A method for determining a dynamic early warning index of mountain flood disaster based on multi-dimensional element coupling

Abstract

The application discloses a kind of mountain torrent disaster dynamic early warning index determination method based on multi-dimensional element coupling, comprising: collecting the real-time and historical risk hidden danger data of associated rainfall station, water level station, flow station, soil moisture station in dangerous area;Combining flood data to construct HEC-RAS two-dimensional grid rainfall model;Determine the flood inundation range by rainfall model simulation, and determine the flood control capacity of specified object, determine mountain torrent disaster early warning rainfall;Introduce river engineering facility hidden danger, early soil moisture content, regional geological data and the multi-dimensional mountain torrent influencing factor in current hydrological state of river channel and construct corresponding data model;Based on a group of dynamic early warning rainfall and coupling associated multi-dimensional mountain torrent influencing factor data model, used to correct initial early warning rainfall threshold, to form dynamic perception mountain torrent disaster early warning index.The application can form dynamic situation perception early warning rainfall index, realize the transition from static risk to dynamic risk.

Description

Technical Field

[0001] This invention relates to the field of flash flood disaster prevention technology, specifically to a method, equipment, and storage medium for determining dynamic early warning indicators for flash flood disasters based on multi-dimensional factor coupling. Background Technology

[0002] The factors that cause flash floods have both natural and social attributes. Their formation, development and degree of harm are jointly influenced by natural conditions such as rainfall, topography, geology and soil moisture content, as well as social factors such as human economic activities. They are characterized by wide distribution, frequent occurrence, suddenness, difficulty in prediction and prevention, rapid disaster formation, strong destructiveness, strong seasonality and obvious regionality.

[0003] Flash flood warning indicators are crucial for flash flood warnings and decision-making. Numerous methods exist for analyzing and determining these indicators, primarily including the critical rainfall curve method, the water level / discharge inverse estimation method, and the rainfall-driven indicator method. The critical rainfall curve method constructs critical rainfall thresholds for different durations based on historical flood events. It is simple to operate but highly dependent on local historical samples and easily ignores initial conditions such as soil moisture content. It also has problems such as fixed thresholds, difficulty in adapting to climate and land use changes, and difficulty in migrating to areas without data. The water level / flow backward method targets dangerous water levels or flows and uses hydrological calculations and water level-flow relationships to backward infer the corresponding rainfall or rainfall conditions. However, the calculation and data processing volume is large, which is not conducive to the speed required for real-time early warning. Moreover, the backward inference results are highly sensitive to parameter errors, and the applicability is limited in small watersheds and in the absence of regular runoff observations. The lead time is usually short. Rainfall-driven index methods use cumulative rainfall, short-duration rainfall intensity, or composite rainfall indices to set thresholds. They are easy to integrate with radar and forecast products, but require a large number of samples for calibration. They are highly dependent on high spatiotemporal resolution rainfall data, and the indices are difficult to cover multiple disaster-causing mechanisms. They also have limited lead time for sudden heavy rainfall.

[0004] Currently, the determination of flash flood warning indicators mostly considers factors such as previous rainfall, soil moisture content, and the confluence time of small watersheds. Through the flash flood disasters that have occurred in recent years, the risk and hidden dangers of bridges, weirs and dams on the river channels have been congested, resulting in the disaster amplification effect. At the same time, the changes in surface permeability of different geological cover types (sand, silt, rock, vegetation, etc.) affect the formation of surface runoff. The various domestic and social water use attributes of mountain rivers that are prone to flash floods, especially the current state of the river channel (water storage, water level, drainage capacity), are important factors affecting the risk of flood disasters.

[0005] Therefore, factors such as potential risks, soil moisture content in the early stage, regional geological data, and the relationship between rivers and tributaries should be included in the elements for determining early warning indicators. The suddenness and nonlinearity of flash floods make it difficult for early warning systems based on static models to respond to the ever-changing risk environment in real time. In particular, when it comes to the high uncertainty of multiple factors affecting the development of flash floods, the accuracy of early warnings using existing technologies is greatly reduced.

[0006] To address these technical problems, this application proposes a method for determining dynamic early warning indicators for flash floods based on multi-dimensional factor coupling. Summary of the Invention

[0007] The main objective of this invention is to provide a method for determining dynamic early warning indicators for flash floods based on multi-dimensional factor coupling. By comprehensively considering dynamic variables such as rainfall timing characteristics, changes in surface permeability, and soil moisture content, as well as the spatiotemporal coupling relationship of risks and hazards, an uncertainty distribution model is constructed. Combined with a two-dimensional numerical model, the flash flood inundation process and flow response are simulated and analyzed, realizing the transition from static risk to dynamic risk, supporting early identification, graded response, and early warning issuance of flash floods, thereby solving the technical problems mentioned in the background art.

[0008] The present invention solves the above-mentioned technical problems by adopting the following technical solutions: A method for determining dynamic early warning indicators for flash floods based on multi-dimensional factor coupling is proposed. The method consists of three modules: "data acquisition and preprocessing → grid rainfall model construction and inundation range simulation → flash flood influencing factor modeling and early warning rainfall coupling prediction of flash flood early warning indicators". Through standardized data processing, the method solves the problems that existing flash flood early warning indicators rely heavily on experience, provide static early warnings, and are difficult to adapt to complex terrain and engineering constraints, leading to missed or false alarms.

[0009] The method for determining dynamic early warning indicators for flash floods based on multi-dimensional factor coupling specifically executes the following steps using computer equipment: Step S1. Collect real-time and historical risk and hazard data from associated rain gauges, water level stations, flow stations, and soil moisture stations within the designated hazardous area under study. Specifically, this includes data on precipitation, land use, soil characteristics, and other types of risk and hazard data within the area. Step S2. Construct an HEC-RAS two-dimensional grid rainfall model based on the collected data and the simulated flood data; Step S3. The flood inundation range is determined by simulating the HEC-RAS two-dimensional grid rainfall model, and the flood control capacity of the specified object is determined based on the flood inundation range. The flood data designed by the simulation is used to determine the early warning rainfall for flash floods. Step S4. Incorporate multidimensional flash flood influencing factors, including potential hazards of river engineering facilities, previous soil moisture content, regional geological data, and current hydrological status of the river, into the rainfall model and construct the corresponding data model; Step S5. Based on a set of dynamic early warning rainfall data coupled with multidimensional flash flood influencing factors, a data model is used to correct the initial early warning rainfall threshold to form a dynamically perceived flash flood disaster early warning index.

[0010] Preferably, the construction process of the HEC-RAS two-dimensional grid rainfall model in step S2 includes: The core task of constructing a two-dimensional grid computing framework in the HEC-RAS environment to support the simulation of the entire process of rainfall-runoff-confluence is to integrate the real terrain, engineering structure and design net rainfall input into a unified numerical model structure, so as to provide a computing carrier for subsequent parameter configuration, flood simulation and back-calculation of early warning indicators. Specifically, a set of initial two-dimensional grid rainfall models are constructed based on the verified digital elevation model. The preset time-series rainfall is transformed into water depth variation and radial flow in the grid cells through the mass conservation equation and added to each two-dimensional grid cell of the rainfall model. When the rainfall intensity ignores porosity, the continuity equation can be written as:

[0011] in, Because of the water depth, For flow rate, This is the depth-averaged flow velocity vector. For rainfall (positive source) or other source-sink items, For based on The calculated porosity function; The momentum conservation described here takes the form of a two-dimensional shallow water equation (Saint-Venant equation), which includes terms such as local and convective acceleration, gravitational gradient, turbulent viscous diffusion, and bottom friction. At this point, the momentum equation (vector form) given internally by HEC-RAS in volume form is:

[0012] in, This is the water level elevation. For flow rate, This is a unit vector perpendicular to the plane of water flow, used to define the vector direction of the Coriolis force. The bottom frictional shear stress is calculated using the Manning formula. It is the acceleration due to gravity. The density of water, For hydraulic radius, Coriolis coefficient of the Coriolis term For turbulent kinematic viscosity coefficient, For surface wind shear stress and wave wind stress, Let be the wave radiation stress tensor. For gradient operators, Atmospheric pressure. and These are the additional drag coefficient and empirical parameters, respectively, used to characterize the additional nonlinear drag in the water flow. The above set of equations embodies the fluid conservation and kinematic equations, and can be solved by the implicit finite volume algorithm, which can automatically handle the flooding and drying process of the grid. After deducting infiltration losses according to the selected loss model, the remaining portion is included as surface runoff in the two-dimensional hydrodynamic calculation of the rainfall model; The infiltration loss is often represented by the empirical SCS-CN model, which assumes an initial loss. With potential maximum soil retention proportional ( (0.05≤r≤0.2), and there is accumulated excess rainfall:

[0013] in For cumulative rainfall, Related to the runoff curve number CN, the HEC-RAS uses: At this point, the actual runoff is The infiltration is calculated as rainfall minus runoff. At this time, the rainfall model receives the grid rainfall at each moment, deducts it according to the loss model, and then uses the shallow water equation to convert the remaining rainfall into runoff and promotes the flow on the grid with hydraulic balance and momentum conservation. After completing the basic data preprocessing and constructing net rainfall processes for different designs, the uniformly constructed net rainfall process is configured as the areal rainfall input into the rainfall model. The rainfall model framework reserves an interface for loss model parameters, but it is not configured at this time, so that the net rainfall process is seamlessly connected with the subsequent parameter setting part. The net rainfall is designed to be applied to the global grid in a uniform time sequence to ensure the comparability between subsequent simulations of different return periods. To ensure the operability of the rainfall model, the initial and boundary conditions of the rainfall model are set here, including setting water level or flow boundary at the watershed outlet, setting the initial wet and dry state in the slope area, and establishing the default time step and stability control terms. Only basic configuration is performed here. Once the rainfall model is built, a set of HEC-RAS engineering files that can be run independently is generated, including geometry files, mesh files, rainfall input files, and basic boundary condition files. Preferably, a hierarchical partitioning strategy is adopted during the initial two-dimensional mesh model construction process: A grid of a specified resolution is deployed around river channels, important control sections, concentrated village areas, and potential risk points to ensure the accuracy of depicting local hydrodynamic gradients, velocity changes, and inundation boundaries. In slopes and non-critical areas, the grid density is reasonably relaxed to balance computational needs and efficiency. The computational domain boundary is strictly set according to the watershed boundary line to avoid boundary offset causing errors in runoff area and slope. The hydraulic structures such as river cross-sections, bridges, culverts, dikes, and embankments are incorporated into the model's geometric system one by one; A coupling method of one-dimensional cross-section and two-dimensional grid is adopted for the river channel, so that the hydraulic processes of the main channel and the beach can be expressed in a coordinated manner; Bridges, culverts, and embankments are laid out using geometric objects or linear structures to reflect their actual role in water flow propagation, energy loss, and flood blocking. The spatial location, dimensions, and elevation information of the above elements are directly derived from the engineering element files and field measurement data compiled in the first module, ensuring that the geometric framework of the model is consistent with the actual terrain and engineering conditions.

[0014] Preferably, in the specific process of configuring the net rainfall process as areal rainfall input into the rainfall model, based on the completed rainfall model framework, the specified physical and numerical parameters required for numerical calculation are systematically configured and calibrated to ensure that the model can realistically reproduce the runoff generation and confluence process of the small watershed, providing a reliable parameter basis for subsequent flood simulation under design storm conditions. Specifically, this includes: Assignment of spatial roughness: Based on the differences in land use classification, vegetation type, surface roughness and renovation materials, Manning roughness is assigned to different land types and differentiated treatment is implemented for main river channels, beaches, villages and roads so that the simulation can reflect the spatial changes in the resistance of land features. Set up infiltration and runoff loss models: Based on the design net rainfall generated in the previous process, select the appropriate specified loss model (such as SCS-CN method or regional empirical formula) according to the watershed soil properties, cover type and previous water content, and configure its parameters so that the rainfall input can be converted into actual effective runoff. The numerical calculation parameters are configured, including time step, stability factor, wet-dry discrimination condition, iterative convergence control and boundary condition numerical format. These settings determine the stability of the model under heavy rain intensity and rapid response conditions, and need to be verified and adjusted through small-scale trial calculations. Parameter calibration: Using historical rainstorm and flood events as calibration samples, specified parameters, including roughness, loss model, and boundary conditions, are iteratively adjusted to match the simulated peak flow, peak arrival time, water level at key sections, and flooded area with actual records or survey results. After calibration, a parameter set report is generated, and this parameter set is fixed as the standard parameters for subsequent design of rainstorm simulations.

[0015] Preferably, after the HEC-RAS two-dimensional grid rainfall model is constructed, a set of two-dimensional flood simulations corresponding to the design return period are formally run to generate key results, including inundation depth, water level, flow rate, and flood peak process lines, under different return periods (set return periods of 5, 10, 20, 50, and 100 years). Specifically, these include: The net rainfall time series of the design storm is input. The net rainfall time series is determined based on rainfall parameters, coefficient of variation, and time series allocation method. Here, the time series allocation method adopts the regional storm time series allocation empirical method of the local hydrological manual. That is, based on the regional empirical allocation pattern extracted from the measured storm data in the watershed hydrological manual, the typical rainfall pattern and time series allocation weight of the corresponding duration and return period are retrieved, and the weight ratio of each time period is adjusted by combining the coefficient of variation. After calibration and optimization with measured data, the net rainfall time series is obtained, which is suitable for scenarios with complete hydrological manual data. If the study area lacks a local hydrological manual, the Chicago rainfall pattern method is adopted. The time series is constructed based on the regional storm intensity formula and the peak location coefficient. By dividing the time period before and after the peak and allocating the rainfall intensity according to the exponential law, it is converted into the net rainfall time series by the initial loss and subsequent loss method, realizing the rapid generation of standardized time series. The two methods can be selected as needed to accurately construct the time-by-time net rainfall process. Then, the rainfall model applies the net rainfall time series to the two-dimensional grid and calculates the effective runoff at each time under the action of the determined infiltration loss parameters. The rainfall model performs two-dimensional unsteady hydrodynamic calculations, using shallow water dynamic equations to describe the processes of slope confluence, river channel runoff, and flood diffusion in the floodplain. It outputs flow velocity, water depth, flood peak process line, and flooded area hourly. The water level and flow time series of the watershed outlet or key section are also calculated and output simultaneously to provide data support for subsequent hazard zone identification and capacity analysis. The simulation results are subjected to quality checks and rationality analyses, including mass conservation checks, peak arrival time rationality checks, and comparison checks with historical flood events, to ensure that the simulation under the design scenario is stable and reliable. If the verification fails, a parameter adjustment cycle is formed with the rainfall model parameter configuration until the simulation results meet the engineering and accuracy requirements.

[0016] Preferably, the flash flood impact factor data model in step S4 includes an impact factor for hidden dangers in river engineering facilities. This factor is used to address the impact of hidden dangers in river engineering facilities (bridges, culverts, weirs, dams, and land occupation in ditches and beaches) on flash flood risk. When the number of hidden danger projects increases, the marginal growth rate of risk slows down, i.e., a saturation effect of hidden dangers. Simultaneously, different types of hidden dangers have different water-blocking effects due to different engineering structures, i.e., water-blocking area effects. This effect is obtained by combining the water-blocking area ratio with the existing hidden danger quantity parameter and assigning weights. Its definition is:

[0017]

[0018] in, This is a risk hazard factor, with values ​​normalized to [0,1]. >0 represents the shape adjustment coefficient, used to adjust the rate of change of the function curve. >0 represents the water-blocking area enhancement coefficient. A value greater than 1 will amplify the contribution of the potential risks associated with a large water-blocking area to the risk factors. If <1, then the effect is weakened. Index for hazard types, =1, 2, 3, and 4 represent bridges, weirs, culverts, and ditches, respectively. For the first Basic risk weights for similar hidden dangers For the first The normalized water-blocking area ratio of similar hidden dangers, and , The actual water-blocking area (m²) of a single hidden danger 2 ), The maximum water-blocking area (m²) of this type of hidden danger 2 ), No. The actual number of such hidden dangers This represents the total number of weighted potential hazards based on the combined water-blocking area (the larger the water-blocking area, the greater the potential hazard). The larger the value, the higher the risk factor (the risk factor increases accordingly). The coefficients are dynamically updated, taking the maximum number of facilities that may appear within the study area to ensure... ≤1.

[0019] Preferably, since soil moisture content significantly affects rainfall runoff, lower moisture content requires more rainfall to form the same flood, the flash flood impact factor data model in step S4 also includes the anterior soil moisture content impact factor, defined as:

[0020] in, Factors affecting soil moisture content in the early stages This is the slope adjustment coefficient, used to control the steepness and saturation rate of the function curve, adapting to the water-holding characteristics of different soil types. This represents the effective rainfall in the early stages of the watershed (which can be the cumulative rainfall from upstream or the corresponding soil infiltration). To determine the maximum water storage capacity of the watershed soil (dynamically calibrated using remote sensing data), soil moisture content is categorized into three typical conditions: relatively dry, moderate, and relatively wet, each corresponding to a specific previous rainfall level. .

[0021] Preferably, since different geological cover types (sand, silt, rock, vegetation, etc.) correspond to different soil permeability, the flash flood impact factor data model in step S4 also includes regional geological data impact factors. Defined as:

[0022] in, >0 indicates a non-linear exponent. A value greater than 1 is used to indicate the amplified impact of high-permeability landforms. <1 is used to indicate the amplified impact of low-permeability land types. Indexes geological or surface cover types (e.g., sandy, silty, rocky, vegetation). Land category The empirical permeability coefficient (assigned to each land type based on soil permeability classification), and This value reflects the permeability under different geological types; the higher the value, the stronger the permeability, such as sandy surfaces. High, rock Low-level, specifically take =0.8, =0.5, =0.3, =0.1, Corresponding land category The area proportion within a watershed, used to represent the proportion of different geological types, can be obtained through geological surveys or remote sensing image analysis. =1; At this point, it is also possible to combine actual soil infiltration experiments or machine learning methods to... Calibration is performed to make the model more dynamically data-driven, and it also allows for... The perturbation weights are designed to reflect the uncertainty of measurement errors or geological changes.

[0023] Preferably, since the current state of the river channel (water storage, water level, drainage capacity) is an important factor affecting flood risk, the flash flood impact factor data model in step S4 also includes the influence factor of the relationship between the river channel and its tributaries based on the current hydrological state of the river channel, defined as:

[0024] in, The factors influencing the relationship between the river channel and its tributaries are determined by weighted averaging, which combines three proportional indicators—water storage, relative water level, and drainage—with the range normalized to between 0 and 1. , , These are respectively represented as the initial river channel water storage, the safe water level (warning water level, m), and the maximum safe drainage flow (m³). 3 ( / s), and all are estimated by telemetry or model. >0 indicates a relational factor used to adjust the power function, when If the value is greater than 1, then the high-risk sub-items are amplified. , These are the weights for the sub-items: water storage, water level, and flow rate. This represents the maximum water storage capacity of the river channel (or can be approximated by the current geometric capacity of the river channel). The current river level (m) Current river flow (m³) 3 / s); The meanings of each sub-item are: current water storage ratio, current water level to safe water level ratio, and current flow rate to drainage capacity ratio, and are dynamically set according to the actual situation of the watershed. and And adjust through real-time monitoring , , .

[0025] Preferably, the specific operation process of step S5 includes: Let the baseline critical rainfall (critical rainfall value from the HEC-RAS model) be... A dynamic weighting system integrating subjective and objective factors is designed based on multidimensional flash flood influencing factors to avoid the one-sidedness of a single weight. This system is used to adjust the initial critical rainfall, where the subjective weights are... The objective weights are calculated based on the relative importance scores, with each multidimensional flash flood influencing factor based on the coefficient of variation in historical samples. Calculations show that objective weights exist. The calculation formula is: , The number of influencing factors for multidimensional flash flood influencing factors. For this impact factor standard deviation For this impact factor The mean, finally calculated using a specified coefficient. Balancing subjective and objective weights, the dynamic weight calculation formula is as follows: ; By combining the impact factor response with dynamic weights, the coupling function is constructed as follows:

[0026] in, This is the adjusted critical rainfall. The initial critical rainfall (m) output by the HEC-RAS model. For the aforementioned factors, a larger value generally indicates a higher risk. A value greater than 0 indicates a coupling index, used to agitate high-risk factors. >0 represents the intensity coefficient of the cross-term factor, used to highlight the synergistic effect of various flash flood influencing factors. The weights of the corresponding impact factors can be set by expert experience or optimized through data-driven methods. For cross weights, such as =0.3, used to reflect the interaction between soil moisture and geological infiltration. If the factor value in the coupling function increases (increases the risk), then... Reduce (issue early warnings), and vice versa. Furthermore, the coupling function also needs to ensure... It has a reasonable physical meaning, therefore it can be used for value range or Apply constraints; Finally, a tiered warning level triggering logic is introduced, which escalates warnings based on the comprehensive factor value exceeding a threshold. When the comprehensive risk indicator or a single factor exceeds a preset threshold, the warning level is automatically raised, and the rainfall threshold is adjusted accordingly. Specifically, this includes: Constructing a power-means comprehensive risk index To achieve non-linear hierarchical triggering, the following indicators are constructed:

[0027] in, This indicates the power parameter of the risk indicator; Set level threshold The following warning rules are specified to trigger the warning level:

[0028] in, This indicates the triggered warning level, and the various thresholds. It can be formulated or optimized based on historical cases, and restrictions can also be set on single factors, such as arbitrary... If the standard is exceeded, the increase will be directly applied.

[0029] The critical rainfall level was further adjusted according to different levels. This can be given by the following empirical rules or functions:

[0030] in, >0 indicates the warning threshold trigger coefficient. These are level numbers (1–4), corresponding to the warning colors. Different risk thresholds; This logic enables tiered triggering and escalation when conditions exceed limits: when weather or real-time monitoring conditions deteriorate and a certain threshold is exceeded, the system automatically upgrades the warning and recalculates a more stringent rainfall index. The threshold setting here is continuously optimized using big data backtesting to make the warning level more consistent with the actual disaster situation. In addition, the initial critical rainfall can be obtained using the HEC-RAS two-dimensional grid rainfall model. Based on this, and combined with the adjustment coefficients calculated from the above dynamic model results, the final warning rainfall threshold is obtained. As a dynamic and context-sensitive early warning indicator.

[0031] In another aspect, the present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method described above.

[0032] In another aspect, the present invention also discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method described above.

[0033] As can be seen from the above technical solution, the present invention provides a method for determining dynamic early warning indicators for flash flood disasters based on multi-dimensional element coupling. Compared with the prior art, the present invention has the following advantages: 1. By comprehensively considering dynamic variables such as rainfall timing characteristics, changes in surface permeability, and soil moisture content, as well as the spatiotemporal coupling relationship of risks and hazards, this invention can construct an uncertainty distribution model. This model can be used in conjunction with a two-dimensional numerical model to simulate and analyze the flash flood inundation process and flow response, ultimately achieving the transition from static risk to dynamic risk. It can also be used to support early identification, graded response, and early warning issuance of flash floods.

[0034] 2. This invention, through multi-dimensional factor coupling during dynamic rainfall adjustment, can systematically correct the initial threshold of HEC-RAS. It has the characteristics of being structured and controllable. When combined with real-time monitoring or simulation data, dynamic learning and data-driven adjustment can be achieved. It can effectively couple the traditional HEC-RAS model with the actual situation on site to form a dynamic situation-aware early warning rainfall index, and enhance the scientificity, practicality and innovation of the early warning method. It is conducive to building a more sensitive and reliable flash flood disaster early warning system.

[0035] It should be understood that the descriptions in this section are not intended to identify key or essential features of embodiments of the invention, nor are they intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Of course, implementing any product of the invention does not necessarily require achieving all of the advantages described above simultaneously. Attached Figure Description

[0036] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a schematic diagram of the overall process of the present invention; Figure 2 This is a schematic diagram illustrating the coupling adjustment of the critical rainfall warning index in this invention. Detailed Implementation

[0037] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. 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.

[0038] For details in the embodiments, please refer to Figures 1 to 2 .

[0039] like Figure 1 and Figure 2 As shown in the figure, the method for determining dynamic early warning indicators for flash floods based on multi-dimensional factor coupling proposed in this embodiment of the invention includes the following steps: Step L1. Data Acquisition and Preprocessing.

[0040] Through risk and hazard investigation, we can identify potential risks and hazards such as bridges, weirs, dams, and gullies within the danger zone; and collect high-precision geographical data on topography, land use, and vegetation within the danger zone.

[0041] Specifically, basic hydrological and geographical data of the study area are obtained through a multi-source channel system, including hourly or daily rainfall observation sequences, rainfall data from representative stations, regional rainstorm parameter atlases, digital elevation models, land use and vegetation cover maps, soil infiltration and runoff characteristic parameters, historical water level and flow records, control section measurement data, past flood disaster records, and administrative boundary and element information related to the prevention and control targets.

[0042] The aforementioned data must be spatially unified using a coordinate system, and in terms of attributes, missing data must be filled in, outliers must be identified, and statistical consistency checks must be performed to ensure that the data can serve as a common base for subsequent calculations.

[0043] Subsequently, a high-precision digital elevation model was used to systematically verify the watershed boundary, river topography, and hill slopes, including filling depressions, filtering, and correcting the longitudinal and cross sections of the river based on measured point cloud data, so that the hydrodynamic model can realistically depict the micro-topographic differences within the small watershed. In terms of land use and soil parameter processing, different data sources were clipped to a consistent watershed boundary range, and raster resolution was unified and classification recoding was performed to ensure that runoff generation and runoff calculations adopted the same spatial scale and classification system.

[0044] After the data base was constructed, representative rainfall amounts and coefficients of variation for different durations such as 1h and 24h were derived from the analysis of rainstorm frequency and regional rainstorm parameter atlas. The temporal structure of each return period rainfall process was allocated according to relevant flood calculation methods, and finally the original rainfall statistics parameters were transformed into a complete design rainfall process.

[0045] Subsequently, in accordance with the requirements of the runoff generation model and based on the laws of soil infiltration, initial loss and runoff generation, the design rainfall process is converted into the design net rainfall process line, which constitutes the time-series input conditions of the hydrodynamic model. At this time, the multiple sets of net rainfall processes at different time periods correspond to different design return periods (5, 10, 20, 50, 100 years) to ensure that the subsequent model operation can systematically present the flood response under various levels of rainstorm conditions.

[0046] Finally, all the preprocessed data are packaged into a standardized data package that can be directly read by hydrodynamic models, geographic information systems, and analysis modules. The data package includes digital elevation models, land use layers, runoff and infiltration parameters, design net rainfall process documents, control section information, and historical verification data. Each type of data is accompanied by complete metadata descriptions, including data source, time range, processing method, and spatial accuracy, to ensure that model simulation, parameter calibration, and result verification are all conducted using the same technical standards.

[0047] Step L2. Construct the HEC-RAS two-dimensional grid rainfall model.

[0048] Based on the collected data and the simulated flood data, a two-dimensional grid computing framework capable of supporting the simulation of the entire process of rainfall-runoff-confluence is constructed in the HEC-RAS environment. Its core task is to integrate the real terrain, engineering structure and design net rainfall input into a unified numerical model structure, providing a computing carrier for subsequent parameter configuration, flood simulation and back-calculation of early warning indicators. Specifically, an initial two-dimensional grid rainfall model is constructed based on the verified digital elevation model.

[0049] At this point, a hierarchical partitioning strategy is used during the initial two-dimensional mesh model construction process: (1) Layout a grid of specified resolution around the river channel, important control sections, concentrated village areas and potential risk points to ensure the accuracy of depiction of local hydrodynamic gradient, velocity change and inundation boundary. (2) In the slope and non-critical areas, the grid density is reasonably relaxed to balance the computational needs and efficiency. The boundary of the computational domain is strictly set according to the watershed boundary line to avoid the boundary offset causing errors in the runoff area and slope. (3) Incorporate the river cross-section, bridges, culverts, dikes, and other hydraulic structures into the model's geometric system one by one; (4) The river channel is coupled with a one-dimensional cross-section and a two-dimensional grid so that the hydraulic processes of the main channel and the beach can be expressed in a coordinated manner; (5) Bridges, culverts and dikes are laid out using geometric objects or linear structures to reflect their real role in water flow propagation, energy loss and flood blocking.

[0050] The spatial location, dimensions, and elevation information of the above elements are directly derived from the engineering element files and field measurement data compiled in the first module, ensuring that the geometric framework of the model is consistent with the actual terrain and engineering conditions.

[0051] The preset time-series rainfall is then transformed into water depth variation and radial flow in the grid cells using the mass conservation equation and added to each two-dimensional grid cell of the rainfall model. When the rainfall intensity ignores porosity, the continuity equation can be written as:

[0052] in, Because of the water depth, This is the depth-averaged flow velocity vector. For rainfall (positive source) or other source / sink items.

[0053] The momentum conservation here is in the form of a two-dimensional shallow water equation (Saint-Venant equation), which includes terms such as local and convective acceleration, gravitational gradient, turbulent viscous diffusion, and bottom friction.

[0054] At this point, the momentum equation (vector form) given internally by HEC-RAS in volume form is:

[0055] in This is the water level elevation. The bottom frictional shear stress is calculated using the Manning formula. Coriolis coefficient of the Coriolis term For turbulent kinematic viscosity coefficient, For surface wind shear stress and wave wind stress, The above equations, including wave radiation stress tensor, embody fluid conservation and kinematic equations. Solving them using an implicit finite volume algorithm can automatically handle the submergence and drying process of the mesh.

[0056] Subsequently, infiltration losses are deducted according to the selected loss model, and the remaining portion is incorporated as surface runoff into the two-dimensional hydrodynamic calculation of the rainfall model.

[0057] The infiltration loss is often represented by the empirical SCS-CN model, which assumes an initial loss. With potential maximum soil retention proportional ( (0.05≤r≤0.2), and there is accumulated excess rainfall:

[0058] in For cumulative rainfall, Related to the runoff curve number CN, the HEC-RAS uses: At this point, the actual runoff is Infiltration is calculated as rainfall minus runoff. Therefore, the rainfall model receives the grid rainfall at each moment, deducts it according to the loss model, and then uses the shallow water equation to convert the remaining rainfall into runoff and promotes the flow on the grid with hydraulic balance and momentum conservation. At this time, for mountainous areas, the SCS-CN model gives the characteristic parameters of the catchment area through soil type, cover and previous soil moisture content, which can reflect the characteristics of shallow soil and steep slope in mountainous areas.

[0059] After completing the basic data preprocessing and constructing net rainfall processes for different designs, the uniformly constructed net rainfall process is configured as the areal rainfall input into the rainfall model.

[0060] In the specific process of designing the net rainfall process as the areal rainfall input and configuring it into the rainfall model, based on the completed rainfall model framework, the specified physical and numerical parameters required for numerical calculations are systematically configured and calibrated to ensure that the model can realistically reproduce the runoff generation and confluence processes of the small watershed. This provides a reliable parameter basis for subsequent flood simulation under design storm conditions, specifically including: (a) Assignment of spatial roughness: Based on the differences in land use classification, vegetation type, surface roughness and renovation materials, Manning roughness is assigned to different land types and differentiated treatment is implemented for main river channels, beaches, villages and roads so that the simulation can reflect the spatial changes of ground resistance. (b) Setting up infiltration and runoff loss models: Based on the design net rainfall generated in the previous process, select the appropriate specified loss model (such as the SCS-CN method or regionalized empirical formula) according to the watershed soil properties, cover type and previous water content, and configure its parameters so that the rainfall input can be converted into actual effective runoff. (c) Configure the numerical calculation parameters, including time step, stability factor, wet-dry discrimination condition, iterative convergence control and boundary condition numerical format. These settings determine the stability of the model under heavy rain intensity and rapid response conditions, and need to be verified and adjusted through small-scale trial calculations. (d) Parameter calibration: Using historical rainstorm and flood events as calibration samples, specified parameters, including roughness, loss model, and boundary conditions, are iteratively adjusted to match the simulated peak flow, peak arrival time, water level at key sections, and flooded area with actual records or survey results. After calibration, a parameter set report is generated, and this parameter set is fixed as the standard parameters for subsequent design of rainstorm simulations.

[0061] The rainfall model framework reserves an interface for loss model parameters, but these are not configured at this time. This ensures a seamless connection between the net rainfall process and subsequent parameter settings. The net rainfall is designed to be applied to the global grid in a uniform time sequence to guarantee comparability between simulations of different return periods.

[0062] In addition, to ensure the operability of the rainfall model, the initial and boundary conditions of the rainfall model are set here, including setting water level or flow boundary at the watershed outlet, setting the initial dry and wet state in the slope area, and establishing the default time step and stability control terms. Only basic configuration is performed here.

[0063] Once the final rainfall model is built, a set of HEC-RAS engineering files that can be run independently will be formed, including geometry files, mesh files, rainfall input files, and basic boundary condition files.

[0064] In addition, after the HEC-RAS two-dimensional grid rainfall model is constructed, a set of two-dimensional flood simulations corresponding to the design return period needs to be formally run to generate key results, including inundation depth, water level, flow rate, and peak flow process lines, under different return periods (set return periods of 5, 10, 20, 50, and 100 years). Specifically, these include: a. Input the net rainfall time series of the design storm. The net rainfall time series is determined based on rainfall parameters, coefficient of variation, and time series allocation method. One time series allocation method is to adopt the regional storm time series allocation experience method of the local hydrological manual. That is, based on the regional experience allocation pattern extracted from the measured storm data in the watershed hydrological manual, the typical rainfall pattern and time series allocation weight of the corresponding duration and return period are retrieved, and the weight ratio of each time period is adjusted by combining the coefficient of variation. After calibration and optimization by measured data, the net rainfall time series is obtained, which is suitable for scenarios with complete hydrological manual data. If the study area lacks a local hydrological manual, the Chicago rainfall pattern method is adopted. The time series is constructed based on the regional storm intensity formula and the peak location coefficient. By dividing the time period before and after the peak and allocating the rainfall intensity according to the exponential law, it is converted into the net rainfall time series by the initial loss and subsequent loss method, realizing the rapid generation of standardized time series. The two methods can be selected as needed to accurately construct the time-by-time net rainfall process. Then the rainfall model applies the net rainfall time series to the two-dimensional grid and calculates the effective runoff at each moment under the action of the determined infiltration loss parameters. b. The rainfall model performs two-dimensional unsteady hydrodynamic calculations, using shallow water dynamic equations to describe the process of slope confluence, river channel runoff and flood diffusion in the beach area. It outputs flow velocity, water depth, flood peak process line and flooded area hourly. The water level and flow time series of the watershed outlet or key section are also calculated and output simultaneously to provide data support for subsequent hazard area identification and capacity analysis. c. Conduct quality checks and rationality analyses on the simulation results, including mass conservation checks, peak arrival time rationality checks, and comparison checks with historical flood events, to ensure that the simulation under the design scenario is stable and reliable. If the verification fails, a parameter adjustment cycle is formed with the rainfall model parameter configuration until the simulation results meet the engineering and accuracy requirements.

[0065] For the constructed model, the flood inundation range is determined by simulating the HEC-RAS two-dimensional grid rainfall model during use, and the flood control capacity of the specified object is determined based on the flood inundation range. The early warning rainfall for flash flood disasters is determined by the simulated flood data.

[0066] Step L3. Predict flash flood warning indicators by modeling flash flood influencing factors and coupling them with early warning rainfall.

[0067] Multidimensional flash flood influencing factors, including potential hazards of river engineering facilities, previous soil moisture content, regional geological data, and current hydrological status of the river, are introduced into the rainfall model, and corresponding data models are constructed. Each factor represents the flood risk under dimensions such as hydrodynamic response and geological environment.

[0068] in: A. The flash flood impact factor data model in step S4 includes the impact factor of potential hazards in engineering facilities within the river channel. This parameter is used to assess the impact of potential hazards in river engineering facilities (bridges, culverts, weirs, dams, and land occupation in ditches) on flash flood risk. As the number of hazard projects increases, the marginal rate of risk growth slows down, indicating a hazard saturation effect. Simultaneously, different types of hazards exhibit varying water-blocking effects due to their different engineering structures, i.e., water-blocking area effects. This parameter is obtained by combining the water-blocking area ratio with the existing hazard quantity parameter and assigning weights. Its definition is:

[0069]

[0070] in, This is a risk hazard factor, with values ​​normalized to [0,1]. >0 represents the shape adjustment coefficient, used to adjust the rate of change of the function curve. >0 represents the water-blocking area enhancement coefficient. A value greater than 1 will amplify the contribution of the potential risks associated with a large water-blocking area to the risk factors. If <1, then the effect is weakened. Index for hazard types, =1, 2, 3, and 4 represent bridges, weirs, culverts, and ditches, respectively. For the first Basic risk weights for similar hidden dangers For the first The normalized water-blocking area ratio of similar hidden dangers, and , The actual water-blocking area (m²) of a single hidden danger 2 ), The maximum water-blocking area (m²) of this type of hidden danger 2 ), No. The actual number of such hidden dangers This represents the total number of weighted potential hazards based on the combined water-blocking area (the larger the water-blocking area, the greater the potential hazard). The larger the value, the higher the risk factor (the risk factor increases accordingly). The coefficients are dynamically updated, taking the maximum number of facilities that may appear within the study area to ensure... ≤1.

[0071] B. Since soil moisture content significantly affects rainfall runoff, lower moisture content requires more rainfall to form the same flood. Therefore, the flash flood impact factor data model in step S4 also includes the impact factor of previous soil moisture content. Defined as:

[0072] in, Factors affecting soil moisture content in the early stages This is the slope adjustment coefficient, used to control the steepness and saturation rate of the function curve, adapting to the water-holding characteristics of different soil types. This represents the effective rainfall in the early stages of the watershed (which can be the cumulative rainfall from upstream or the corresponding soil infiltration). To determine the maximum water storage capacity of the watershed soil (dynamically calibrated using remote sensing data), soil moisture content is categorized into three typical conditions: relatively dry, moderate, and relatively wet, each corresponding to a specific previous rainfall level. .

[0073] In a specific set of examples, when the soil is relatively dry ( =0.2 )hour, Approaching 0, the risk of runoff is extremely low; when the soil is in a normal state ( =0.5 )hour, =0.5, with a moderate risk of runoff, especially when the soil is relatively moist ( =0.8 )hour, A rapid rise to above 0.9 significantly increases the risk of runoff; when the soil is close to saturation ( When approaching Im), The value approaches 1 and the growth rate slows down, which is consistent with the soil water saturation effect. This can be calibrated using historical runoff data based on the watershed soil type (sandy, clay, loam). Alternatively, remote sensing soil moisture index correction could be introduced. It supports data-driven dynamic updates.

[0074] C. Since different geological cover types (sand, silt, rock, vegetation, etc.) correspond to different soil permeability, the flash flood impact factor data model in step S4 also includes regional geological data impact factors. Defined as:

[0075] in, >0 indicates a non-linear exponent. A value greater than 1 is used to indicate the amplified impact of high-permeability landforms. <1 is used to indicate the amplified impact of low-permeability land types. Indexes geological or surface cover types (e.g., sandy, silty, rocky, vegetation). Land category The empirical permeability coefficient (assigned to each land type based on soil permeability classification), and This value reflects the permeability under different geological types; the higher the value, the stronger the permeability, such as sandy surfaces. High, rock Low-level, specifically take =0.8, =0.5, =0.3, =0.1, Corresponding land category The area proportion within a watershed, used to represent the proportion of different geological types, can be obtained through geological surveys or remote sensing image analysis. =1; At this point, it is also possible to combine actual soil infiltration experiments or machine learning methods to... Calibration is performed to make the model more dynamically data-driven, and it also allows for... The perturbation weights are designed to reflect the uncertainty of measurement errors or geological changes.

[0076] D. Since the current state of the river channel (water storage, water level, drainage capacity) is an important factor affecting flood risk, the flash flood impact factor data model in step S4 also includes the influence factors of the relationship between the river channel and its tributaries based on the current hydrological state of the river channel. Defined as:

[0077] in, The factors influencing the relationship between the river channel and its tributaries are determined by weighted averaging, which combines three proportional indicators—water storage, relative water level, and drainage—with the range normalized to between 0 and 1. , , These represent the initial river channel storage, safe water level, and safe drainage flow, respectively, all estimated by telemetry or modeling. >0 indicates a relational factor used to adjust the power function, when If the value is greater than 1, then the high-risk sub-items are amplified. , These are the weights for the sub-items: water storage, water level, and flow rate. This represents the maximum water storage capacity of the river channel (or can be approximated by the current geometric capacity of the river channel). The current river level (m) The safe water level (warning water level, m) is the warning water level. Current river flow (m³) 3 / s), Maximum safe drainage flow rate (m 3 / s); The meanings of each sub-item are: current water storage ratio, current water level to safe water level ratio, and current flow rate to drainage capacity ratio, and are dynamically set according to the actual situation of the watershed. and And adjust through real-time monitoring , , .

[0078] Finally, the above four types of factors are integrated and used to correct the initial critical rainfall threshold for early warning, forming a dynamic perception flash flood disaster early warning indicator. The specific operational procedures include: Step 1: Set the baseline critical rainfall (critical rainfall value from the HEC-RAS model) as follows: A dynamic weighting system integrating subjective and objective factors is designed based on multidimensional flash flood influencing factors to avoid the one-sidedness of a single weight. This system is used to adjust the initial critical rainfall, where: By ranking the four types of factors according to their watershed risk sensitivity, a priority graph matrix was constructed, and the relative importance scores were calculated to obtain the subjective weights. The subjective weight vector is:

[0079] in, These are the influencing factors of potential hazards in engineering facilities within the river channel. Factors affecting soil moisture content in the early stage Regional geological data impact factors Factors influencing the relationship between river channels and tributaries Subjective weighting; The objective weights are based on the coefficient of variation of each multidimensional flash flood influencing factor in the historical sample. Calculations show that objective weights exist. The calculation formula is:

[0080] The number of influencing factors for multidimensional flash flood influencing factors. For this impact factor standard deviation For this impact factor The mean; The objective weight vector is:

[0081] in, These are the influencing factors of potential hazards in engineering facilities within the river channel. Factors affecting soil moisture content in the early stage Regional geological data impact factors Factors influencing the relationship between river channels and tributaries Objective weight; Finally, use the specified coefficients Balancing subjective and objective weights, the dynamic weight calculation formula is as follows: ; The final weight vector is: .

[0082] Step 2: Combine the impact factor response with dynamic weights to construct the coupling function as follows:

[0083] in, This is the adjusted critical rainfall. The initial critical rainfall (m) output by the HEC-RAS model. For the aforementioned factors, a larger value generally indicates a higher risk. A value greater than 0 indicates a coupling index, used to agitate high-risk factors. >0 represents the intensity coefficient of the cross-term factor, used to highlight the synergistic effect of various flash flood influencing factors. The weights of the corresponding impact factors can be set by expert experience or optimized through data-driven methods. For cross weights, such as =0.3, used to reflect the interaction between soil moisture and geological infiltration. If the factor value in the coupling function increases (increases the risk), then... Reduce (issue early warnings), and vice versa. Furthermore, the coupling function also needs to ensure... It has a reasonable physical meaning, therefore it can be used for value range or Apply constraints; Step 3: Finally, introduce a tiered warning level triggering logic. This logic escalates warnings based on the overall factor value exceeding a threshold. When the overall risk indicator or a single factor exceeds a preset threshold, the warning level is automatically raised, and the rainfall threshold is adjusted accordingly. Specifically, this includes: A power-means type comprehensive risk index is constructed to achieve non-linear hierarchical triggering. The index is constructed as follows:

[0084] in, This indicates the power parameter of the risk indicator; Set level threshold The following warning rules are specified to trigger the warning level:

[0085] in, This indicates the triggered warning level, and the various thresholds. It can be formulated or optimized based on historical cases, and restrictions can also be set on single factors, such as arbitrary... If the standard is exceeded, the increase will be directly applied.

[0086] The critical rainfall level was further adjusted according to different levels. This can be given by the following empirical rules or functions:

[0087] in, >0 indicates the warning threshold trigger coefficient. These are level numbers (1–4), corresponding to the warning colors. Different risk thresholds; This logic enables tiered triggering and escalation when conditions exceed limits: when weather or real-time monitoring conditions deteriorate and a certain threshold is exceeded, the system automatically upgrades the warning and recalculates a more stringent rainfall index. The threshold setting here is continuously optimized using big data backtesting to make the warning level more consistent with the actual disaster situation. In addition, the initial critical rainfall can be obtained at this time using the HEC-RAS two-dimensional grid rainfall model. Based on this, and combined with the adjustment coefficients calculated from the above dynamic model results, the final warning rainfall threshold is obtained. As a dynamic and context-sensitive early warning indicator.

[0088] This coupling enables the fusion of traditional static thresholds based on engineering hydrodynamics with real-time information from multiple on-site factors: under normal conditions, ≈ Under adverse conditions such as high risk of accidents, high humidity, and low permeability, The alert level is automatically lowered to improve its sensitivity. If a tiered triggering system is used, the higher the alert level, the greater the reduction in alert level, forming a dynamic response mechanism that links the number of alerts.

[0089] In summary, this method systematically corrects the initial threshold of HEC-RAS through multidimensional factor coupling, and has the characteristics of being structured and adjustable.

[0090] Furthermore, this method combines real-time monitoring or simulation data, enabling dynamic learning and data-driven adjustments, which enhances the scientific rigor, practicality, and innovation of the early warning method, and is conducive to building a more sensitive and reliable flash flood disaster early warning system.

[0091] In another aspect, the present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method described above.

[0092] In another aspect, the present invention also discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method described above.

[0093] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the above embodiments of the method for determining dynamic early warning indicators of flash flood disasters based on multi-dimensional element coupling.

[0094] It is understood that the system provided in the embodiments of the present invention corresponds to the method provided in the embodiments of the present invention, and the explanation, examples and beneficial effects of the relevant content can be referred to the corresponding parts of the above method.

[0095] This application also provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, communication interface, and memory communicate with each other via the communication bus. Memory, used to store computer programs; When the processor executes the program stored in the memory, it implements the above-mentioned method for determining dynamic early warning indicators for flash flood disasters based on multi-dimensional factor coupling.

[0096] The communication bus mentioned in the above-mentioned electronic devices can be a standard bus for interconnecting peripheral components or an extended industrial standard structure bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc.

[0097] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0098] The memory may include random access memory or non-volatile memory, such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0099] The processors mentioned above can be general-purpose processors, including central processing units, network processors, etc.; they can also be digital signal processors, application-specific integrated circuits, field-programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0100] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, an optical medium, or a semiconductor medium, etc.

[0101] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0102] Furthermore, it should be noted that if any directional indication (such as up, down, left, right, front, back, etc.) is involved in the embodiments of the present invention, the directional indication is only used to explain the relative positional relationship and movement of each component in a specific posture. If the specific posture changes, the directional indication will also change accordingly.

[0103] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the meaning of "and / or" throughout the text includes three parallel solutions; for example, "A and / or B" includes solution A, solution B, or a solution where both A and B are satisfied simultaneously. Furthermore, in the embodiments of this invention, "multiple" refers to two or more. Moreover, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

Claims

1. A method for determining a dynamic early warning index of mountain flood disaster based on multi-dimensional element coupling, characterized in that, include: Step S1. Collect real-time and historical risk and hazard data from associated rain gauges, water level stations, flow stations, and soil moisture stations within the hazardous area; Step S2. Construct an HEC-RAS two-dimensional grid rainfall model based on the collected data and the simulated flood data; Step S3. The flood inundation range is determined by simulating the HEC-RAS two-dimensional grid rainfall model, and the flood control capacity of the specified object is determined based on the flood inundation range. The flood data designed by the simulation is used to determine the early warning rainfall for flash floods. Step S4. Incorporate multidimensional flash flood influencing factors, including potential hazards of river engineering facilities, previous soil moisture content, regional geological data, and current hydrological status of the river, into the rainfall model and construct the corresponding data model; Step S5. Based on a set of dynamic early warning rainfall data coupled with multidimensional flash flood influencing factors, a data model is used to correct the initial early warning rainfall threshold to form a dynamically perceived flash flood disaster early warning index.

2. The method for determining dynamic early warning indicators for flash floods based on multi-dimensional factor coupling as described in claim 1, characterized in that, The construction process of the HEC-RAS two-dimensional grid rainfall model in step S2 includes: In the HEC-RAS environment, an initial two-dimensional grid rainfall model is constructed based on the verified digital elevation model. The preset time-series rainfall is transformed into water depth changes and radial flow in the grid cells through the mass conservation equation and added to each two-dimensional grid cell of the rainfall model; After deducting infiltration losses according to the selected loss model, the remaining portion is included as surface runoff in the two-dimensional hydrodynamic calculation of the rainfall model; The infiltration loss was modeled using the SCS-CN model, with an initial loss of... With potential maximum soil retention Proportional, and with accumulated excess rainfall: in For cumulative rainfall, Related to the runoff curve number CN, the HEC-RAS uses: At this time, the rainfall model receives the grid rainfall at each moment, deducts it according to the loss model, and then uses the shallow water equation to convert the remaining rainfall into runoff and promotes the flow on the grid with hydraulic balance and momentum conservation. After completing the basic data preprocessing and constructing net rainfall processes for different designs, the uniformly constructed net rainfall process is configured as the areal rainfall input into the rainfall model. The initial and boundary conditions of the rainfall model are set up, including setting water level or flow boundary at the watershed outlet, setting initial wet and dry conditions in the slope area, and establishing default time steps and stability control terms. Once the rainfall model is built, a set of HEC-RAS engineering files that can be run independently is generated, including geometry files, mesh files, rainfall input files, and basic boundary condition files.

3. The method for determining dynamic early warning indicators for flash floods based on multi-dimensional factor coupling as described in claim 2, characterized in that, The initial two-dimensional mesh model is constructed using a hierarchical partitioning strategy. A grid of a specified resolution is deployed around river channels, important control sections, concentrated village areas, and potential risk points. On slopes and in non-critical areas, the grid density should be reasonably relaxed to balance computational needs and efficiency. The model incorporates hydraulic structures such as river cross-sections, bridges, culverts, dikes, and embankments into its geometric system. A coupling method of one-dimensional cross-section and two-dimensional grid is adopted for the river channel, so that the hydraulic processes of the main channel and the beach can be expressed in a coordinated manner; Bridges, culverts, and embankments are laid out using geometric objects or linear structures.

4. The method for determining dynamic early warning indicators for flash floods based on multi-dimensional factor coupling as described in claim 3, characterized in that, The process of configuring the net rainfall process as an isal rainfall input into the rainfall model involves, based on the rainfall model framework, systematically configuring and calibrating the specified physical and numerical parameters required for numerical calculations. Specifically, this includes: Based on the differences in land use classification, vegetation type, surface roughness and renovation materials, Manning roughness is assigned to different land types and differentiated treatment is implemented for main river channels, beaches, villages and roads. Based on the generated design net rainfall, the corresponding loss model is selected and its parameters are configured according to the watershed soil properties, cover type and previous water content. Configure the numerical calculation parameters, including time step, stability factor, wet / dry discrimination condition, iterative convergence control and boundary condition numerical format; Parameter calibration: Using historical rainstorm and flood events as calibration samples, specified parameters, including roughness, loss model, and boundary conditions, are iteratively adjusted to match the simulated peak flow, peak arrival time, water level at key sections, and flooded area with actual records or survey results. After calibration, a parameter set report is generated, and this parameter set is fixed as the standard parameters for subsequent design of rainstorm simulations.

5. The method for determining dynamic early warning indicators for flash floods based on multi-dimensional factor coupling as described in claim 3, characterized in that, After the HEC-RAS two-dimensional grid rainfall model was constructed, a set of two-dimensional flood simulations corresponding to the design return period were officially run to generate inundation depth, water level, flow rate, and peak flow process lines under different return period conditions, specifically including: Input the net rainfall time series of the design storm. The net rainfall time series is determined based on rainfall parameters, coefficient of variation and time series allocation method. Then the rainfall model applies the net rainfall time series to the two-dimensional grid and calculates the effective yield at each moment under the influence of the determined infiltration loss parameters. The rainfall model performs two-dimensional unsteady hydrodynamic calculations, using shallow water dynamic equations to describe the processes of slope confluence, river channel runoff, and flood zone diffusion, and outputs flow velocity, water depth, flood peak process line, and flooded area hourly. The simulation results are subjected to quality checks and rationality analyses, including mass conservation checks, peak arrival time rationality checks, and comparison checks with historical flood events. If the verification fails, a parameter adjustment cycle is formed with the rainfall model parameter configuration until the simulation results meet the verification requirements.

6. The method for determining dynamic early warning indicators for flash floods based on multi-dimensional factor coupling as described in claim 2, characterized in that, The flash flood impact factor data model in step S4 includes the impact factor of potential hazards in river engineering facilities. This is obtained by weighting and synthesizing the water-blocking area ratio and the existing hazard quantity parameter, and is defined as follows: in, As a risk factor, >0 represents the shape adjustment coefficient. >0 represents the water-blocking area enhancement coefficient. Index for hazard types, =1, 2, 3, and 4 represent bridges, weirs, culverts, and ditches, respectively. For the first Basic risk weights for similar hidden dangers For the first The normalized water-blocking area ratio of similar hidden dangers, and , The actual water-blocking area of ​​a single type of hidden danger. This represents the maximum water-blocking area for this type of potential hazard. No. The actual number of such hidden dangers This represents the total number of weighted potential hazards based on the combined water-blocking area. The coefficient is dynamically updated, taking the maximum number of facilities that may appear within the study area.

7. The method for determining dynamic early warning indicators for flash floods based on multi-dimensional factor coupling as described in claim 2, characterized in that, The flash flood impact factor data model in step S4 includes the anterior soil moisture content impact factor, defined as: in, Factors affecting soil moisture content in the early stages This is the slope adjustment coefficient. This represents the effective rainfall in the basin in the early stages. To determine the maximum water storage capacity of the watershed soil, soil moisture content is categorized into three typical conditions: relatively dry, moderate, and relatively wet, each corresponding to a specific amount of prior rainfall. .

8. The method for determining dynamic early warning indicators for flash floods based on multi-dimensional factor coupling as described in claim 2, characterized in that, The flash flood impact factor data model in step S4 includes regional geological data impact factors, defined as: in, >0 indicates a non-linear exponent. Index for geological or surface cover types. Land category The empirical penetration coefficient, and This value reflects the permeability under different geological conditions; a higher value indicates greater permeability. Corresponding land category The area proportion within a watershed is used to represent the proportion of different geological types.

9. The method for determining dynamic early warning indicators for flash floods based on multi-dimensional factor coupling as described in claim 2, characterized in that, The flash flood impact factor data model in step S4 includes impact factors related to the relationship between the river channel and its tributaries, based on the current hydrological state of the river channel. These factors are defined as follows: in, Factors influencing the relationship between river channels and tributaries. , , These represent the initial river channel water storage, safe water level, and safe drainage flow, respectively. >0 indicates a relational factor. , These are the weights for the sub-items: water storage, water level, and flow rate. The maximum water storage capacity of the river channel This is the current river level. For safe water level, The current river flow rate, This is the maximum safe drainage flow rate.

10. The method for determining dynamic early warning indicators for flash floods based on multi-dimensional factor coupling as described in claim 2, characterized in that, The specific operation process of step S5 includes: Let the baseline critical rainfall be A dynamic weighting system integrating subjective and objective factors was designed based on multidimensional flash flood influencing factors, where the subjective weights... The objective weights are calculated based on the relative importance scores, with each multidimensional flash flood influencing factor based on the coefficient of variation in historical samples. Calculations show that objective weights exist. The calculation formula is: , The number of influencing factors for multidimensional flash flood influencing factors. For this impact factor standard deviation For this impact factor The mean, finally calculated using a specified coefficient. Balancing subjective and objective weights, the dynamic weight calculation formula is as follows: ; By combining the impact factor response with dynamic weights, the coupling function is constructed as follows: in, This is the adjusted critical rainfall. The initial critical rainfall output by the HEC-RAS model. For the aforementioned factors, >0 represents the coupling index. >0 indicates the intensity coefficient of the cross-term factor. For the weights of the corresponding impact factors, For cross-weighting, if the factor value in the coupling function increases, then... Decrease, or vice versa Increase; A power-means type comprehensive risk index is constructed to achieve non-linear hierarchical triggering. The index is constructed as follows: in, This indicates the power parameter of the risk indicator; Set level threshold The following warning rules are specified to trigger the warning level: in, This indicates the level of warning that has been triggered. The critical rainfall level was further adjusted according to different levels. The following empirical rules or functions are given: in, >0 indicates the warning threshold trigger coefficient. These are risk thresholds for different levels.

Images