Numerical weather forecasting method and system based on river network convergence calculation

Abstract

The invention discloses a numerical weather forecasting method and system based on river network convergence calculation, and the method comprises the following steps: configuring an initial numerical weather forecasting model, carrying out the regional forecasting simulation based on the initial numerical weather forecasting model to obtain runoff depth data, and converting the runoff depth data into river inflow data; constructing a river network convergence calculation scheme, and carrying out river network convergence calculation on the initial numerical weather forecast model based on the river inflow data to obtain a plurality of flow simulation values of all river reaches in the river network; and comparing the flow simulation value with the historical observation value, selecting a configuration scheme with the closest two values to construct a final numerical weather forecast model, and performing weather forecast by using the numerical weather forecast model. According to the method, river flow observation data can be used in an area with complex topographic conditions and rich water energy resources, an objective basis is provided for optimization of a numerical weather forecasting model, and the accuracy of numerical weather forecasting, especially numerical weather forecasting in a mountainous area, is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of numerical weather forecasting, and in particular relates to a numerical weather forecasting method and system based on river network confluence calculation. Background Art

[0002] Mountainous areas are a key focus for numerical weather forecasting, with significant practical implications for flash flood prevention and hydropower generation. On the one hand, mountainous areas are characterized by steep terrain and concentrated precipitation, making them prone to flash floods. Accurate numerical weather forecasts can provide critical information for flash flood warnings, enabling relevant departments to take timely measures and minimize casualties and property losses. On the other hand, mountainous areas, with their concentrated water flows and steep terrain, offer abundant hydropower resources. Accurate numerical weather forecasts can improve the effectiveness and economy of reservoir operation and the safety of hydropower projects.

[0003] The accuracy of numerical weather prediction models is crucial for accurate weather forecasts, and sufficient observational data is required for model evaluation and optimization. However, due to the isolation of mountainous terrain, the spatial range that can be represented by observation stations is limited. Furthermore, the harsh mountainous environment and insufficient number of observation sites exacerbate the problem of limited and underrepresented data for evaluating and optimizing numerical weather prediction models in mountainous areas. River discharge observations offer broad spatial representation, reflecting the overall weather conditions within the catchment area upstream of the observation station, but have not yet been used to evaluate and optimize numerical weather prediction models. Summary of the Invention

[0004] The present invention aims to solve the deficiencies of the prior art and provides the following solutions:

[0005] A numerical weather forecast method based on river network confluence calculation includes the following steps:

[0006] configuring an initial numerical weather forecast model, performing a regional forecast simulation based on the initial numerical weather forecast model to obtain runoff depth data, and converting the runoff depth data into river inflow data;

[0007] Constructing a river network flow calculation scheme, and performing a river network flow calculation on the initial numerical weather forecast model based on the river inflow data to obtain a number of flow simulation values for all river sections in the river network;

[0008] The flow simulation value and the historical observation value are compared, and the configuration scheme with the two values closest to each other is selected to construct a final numerical weather prediction model, and the numerical weather prediction model is used to perform weather forecasting.

[0009] Preferably, the method for carrying out regional forecast simulation includes:

[0010] Setting a simulation area of the initial numerical weather forecast model, the simulation area encompassing the entire range of the catchment area upstream of the flow observation station, and dividing the simulation area into a plurality of model calculation grids;

[0011] Aiming at the key physical processes that affect the simulation and forecasting results, several physical process parameterization schemes are selected to form several model configuration schemes;

[0012] Collect global-scale atmospheric reanalysis data and forecast data, select several model configuration schemes, drive the initial numerical weather forecast model to carry out regional forecast simulation, and obtain the runoff depth data.

[0013] Preferably, the method for converting the runoff depth data into river inflow data includes:

[0014] Selecting vectorized river network geographic information data, wherein the river network geographic information data includes a river centerline and a watershed area;

[0015] Determine each of the model calculation grids intersecting with the catchment area, calculate the intersection area, and calculate the runoff volume input from each of the model calculation grids to the catchment area corresponding thereto based on the intersection area and the runoff depth data;

[0016] Each of the runoff amounts is added together to obtain the total runoff amount of the catchment area, and the total runoff amount is used as the river inflow data.

[0017] Preferably, the method for constructing the river network confluence calculation scheme includes:

[0018] According to the spatial location of the river flow observation station, determining the river section closest to the river flow observation station in the river network geographic information data as the river section where the station is located;

[0019] Search upstream and downstream along the river section where the station is located to determine the river section upstream of the station and the river section downstream of the station;

[0020] Arrange the river flow observation stations in order from upstream to downstream, and determine the river section between every two adjacent stations;

[0021] For all river sections in the river network geographic information data, the Muskingum method is used to perform flow routing:

[0022]

[0023] Among them, Q t Indicates the flow of the river section at the current calculation time step, Q t-1 represents the river flow at the previous time step, represents the upstream river flow at the current calculation time step, represents the upstream river flow at the previous time step, represents the river inflow data, Δt represents the time step, Δl represents the river section length, both the time step and the river section length are greater than 0, x represents the weight coefficient, the weight coefficient value is greater than or equal to 0 and less than or equal to 0.5, c represents the wave speed, the wave speed value is greater than 0 and less than half of the quotient of the river section length and the time step.

[0024] Preferably, the method for obtaining the plurality of flow simulation values includes:

[0025] Step 1: Select the river flow observation stations in order from upstream to downstream;

[0026] Step 2: Select a series of different wave velocities and weight coefficients, use the first flow simulation value of the river section above the upstream station and the corresponding river inflow data as input, perform confluence calculation on all river sections from the station to its upstream station, and obtain the second flow simulation value of the current station;

[0027] Step 3: Compare the second flow simulation value of the current station with the observed value, select the wave velocity and weight coefficient corresponding to the second flow simulation value closest to the observed value as the confluence parameter value of the river section between the current station and the upstream station, and use the corresponding second flow simulation value as the input of the downstream station;

[0028] Repeat steps 1 to 3 until all the river flow observation stations are used to obtain the flow simulation values on all river sections.

[0029] Preferably, the method for comparing the flow simulation value and the historical observation value includes:

[0030] Calculate the KGE indicator value:

[0031]

[0032] Among them, r represents the correlation coefficient between the flow simulation value and the historical observation value, σ represents the standard deviation of the flow simulation value, and σ o represents the standard deviation of historical observations, μ represents the average value of flow simulation values, and μ o represents the average of historical observations;

[0033] The KGE index values of each configuration scheme are compared, the configuration scheme corresponding to the KGE index value closest to 1 is selected, and the final numerical weather forecast model is constructed based on the selected configuration scheme.

[0034] The present invention also provides a numerical weather forecast system based on river network confluence calculation, wherein the system applies any of the above methods and comprises: a model configuration module, a confluence calculation module and a forecast module;

[0035] The model configuration module is used to configure an initial numerical weather forecast model, conduct regional forecast simulation based on the initial numerical weather forecast model to obtain runoff depth data, and convert the runoff depth data into river inflow data;

[0036] The confluence calculation module is used to construct a river network confluence calculation scheme, and perform river network confluence calculation on the initial numerical weather forecast model based on the river inflow data to obtain a number of flow simulation values for all river sections in the river network;

[0037] The forecast module is used to compare the flow simulation value and the historical observation value, select the configuration scheme with the two values closest to build a final numerical weather forecast model, and use the numerical weather forecast model to perform weather forecasting.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] The present invention utilizes river flow observation data that can represent the overall weather conditions of the upstream catchment area to evaluate and optimize the numerical weather forecast model. In areas with complex terrain conditions but rich water resources, the river flow observation data can be used to provide an objective basis for the optimization of the numerical weather forecast model, thereby improving the accuracy of numerical weather forecasts, especially numerical weather forecasts in mountainous areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0041] Figure 1 Schematic diagram of a method flow in an embodiment of the present invention;

[0042] Figure 2 This is a flowchart of a method according to an embodiment of the present invention. DETAILED DESCRIPTION

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

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

[0045] Example 1

[0046] In this embodiment, if Figure 1 、 Figure 2 As shown in FIG, a numerical weather forecast method based on river network confluence routing includes the following steps:

[0047] S1. Initial numerical weather forecast model. Based on the initial numerical weather forecast model, regional forecast simulation is carried out to obtain runoff depth data, and the runoff depth data is converted into river inflow data.

[0048] The method for carrying out regional forecast simulation includes: setting the simulation area of the initial numerical weather forecast model, the simulation area includes the entire range of the catchment area upstream of the flow observation station, and dividing the simulation area into several model calculation grids; for the key physical processes that affect the simulation forecast effect, selecting several physical process parameterization scheme combinations to form several model configuration schemes; collecting global-scale atmospheric reanalysis data and forecast data, selecting several model configuration schemes, driving the initial numerical weather forecast model to carry out regional forecast simulation, and obtaining runoff depth data.

[0049] The method for converting runoff depth data into river inflow data includes: selecting vectorized river network geographic information data, which includes the river centerline and the catchment area; determining each model calculation grid that intersects with the catchment area, and calculating the intersection area, based on the intersection area and the runoff depth data, calculating the runoff of each model calculation grid input to the corresponding catchment area; adding each runoff to obtain the total runoff of the catchment area, and using the total runoff as the river inflow data.

[0050] In this example, a numerical weather prediction model simulation region is first set up, encompassing the entire catchment area upstream of the flow observation station. The simulation region is then divided into model computational grids. Next, several physical process parameterization schemes are selected for key physical processes that influence the simulation and forecast, such as cloud microphysics, the planetary boundary layer, radiation transfer, and land surface processes, to form several model configurations. Global-scale atmospheric reanalysis data or forecast data are then collected to drive the numerical weather prediction model, conduct regional forecast simulations, and obtain runoff depth data. Next, vectorized river network geographic information data with river centerlines and catchment boundaries are selected, and the rivers and catchments correspond one to one. For each catchment in the river network, the model calculation grid of the numerical weather forecast model that intersects with it is determined, the intersection area is calculated, and the runoff input to the catchment by the corresponding grid is obtained by multiplying the intersection area by the runoff depth output by the numerical weather forecast model; finally, the runoff input from all the numerical weather forecast model grids that intersect with the catchment is added together to obtain the total runoff of the catchment, which is used as the river inflow data for the corresponding river section of the catchment.

[0051] S2. River network flow calculation scheme, and carry out river network flow calculation on the initial numerical weather forecast model based on river inflow data to obtain several flow simulation values for all river sections in the river network.

[0052] The method for constructing a river network confluence calculation scheme includes: according to the spatial location of a river flow observation station, determining the river section closest to the river flow observation station in the river network geographic information data as the river section where the station is located; searching upstream and downstream along the river section where the station is located to determine the river section upstream of the station and the river section downstream of the station; arranging the river flow observation stations in order from upstream to downstream, and determining the river section between every two adjacent stations. In this embodiment, the river sections within the station interval have the same confluence parameters, and different intervals have different confluence parameters, where the confluence parameters include wave velocity c and weight coefficient x; for all river sections in the river network geographic information data, using the Muskingum method to perform confluence calculation. In this embodiment, the calculation formula of the Muskingum confluence method is:

[0053]

[0054] Among them, Q t Indicates the flow of the river section at the current calculation time step, Q t-1 represents the river flow at the previous time step, represents the upstream river flow at the current calculation time step, represents the upstream river flow at the previous time step, represents the river inflow data, Δt represents the time step, x represents the weight coefficient, Δl represents the river section length, and c represents the wave velocity. In this embodiment, the time step and the river section length are both greater than 0, the weight coefficient x is greater than or equal to 0 and less than or equal to 0.5, and the wave velocity c is greater than 0 and less than half the quotient of the river section length and the time step.

[0055] The method for obtaining a plurality of flow simulation values includes: step 1, selecting river flow observation stations in order from upstream to downstream; step 2, selecting a series of different convergence parameter values (wave velocity and weight coefficient), using the flow simulation value of the river section above the upstream station and the corresponding river inflow data as input, performing convergence calculation on all river sections between the station and its upstream station, and obtaining the flow simulation value of the current station; step 3, comparing the flow simulation value of the current station with the observed value, selecting the wave velocity and weight coefficient corresponding to the flow simulation value closest to the observed value as the convergence parameter value of the river section between the current station and the upstream station, and using the corresponding flow simulation value as the input of the downstream station; in this embodiment, it also includes step 4, selecting an evaluation index such as a correlation coefficient, comparing the flow simulation value of the current station with the observation station, selecting the convergence parameter value when the convergence parameter value is closest to the observation value as the convergence parameter value of the river section between the current station and the upstream station, and using the corresponding river section flow simulation value as the input of the downstream station; repeating steps 1 to 4 until all river flow observation stations are used, and obtaining the flow simulation values on all river sections.

[0056] S3. The flow simulation value and the historical observation value are selected, and the configuration scheme with the two values closest to each other is selected to construct a final numerical weather prediction model, and the numerical weather prediction model is used to perform weather forecasting.

[0057] The method for comparing the flow simulation value and the historical observation value includes: selecting the correlation coefficient and the KGE index value as the comparison parameters, and calculating the KGE index value:

[0058]

[0059] Among them, r represents the correlation coefficient between the flow simulation value and the historical observation value, σ represents the standard deviation of the flow simulation value, and σ o represents the standard deviation of historical observations, μ represents the average value of flow simulation values, and μ o Represents the average value of historical observations; compare the KGE index values of each configuration scheme, select the configuration scheme corresponding to the KGE index value closest to 1, and build the final numerical weather forecast model based on the selected configuration scheme.

[0060] The constructed numerical weather forecast model is then used to make weather forecasts for mountainous areas.

[0061] Example 2

[0062] In this embodiment, a numerical weather forecasting system based on river network runoff calculation includes: a model configuration module, a runoff calculation module, and a forecasting module.

[0063] The model configuration module is used to configure the initial numerical weather forecast model, conduct regional forecast simulation based on the initial numerical weather forecast model to obtain runoff depth data, and convert the runoff depth data into river inflow data.

[0064] In the model configuration module, the simulation area of the initial numerical weather forecast model is set. The simulation area includes the entire range of the catchment area upstream of the flow observation station and is divided into several model calculation grids. For the key physical processes that affect the simulation and forecast results, several physical process parameterization scheme combinations are selected to form several model configuration schemes. Global-scale atmospheric reanalysis data and forecast data are collected, and several model configuration schemes are selected to drive the initial numerical weather forecast model to carry out regional forecast simulations and obtain runoff depth data. Vectorized river network geographic information data is selected, including river centerlines and catchment areas. Each model calculation grid that intersects the catchment area is determined and the intersection area is calculated. Based on the intersection area and runoff depth data, the runoff input to the corresponding catchment area of each model calculation grid is calculated. Each runoff volume is added together to obtain the total runoff volume of the catchment area, and the total runoff volume is used as the river inflow data.

[0065] In this embodiment, the model configuration module first sets the numerical weather forecast model simulation area, which includes the entire range of the catchment area upstream of the flow observation station, and then divides the simulation area into model calculation grids. Then, for key physical processes that affect the simulation and forecast effect, such as cloud microphysics, planetary boundary layer, radiation transfer, and land surface processes, several physical process parameterization scheme combinations are selected to form several model configuration schemes. Then, global-scale atmospheric reanalysis data or forecast data are collected to drive the numerical weather forecast model, carry out regional forecast simulation, and obtain runoff depth data. Next, vectorized river network geographic information data with river centerlines and catchment area boundaries is selected, and the river channels correspond one-to-one with the catchment areas. For each catchment area in the river network, the model calculation grid of the numerical weather forecast model that intersects with it is determined, the intersection area is calculated, and the runoff input to the catchment by the corresponding grid is obtained by multiplying the intersection area by the runoff depth output by the numerical weather forecast model. Finally, the runoff input from all numerical weather forecast model grids intersecting the catchment is added together to obtain the total runoff of the catchment, which is used as the river inflow data for the corresponding river section of the catchment.

[0066] The confluence calculation module is used to construct a river network confluence calculation scheme and perform river network confluence calculation on the initial numerical weather forecast model based on river inflow data to obtain several flow simulation values for all river sections in the river network.

[0067] In the confluence calculation module, based on the spatial location of the river flow observation station, the river section closest to the river flow observation station is determined in the river network geographic information data as the river section where the station is located. A search is performed upstream and downstream along the river section where the station is located to determine the river section upstream of the station and the river section downstream of the station. The river flow observation stations are arranged in order from upstream to downstream, and the river section between every two adjacent stations is determined. In this embodiment, the river sections within the station interval have the same confluence parameters, while different intervals have different confluence parameters, where the confluence parameters include wave velocity c and weight coefficient x. For all river sections in the river network geographic information data, the Muskingum method is used to perform confluence calculation. In this embodiment, the calculation formula of the Muskingum confluence method is:

[0068]

[0069] Among them, Q t Indicates the flow of the river section at the current calculation time step, Q t-1 represents the river flow at the previous time step, represents the upstream river flow at the current calculation time step, represents the upstream river flow at the previous time step, represents the river inflow data, Δt represents the time step, x represents the weight coefficient, Δl represents the river section length, and c represents the wave velocity. In this embodiment, the time step and the river section length are both greater than 0, the weight coefficient x is greater than or equal to 0 and less than or equal to 0.5, and the wave velocity c is greater than 0 and less than half the quotient of the river section length and the time step.

[0070] The process of obtaining a plurality of flow simulation values includes: step 1, selecting river flow observation stations in order from upstream to downstream; step 2, selecting a series of different convergence parameter values (wave velocity and weight coefficient), using the flow simulation value of the river section above the upstream station and the corresponding river inflow data as input, performing convergence calculation on all river sections between the station and its upstream station, and obtaining the flow simulation value of the current station; step 3, comparing the flow simulation value of the current station with the observed value, selecting the wave velocity and weight coefficient corresponding to the flow simulation value closest to the observed value as the convergence parameter value of the river section between the current station and the upstream station, and using the corresponding flow simulation value as the input of the downstream station; in this embodiment, step 4 is further included, selecting an evaluation index such as a correlation coefficient, comparing the flow simulation value of the current station with the observation station, selecting the convergence parameter value when the convergence parameter value is closest to the observation value as the convergence parameter value of the river section between the current station and the upstream station, and using the corresponding river section flow simulation value as the input of the downstream station; repeating steps 1 to 4 until all river flow observation stations are used, and the flow simulation values on all river sections are obtained.

[0071] The forecast module is used to compare the flow simulation value and the historical observation value, select the configuration scheme with the two values closest to build the final numerical weather forecast model, and use the numerical weather forecast model to make weather forecasts.

[0072] In the forecast module, select the correlation coefficient and KGE index value as the comparison parameters and calculate the KGE index value:

[0073]

[0074] Among them, r represents the correlation coefficient between the flow simulation value and the historical observation value, σ represents the standard deviation of the flow simulation value, and σ o represents the standard deviation of historical observations, μ represents the average value of flow simulation values, and μ o Represents the average value of historical observations; compare the KGE index values of each configuration scheme, select the configuration scheme corresponding to the KGE index value closest to 1, and build the final numerical weather forecast model based on the selected configuration scheme.

[0075] The constructed numerical weather forecast model is then used to make weather forecasts for mountainous areas.

[0076] The embodiments described above are only descriptions of the preferred embodiments of the present invention and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made by ordinary technicians in this field to the technical solution of the present invention, including the selection of evaluation indicators such as correlation coefficient, KGE, Nash efficiency coefficient, and root mean square error, should all fall within the scope of protection determined by the claims of the present invention.

Claims

1. A numerical weather forecast method based on river network confluence calculation, characterized in that: The following steps are involved: configuring an initial numerical weather forecast model, performing a regional forecast simulation based on the initial numerical weather forecast model to obtain runoff depth data, and converting the runoff depth data into river inflow data; Constructing a river network flow calculation scheme, and performing a river network flow calculation on the initial numerical weather forecast model based on the river inflow data to obtain a number of flow simulation values for all river sections in the river network; The flow simulation value and the historical observation value are compared, and the configuration scheme with the two values closest to each other is selected to construct a final numerical weather prediction model, and the numerical weather prediction model is used to perform weather forecasting.

2. The numerical weather forecasting method based on river network confluence calculation according to claim 1, characterized in that: The method for carrying out regional forecast simulation comprises: Setting a simulation area of the initial numerical weather forecast model, the simulation area encompassing the entire range of the catchment area upstream of the flow observation station, and dividing the simulation area into a plurality of model calculation grids; Aiming at the key physical processes that affect the simulation and forecasting results, several physical process parameterization schemes are selected to form several model configuration schemes; Collect global-scale atmospheric reanalysis data and forecast data, select several model configuration schemes, drive the initial numerical weather forecast model to carry out regional forecast simulation, and obtain the runoff depth data.

3. The numerical weather forecasting method based on river network confluence calculation according to claim 2, characterized in that: The method of converting the runoff depth data into river inflow data includes: Selecting vectorized river network geographic information data, wherein the river network geographic information data includes a river centerline and a watershed area; Determine each of the model calculation grids intersecting with the catchment area, calculate the intersection area, and calculate the runoff volume input from each of the model calculation grids to the catchment area corresponding thereto based on the intersection area and the runoff depth data; Each of the runoff amounts is added together to obtain the total runoff amount of the catchment area, and the total runoff amount is used as the river inflow data.

4. The numerical weather forecasting method based on river network confluence calculation according to claim 3, characterized in that: The method for constructing the river network confluence routing scheme includes: According to the spatial location of the river flow observation station, determining the river section closest to the river flow observation station in the river network geographic information data as the river section where the station is located; Search upstream and downstream along the river section where the station is located to determine the river section upstream of the station and the river section downstream of the station; Arrange the river flow observation stations in order from upstream to downstream, and determine the river section between every two adjacent stations; For all river sections in the river network geographic information data, the Muskingum method is used to perform flow routing: Among them, Q t Indicates the flow of the river section at the current calculation time step, Q t-1 represents the river flow at the previous time step, represents the upstream river flow at the current calculation time step, represents the upstream river flow at the previous time step, represents the river inflow data, Δt represents the time step, Δl represents the river section length, both the time step and the river section length are greater than 0, x represents the weight coefficient, the weight coefficient value is greater than or equal to 0 and less than or equal to 0.5, c represents the wave speed, the wave speed value is greater than 0 and less than half of the quotient of the river section length and the time step.

5. The numerical weather forecasting method based on river network confluence calculation according to claim 4, characterized in that: The method for obtaining the plurality of flow simulation values includes: Step 1: Select the river flow observation stations in order from upstream to downstream; Step 2: Select a series of different wave velocities and weight coefficients, use the first flow simulation value of the river section above the upstream station and the corresponding river inflow data as input, perform confluence calculation on all river sections from the station to its upstream station, and obtain the second flow simulation value of the current station; Step 3: Compare the second flow simulation value of the current station with the observed value, select the wave velocity and weight coefficient corresponding to the second flow simulation value closest to the observed value as the confluence parameter value of the river section between the current station and the upstream station, and use the corresponding second flow simulation value as the input of the downstream station; Repeat steps 1 to 3 until all the river flow observation stations are used to obtain the flow simulation values on all river sections.

6. The numerical weather forecasting method based on river network confluence calculation according to claim 1, characterized in that: The method for comparing the flow simulation value and the historical observation value includes: Calculate the KGE indicator value: Among them, r represents the correlation coefficient between the flow simulation value and the historical observation value, σ represents the standard deviation of the flow simulation value, and σ o represents the standard deviation of historical observations, μ represents the average value of flow simulation values, and μ o represents the average of historical observations; The KGE index values of each configuration scheme are compared, the configuration scheme corresponding to the KGE index value closest to 1 is selected, and the final numerical weather forecast model is constructed based on the selected configuration scheme.

7. A numerical weather forecast system based on river network confluence calculation, the system applying the method according to any one of claims 1 to 6, characterized in that: include: Model configuration module, flow calculation module and forecast module; The model configuration module is used to configure an initial numerical weather forecast model, conduct regional forecast simulation based on the initial numerical weather forecast model to obtain runoff depth data, and convert the runoff depth data into river inflow data; The confluence calculation module is used to construct a river network confluence calculation scheme, and perform river network confluence calculation on the initial numerical weather forecast model based on the river inflow data to obtain a number of flow simulation values for all river sections in the river network; The forecast module is used to compare the flow simulation value and the historical observation value, select the configuration scheme with the two values closest to build a final numerical weather forecast model, and use the numerical weather forecast model to perform weather forecasting.

Images