Large river basin flood forecasting method based on improved Xinanjiang model
By improving the Xin'anjiang model, introducing vegetation and soil correction coefficients and a variable parameter runoff model, and adopting the particle swarm optimization algorithm, the problem of low flood forecast accuracy in the Dawen River Basin was solved, achieving higher forecast accuracy and adaptability, and supporting flood control and disaster reduction decision-making.
Patent Information
- Application Number
- CN202511491409.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-18
- Publication Date
- 2026-03-17
AI Technical Summary
Traditional flood forecasting models in the Dawen River Basin have failed to fully consider the complex underlying surface conditions and the spatiotemporal distribution characteristics of precipitation, resulting in low forecast accuracy and an inability to effectively support flood control and disaster reduction efforts.
The Xin'anjiang model was improved by introducing vegetation cover and soil texture correction coefficients, adopting a variable parameter runoff generation model, and optimizing parameters through particle swarm optimization algorithm to improve model adaptability and accuracy.
It significantly improves the accuracy and reliability of flood forecasting, reduces the root mean square error by about 30%, increases the coefficient of certainty by about 20%, adapts to different seasons and precipitation patterns, and enhances the scientific nature of flood control and disaster reduction decision-making.
Smart Images
Figure CN121683418A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of hydrology and water resources, specifically relating to a flood forecasting method for the Dawen River Basin based on an improved Xin'anjiang model. Background Technology
[0002] In the field of hydrology and water resources, flood forecasting is crucial for watershed flood control and disaster reduction. The Xin'anjiang model, as a commonly used hydrological model, is widely applied in flood forecasting. However, traditional flood forecasting models have significant shortcomings when applied to the Dawen River basin.
[0003] Existing flood forecasting technologies have the following problems: 1. They fail to fully consider the complex underlying surface conditions of the Dawen River basin, such as the impact of vegetation cover and soil texture on evapotranspiration and runoff generation; 2. They do not take into account the spatiotemporal distribution characteristics of precipitation in the basin, resulting in inaccurate runoff generation calculations; 3. Traditional model parameter calibration methods are difficult to effectively optimize model parameters, ultimately leading to low flood forecast accuracy and an inability to provide reliable support for flood control and disaster reduction efforts in the Dawen River basin. Summary of the Invention
[0004] The purpose of this invention is to provide a flood forecasting method for the Dawen River Basin based on an improved Xin'anjiang model. This method addresses the problem of low forecast accuracy in the Dawen River Basin due to the failure of traditional flood forecasting models to fully consider the complex underlying surface conditions and spatiotemporal distribution characteristics of precipitation. By making targeted improvements to the Xin'anjiang model, the accuracy and reliability of flood forecasting are enhanced, overcoming the shortcomings of previous technologies in this specific basin, and providing more effective technical support for flood control and disaster reduction efforts in the Dawen River Basin.
[0005] To solve the above-mentioned technical problems, the present invention adopts the following solution:
[0006] A flood forecasting method for the Dawen River basin based on an improved Xin'anjiang model includes the following steps:
[0007] Step 1: Data Collection and Preprocessing; Collect historical precipitation, evaporation, and flow data, as well as topographic and soil type data for the Dawen River basin. Organize precipitation data by time series, filling in missing values using Kriging interpolation. Perform spatial interpolation of evaporation data using the inverse distance weighting method based on observations from different meteorological stations. Perform quality control on flow data according to the formula... Remove outliers; let the precipitation data set be P = p1, p2, ..., p n The evaporation data set is E = e1, e2, ..., e m The traffic data set is Q = q1, q2, ..., q k , where p i Let e represent the precipitation at time point i. jThis represents the evaporation rate at the j-th spatial location. Let q be the mean flow rate, σ be the standard deviation, and q be the mean flow rate. l This is the l-th traffic data;
[0008] Step 2: Improve the structure of the Xin'anjiang model; considering the complex underlying surface of the Dawen River basin, a correction coefficient that takes into account vegetation cover and soil texture is introduced into the evapotranspiration calculation module. The formula for calculating evapotranspiration ET is improved to ET=K1×K2×E p E p K1 represents potential evapotranspiration, K2 represents vegetation cover correction factor, and K2 represents soil texture correction factor. Vegetation cover was obtained through satellite remote sensing data, and the empirical relationship between Normalized Difference Vegetation Index (NDVI) and vegetation cover was utilized. Calculation, NDVI min and NDVI max These represent the minimum and maximum NDVI values within the study area, respectively. The soil texture correction factor K2 is determined based on soil particle composition analysis, with different K2 values corresponding to different soil textures.
[0009] In the runoff calculation module, considering the uneven spatial and temporal distribution of precipitation in the watershed, a variable-parameter full-storage runoff model is adopted. The formula for calculating the runoff volume R is:
[0010] when
[0011] When R=0
[0012] Where R is surface runoff, P is precipitation, and W is... m Let W be the water storage capacity of the basin, W0 be the initial water storage, and W m W0 changes dynamically with spatial location and time.
[0013] Step 3: Parameter calibration and optimization. Using the collected historical data, the particle swarm optimization algorithm is used to calibrate the parameters of the improved Xin'anjiang model.
[0014] In particle swarm optimization, the position of each particle represents a set of model parameters, and the velocity determines the update of the particle's position. Let x be the position of the i-th particle in the d-th dimension. id The speed is v id Its update formula is v id (t+1)=ωv id (t)+c1r1(p id -x id (t))+c2r2(g d -x id (t)), x id (t+1)=x id(t)+v id (t+1), where ω is the inertia weight, c1 and c2 are learning factors, r1 and r2 are random numbers between 0 and 1, and p id For the particle's own historical optimal position, g d To determine the globally optimal location, the objective function is to minimize the root mean square error (RMSE) between the measured and simulated flow rates. Q sim,i To simulate traffic, Q obs,i The measured flow rate is N, and the number of data samples is N. The optimal model parameters are obtained by iteratively optimizing the particle positions.
[0015] Step 4: Flood Forecasting and Validation. Real-time collected precipitation and evaporation data are input into the improved Xin'anjiang model after parameter calibration and optimization. Runoff generation and confluence calculations are performed according to the improved model structure to obtain the flood discharge forecast. The coefficient of determination (DC) is used to validate the forecast results. in The DC value is the average of the measured flow rate. If the DC value is greater than the set threshold, the forecast result is considered reliable; otherwise, the data needs to be checked again or the model parameters need to be adjusted.
[0016] Furthermore, in step 1, organizing the precipitation data by time series includes integrating and sorting data from different meteorological stations.
[0017] Furthermore, in step 2, the K2 values corresponding to different soil textures are experimentally determined and a corresponding relationship table is formed.
[0018] Furthermore, in step 3, the initial parameter settings of the particle swarm optimization algorithm are determined based on experience and experiments.
[0019] Furthermore, in step 4, the threshold value is set to a range of 0.7-0.9.
[0020] Furthermore, the topographic data in step 1 includes information such as slope and aspect.
[0021] Furthermore, in step 2, W in the production flow calculation module m The dynamic changes of W0 are determined by a combination of factors such as topography, soil, and precipitation.
[0022] Furthermore, if the forecast results are unreliable, the data should be re-examined, including checking the accuracy, completeness, and reliability of the data source.
[0023] The beneficial effects of this invention are as follows:
[0024] 1. Significantly improved forecast accuracy: By considering the complex underlying surface conditions and the spatiotemporal distribution characteristics of precipitation in the Dawen River Basin, the Xin'anjiang model was improved, and the parameters were calibrated using the particle swarm optimization algorithm. This resulted in a reduction of the root mean square error of flood forecasts by about 30% and an increase of the coefficient of determination by about 20% compared to the traditional Xin'anjiang model. In simulations of multiple historical flood events, the predicted flow rate showed a significantly better fit with the measured flow rate, and the model was able to capture the peak flood value and peak occurrence time more accurately.
[0025] 2. Enhanced Adaptability: The improved model is better able to adapt to the flood forecasting needs of the Dawen River Basin under different seasons and precipitation patterns. For example, during periods of concentrated summer rainstorms, traditional models have larger forecast errors, while the model of this invention can still maintain high forecast accuracy, effectively improving the scientific nature of flood control and disaster reduction decisions in this basin.
[0026] 3. Expanding the scope of application: The method of this invention is not only applicable to the Dawen River Basin, but also has certain reference and promotion value for other basins with similar underlying surface conditions and precipitation characteristics, which can promote the development of flood forecasting technology in the field of hydrology and water resources to a certain extent. Attached Figure Description
[0027] Figure 1 This is a schematic diagram of the overall process of the invention method;
[0028] Figure 2 This is a schematic diagram of the structure of the improved evapotranspiration calculation module of the Xin'anjiang model in the invention. Detailed Implementation
[0029] To make the objectives, technical solutions, and technical effects of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments.
[0030] Example 1
[0031] like Figure 1 As shown, the flood forecasting method for the Dawen River basin based on the improved Xin'anjiang model includes the following steps:
[0032] Step 1: Data Collection and Preprocessing
[0033] In the Ningyang County section of Tai'an City, historical precipitation, evaporation, and flow data, as well as underlying surface data such as topography and soil type, were collected for the Dawen River basin. For precipitation data, staff meticulously organized it according to time series. If missing values were found, Kriging interpolation was used to fill them. For example, there are multiple meteorological monitoring stations in Ningyang County; when precipitation data for a particular time period is missing at a station, the missing precipitation amount for that period was estimated using data from surrounding stations through Kriging interpolation. Regarding evaporation data, spatial interpolation was performed using the inverse distance weighting method based on observation data from different meteorological stations in Ningyang County to obtain the evaporation distribution throughout the Dawen River basin in Ningyang County. Flow data underwent strict quality control to remove outliers. Let the precipitation dataset be P = p1, p2, ..., p n , where p i This represents the precipitation at the i-th time point; the evaporation data set is E = e1, e2, ..., e m e j This represents the evaporation rate at the j-th spatial location. The flow rate dataset is Q = q1, q2, ..., q k The judgment of outliers is based on the formula. in Let q be the mean flow rate, σ be the standard deviation, and q be the mean flow rate. l For the l-th flow data point, if it satisfies this formula, it is determined to be an outlier and removed. For example, in the flow data recorded by a hydrological monitoring station in Ningyang County, the flow rate at a certain moment is determined to be an outlier through calculation and is removed from the dataset.
[0034] Step 2: Improve the structure of the Xin'anjiang model
[0035] To address the complex underlying surface characteristics of the Dawen River basin in Ningyang County, Tai'an City, a correction coefficient considering vegetation cover and soil texture is introduced into the evapotranspiration calculation module of the Xin'an River model. The formula for calculating evapotranspiration ET is thus improved to ET = K1 × K2 × E. p E p K1 represents potential evapotranspiration, K2 is the vegetation cover correction factor, and K1 is the soil texture correction factor. Vegetation cover was obtained through satellite remote sensing data, utilizing the empirical relationship between the Normalized Difference Vegetation Index (NDVI) and vegetation cover. Calculation, NDVI min and NDVI max These represent the minimum and maximum NDVI values within the study area of Ningyang County, respectively.
[0036] For example, NDVI data for the Dawen River basin in Ningyang County was acquired through satellite remote sensing, and the vegetation cover correction coefficient K1 for the region was calculated. The soil texture correction coefficient K2 was determined based on soil particle composition analysis in Ningyang County, with different K2 values corresponding to different soil textures. In the runoff calculation module, considering the uneven spatial and temporal distribution of precipitation in the region, a variable-parameter saturation runoff model was adopted, and the formula for calculating the runoff volume R is as follows: Among them W m Let W be the water storage capacity of the basin, W0 be the initial water storage, and W m W0 changes dynamically with spatial location and time. For example, in different sub-basins of Ningyang County, different values of W are determined based on local topography, soil, and other conditions. m And W0 value.
[0037] Step 3: Parameter Calibration and Optimization
[0038] Using historical data collected from the Dawen River basin in Ningyang County, the improved Xin'anjiang model parameters were calibrated using the Particle Swarm Optimization (PSO) algorithm. In PSO, the position of each particle represents a set of model parameters, and the velocity determines the update of the particle's position. Let x be the position of the i-th particle in the d-th dimension. id The speed is v id Its update formula is:
[0039] v id (t+1)=ωv id (t)+c1r1(p id -x id (t))+c2r2(g d -x id (t)),
[0040] x id (t+1)=x id (t)+v id (t+1),
[0041] Where ω is the inertia weight, c1 and c2 are learning factors, r1 and r2 are random numbers between 0 and 1, and p id For the particle's own historical optimal position, g d The global optimal location is determined by minimizing the root mean square error (RMSE) between the measured and simulated flow rates.
[0042]
[0043] Q sim,i To simulate traffic, Q obs,iLet N be the measured flow rate and N be the number of data samples. The optimal model parameters are obtained by iteratively optimizing the particle positions. For example, in the flood simulation in Ningyang County, multiple iterations of calculations continuously reduced the RMSE between the simulated and measured flow rates, ultimately determining the optimal parameters.
[0044] Step 4: Flood Forecasting and Verification
[0045] Real-time precipitation and evaporation data collected from Ningyang County, Tai'an City, were input into the improved Xin'anjiang River model after parameter calibration and optimization. Runoff generation and confluence calculations were performed according to the improved model structure to obtain flood flow forecasts. The forecast results were verified using the deterministic coefficient DC.
[0046]
[0047] in The DC value represents the average measured flow rate. If the DC value is greater than a set threshold (e.g., 0.8), the forecast result is considered reliable; otherwise, the data needs to be rechecked or the model parameters adjusted. For example, in a flood forecast for Ningyang County, the calculated DC value was 0.85, indicating that the forecast result was reliable and could provide effective data support for local flood prevention and disaster reduction.
[0048] It should be understood that any parts not described in detail in this invention belong to the prior art.
[0049] The above description, in conjunction with the accompanying drawings, is merely a specific implementation method and process of the present invention. However, the scope of protection of the present invention is not limited thereto. Any person skilled in the art should understand that this is only an illustrative example, and various changes and substitutions can be made to this implementation method without departing from the essence of the present invention. The scope of the present invention is defined only by the appended claims.
Claims
1. A flood forecasting method for the Dawen River basin based on an improved Xin'anjiang model, characterized in that, Includes the following steps: Step 1: Data Collection and Preprocessing. Historical precipitation, evaporation, and flow data, as well as topographic and soil type data for the Dawen River basin, were collected. Precipitation data were organized by time series, and missing values were filled using Kriging interpolation. Evaporation data were spatially interpolated using the inverse distance weighting method based on observations from different meteorological stations. Flow data underwent quality control according to the formula... Remove outliers; let the precipitation data set be P = p1, p2, ..., p n The evaporation data set is E = e1, e2, ..., e m The traffic data set is Q = q1, q2, ..., q k , where p i Let e represent the precipitation at time point i. j This represents the evaporation rate at the j-th spatial location. Let q be the mean flow rate, σ be the standard deviation, and q be the mean flow rate. l This is the l-th traffic data; Step 2: Improve the structure of the Xin'anjiang model. In the evapotranspiration calculation module, introduce correction coefficients that consider vegetation cover and soil texture. The formula for calculating evapotranspiration ET is improved to ET=K1×K2×E p E p K1 represents potential evapotranspiration, K2 represents vegetation cover correction factor, and K2 represents soil texture correction factor. Vegetation cover was obtained through satellite remote sensing data, and the empirical relationship between Normalized Difference Vegetation Index (NDVI) and vegetation cover was utilized. Calculation, NDVI min and NDVI max These are the minimum and maximum values of NDVI within the study area, respectively. The soil texture correction factor K2 is determined based on soil particle composition analysis. In the flow generation calculation module, a variable parameter full flow generation model is adopted, and the formula for calculating the flow generation rate R is: when When R=0 Where R is surface runoff, P is precipitation, and W is... m Let W be the water storage capacity of the basin, W0 be the initial water storage, and W m W0 changes dynamically with spatial location and time; Step 3: Parameter calibration and optimization. Using the collected historical data, the particle swarm optimization algorithm is used to calibrate the parameters of the improved Xin'anjiang model. Step 4: Flood Forecasting and Validation. Real-time collected precipitation and evaporation data are input into the improved Xin'anjiang model after parameter calibration and optimization. Runoff generation and confluence calculations are performed according to the improved model structure to obtain the flood discharge forecast. The coefficient of determination (DC) is used to validate the forecast results. in The DC value is the average measured flow rate. If the DC value is greater than the set threshold, the forecast result is considered reliable; otherwise, the data needs to be checked again or the model parameters need to be adjusted.
2. The flood forecasting method for the Dawen River basin based on the improved Xin'anjiang model according to claim 1, characterized in that, In step 1, organizing precipitation data by time series includes integrating and sorting data from different meteorological stations.
3. The flood forecasting method for the Dawen River basin based on the improved Xin'anjiang model according to claim 2, characterized in that, In step 2, the K2 values corresponding to different soil textures are determined experimentally and a corresponding relationship table is formed.
4. The flood forecasting method for the Dawen River basin based on the improved Xin'anjiang model according to claim 3, characterized in that, In step 3, the initial parameter settings of the particle swarm optimization algorithm are determined based on experience and experiments.
5. The flood forecasting method for the Dawen River basin based on the improved Xin'anjiang model as described in claim 4, characterized in that, In step 4, the threshold value is set to a range of 0.7-0.
9.
6. The flood forecasting method for the Dawen River basin based on the improved Xin'anjiang model as described in claim 5, characterized in that, The topographic data in step 1 includes information such as slope and aspect.
7. The flood forecasting method for the Dawen River basin based on the improved Xin'anjiang model as described in claim 6, characterized in that, In step 2, W in the production flow calculation module m The dynamic changes of W0 are determined by a combination of factors such as topography, soil, and precipitation.
8. The flood forecasting method for the Dawen River basin based on the improved Xin'anjiang model as described in claim 7, characterized in that, In step 4, if the forecast result is unreliable, the data is re-examined, including checking the accuracy, completeness, and reliability of the data source.
9. A flood forecasting method for the Dawen River basin based on an improved Xin'anjiang model according to claim 1, characterized in that, Step 3 specifically involves: Let x be the position of the i-th particle in the d-th dimension. id The speed is v id Its update formula is v id (t+1)=ωv id (t)+c1r1(p id -x id (t))+c2r2(g d -x id (t)), x id (t+1)=x id (t)+v id (t+1), where ω is the inertia weight, c1 and c2 are learning factors, r1 and r2 are random numbers between 0 and 1, and p id For the particle's own historical optimal position, g d To determine the globally optimal location, the objective function is to minimize the root mean square error (RMSE) between the measured and simulated flow rates. Q sim,i To simulate traffic, Q obs,i The measured flow rate is N, and the number of data samples is N. The optimal model parameters are obtained by iteratively optimizing the particle positions.