Rainfall loss parameter verification optimization method based on soil humidity grading

Abstract

The invention provides a rainfall loss parameter verification optimization method based on soil humidity grading, and relates to the technical field of disaster assessment, and the method comprises the steps: obtaining a plurality of soil hydrodynamic parameter units; constructing a dynamic parameter mapping matrix including an initial loss coefficient and an initial loss index; performing disturbance combination to generate a plurality of regional soil composition scenes; after multiple sets of initial loss parameters are called, weighted linear fusion is executed, and multiple fusion loss parameters are output; executing rainstorm peak discharge simulation, and outputting a plurality of peak discharge predicted values; calling a plurality of actual measurement values of the flood peak flow, and calculating a plurality of prediction deviation vectors; executing multi-scene parallel loss parameter iterative correction, and outputting multiple groups of corrected loss parameters; and carrying out parameter directional replacement. The technical problems that most models in the prior art are calculated based on static soil parameters and empirical formulas, real-time monitoring and dynamic changes of the soil humidity are ignored, and the space-time precision and adaptability of disaster early warning are poor are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of disaster assessment, in particular to a rainfall loss parameter verification optimization method based on soil moisture grading. BACKGROUND

[0002] The formation of mountain flood disaster involves complex meteorological, hydrological, geological, social and economic factors. In mountainous areas, due to complex terrain, large changes in rainfall intensity and geological conditions, mountain flood disasters often have high suddenness and severity. Specifically in Beijing and its surrounding areas, due to the mountainous terrain and frequent geological disasters, extreme heavy rainfall often triggers debris flow, landslide, mountain flood and other disasters, posing a major threat to people's life and property safety. In order to effectively respond to mountain flood disasters, accurate disaster monitoring and early warning are crucial, and accurate storm flood peak flow calculation is the key to predicting mountain flood disasters.

[0003] Traditional rainfall-flow models are mostly based on static soil parameters and empirical formulas for calculation, without considering real-time monitoring and dynamic changes of soil moisture. Soil moisture is a dynamic variable affected by factors such as rainfall, evaporation and vegetation. Therefore, the performance of flood peak flow prediction based on fixed parameters in heavy rain events often deviates, especially when soil moisture fluctuates greatly.

[0004] This means that the existing technology cannot effectively dynamically adjust the model parameters according to real-time soil moisture monitoring data, resulting in poor spatial and temporal accuracy and adaptability of the early warning system, especially when soil moisture and rainfall conditions change dramatically, the accuracy of disaster warning will be severely affected.

[0005] It should be noted that the information disclosed in this BACKGROUND section is intended only to increase an understanding of the general context of the present application, and is not to be taken in any way as an acknowledgment or any form of suggestion that this information forms prior art presently known to those of ordinary skill in the art. SUMMARY

[0006] In view of the above defects or improvement needs of the prior art, the present application provides a rainfall loss parameter verification optimization method based on soil moisture grading, which solves the technical problem that the model of the prior art is mostly based on static soil parameters and empirical formulas for calculation, ignoring real-time monitoring and dynamic changes of soil moisture, resulting in poor spatial and temporal accuracy and adaptability of disaster warning.

[0007] The specific technical scheme is as follows:

[0008] The application provides a rainfall loss parameter verification optimization method based on soil humidity grading, which comprises the following steps: combining a plurality of sample soil types and a plurality of pre-divided soil humidity states to obtain a plurality of soil water dynamic parameter units; after matching a plurality of initial loss parameters of the plurality of soil water dynamic parameter units, a dynamic parameter mapping matrix is constructed by establishing a mapping relationship, wherein the initial loss parameters include an initial loss coefficient and an initial loss index; based on a preset soil property correlation threshold, disturbance combination of the plurality of sample soil types and the plurality of soil humidity states is performed to generate a plurality of regional soil composition scenarios; a plurality of initial loss parameters are retrieved from the dynamic parameter mapping matrix by taking the plurality of regional soil composition scenarios as retrieval conditions, and weighted linear fusion is performed based on soil area proportion to output a plurality of fused loss parameters; the plurality of fused loss parameters are loaded into a storm flood peak flow model to perform storm flood peak flow simulation, and a plurality of flood peak flow prediction values are output; a plurality of flood peak flow measured values of the plurality of regional soil composition scenarios are retrieved from historical disaster records, and a plurality of prediction deviation vectors of the plurality of flood peak flow prediction values are calculated; taking the plurality of initial loss parameters as a starting point and minimizing the plurality of prediction deviation vectors as an optimization target, multi-scenario parallel loss parameter iterative correction is performed in a loss parameter feasible region to output a plurality of corrected loss parameters; and the plurality of corrected loss parameters are used for parameter directional replacement in the dynamic parameter mapping matrix.

[0009] In an embodiment, the method further comprises:

[0010] A pre-constructed storm intensity calculation module, a net rain calculation module, a slope confluence module, a channel confluence module and a flood peak flow output layer are provided; the storm flood peak flow model is constructed by unidirectional cascading the storm intensity calculation module, the net rain calculation module, the slope confluence module, the channel confluence module and the flood peak flow output layer.

[0011] In an embodiment, the method further comprises:

[0012] The Nth sample storm parameter, the Nth terrain parameter, the Nth slope flow velocity coefficient and the Nth channel flow velocity coefficient of the Nth regional soil composition scenario are matched, wherein the Nth terrain parameter comprises an Nth slope geometric parameter and an Nth channel geometric parameter;

[0013] The Nth sample storm parameter, the Nth terrain parameter, the Nth slope flow velocity coefficient and the Nth channel flow velocity coefficient are loaded into the storm flood peak flow model to perform:

[0014] S1: input the Nth sample storm parameter into the storm intensity calculation module, and calculate and output the Nth storm intensity; S2: after coupling the Nth group of initial loss parameters and the Nth storm intensity, input into the net rain calculation module, and calculate and output the Nth net rain intensity; S3: after coupling the Nth net rain intensity, the Nth slope surface geometric parameter and the Nth slope surface flow velocity coefficient, input into the slope surface confluence module, and calculate and output the Nth slope surface runoff; S4: after coupling the Nth slope surface runoff, the Nth channel geometric parameter and the Nth channel flow velocity coefficient, input into the channel confluence module, and calculate and output the Nth channel runoff; S5: nonlinear coupling calculation of the Nth slope surface runoff and the Nth channel runoff is performed in the flood peak flow output layer, and the Nth flood peak flow prediction value is output.

[0015] In an embodiment, starting from the multiple groups of initial loss parameters, taking minimizing the multiple prediction deviation vectors as the optimization target, performing multi-scenario parallel loss parameter iterative correction in the loss parameter feasible region, and outputting multiple groups of corrected loss parameters, including:

[0016] decompose the Nth regional soil composition scenario to obtain K sample soil types, K soil area proportion parameters and the Nth soil moisture state; decompose the Nth group of initial loss parameters to obtain K initial loss parameters of the K sample soil types under the Nth soil moisture state; construct K initial optimization spaces according to the physical feasible region boundary of the K sample soil types under the Nth soil moisture state; perform feasible region reduction of the K initial loss parameters in the K initial optimization spaces to obtain K convergent optimization spaces; take minimizing the Nth prediction deviation vector as the optimization target, and perform adaptive step particle swarm optimization in the K convergent optimization spaces, starting from the K initial loss parameters corresponding to K initial loss particles, to output K corrected loss parameters, which constitute the Nth group of corrected loss parameters.

[0017] In an embodiment, taking minimizing the Nth prediction deviation vector as the optimization target, performing adaptive step particle swarm optimization in the K convergent optimization spaces, starting from the K initial loss parameters corresponding to K initial loss particles, to output K corrected loss parameters, which constitute the Nth group of corrected loss parameters, including:

[0018] According to the K soil area proportion parameters, K initial search steps of the K convergent optimization spaces are set; multi-directional random updates of the K initial search steps are performed in the K convergent optimization spaces starting from the K initial loss particles, to obtain K updated loss particles; the K updated loss particles are linearly fused according to the K soil area proportion parameters, to obtain a first fused updated parameter; the first fused updated parameter is loaded to the storm flood peak flow model, storm flood peak flow simulation is performed, and a first updated flow prediction value is output; a first updated deviation vector of the Nth flood peak flow measured value and the first updated flow prediction value is calculated; if a first deviation percentage of the first updated deviation vector and the Nth prediction deviation vector is greater than a preset tolerance threshold, then after the K initial search steps are dynamically adjusted according to the first deviation percentage, adaptive step particle swarm optimization is performed in the K convergent optimization spaces starting from the K updated loss particles, until the K corrected loss parameters meeting the preset tolerance threshold are output.

[0019] In an embodiment, based on a preset soil property association threshold, disturbance combination of the plurality of sample soil types and the plurality of soil moisture states is performed, to generate a plurality of regional soil composition scenarios, including:

[0020] The soil area proportion constraint and the soil type quantity constraint are predefined to obtain the soil property association threshold; non-exhaustive combination of the plurality of sample soil types is performed according to the soil type quantity constraint, to obtain a plurality of initial soil combination scenarios; within the soil area proportion constraint, proportion disturbance of a dominant soil type of the plurality of initial soil combination scenarios is performed, to obtain a plurality of dominant soil combination scenarios; the plurality of dominant soil combination scenarios and the plurality of soil moisture states are combined and enumerated, to output the plurality of regional soil composition scenarios, wherein each regional soil composition scenario is provided with a soil area proportion identifier.

[0021] In an embodiment, further including:

[0022] After receiving the real-time volume water content of the target area monitored by the soil water content sensor, a soil water potential state classification is performed to obtain a real-time soil moisture state grade; a soil area proportion composition of the target area is searched to generate a target soil combination scenario; the real-time soil moisture state grade and the target soil combination scenario are taken as double-dimensional search conditions to retrieve a real-time optimization loss parameter group from the dynamic parameter mapping matrix; the real-time optimization loss parameter group is linearly weighted and fused according to a target soil area proportion of the target soil combination scenario to output a real-time fused loss parameter; real-time storm parameters, target terrain parameters, target slope flow velocity coefficients and target channel flow velocity coefficients are searched according to a region boundary of the target area; and the real-time fused loss parameter, the real-time storm parameters, the target terrain parameters, the target slope flow velocity coefficients and the target channel flow velocity coefficients are dynamically loaded into the storm flood peak flow model to simulate and output a real-time flood peak flow prediction value.

[0023] In an embodiment, the method further comprises:

[0024] After the rainfall event in the target area ends, a real-time flood peak flow measured value is retrieved; if a real-time deviation vector of the real-time flood peak flow prediction value and the real-time flood peak flow measured value exceeds the preset tolerance threshold, a dynamic feedback correction of the dynamic parameter mapping matrix is triggered.

[0025] The embodiment of the application has the following beneficial effects:

[0026] According to the different states of soil moisture and the combination of soil types, the mapping relationship between the soil moisture level and the rainfall loss coefficient is established, and through this mapping relationship, the model parameters can be dynamically adjusted under different humidity conditions, so as to more accurately simulate the storm flood flow; through the establishment and optimization of the dynamic parameter mapping matrix, the real-time updating and adjustment of the parameters can be realized, and the dynamic nature of this matrix enables the model to automatically correct and optimize the prediction of the flood peak flow according to the real-time monitoring of the soil moisture change and the rainfall intensity change; by optimizing the initial loss parameter, minimizing the target of the prediction deviation vector, optimizing the rainfall loss coefficient, and finding the optimal solution in the process of parallel loss parameter correction in multiple scenarios, the prediction accuracy is improved; with the updating of the real-time soil moisture monitoring data, the loss parameter is updated through the dynamic correction mechanism, and the flood peak flow model is always in the optimal state, and this dynamic correction mechanism based on the humidity level is helpful for timely adjusting the model prediction in actual rainfall events; through the disturbance combination of multiple sample soil types and humidity states, multiple regional soil composition scenarios are generated, and the loss parameter optimization is further refined in combination with the soil area proportion and the humidity level, and based on the analysis of these regional soil composition scenarios, the flood peak flow prediction can be more accurate; by combining the soil humidity grading with the loss parameter dynamic correction mechanism, a closed-loop feedback system is formed, which can dynamically adjust the relationship between the soil humidity level and the loss parameter through the prediction deviation value of the flood peak flow after the actual disaster, and further improve the accuracy of the model.

[0027] Of course, implementing any product or method of the present application does not necessarily require achieving all the advantages described above at the same time. BRIEF DESCRIPTION OF DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0029] Figure 1 A flowchart of a rainfall loss parameter verification and optimization method based on soil moisture grading provided by the present application is shown;

[0030] Figure 2 A flowchart of constructing a storm flood peak flow model in a rainfall loss parameter verification and optimization method based on soil moisture grading provided by the present application is shown. DETAILED DESCRIPTION

[0031] In order to facilitate the understanding of the present application, a more full description of the present application will be described below with reference to the accompanying drawings, in which the preferred embodiments of the present application are given, however, the present application can be realized in many different forms and is not limited to the embodiments described herein; on the contrary, these embodiments are provided to make the disclosure of the present application more thorough and comprehensive.

[0032] In addition, the technical features involved in each of the embodiments of the application described below can be combined with each other as long as there is no conflict between them.

[0033] In the description of the present application, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the purpose of facilitating the description of the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application.

[0034] Unless otherwise explicitly stated, throughout the specification and claims, the term "comprise" or its variants such as "contain" or "include" and the like will be understood to include the stated element or component, but not to exclude other elements or components.

[0035] The soil moisture grading-based rainfall loss parameter verification optimization method provided by the present application is used to solve the technical problem that the existing technology is mostly based on static soil parameters and empirical formula for calculation, ignoring real-time monitoring and dynamic changes of soil moisture, resulting in poor spatiotemporal accuracy and adaptability of disaster warning.

[0036] Embodiment one: see Figure 1 The flowchart of the soil moisture grading-based rainfall loss parameter verification optimization method provided by the embodiments of the present application, the method comprises:

[0037] Y100: combining a plurality of sample soil types and a plurality of pre-divided soil moisture states to obtain a plurality of soil hydrodynamic parameter units.

[0038] A plurality of soil types are selected as basic samples, for example, different types of soil such as clay, sandy soil, loam soil, etc. Each type of soil has different hydrodynamic characteristics, including water permeability and water holding capacity of the soil, which affect the movement and runoff of water flow. The soil moisture state can be characterized by soil water potential or water content, and the soil is classified into three categories of high, medium and low moisture, which not only affects the water holding capacity of the soil, but also directly affects the initial loss of rainfall, i.e. the loss of water at the beginning of rainfall. By combining the above soil types and soil moisture states, a plurality of soil hydrodynamic parameter units are obtained, each corresponding to a specific soil type and moisture level, and having a set of related hydrodynamic characteristics.

[0039] Y200: After matching the plurality of initial loss parameters of the plurality of soil hydrodynamic parameter units, a dynamic parameter mapping matrix is constructed by establishing a mapping relationship, wherein the initial loss parameters include an initial loss coefficient and an initial loss index.

[0040] Each soil hydrodynamic parameter unit has a set of initial loss parameters, wherein the initial loss coefficient refers to the proportion of water retained or penetrated on the soil surface during the initial rainfall process, which is affected by factors such as soil moisture, soil type and rainfall intensity; the initial loss index indicates the index of the change of initial loss with rainfall intensity or moisture, which is used to simulate the change rule of initial loss under different rainfall intensities. By matching the initial loss parameters of different soil hydrodynamic parameter units, a set of dynamic mapping relationships is obtained, which binds each soil type and moisture state with the corresponding initial loss parameters and organizes them into a dynamic parameter mapping matrix. The rows of this matrix represent different soil hydrodynamic parameter units, and the columns represent different loss parameters. The matrix is dynamic because the loss parameters are adjusted as the soil moisture state and soil type change.

[0041] Y300: Based on the preset soil property association threshold, the disturbance combination of the plurality of sample soil types and the plurality of levels of soil moisture state is performed to generate a plurality of regional soil composition scenarios.

[0042] The preset soil property correlation threshold is used to determine which soil type combination and humidity state combination is reasonable, for example, certain soil types do not conform to the actual situation under certain humidity, such as sand soil becomes too thick and loses water permeability under very high humidity, therefore, the threshold is set to ensure the reasonableness of the soil type and humidity state. By disturbing the combination of soil type and humidity state, a plurality of possible regional soil composition scenarios are obtained, each scenario including a plurality of soil types and each soil type being assigned a certain area ratio. For a plurality of regional soil composition scenarios, each scenario includes a plurality of soil types-humidity, so that each scenario is a set of loss parameters when matching the loss parameters, and the plurality of soil types in each scenario have an area ratio.

[0043] Y400: After retrieving a plurality of initial loss parameters from the dynamic parameter mapping matrix using the plurality of regional soil composition scenarios as retrieval conditions, performing weighted linear fusion based on the soil area ratio to output a plurality of fused loss parameters.

[0044] The regional soil composition scenario is used as a retrieval condition to obtain the corresponding initial loss parameter from the dynamic parameter mapping matrix. In each regional soil composition scenario, there are multiple soil types, so each soil type has an area ratio, such as 50% of the area in a certain region being sand, 30% of the area being loam, and 20% of the area being clay. For each regional soil composition scenario, the loss parameters (initial loss coefficient and initial loss index) of all soil types are weighted and averaged according to their area ratio, and a new fused loss parameter is output for each regional soil composition scenario through this weighted linear fusion process, which can reflect the initial loss of the entire region.

[0045] Y500: Load the plurality of fused loss parameters into the storm flood peak flow model to perform storm flood peak flow simulation and output a plurality of flood peak flow prediction values.

[0046] The obtained plurality of fused loss parameters will be loaded as input into the storm flood peak flow model, which includes modules such as storm intensity calculation, net rain calculation, slope confluence, and channel confluence, to simulate flood flow after a storm. In the storm flood peak flow model, the fused loss parameters are used for simulation to calculate the flood flow (i.e., flood peak flow) caused by rainfall under different soil composition scenarios. After the model is executed, a plurality of flood peak flow prediction values are obtained for different regional soil composition scenarios, which are simulation results based on different soil humidity states, soil type combinations, and rainfall conditions.

[0047] Y600: Retrieve a plurality of flood peak flow measured values for the plurality of regional soil composition scenarios from historical disaster records, and calculate a plurality of prediction deviation vectors for the plurality of flood peak flow prediction values.

[0048] Through historical disaster records, a plurality of measured values of flood peak flow corresponding to a plurality of simulated regional soil composition scenarios are obtained, the measured values are from actual rainstorm and flood events, and reflect flood flow of the region in historical disasters. The prediction deviation vector refers to the difference between the predicted flood peak flow of the model and the actually measured flood peak flow. For each regional soil composition scenario, the simulated flood peak flow is compared with the measured value, and the prediction deviation vector is calculated. The deviation vector can reflect the prediction error of the model under different scenarios. The smaller the deviation, the more accurate the prediction of the model. The larger the deviation, the greater the prediction error of the model.

[0049] Y700: Taking the plurality of sets of initial loss parameters as a starting point, taking minimizing the plurality of prediction deviation vectors as an optimization target, performing multi-scenario parallel loss parameter iterative correction in a loss parameter feasible region, and outputting a plurality of sets of corrected loss parameters.

[0050] The plurality of sets of initial loss parameters are taken as a starting point of an optimization algorithm, and subsequent adjustment and optimization are performed. The optimization target is to minimize the plurality of prediction deviation vectors, so that the prediction result of the model is closer to the measured value, thereby improving the accuracy of the model. The optimization target can be achieved by minimizing the sum of squares of errors. When optimization is performed, all parameters are adjusted in the loss parameter feasible region, which is limited by physical and engineering actual conditions, such as the upper and lower limits of the loss coefficient, the range of the initial loss index, etc. The feasible region ensures that the parameter correction in the optimization process will not exceed the actual possible range. In order to improve the optimization efficiency, the optimization process adopts a multi-scenario parallel mode, that is, the iterative correction of the loss parameters is performed in parallel under a plurality of soil composition scenarios. The correction process of each scenario is independent of each other, so the optimization process can be accelerated in parallel computing. Through parallel computing, the initial loss parameters are gradually adjusted, so that the prediction error of the flood peak flow under each scenario is continuously reduced, and the optimization target is finally reached. After a plurality of iterations, a plurality of sets of corrected loss parameters are outputted. These parameters are more accurate than the initial values and can better fit the actual flood peak flow measured data.

[0051] Y800: Using the plurality of sets of corrected loss parameters to perform parameter directional replacement in the dynamic parameter mapping matrix.

[0052] The obtained multiple sets of corrected loss parameters are taken as new inputs, replace the original initial loss parameters, and are directionally replaced in the dynamic parameter mapping matrix. Directional replacement means that, for each combination of soil type and humidity state, the original parameters are replaced by the newly obtained corrected loss parameters. Through this replacement, all parameters in the matrix can be updated according to the optimization result, thereby improving the prediction accuracy of the model. The replaced corrected loss parameters affect the simulation of the storm flood peak flow, and especially in the actual storm event, the parameters can be dynamically updated according to the real-time monitoring data, so that the flood peak flow prediction is more accurate.

[0053] In an implementation manner, referring to Figure 2 Further comprising:

[0054] Y510: a pre-constructed storm intensity calculation module, a net rain calculation module, a slope confluence module, a channel confluence module, and a flood peak flow output layer; Y520: the construction of the storm flood peak flow model is completed by unidirectional cascading the storm intensity calculation module, the net rain calculation module, the slope confluence module, the channel confluence module, and the flood peak flow output layer.

[0055] The storm intensity calculation module is used to calculate the storm intensity in a specific time period, that is, the precipitation in a unit time. The storm intensity is closely related to factors such as soil type and humidity state, because different soils respond differently to precipitation under different humidity conditions. The module is generally based on precipitation data, meteorological data, etc. to calculate, and the storm intensity calculation method includes statistical method, empirical formula, etc. Net rain refers to the part of precipitation that effectively causes runoff, and the water absorbed or evaporated by soil is subtracted. Net rain calculation is based on factors such as initial moisture of soil and permeability of soil. The net rain calculation module calculates the net rainfall based on input data such as storm intensity and soil humidity, which is used for subsequent runoff calculation. The slope confluence module calculates how water flow on the slope converges and is converted into runoff. Factors such as slope flow rate, slope, and soil properties will affect the water flow convergence process. This module uses empirical formula or physical model to estimate the slope confluence flow. The channel confluence module calculates the movement of water flow in the channel. Key factors affecting the flood peak flow include channel morphology, channel flow rate, slope, etc. This module further calculates the convergence of water flow after passing through the channel based on the slope confluence water flow, which ultimately affects the flood peak flow. The flood peak flow output layer collects water flow data at different stages through the calculation of each module. This layer is responsible for outputting the maximum flow prediction value of flood, that is, the flood peak flow.

[0056] These modules are connected in a unidirectional cascade manner, that is, the output data of each module is the input of the next module, and the output of each module affects the calculation of the next module. Finally, the output is the flood peak flow. Through the connection of the above modules, a complete storm flood peak flow model is constructed to simulate the flood flow after the storm and output the flood peak flow of the region.

[0057] In an implementation, further comprising:

[0058] Y530: match the Nth sample storm parameter, the Nth terrain parameter, the Nth overland flow velocity coefficient and the Nth channel flow velocity coefficient with the Nth regional soil composition scenario, wherein the Nth terrain parameter includes Nth overland geometry parameter and Nth channel geometry parameter; Y540: load the Nth sample storm parameter, the Nth terrain parameter, the Nth overland flow velocity coefficient and the Nth channel flow velocity coefficient into the storm flood peak flow model, and perform: S1: input the Nth sample storm parameter into the storm intensity calculation module, and calculate and output the Nth storm intensity; S2: after coupling the Nth set of initial loss parameters and the Nth storm intensity, input into the net rain calculation module, and calculate and output the Nth net rain intensity; S3: after coupling the Nth net rain intensity, the Nth overland geometry parameter and the Nth overland flow velocity coefficient, input into the overland flow concentration module, and calculate and output the Nth overland runoff; S4: after coupling the Nth overland runoff, the Nth channel geometry parameter and the Nth channel flow velocity coefficient, input into the channel flow concentration module, and calculate and output the Nth channel runoff; S5: perform nonlinear coupling calculation of the Nth overland runoff and the Nth channel runoff at the flood peak flow output layer, and output the Nth flood peak flow prediction value.

[0059] The Nth regional soil composition scenario includes the soil types and moisture states within the Nth region, as well as the area proportions of these soil types within the region. The Nth sample storm parameter includes the intensity, duration, precipitation, etc. of the storm, which is used for flood simulation. The storm parameter is set based on historical meteorological data or real-time meteorological prediction data, and can be statistical data of a storm event or storm characteristics under certain simulation conditions. The Nth terrain parameter affects water flow concentration and flow, including: overland geometry parameters, which describe the shape, slope, etc. of the slope, and affect the speed and direction of runoff; channel geometry parameters, which describe the characteristics of water flow in the channel, including the width, depth, slope, etc. of the channel, and determine the flow velocity and flow of water flow. The Nth overland flow velocity coefficient is used to describe the speed of water flow on the slope, which is related to factors such as slope, soil type, moisture, etc. The overland flow velocity coefficient is used to calculate the runoff on the slope, and affects the distribution and convergence of water flow after the storm. The Nth channel flow velocity coefficient describes the flow velocity of water flow in the channel, which depends on factors such as the geometric shape of the channel, soil type and soil moisture state, etc. The channel flow velocity coefficient plays a key role in calculating the transmission of water flow from the slope to the channel.

[0060] The storm parameters, terrain parameters, overland flow velocity coefficients and channel flow velocity coefficients of the Nth region are loaded into the storm flood peak flow model to start the calculation of the storm flood peak flow. This process includes the cooperative work of multiple calculation modules, from storm intensity calculation to final flood peak flow output.

[0061] Specifically, the Nth sample storm parameter is input into the storm intensity calculation module, which calculates the storm intensity under the storm event in the area. The storm intensity refers to the intensity of precipitation per unit time, expressed in millimeters per hour. The module estimates the storm intensity through meteorological data of the storm event. The greater the storm intensity, the greater the possible flood flow it may cause.

[0062] The net rain calculation module is used to calculate the amount of water in precipitation that is actually converted into runoff. Since the soil can absorb part of the precipitation, only the remaining water can form runoff. The initial loss parameter reflects the soil's ability to absorb water at the beginning of the storm. For example, when the soil moisture is high, the initial loss coefficient is large, resulting in more precipitation being absorbed by the soil, and the remaining water is used to generate runoff. By coupling the storm intensity with the Nth set of initial loss parameters, the Nth net rain intensity is calculated.

[0063] The net rain intensity, slope geometry parameters, and slope flow velocity coefficient are combined and input into the slope confluence module for calculation. The slope confluence module calculates how water flows converge along the slope based on the effective precipitation on the slope and the geometric characteristics of the slope, and ultimately forms the slope runoff. The calculation of slope confluence takes into account factors such as slope, flow velocity, and soil type. Higher slope or larger flow velocity coefficient will result in faster convergence and flow of water. The output of this module is the Nth slope runoff, which is the water flow that converges from the slope to the next stage and affects the calculation of subsequent channel confluence.

[0064] The slope runoff, channel geometry parameters, and channel flow velocity coefficient are combined and input into the channel confluence module. This module calculates how the water flow from the slope into the channel further converges and propagates along the channel. The river flow velocity formula and flow formula are used for water flow calculation. Here, the shape of the channel and the flow velocity coefficient directly affect the final flow of water. The output of this module is the Nth channel runoff, which is the final water flow in the channel. This quantity represents the propagation of water flow in the channel and affects the final flood peak flow.

[0065] The slope runoff and channel runoff are calculated through nonlinear coupling. This takes into account the complex interaction between slope and channel water flow, i.e., the mutual influence between slope and channel water flow, as well as the nonlinear relationship between different flow velocities and flows, such as superposition and mutual interference of slope runoff and channel water flow. This influence cannot be represented by a simple linear relationship, so nonlinear coupling is required. Through nonlinear coupling calculation, the Nth flood peak flow prediction value is obtained. This value represents the maximum flood flow generated under specific storm and soil conditions and is used as a key indicator for flood warning and disaster assessment in hydrological models.

[0066] These steps constitute the core calculation part of the storm flood peak flow model, focusing on simulating the changes of water flow in different stages step by step to ultimately predict the flood peak flow. It should be understood that throughout the process, the physical structure and calculation process of the model are fixed, and the only change is the initial loss parameter, including the initial loss coefficient and the loss exponent. The change of these parameters will affect the input of each module, thereby affecting the simulation of water flow and the final prediction of the flood peak flow.

[0067] In one implementation, starting from the plurality of sets of initial loss parameters, the optimization objective is to minimize the plurality of prediction deviation vectors, and the multi-scenario parallel loss parameter iterative correction is performed in the loss parameter feasible region. Output multiple sets of corrected loss parameters, including:

[0068] Y710: Decompose the Nth regional soil composition scenario to obtain K sample soil types, K soil area proportion parameters, and the Nth soil moisture state; Y720: Decompose the Nth set of initial loss parameters to obtain K initial loss parameters for the K sample soil types under the Nth soil moisture state; Y730: Construct K initial optimization spaces according to the physical feasible region boundaries of the K sample soil types under the Nth soil moisture state; Y740: Perform feasible region reduction of the K initial loss parameters in the K initial optimization spaces to obtain K convergent optimization spaces; Y750: Perform adaptive step particle swarm optimization in the K convergent optimization spaces, starting from the K initial loss parameters corresponding to the K initial loss particles, with the optimization objective of minimizing the Nth prediction deviation vector, and output K corrected loss parameters to constitute the Nth set of corrected loss parameters.

[0069] According to the soil composition in the region, K sample soil types are extracted, which affect the rainfall absorption capacity and water flow characteristics; the proportion of each soil type in the region, i.e., the soil area proportion parameter, affects the water absorption capacity of the soil type and its response to rainfall. By quantifying the area proportion of each soil type in the region, the soil moisture characteristics of the region can be more accurately represented; the soil moisture state of the region determines the initial water content of the soil, which in turn affects the initial loss parameter. Under different moisture states, the water absorption capacity and water flow response of the soil will be different.

[0070] The Nth set of initial loss parameters is decomposed into each soil type and moisture state, and the initial loss parameters are closely related to the soil type and moisture state, for example, for dry soil, the initial loss coefficient is higher because more precipitation will be absorbed by the soil; while the wet soil has a lower initial loss coefficient. The initial loss coefficient and the loss exponent reflect the soil's ability to absorb precipitation and the rate of loss over time. For each soil type, the initial loss parameters of each soil type are calculated under the Nth soil moisture state, that is, for each sample soil type, the initial loss coefficient and the loss exponent of the soil are obtained separately under a given soil moisture state, reflecting the water retention capacity and runoff characteristics of different soil types under specific humidity conditions.

[0071] The initial optimization space is used to define the search range of the initial loss parameters, and the initial loss coefficient and the loss exponent of each soil type under a specific moisture state have certain physical boundaries, which do not conform to the actual situation if exceeded, so reasonable physical feasible domain boundaries, i.e. upper and lower limits of the loss coefficient and the loss exponent, are set for each soil type under a specific moisture state. The optimization space of each soil type is two-dimensional, with the X-axis representing the initial loss coefficient and the Y-axis representing the loss exponent, and each point in the space represents a possible combination of initial loss parameters.

[0072] Due to the actual physical meaning of the loss parameters and the requirements of the model, some areas in the initial optimization space may not meet the actual conditions or cannot provide effective flood peak flow prediction, so these spaces need to be limited, i.e. through physical constraints, empirical data or historical data, these spaces are further narrowed down to a feasible domain. The process of feasible domain reduction is actually to set reasonable upper and lower limits for each optimization space according to experience, soil physical properties, historical disaster records, etc., and to eliminate unreasonable areas. The K initial optimization spaces correspond to different soil types and moisture states, and after limiting, K convergent optimization spaces are obtained, each of which contains a narrower search range, allowing the optimization algorithm to search for the best loss parameters within this range.

[0073] The Nth prediction deviation vector represents the difference between the model-predicted flood peak flow and the actual measured flood peak flow, and the minimization of the Nth prediction deviation vector is the optimization objective, that is, by adjusting the initial loss parameters through the optimization process, the error between the prediction result and the actual result is reduced. Particle swarm optimization is a swarm intelligence optimization algorithm, in which each particle represents a possible solution, that is, a set of initial loss parameters, and the best solution is found by searching in the optimization space. The step size in particle swarm optimization refers to the step size of the particle moving in the search space, and the adaptive step size means that the particle adjusts the step size according to the current search situation, and the step size is larger in the initial search stage, and the step size is gradually reduced to refine the search process. Through adaptive step size, the particle can search the feasible region in a large range, and can accurately adjust the parameters when approaching the optimal solution, avoiding excessive oscillation or missing the optimal solution. In each convergent optimization space, the particle swarm optimization algorithm is started with the initial loss parameters as the starting point, each initial loss parameter corresponds to a particle, and the particle swarm adjusts the position of each particle according to the objective function (i.e., minimizing the prediction deviation vector). After multiple iterations of the particle swarm optimization algorithm, each final particle provides a modified loss parameter, that is, a modified initial loss coefficient and a loss index. These modified loss parameters are the optimal solutions for K soil types and humidity states, corresponding to different initial loss parameter combinations, which can make the model prediction results closest to the measured flood peak flow data. The K modified loss parameters constitute the Nth set of modified loss parameters.

[0074] In one implementation, the adaptive step particle swarm optimization is performed in the K convergent optimization spaces with the K initial loss parameters corresponding to K initial loss particles as the starting point, with the optimization objective of minimizing the Nth prediction deviation vector, and the K modified loss parameters are output to constitute the Nth set of modified loss parameters, including:

[0075] Y751: setting K initial search steps of the K convergent optimization spaces according to the K soil area proportion parameters; Y752: performing multi-directional random updates of the K initial search steps in the K convergent optimization spaces starting from the K initial loss particles, to obtain K updated loss particles; Y753: obtaining a first fused updated parameter by linearly fusing the K updated loss particles according to the K soil area proportion parameters; Y754: loading the first fused updated parameter to the storm flood peak flow model to perform storm flood peak flow simulation, and outputting a first updated flow prediction value; Y755: calculating a first updated deviation vector of the Nth flood peak flow measured value and the first updated flow prediction value; Y756: if a first deviation percentage of the first updated deviation vector and the Nth prediction deviation vector is greater than a preset tolerance threshold, then performing adaptive step particle swarm optimization in the K convergent optimization spaces starting from the K updated loss particles after dynamically adjusting the K initial search steps according to the first deviation percentage, until the K corrected loss parameters meeting the preset tolerance threshold are output.

[0076] In particle swarm optimization, each particle (i.e., a loss parameter combination) has its own search step, which determines the speed and range of the particle's movement in the optimization space. The initial search step for each convergent optimization space is set according to the area proportion of each soil type. Specifically, the area proportion of a soil type reflects the influence weight of different soil types in the total area. Soil types with larger area proportions play a more important role in the optimization process, so they should be given larger search steps to quickly search for the optimal solution. By setting appropriate initial search steps for each soil type according to the area proportion, the optimization algorithm can flexibly adjust the parameter search speed according to the influence weight of different soil types. This method ensures that the optimization process is both fast and accurate, while avoiding excessive adjustment of soil types with less influence.

[0077] In particle swarm optimization, particles explore the optimal solution by randomly updating their positions in the search space. In this stage, the update of a particle is influenced not only by its own experience (individual best position) and group experience (global best position), but also by the set search step. For each convergent optimization space, the particle performs multi-directional random updates within that space starting from the initial loss parameter. Specifically, the particle randomly adjusts its position (i.e., loss parameter) in multiple directions to explore possible optimization solutions. After this multi-directional random update, the position of each particle (i.e., the loss parameter combination) changes, resulting in K updated loss particles. These updated loss particles represent the current loss parameter combination in the optimization process.

[0078] For each soil type, a weighted average of its corresponding loss coefficient and loss index is calculated, specifically, the soil type with a larger proportion of soil area occupies a larger weight in the fusion, and the degree of influence on the fusion result is also greater. Through this weighted fusion, the loss parameters of each soil type are adjusted according to their importance in the region, ensuring that the optimization result is more in line with the actual soil distribution. After weighted fusion, the first fused update parameter is obtained, which represents the optimal loss parameter combination of the region.

[0079] The first fused update parameter is input into the storm flood peak flow model for simulation. The storm flood peak flow model calculates the flood peak flow in the region according to the input rainfall intensity, soil properties, topographic parameters, etc., and outputs the first updated flow prediction value, which is the flood peak flow predicted by the model based on the current optimization parameters.

[0080] According to the previous calculation method, the first update deviation vector of the Nth flood peak flow measured value and the first update flow prediction value is calculated to evaluate the error between the predicted flood peak flow output by the current model after the first optimization and the actual situation.

[0081] The first deviation percentage represents the error percentage between the first update deviation vector and the Nth prediction deviation vector. This value represents the degree of deviation after the first optimization, and whether the previous prediction error has been significantly improved.

[0082] The preset tolerance threshold is a maximum error range set according to the model accuracy requirement. If the calculated first deviation percentage is greater than the preset tolerance threshold, it means that the difference between the current optimization result and the actual flood peak flow is still large, and the optimization result does not meet the accuracy requirement. Therefore, the initial search step size needs to be adjusted to improve the optimization result. By adjusting the step size, the optimization process can be accelerated, and the convergence near the optimal solution can be avoided. If the error is large, the step size can be increased to ensure a more extensive search process and explore more potential solutions. If the error is small, the step size can be reduced to improve the search accuracy.

[0083] After adjusting the step size, the optimization process continues, and the particle swarm continues to iterate in the optimization space with the updated search step size to find the optimal loss parameter that minimizes the deviation. The optimization process continues until the model prediction flow deviation is within the preset tolerance threshold, ensuring that the prediction result meets the accuracy requirement. After iterative optimization, K corrected loss parameters, i.e., the optimal initial loss coefficient and loss index, are finally output.

[0084] In one implementation, based on a preset soil property association threshold, a disturbance combination of the plurality of sample soil types and the plurality of soil moisture states is performed to generate a plurality of regional soil composition scenarios, including:

[0085] Y310: obtaining the soil attribute correlation threshold value according to the predefined soil area proportion constraint and the soil type number constraint; Y320: performing non-exhaustive combination of the multiple sample soil types according to the soil type number constraint to obtain multiple initial soil combination scenarios; Y330: performing dominant soil type proportion perturbation of the multiple initial soil combination scenarios within the soil area proportion constraint to obtain multiple dominant soil combination scenarios; Y340: combining enumeration of the multiple dominant soil combination scenarios and multiple levels of soil moisture states to output the multiple regional soil composition scenarios, wherein each regional soil composition scenario is provided with a soil area proportion identifier.

[0086] The soil area proportion constraint refers to the proportion range of the area occupied by each soil type in the region. This constraint ensures that the distribution of each soil type conforms to the actual geographical environment distribution and does not have excessively high or low proportions of a certain soil type. The soil type number constraint limits the number of soil types to ensure that the soil type combination of each region is not excessive or insufficient, ensuring a certain rationality. By setting these area proportion constraints and soil type number constraints, a soil attribute correlation threshold value is obtained to ensure that the generated soil combination scenarios conform to the actual soil distribution conditions.

[0087] Non-exhaustive combination is not simply listing all possible combinations. The exhaustive method generates all possible combinations, but results in a large number of combinations, and some combinations do not conform to the actual soil distribution rules in practical applications. Therefore, a non-exhaustive combination method is selected to reduce the number of combinations and ensure their rationality. For example, if the soil type number constraint is 3 soil types, different combinations are selected from all soil types to ensure that each combination conforms to the actual situation and avoids producing duplicate or unreasonable combinations. Through the non-exhaustive combination method, multiple initial soil combination scenarios are generated that conform to the soil type number constraint. These combination scenarios provide different soil type distribution scenarios for subsequent optimization, reflecting different changes in regional soil types and distribution.

[0088] For each initial soil combination scenario, perturbation is performed according to the soil area proportion constraint, i.e., adjusting the proportion of the dominant soil type in each scenario. The perturbation is to introduce certain changes so that each soil combination scenario is closer to the actual soil distribution situation. For example, if the initial scenario has a soil type A proportion of 70% and types B and C each have a proportion of 15%, after perturbation, the proportion of A is adjusted to 60% or 80%, and the proportions of B and C are adjusted accordingly to ensure that the proportions of each soil type still conform to the constraint conditions. Through perturbation, multiple dominant soil combination scenarios are obtained, which have more changes and diversity and can better reflect the actual fluctuations of soil types and area proportions.

[0089] The multiple dominant soil composition scenarios are combined with multiple soil moisture states to form more complex regional soil composition scenarios, each of which is marked with a soil area proportion, i.e., the relative area proportion of each soil type in the region, which helps to weight the effects of different soil types in subsequent modeling.

[0090] In an implementation manner, the method further comprises:

[0091] Y510: After receiving the real-time volume water content of the target region monitored by the soil water content sensor, performing soil water potential state classification to obtain a real-time soil moisture state grade; Y520: Retrieving the soil area proportion composition of the target region to generate a target soil combination scenario; Y530: Taking the real-time soil moisture state grade and the target soil combination scenario as a two-dimensional retrieval condition to retrieve a real-time optimization loss parameter group from the dynamic parameter mapping matrix; Y540: Linearly weighting and fusing the real-time optimization loss parameter group according to the target soil area proportion of the target soil combination scenario to output a real-time fused loss parameter; Y550: Retrieving real-time storm parameters, target terrain parameters, target slope flow velocity coefficients and target channel flow velocity coefficients according to the regional boundary of the target region; Y560: Dynamically loading the real-time fused loss parameter, real-time storm parameter, target terrain parameter, target slope flow velocity coefficient and target channel flow velocity coefficient into the storm flood peak flow model to simulate and output a real-time flood peak flow prediction value.

[0092] The soil water content sensor is used to monitor the soil moisture in the target region in real time, and the real-time volume water content represents the amount of water in a unit volume of soil, which is expressed in percentage or volume ratio. According to the received real-time volume water content, soil water potential state classification is performed to convert soil moisture into a specific moisture state grade. Soil water potential is a physical quantity that describes the water absorption capacity and water state of soil. By calculating the relationship between soil water content and water potential, the soil moisture state is divided. For example, high humidity indicates that the soil has a high water content and is close to saturation; medium humidity indicates that the soil moisture is moderate, which is suitable for crop growth and hydrological processes; and low humidity indicates that the soil is relatively dry, which affects hydrological cycle and rainfall absorption.

[0093] The soil area proportion composition refers to the proportion of each soil type in the target region, which reflects the area share of different soil types in the region. The target soil combination scenario is generated based on the soil type area proportion composition of the target region and the soil moisture state.

[0094] The real-time soil moisture state level and the target soil combination scenario are input into the dynamic parameter mapping matrix as double-dimensional retrieval conditions, and the corresponding real-time optimization loss parameter group is called. The real-time optimization loss parameter group refers to the optimization parameters for the target soil combination scenario under the actual soil moisture state. These parameters help to more accurately predict the storm flood flow

[0095] According to the area proportion of each soil type, the loss parameters corresponding to each soil type are weighted through linear weighted fusion. The purpose is to weight and average the loss parameters according to the actual distribution of the soil types in the region, so as to obtain the real-time fused loss parameters.

[0096] The regional boundary defines the spatial range of the storm simulation. Through the regional boundary, other environmental parameters in the region are retrieved. The real-time storm parameters refer to the rainfall intensity, rainfall duration and other key parameters of the current rainfall event. These parameters come from meteorological monitoring or prediction systems and reflect the actual storm situation in the target region. The target terrain parameters include the slope, slope direction and terrain relief of the region. These factors directly affect the water flow distribution and runoff speed after the storm. For example, the region with steeper slope will have higher runoff speed. The target slope flow velocity coefficient and the target channel flow velocity coefficient are parameters describing the movement speed of water flow on the slope and in the channel. These coefficients reflect the influence of soil and surface permeability, roughness and other factors on the water flow speed.

[0097] All the parameters obtained as described above, including the real-time fused loss parameters, the real-time storm parameters, the target terrain parameters, the target slope flow velocity coefficient and the target channel flow velocity coefficient, are dynamically loaded into the storm flood flow model. This means that the model automatically updates its input data every time a storm occurs, ensuring real-time and accuracy. After loading all the necessary parameters, the storm flood flow model simulates the propagation process of water flow, evaluates the changes of runoff, confluence and flood flow after the storm, and outputs the real-time flood flow prediction value, i.e. the predicted maximum flood flow under the current soil, weather and terrain conditions.

[0098] In one implementation, it further includes:

[0099] Y910: After the target region ends the rainfall event, the real-time flood flow measured value is retrieved; Y920: If the real-time deviation vector of the real-time flood flow prediction value and the real-time flood flow measured value exceeds the preset tolerance threshold, the dynamic feedback correction of the dynamic parameter mapping matrix is triggered.

[0100] The end of the rainfall event can be determined by meteorological monitoring data, or according to the rainfall intensity reducing to a certain preset threshold as an end mark, after the rainfall event ends in the target area, the real-time flood peak flow measured value is called, which is the actual measured flood peak flow after the rainfall event ends, and this value is obtained through the flow monitoring station, river or channel hydrological monitoring facility.

[0101] The real-time deviation vector refers to the difference between the predicted value of the storm flood peak flow model and the actual observation value, if the real-time deviation vector exceeds the preset tolerance threshold, it means that there is a large deviation between the prediction result of the model and the actual disaster, and the model needs to be adjusted, at this time, the dynamic feedback correction is triggered, the model parameters are corrected by using the actual observation data, so as to improve the accuracy in future prediction, specifically, the model adjusts the related loss parameters in the dynamic parameter mapping matrix according to the difference between the measured flood peak flow and the predicted flood peak flow, so as to make it more consistent with the actual hydrological process. After dynamic feedback correction, a new set of loss parameters is obtained, which improves the accuracy of subsequent flood peak flow prediction. This feedback correction process is a continuous iteration process, after each rainfall event ends, the model is adjusted according to the deviation between the measured data and the predicted data, to ensure that the prediction ability of the model is continuously improved in the long-term operation.

[0102] The above is only a preferred embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be limited by the protection scope of the claims.

[0103] The foregoing description of specific exemplary embodiments of the application is intended to be illustrative only and is not intended to limit the application to the precise forms described. Many modifications and variations are possible in light of the above teachings without departing from the spirit or essential characteristics of the application. The exemplary embodiments were chosen and described in order to explain the principles of the application and its practical application and to allow others skilled in the art to understand the application for various exemplary embodiments with various modifications being made to the application. The scope of the application is intended to be limited only by the claims and their equivalents.

Claims

1. A method for verifying and optimizing rainfall loss parameters based on soil moisture classification, characterized in that, The method includes: By combining and enumerating multiple sample soil types and pre-divided multi-level soil moisture states, multiple soil hydrodynamic parameter units are obtained. After matching multiple initial loss parameters of the multiple soil hydrodynamic parameter units, a dynamic parameter mapping matrix is ​​constructed by establishing a mapping relationship, wherein the initial loss parameters include initial loss coefficients and initial loss exponents; Based on a preset soil attribute association threshold, the perturbation combination of the various sample soil types and multi-level soil moisture states is performed to generate multiple regional soil composition scenarios. Using the soil composition scenarios of the multiple regions as search conditions, after retrieving multiple sets of initial loss parameters from the dynamic parameter mapping matrix, weighted linear fusion is performed based on the soil area ratio to output multiple fusion loss parameters. The multiple fusion loss parameters are loaded into the rainstorm peak flow model, rainstorm peak flow simulation is performed, and multiple peak flow prediction values ​​are output. Multiple measured peak flow values ​​for soil composition scenarios in the multiple regions are retrieved from historical disaster records, and multiple prediction deviation vectors for the multiple predicted peak flow values ​​are calculated. Starting with the multiple sets of initial loss parameters and minimizing the multiple prediction deviation vectors as the optimization objective, multi-scenario parallel iterative correction of loss parameters is performed in the feasible region of loss parameters, and multiple sets of corrected loss parameters are output. The multiple sets of corrected loss parameters are used to perform parameter-oriented replacement on the dynamic parameter mapping matrix.

2. The method for verifying and optimizing rainfall loss parameters based on soil moisture classification as described in claim 1, characterized in that, Also includes: The system includes pre-built modules for calculating rainfall intensity, net rainfall, slope runoff, gully runoff, and peak flow output. The rainstorm peak flow model is constructed by cascading the rainstorm intensity calculation module, net rainfall calculation module, slope runoff module, gully runoff module, and peak flow output layer in a one-way manner.

3. The method for verifying and optimizing rainfall loss parameters based on soil moisture classification as described in claim 2, characterized in that, Also includes: The Nth sample rainfall parameters, Nth topographic parameters, Nth slope velocity coefficient, and Nth gully velocity coefficient are matched to the soil composition scenario of the Nth region, wherein the Nth topographic parameters include the Nth slope geometric parameters and the Nth gully geometric parameters; The Nth sample rainfall parameters, Nth topographic parameters, Nth slope velocity coefficient, and Nth gully velocity coefficient are loaded into the rainfall peak discharge model, and the following steps are executed: S1: Input the Nth sample rainstorm parameters into the rainstorm intensity calculation module, and calculate and output the Nth rainstorm intensity; S2: After coupling the Nth set of initial loss parameters and the Nth rainfall intensity, input them into the net rainfall calculation module to calculate and output the Nth net rainfall intensity; S3: After coupling the Nth net rainfall intensity, the Nth slope geometric parameters and the Nth slope velocity coefficient, input them into the slope runoff module to calculate and output the Nth slope runoff. S4: After coupling the Nth slope runoff, the Nth gully geometric parameters and the Nth gully velocity coefficient, input them into the gully confluence module to calculate and output the Nth gully runoff; S5: Perform nonlinear coupling calculation of the Nth slope runoff and the Nth gully runoff at the peak flow output layer, and output the predicted value of the Nth peak flow.

4. The method for verifying and optimizing rainfall loss parameters based on soil moisture classification as described in claim 3, characterized in that, Starting with the multiple sets of initial loss parameters and aiming to minimize the multiple prediction deviation vectors, multi-scenario parallel iterative correction of loss parameters is performed within the feasible region of the loss parameters, outputting multiple sets of corrected loss parameters, including: By decomposing the soil composition scenario of the Nth region, we obtain K sample soil types, K soil area ratio parameters, and the soil moisture status of the Nth region. Decompose the Nth group of initial loss parameters to obtain K initial loss parameters for the K soil types under the Nth soil moisture state; Based on the physical feasible domain boundaries of the K types of soil samples under the Nth soil moisture state, K initial optimization spaces are constructed. By narrowing the feasible region of the K initial loss parameters within the K initial optimization spaces, K convergent optimization spaces are obtained. With minimizing the Nth prediction deviation vector as the optimization objective, adaptive step-size particle swarm optimization is performed in the K convergent optimization spaces, starting with the K initial loss particles corresponding to the K initial loss parameters, and outputting K corrected loss parameters to form the Nth set of corrected loss parameters.

5. The method for verifying and optimizing rainfall loss parameters based on soil moisture classification as described in claim 4, characterized in that, With minimizing the Nth prediction deviation vector as the optimization objective, adaptive step-size particle swarm optimization is performed in the K convergent optimization spaces, starting with the K initial loss particles corresponding to the K initial loss parameters, and outputting K corrected loss parameters, which constitute the Nth set of corrected loss parameters, including: Based on the K soil area ratio parameters, set K initial search step sizes for the K convergence optimization spaces; Starting with the K initial loss particles, perform multi-directional random updates of the K initial search step sizes in the K convergent optimization spaces to obtain K updated loss particles; The first fusion update parameter is obtained by weighted linearly fusing the K update loss particles according to the K soil area ratio parameters; The first fusion update parameter is loaded into the rainstorm peak flow model, rainstorm peak flow simulation is performed, and the first updated flow prediction value is output. Calculate the first update deviation vector between the measured value of the Nth flood peak flow and the first updated predicted flow value; If the percentage of the first deviation between the first updated deviation vector and the Nth predicted deviation vector is greater than a preset tolerance threshold, then the K initial search steps are dynamically adjusted according to the first deviation percentage, and the K updated loss particles are used as the update starting point for adaptive step-size particle swarm optimization in the K convergence optimization space until the K corrected loss parameters that satisfy the preset tolerance threshold are output.

6. The method for verifying and optimizing rainfall loss parameters based on soil moisture classification as described in claim 1, characterized in that, Based on a preset soil attribute association threshold, perturbation combinations are performed on the various sample soil types and multi-level soil moisture states to generate multiple regional soil composition scenarios, including: By predefining soil area ratio constraints and soil type quantity constraints, the soil attribute association threshold is obtained; Based on the soil type quantity constraint, a non-exhaustive combination of the various sample soil types is performed to obtain multiple initial soil combination scenarios; Within the soil area ratio constraint, the dominant soil type ratio perturbation of the multiple initial soil combination scenarios is performed to obtain multiple dominant soil combination scenarios. The system combines and enumerates the multiple dominant soil composition scenarios and multi-level soil moisture states to output the multiple regional soil composition scenarios, wherein each regional soil composition scenario is labeled with a soil area ratio.

7. The method for verifying and optimizing rainfall loss parameters based on soil moisture classification as described in claim 5, characterized in that, Also includes: After receiving the real-time volumetric water content of the target area monitored and transmitted by the soil moisture sensor, the soil water potential status classification is performed to obtain the real-time soil moisture status level. The soil area ratio of the target area is retrieved to generate a target soil combination scenario. The real-time soil moisture state level and the target soil combination scenario are used as two-dimensional search conditions to retrieve the real-time optimized loss parameter group from the dynamic parameter mapping matrix. Based on the target soil area ratio of the target soil combination scenario, the real-time optimized loss parameter group is linearly weighted and fused to output the real-time fused loss parameter. Based on the regional boundary of the target area, real-time rainstorm parameters, target terrain parameters, target slope velocity coefficient, and target gully velocity coefficient are retrieved. The real-time fusion loss parameters, real-time rainstorm parameters, target terrain parameters, target slope velocity coefficient, and target gully velocity coefficient are dynamically loaded into the rainstorm peak flow model to simulate and output the real-time peak flow prediction value.

8. The method for verifying and optimizing rainfall loss parameters based on soil moisture classification as described in claim 7, characterized in that, Also includes: After the rainfall event ends in the target area, the real-time peak flow measurement value is retrieved; If the real-time deviation vector between the predicted real-time peak flow and the measured real-time peak flow exceeds the preset tolerance threshold, then the dynamic feedback correction of the dynamic parameter mapping matrix is ​​triggered.

Images