Rainstorm-debris flow disaster chain threshold identification method based on causal decoupling

Abstract

The application provides a storm-mudflow-debris flow disaster chain threshold identification method based on causal decoupling. A MassFlow physical mechanism model is used to perform evolution simulation on a storm-mudflow-debris flow disaster chain prone area, and a disaster chain sample database is obtained. Based on the disaster chain sample database, a first structural equation and a second structural equation are respectively constructed, and standardized path coefficients of different driving factors are determined. According to the standardized path coefficients, direct effect values, interaction effect values and total effect values in the occurrence and evolution process of the storm-mudflow-debris flow disaster chain, main driving factors are identified. The main driving factors are input into a generalized additivity model to obtain debris flow intensity, and a three-dimensional disaster-causing response surface of the storm-mudflow-debris flow disaster chain is established. Based on the three-dimensional disaster-causing response surface, a critical threshold value for inducing the storm-mudflow-debris flow disaster chain is determined. Through the above technical solution, the accuracy of threshold determination is improved.

Description

Technical Field

[0001] This application relates to the field of natural disaster management technology, and in particular to a method for identifying the threshold of a rainstorm-flash flood-debris flow disaster chain based on causal decoupling. Background Technology

[0002] Floods are among the most frequent and widespread natural disasters. Against the backdrop of global change, extreme precipitation triggers frequent and intense flash floods and basin-wide floods, which can easily induce secondary disasters such as debris flows and landslides, forming a multi-hazard chain of flood disasters and posing a new challenge to flood prevention and control. Furthermore, extreme precipitation, while easily leading to floods, can also easily trigger geological disasters such as debris flows in affected areas with steep terrain and loose soil, forming a rainstorm-flash flood-debris flow disaster chain, triggering a "multiple disasters in one area" phenomenon and a snowballing accumulation of damage.

[0003] Currently, while there is considerable research both domestically and internationally on the "occurrence-evolution" of two-hazard flood disaster chains, research on three-hazard disaster chains—rainstorm-flash flood-debris flow—is relatively limited. Existing technologies still have the following shortcomings in researching the main driving factors of disaster chains and their multivariate trigger thresholds: (1) The rainstorm-flash flood-debris flow disaster chain has extremely strong instantaneousness and high destructiveness, but its historical sample data is scarce and the research data foundation is weak, resulting in very limited reliability of the identified main driving factors and multivariate trigger thresholds; (2) In related technologies, mathematical statistical methods such as causal Bayesian networks are usually used to identify the driving factors and multivariate trigger thresholds of the disaster chain. However, such mathematical statistical methods are essentially black box models that lack explicit causal relationships. They cannot clearly characterize the disaster chain formation mechanism induced by factors such as rainfall infiltration, increased soil moisture, pore water pressure response, and slope instability, and it is difficult to accurately identify the main driving factors of the disaster chain; (3) Related mathematical statistical methods often ignore the interaction between driving factors and regard multiple factors as independent input features, resulting in the disaster chain trigger threshold analysis being disconnected from the internal physical mechanism, and the calculated threshold is not accurate.

[0004] Therefore, there is an urgent need for a multivariate triggering threshold identification method for rainstorm-flash flood-debris flow disaster chains. By decoupling the direct disaster-causing effects and interactions of each driving factor, the main driving factors can be identified, and the multivariate triggering threshold of the rainstorm-flash flood-debris flow disaster chain can be calculated, providing technical support for the early warning and prevention of flood chain disasters. Summary of the Invention

[0005] This application provides a threshold identification method for rainstorm-flash flood-debris flow disaster chains based on causal decoupling. This method solves the problems in the prior art of accurately identifying the main driving factors of disaster chains and inaccurate calculation of disaster chain trigger thresholds. It achieves accurate identification of the main driving factors of disaster chains and improves the accuracy of determining the multivariate trigger thresholds of rainstorm-flash flood-debris flow disaster chains.

[0006] To achieve the above objectives, embodiments of this application provide a threshold identification method for rainstorm-flash flood-debris flow disaster chains based on causal decoupling, the method comprising: Based on the MassFlow physical mechanism model, the evolution of the disaster-prone areas of the rainstorm-flash flood-debris flow disaster chain was simulated, and a disaster chain sample database under various scenarios of rainfall and soil moisture content was obtained. Based on the disaster chain sample database, the first structural equation for the triggering path of the rainstorm-flash flood disaster chain and the second structural equation for the cascading path of the rainstorm-flash flood-debris flow disaster chain are constructed respectively. Solve the first and second structural equations to determine the standardized path coefficients for different driving factors; Based on the direct effect value, interaction effect value, and total effect value of the standardized path coefficient in the occurrence and evolution of the rainstorm-flash flood-debris flow disaster chain, the main driving factors of the rainstorm-flash flood-debris flow disaster chain are identified. The main driving factors are input into a generalized additivity model to obtain the debris flow intensity under the combined action of the main driving factors; and a three-dimensional disaster response surface of the rainstorm-flash flood-debris flow disaster chain is established based on the debris flow intensity. Based on the three-dimensional disaster response surface, the critical space of driving factors that trigger the rainstorm-flash flood-debris flow disaster chain is identified, and then the critical threshold for inducing the rainstorm-flash flood-debris flow disaster chain is determined.

[0007] In one possible implementation, the MassFlow physics mechanism model is used to simulate the evolution of the rainstorm-flash flood-debris flow disaster chain in prone areas, resulting in a disaster chain sample database with rainfall and soil moisture content under various scenarios, including: Determine the slope stability coefficient of the target area under heavy rain conditions, and identify the vulnerable areas based on the slope stability coefficient; Based on historical information on rainstorms, floods, and debris flows in the prone areas, as well as DEM data and soil physical and mechanical parameters, the boundary and dynamic parameters of the MassFlow physical mechanism model are determined. Based on historical disaster data of the prone areas, the friction coefficient, turbulence coefficient, and debris flow unit weight parameters of the MassFlow physical mechanism model are determined. Rainfall and soil moisture scenarios with different return periods were designed to drive the MassFlow physical mechanism model, simulating the occurrence and evolution of the rainstorm-flash flood-debris flow disaster chain, and obtaining the disaster chain sample database.

[0008] In one possible implementation, the construction of a first structural equation for the triggering path of a rainstorm-flash flood disaster chain and a second structural equation for the cascading path of a rainstorm-flash flood-debris flow disaster chain, based on the disaster chain sample database, includes: Extract the characteristic variables of disasters in the evolution process from the disaster chain sample database, and classify the characteristic variables into exogenous variables, mediating variables and endogenous variables; Based on the aforementioned feature variables, the first structural equation and the second structural equation are constructed respectively.

[0009] In one possible implementation, solving the first structural equation and the second structural equation to determine the standardized path coefficients of different driving factors includes: The first structural equation and the second structural equation are solved iteratively using the maximum likelihood estimation method to determine the standardized path coefficients; wherein the standardized path coefficients are the first coefficient, the second coefficient, and the third coefficient; the objective is to minimize the difference between the sample covariance matrix and the model latent covariance matrix.

[0010] In one possible implementation, identifying the main driving factors of the rainstorm-flash flood-debris flow disaster chain based on the direct effect values, interaction effect values, and total effect values ​​of the standardized path coefficients during the occurrence and evolution of the rainstorm-flash flood-debris flow disaster chain includes: The second coefficient is used as the direct effect value; the direct effect value is used to reflect the strength of the direct influence of the exogenous variable or the mediating variable on the endogenous variable; The product of the first coefficient and the third coefficient is taken as the interaction effect value; the interaction effect value is used to reflect the interaction effect of the exogenous variable on the endogenous variable through multiple mediating variables; The sum of the direct effect and the interaction effect is taken as the total effect value; Arrange the multiple total effect values ​​from highest to lowest, and then... n The total effect value is used as the main driving factor.

[0011] In one possible implementation, the generalized additivity model is determined based on the range of variation of the main driving factors in the sample database of rainstorm-flash flood-debris flow disaster chain.

[0012] In one possible implementation, the step of identifying the critical space of driving factors triggering the rainstorm-flash flood-debris flow disaster chain based on the three-dimensional disaster response surface, and then determining the critical threshold for inducing the rainstorm-flash flood-debris flow disaster chain, includes: Based on the extreme points of the sum of the second-order partial derivatives of the main driving factors on the three-dimensional disaster response surface, the abrupt change region of debris flow disaster is determined as the critical space of the driving factors. Within the critical space of driving factors, and in conjunction with the debris flow disaster level classification standard, critical values ​​for debris flow volume are set for different levels. Based on the critical value of debris flow volume, contour lines are drawn on the three-dimensional disaster response surface and projected onto the coordinate system of the main driving factors to obtain the critical thresholds under different levels of debris flow disasters.

[0013] The technical solutions provided in this application embodiment have at least the following technical effects or advantages: (1) Based on historical disaster data of prone areas, the MassFlow physical mechanism model was determined to simulate the occurrence and evolution of the rainstorm-flash flood-debris flow disaster chain under different design rainfall-soil moisture content combination scenarios, and a disaster chain sample database was obtained, which solved the problem of scarce research data; (2) By constructing the structural equations of the triggering path of the rainstorm-flash flood disaster chain and the cascading path of the rainstorm-flash flood-debris flow disaster chain, the standardized path coefficients are calculated, thereby quantifying the direct effects, interaction effects and total effects of each driving factor, decoupling the direct impact and interaction of the driving factors on the disaster chain, and accurately identifying the main driving factors of the rainstorm-flash flood-debris flow disaster chain; (3) By using the generalized additivity model and the three-dimensional disaster response surface, the multivariate triggering thresholds under different levels of debris flow disasters were determined, which improved the accuracy of threshold determination. Attached Figure Description

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

[0015] Figure 1 A flowchart illustrating a threshold identification method for a rainstorm-flash flood-debris flow disaster chain based on causal decoupling, provided in this application embodiment; Figure 2 A flowchart of another threshold identification method for rainstorm-flash flood-debris flow disaster chain based on causal decoupling provided in this application embodiment. Detailed Implementation

[0016] The technical solutions of 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, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0017] In the description of the embodiments of this application, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the embodiments of this application and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application. The terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, the terms "installed," "connected," and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; a mechanical connection or an electrical connection; a direct connection or an indirect connection through an intermediate medium; or a connection within two components. Those skilled in the art can understand the specific meaning of the above terms in the embodiments of this application according to the specific circumstances.

[0018] Figure 1 A flowchart illustrating a threshold identification method for a rainstorm-flash flood-debris flow disaster chain based on causal decoupling, provided as an embodiment of this application. Figure 1 As shown, the method may include the following steps.

[0019] S101. Based on the MassFlow physical mechanism model, the evolution of the disaster-prone areas of the rainstorm-flash flood-debris flow disaster chain is simulated, and a disaster chain sample database under various scenarios of rainfall-soil moisture content is obtained.

[0020] S102. Based on the disaster chain sample database, construct the first structural equation for the triggering path of the rainstorm-flash flood disaster chain and the second structural equation for the cascading path of the rainstorm-flash flood-debris flow disaster chain.

[0021] S103. Solve the first structural equation and the second structural equation to determine the standardized path coefficients of different driving factors.

[0022] S104. Based on the direct effect value, interaction effect value, and total effect value of the standardized path coefficient in the occurrence and evolution of the rainstorm-flash flood-debris flow disaster chain, identify the main driving factors of the rainstorm-flash flood-debris flow disaster chain.

[0023] S105. Input the main driving factor into the generalized additivity model to obtain the debris flow intensity under the combined action of the main driving factor; and based on the debris flow intensity, establish a three-dimensional disaster response surface of the rainstorm-flash flood-debris flow disaster chain.

[0024] S106. Based on the three-dimensional disaster response surface, identify the critical space of driving factors that trigger the rainstorm-flash flood-debris flow disaster chain, and then determine the critical threshold for inducing the rainstorm-flash flood-debris flow disaster chain.

[0025] Based on the above technical solution, structural equations can be constructed to establish the triggering path of the rainstorm-flash flood disaster chain and the cascading path of the rainstorm-flash flood-debris flow disaster chain. Standardized path coefficients can then be calculated to quantify the direct effects, interaction effects, and total effects of each driving factor. This decouples the direct impact and interaction of driving factors on the disaster chain, accurately identifying the main driving factors of the rainstorm-flash flood-debris flow disaster chain. Furthermore, by using a generalized additivity model and a three-dimensional disaster-causing response surface, multivariate triggering thresholds under different levels of debris flow disasters can be determined, improving the accuracy of threshold determination.

[0026] In one possible implementation, S101 includes: determining the slope stability coefficient of the target area under rainstorm conditions, and identifying the disaster-prone area based on the slope stability coefficient; determining the boundary and dynamic parameters of the MassFlow physical mechanism model based on historical rainstorm, flood, and debris flow disaster information, DEM (Digital Elevation Model) data, and soil physical and mechanical parameters in the disaster-prone area; determining the friction coefficient, turbulence coefficient, and debris flow unit weight parameters of the MassFlow physical mechanism model based on historical disaster data of the disaster-prone area; designing rainfall scenarios and soil moisture content scenarios under different return periods to drive the MassFlow physical mechanism model, simulating the occurrence and evolution process of the rainstorm-flash flood-debris flow disaster chain, and obtaining a disaster chain sample database.

[0027] For example, the slope stability coefficient It can be determined using the following formula: ;in, Indicates soil thickness; Indicates the duration of rainfall; Indicates the soil friction angle; Indicates the slope angle; Indicates soil cohesion; Indicates the pore water pressure head; This refers to the unit weight of water. This represents the unit weight of the soil.

[0028] For example, this MassFlow physical mechanism model, based on the Voellmy model, determines slope confluence, simulates water transport into channels, and the evolution process within channels; the governing equations of the Voellmy model are as follows: , ; in, Indicates the depth of the mudslide; Indicates time; Represents gravitational acceleration; , These represent mudslides in , Average flow velocity in the direction; Indicates rainfall runoff; , express , Resistance shear stress in the direction; Indicates the density of debris flow; Indicates terrain elevation; Indicates the slope angle; Indicates the Coulomb coefficient of friction; This represents the turbulent friction parameters.

[0029] In one possible implementation, S102 may include: extracting characteristic variables of disasters in the evolution process from the disaster chain sample database, and classifying the characteristic variables into exogenous variables, mediating variables and endogenous variables; and constructing the first structural equation and the second structural equation based on the characteristic variables.

[0030] For example, this exogenous variable can characterize the external triggering conditions of a disaster chain, including the maximum hourly rainfall intensity. (mm / h) Rainfall (mm), duration of rainfall (h) Soil moisture content in the early stage (%),slope (°), soil type, etc. This mediating variable can characterize the variables in the evolution of the disaster chain, including soil moisture content. (%) Surface runoff (m³ / s), slope stability ditch peak flow (m³ / s), maximum mud depth (m) Maximum base shear stress (kPa). This endogenous variable can characterize the final output of the disaster chain, including the peak flow rate of debris flows. (m³ / s), destroyed area (㎡), debris flow velocity (m / s) and debris flow volume (m³).

[0031] For example, the first structural equation can be: The second structure equation can be: ;in, The dynamic characteristics of flash floods in the sample data; For exogenous variables, the path coefficients of flash floods; For residual terms; This represents the maximum hourly rainfall intensity. Rainfall; This refers to the initial soil moisture content; Represents the scale of a debris flow; The path coefficient for rainwater directly triggering debris flows; The path coefficient of debris flow is the effect of flash flood dynamic characteristics on the path coefficient of debris flow. This is the residual term.

[0032] In one possible implementation, S103 may include: using the maximum likelihood estimation method to iteratively solve the first structural equation and the second structural equation to determine the standardized path coefficients; wherein the standardized path coefficients are the first coefficient, the second coefficient, and the third coefficient; the solution objective is to minimize the difference between the sample covariance matrix and the model latent covariance matrix.

[0033] For example, the objective function can be expressed by the formula: ; This represents the maximum likelihood objective function value; The sample covariance matrix; This represents the model's implicit covariance matrix; The number of observed variables; the result is obtained through iterative calculation. The smallest parameter matrix yields all standardized path coefficients.

[0034] In one possible implementation, S104 may include: using the second coefficient as the direct effect value; the direct effect value being used to reflect the strength of the direct influence of the exogenous variable or the mediating variable on the endogenous variable; using the product of the first coefficient and the third coefficient as the interaction effect value; the interaction effect value being used to reflect the interaction effect of the exogenous variable on the endogenous variable through multiple mediating variables; using the sum of the direct effect and the interaction effect as the total effect value; arranging the multiple total effect values ​​from high to low, and placing the top... n The total effect value is used as the main driving factor.

[0035] For example, the first, second, and third coefficients can be represented as follows: The direct effect value can be expressed as: For example, rainfall directly impacting a slope can trigger a landslide, which then transforms into a debris flow; this interaction effect can be expressed as... For example, rainfall can cause flash floods, which further erode gully beds and trigger debris flows; the total effect can be expressed as This represents the total effect of the driving factors in the rainstorm-flash flood-debris flow disaster chain. The process of determining the direct effect value, interaction effect value, and total effect value can be expressed as: .

[0036] In one possible implementation, the generalized additivity model is determined based on the range of variation of the main driving factors in a sample database of rainstorm-flash flood-debris flow disaster chains.

[0037] For example, the governing equations of this generalized additivity model (GAM-te) are as follows: ;in, Represents the intercept term; For the tensor product smoothing term, The operator can capture the nonlinear relationship between multiple variables simultaneously through the product of Laplace smoothing basis functions; The sum of the additional smoothing terms represents other environmental variables that affect the disaster; Indicates random error; Represents the volume of debris flow among exogenous variables; express One of the main driving factors.

[0038] In one possible implementation, S106 may include: determining the abrupt change region of debris flow disaster based on the extreme points of the sum of the second-order partial derivatives of the main driving factor on the three-dimensional disaster response surface, as the critical space of the driving factor; setting critical values ​​for debris flow volume of different levels in the critical space of the driving factor, in conjunction with the debris flow disaster level classification standard; drawing contour lines on the three-dimensional disaster response surface based on the critical value of debris flow volume, projecting them onto the coordinate system of the main driving factor, and obtaining the critical threshold under different levels of debris flow disaster.

[0039] For example, the classification standard for debris flow disaster levels can be shown in Table 1 below.

[0040] Table 1: Figure 2 A flowchart of another threshold identification method for rainstorm-flash flood-debris flow disaster chains based on causal decoupling provided in this application embodiment. The following is in conjunction with... Figure 2 The workflow of the technical solution in this application is described.

[0041] S201. Simulation of a rainstorm-flash flood-debris flow disaster chain based on the MassFlow physics mechanism model.

[0042] By using the MassFlow physics mechanism model, the formation process of the rainstorm-flash flood-debris flow disaster chain is simulated, and disaster chain samples under various scenarios of rainfall and soil moisture content are obtained. This solves the problem of scarce disaster chain sample data and provides data support for decoupling the complex causal path and identifying the trigger threshold of the rainstorm-flash flood-debris flow disaster chain.

[0043] The MassFlow physical mechanism model can quantitatively characterize the transport, scouring, and deposition processes of materials such as water flow and debris along channels. Through numerical solutions, it can simulate the formation process of a rainstorm-flash flood-debris flow disaster chain. Specifically, the MassFlow physical mechanism model, based on the Voellmy model, calculates slope runoff, simulates water transport into channels, and the evolution process within channels. When a rainstorm causes a peak water depth in a channel unit... Exceeding the warning depth for this area Or peak bandwidth Exceeding the warning threshold set for this area At this time, torrential rain will lead to flash floods. After a flash flood occurs, in unstable slope units such as mountain valleys, if the input energy of the flash flood exceeds the critical kinetic energy of unstable material sources such as mud, sand, and rocks, these unstable material sources will enter the gully and maintain material transport and gully erosion. At this point, the torrential rain-flash flood disaster chain will continue to evolve into a torrential rain-flash flood-debris flow disaster chain. The governing equations of the Voellmy model are as follows: , ; in: Indicates the depth of the mudslide; , These represent mudslides in , Average flow velocity in the direction; Indicates rainfall runoff; , express , Resistance shear stress in the direction; Indicates the density of debris flow; Indicates terrain elevation; Indicates the slope angle; Indicates the Coulomb coefficient of friction; This represents the turbulent friction parameters.

[0044] The above S201 may include the following steps.

[0045] (1) Identify the areas prone to the rainstorm-flash flood-debris flow disaster chain. Taking a river basin as an example, determine the slope stability coefficient of the target area under rainstorm conditions. The dynamic changes in pore water pressure are used to characterize slope stability, thereby identifying areas prone to the rainstorm-flash flood-debris flow disaster chain.

[0046] In the soil saturated infiltration zone, when the slope stability coefficient A slope stability coefficient <1 indicates that after rainfall enters the soil, the sliding force acting on the slope surface has exceeded the maximum shear strength that the soil can provide, and the slope is in an unstable state. Under continuous rainfall conditions, the soil will shift, leading to derivative disasters such as flash floods and debris flows, and this area can be identified as a high-risk zone for the rainstorm-flash flood-debris flow disaster chain. The calculation formula is as follows: ;in: Indicates soil thickness; Indicates the duration of rainfall; Indicates the soil friction angle; Indicates the slope angle; Indicates soil cohesion; Indicates the pore water pressure head; This refers to the unit weight of water. This represents the unit weight of the soil.

[0047] (2) In the identified areas prone to the rainstorm-flash flood-debris flow disaster chain, collect historical rainstorm, flood, and debris flow disaster information, high-precision DEM data, and physical and mechanical parameters of the soil to determine the boundary and dynamic parameters (e.g., cohesion, earth pressure coefficient, etc.) of the model simulation. Based on historical disaster data, determine the friction coefficient of the MassFlow physical mechanism model. turbulence coefficient and the density of mudslides Parameters such as these.

[0048] (3) Set up different return periods from 2 years to 200 years for design rainfall scenarios (e.g., different rainfall durations, rainfall intensity and rainfall amount from 1h to 72h) and soil moisture scenarios to drive the MassFlow physical mechanism model to simulate the occurrence and evolution of the rainstorm-flash flood-debris flow disaster chain.

[0049] Based on the peak flow rate output by the model Depth of mudslide Flow rate The spatiotemporal evolution process and final accumulation range were studied, and the characteristics of rainstorm, flash flood and debris flow disasters under different scenarios were extracted to construct a sample database of rainstorm-flash flood-debris flow disaster chains under different design rainfall-soil moisture content combination scenarios.

[0050] S202. Decoupling of the main driving factors of the rainstorm-flash flood-debris flow disaster chain based on structural equation modeling.

[0051] Structural Equation Models (SEM) are used to construct triggering paths for rainstorm-flash flood disaster chains and cascading paths for rainstorm-flash flood-debris flow disaster chains. This allows for the determination of the direct, interaction, and total effects of different driving factors in the occurrence and evolution of these chains, identifying the main driving factors. By constructing causal paths, SEM can simultaneously calculate the relationships between multiple input and output variables, thereby clarifying the direct destructive effects and interactions of each driving factor in the evolution of the rainstorm-flash flood-debris flow disaster chain.

[0052] The above S202 may include the following steps.

[0053] (1) Extract the characteristic variables of the disaster in the evolution process from the sample database of rainstorm-flash flood-debris flow disaster chain, divide the characteristic variables into three categories, and provide input data for structural equation path calculation.

[0054] Feature variables can be categorized into the following three types.

[0055] A. Exogenous variables: These represent the external triggering conditions of a disaster chain. Specifically, they include maximum hourly rainfall intensity. (mm / h) Rainfall (mm), duration of rainfall (h) Soil moisture content in the early stage (%),slope (°), soil type, etc.

[0056] B. Mediating variables: Variables that characterize the evolution of the disaster chain. Specifically, these include soil moisture content. (%) Surface runoff (m³ / s), slope stability ditch peak flow (m³ / s), maximum mud depth (m) Maximum base shear stress (kPa).

[0057] C. Endogenous variables: Characterizing the final output of the disaster chain. Specifically, these include peak debris flow. (m³ / s), destroyed area (㎡), debris flow velocity (m / s) and debris flow volume (m³).

[0058] (2) Construct the structural equations for the triggering path of the rainstorm-flash flood disaster chain and the cascading path of the rainstorm-flash flood-debris flow disaster chain, respectively, with the following specific settings: ; ;in: The dynamic characteristics of flash floods in the sample data; For exogenous variables, the path coefficients of flash floods; For residual terms; Represents the scale of a debris flow; The path coefficient for rainwater directly triggering debris flows; The path coefficient of debris flow is the effect of flash flood dynamic characteristics on the path coefficient of debris flow. This is the residual term.

[0059] (3) The structural equations are solved iteratively using the maximum likelihood estimation (MLE) method to obtain the objective function. It is to minimize the sample covariance matrix With the model's implicit covariance matrix The differences between them were determined, and the standardized path coefficients obtained from the structural equations of different paths were identified. .

[0060] The objective can be expressed as: ;in: This represents the maximum likelihood objective function value; The sample covariance matrix; This represents the model's implicit covariance matrix; The number of observed variables is determined through iterative calculation. The smallest parameter matrix yields all standardized path coefficients.

[0061] (4) Based on the standardized path coefficient This study aims to decouple the direct impact of different driving factors on the rainstorm-flash flood-debris flow disaster chain and the interaction among multiple factors, and to identify the main driving factors of the rainstorm-flash flood-debris flow disaster chain.

[0062] Among them, direct effect value From path coefficients This indicates the strength of the direct influence of exogenous variables (or mediating variables) on endogenous variables; interaction effect value. By path coefficient and The product calculation reflects the interaction between exogenous variables and endogenous variables through multiple mediating variables; the sum of the direct effect and the interaction effect is the total effect value of the driving factor on the final disaster chain. The direct effect value, interaction effect value, and total effect value of all characteristic variables among the exogenous and mediating variables are determined, and the total effect value is arranged from high to low to identify the main driving factors of the rainstorm-flash flood-debris flow disaster chain, providing input variables for constructing a generalized additivity model (GAM-te) and identifying multivariate trigger thresholds.

[0063] The process of determining the direct effect size, interaction effect size, and total effect size described above can be expressed by the following formula: ;in: This indicates the direct impact of exogenous variables (or mediating variables) on endogenous variables, such as the effect of rainfall directly on the slope, causing landslides to transform into debris flows; This indicates the interaction between exogenous variables and endogenous variables through one or more mediating variables. For example, rainfall can induce debris flows by forming flash floods and further eroding the gully bed. This represents the total effect of driving factors in the rainstorm-flash flood-debris flow disaster chain; These represent the standardized path coefficients for different driving factors.

[0064] S203. Multivariate threshold identification of rainstorm-flash flood-debris flow disaster chain based on generalized additivity model.

[0065] Selecting the total effect value TE Larger front n Using the main driving factors as input variables, construct a system containing n The generalized additivity model (GAM-te) of the tensor product operator is used to perform nonlinear fitting and parameter determination on sample data of rainstorm-flash flood-debris flow disaster chains, thereby determining the critical threshold for driving factors to induce rainstorm-flash flood-debris flow disaster chains. The governing equations of the generalized additivity model are as follows: In the formula: Represents the intercept term; For the tensor product smoothing term, The operator can capture the nonlinear relationship between multiple variables simultaneously through the product of Laplace smoothing basis functions; The sum of the additional smoothing terms represents other environmental variables that affect the disaster; Indicates random error; Represents the volume of debris flow among exogenous variables; express One of the main driving factors.

[0066] The above S203 may include the following steps.

[0067] (1) For the rainstorm-flash flood-debris flow disaster chain, the main driving factor is rainfall. and the soil moisture content in the early stage For example, based on a sample database of rainstorm-flash flood-debris flow disaster chains, the main driving factors (such as rainfall) are analyzed. Compared with the previous soil moisture content The range of variation of the rainstorm-flash flood-debris flow disaster chain is used to construct a generalized additivity model to determine the intensity of debris flow under the combined action of the main driving factors, thereby forming a three-dimensional disaster response surface. .

[0068] (2) Calculate the three-dimensional disaster response surface of the rainstorm-flash flood-debris flow disaster chain respectively. For the main driving factors (rainfall) Early soil moisture content The second-order partial derivative of ) and The extreme points are used to identify the abrupt change zone from the gestation to the outbreak of debris flow disasters, which is the critical space of the driving factors that trigger the rainstorm-flash flood-debris flow disaster chain.

[0069] (3) In the critical space of driving factors, according to the debris flow disaster level classification standard (Table 1 above), set the critical volume value of debris flow for different levels. Three-dimensional disaster response surface of rainstorm-flash flood-debris flow disaster chain Draw contour lines on the top and project them onto the main driving factor (rainfall). Early soil moisture content Using a coordinate system, we can obtain the multivariate triggering thresholds for different levels of debris flow disasters. , ).

[0070] Based on the above technical solution, structural equations can be constructed to establish the triggering path of the rainstorm-flash flood disaster chain and the cascading path of the rainstorm-flash flood-debris flow disaster chain. Standardized path coefficients can then be calculated to quantify the direct effects, interaction effects, and total effects of each driving factor. This decouples the direct impact and interaction of driving factors on the disaster chain, accurately identifying the main driving factors of the rainstorm-flash flood-debris flow disaster chain. Furthermore, by using a generalized additivity model and a three-dimensional disaster-causing response surface, multivariate triggering thresholds under different levels of debris flow disasters can be determined, improving the accuracy of threshold determination.

[0071] The various embodiments in this specification are described in a progressive manner. For the same or similar parts between the various embodiments, please refer to each other. Each embodiment focuses on describing the differences from other embodiments.

[0072] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of this application.

Claims

1. A threshold identification method for a rainstorm-flash flood-debris flow disaster chain based on causal decoupling, characterized in that, The method includes: Based on the MassFlow physical mechanism model, the evolution of the disaster-prone areas of the rainstorm-flash flood-debris flow disaster chain was simulated, and a disaster chain sample database under various scenarios of rainfall and soil moisture content was obtained. Based on the disaster chain sample database, the first structural equation for the triggering path of the rainstorm-flash flood disaster chain and the second structural equation for the cascading path of the rainstorm-flash flood-debris flow disaster chain are constructed respectively. Solve the first and second structural equations to determine the standardized path coefficients for different driving factors; Based on the direct effect value, interaction effect value, and total effect value of the standardized path coefficient in the occurrence and evolution of the rainstorm-flash flood-debris flow disaster chain, the main driving factors of the rainstorm-flash flood-debris flow disaster chain are identified. The main driving factors are input into a generalized additivity model to obtain the debris flow intensity under the combined action of the main driving factors; and a three-dimensional disaster response surface of the rainstorm-flash flood-debris flow disaster chain is established based on the debris flow intensity. Based on the three-dimensional disaster response surface, the critical space of driving factors that trigger the rainstorm-flash flood-debris flow disaster chain is identified, and then the critical threshold for inducing the rainstorm-flash flood-debris flow disaster chain is determined.

2. The method according to claim 1, characterized in that, The MassFlow physics model was used to simulate the evolution of the rainstorm-flash flood-debris flow disaster chain in prone areas, resulting in a disaster chain sample database with rainfall and soil moisture content under various scenarios, including: Determine the slope stability coefficient of the target area under heavy rain conditions, and identify the vulnerable areas based on the slope stability coefficient; Based on historical information on rainstorms, floods, and debris flows in the prone areas, as well as DEM data and soil physical and mechanical parameters, the boundary and dynamic parameters of the MassFlow physical mechanism model are determined. Based on historical disaster data of the prone areas, the friction coefficient, turbulence coefficient, and debris flow unit weight parameters of the MassFlow physical mechanism model are determined. Rainfall and soil moisture scenarios with different return periods were designed to drive the MassFlow physical mechanism model, simulating the occurrence and evolution of the rainstorm-flash flood-debris flow disaster chain, and obtaining the disaster chain sample database.

3. The method according to claim 1, characterized in that, Based on the disaster chain sample database, the first structural equation for the triggering path of the rainstorm-flash flood disaster chain and the second structural equation for the cascading path of the rainstorm-flash flood-debris flow disaster chain are constructed, including: Extract the characteristic variables of disasters in the evolution process from the disaster chain sample database, and classify the characteristic variables into exogenous variables, mediating variables and endogenous variables; Based on the aforementioned feature variables, the first structural equation and the second structural equation are constructed respectively.

4. The method according to claim 3, characterized in that, Solving the first structural equation and the second structural equation to determine the standardized path coefficients of different driving factors includes: The first structural equation and the second structural equation are solved iteratively using the maximum likelihood estimation method to determine the standardized path coefficients; wherein the standardized path coefficients are the first coefficient, the second coefficient, and the third coefficient; the objective is to minimize the difference between the sample covariance matrix and the model latent covariance matrix.

5. The method according to claim 4, characterized in that, The main driving factors of the rainstorm-flash flood-debris flow disaster chain are identified based on the direct effect values, interaction effect values, and total effect values ​​of the standardized path coefficients in the occurrence and evolution of the rainstorm-flash flood-debris flow disaster chain, including: The second coefficient is used as the direct effect value; the direct effect value is used to reflect the strength of the direct influence of the exogenous variable or the mediating variable on the endogenous variable; The product of the first coefficient and the third coefficient is taken as the interaction effect value; the interaction effect value is used to reflect the interaction effect of the exogenous variable on the endogenous variable through multiple mediating variables; The sum of the direct effect and the interaction effect is taken as the total effect value; Arrange the multiple total effect values ​​from highest to lowest, and then... n The total effect value is used as the main driving factor.

6. The method according to claim 1, characterized in that, The generalized additivity model is determined based on the range of variation of the main driving factors in the sample database of rainstorm-flash flood-debris flow disaster chain.

7. The method according to claim 1, characterized in that, The process of identifying the critical space of driving factors triggering the rainstorm-flash flood-debris flow disaster chain based on the three-dimensional disaster response surface, and then determining the critical threshold for inducing the rainstorm-flash flood-debris flow disaster chain, includes: Based on the extreme points of the sum of the second-order partial derivatives of the main driving factors on the three-dimensional disaster response surface, the abrupt change region of debris flow disaster is determined as the critical space of the driving factors. Within the critical space of driving factors, and in conjunction with the debris flow disaster level classification standard, critical values ​​for debris flow volume are set for different levels. Based on the critical value of debris flow volume, contour lines are drawn on the three-dimensional disaster response surface and projected onto the coordinate system of the main driving factors to obtain the critical thresholds under different levels of debris flow disasters.

Images