function, and hydraulic conductivity function, then input them into the drawing material.

[0040] Set and input the initial water level of the landslide and set the input boundary conditions, add rainfall boundary conditions to the boundary conditions, and note to check "potential seepage surface inspection" to prevent overflow;

[0041] (4) Add a SLOPE / W module to the SEEP / W module for stability analysis. After inheriting the seepage analysis results described above, the sliding surface position and stability coefficient of the landslide hazard point under this scenario are calculated. The sliding surface position above is the unstable failure range of the landslide, and then the failure range (depth) of the landslide is determined;

[0042] (5) For the earthquake scenario, after meshing is completed, set the initial stress of the slope, the initial pore water pressure, and input the transverse seismic acceleration data as the upper directory through the QUAKE / W stress analysis module. Then, combine the Newmark deformation module to analyze the slope stability under different seismic scenarios, and obtain the sliding surface position and stability coefficient of the landslide hazard point under this scenario. The sliding surface position above is the unstable failure range of the landslide, and then the failure range of the landslide under different seismic scenarios is determined.

[0043] S3, using the high-resolution terrain data of the landslide hazard point obtained in S1, a three-dimensional geological model of the landslide is established in PFC3D software using ball-wall contact. The sliding surface in the three-dimensional geological model of the landslide is determined by S2, and the sliding bed in the three-dimensional geological model of the landslide is determined based on the failure range of the landslide obtained in S2;

[0044] After obtaining the stability coefficients of the landslide hazard points under different rainfall scenarios and different earthquake scenarios in step S2, the stability coefficients are taken as the strength reduction factors of the rock-soil strength parameters of the landslide, and the strength parameter values under different rainfall scenarios and different earthquake scenarios are obtained by multiplying the rock-soil strength parameter values under natural conditions by the corresponding strength reduction factors. The rock-soil strength parameter values under natural conditions are the data determined by experience or experiment without rainfall and earthquake. For example, if the stability coefficient of the landslide hazard point under rainfall conditions calculated by Geostudio is 0.9, the rock-soil strength parameter under natural conditions (without rainfall, without earthquake, etc.) is multiplied by 0.9 to obtain the rock-soil strength parameter under rainfall conditions.

[0045] The landslide body is discretized into particles, and the discrete element particles in PFC are assigned according to the strength parameter values under corresponding scenarios to obtain the three-dimensional geological model of the landslide under different scenarios.

[0046] The specific process of step S3 is as follows:

[0047] (1) High-resolution terrain data is processed using ArcGIS software, and digital elevation model (DEM) data obtained by a UAV is imported into ArcGIS. Based on the landslide sliding surface and damage range determined by Geostudio software, the area range of the slide bed and the slide body is demarcated to prepare the basic data for subsequent modeling.

[0048] (2) The tif files of the slide bed and the slide body are imported into GlobalMapper for processing into dem format files that can be recognized by 3Dmax software. After importing them into 3Dmax software, a 3D model is established. The 3D model is exported as an stl format file that can be recognized by PFC3D software, and then imported into PFC3D to form the final three-dimensional geological model of the landslide.

[0049] (3) The landslide body is discretized into particles, and the discrete element particles in PFC are assigned according to the strength parameter values under corresponding scenarios. By defining particle parameters, including particle size, damping coefficient, friction coefficient, and some mesoscopic parameters, the damping coefficient, friction coefficient, and some mesoscopic parameters are determined by the stability coefficient reduction to generate the particles of the slide body part under the current scenario.

[0050] S4, based on the three-dimensional geological model of the landslide under different scenarios obtained in step S3, the initial conditions and boundary conditions of the landslide are set in PFC3D software, and the landslide is numerically simulated. The speed, displacement, and three-dimensional motion range under different scenarios are output, and the three-dimensional motion characteristics of the landslide are analyzed. By analyzing the indicators in the motion process, the danger of the landslide under different rainfall and earthquake conditions (such as motion speed, displacement, and influence range) is reflected, thereby effectively predicting the landslide.

[0051] The possible movement range and maximum movement speed of the landslide in different scenarios are compared, so as to compare the landslide danger in different scenarios.

[0052] (1) gravity load is given to the landslide particles, and the unbalanced force between the particles and the constraint wall is constrained, so that the mechanical balance of the landslide particles is achieved, and the phenomenon of the flying of the landslide particles is avoided;

[0053] (2) the movement process of the landslide particles under the action of gravity and the influence range and accumulation form of the landslide are simulated. By setting monitoring particles at different parts of the landslide, the speed, displacement and accumulation form of the landslide particles in different scenarios are output, so that the movement process of the landslide is analyzed.

[0054] The different scenarios in the application mainly include the change of rainfall conditions and earthquake conditions, such as different rainfall amounts of different recurrence periods for rainfall conditions, and different earthquake accelerations for earthquake conditions.

[0055] Since the application performs three-dimensional movement analysis of the landslide, the movement speed, displacement, movement range and accumulation depth of the landslide particles in different rainfall scenarios (such as 50-year and 100-year rainfall) and different earthquake scenarios can be compared. The above indexes can be used as physical quantities representing the characteristics of the landslide movement, so that the characteristics of the landslide movement in different scenarios can be compared.

[0056] The application uses Geostudio software to calculate the stability coefficients of different landslide hidden points, overcomes the disadvantages of the discrete element PFC software in realizing rainfall infiltration and fluid-solid coupling analysis, establishes Geostudio-PFC coupling analysis of finite elements and discrete elements, reduces the strength parameters of the rock-soil mass through the stability coefficients in different scenarios, and gives the particles in PFC accurate strength parameters. At the same time, the PFC software realizes the three-dimensional movement process analysis of the landslide hidden points.

[0057] Embodiment:

[0058] The research example is the Zhuangguoyu landslide, which is located in Jizhou District, Tianjin City, China. After a heavy rain, a small-scale landslide with a volume of more than 1000m3 was found at the toe of the slope in this area. The plants here have been destroyed, but other parts of the slope are still covered with trees. The local civil affairs department has set up a geological disaster warning point here, but no professional monitoring instruments have been implemented except for a simple protective net.

[0059] For Zhuoguoshan landslide, rainfall is the most important factor to trigger landslide instability, so it is necessary to analyze the extreme rainfall scenario to predict the potential stability and danger of landslide in advance. The extreme rainfall of the study area in 1980 and 2017 was generated, and the analysis of the number of rainfall days in each month showed that the continuous 3-day rainfall was the most frequent event in the rainy season of the study area. The maximum rainfall of each month for 3 consecutive days was calculated, and then the relationship between rainfall and return period was calculated by Matlab program according to Gumbel distribution.

[0060] The probability density function f(x) and the cumulative density function F(x) of Gumbel distribution are respectively:

[0061]

[0062]

[0063] Where x refers to the analysis variable, such as the analysis of rainfall conditions, x refers to the rainfall. Alpha is the scale parameter, and beta is the location parameter. Their relationship can be expressed as:

[0064]

[0065] Where mu is the mean value of the distribution, and parameters alpha and beta can be obtained by maximum likelihood estimation.

[0066] The results of extreme rainfall and return period obtained from Gumbel distribution show that the rainfall of 20-year, 50-year and 100-year return period is 151.5 mm, 184.6 mm and 209.3 mm respectively.

[0067] Step S1: Obtain high-resolution remote sensing of landslide by DJI Mavic3E unmanned aerial vehicle with RTK module. The unmanned aerial vehicle obtains point cloud data through two flight paths with 83% overlap in the direction and 75% overlap in the lateral direction, and adopts Pix4Dmapper software to process, to generate digital surface model (DEM), orthographic image and oblique model of the survey area, as shown in Figure 1 The two lines ABCDE and A'B'C'D'E' in the figure have an overlap degree of about 75% in the embodiment. Process the terrain model by ArcGIS to convert the digital elevation model (DEM) into projection coordinate data recognizable by AutoCAD. In AutoCAD, use the boundary command to generate a dwg file, and then import it into the Geostudio software SEEP / W module to generate a Geostudio landslide model. Considering the geometric shape and volume of the slope, the size of the grid is set to 0.5 m, thereby determining a total of 6027 nodes and 5866 elements.

[0068] Step S2: Re-annual rainfall is used as the hydraulic boundary condition. Since the field engineering survey does not show the groundwater level, the water level line is not set. The seepage field of the slope under the rainfall scenario is calculated in the SEEP / W module of the software, and the results of this step are inherited, the deterministic parameters are input in the SLOPE / W module for stability analysis, and finally the landslide failure range and the stability coefficient under this scenario are obtained, as shown in Figure 2

[0069] Step S3: The terrain data obtained in S1 is processed by ArcGIS, and the landslide failure range obtained in S2 is determined as the sliding body, and the range that does not occur failure is determined as the sliding bed. The above sliding body and sliding bed range are respectively exported into tif format file to be imported into GlobalMapper software. The tif file is imported into GlobalMapper for processing into dem format file, and the file is opened in 3Dmax file to form a 3D model of Zhuangguo Valley landslide. The file is saved as stl format, which is imported into PFC3D software to generate the final three-dimensional geological model of the landslide.

[0070] In PFC3D, the three-dimensional geological model of the landslide is established by using ball-wall contact. First, the sliding bed part is set as wall, and then the sliding body particles are generated by filling the complex closed surface formed by the upper surface of the landslide body and the sliding bed. The porosity is set to 0.2, the particle radius range is between 0.4-0.7m, and the bedrock particle size is 0.5-1m. The selected particle size is from 0.8m to 1.0m, a total of 8589 balls are generated, and the ball unit is written into PFC3D, and finally the three-dimensional modeling of Zhuangguo Valley landslide is obtained.

[0071] The stability coefficient of the landslide hidden danger point under the condition of 50-year return period rainfall is 0.897, so the strength reduction coefficient is 0.897. The rock and soil strength parameter value under the natural scenario is multiplied by 0.897, which is the rock and soil strength parameter value under the condition of 50-year return period rainfall. Assign these parameter values to the particles in PFC3D software to obtain the three-dimensional geological model of the landslide under the current scenario. The rock and soil strength parameter values include damping coefficient, friction coefficient, cohesion, internal friction angle, etc.

[0072] Step S4: Based on the three-dimensional geological model of the landslide under different scenarios obtained in step S3, the initial conditions and boundary conditions of the landslide are set in PFC3D software, and the landslide is numerically simulated, and the three-dimensional motion characteristics of the landslide under different scenarios are analyzed.

[0073] ​The gravity load is given to the landslide particle, and the mechanical balance of the landslide particle is achieved by restraining the unbalanced force between the particle and the restraint wall, so that the phenomenon of the flying of the landslide particle is avoided.In order to better show the three-dimensional motion process of the Zhuangguo Valley landslide under the rainfall and earthquake scenarios, six time nodes T=5s, 10s, 20s, 30s, 40s and 60s are selected to analyze the motion characteristics, as shown in Figure 3 and Figure 4 as shown in

[0074] In order to more accurately understand the detailed motion state of the particle under the three-dimensional simulation of the Zhuangguo Valley landslide under different scenarios, four monitoring balls are also arranged at different positions of the landslide. The monitoring particle velocity and time variation characteristics are as shown in Figure 5 The particle displacement changes with time, and the results are as shown in Figure 6

[0075] Finally, the simulation results of the landslide motion under different scenarios can be obtained, and the influence range of the landslide on the surrounding area under the 50-year return period rainfall condition is larger than that of the earthquake-induced landslide, the particle velocity is faster, and the influence range on the residential buildings is wider.

[0076] The above only describes the preferred embodiments of the present application, and is not used to limit the present application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

[0077] The unmentioned part of the present application is applicable to the prior art.

Claims

1. A landslide movement analysis method based on unmanned aerial vehicle remote sensing aerial survey and three-dimensional geological modeling, characterized in that, The process of the method is: S1, using a UAV to obtain high-resolution remote sensing images of a landslide hazard investigation area, generating a digital elevation model DEM, an orthographic image and an inclined model of the investigation area, obtaining the topographic conditions and high-resolution topographic data of the landslide hazard point; S2, converting the digital elevation model of step S1 and importing it into Geostudio software, generating a Geostudio landslide model, calculating the seepage field of the slope under different rainfall scenarios in the SEEP / W module of the Geostudio software, and using the SLOPE / W module for stability analysis to obtain the landslide sliding surface position and stability coefficient of the landslide hazard point under different rainfall scenarios, and then determine the damage range of the landslide; set the initial stress, initial pore water pressure of the slope and input the transverse seismic acceleration data as the upper directory in the QUAKE / W stress analysis module of the Geostudio software, and then combine the Newmark deformation module to analyze the stability of the slope under different earthquake scenarios, obtain the sliding surface position and stability coefficient of the landslide hazard point under the scenario, and then determine the damage range of the landslide under different earthquake scenarios; S3, using the high-resolution topographic data of the landslide hazard point obtained in step S1, a three-dimensional geological model of the landslide is established in PFC3D software using ball-wall contact, the sliding surface in the three-dimensional geological model of the landslide is determined by S2, and the sliding bed in the three-dimensional geological model of the landslide is determined based on the damage range of the landslide obtained by S2; After obtaining the stability coefficients of the landslide hazard point under different rainfall scenarios and different earthquake scenarios in step S2, the stability coefficients are used as the strength reduction factor of the rock-soil strength parameter of the landslide, and the rock-soil strength parameter value under natural working conditions is multiplied by the corresponding strength reduction factor to obtain the strength parameter value under different rainfall scenarios and different earthquake scenarios; The rock-soil strength parameter value under natural working conditions is the data determined by experience or experiment without rainfall and earthquake; The landslide body is discretized into particles, and the discrete element particles in PFC are assigned according to the strength parameter value of the corresponding scenario to obtain the three-dimensional geological model of the landslide under different scenarios; S4, based on the three-dimensional geological model of the landslide under different scenarios obtained in step S3, set the initial conditions and boundary conditions of the landslide in PFC3D software, and then simulate the landslide, output the velocity, displacement, and three-dimensional motion range diagram under different scenarios, and analyze the three-dimensional motion characteristics of the landslide under different scenarios.

2. The method of claim 1, wherein, The analysis of the three-dimensional motion characteristics of the landslide under different scenarios includes the analysis of the motion speed, displacement, motion range and accumulation depth of the landslide particles, and the comparison of the possible motion range and maximum motion speed of the landslide under different scenarios, so as to compare the landslide danger under different scenarios.

3. The method of claim 1, wherein, The rock-soil strength parameter value includes damping coefficient, friction coefficient, cohesion and internal friction angle.

4. The method of claim 1, wherein, The UAV is a DJI Mavic3E UAV equipped with an RTK module, and when planning the route of the UAV, two routes with a certain overlap degree are used to realize oblique photography, and the overlap degree is 50-80%.

Images