Method and device for predicting co-seismic landslide in meizoseismal region based on terrain amplification coefficient

Abstract

The invention discloses a meizoseismal region co-seismic landslide prediction method and device based on a terrain amplification coefficient, and the method comprises the steps: dividing a meizoseismal region into a plurality of grid units, and determining the initial peak acceleration of each grid unit; then determining a corresponding terrain amplification coefficient based on the terrain gradient and the lithologic parameter of the grid unit; determining the peak acceleration of the corresponding grid unit according to the initial peak acceleration and the terrain amplification coefficient; and finally, inputting the influence factor and the peak acceleration into a landslide prediction model to obtain a prediction result, the reliability of the scheme is superior to that of a traditional scheme, the method has high scientificity, and the co-seismic landslide can be predicted more accurately, so that the disaster prevention and reduction capability is improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of co-seismic landslide, and particularly relates to a co-seismic landslide prediction method and device based on a terrain amplification coefficient in an extreme earthquake zone. BACKGROUND

[0002] In some areas with frequent geological activities, strong earthquakes frequently occur in history, and often induce large-scale co-seismic landslides in the extreme earthquake zone, causing serious casualties and property losses. The co-seismic landslide is a landslide phenomenon directly induced by an earthquake. The prediction or susceptibility evaluation of the co-seismic landslide in the prior art ignores the amplification effect of complex terrain, resulting in insufficient accuracy and reliability of the prediction result or evaluation result.

[0003] Therefore, how to accurately predict the co-seismic landslide so as to improve the ability of disaster prevention and reduction is a technical problem to be solved by the person skilled in the art. SUMMARY

[0004] The application aims to solve the technical problem of low prediction accuracy of the co-seismic landslide in the prior art.

[0005] To achieve the above technical purpose, in one aspect, the application provides a co-seismic landslide prediction method based on a terrain amplification coefficient in an extreme earthquake zone, which comprises the following steps: dividing the extreme earthquake zone into a plurality of grid units and determining initial peak accelerations of the grid units; determining corresponding terrain amplification coefficients based on terrain slopes and lithology parameters of the grid units; determining peak accelerations of the corresponding grid units according to the initial peak accelerations and the terrain amplification coefficients; inputting an influence factor and the peak accelerations into a landslide prediction model to obtain a prediction result.

[0006] Further, the peak acceleration is determined by the following formula: ; In the formula, is the peak acceleration, is the terrain amplification coefficient, is the initial peak acceleration.

[0007] Further, the terrain amplification coefficient is determined by the following formula: ; In the formula, is the terrain amplification coefficient, is a constant set based on geology, i.e., a lithology parameter, is an included angle between a terrain slope and a horizontal plane.

[0008] Furthermore, the initial peak acceleration is determined by the following formula: ; In the formula, For the magnitude term, For distance, For site response items, For fault type items, This is the upper plate effect term.

[0009] Furthermore, the influencing factors include at least: elevation, slope, aspect, lithology, distance from the fault, and normalized vegetation index.

[0010] Furthermore, before inputting the peak acceleration and influencing factors into the landslide prediction model to obtain the prediction result, the method further includes: Multiple pairs of related factors are obtained by performing pairwise correlation calculations on each training influencing factor. Remove one factor from each of all relevant factor pairs to obtain multiple initial factor selections; After removing unqualified factors from all the initial selection factors, the landslide prediction model is trained by combining the peak acceleration training to obtain a well-trained landslide prediction model.

[0011] Furthermore, the landslide prediction model is specifically a random forest machine learning model, and the unqualified influencing factor is specifically an influencing factor whose impurity decreases below a preset threshold. The amount of impurity decrease is determined by the following formula: ; In the formula, The amount of decrease in impurity, The impurity of the parent node. This represents the proportion of samples in the left child node to samples in the parent node. The impurity of the left child node. This represents the proportion of samples in the right child node to the total number of samples in the parent node. The impurity of the right child node.

[0012] On the other hand, the present invention also provides a device for predicting coseismic landslides in megaseismic zones based on terrain amplification factors, the device comprising: The partitioning module is used to divide the metronome zone into multiple grid cells and determine the initial peak acceleration of each grid cell. The terrain module is used to determine the corresponding terrain magnification factor based on the terrain slope and lithology parameters of the grid cells; A peak acceleration module is used to determine the peak acceleration of the corresponding grid cell based on the initial peak acceleration and the terrain magnification factor; The prediction module is used to input the influencing factors and the peak acceleration into the landslide prediction model to obtain the prediction results.

[0013] This invention provides a method and apparatus for predicting coseismic landslides in megaseismic zones based on topographic amplification factors. Compared with existing technologies, this method first divides the megaseismic zone into multiple grid cells and determines the initial peak ground acceleration (PGA) of each grid cell. Then, it determines the corresponding topographic amplification factor based on the topographic slope and lithological parameters of the grid cell. Next, it determines the peak ground acceleration of the corresponding grid cell based on the initial PGA and the topographic amplification factor. Finally, it inputs the influencing factors and the peak ground acceleration into the landslide prediction model to obtain the prediction results. This scheme has better reliability than traditional schemes, higher scientific validity, and can more accurately predict coseismic landslides, thereby improving disaster prevention and mitigation capabilities. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments or prior art of this specification, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 The diagram shown is a flowchart illustrating the method for predicting coseismic landslides in megaseismic zones based on terrain amplification factor, as provided in the embodiments of this specification. Figure 2 The diagram shown is a structural schematic of the coseismic landslide prediction device based on terrain amplification factor provided in the embodiments of this specification. Detailed Implementation

[0016] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0017] like Figure 1The diagram shows a flowchart of the method for predicting coseismic landslides in meridional zones based on terrain amplification factors, as provided in the embodiments of this specification. Although this specification provides the method operation steps or device structures shown in the embodiments or figures below, based on conventional methods or without creative effort, the method or device may include more or fewer operation steps or module units after partial merging. In steps or structures where there is no logically necessary causal relationship, the execution order of these steps or the module structure of the device is not limited to the execution order or module structure shown in the embodiments or figures of this specification. When the method or module structure is applied in actual devices, servers, or terminal products, it can be executed sequentially or in parallel according to the method or module structure shown in the embodiments or figures (e.g., in a parallel processor or multi-threaded processing environment, or even in a distributed processing or server cluster implementation environment).

[0018] The method for predicting coseismic landslides in metronome zones based on terrain amplification factors provided in the embodiments of this specification is as follows: Figure 1 As shown, the method specifically includes the following steps: Step S101: Divide the metronome zone into multiple grid cells and determine the initial peak acceleration of each grid cell.

[0019] Specifically, the metronome zone is divided into grid cells with an accuracy of 30m, and the initial peak ground acceleration is calculated for each grid cell. This scheme determines the initial peak ground acceleration using the following formula. : ; In the formula, For the magnitude term, For distance, For site response items, For fault type items, This is the upper plate effect term.

[0020] In specific application scenarios, the calculation is based on the fault distance. Instead of the epicentral distance, the vertical distance from the point to the fault is calculated using the fault plane geometry equations. The specific process is as follows: (1) Take M=8.0 to calculate the magnitude term ; (2) Through ; (3) Based on the site item of the mesothermic zone Calculate the linear site effect and solve the nonlinear term iteratively (initial assumptions) ); (4) Calculate the dummy variable RS=1 for the reverse fault. Since the fault is nearly vertical, the hanging wall effect term is simplified to 0.

[0021] In the above, a, b, and h are constants obtained from the fitting, and the site reaction term... Calculate linear site effects using VS30 and iteratively find nonlinear terms; fault type term. The dummy variable RS for thrust faults is set to 1; for approximately vertical faults, the hanging wall effect term can be ignored. The coefficients mentioned above are obtained by regression fitting of strong ground motion records of strong earthquakes in the training area over the past few decades, i.e., the training data of the training area.

[0022] This method solves the prediction bias problem under high magnitude by coupling fault plane parameters with regional site characteristics.

[0023] Specifically, the metronome zone is divided into grid cells with an accuracy of 30m, and the peak ground acceleration (KPGA) of each grid cell in the study area, taking into account the topographic amplification effect, is calculated one by one to achieve regional characterization of historical ground motion parameters.

[0024] Step S102: Determine the corresponding terrain amplification factor based on the terrain slope and lithology parameters of the grid unit.

[0025] Specifically, in the embodiments of this application, the terrain magnification factor Specifically, it is determined using the following formula: ; In the formula, This is the terrain magnification factor. These are lithological parameters, that is, constants set based on geology. It is the angle between the terrain slope and the horizontal plane.

[0026] Specifically, This is a geologically based constant that reflects the sensitivity of stratigraphic lithology to topographic amplification, typically ranging from 0.5 to 1.0. Specifically, a lower value is recommended for bedrock or hard soil. For softer sedimentary soils, the a value needs to be increased (e.g., 0.5), while for softer sedimentary soils, the a value needs to be increased (e.g., 1.0) to reflect a more significant topographic amplification effect. This represents the angle (slope) between the terrain slope and the horizontal plane, and the unit can be degrees. By coupling the tangent value of the terrain slope with empirical coefficients, the degree of seismic amplification can be effectively quantified.

[0027] Step S103: Determine the peak acceleration of the corresponding grid cell based on the initial peak acceleration and the terrain magnification factor.

[0028] In this embodiment of the application, the peak acceleration is specifically determined by the following formula: ; In the formula, Peak acceleration, This is the terrain magnification factor. This represents the initial peak acceleration.

[0029] Step S104: Input the influencing factor and the peak acceleration into the landslide prediction model to obtain the prediction result.

[0030] Specifically, the influencing factors include at least: elevation, slope, aspect, lithology, distance from the fault, and normalized difference vegetation index. Before inputting the peak ground acceleration and influencing factors into the landslide prediction model to obtain the prediction result, the method further includes: Multiple pairs of related factors are obtained by performing pairwise correlation calculations on each training influencing factor. Remove one factor from each of all relevant factor pairs to obtain multiple initial factor selections; After removing unqualified factors from all the initial selection factors, the landslide prediction model is trained by combining the peak acceleration training to obtain a well-trained landslide prediction model.

[0031] Pearson correlation analysis was used to calculate the correlation coefficient of each initially selected influencing factor to obtain pairs of strongly correlated factors, i.e., related factor pairs. Then, based on the Gini coefficient of random forest, factors with lower feature importance were eliminated.

[0032] Furthermore, based on the principles of objective existence, significance, and inheritance, a total of 18 influencing factors were initially selected from seven aspects, including topography, stratigraphy, geological structure, rivers, roads, environmental geological characteristics, and seismic parameters, to participate in the regional landslide susceptibility assessment.

[0033] A rich factor dataset can avoid evaluation bias due to the omission of key factors, but it is necessary to ensure the independence between factors and avoid information redundancy that may cause errors in the machine learning model. Pearson correlation analysis showed that slope and topographic relief were strongly correlated (correlation coefficient: 0.96). In addition, factors with lower importance (topographic relief) were eliminated based on the Gini coefficient of random forest. The remaining 17 influencing factors, including elevation, slope, NDVI, KPGA, distance from fault, SPI, TWI, MNDWI, stratigraphic lithology, distance from first- or second-level rivers, distance from first- or second-level roads, aspect, plan / profile curvature, and land use type, participated in the coseismic landslide susceptibility assessment.

[0034] The Pearson correlation is calculated as follows: Obtain the raster values ​​of each initially selected influencing factor from the landslide-non-landslide sample set, denoted as . For any two influencing factors and The Pearson correlation coefficient is calculated using the following formula. : ; in, Impact Factor and The Pearson correlation coefficient between them; and The k-th sample is in the influence factor and The value of ; and Impact Factors and The average value across all samples; N is the total number of samples.

[0035] The landslide prediction model is specifically a random forest machine learning model, and the unqualified influencing factor is specifically an influencing factor whose impurity decrease is lower than a preset threshold. The impurity decrease is determined by the following formula: ; In the formula, The amount of decrease in impurity, The impurity of the parent node. This represents the proportion of samples in the left child node to samples in the parent node. The impurity of the left child node. This represents the proportion of samples in the right child node to the total number of samples in the parent node. The impurity of the right child node.

[0036] In the Gini impurity formula, a sample refers to a single data point used for modeling during the training phase, typically a raster cell containing features and labels. During the partitioning of each node, the parent node contains a certain number of training samples, which are used to calculate impurity and determine how samples are assigned to child nodes. Specifically, parent node samples refer to all training data used in the current node, while child node samples are data assigned to the left and right child nodes according to a splitting rule based on a certain feature. Each sample typically consists of its features (such as elevation, slope, KPGA, NDVI, etc.) and a label (landslide or non-landslide). By calculating the distribution of samples across different nodes, Gini impurity measures the uncertainty of samples after node splitting, thus determining the optimal features and splitting method. Therefore, a sample here refers to the data unit participating in model training and evaluation, typically a raster cell containing features and labels.

[0037] Elevation (13.4%), slope (8.0%), NDVI (7.9%), and KPGA (7.3%) have a significant impact on landslide development, contributing a cumulative 36.6% and becoming the dominant factors controlling landslide development. ① Elevation, as the most important influencing factor, directly reflects the degree of topographic relief. In high-altitude areas, the soil and rock masses are significantly affected by tectonic movements and weathering, resulting in the development of unloading fissures. In addition, the amplification effect of inertial forces under earthquakes (the higher the elevation, the more significant the component of seismic inertial force along the slope direction) provides inherent potential energy conditions for landslides. ② Slope is a key topographic parameter controlling the stability of soil and rock masses. In steep slope areas (e.g., >30°), the resistance to sliding of soil and rock masses decreases with increasing slope, while the component of gravity along the slope direction increases significantly. The shear force under earthquakes is more likely to exceed the resistance threshold, confirming the classic theory that "the steeper the slope, the higher the landslide susceptibility," directly determining the spatial susceptibility of landslides. ③ Vegetation enhances the anti-slide capacity of soil and rock masses through root stabilization and reduction of slope runoff erosion. In areas with low NDVI values, the soil and rock masses have poor resistance to erosion and shear, making them more prone to instability under earthquake triggering. ④ KPGA reflects the amplification effect of topography on seismic waves. In areas with complex topography (such as deep valleys and around steep cliffs), the ground motion acceleration is significantly amplified, increasing the dynamic load on the soil and rock masses and making them more likely to exceed the stability limit (Xu Qiang et al., 2015). The study area is affected by tectonic movement, with dramatic topographic undulations, resulting in a prominent topographic amplification effect of ground motion. The high contribution of KPGA reveals the direct control of landslides by the topographic-ground motion coupling effect. That is, topography not only provides potential energy conditions (elevation, slope) but also further enhances landslide susceptibility by amplifying the intensity of ground motion.

[0038] After removing strongly correlated factors and determining the final influencing factors for modeling, the random forest model is retrained. For each influencing factor... The total importance of this factor is obtained by summing the decreases in Gini impurity across all trees and all nodes and averaging the results. Then normalize it to a percentage of contribution using the following formula. : ; Where n represents the final number of influencing factors retained, i.e., all initially selected factors. Thus, the contributions of factors such as elevation, slope, NDVI, and KPGA are approximately 13.4%, 8.0%, 7.9%, and 7.3%, respectively.

[0039] Factors with lower rankings (such as land use type, 1.8%; TWI, 2.8%) all contributed less than 3%, indicating a weak direct impact on landslide development. The study area is predominantly natural landform, with limited human engineering activities altering the terrain. Land use change has a limited direct impact on landslides. While TWI characterizes runoff concentration capacity, its impact on landslides is only apparent through long-term groundwater infiltration (softening of soil and rock). In contrast, coseismic landslides are primarily controlled by instantaneous dynamic conditions, and the long-term effects of human land use are not fully reflected in the short timescale (the instant of earthquake triggering). The short-term mechanical effects of hydrological factors are limited, hence their low contribution.

[0040] Specifically, the random forest model is more widely used and performs better than other machine learning models (Park et al. 2021; Huang et al. 2022), achieving higher prediction accuracy (Guilherme et al. 2019). This study aims to conduct landslide susceptibility assessment based on an optimized random forest model from three aspects: landslide-non-landslide sample selection, influencing factor selection, and hyperparameter optimization. Specifically: (1) In the selection of landslide-non-landslide samples, a 50m buffer zone is established based on the post-disaster interpretation of the mass landslide surface to determine the non-landslide selection area. An equal amount of non-landslide data is selected by random sampling to form a landslide-non-landslide modeling sample. The method of establishing a buffer zone can effectively prevent the selection of non-landslide samples from falling into the landslide area.

[0041] (2) In terms of the selection of influencing factors, Pearson correlation analysis was used to calculate the correlation coefficient of each initially selected influencing factor to obtain pairwise strongly correlated factors. Furthermore, the Gini coefficient of random forest was used to remove those with less important features, thereby solving the problem of autocorrelation of each influencing factor and avoiding information redundancy caused by excessive or repeated consideration of related factors when selecting parameters.

[0042] (3) In terms of hyperparameter optimization, the genetic algorithm is used to adaptively and heuristically optimize the hyperparameters of the random forest model based on the input samples, avoiding the uncertainty caused by subjective adjustment of the model hyperparameter settings.

[0043] 1. Selection of landslide and non-landslide samples Based on the landslide surface grid interpreted after the disaster, a buffer zone of a predetermined width (e.g., 50m) is constructed outside the landslide boundary. Grid cells within this buffer zone are marked as no-sampling areas to prevent potentially disturbed areas adjacent to the landslide zone from being mistakenly identified as non-landslide samples. In the remaining area outside the buffer zone, non-landslide sample grid cells are selected using random sampling, resulting in a balanced landslide-non-landslide sample set. This can be considered a technical effect in substantive review of "improving the reliability of non-landslide sample selection and reducing labeling noise."

[0044] 2. Optimizing Random Forest with Genetic Algorithm The key hyperparameters of the random forest model (such as the number of decision trees, maximum tree depth, minimum number of samples per leaf node, etc.) are encoded as chromosome vectors, and adaptive optimization is performed using a genetic algorithm through the following steps: 1) Randomly initialize several sets of hyperparameter combinations as the initial population; 2) Calculate the fitness of each individual using the AUC value on the validation set as the fitness function; 3) Select individuals with high fitness using selection operators such as roulette or tournament selection; 4) Perform crossover and mutation operations on the selected individuals to generate a new generation population; 5) Repeat steps 2)–4) until the preset number of iterations or fitness convergence is reached; 6) Use the hyperparameter combination with the highest fitness as the hyperparameter setting for the final random forest model.

[0045] More specifically, for the random forest model, the parameter candidate range is set as follows: based on the landslide susceptibility evaluation scenario and factor dimensions, the key hyperparameter candidate intervals are set as follows: number of decision trees n_estimators (50-500, step size 50), maximum tree depth max_depth (5-30, step size 5), minimum number of sample splits min_samples_split (2-20, step size 2), minimum number of leaf node samples min_samples_leaf (1-10, step size 1), and feature sampling ratio max_features (0.3-1.0, step size 0.1).

[0046] Genetic algorithm parameter settings: Combining the sample size (≥20,000) and the hyperparameter dimensions (5), the genetic algorithm parameters are set as follows: population size 50, number of iterations 100, crossover probability 0.8, mutation probability 0.1, and the fitness function is the AUC value (Area Under ROC Curve, representing the model's prediction accuracy) of the random forest model on the validation set. Hyperparameter encoding and optimization: The five hyperparameters are converted into chromosomes using binary encoding (the encoding length is determined according to the candidate range of each hyperparameter). Through selection (roulette wheel selection), crossover (single-point crossover), and mutation (bit mutation) operations, a new generation of population is generated. Each individual in each generation of population corresponds to a set of hyperparameter combinations. Model training and fitness calculation: For each combination of hyperparameters in each generation of the population, a random forest model is constructed, trained with the training set as input, and the model AUC value is calculated using the validation set as the fitness value; Determining the optimal hyperparameters: When the genetic algorithm reaches the maximum number of iterations (100 generations) or the fitness value change is ≤0.001 for 10 consecutive generations, the optimization is stopped, and the hyperparameter combination corresponding to the highest fitness value (highest AUC value) is selected as the optimal hyperparameters. The specific hyperparameter settings are shown in Table 1 below: Table 1 Parameter name Optimized parameter value Default parameter value Training time 46 hours 34 minutes 2s Number of decision trees 70 100 Minimum number of samples for internal node splits 54 2 Minimum number of samples for leaf nodes 45 1 Maximum depth of trees 20 10 Maximum number of leaf nodes 103 50 Final model construction: Based on the optimal combination of hyperparameters, the random forest model is retrained using the training set to obtain the optimized landslide susceptibility evaluation model.

[0047] The construction process of a random forest specifically includes: S201. Randomly select a portion of the samples (with replacement).

[0048] S202. Randomly select some features as candidate features.

[0049] S303. Use the Gini index to determine the test features from the candidate features.

[0050] S304. Generate nodes based on test features.

[0051] S305. Determine if the generated node can become a leaf node. If yes, execute S306; otherwise, branch and execute S302.

[0052] S306. Determine if the decision tree has stopped growing. If yes, execute S307; otherwise, branch and execute S302.

[0053] S307, Store the decision tree.

[0054] S308. Determine if the number of decision trees meets the requirement. If yes, generate a random forest; otherwise, execute S301.

[0055] Based on the above-described method for predicting coseismic landslides in mesastric zones based on terrain amplification factors, one or more embodiments of this specification also provide a platform or terminal for predicting coseismic landslides in mesastric zones based on terrain amplification factors. This platform or terminal may include devices, software, modules, plug-ins, servers, clients, etc., using the methods described in the embodiments of this specification, combined with necessary hardware implementation. Based on the same innovative concept, the systems in one or more embodiments provided in this specification are as described in the following embodiments. Since the implementation schemes and methods for solving the system problem are similar, the specific system implementation in the embodiments of this specification can refer to the implementation of the aforementioned methods. Repeated descriptions will not be repeated. The terms "unit" or "module" used below can refer to a combination of software and / or hardware that achieves a predetermined function. Although the systems described in the following embodiments are preferably implemented in software, hardware implementation, and a combination of software and hardware, are also possible and contemplated.

[0056] Specifically, Figure 2 This is a schematic diagram of the module structure of an embodiment of the coseismic landslide prediction device based on terrain amplification factor in the mesastric zone provided in this specification, as shown below. Figure 2As shown, the coseismic landslide prediction device based on topographic amplification factor in the mesastric zone provided in this specification includes: The partitioning module 201 is used to divide the metronome zone into multiple grid cells and determine the initial peak acceleration of each grid cell. Terrain module 202 is used to determine the corresponding terrain magnification factor based on the terrain slope and lithology parameters of the grid cell; Peak acceleration module 203 is used to determine the peak acceleration of the corresponding grid cell based on the initial peak acceleration and the terrain magnification factor; The prediction module 204 is used to input the influencing factors and the peak acceleration into the landslide prediction model to obtain the prediction results.

[0057] It should be noted that the system described above may include other implementation methods based on the description of the corresponding method embodiments. The specific implementation methods can be referred to the description of the corresponding method embodiments above, and will not be elaborated here.

[0058] This application also provides an electronic device, including: processor; Memory used to store the processor's executable instructions; The processor is configured to perform the methods provided in the embodiments described above.

[0059] The electronic device provided in this application embodiment stores executable instructions of the processor in a memory. When the processor executes the executable instructions, it can first divide the mesastric zone into multiple grid cells and determine the initial peak acceleration of each grid cell; then, it determines the corresponding topographic amplification factor based on the topographic slope and lithological parameters of the grid cell; next, it determines the peak acceleration of the corresponding grid cell based on the initial peak acceleration and the topographic amplification factor; finally, it inputs the influencing factor and the peak acceleration into the landslide prediction model to obtain the prediction result. This scheme has better reliability than traditional schemes, has higher scientific validity, and can more accurately predict coseismic landslides, thereby improving the ability to prevent and mitigate disasters.

[0060] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0061] The methods or apparatus described in the embodiments provided in this specification can implement business logic through a computer program and record it on a storage medium. The storage medium can be read and executed by a computer to achieve the effects of the solutions described in the embodiments of this specification, such as: The metronome zone is divided into multiple grid cells, and the initial peak acceleration of each grid cell is determined. The corresponding terrain amplification factor is determined based on the terrain slope and lithological parameters of the grid unit; The peak acceleration of the corresponding grid cell is determined based on the initial peak acceleration and the terrain magnification factor; The influencing factors and the peak acceleration are input into the landslide prediction model to obtain the prediction results.

[0062] The storage medium can include physical devices for storing information, typically digitizing the information and then storing it using electrical, magnetic, or optical methods. The storage medium can include: devices that store information using electrical energy, such as various types of memory, like RAM and ROM; devices that store information using magnetic energy, such as hard disks, floppy disks, magnetic tapes, magnetic core memory, bubble memory, and USB flash drives; and devices that store information using optical methods, such as CDs or DVDs. Of course, there are other readable storage media, such as quantum memories and graphene memories.

[0063] The embodiments in this specification are not limited to conforming to industry communication standards, standard computer resource data update and data storage rules, or the situations described in one or more embodiments of this specification. Slightly modified implementations based on certain industry standards or custom methods or embodiments can also achieve the same, equivalent, or similar, or predictable, implementation effects as described above. Embodiments that utilize these modified or modified methods for data acquisition, storage, judgment, and processing still fall within the scope of optional implementations of the embodiments in this specification.

[0064] The controller can be implemented in any suitable manner. For example, it can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code form, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, ASICs, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included therein for implementing various functions can also be considered as structures within the hardware component. Alternatively, the means for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0065] The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or plug-ins may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection between devices or units, and may be electrical, mechanical, or other forms.

[0066] These computer program instructions can also be loaded onto a computer or other programmable resource data updating device, causing a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable device for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0067] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, system embodiments are basically similar to method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. In the description of this specification, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this specification. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described can be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0068] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A method for predicting coseismic landslides in megaseismic zones based on topographic amplification factors, characterized in that, The method includes: The metronome zone is divided into multiple grid cells, and the initial peak acceleration of each grid cell is determined. The corresponding terrain amplification factor is determined based on the terrain slope and lithological parameters of the grid unit; The peak acceleration of the corresponding grid cell is determined based on the initial peak acceleration and the terrain magnification factor; The influencing factors and the peak acceleration are input into the landslide prediction model to obtain the prediction results.

2. The method for predicting coseismic landslides in megaseismic zones based on topographic amplification factor as described in claim 1, characterized in that, The peak acceleration is determined by the following formula: ； In the formula, Peak acceleration, This is the terrain magnification factor. This represents the initial peak acceleration.

3. The method for predicting coseismic landslides in megaseismic zones based on topographic amplification factor as described in claim 2, characterized in that, The terrain magnification factor is specifically determined by the following formula: ； In the formula, This is the terrain magnification factor. These are lithological parameters, that is, constants set based on geology. It is the angle between the terrain slope and the horizontal plane.

4. The method for predicting coseismic landslides in megaseismic zones based on topographic amplification factor as described in claim 2, characterized in that, The initial peak acceleration is determined by the following formula: ； In the formula, For the magnitude term, For distance, For site response items, For fault type items, This is the upper plate effect term.

5. The method for predicting coseismic landslides in megaseismic zones based on topographic amplification factor as described in claim 1, characterized in that, The influencing factors include at least: elevation, slope, aspect, lithology, distance from the fault, and normalized vegetation index.

6. The method for predicting coseismic landslides in megaseismic zones based on topographic amplification factor as described in claim 1, characterized in that, Before inputting the peak acceleration and influencing factors into the landslide prediction model to obtain the prediction result, the method further includes: Multiple pairs of related factors are obtained by performing pairwise correlation calculations on each training influencing factor. Remove one factor from each of all relevant factor pairs to obtain multiple initial factor selections; After removing unqualified factors from all the initial selection factors, the landslide prediction model is trained by combining the peak acceleration training to obtain a well-trained landslide prediction model.

7. The method for predicting coseismic landslides in megaseismic zones based on topographic amplification factor as described in claim 6, characterized in that, The landslide prediction model is specifically a random forest machine learning model, and the unqualified influencing factor is specifically an influencing factor whose impurity decrease is lower than a preset threshold. The impurity decrease is determined by the following formula: ； In the formula, The amount of decrease in impurity, The impurity of the parent node. This represents the proportion of samples in the left child node to samples in the parent node. The impurity of the left child node. This represents the proportion of samples in the right child node to the total number of samples in the parent node. The impurity of the right child node.

8. A device for predicting coseismic landslides in megaseismic zones based on terrain amplification factor, characterized in that, The device includes: The partitioning module is used to divide the metronome zone into multiple grid cells and determine the initial peak acceleration of each grid cell. The terrain module is used to determine the corresponding terrain magnification factor based on the terrain slope and lithology parameters of the grid cells; A peak acceleration module is used to determine the peak acceleration of the corresponding grid cell based on the initial peak acceleration and the terrain magnification factor; The prediction module is used to input the influencing factors and the peak acceleration into the landslide prediction model to obtain the prediction results.

Images