Debris flow runoff threshold probability forecasting system construction method and small watershed debris flow early warning method

Abstract

The invention discloses a debris flow runoff threshold probability forecasting system construction method and a small watershed debris flow early warning method. The invention provides a debris flow runoff threshold probability forecasting system construction method for overcoming the defects of an existing dichotomy forecasting logic monitoring technology. According to the method, rainfall is regarded as a disaster clue, dimensionless single-width flow q * is regarded as a core variable, and a mapping relation among terrain parameters, q * and proportionality coefficients is established according to a model; the topographic parameters represent the influence of the topographic features on the disaster threshold in local and global meanings; the proportionality coefficient represents a q * induced soil position point starting proportion through a curve shape; and extracting a threshold corresponding probability from key parameter cumulative frequency analysis by using fitting generated data. According to the technical scheme, a complete rainfall runoff confluence input conduction model is established to be connected into a system construction core link, and a median particle size dynamic valuing method is adopted in the optimization scheme. The invention adopts a monitoring system threshold setting technical concept opposite to the prior art, is more scientific and rigorous, and has higher utilization value.

Description

Technical Field

[0001] The present invention relates to rainfall-induced debris flow monitoring, forecasting, and early warning technologies, particularly to the construction and application of a debris flow early warning signal system using rainfall data monitoring, as well as a debris flow early warning method. This technology relates to the generation and transmission of forecast signal systems, digital computing devices or data processing devices or methods specifically adapted for specific functions, and debris flow disaster prevention and control technologies. Background Art

[0002] Rainfall is a primary factor in triggering debris flows, but it is not a deterministic factor. Therefore, debris flow monitoring and early warning methods based on binary forecasting logic using rainfall monitoring indicators, whether based on statistical data or physical processes, often have significant forecast errors because they directly correlate the satisfaction of critical rainfall conditions with the level of debris flow risk, reducing the application value of monitoring and early warning technologies. A debris flow probabilistic forecast system that considers rainfall as a clue to debris flow occurrence and uses rainfall data as initial values ​​can reduce the false alarm rate compared to deterministic forecast systems due to its nonlinear perspective on the relationship between rainfall and debris flow occurrence, thereby improving the utility of forecast tools.

[0003] Prior art 2020113548912 provides a method for forecasting the probability and scale of debris flows. Specifically, for forecasting the probability and scale of debris flow disasters, the initial influencing factor data of the corresponding debris flow disasters are obtained through field investigations and surveys, and sample data are sorted. The corresponding main influencing factor data are extracted based on the FMPCE algorithm. Debris flow disaster probability and scale forecasting models based on the optimal path forest and matrix random approximate singular value decomposition optimized width learning are respectively constructed. The test sample data are input into the established debris flow disaster probability forecasting model and debris flow disaster scale forecasting model, and the forecast information of the debris flow probability and scale is output. The main defects of this technical solution are: First, the method based on the optimal path forest-based debris flow disaster probability forecasting and the matrix random approximate singular value decomposition optimized width learning debris flow disaster scale forecasting essentially use the collective effect of multiple indicator data to solve the "nonlinear" relationship model expression between rainfall indicators and debris flow occurrence, which greatly increases the data requirements, computing requirements, and monitoring costs of forecasting and warning. Second, the model is based on historical data statistics, which has shortcomings such as reflecting "past experience" and having limited predictive power for extreme conditions. Third, the forecast probability thresholds are derived from artificial grading of statistical data results, which lacks scientific validity. Summary of the Invention

[0004] The purpose of the present invention is to address the deficiencies of the existing technology and provide a method for constructing a debris flow threshold probability prediction system, as well as a basin debris flow prediction method implemented based on this method.

[0005] To achieve the above objectives, the present invention first provides a method for constructing a debris flow runoff threshold probability forecasting system, and its technical solution is as follows.

[0006] A method for constructing a debris flow runoff threshold probability forecasting system, characterized by: Step S100: Delineate the debris flow forecast area, obtain basic data of the forecast area, and set several soil start monitoring sites for debris flow forecast in the forecast area based on the basic data. p ; Step S200: Determine the design monitoring threshold interval K and within the range x 1 threshold level K n , K Corresponding to all monitoring sites p The percentage of sites where soil initiation occurs; Step S300: Use the basic data of the forecast area to simulate the rainfall process and surface runoff process in the forecast area, and calculate the p Dimensionless single-width flow generated during rainfall q * , obtain the predicted area location p Dimensionless single-width traffic data set {( p , q * )}, Using the dataset {( p , q * )}, according to the equation 1 model fitting equation, extract the global constant in the equation 1 model N , determine the parameters N Fixed value, parameter N is the morphological coefficient of the forecast area; Formula 1 Where, i - Site p The channel slope, unit is °, is determined based on basic data. C - proportionality factor; Step S400, using the data set {( p , q * )}、 N value, given scale factor C Take values ​​to design monitoring threshold intervals K Different threshold levels withinK n , according to the equation 1 model fitting equation, generate sample data (tan i , q * , K n , C ),Will C Value data is stored in the data set {(tan i , q * , C )} until the data size m Meet the monitoring threshold K The amount of data required for probabilistic forecasts; Step S500: For the dataset {(tan i , q * , C )} to conduct cumulative frequency analysis and count any C The cumulative frequency P ; Based on sample data (tan i , q * , C )and C → P Mapping relationship, construct sample data (tan i , q * , P ), generate sample data set {(tan i , q * , P )}; Step S600, using the sample data set {(tan i , q * , P )}Build a three-dimensional coordinate system, x, y , z-axis are tan i value, q * value, P Value, the resulting coordinate system tan i - q * - P That is the debris flow runoff threshold probability forecast system in the debris flow forecast area.

[0007] The above-mentioned method for constructing a debris flow runoff threshold probability prediction system uses dimensionless single-width flow as the core variable and uses the formula 1 model to establish a mapping relationship between terrain parameters, dimensionless single-width flow, and proportionality coefficient. Among them, the terrain parameters are divided into site-level parameters and forecast area-level parameters, which characterize the impact of terrain characteristics on the actual significance of the debris flow threshold in a local and global sense. Therefore, by extracting the global terrain parameter constant term, the data analysis constructed based on the formula 1 model can be adapted to a specific forecast area; at the same time, through preliminary research, the present invention has determined that the starting ratio of soil sites induced by single-width flow in the forecast area can be characterized by the curve shape determined by the proportionality coefficient. Therefore, it is possible to construct a forecast system for debris flow threshold around the forecast area. Based on the fitted generated data, the proportionality coefficient is subjected to frequency analysis results, and its frequency is converted into probability, thereby determining the threshold probability.

[0008] The present invention provides an optimization scheme for the above-mentioned method for constructing a debris flow runoff threshold probability forecasting system. Each optimization scheme can be implemented separately or simultaneously without conflict.

[0009] Optimization 1: In step S300, the rainfall process data is used to first calculate any point p Single-width flow generated during rainfall q , and then q Dimensionless q * , specifically, the dimensionless processing is performed according to the model of formula 2. Among them, r s and D 50 They are respectively the density of solid particles in debris flow (kg / m 3 ) and median particle size (m), r is the density of water (kg / m 3 ).

[0010] Formula 2 Using the model of Equation 2 to make it dimensionless, the characteristic value of solid particles in debris flow can be D 50 The threshold system construction process is introduced to take into account the possible changes in the distribution characteristics of fixed particles in debris flows during the hydrological process. D 50 The dimensionless process of the model in formula 2 is added in a dynamic value manner. Specifically, it is determined according to the underlying surface conditions of the forecast area. D 50 Range value, then design the value step size during the calculation process to achieve D 50 Dynamic value generation to generate multiple sets of the same monitoring site in a rainfall q * value.

[0011] Optimization 2: In step S400, the fitting is determined in two steps N Value and C The first step is to use some sample data to fit the formula 1 and extract the global constant term N For example, the model of Equation 1 is logarithmically transformed to eliminate C value, and then use a small number of random samples to calculate and determine N value, and submit all samples for verification; the second step is to N Under the value condition, C Value fitting calculation. Through two-step splitting, the first step only requires a small amount of data calculation, and the second step only requires simple quantile calculation, which can significantly reduce the amount of calculation and computational complexity while ensuring physical consistency.

[0012] Optimization three: When simulating the surface runoff process, the SCS-CN runoff model is used to simulate the runoff process under different rainfall conditions, and the motion wave equation confluence model is used to simulate the confluence process under different runoff conditions, thereby establishing an input model for surface runoff single-width flow calculation.

[0013] Optimization 4: Use designed rain patterns to access the surface runoff simulation process to achieve precise control of rainfall input conditions in the simulation calculation.

[0014] The present invention also provides a method for predicting debris flow in a watershed based on the method for constructing a debris flow runoff threshold probability prediction system. The specific technical solution is as follows.

[0015] A method for predicting debris flow in a small watershed is characterized by: defining the debris flow forecast area of ​​the small watershed, constructing a debris flow runoff threshold probability forecast system in the forecast area using the above-mentioned debris flow runoff threshold probability forecast system construction method, deploying a runoff monitoring system in the forecast area, placing the constructed debris flow runoff threshold probability forecast system into the runoff monitoring system, and during rainfall, according to real-time q * Data input debris flow forecast results.

[0016] The field investigation referred to in this technology includes various geological surveys, reconnaissance, mapping, and measurement work at the alluvial fan site at the outlet of the river channel where the project is located, as well as existing simulation experiments, testing experiments, observation experiments, and analysis experiments in the field, as well as the acquisition of historical disaster records, relevant technical specifications, and empirical methods and data acquisition for reference. The data obtained from the field investigation is collectively referred to as the basic engineering data of this technical solution.

[0017] Compared with the existing technology, the beneficial effects of the present invention are: (1) The method for constructing a debris flow runoff threshold probability forecasting system of the present invention provides a new technical concept for predicting the probability of debris flow occurrence. Taking the dimensionless single-width flow as the core variable, the model of Formula 1 is used to establish the mapping relationship between terrain parameters, dimensionless single-width flow, and proportional coefficient, making it possible to construct a forecasting system for the debris flow occurrence threshold around the forecast area and obtain a practical technical solution. The existing probability forecasting model is usually artificially divided into probability thresholds and is affected by experience. Starting from the frequency analysis of key parameters, the present invention converts its frequency into probability, which is a technical concept opposite to the existing technology and also makes the probability threshold of the forecast volume more scientific and rigorous. (2) Surface runoff changes with the rainfall process, dynamically shaping the material composition and particle size distribution characteristics of the debris flow. In the optimization scheme of the present invention, by introducing the model of Formula 2, the characteristic value of the solid particles of the debris flow is converted into probability. D 50 Introducing the threshold system construction process to achieve the median particle size in the system construction process D 50 The dynamic value of makes the system construction process more consistent with the debris flow initiation and mixing principle. (3) The present invention establishes a complete rainfall-runoff-confluence input conduction model to connect the core link of the runoff threshold system construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the technical route for the construction of a debris flow runoff threshold probability forecasting system.

[0019] Figure 2 It is a debris flow forecast area and soil initiation monitoring site p Layout diagram.

[0020] Figure 3 It is a monitoring site p Schematic diagram showing the meaning of some terrain data.

[0021] Figure 4 It is a probability forecast system for debris flow runoff threshold in a small watershed in a certain mountainous area. DETAILED DESCRIPTION

[0022] The preferred embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0023] Example 1 A small mountainous watershed in Wenchuan County, Aba Tibetan and Qiang Autonomous Prefecture, Sichuan Province, is prone to rainfall-induced debris flows. The method of the present invention was used to construct a debris flow runoff threshold probability forecasting system for this small watershed, which was used for debris flow monitoring and early warning in the area.

[0024] Figure 1 It is a schematic diagram of the technical route for the construction of a debris flow runoff threshold probability forecasting system.

[0025] 1. Soil start-up monitoring points in the forecast area p Layout A small stream in this mountainous area was designated as a debris flow forecast area. An on-site survey was conducted to obtain basic data for the forecast area. This data included basin topography, meteorological data, geological data, and historical debris flow records. Some of this basic data is shown in Table 1.

[0026] Table 1 Basic data of the forecast area (partial)

[0027] Based on basic data, possible soil starting locations in the forecast area are analyzed and set as soil starting monitoring locations for debris flow forecasting. p In order to improve the accuracy of the constructed system, the soil start monitoring site p To be laid out at multiple points along the channel, M =35, numbered p i ( i =0,1,2,…,34), each site monitors p Coordinates (x, y, z) ( Figure 2 ). Match the terrain data in the basic data to each monitoring site p Table 2 shows the soil start monitoring locations p Basic information, including slope width w Finger point p The width of the channel section, in m; the slope i Refers to the site on the longitudinal section of the channel p Unit slope distance before and after l (In this example, l =0.5m) i , unit ° ( Figure 3 ).

[0028] Figure 2 It is a debris flow forecast area and soil initiation monitoring site p Layout diagram; Figure 3 It is a monitoring site p slope i Meaning diagram.

[0029] Table 2 Soil start-up monitoring sites in the forecast area p Basic terrain data (35)

[0030] 2. Design monitoring threshold interval K Design monitoring threshold intervals based on the risk warning level during operation of the project probability forecast systemK =[10%,50%]; set the step size based on the expected data volume and previous experience d 1=0.1. Determine the specific calculation of 5 threshold level states in the simulation calculation, which are K 1=10%, K 2=20%, K 3=30%, K 4=40%, K 5=50%, then there is a threshold amount of data x 1 = 5. Threshold K Represents all monitoring sites at a certain moment in the simulation calculation process p The percentage of sites where soil initiation occurs.

[0031] 3. Get the dataset { q i *} and determine the parameters of model 1 N To obtain each point p i Dimensionless single-width flow q i * , and the dataset { q i * In this example, the surface runoff simulation and runoff simulation in the forecast area are combined to simulate the surface runoff process and calculate the p i Single-width flow q i , and then q i Sure q i * .

[0032] 3.1 Construction of rainfall input model for runoff generation According to the rainfall conditions that induce debris flow in Table 1, combined with the rainfall distribution characteristics of the rainfall process in the forecast area shown by the basic data, the rain pattern design method is used to design the rain pattern (Table 3). A total of 16 rain patterns are designed, so there are rain pattern data. x 4=16. Among them, the peak rainfall intensity of rain type I RZ At 6:00, the peak rainfall intensity of rain type II RZ At 7:00, the peak rainfall intensity of rain type III RZ At 8:00, and so on, the peak rainfall intensity of rain type XVI RZ At 21:00.

[0033] Table 3 P-RZ-T rain pattern design method

[0034] 3.2 Simulating the rainfall process and locations in the forecast area p Dynamic single-width traffic q In this example, the SCS-CN runoff model is used to simulate the runoff process in the forecast area under different rainfall conditions (i.e. different rain types). The key parameters of the SCS-CN runoff model are determined based on the basic data of the forecast area. CN The value range is [68,76]. According to previous experience, the step size is set d 2=2, total, CN =68, 70, 72, 74, 76, a total of 5 values, then there are CN Value data volume x 2=5.

[0035] In this example, the Kinematic Wave equation is used as the runoff model to simulate the runoff process in the forecast area under different runoff conditions. The Manning coefficient in the Kinematic Wave equation runoff model is determined based on the basic data of the forecast area. n The value range is [0.06, 0.10]. Based on previous experience, set the step size d 3=0.02, total, n =0.06, 0.08, 0.10, a total of 3 values, then n Value data volume x 3=3.

[0036] The simulation calculation of the specific surface runoff generation and convergence process refers to the existing technology (Sun Yuqing et al., Flood simulation and uncertainty analysis of small watersheds in the Wenchuan earthquake-stricken area, "Journal of Natural Disasters", Vol. 31, No. 1, February 2022) methods.

[0037] During the calculation process, a set of simulation processes from 0:00 to 24:00 is considered as a rainfall, and a set of [rain type, CN , n The parameter condition simulation process is one rainfall, and the calculation process has a total of 240 rainfalls.

[0038] 3.3 Determining the dynamic dimensionless single-width flow rate of the rainfall process in the forecast area q * During the simulated rainfall process, each point p Single-width traffic generated q Dimensionless processing. Specifically, use the formula 3 model to convert each data q Dimensionless, obtain the corresponding dimensionless single-width flow q * , calculating D 50 The value is 0.03m.

[0039] Take one rainfall (rain type VII, CN =72, n =0.08) as an example, some relevant calculation data are shown in Table 4. In Table 4, Q - Peak flow, q - Single-width traffic.

[0040] Table 4 A rainfall (rain type VII, CN =72, n =0.08) Simulation calculation of some data

[0041] Calculate all sites p , obtain the dimensionless single-width flow data set of 240 rainfall events in the forecast area { q * j,i ∣ j =1,2,3,…, 240; i= 0,1,2,…,34}, q * j,i Indicates the j The first rainfall i loci p of q * data.

[0042] 3.4 Determine the terrain characteristic parameters of the forecast area N Each rainfall data q * j,i Data and sites p i slope i i Composition of sample data ( q * j,i , i i ), with some sample data ( q * j,i , i i ) According to the model fitting equation of formula 1, the parameters N is the morphological coefficient of the forecast area, and is the global constant of the model in Equation 1. The global constant is extracted through fitting.

[0043] In this example, the fitting results determine the terrain morphological parameters of the forecast area N =0.87.

[0044] 3.5 Determine the dataset Kn , C} Use the j =1,2,3,…, 240 rainfall sample data ( q * i , i i ), i =0,1,2,…,34 and N =0.87, the proportional coefficient C takes different monitoring threshold levels K n (In this example, K n They are K 1=10%, K 2=20%, K 3=30%, K 4=40%, K 5=50%,)According to the equation of formula 1, the data of this rainfall is determined ( K n , C ).

[0045] This example uses the quantile regression method to determine the proportionality coefficient C , specifically: In one rainfall, based on 35 site data ( q * i , i i ), when C= K n When the C value fitting calculation is performed, 35 data ( q * i , i i ) K n The data points satisfy Equation 3. The C value and the independent variable form the sample data (tan i i , q * i , K n , C ).

[0046] Formula 3 In this example, using { q * j,i ∣ j =1,2,3,…, 240; i= 0,1,2,…,34} and { i i , i= 0,1,2,…,34} m = x 1* x 2* x 3* x 4 = 1200 fits, generating the data set {(tan i i , q * j,i , C )}.

[0047] 4. Dataset {(tan i i , q * j,i , C Cumulative frequency analysis For the dataset {(tan i , q * , C )} Perform cumulative frequency analysis to determine each C Cumulative frequency of values ​​(intervals) P ,by C The cumulative frequency of the value represents the corresponding threshold K n The probability of occurrence, construct sample data (tan i , q * , P ), generate sample data set {(tan i , q * , P )}. Table 5 is the sample data set {(tan i , q * , P )}Part of the data.

[0048] Table 5 Partial sample data (tan i , q * , P )

[0049] 5. Establish a probability forecast system for runoff thresholds in the forecast area Based on the sample data set {(tan i , q * , P)}, and supplement the sample data by interpolation calculation until the sample data volume meets the requirements of two adjacent P The difference between them is no more than 0.01.

[0050] This example uses linear interpolation.

[0051] Using all sample data (tan i , q * , P ) Construct a three-dimensional coordinate system, where the x, y, and z axes are tan i value, q * value, P Value, the resulting coordinate system tan i - q * - P This is the probability forecast system for debris flow runoff threshold in the debris flow forecast area. In this probability forecast system, for any monitoring point A in the forecast area, its tan i Value (x-axis), and during rainfall q * Value (y-axis), then the xy plane projection q * -tan( i ) corresponds to the monitoring threshold level (called q * -tan(θ) threshold), represents the runoff monitoring threshold in the forecast area. In the system, each q * -tan( i ) The threshold corresponds to a certain probability (z axis), the higher the z axis position q * -tan( i ) threshold, which means the greater the probability of debris flow.

[0052] Figure 4 It is a probability forecast system for debris flow runoff threshold in a small watershed in a mountainous area. Using this system, in the debris flow monitoring and early warning of a small watershed in a mountainous area, the xy plane threshold is adjusted from the lowest to the highest. q *=7.61tan( i ) -0.87 To the highest q *=70.61tan( i ) -0.87 , the corresponding occurrence probabilities are 0% and 100% respectively; and the runoff threshold corresponding to the 50% occurrence probability is q *=25.58tan( i ) -0.87 ; Commonly used thresholds in the example study area q*=24.40tan( i ) -0.87 The corresponding probability of occurrence is 46.25%.

[0053] Example 2 Based on the technical route of Example 1, the median particle size is introduced into the model of formula 3 D 50 Dynamic value of .

[0054] According to the basic data of the forecast area (see Table 1), D 50 =0.02~0.04. In this range, the step size is set to 0.005, that is, D 50 =0.02, 0.025, 0.03, 0.035, 0.04.

[0055] Similarly, the rainfall in Example 1 (rain type VII, CN =72, n =0.08) as an example, some relevant calculation data are shown in Table 6 below.

[0056] Table 6 Partial data of a rainfall simulation calculation ( D 50 Dynamic value)

[0057] Same as the first embodiment, using { q * j,i ∣ j =1,2,3,…, 240; i= 0,1,2,…,34} and { i i , i= 0,1,2,…,34} m = 1200 fittings, generating the data set {(tan i i , q * j,i , C )}. Continue with the subsequent analysis steps, and the relevant data are omitted.

Claims

1. A method for constructing a debris flow runoff threshold probability forecasting system, characterized by: Step S100: Delineate the debris flow forecast area, obtain basic data of the forecast area, and set several soil start monitoring sites for debris flow forecast in the forecast area based on the basic data. p ; Step S200: Determine the design monitoring threshold interval K and within the range x 1 threshold level K n , K Corresponding to all monitoring sites p The percentage of sites where soil initiation occurs; Step S300: Use the basic data of the forecast area to simulate the rainfall process and surface runoff process in the forecast area, and calculate the p Dimensionless single-width flow generated during rainfall q * , obtain the predicted area location p Dimensionless single-width traffic data set {( p , q * )}; Step S400, using the data set {( p , q * )}, according to the equation 1 model fitting equation, extract the global constant in the equation 1 model N , and give the proportional coefficient C Take values ​​to design monitoring threshold intervals K Different threshold levels within K n , generate sample data (tan θ , q * , K n , C ),Will C Value data is stored in the data set {(tan θ , q * , C )} until the data size m Meet the monitoring threshold K The amount of data required for probabilistic forecasts; Formula 1 Where, θ - Site p The channel slope, unit is °, is determined based on basic data. C - proportionality factor; Step S500: For the dataset {(tan θ , q * , C )} Perform cumulative frequency analysis and count any C The cumulative frequency P ; Based on sample data (tan θ , q * , C )and C → P Mapping relationship, construct sample data (tan θ , q * , P ), generate sample data set {(tan θ , q * , P )}; Step S600, using the sample data set {(tan θ , q * , P )} Construct a three-dimensional coordinate system, where the x, y, and z axes are tan θ value, q * value, P Value, the resulting coordinate system tan θ - q * - P That is the debris flow runoff threshold probability forecast system in the debris flow forecast area.

2. The construction method according to claim 1, characterized in that: The step S400 includes: Step S410, using the data set {( p , q * )}, according to the equation 1 model fitting equation, extract the global constant in the equation 1 model N , determine the parameters N Fixed value, parameter N is the morphological coefficient of the forecast area; Step S420, using the data set {( p , q * )}、 N value, given scaling factor C Take values ​​to design monitoring threshold intervals K Different threshold levels within K n , according to the equation 1 model fitting equation, generate sample data (tan θ , q * , K n , C ).

3. The construction method according to claim 1 or 2, characterized in that: In step S400, the proportionality coefficient is determined by quantile regression C ,when C Value K n When , the fitting result C Valuer K n Data points (tan θ , q * ) satisfies the following formula, Formula 3.

4. The construction method according to claim 1, wherein: In step S600, using the sample data set {(tan θ , q * , P )} Perform interpolation calculations to expand the sample data and then construct a three-dimensional coordinate system.

5. The construction method according to claim 1, wherein: In step S300, when simulating the surface runoff process using the basic data of the forecast area, the SCS-CN runoff generation model is first established, and the model is determined based on the soil and soil utilization characteristics of the debris flow forecast area. CN The range of values, as well as the simulation calculation process CN Calculation step length for values ​​within the range d 2. Use interval CN Value range and step size d 2 Determine the simulation calculation CN Value data volume x 2. Then establish the confluence model of the motion wave equation and determine the Manning coefficient in the model according to the roughness characteristics of the debris flow forecast area. n The value range of n Calculation step length within the value range d 3. Utilize n Range value and step size d 3. Determine the simulation calculation n Data volume x 3; In step S400, the data size m yes x 1. x 2. x The product of 3.

6. The construction method according to claim 5, characterized in that: In step S300, when simulating the rainfall process in the forecast area using the basic data of the forecast area, the number of rain types that may induce debris flow in the forecast area is determined according to the basic data of the forecast area. x 4; In step S400, the data size m yes x 1. x 2. x 3. x The product of 4.

7. The construction method according to claim 6, characterized in that: In step S300, the number of rain patterns is determined using the rain pattern design method. x 4. Determine the total rainfall based on the basic data of the forecast area P , peak rainfall intensity RZ , total design rainfall duration T , adjust peak rainfall intensity RZ Rain pattern design is carried out according to the appearance time.

8. The construction method according to any one of claims 1 to 7, characterized in that: Step S300, calculate any site p Single-width flow generated during rainfall q , and then q According to the model of formula 2, dimensionless processing is performed to obtain q * , Formula 2 Where, ρ s 、 ρ - are respectively the density of debris flow solid particles and the density of water, in kg / m 3 , are determined based on basic data. D 50 - Median particle size of debris flow solid particles, in meters, determined based on basic data.

9. The construction method according to claim 8, characterized in that: described D 50 exist D 50 Dynamic value within the range, D 50 The range value is determined based on basic data.

10. A method for predicting debris flows in a small watershed, characterized by: Delineate the debris flow forecast area of ​​the small watershed, use the debris flow runoff threshold probability forecast system construction method of any one of claims 1 to 9 to construct the debris flow runoff threshold probability forecast system of the forecast area, deploy the runoff monitoring system of the forecast area, place the constructed debris flow runoff threshold probability forecast system into the runoff monitoring system, and during the rainfall process, according to the real-time q * Data input debris flow forecast results.

Images