A basin runoff forecasting method based on dynamic calibration of river flow propagation time

Abstract

The present invention discloses a method for forecasting river basin runoff based on dynamically calibrating river flow propagation time, comprising: S1, acquiring and processing river basin meteorological and hydrological data, selecting rainfall runoff data for dynamically calibrating river flow propagation time; S2, dividing the rainfall runoff data into a calibration data set and a verification data set, and using an optimization algorithm to calibrate and verify the river flow propagation time; S3, bringing the calibrated river flow propagation time into a river confluence model, and using the river basin runoff generation and confluence model and the river confluence model to forecast the river basin runoff; the present invention significantly improves the accuracy of the forecast, has the advantages of high precision, high reliability, strong adaptability, etc., and can be applied to important scenarios such as flood control, water resource utilization, and large reservoir scheduling.

Description

Technical Field

[0001] The invention relates to the technical field of hydrological forecasting, and in particular to a basin runoff forecasting method based on dynamically calibrating river flow propagation time. Background Art

[0002] River basin runoff forecast is a hydrological forecasting method that aims to estimate the runoff in a specific river basin over a period of time in the future. It is usually used in flood warning, water resources management and hydropower station operations to help decision makers take appropriate measures to deal with possible floods or droughts. River basin runoff forecasts can predict the possibility and scale of flood and drought events in advance, enabling relevant departments to take necessary preventive measures to reduce the threat of flood and drought disasters to people's lives and property. At the same time, river basin runoff forecasts can help decision makers optimize the operation of reservoirs and hydropower stations, ensure that water users have sufficient water supply, and avoid excessive use and waste of water resources.

[0003] The river flow propagation time refers to the time required for water flow to propagate from one point to another in the river system. It reflects the speed of water flow in the river and its propagation speed over a specific distance. The calculation of river flow propagation time can help us understand the dynamic characteristics of water flow, which is of great significance to flood prediction, water resources management and environmental protection. The basin runoff forecast model generally includes two parts: basin runoff calculation and river confluence calculation. The river flow propagation time is a very critical parameter in river confluence calculation. The accuracy of this parameter directly affects the reliability of basin runoff forecast. At present, the basin runoff forecast model generally uses a fixed river flow propagation time to calculate river confluence. However, the water volume in the river is different, and the river flow propagation time is also different. Therefore, using a fixed river flow propagation time to forecast basin runoff will inevitably reduce the reliability of basin runoff forecast. Summary of the invention

[0004] The purpose of the present invention is to overcome the above-mentioned shortcomings and provide a basin runoff forecasting method based on dynamically calibrated river water flow propagation time, so as to make full use of historical rainfall runoff data to dynamically calibrate river water flow propagation time, and use the optimized river water flow propagation time to forecast basin runoff, thereby significantly improving the accuracy of the forecast and being able to be widely used in actual production.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0006] A basin runoff forecasting method based on dynamic calibration of river flow propagation time comprises the following steps:

[0007] S1. Acquire and process basin meteorological and hydrological data, and select rainfall-runoff data for dynamic calibration of river flow propagation time;

[0008] S2, dividing the rainfall runoff data into a calibration data set and a validation data set, and using an optimization algorithm to calibrate and validate the river flow propagation time;

[0009] S3. Bring the calibrated river flow propagation time into the river confluence model, and use the basin runoff generation and confluence model and the river confluence model to predict the basin runoff.

[0010] Preferably, the step S1 comprises the following steps:

[0011] S11. According to the meteorological and hydrological data of the watershed, select rainfall runoff data that meets the calculation requirements of the watershed runoff model and the river confluence model;

[0012] S12. Obtain the forecast rainfall time series data of the basin in the next 1 to 10 days through precipitation forecast products.

[0013] Preferably, step S2 comprises the following steps:

[0014] S21. Randomly select and construct calibration and validation datasets from rainfall-runoff data.

[0015] S22. Calibrate and verify the river water flow propagation time using the optimization algorithm, specifically: determine the optimization variables, constraints, and objective functions involved in calibrating the river water flow propagation time, then use the optimization algorithm SCE-UA to solve the optimal river water flow propagation time, and verify the optimized river water flow propagation time.

[0016] Preferably, the optimization variable in step S22 is the forecast basin The flow propagation time of a river channel is expressed as:

[0017] ;

[0018] The constraints are:

[0019] ;

[0020] In the formula, is the minimum flow propagation time in the river channel, is the maximum water flow propagation time in the river; and It can be obtained by measuring in dry season and flood season respectively, or it can be preliminarily determined by artificial experience for the minimum and maximum water flow propagation time of a river channel, and then dynamically adjusted according to the specific situation during calibration;

[0021] The Nash-Sutcliffe efficiency coefficient is selected as the objective function, and its expression is:

[0022] ;

[0023] In the formula, is the measured runoff of the forecast section of the basin (unit, m³ / s), Simulated runoff for the forecast section of the basin (unit, m³ / s), is the average measured runoff of the forecast section of the basin (unit, m³ / s); the simulated runoff of the forecast section of the basin It is calculated through the basin runoff model and the river runoff model, where the input of the basin runoff model is the rainfall data in the calibration data set, and the parameters of the basin runoff model are determined by combining the hydrological model parameter calibration and artificial experience.

[0024] Preferably, the verification of the optimized river flow propagation time in step S22 is specifically as follows:

[0025] ①Using the rainfall data in the validation data set as the model input, the runoff process of each sub-basin outlet in the basin is calculated and predicted using the basin runoff model;

[0026] ② The calibrated river flow propagation time is brought into the river confluence model, and the river confluence calculation is performed according to the basin topology structure to obtain the forecast runoff process at the forecast basin outlet. The measured runoff process at the forecast basin outlet corresponding to the rainfall data in the validation data set can be calculated. The value is used to verify whether the calibrated river flow propagation time meets the accuracy requirements. Value greater than 0.7.

[0027] Preferably, the forecast basin is composed of multiple sub-basins, each sub-basin is connected by a river channel, the sub-basin outlet flow process is calculated by a basin runoff generation and confluence model, and the topological structure refers to the upstream and downstream position relationship of each sub-basin and river channel in the forecast basin.

[0028] Preferably, the step S22 comprises the following steps:

[0029] S221. Assume that the dimension of the optimization problem is , The number is equal to the number of rivers in the forecast basin. and The propagation time of river water flow is calculated by generating random numbers within a limited range. Initialize and select the number of complex types participating in the evolution and the number of vertices contained in each complex , calculate the number of sample points ;

[0030] S222, distributed in the feasible space with uniform probability Randomly generate sample points indivual ;

[0031] S223, calculate the target value of each point And arrange them from small to large, storage space ,in For the best, is the worst;

[0032] S224, Division The composite type is , each composite type includes points, that is ;

[0033] S225, perform CCE evolution on each composite type;

[0034] S226, all points of each composite type after evolution form a new set and sort them again according to the target value from small to large. The storage space after sorting Also sorted in ascending order of target value;

[0035] S227, stop evolution judgment, if the evolution number limit or the target value convergence range limit condition is met, then stop evolution, otherwise return to step S224 to continue evolution.

[0036] Preferably, the step S225 includes the following steps:

[0037] S2251, give Assign weight ;

[0038] S2252, select the parent; from Choose by probability Individuals ( , , ,..., ) is stored in middle, ,in, yes The function value of The location is stored in inside;

[0039] S2253, produce offspring; including Calculate the centroid , in order from smallest to largest and To sort; to reflect on individuals , mutation, shrinkage or mutation steps and repeat the process.

[0040] Preferably, the step S3 comprises the steps of:

[0041] S31, using the forecast rainfall time series data obtained in step S12 as model input, and using the basin runoff generation and confluence model to calculate the runoff process at the outlet of each sub-basin in the forecast basin;

[0042] S32. Bring the calibrated river flow propagation time into the river confluence model, perform river confluence calculation according to the basin topology structure, and obtain the predicted runoff process at the basin outlet.

[0043] Beneficial effects of the present invention:

[0044] The present invention is based on the historical rainfall runoff process and utilizes the runoff forecasting method of dynamically calibrating the river flow propagation time to predict the runoff process caused by future rainfall. Compared with the traditional runoff forecasting model, it takes into account the scientific fact that the river flow propagation time is different under different water inflow conditions, realizes the dynamic calibration of the river flow propagation time, and greatly improves the runoff forecast accuracy. The method of the present invention has the advantages of high accuracy, high reliability, strong adaptability, etc., and can be applied to important scenarios such as flood control, water resource utilization and large reservoir scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] FIG1 is a flow chart of a basin runoff forecasting method based on dynamic calibration of water flow propagation time;

[0046] Figure 2 is a water system map of the Qujiang River Basin;

[0047] Figure 3 is a diagram of the pre-simulated flow process in the Qujiang River Basin from August 29, 2022 to September 10, 2022;

[0048] Figure 4 is the forecast rainfall and runoff process diagram for the Qujiang River Basin from September 15, 2021 to October 2, 2021. DETAILED DESCRIPTION

[0049] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0050] Example 1: Figure 1 As shown, a basin runoff forecasting method based on dynamic calibration of river flow propagation time includes the following steps:

[0051] S1. Acquire and process basin meteorological and hydrological data, and select rainfall-runoff data for dynamic calibration of river flow propagation time;

[0052] S11. According to the meteorological and hydrological data of the watershed, select rainfall runoff data that meets the calculation requirements of the watershed runoff model and the river confluence model;

[0053] S12. Obtain the forecast rainfall time series data of the basin in the next 1 to 10 days through precipitation forecast products (such as MM5, NCEP, ECMWF, WRF, etc.).

[0054] S2, dividing the rainfall runoff data into a calibration data set and a validation data set, and using an optimization algorithm to calibrate and validate the river flow propagation time;

[0055] The step S2 comprises the following steps:

[0056] S21. Randomly select and construct calibration and validation datasets from rainfall-runoff data.

[0057] S22. Calibrate and verify the river water flow propagation time using the optimization algorithm, specifically: determine the optimization variables, constraints, and objective functions involved in calibrating the river water flow propagation time, then use the optimization algorithm SEC-UA to solve the optimal river water flow propagation time, and verify the optimized river water flow propagation time.

[0058] The optimization variable in step S22 is the forecast basin The flow propagation time of a river channel is expressed as:

[0059] ;

[0060] The constraints are:

[0061] ;

[0062] In the formula, is the minimum flow propagation time in the river channel, is the maximum water flow propagation time in the river channel. and It can be measured separately during the dry season and the extremely flood season, or it can be preliminarily determined through artificial experience for the minimum and maximum water flow propagation time of a river channel, and then dynamically adjusted according to the specific situation during calibration.

[0063] The Nash-Sutcliffe efficiency coefficient is selected as the objective function, and its expression is:

[0064] ;

[0065] In the formula, is the measured runoff of the forecast section of the basin (unit, m³ / s), Simulated runoff for the forecast section of the basin (unit, m³ / s), The average measured runoff of the forecast section of the basin (unit, m³ / s). The simulated runoff of the forecast section of the basin It is calculated through basin runoff models (such as the Xin'anjiang model, NAM model, TANK model, etc.) and river confluence models (such as the Muskingum model and one-dimensional hydrodynamic model). The input of the basin runoff model is the rainfall data in the calibration data set, and the parameters of the basin runoff model are determined by combining hydrological model parameter calibration and artificial experience.

[0066] The verification of the optimized river flow propagation time in step S22 is specifically as follows:

[0067] ①Using the rainfall data in the validation data set as the model input, the runoff process of each sub-basin outlet in the basin is calculated and predicted using the basin runoff model;

[0068] ② The calibrated river flow propagation time is brought into the river confluence model, and the river confluence calculation is performed according to the basin topology structure to obtain the forecast runoff process at the forecast basin outlet. The measured runoff process at the forecast basin outlet corresponding to the rainfall data in the validation data set can be calculated. The value is used to verify whether the calibrated river flow propagation time meets the accuracy requirements (generally required value greater than 0.7).

[0069] The forecast basin is composed of multiple sub-basins, each sub-basin is connected by a river channel, and the sub-basin outlet flow process is calculated by a basin flow generation and confluence model. The topological structure refers to the upstream and downstream position relationship of each sub-basin and river channel in the forecast basin.

[0070] The step S22 comprises the following steps:

[0071] S221. Dynamic calibration of river flow propagation time is a mathematical optimization problem. Assume that the dimension of this optimization problem is , The number is equal to the number of rivers in the forecast basin. and The propagation time of river water flow is calculated by generating random numbers within a limited range. Initialize and select the number of complex types participating in the evolution and the number of vertices contained in each complex , calculate the number of sample points ;

[0072] S222, distributed in the feasible space with uniform probability Randomly generate sample points indivual ;

[0073] S223, calculate the target value of each point And arrange them from small to large, storage space ,in For the best, is the worst;

[0074] S224, Division The composite type is , each composite type includes points, that is ;

[0075] S225, perform CCE evolution on each composite type;

[0076] S226, all points of each composite type after evolution form a new set and sort them again according to the target value from small to large. The storage space after sorting Also sorted in ascending order of target value;

[0077] S227, stop evolution judgment, if the evolution number limit or the target value convergence range limit condition is met, then stop evolution, otherwise return to step S224 to continue evolution.

[0078] Preferably, the step S225 includes the following steps:

[0079] S2251, give Assign weight ;

[0080] S2252, select the father; from Choose by probability Individuals ( , , ,..., ) is stored in middle, ,in, yes The function value of The location is stored in inside;

[0081] S2253, produce offspring; including Calculate the centroid , in order from small to large and To sort; to reflect on individuals , mutation, shrinkage or mutation steps and repeat the process.

[0082] S3. Bring the calibrated river flow propagation time into the river confluence model, and use the basin runoff generation and confluence model and the river confluence model to predict the basin runoff.

[0083] The step S3 comprises the steps of:

[0084] S31, using the forecast rainfall time series data obtained in step S12 as model input, and using the basin runoff generation and confluence model to calculate the runoff process at the outlet of each sub-basin in the forecast basin;

[0085] S32. Bring the calibrated river flow propagation time into the river confluence model, perform river confluence calculation according to the basin topology structure, and obtain the predicted runoff process at the basin outlet.

[0086] The idea of ​​Example 1 is: first, obtain and process the meteorological and hydrological data of the watershed, and select rainfall-runoff data for dynamically calibrating the propagation time of river water flow. Then, divide the rainfall-runoff data into a calibration data set and a verification data set, and use the optimization algorithm to calibrate and verify the propagation time of river water flow. Finally, bring the calibrated river water flow propagation time into the river confluence model, and use the watershed runoff generation and confluence model and the river confluence model to predict the watershed runoff. This runoff forecasting method can make a relatively accurate prediction of the runoff process generated by future rainfall. Compared with the traditional hydrological forecasting model, the runoff forecasting method has the advantages of high accuracy, high reliability, and strong adaptability, and can be applied to important scenarios such as flood control, water resource utilization, and large reservoir scheduling.

[0087] Embodiment 2: This embodiment is as shown in the attached Figure 2 The application of this method is illustrated by taking the Qujiang River Basin, a tributary of the left bank of the Jialing River, as an example. The terrain of this basin is high in the northeast and low in the southwest, and the annual rainfall distribution is extremely uneven. There are 23 rain gauges and 12 hydrological stations in the basin. This embodiment uses a basin runoff forecasting method based on dynamic calibration of river flow propagation time, which can be implemented in the following steps:

[0088] Step 1: Acquire and process the basin meteorological and hydrological data, and select rainfall-runoff data for dynamic calibration of river flow propagation time;

[0089] Meteorological and hydrological data from January 2020 to August 2023 in the study basin were collected, and the acquired data were divided into rainfall runoff data. The time step of the rainfall runoff data in this embodiment is 1 hour. In actual applications, the time scale of rainfall runoff data can be changed according to the specific collected data, but it is at least a daily scale or a shorter time scale.

[0090] Step 2: Divide the rainfall-runoff data into a calibration data set and a validation data set, and use the optimization algorithm SCE-UA to calibrate and validate the river flow propagation time;

[0091] A portion of rainfall runoff data is randomly selected as calibration data, and another portion of rainfall runoff data is used as verification data. Based on the calibration data, this example uses the optimization algorithm SCE-UA to calibrate the river flow propagation time parameters, selects the Xinanjiang model for basin runoff calculation, and selects the Muskingum model for river confluence calculation. The river flow propagation time shown in Table 2 is obtained through optimization. The rainfall data in the verification data set is used as the model input, and the basin runoff model is used to calculate the outlet runoff process of each sub-basin in the forecast basin. Then, the calibrated river flow propagation time is brought into the river confluence model, and the river confluence calculation is performed according to the basin topology structure to obtain the forecast runoff process at the basin outlet. As shown in the attached figure. Figure 3 As shown, the runoff forecast results for the rainfall from September 15, 2021 to October 2, 2021 are basically consistent with the measured flow process lines. The NSE coefficient of the Luojiang River Basin is 0.92, the NSCE coefficient of the Sanhui River Basin is 0.93, and the NSE coefficient of the Luoduxi River Basin is 0.96. The simulated runoff results basically meet the forecast accuracy requirements, and the calibrated river water propagation time has been verified.

[0092] Table 1 Dataset division and flood element statistics

[0093]

[0094] Table 2 Statistics of confluence time parameters in the Qujiang River Basin

[0095]

[0096] Step 3: Bring the calibrated river flow propagation time into the river confluence model, and use the basin runoff generation and confluence model and the river confluence model to predict the basin runoff.

[0097] First, attach Figure 4 The forecast rainfall time series data shown is used as the model input, and the basin runoff model is used to calculate the runoff process at the outlet of each sub-basin in the forecast basin. Then, the calibrated river flow propagation time is brought into the river confluence model, and the river confluence calculation is performed according to the basin topology to obtain the forecast runoff process at the outlet of the forecast basin. The forecast results show that the runoff forecast results for the rainfall from August 29, 2022 to September 10, 2022 are basically consistent with the measured flow process line. The NSCE coefficient of the Luojiang Basin is 0.95, the NSCE coefficient of the Sanhui Basin is 0.88, and the NSE coefficient of the Luoduxi Basin is 0.81. The forecast results basically meet the accuracy requirements.

[0098] The above-mentioned embodiments are only preferred technical solutions of the present invention and should not be regarded as limitations of the present invention. The embodiments and features in the embodiments of the present application can be arbitrarily combined with each other without conflict. The protection scope of the present invention shall be the technical solutions recorded in the claims, including the equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A basin runoff forecasting method based on dynamic calibration of river flow propagation time, characterized by: It includes the following steps: S1. Acquire and process basin meteorological and hydrological data, and select rainfall-runoff data for dynamic calibration of river flow propagation time; S2, dividing the rainfall runoff data into a calibration data set and a validation data set, and using an optimization algorithm to calibrate and validate the river flow propagation time; S3, bring the calibrated river flow propagation time into the river confluence model, and use the basin runoff generation and confluence model and the river confluence model to predict the basin runoff; The step S2 comprises the following steps: S21. Randomly select and construct calibration and validation datasets from rainfall-runoff data. S22. Calibrate and verify the river water flow propagation time using the optimization algorithm, specifically: determine the optimization variables, constraints, and objective functions involved in calibrating the river water flow propagation time, then use the optimization algorithm SCE-UA to solve and obtain the optimal river water flow propagation time, and verify the optimized river water flow propagation time; The optimization variable in step S22 is the forecast basin The flow propagation time of a river channel is expressed as: ； The constraints are: ； In the formula, is the minimum flow propagation time in the river channel, is the maximum water flow propagation time in the river; and It can be obtained by measuring in dry season and flood season respectively, or it can be preliminarily determined by artificial experience for the minimum and maximum water flow propagation time of a river channel, and then dynamically adjusted according to the specific situation during calibration; The Nash-Sutcliffe efficiency coefficient is selected as the objective function, and its expression is: ； In the formula, is the measured runoff of the forecast section of the basin (unit, m³ / s), Simulated runoff for the forecast section of the basin (unit, m³ / s), is the average measured runoff of the forecast section of the basin (unit, m³ / s); the simulated runoff of the forecast section of the basin It is calculated through the basin runoff model and the river runoff model. The input of the basin runoff model is the rainfall data in the calibration data set. The parameters of the basin runoff model are determined by combining the hydrological model parameter calibration and artificial experience. The verification of the optimized river flow propagation time in step S22 is specifically as follows: ①Using the rainfall data in the validation data set as the model input, the runoff process of each sub-basin outlet in the basin is calculated and predicted using the basin runoff model; ② The calibrated river flow propagation time is brought into the river confluence model, and the river confluence calculation is performed according to the basin topology structure to obtain the forecast runoff process at the forecast basin outlet. The measured runoff process at the forecast basin outlet corresponding to the rainfall data in the validation data set can be calculated. The value is used to verify whether the calibrated river flow propagation time meets the accuracy requirements. Value greater than 0.7; The step S22 comprises the following steps: S221. Dynamic calibration of river flow propagation time is a mathematical optimization problem. Assume that the dimension of this optimization problem is , The number is equal to the number of rivers in the forecast basin. and The propagation time of river water flow is calculated by generating random numbers within a limited range. Initialize and select the number of complex types participating in the evolution and the number of vertices contained in each complex , calculate the number of sample points ; S222, distributed in feasible space with uniform probability Randomly generate sample points indivual ; S223, calculate the target value of each point And arrange them from small to large, storage space ,in For the best, is the worst; S224, Division The composite type is , each composite type includes points, that is ; S225, perform CCE evolution on each composite type; S226, all points of each composite type after evolution form a new set and sort them again according to the target value from small to large. The storage space after sorting Also sorted in ascending order of target value; S227, stop evolution judgment, if the evolution number limit or the target value convergence range limit condition is met, then stop evolution, otherwise return to step S224 to continue evolution.

2. A method for basin runoff forecasting based on dynamic calibration of river flow propagation time according to claim 1, characterized in that: The step S1 comprises the following steps: S11. According to the meteorological and hydrological data of the watershed, select rainfall runoff data that meets the calculation requirements of the watershed runoff model and the river confluence model; S12. Obtain the forecast rainfall time series data of the basin in the next 1 to 10 days through precipitation forecast products.

3. A method for basin runoff forecasting based on dynamic calibration of river flow propagation time according to claim 1, characterized in that: The forecast basin is composed of multiple sub-basins, each sub-basin is connected by a river channel, and the sub-basin outlet flow process is calculated by a basin runoff generation and confluence model. The topological structure refers to the upstream and downstream position relationship of each sub-basin and river channel in the forecast basin.

4. A method for basin runoff forecasting based on dynamic calibration of river flow propagation time according to claim 1, characterized in that: The step S225 comprises the following steps: S2251, give Assign weight ; S2252, select the parent; from Choose by probability Individuals ( , , ,..., ) is stored in middle, ,in, yes The function value of The location is stored in inside; S2253, produce offspring; including Calculate the centroid , in order from smallest to largest and To sort; to reflect on individuals , mutation, shrinkage or mutation steps and repeat the process.

5. The method for basin runoff forecasting based on dynamic calibration of river flow propagation time according to claim 1, characterized in that: The step S3 comprises the steps of: S31, using the forecast rainfall time series data obtained in step S12 as model input, and using the basin runoff generation and confluence model to calculate the runoff process at the outlet of each sub-basin in the forecast basin; S32. Bring the calibrated river flow propagation time into the river confluence model, perform river confluence calculation according to the basin topology structure, and obtain the predicted runoff process at the basin outlet.

Images