Medium and long term runoff forecasting method coupled with weather forecasting product and related device

Abstract

The invention discloses a medium-and-long-term runoff forecasting method coupled with a weather forecasting product and a related device, and relates to the technical field of runoff forecasting, the method comprises the following steps: firstly, acquiring runoff data, weather data and climate index data of a plurality of time periods before a to-be-predicted time period and rainfall correction forecasting data of the to-be-predicted time period; taking as an input data set; after preprocessing and redundant variable screening are carried out, an obtained input data set is input into an optimal runoff forecasting model which is determined through training and comparison, and runoff forecasting data of a to-be-predicted time period are obtained; compared with an existing physical hydrological model, runoff prediction is carried out by coupling rainfall correction forecast data, complex hydrological physical calculation is avoided, and the adaptability and generalization ability of the model are improved; compared with an existing data driving method, the method has the advantages that medium-and-long-term runoff forecasting is carried out by coupling rainfall correction forecasting data, and the defect that the influence of future weather on medium-and-long-term runoff is not fully considered in an existing method is overcome.

Description

Technical Field

[0001] The present application relates to the technical field of runoff forecasting, and in particular to a medium- and long-term runoff forecasting method and related devices coupled with meteorological forecast products. Background Art

[0002] Medium- and long-term runoff forecasts are a key tool for water resources management, playing a vital role in resolving the imbalance between water supply and demand, and in efficiently developing and optimizing water resources allocation. First, they provide valuable long-term reference information for the operation and management of water conservancy projects, helping to alleviate the conflict between water storage and water abandonment and improve the overall utilization efficiency of water resources. Second, accurate medium- and long-term forecasts provide a scientific basis for optimizing reservoir operation, helping water conservancy departments to rationally allocate water, achieve a balance between supply and demand, and maximize the economic and social benefits of reservoirs. Medium- and long-term runoff forecasts are an indispensable foundation for formulating reservoir water supply plans, basin water resources scheduling schemes, and promoting modern water conservancy construction. Therefore, strengthening medium- and long-term runoff forecasts and improving their accuracy are key to achieving optimal water resource allocation and improving water resources management.

[0003] Currently, two main approaches are used for medium- and long-term runoff forecasting: process-driven and data-driven. Process-driven approaches describe the basin's hydrological cycle through physical equations, offering strong scientific explanatory power. However, the modeling process is complex, computationally expensive, and limited in its ability to address the nonstationary nature of future climate change. Data-driven approaches can effectively exploit nonlinear relationships between data and improve forecast accuracy. However, they generally rely on historical data for training, fail to fully utilize future climate forecast information, and are therefore inadequately adaptable to long-term runoff trends. Therefore, there is an urgent need for a forecasting method that can integrate meteorological forecast products to rapidly and accurately predict monthly reservoir inflows. Summary of the Invention

[0004] The purpose of this application is to provide a medium- and long-term runoff forecasting method and related devices coupled with meteorological forecast products, which can be combined with meteorological forecast products to achieve rapid and accurate forecasting of monthly reservoir inflow data flow.

[0005] To achieve the above objectives, this application provides the following solutions:

[0006] In a first aspect, the present application provides a medium- and long-term runoff forecasting method coupled with meteorological forecast products, comprising:

[0007] The runoff data, meteorological data and climate index data of several time periods before the time period to be predicted and the precipitation correction forecast data of the time period to be predicted are obtained to obtain an input data set; the runoff data, meteorological data and climate index data all include several input variables.

[0008] The meteorological data and precipitation correction forecast data are downscaled, and the runoff data and climate index data are interpolated and completed to obtain the preprocessed input dataset.

[0009] The LASSO algorithm is used to screen the input variables in the preprocessed input data set, remove redundant input variables, and obtain the screened input data set.

[0010] The screened input data set is input into the optimal runoff forecast model to obtain the medium- and long-term runoff forecast data for the forecast period; the optimal runoff forecast model is the runoff forecast model with the highest prediction accuracy among several runoff forecast models obtained by training the initial runoff forecast model using different training sample sets; the input variables included in the samples in different training sample sets are not exactly the same; the input variables included in the samples in the training sample set for training the optimal runoff forecast model are exactly the same as the input variables in the screened input data set.

[0011] Optionally, the method further comprises the following steps:

[0012] A training data set is constructed based on historical observation data; the training data set includes precipitation-corrected forecast data and runoff data for the target period, as well as runoff data, meteorological data, and climate index data for several periods before the target period; among them, the runoff data, meteorological data, and climate index data for several periods before the target period, as well as the precipitation-corrected forecast data for the target period are the input data of the model, and the runoff data for the target period are the target output data of the model; the target period is any historical period with known runoff data.

[0013] The meteorological data and precipitation correction forecast data are downscaled, and the runoff data of the target period, the runoff data of several periods before the target period, and the climate index data are interpolated and completed to obtain the preprocessed training data set.

[0014] The LASSO algorithm is used to screen the input variables in the preprocessed training data set, remove redundant input variables, and obtain the screened training data set.

[0015] According to different combinations of input data in the screened training data set, several input data combination methods are determined.

[0016] For any combination of input data, the filtered training data set is used to construct the training sample set and the validation sample set.

[0017] The initial runoff forecast model is trained using the training sample set to obtain a trained runoff forecast model.

[0018] The trained runoff forecast model is verified using the validation sample set to determine the prediction accuracy of the runoff forecast model.

[0019] Several runoff forecast models are arranged in descending order of prediction accuracy, and the runoff forecast model with the highest prediction accuracy is taken as the optimal runoff forecast model.

[0020] Optionally, the runoff data includes the reservoir inflow runoff; the meteorological data includes the precipitation, minimum temperature and maximum temperature in the catchment area upstream of the reservoir; the climate index data includes the atmospheric circulation index and the sea temperature index; the precipitation correction forecast data is the data obtained by averaging and fusing several precipitation forecast products through the Bayesian model.

[0021] Optionally, the initial runoff forecast model is an XGBoost machine learning model. When the initial runoff forecast model is trained using the training sample set, the hyperparameters of the initial runoff forecast model are automatically optimized through the Bayesian optimization algorithm; the hyperparameters of the initial runoff forecast model include the number of trees, learning rate, maximum tree depth, minimum child node weight, L1 regularization term weight and L2 regularization term weight.

[0022] Optionally, after obtaining the filtered training data set, the method further includes: normalizing the data in the filtered training data set according to the following formula:

[0023]

[0024] Among them, x t ' is the value of the normalized data x at time t, x t is the value of data x at time t before normalization, a is the lower limit of data x, and b is the upper limit of data x.

[0025] After obtaining the medium- and long-term runoff forecast data for the period to be predicted, the method further includes: performing a denormalization process on the medium- and long-term runoff forecast data for the period to be predicted according to the following formula:

[0026] x t =x t '(ba)+a.

[0027] Optionally, bilinear interpolation is used to downscale meteorological data and precipitation correction forecast data, and linear interpolation is used to interpolate and complete runoff data and climate index data.

[0028] In a second aspect, the present application provides a medium- and long-term runoff forecast system coupled with meteorological forecast products, comprising:

[0029] The input data set acquisition module is used to obtain runoff data, meteorological data and climate index data for several time periods before the forecast period and precipitation correction forecast data for the forecast period to obtain the input data set; the runoff data, meteorological data and climate index data all include several input variables.

[0030] The input data preprocessing module is used to downscale the meteorological data and precipitation correction forecast data, and to interpolate and complete the runoff data and climate index data to obtain the preprocessed input data set.

[0031] The redundant variable screening module is used to screen the input variables in the preprocessed input data set using the LASSO algorithm, remove redundant input variables, and obtain a screened input data set.

[0032] The medium- and long-term runoff forecast module is used to input the screened input data set into the optimal runoff forecast model to obtain the medium- and long-term runoff forecast data for the forecast period; the optimal runoff forecast model is the runoff forecast model with the highest prediction accuracy among several runoff forecast models obtained by training the initial runoff forecast model using different training sample sets; the input variables included in the samples in different training sample sets are not exactly the same; the input variables included in the samples in the training sample set for training the optimal runoff forecast model are exactly the same as the input variables in the screened input data set.

[0033] In a third aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the medium- and long-term runoff forecasting method for coupled meteorological forecast products as described above.

[0034] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the medium- and long-term runoff forecasting method of coupled meteorological forecast products as described above.

[0035] In a fifth aspect, the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of the medium- and long-term runoff forecasting method of coupling meteorological forecast products as described above.

[0036] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0037] The present application provides a medium- and long-term runoff forecasting method and related devices coupled with meteorological forecast products. In the method, runoff data, meteorological data and climate index data for several time periods before the time period to be predicted, as well as precipitation correction forecast data for the time period to be predicted, are first obtained as input data sets; then, after preprocessing and redundant variable screening, the screened input data sets are input into an optimal runoff forecast model determined through training and comparison to obtain medium- and long-term runoff forecast data for the time period to be predicted; compared with existing physical hydrological models, the present application couples meteorological forecast products to perform runoff forecasting, avoids complex hydrological and physical calculations, and improves the adaptability and generalization ability of the model; compared with existing data-driven methods, the present application couples precipitation correction forecast data to perform medium- and long-term runoff forecasting, overcoming the problem that existing methods do not fully consider the impact of future meteorology on medium- and long-term runoff. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0039] Figure 1 A flowchart of a medium- and long-term runoff forecasting method for coupling meteorological forecast products provided in one embodiment of the present application.

[0040] Figure 2 A flowchart of training a model and selecting the optimal runoff forecast model in a medium- and long-term runoff forecast method for coupling meteorological forecast products provided in one embodiment of the present application.

[0041] Figure 3 A schematic diagram of the functional modules of a medium- and long-term runoff forecast system coupled with a meteorological forecast product provided in one embodiment of the present application.

[0042] Figure 4 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0044] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0045] The embodiment of the present application provides a method for medium- and long-term runoff forecasting coupled with a meteorological forecast product. In an exemplary embodiment, Figure 1 As shown, the following steps are included:

[0046] A1. Obtain runoff data, meteorological data, and climate index data for several time periods prior to the forecast period, as well as corrected precipitation forecast data for the forecast period, to form the input dataset. Runoff data, meteorological data, and climate index data each include several input variables. Specifically, runoff data includes reservoir inflow; meteorological data includes precipitation, minimum and maximum temperatures in the reservoir's upstream catchment area; climate index data includes the atmospheric circulation index and sea surface temperature index; and corrected precipitation forecast data is a Bayesian averaging fusion of several precipitation forecast products.

[0047] A2. Downscale the meteorological data and the corrected precipitation forecast data, and interpolate and complete the runoff data and climate index data to obtain the preprocessed input dataset. Specifically, bilinear interpolation is used to downscale the meteorological data and the corrected precipitation forecast data, and linear interpolation is used to complete the runoff data and climate index data. It is understood that if there are no missing values in the runoff data, interpolation and completion of the runoff data can be omitted.

[0048] A3. Use the LASSO algorithm to filter the input variables in the preprocessed input data set, remove redundant input variables, and obtain a filtered input data set. Before the next step, the Min-Max normalization method can be used to normalize all input variables to reduce the scale effect between different data and improve the training convergence speed. Specifically, after obtaining the filtered input data set, the data in the filtered input data set is normalized according to the following formula:

[0049]

[0050] Among them, x t ' is the value of the normalized data x at time t, x t is the value of data x at time t before normalization, a is the lower limit of data x, and b is the upper limit of data x.

[0051] A4. Input the filtered input data set into the optimal runoff forecast model to obtain medium- and long-term runoff forecast data for the forecast period; the optimal runoff forecast model is the runoff forecast model with the highest prediction accuracy among several runoff forecast models obtained by training the initial runoff forecast model using different training sample sets; the input variables included in the samples in different training sample sets are not exactly the same; the input variables included in the samples in the training sample set used to train the optimal runoff forecast model are exactly the same as the input variables in the filtered input data set.

[0052] Correspondingly, after obtaining the medium- and long-term runoff forecast data for the forecast period, the method further includes: performing denormalization processing on the medium- and long-term runoff forecast data for the forecast period according to the following formula:

[0053] x t =x t '(ba)+a.

[0054] Among them, x t is the value of the data x at time t after denormalization, x t ' is the value of data x at time t before denormalization, a is the lower limit of data x, and b is the upper limit of data x.

[0055] Specifically, before using the optimal runoff forecast model for medium- and long-term runoff forecasting in step A4, it is obviously necessary to train the model and select the optimal runoff forecast model, such as Figure 2 As shown, the process includes the following steps:

[0056] B1. Construct a training dataset based on historical observation data. The training dataset includes the precipitation-corrected forecast data and runoff data for the target period, as well as runoff data, meteorological data, and climate index data for several periods before the target period. The runoff data, meteorological data, and climate index data for several periods before the target period, as well as the precipitation-corrected forecast data for the target period, serve as the model's input data, and the runoff data for the target period serve as the model's target output data. The target period is any historical period for which runoff data is known.

[0057] Specifically, in this embodiment, the time lag effect of past runoff data, meteorological data, and climate index data on future monthly runoff predictions is taken into account. To ensure that the model can fully utilize historical information and improve prediction accuracy, the several time periods before the target period are limited to the past 18 months before the target period.

[0058] Runoff data for the target period and previous periods represent inflows to the reservoir, derived from hydrological monitoring station records. Meteorological data, including precipitation, minimum and maximum temperatures in the reservoir's upstream catchment area, are sourced from the CN05.1 gridded observation dataset, with a resolution of 0.25° × 0.25°. Bilinear interpolation is then used to regrid the data to match the spatial scale of the reservoir catchment area. Climate index data, including the atmospheric circulation index and the sea surface temperature index, are sourced from the National Climate Center of the China Meteorological Administration. GCM precipitation correction forecast data are derived from the NMME dataset. Precipitation forecast products from five global climate models, CanSIPS-IC3, CCSM4, CFSv2, CM2p1-aer04, and CM2p5-FLOR-B01, are selected at a resolution of 1° and fused using the Bayesian model averaging (BMA) method to generate GCM precipitation correction forecast data.

[0059] B2. Downscale the meteorological data and the corrected precipitation forecast data. Interpolate and complete the runoff data for the target period, as well as runoff data for several periods preceding the target period and climate index data, to obtain the preprocessed training dataset. Specifically, bilinear interpolation is used to regrid the CN05.1 meteorological data and GCM data to match the spatial resolution of the study basin and the spatial scale of the study area. Linear interpolation is used to fill missing values in the runoff data and climate index data. It is understood that if there are no missing values in the runoff data, interpolation of the runoff data can be omitted.

[0060] B3. Use the LASSO algorithm to filter the input variables in the preprocessed training data set, remove redundant input variables, and obtain a filtered training data set. Use the LASSO (Least Absolute Shrinkage and Selection Operator) algorithm to filter the input variables in the data set, remove redundant variables, and improve model calculation efficiency and prediction accuracy.

[0061] B4. Based on the different combinations of input data in the screened training dataset, several input data combinations were determined. Specifically, runoff data, meteorological data, climate index data, and GCM precipitation correction forecast data were permuted and combined to construct multiple sample datasets. These four input data combinations were then used to analyze the impact of the presence or absence of GCM precipitation correction forecast data on prediction accuracy, as shown in Table 1.

[0062] Table 1 The impact of GCM data on prediction accuracy with and without precipitation correction forecast data

[0063]

[0064] B5. For any input data combination, use the filtered training dataset to construct a training sample set and a validation sample set. The ratio of the training sample set to the validation sample set is 3:1 to ensure the generalization ability of the model.

[0065] B6. Train the initial runoff forecast model using the training sample set to obtain a trained runoff forecast model. In this embodiment, the initial runoff forecast model is an XGBoost machine learning model. When the initial runoff forecast model is trained using the training sample set, the hyperparameters of the initial runoff forecast model are automatically optimized using a Bayesian optimization algorithm; the hyperparameters of the initial runoff forecast model include the number of trees, the learning rate, the maximum tree depth, the minimum child node weight, the L1 regularization term weight, and the L2 regularization term weight.

[0066] B7. Use the validation sample set to validate the trained runoff forecast model and determine the prediction accuracy of the runoff forecast model.

[0067] B8. Arrange several runoff forecast models in descending order of prediction accuracy, and select the runoff forecast model with the highest prediction accuracy as the optimal runoff forecast model.

[0068] The following describes a medium- and long-term runoff forecast method for a coupled meteorological forecast product, provided in this embodiment, using the Biliu River Reservoir as a research reservoir. The Biliu River's main stream is 156 kilometers long, 100 kilometers of which are located in Dalian. The total drainage area is 2,814 square kilometers, and the reservoir's catchment area is 2,085 square kilometers, accounting for 74.1% of the total drainage area. The medium- and long-term runoff forecast method for the Biliu River Reservoir provided in this example specifically includes the following steps:

[0069] Corresponding to step B1 of the previous solution, the historical observation data collected in this example covers the period from January 1, 1982, to December 31, 2020. The reservoir's pre-inflow runoff data are derived from hydrological monitoring station records. Monthly precipitation, maximum temperature, and minimum temperature data for the reservoir's upstream catchment area are available from the CN05.1 gridded observational dataset. Data for 111 indices, including the atmospheric circulation index and sea surface temperature index, are obtained from the National Climate Center of the China Meteorological Administration. GCM precipitation correction forecast data are derived from the NMME dataset.

[0070] Corresponding to step B2 of the previous solution, bilinear interpolation was used in this example to regrid the CN05.1 meteorological data and GCM data to match the spatial resolution of the study basin (adjusted to 0.1° for the Biliu River Reservoir), thus matching the spatial scale of the study area. Linear interpolation was used to fill missing values in the meteorological data, runoff data, and climate index data.

[0071] Corresponding to step B5 of the aforementioned scheme, the distribution dates of the data in the constructed training sample set are January 1, 1982 - December 31, 2010, and the distribution dates of the data in the verification sample set are January 1, 2011 - December 31, 2020.

[0072] In this embodiment, the forecast period is set to 1 month and 3 months respectively, so as to predict the monthly reservoir runoff data of the target months through the above scheme, that is, the medium- and long-term runoff forecast data of the test time period.

[0073] Corresponding to step B7 of the above scheme, the model calibration period is from January 1982 to December 2010, and the model validation period is from January 2011 to December 2020. The evaluation indicators used are Nash efficiency coefficient (NSE), Kling-Gupta efficiency (KGE), root mean square error (RMSE) and Pearson correlation coefficient (R). The specific evaluation indicators are shown in the following formula:

[0074]

[0075] Where, is the observation value at the tth time step; is the predicted value at the tth time step; is the mean of the observed values; n is the total number of observations; r is the Pearson correlation coefficient between the model simulation value and the observation value, which indicates the linear correlation between the predicted value and the observed value; α is the deviation coefficient between the predicted and observed values, which is used to measure the proportional relationship between the model output and the observed data. σ sim is the standard deviation of the forecast data, σ obs is the standard deviation of the observed data; β is the mean ratio between the predicted data and the observed data, μ sim is the mean of the forecast data, μ obs is the mean of the observed data.

[0076] When the forecast period is 1 month, the evaluation coefficients are shown in Table 2:

[0077] Table 2 Evaluation coefficients for different input data with a forecast period of 1 month

[0078]

[0079] When the forecast period is 3 months, the evaluation coefficients are shown in Table 3:

[0080] Table 3 Evaluation coefficients for different input data with a 3-month forecast period

[0081]

[0082] In summary, the four evaluation indicators for combination 1 in this example show consistent trends. However, the addition of GCM forecast data improves forecast accuracy for both the one-month and three-month forecast periods, with the improvement being particularly significant for the three-month forecast period, where the KGE index increases by 19%. A comprehensive analysis of the four combinations reveals that the inclusion of GCM forecast information significantly improves the forecast accuracy for both the one-month and three-month forecast periods for the Biliuhe Reservoir's monthly inflow. Furthermore, the forecast accuracy of the combinations varies significantly across different forecast periods: for the one-month forecast period, combination 4 (with GCM) performs best, with an NSE of 0.646; for the three-month forecast period, combination 1 (with GCM) achieves the highest NSE of 0.721. This demonstrates that coupling meteorological forecast products can achieve rapid and accurate forecasts of monthly reservoir inflow data while improving adaptability to long-term runoff trends.

[0083] Based on the same inventive concept, the embodiment of the present application also provides a system for implementing the above-mentioned method for coupling medium- and long-term runoff forecasting with meteorological forecast products. The solution provided by the system is similar to the solution described in the above-mentioned method. In an exemplary embodiment, Figure 3 As shown in the figure, a medium- and long-term runoff forecast system coupled with meteorological forecast products is provided, including the following functional modules:

[0084] The input data set acquisition module is used to obtain runoff data, meteorological data and climate index data for several time periods before the forecast period and precipitation correction forecast data for the forecast period to obtain the input data set; the runoff data, meteorological data and climate index data all include several input variables.

[0085] The input data preprocessing module is used to downscale the meteorological data and precipitation correction forecast data, and to interpolate and complete the runoff data and climate index data to obtain the preprocessed input data set.

[0086] The redundant variable screening module is used to screen the input variables in the preprocessed input data set using the LASSO algorithm, remove redundant input variables, and obtain a screened input data set.

[0087] The medium- and long-term runoff forecast module is used to input the screened input data set into the optimal runoff forecast model to obtain the medium- and long-term runoff forecast data for the forecast period; the optimal runoff forecast model is the runoff forecast model with the highest prediction accuracy among several runoff forecast models obtained by training the initial runoff forecast model using different training sample sets; the input variables included in the samples in different training sample sets are not exactly the same; the input variables included in the samples in the training sample set for training the optimal runoff forecast model are exactly the same as the input variables in the screened input data set.

[0088] certainly, Figure 3 The architecture shown is only exemplary and can be omitted according to actual needs when implementing different functions. Figure 3 One or at least two components of the system shown.

[0089] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 4 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a medium- and long-term runoff forecast method for coupling meteorological forecast products mentioned in the above embodiment can be implemented.

[0090] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0091] In an exemplary embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0092] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0093] In an exemplary embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0094] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0095] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0096] The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may include, but are not limited to, general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units based on quantum computing, and the like.

[0097] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0098] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A medium- and long-term runoff forecast method coupled with meteorological forecast products, characterized in that: include: Obtaining runoff data, meteorological data, and climate index data for a plurality of time periods prior to a time period to be predicted, as well as precipitation correction forecast data for the time period to be predicted, to obtain an input data set; the runoff data, the meteorological data, and the climate index data each include a plurality of input variables; Downscaling the meteorological data and the precipitation correction forecast data, and interpolating and completing the runoff data and the climate index data to obtain a preprocessed input data set; Using the LASSO algorithm to screen the input variables in the preprocessed input data set, removing redundant input variables, and obtaining a screened input data set; The screened input data set is input into the optimal runoff forecast model to obtain medium- and long-term runoff forecast data for the forecast period; the optimal runoff forecast model is the runoff forecast model with the highest prediction accuracy among several runoff forecast models obtained by training the initial runoff forecast model using different training sample sets; the input variables included in the samples in different training sample sets are not exactly the same; the input variables included in the samples in the training sample set for training the optimal runoff forecast model are exactly the same as the input variables in the screened input data set.

2. The medium- and long-term runoff forecasting method of coupled meteorological forecast products according to claim 1 is characterized in that: Also includes: Construct a training dataset based on historical observation data; The training data set includes precipitation-corrected forecast data and runoff data for a target period, as well as runoff data, meteorological data, and climate index data for several periods before the target period; wherein the runoff data, meteorological data, and climate index data for several periods before the target period, as well as the precipitation-corrected forecast data for the target period, are input data for the model, and the runoff data for the target period are target output data for the model; the target period is any historical period for which runoff data is known; Downscaling the meteorological data and the precipitation correction forecast data, and interpolating and completing the runoff data of the target period, the runoff data of several periods before the target period, and the climate index data to obtain a preprocessed training data set; The LASSO algorithm is used to filter the input variables in the preprocessed training data set, remove redundant input variables, and obtain a filtered training data set; Determining several input data combination methods based on different combinations of input data in the screened training data set; For any input data combination, the filtered training data set is used to construct a training sample set and a validation sample set; Using the training sample set to train the initial runoff forecast model to obtain a trained runoff forecast model; Using the validation sample set to validate the trained runoff forecast model to determine the prediction accuracy of the runoff forecast model; Several runoff forecast models are arranged in descending order of prediction accuracy, and the runoff forecast model with the highest prediction accuracy is taken as the optimal runoff forecast model.

3. The medium- and long-term runoff forecasting method of coupled meteorological forecast products according to claim 2 is characterized in that: The runoff data includes the reservoir inflow runoff; the meteorological data includes the precipitation, minimum temperature and maximum temperature in the upstream catchment area of the reservoir; the climate index data includes the atmospheric circulation index and the sea temperature index; the precipitation correction forecast data is the data obtained by averaging and fusing several precipitation forecast products through the Bayesian model.

4. The medium- and long-term runoff forecasting method for coupling meteorological forecast products according to claim 2 is characterized in that: The initial runoff forecast model is an XGBoost machine learning model. When the initial runoff forecast model is trained using the training sample set, the hyperparameters of the initial runoff forecast model are automatically optimized by the Bayesian optimization algorithm; the hyperparameters of the initial runoff forecast model include the number of trees, learning rate, maximum tree depth, minimum child node weight, L1 regularization term weight and L2 regularization term weight.

5. The medium- and long-term runoff forecasting method of coupled meteorological forecast products according to claim 2 is characterized in that: After obtaining the filtered training data set, the method further includes: performing normalization processing on the data in the filtered training data set according to the following formula: Among them, x t ' is the value of the normalized data x at time t, x t is the value of data x at time t before normalization, a is the lower limit of data x, and b is the upper limit of data x; After obtaining the medium- and long-term runoff forecast data for the period to be predicted, the method further includes: performing a denormalization process on the medium- and long-term runoff forecast data for the period to be predicted according to the following formula: x t =x t '(b-a)+a。 6. The medium- and long-term runoff forecasting method for coupling meteorological forecast products according to claim 1 is characterized in that: The meteorological data and the precipitation correction forecast data are downscaled using a bilinear interpolation method, and the runoff data and the climate index data are interpolated and completed using a linear interpolation method.

7. A medium- to long-term runoff forecast system coupled with meteorological forecast products, characterized in that: include: An input data set acquisition module is used to acquire runoff data, meteorological data, and climate index data for a number of time periods before the time period to be predicted, as well as precipitation correction forecast data for the time period to be predicted, to obtain an input data set; the runoff data, the meteorological data, and the climate index data each include a number of input variables; An input data preprocessing module is used to perform downscaling processing on the meteorological data and the precipitation correction forecast data, and to perform interpolation and completion on the runoff data and the climate index data to obtain a preprocessed input data set; A redundant variable screening module is used to screen the input variables in the preprocessed input data set using the LASSO algorithm, remove redundant input variables, and obtain a screened input data set; The medium- and long-term runoff forecast module is used to input the screened input data set into the optimal runoff forecast model to obtain medium- and long-term runoff forecast data for the forecast period; the optimal runoff forecast model is the runoff forecast model with the highest prediction accuracy among several runoff forecast models obtained by training the initial runoff forecast model using different training sample sets; the input variables included in the samples in different training sample sets are not exactly the same; the input variables included in the samples in the training sample set for training the optimal runoff forecast model are exactly the same as the input variables in the screened input data set.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the medium- and long-term runoff forecasting method for coupled meteorological forecast products according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the medium- and long-term runoff forecasting method of the coupled meteorological forecast product according to any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the medium- and long-term runoff forecasting method of the coupled meteorological forecast product according to any one of claims 1 to 6 is implemented.

Images