High-precision and high-efficiency simulation and risk prediction method and system for dam-break flood of barrier lake

Abstract

The application discloses a high-precision and high-efficiency dammed lake flood outburst simulation and risk prediction method and system, breaks through the limitations of traditional methods in the balance of calculation accuracy and calculation cost, establishes a dammed lake flood outburst disaster chain coupling calculation model considering an erodible bed surface, and is used for simulating the whole process of a disaster chain from dammed lake water level rising-dammed body overtopping erosion outburst-flood outburst evolution-downstream engineering; the method is based on shallow water-erosion-riverbed morphology coupling dynamics theory, considers the collapse mechanism in the dammed body outburst process, can be applied to the simulation of the dammed body outburst process under the erosion action and the whole dynamics process of the outburst flood spreading to the downstream engineering and further affecting the engineering buildings, can calculate in advance the flood risk factors such as the time, maximum water level and maximum flow rate of the outburst flood spreading to the downstream engineering and the auxiliary buildings, and can provide decision support for the outburst flood risk warning and defense.

Description

Technical Field

[0001] This invention relates to the field of flood simulation and early warning technology, specifically to a high-precision and efficient simulation and hazard prediction method and system for landslide dammed lake outburst floods, and particularly to a method for efficient and high-precision full-process simulation of landslide dammed lake outburst floods and hazard assessment of downstream projects. Background Technology

[0002] A landslide-dammed lake outburst flood is a high-energy, unsteady flow formed when a sudden landslide or collapse blocks a river channel, creating a natural dam (i.e., a landslide dam). Subsequent instability of the dam structure or overtopping causes the instantaneous release of water from the reservoir. It represents the most energy-dense, sudden, and longest-chain-of-damming extreme type of flood disaster. Defense against landslide-dammed lake outburst floods relies heavily on accurate estimations of the outburst and flood propagation processes. The former involves understanding the formation of the unsteady flow, while the latter primarily affects the assessment of the flood's spread and its destructive potential.

[0003] Research on landslide dammed lake outburst floods mainly falls into two categories: physical models and numerical simulations. Physical models can accurately invert key characteristics of dam failure and flood evolution processes, and are often used for mechanism studies. However, due to their long simulation cycles and high costs, they are difficult to use for actual landslide dammed lake outburst flood risk early warning. Therefore, in practical engineering applications and scientific research, simulations of landslide dammed lake outburst floods are usually based on various mathematical models, mainly in the following two aspects.

[0004] On the one hand, traditional simulations of landslide dammed lake outburst floods typically involve two steps. The first step uses a simplified physical mechanism outburst flood model to calculate the outburst process, obtaining the flow profile of the outburst flood at the breach. The second step uses the outburst flood flow profile at the breach as a boundary condition within the modeling area of ​​the potential downstream impact zone, employing a one-dimensional or two-dimensional hydrodynamic model to simulate the evolution of the outburst flood. Landslide dammed lake outbursts involve the interaction between the riverbed and water flow under complex terrain. Traditional methods based on simplified physical mechanisms often have low accuracy, primarily due to poor estimation of the outburst flow profile at the breach. In reality, the outburst process and the evolution of the outburst flood should be continuous and interconnected, forming a complete chain of landslide dammed lake outburst flood disasters. Traditional methods that separate these processes into two distinct numerical models often result in low accuracy.

[0005] On the other hand, two-dimensional and three-dimensional hydrodynamic-sediment-morphology coupled dynamic models based on complete physical mechanisms can simulate the two processes of landslide dam breach and flood evolution in the same model. These models usually consider physical backgrounds such as non-uniform sand and two-phase flow, and can achieve high-precision simulation of the entire process of landslide dam breach flood and its evolution. However, the equations that need to be solved are very complex, the data requirements for modeling are extremely high, the solution efficiency is low, and it is difficult to apply them to the real-time simulation of actual landslide dam breach flood and downstream risk early warning work.

[0006] In addition, due to the suddenness, variability, and chain-like nature of landslide dammed lake outburst floods, the entire disaster chain process—from the rising water level of the dammed lake to its breach, the flood's evolution, and the subsequent impact on downstream engineering structures—often completes within days or even hours. Therefore, the efficiency of numerical model solutions significantly impacts flood prevention and response efforts. Currently, the main computational burden of numerical models lies in solving the fluid dynamics model, typically employing mesh-based methods such as the finite volume method or particle-based methods like smoothed particle fluid dynamics. These methods often require a large number of meshes or particles to achieve the required computational accuracy. Traditional acceleration methods are usually based on distributed parallel computing using multi-node, multi-core CPUs, which often requires dozens or even hundreds of computing cores to barely meet real-time requirements, making them insufficient for practical applications.

[0007] In summary, research on the simulation and risk early warning of landslide dammed lake outburst floods still faces many challenges. On the one hand, numerical models based on relatively sound physical mechanisms can provide high simulation accuracy, but their complex modeling process and high computational cost limit their widespread application in practical engineering. On the other hand, although GPU parallel computing technology has brought new solutions to improve the solution efficiency of numerical models, how to further optimize algorithm performance while ensuring accuracy remains an urgent problem to be solved. Summary of the Invention

[0008] The purpose of this invention is to address the problems existing in the prior art by providing a high-precision and efficient method and system for simulating and predicting the risk of landslide dammed lake breach floods.

[0009] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A high-precision and efficient method for simulating and predicting the hazard of landslide dammed lake outburst floods includes the following steps: A high-precision preprocessing model of the landslide dam body, including the landslide lake and the landslide dam, is constructed. The two-dimensional modeling area is divided into unstructured meshes. The actual river topography and the dam body shape are assigned to the mesh cells. By setting the mesh cell size, high-precision modeling of the topography within the calculation area is achieved. Based on the preprocessing high-precision model, a coupled calculation model considering the water flow and dynamic changes in riverbed morphology on the erodible riverbed is established. By depicting the bidirectional coupling effect between riverbed morphology evolution and topographic slope, the entire process of the landslide dam breaking under erosion is calculated and simulated. The collapse effect of the landslide dam is calculated. By activating erosion and collapse calculations in a fixed area, the gradual collapse process of the dam during the landslide dam failure is simulated. The information obtained from the computational simulation is converted into output and visualization files.

[0010] This high-precision and efficient simulation and hazard prediction method for landslide dam breach floods is based on the coupled dynamics theory of shallow water-erosion-riverbed morphology and considers the collapse mechanism during the breach process. It can be applied to the simulation of the breach process of landslide dams under erosion and the entire dynamic process of the breach flood propagating downstream and further affecting engineering structures. It can pre-calculate flood risk factors such as the time, maximum water level, and maximum flow velocity of the breach flood propagating to downstream engineering and auxiliary structures, and can provide decision support for early warning and defense of breach flood risks.

[0011] Furthermore, in the step of constructing the preprocessing high-precision model, the method for assigning values ​​is as follows: The original terrain of the entire computational domain is linearly interpolated to the grid cells using ground elevation points and grid center coordinates; Select the grid cell containing the landslide dam and mark it as an area of ​​erodible bed surface. Then, use linear interpolation to assign the landslide dam topography to the selected grid cell. Based on the soil characteristics of the dam body, the erosion coefficient, critical shear stress, and angle of repose are set for dam failure simulation. Set the roughness of the entire calculation area according to the underlying surface conditions; Initial and boundary conditions are set based on measured and forecast data.

[0012] Furthermore, the governing equations of the coupled computational model are as follows: , In the formula, t Indicates time, x and y This indicates the horizontal direction of the Cartesian coordinate system. h Indicates water depth. z Indicates the bed surface elevation. u and v They represent x direction and y Flow velocity in direction τ x and τ y They representx and y Riverbed friction stress in the direction of direction, ρ Indicates water flow density. g Represents gravitational acceleration. p Indicates the porosity of the bed surface material. Let kd represent the erosion flux, kc be the erosion modulus, and τc be the critical shear stress. This refers to the shear stress between the water flow and the bed surface.

[0013] Furthermore, the governing equations of the coupled computational model are discretized using the finite volume method, and the discretized form is expressed as the following ordinary differential equations: , In the formula, t Indicates time, subscript i Indicates the grid cell number, subscript k This indicates the number of the grid edge in the grid cell, with values ​​of 1, 2, 3...N. A t Represents grid cells i area, L i,k Represents grid cells i No. k The length of the grid edge, n i,k Represents grid cells i No. k The outward normal vector of each grid edge. For a vector of conserved variables, This represents the numerical flux perpendicular to the cell edge in the local projected coordinate system of the mesh cell edge. The outward normal direction vector of the grid cell edge. For the bottom-breaking source term flux, For friction source terms, This is the flux vector for the exchange between the water flow and the riverbed.

[0014] Furthermore, the calculation method for the collapse effect of the landslide dam is as follows: considering the angle of repose under the ultimate equilibrium of the soil, at the end of each calculation time step, the inundation state of the current grid cell and the neighboring grid cells is judged, and the difference between the elevation slope and the angle of repose between the grid cells is calculated to determine the collapse amount.

[0015] Furthermore, the collapse amount of the current grid cell and its neighboring grid cells. for: , In the formula, The underwater angle of repose is used when the water depth of the current grid cell and its adjacent grid cells are both greater than 0. Otherwise, it is the angle of repose on the water. , For cell mesh i The distance between the center of the grid and the centers of neighboring grid cells; Slope of the current grid cell and adjacent grid cells φ for: , In the formula, Z i Represents grid cells i elevation, Z k The elevation of adjacent grid cells. L i,k This represents the distance from the current grid centroid to the centroid of the nearest neighboring grid. After calculating the collapse amount of the current grid cell and its neighboring grid cells, the change in terrain elevation of the current grid cell due to the collapse is as follows: , In the formula, Z new Represents grid cells i Updated grid elevation, Z This indicates the grid elevation before the update. Z i,k This represents the mesh collapse flux along the k-th edge of mesh cell i.

[0016] Furthermore, it also includes building a CPU-GPU heterogeneous acceleration framework, in which the pre- and post-processing of file input and output, and data processing are placed on the CPU and executed in an asynchronous structure during the computation process, while large-scale and intensive numerical solution work is placed on the GPU for asynchronous computation.

[0017] Furthermore, it also includes an adaptive multi-engine coupled computation method: when there is only limited data in the modeling area, a zero-dimensional water body model and a one-dimensional river network model are used to couple the modeling of the local data-deficient area.

[0018] Furthermore, the zero-dimensional water body model is solved using the water balance equation: , In the formula, V Indicates the water body's capacity. V i n+1 、 V i n This represents the water volume of the two adjacent water bodies in grid cell i. Q i and Q oΔ represents the flow rate entering and exiting the water body, respectively. t Indicates the time difference between the two dates; The one-dimensional river network model is solved using the Saint-Venant equations: , In the formula, A and Q These represent the cross-sectional area and flow rate, respectively. η Indicates water level. τ f Indicates riverbed resistance. q and w These represent the lateral inflow rate and the corresponding velocity, respectively. g Represents gravitational acceleration. x and y This indicates the horizontal direction of the Cartesian coordinate system.

[0019] A high-precision and efficient simulation and hazard prediction system for landslide dammed lake outburst floods is provided to implement the high-precision and efficient simulation and hazard prediction method for landslide dammed lake outburst floods as described above, comprising: The pre-processing high-precision modeling module is used to divide the grid cells and construct a pre-processing high-precision model of the landslide dam. The numerical algorithm implementation module includes a high-precision hydrodynamic numerical calculation module, a shallow water-erosion-riverbed morphology dynamics coupled calculation module, a collapse calculation module, a GPU acceleration framework, and an adaptive multi-engine coupled calculation module. The simulation calculation result post-processing module is used to convert the numerical text information obtained from the calculation into a common binary VTK file, Tecplot file format or shapefile file for visualization output and display.

[0020] Compared with existing technologies, the beneficial effects of this invention are: 1. This method overcomes the limitations of traditional methods in balancing computational accuracy and computational cost, and establishes a coupled computational model for the disaster chain of landslide dam breach floods considering the erodible bed surface. This model is used to simulate the entire disaster chain process from the rise of the landslide dam water level to the erosion and breach of the landslide dam body, the evolution of the breach flood, and the downstream engineering works. This method is based on the coupled dynamics theory of shallow water, erosion, and riverbed morphology, and considers the collapse mechanism during the breach of the landslide dam body. It can be applied to the simulation of the breach process of the landslide dam body under erosion and the entire dynamic process of the breach flood propagating downstream and further affecting engineering structures. It can pre-calculate the time for the breach flood to propagate to downstream engineering works and ancillary structures. 1. Flood risk factors such as maximum water level and maximum flow velocity can provide decision support for early warning and defense against dam breach risks; 2. By performing calculations on erosion and collapse within a fixed area, the problem of achieving high-precision simulation of dam breaches with low computational cost is solved; 3. Through a high-performance CPU-GPU computing framework, the computing tasks on the CPU and GPU sides do not interfere with each other, ensuring efficient operation throughout the entire calculation process; 4. The adaptive multi-engine coupled computing method solves the problem of effectively simulating landslide dam breach floods under conditions of data scarcity in practical applications, and can be used to predict risk factors such as water level, water depth, and flow field of breach floods in downstream projects, providing guidance for early warning and defense work in downstream projects. Attached Figure Description

[0021] Figure 1 This is a schematic diagram showing the before-and-after comparison of the topographic modeling of the landslide dam area (including the landslide dam) according to the present invention; Figure 2 This is a flowchart illustrating the model solving process using CPU-GPU heterogeneous parallel acceleration technology in this invention. Figure 3 This is a schematic diagram of the water depth distribution during the downstream process of the landslide dam breach and the evolution of the breach flood obtained by the calculation and simulation of this invention; Figure 4 This is a schematic diagram of the flow velocity distribution during the downstream process of a landslide dam breach and its flood evolution, obtained through calculation and simulation in this invention. Figure 5 This is a flowchart comparing the flow rate at the breach obtained from the simulation of this invention with the experimental value; Figure 6 This is a schematic diagram of the adaptive multi-engine modeling method used in this invention when data is scarce; Figure 7 This is a schematic diagram comparing the observed and simulated values ​​of the flow process at a downstream hydrological station using a multi-engine model, as per the present invention. Detailed Implementation

[0022] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] A high-precision and efficient method for simulating and predicting the hazard of landslide dammed lake outburst floods includes the following steps: A high-precision preprocessing model of the landslide dam (including the landslide lake and its dam) is constructed. The two-dimensional modeling area is divided into unstructured meshes. The actual river topography and dam morphology are assigned to the mesh cells. By setting the mesh cell size, high-precision modeling of the topography within the calculation area is achieved. Based on the preprocessing high-precision model, a coupled calculation model considering the water flow and dynamic changes in riverbed morphology on the erodible riverbed is established. By depicting the bidirectional coupling effect between riverbed morphology evolution and topographic slope, the entire process of the landslide dam breaking under erosion is calculated and simulated. The collapse effect of the landslide dam is calculated. By activating erosion and collapse calculations in a fixed area, the gradual collapse process of the dam during the landslide dam failure is simulated. The information obtained from the computational simulation is converted into output and visualization files.

[0024] This method overcomes the limitations of traditional approaches in balancing computational accuracy and cost. It establishes a coupled computational model for the disaster chain of landslide dammed lake outburst floods, considering the erodible bed surface. This model is used to simulate the entire disaster chain process from the rise in water level of the landslide dammed lake to the erosion and outburst of the dammed body, the evolution of the outburst flood, and the downstream engineering structures. Based on the coupled dynamics theory of shallow water, erosion, and riverbed morphology, and considering the collapse mechanism during the outburst flood, this method can be applied to the simulation of the outburst process of the dammed body under erosion and the entire dynamic process of the outburst flood propagating downstream and further affecting engineering structures. It can pre-calculate flood risk factors such as the time, maximum water level, and maximum flow velocity of the outburst flood propagating to downstream engineering structures and ancillary structures, and can provide decision support for the early warning and defense of outburst flood risks.

[0025] This prediction method proposes a numerical simulation algorithm that avoids the need to introduce unnecessary empirical parameters such as dam breach widening coefficients and breach shape development in traditional breach flood models, as well as unreasonable hydrodynamic calculation methods such as weir flow formulas used in traditional breach flood models. This method can automatically realize breach development, effectively improving the accuracy of breach flood simulation. Furthermore, this method does not require solving sediment transport equations, thus possessing high computational efficiency.

[0026] Specifically, in the step of constructing a high-precision preprocessing model, the two-dimensional modeling area is first divided into unstructured grids using mesh generation software such as GMash. Then, the actual river topography and dam morphology can be assigned to the grid cells using the parameter assignment method of this patent in a GIS tool. The assignment method is as follows: (1) The original terrain of the entire calculation area is linearly interpolated to the grid center coordinates through ground elevation points and assigned to the grid cells; (2) Select the grid cell where the landslide dam is located and mark it as an area of ​​erodible bed surface. Assign the landslide dam topography to the selected grid cell through linear interpolation. (3) Based on the soil characteristics of the dam body, the erosion coefficient, critical shear stress and angle of repose are set for dam failure simulation; (4) Set the roughness of the entire calculation area according to the underlying surface conditions, usually by setting the Manning coefficient; (5) Set initial and boundary conditions based on measured and forecast data.

[0027] By setting the grid cell size, high-precision modeling of the terrain within the computational area can be achieved. The modeling effect of the grid near the landslide dam is shown below. Figure 1 As shown.

[0028] Furthermore, the model is solved through numerical algorithms. The numerical algorithms of this invention include a high-precision hydrodynamic numerical algorithm, a shallow water-erosion-riverbed morphology dynamic coupling method (the coupled calculation model), a collapse algorithm (dam collapse effect), a GPU acceleration framework, and an adaptive multi-engine coupled calculation method.

[0029] Specifically, the shallow water-erosion-riverbed morphology dynamic coupling method establishes a coupled computational model that considers the dynamic changes in water flow and riverbed morphology on erodible riverbeds. This coupled computational model fully considers the riverbed erosion mechanism caused by water flow shearing. By characterizing the bidirectional coupling between riverbed morphology evolution and topographic slope (the source term driving water flow), it can accurately simulate the entire process of dam failure under erosion.

[0030] The governing equations of the coupled computational model are as follows: , In the formula, t Indicates time, x and y This indicates the horizontal direction of the Cartesian coordinate system. h Indicates water depth. z Indicates the bed surface elevation. u and v They represent x direction and y Flow velocity in direction τ x and τ y They represent x and y Riverbed friction stress in the direction of direction, ρ Indicates water flow density. g Represents gravitational acceleration. p Indicates the porosity of the bed surface material. Let kd represent the erosion flux, kc be the erosion modulus, and τc be the critical shear stress. This refers to the shear stress between the water flow and the bed surface.

[0031] Furthermore, the governing equations of the coupled computational model are discretized using the finite volume method, and the discretized form in the computational unit is expressed as the following ordinary differential equations: , In the formula, t Indicates time, subscript i Indicates the grid cell number, subscript k This indicates the number of the grid edge in the grid cell, with values ​​of 1, 2, 3...N. A t Represents grid cells i area, L i,k Represents grid cells i No. k The length of the grid edge, n i,k Represents grid cells i No. k The outward normal vector of each grid edge. For a vector of conserved variables, This represents the numerical flux perpendicular to the cell edge in the local projected coordinate system of the mesh cell edge. The outward normal direction vector of the grid cell edge. For the bottom-breaking source term flux, For friction source terms, This is the flux vector for the exchange between the water flow and the riverbed.

[0032] For the time term on the left-hand side of the above ordinary differential equation, the second-order Runge-Kutta method is used for time advancement. The model employs a multi-gradient MUSCL scheme for interpolation of the original interface variables, and corrects these variables using a hydrostatic reconstruction method to ensure harmony. Interface flux is calculated using an HLLC approximate Riemann solver, the bottom-breakage source term is calculated using the bottom-breakage source term flux method, and the friction source term is calculated using a fully implicit method to ensure numerical stability.

[0033] Specifically, the calculation method for the collapse effect of a landslide dam is used to simulate the gradual collapse of the dam body during the failure of a landslide dam. The calculation method for the collapse effect of a landslide dam is as follows: considering the angle of repose under the ultimate equilibrium of the soil, at the end of each calculation time step, the inundation state of the current grid cell and the neighboring grid cells is judged, and the difference between the elevation slope and the angle of repose between the grid cells is calculated to determine the amount of collapse.

[0034] Collapse amount of the current grid cell and its neighboring grid cells for: , In the formula, The underwater angle of repose is used when the water depth of the current grid cell and its adjacent grid cells are both greater than 0. Otherwise, it is the angle of repose on the water. , For cell mesh i The distance between the center of the grid and the centers of neighboring grid cells; Slope of the current grid cell and adjacent grid cells φ for: , In the formula, Z i Represents grid cells i elevation, Z k The elevation of adjacent grid cells. L i,k This represents the distance from the current grid centroid to the centroid of the nearest neighboring grid. After calculating the collapse amount of the current grid cell and its neighboring grid cells, the change in terrain elevation of the current grid cell due to the collapse is as follows: , In the formula, Z new Represents grid cells i Updated grid elevation, Z This indicates the grid elevation before the update. Z i,k This represents the mesh collapse flux along the k-th edge of mesh cell i.

[0035] Furthermore, in the computational simulation (computational solution) process, a CPU-GPU heterogeneous acceleration framework is used for parallel computing to achieve high-performance computation for data processing and large-scale intensive solution tasks. During the computation, pre- and post-processing of file input / output and data processing are performed asynchronously on the CPU, while large-scale intensive numerical solution work is performed asynchronously on the GPU. The program comprises three computational modules, and the computational flow is as follows: Figure 2As shown.

[0036] Because the CPU and GPU are implemented entirely through an asynchronous framework, the computing tasks on the CPU and GPU do not interfere with each other, ensuring that the entire computing process is carried out in an efficient manner. This solves the problem that it is difficult to achieve short-term forecasting of the propagation of landslide dam breach floods using the CPU in traditional frameworks.

[0037] Furthermore, it also includes an adaptive multi-engine coupled calculation method: when there is only limited data in the modeling area (such as the reservoir capacity curve of a landslide dam, river cross-section data, etc.), a zero-dimensional water body model and a one-dimensional river network model are used to couple the modeling of the local data-deficient area.

[0038] Specifically, the zero-dimensional water body model is solved using the water balance equation: , In the formula, V Indicates the water body's capacity. V i n+1 、 V i n This represents the water volume of the two adjacent water bodies in grid cell i. Q i and Q o Δ represents the flow rate entering and exiting the water body, respectively. t Indicates the time difference between the two dates; The one-dimensional river network model is solved using the Saint-Venant equations: , In the formula, A and Q These represent the cross-sectional area and flow rate, respectively. η Indicates water level. τ f Indicates riverbed resistance. q and w These represent the lateral inflow rate and the corresponding velocity, respectively. g Represents gravitational acceleration. x and y The horizontal direction is represented by the Cartesian coordinate system. The solution of this computational model adopts the same finite volume format as the two-dimensional model based on the shallow water-erosion-riverbed morphology dynamic coupling theory (hereinafter referred to as the two-dimensional model) introduced above. The difference is that the flux solution is calculated using the HLL Riemann approximation solver, and the bottom slope source term and pressure source term are combined and solved using the central difference method.

[0039] In adaptive multi-engine modeling, water body models can be used to replace the grid of the landslide dammed lake area and the river channel area, respectively. For the former, the modeling approach is to replace the grid of the landslide dammed lake area upstream of the dam with a water body model, and to construct coupling boundary conditions between the water body model and the two-dimensional model (the two-dimensional model provides the flow boundary to the water body model, and the water body model provides the water level boundary to the two-dimensional model). For the latter, for areas of the river channel within the modeling scope that lack two-dimensional topographic data but have cross-sectional data or simplified cross-sectional data, a river network model can be used, and a positive coupling connection can be established between the one-dimensional river network model and the two-dimensional model at both ends (the two-dimensional model provides the flow boundary to the water body model, and the water body model provides the water level boundary to the two-dimensional model).

[0040] The adaptive multi-engine coupled calculation method solves the problem of effectively simulating landslide dam breach floods under conditions of data shortage in practical applications. It can be used to predict risk factors such as water level, water depth, and flow field of breach floods in downstream projects, providing guidance for early warning and defense work in downstream projects.

[0041] This invention also provides a high-precision and efficient simulation and hazard prediction system for landslide dammed lake outburst floods, used to implement the high-precision and efficient simulation and hazard prediction method for landslide dammed lake outburst floods as described above, comprising: The pre-processing high-precision modeling module is used to divide the grid cells and construct a pre-processing high-precision model of the landslide dam. The numerical algorithm implementation module includes a high-precision hydrodynamic numerical calculation module, a shallow water-erosion-riverbed morphology dynamics coupled calculation module, a collapse calculation module, a GPU acceleration framework, and an adaptive multi-engine coupled calculation module. The simulation calculation result post-processing module is used to convert the numerical text information obtained from the calculation into a common binary VTK file, Tecplot file format or shapefile file for visualization output and display.

[0042] The model solved by this prediction method can obtain parameters such as water depth, flow velocity, water level, riverbed elevation, erosion, and frictional stress on all grids. The obtained information is in the form of digital text, which can be converted into common binary VTK files, Tecplot files, or shapefiles for visualization. These files can be displayed in some common open-source software such as Paraview or commercial software such as Tecplot and ArcGIS.

[0043] To further demonstrate the system implementation and application, two case studies are compared. First, a comprehensive two-dimensional model is presented to solve and verify the entire process of the landslide dam failure and its resulting flood disaster chain. This model includes information such as the landslide dam failure process and the downstream water depth, velocity, and flow rate of the flood. Figures 3-5As shown, the water depth distribution during the downstream process of the landslide dam breach and its flood evolution, the flow velocity distribution during the downstream process of the landslide dam breach and its flood evolution, and the flow rate at the breach point are calculated and compared with experimental values. It can be seen that the experimental values ​​and calculated values ​​are in good agreement.

[0044] Secondly, the solution and verification are presented using an adaptive multi-engine approach in the case of missing data. Here, in the case of extreme data scarcity, simplification of certain areas is achieved through water level-reservoir capacity curves, measured river cross-sections, and generalized river cross-sections. The comparison and verification results of downstream station flow values ​​and measured values ​​are also provided. Figure 6 and Figure 7 As shown, the adaptive multi-engine modeling used in the case of data shortage is presented, and the observation and simulation values ​​of the flow process at the downstream hydrological station are compared in the case of multi-engine model. It can be seen that the observation and calculation values ​​are in good agreement.

[0045] Finally, Table 1 shows the computational efficiency of this patent under both CPU and CPU-GPU heterogeneous parallel computing frameworks. As can be seen, this algorithm has high solution efficiency and can meet the efficiency requirements of practical engineering applications.

[0046] Table 1 Comparison of computational performance under CPU and CPU-GPU heterogeneous parallel computing.

[0047] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A high-precision and efficient method for simulating and predicting the hazard of landslide dammed lake outburst floods, characterized in that, Includes the following steps: A high-precision preprocessing model of the landslide dam body, including the landslide lake and the landslide dam, is constructed. The two-dimensional modeling area is divided into unstructured meshes. The actual river topography and the dam body shape are assigned to the mesh cells. By setting the mesh cell size, high-precision modeling of the topography within the calculation area is achieved. Based on the preprocessing high-precision model, a coupled calculation model considering the water flow and dynamic changes in the riverbed morphology on the erodible riverbed is established. By depicting the bidirectional coupling effect between the evolution of the riverbed morphology and the topographic slope, the entire process of the landslide dam breaking under erosion is calculated and simulated. The collapse effect of the landslide dam is calculated. By activating erosion and collapse calculations in a fixed area, the gradual collapse process of the dam during the landslide dam failure is simulated. The information obtained from the computational simulation is converted into output and visualization files.

2. The method for high-precision and efficient simulation and hazard prediction of landslide dammed lake outburst floods according to claim 1, characterized in that, In the step of constructing a high-precision preprocessing model, the method for assigning values ​​is as follows: The original terrain of the entire computational domain is linearly interpolated to the grid cells using ground elevation points and grid center coordinates; Select the grid cell containing the landslide dam and mark it as an area of ​​erodible bed surface. Then, use linear interpolation to assign the landslide dam topography to the selected grid cell. Based on the soil characteristics of the dam body, the erosion coefficient, critical shear stress, and angle of repose are set for dam failure simulation. Set the roughness of the entire calculation area according to the underlying surface conditions; Initial and boundary conditions are set based on measured and forecast data.

3. The method for high-precision and efficient simulation and hazard prediction of landslide dammed lake outburst floods according to claim 1, characterized in that, The governing equations of the coupled computational model are as follows: ， In the formula, t Indicates time, x and y This indicates the horizontal direction of the Cartesian coordinate system. h Indicates water depth. z Indicates the bed surface elevation. u and v They represent x and y Flow velocity in direction τ x and τ y They represent x and y Riverbed friction stress in the direction of direction, ρ Indicates water flow density. g Represents gravitational acceleration. p Indicates the porosity of the bed surface material. Indicates erosion flux, k d For the erosion modulus, τ c The critical shear stress. This refers to the shear stress between the water flow and the bed surface.

4. The method for high-precision and efficient simulation and hazard prediction of landslide dammed lake outburst floods according to claim 3, characterized in that, The governing equations of the coupled computational model are discretized using the finite volume method, and the discretized form is expressed as the following ordinary differential equations: ， In the formula, t Indicates time, subscript i Indicates the grid cell number, subscript k Indicates the number of the grid edge in the grid cell. A t Represents grid cells i area, L i，k Represents grid cells i No. k The length of the grid edge, n i，k Represents grid cells i No. k The outward normal vector of each grid edge. For a vector of conserved variables, This represents the numerical flux perpendicular to the cell edge in the local projected coordinate system of the mesh cell edge. The outward normal direction vector of the grid cell edge. For the bottom-breaking source term flux, For friction source terms, This is the flux vector for the exchange between the water flow and the riverbed.

5. The method for high-precision and efficient simulation and hazard prediction of landslide dammed lake outburst floods according to claim 1, characterized in that, The calculation method for the collapse effect of the landslide dam is as follows: considering the angle of repose under the ultimate equilibrium of the soil, at the end of each calculation time step, the inundation state of the current grid cell and the neighboring grid cells is judged, and the difference between the elevation slope and the angle of repose between the grid cells is calculated to determine the collapse amount.

6. The method for high-precision and efficient simulation and hazard prediction of landslide dammed lake outburst floods according to claim 5, characterized in that, Collapse amount of the current grid cell and its neighboring grid cells for: ， In the formula, The underwater angle of repose is used when the water depth of the current grid cell and its adjacent grid cells are both greater than 0. Otherwise, it is the angle of repose on the water. , For cell mesh i The distance between the center of the grid and the centers of neighboring grid cells; Slope of the current grid cell and adjacent grid cells φ for: ， In the formula, The elevation of adjacent grid cells. This represents the distance from the current grid centroid to the centroid of the nearest neighboring grid. After calculating the collapse amount of the current grid cell and its neighboring grid cells, the change in terrain elevation of the current grid cell due to the collapse is as follows: ， In the formula, Z new Representation unit i Updated grid elevation, Z This indicates the grid elevation before the update. Z i,k Represents grid cells i No. k Grid collapse flux along the strip edge direction.

7. The method for high-precision and efficient simulation and hazard prediction of landslide dammed lake outburst floods according to claim 1, characterized in that, It also includes building a CPU-GPU heterogeneous acceleration framework, in which the pre- and post-processing of file input and output, and data processing are placed on the CPU and executed in an asynchronous structure during the calculation process, while large-scale and intensive numerical solution work is placed on the GPU for asynchronous computation.

8. The method for high-precision and efficient simulation and hazard prediction of landslide dammed lake outburst floods according to claim 1, characterized in that, It also includes an adaptive multi-engine coupled computation method: when there is only limited data in the modeling area, a zero-dimensional water body model and a one-dimensional river network model are used to couple the modeling of the local data-deficient area.

9. The method for high-precision and efficient simulation and hazard prediction of landslide dammed lake breach floods according to claim 8, characterized in that, The zero-dimensional water body model is solved using the water balance equation: ， In the formula, V Indicates the water body's capacity. Q i and Q o Δ represents the flow rate entering and exiting the water body, respectively. t Indicates the time difference between the two dates; The one-dimensional river network model is solved using the Saint-Venant equations: ， In the formula, A and Q These represent the cross-sectional area and flow rate, respectively. η Indicates water level. τ f Indicates riverbed resistance. q and w These represent the lateral inflow rate and the corresponding velocity, respectively. g Represents gravitational acceleration. x and y This indicates the horizontal direction of the Cartesian coordinate system.

10. A high-precision and efficient simulation and hazard prediction system for landslide dammed lake outburst floods, used to implement the high-precision and efficient simulation and hazard prediction method for landslide dammed lake outburst floods as described in any one of claims 1 to 9, characterized in that, include: The pre-processing high-precision modeling module is used to divide the grid cells and construct a pre-processing high-precision model of the landslide dam. The numerical algorithm implementation module includes a high-precision hydrodynamic numerical calculation module, a shallow water-erosion-riverbed morphology dynamics coupled calculation module, a collapse calculation module, a GPU acceleration framework, and an adaptive multi-engine coupled calculation module. The simulation calculation result post-processing module is used to convert the numerical text information obtained from the calculation into a common binary VTK file, Tecplot file format or shapefile file for visualization output and display.

Images