A hydrological and meteorological coupled forecast and early warning system

Abstract

A hydrometeorological coupling forecasting and warning system relates to the technical field of hydrology, water resources and meteorological forecasting. It constructs a visual map of geographical elements in a target area; constructs a hydrometeorological prediction model to output initial hydrometeorological prediction data; performs ultra-short-term difference detection on the initial hydrometeorological prediction data in combination with the hydrometeorological data collected at each real-time monitoring point, and adjusts the initial hydrometeorological data hour by hour based on the ultra-short-term difference detection result; constructs a coupling feature database, performs coupling analysis on multiple feature parameters of several historical collection cycles, and obtains the coupling coefficient between different precipitation feature parameters and flood flow level under different terrain feature parameters and water level feature parameters; obtains the flood flow level of each grid in the target area; and constructs a risk warning visual map based on the flood flow level of each grid, so that decision makers can quickly grasp the overall risk distribution, formulate emergency plans and allocate resources more efficiently, and improve the overall emergency management level.

Description

Technical Field

[0001] The present invention relates to the technical field of hydrology, water resources and meteorological forecasting, and in particular to a hydrological and meteorological coupled forecasting and early warning system. Background Art

[0002] Chinese patent publication number CN111260159A discloses a meteorological-hydrological coupled flood monitoring and reporting method, which includes: obtaining the relationship between the catchment area, precipitation duration, flood duration, and flood peak duration, that is, the coupled relationship between meteorology and hydrology. This relationship is formed by the combined effect of various factors in the basin. Therefore, this relationship can greatly simplify the calculation process. When measuring and reporting, grassroots staff or remote control personnel only need to obtain the conventionally observed precipitation duration, precipitation, flood rise time and flow, runoff coefficient, etc. of the monitoring section to calculate the flood process, peak flow and time of the flood peak in the section.

[0003] A Chinese patent with publication number CN119849343A discloses a meteorological-hydrological-hydrodynamic coupled small-watershed flash flood disaster forecasting method, including: constructing a multidimensional quantitative indicator system of disaster-causing factors and disaster-prone environments, combining a three-dimensional water system model and a smoothed particle hydrodynamics (SPH) hydrodynamic model, building a disaster prediction ontology and extracting feature data, optimizing features and target variables through machine learning and deep learning, training the extracted features, and constructing disaster scenarios based on historical flash flood disaster data, thereby establishing a chain relationship between hydrological and meteorological data, hydrodynamic characteristics, water level and flow representation, and disaster conditions.

[0004] Existing technologies will face many technical problems: at the data level, it is difficult to fully integrate socioeconomic data with GIS geographic data, the layout of monitoring points lacks scientific basis, and there is waste of resources or monitoring blind spots; at the prediction level, traditional models cannot effectively cope with complex terrain and extreme weather, the prediction accuracy is low and difficult to update in real time; at the analysis level, there is a lack of in-depth exploration of the coupling relationship between hydrological and meteorological elements, making it difficult to accurately assess flood flow levels; at the early warning level, the degree of visualization is low and the risk classification is unscientific, which cannot provide intuitive and reliable support for emergency decision-making, resulting in delayed early warning response and inefficient emergency resource allocation. Summary of the Invention

[0005] In order to solve the above technical problems, the object of the present invention is to provide a hydro-meteorological coupled forecasting and warning system, comprising a monitoring center, wherein the monitoring center is communicatively connected to a regional feature analysis module, a data acquisition module, a hydro-meteorological prediction module, a fast and accurate detection module, a feature extraction module, a coupling analysis module, a coupling evaluation module, and a risk warning module;

[0006] The regional feature analysis module is used to perform feature analysis on the socio-economic data and GIS geographic data of the target area and generate a visual map of geographic elements;

[0007] The data acquisition module is used to arrange real-time monitoring points based on the visual map of geographic elements and collect hydrological and meteorological data;

[0008] The hydrometeorological forecast module is used to build a hydrometeorological forecast model and output the initial hydrometeorological forecast data for the current collection period;

[0009] The fast and accurate detection module is used to perform ultra-short-term difference detection on the initial hydrological and meteorological forecast data in combination with the hydrological and meteorological data collected from each real-time monitoring point, and to adjust the initial hydrological and meteorological data hour by hour based on the ultra-short-term difference detection results;

[0010] The feature extraction module is used to extract features from the hydrological and meteorological forecast data after ultra-short-time difference detection to obtain multiple feature parameters;

[0011] The coupling analysis module is used to build a coupling feature database, conduct coupling analysis on multiple feature parameters of several historical acquisition periods, obtain the coupling coefficients between different precipitation feature parameters and flood flow levels under different terrain feature parameters and water level feature parameters, and store them in the coupling feature database;

[0012] The coupling evaluation module is used to perform coupling evaluation based on the coupling feature database to obtain the flood flow level of each grid in the target area;

[0013] The risk warning module is used to construct a risk warning visual graph based on the flood flow level of each grid.

[0014] Furthermore, the regional feature analysis module performs feature analysis on the socio-economic data and GIS geographic data of the target area. The process of generating a visual map of geographic features includes:

[0015] Obtain socioeconomic data and GIS geographic data for the target area, pre-process the data format, convert all vector data (such as rivers and administrative divisions) into raster data, and unify the coordinate system (national geodetic coordinate system), projection (Gauss-Krüger projection), and resolution (100m×100m). Convert indicators of different dimensions to standardized values ​​in the interval [0,1]. Obtain grid data layers corresponding to each type of indicator in the socioeconomic and GIS geographic data (composed of the standardized values ​​corresponding to the indicators in different grids). The various types of indicators in the socioeconomic and GIS geographic data include: physical geographic indicators (precipitation frequency, river slope, elevation, river network density, soil permeability, vegetation cover, and land use type); socioeconomic indicators (population density, GDP density, infrastructure density, historical flood frequency, and drainage system integrity). Set indicator weights for each type of indicator (the indicator weights for different indicators are determined based on expert experience to reduce uncertainty in the fuzzy comprehensive evaluation process). Perform a weighted overlay on the grid data layers corresponding to all types of indicators to generate a comprehensive risk index raster layer.

[0016] Further, The calculation process of weighted overlaying of grid data layers to generate a comprehensive risk index grid layer includes: ;

[0017] in, is the comprehensive risk index, For the The indicator weight of the type indicator, For the The normalized value corresponding to the type indicator, is the total number of indicator types;

[0018] Based on GIS geographic data, several geographic element vector layers (including rivers, settlements, roads, etc.) of the target area are constructed, and several geographic element vector layers are superimposed on the comprehensive risk index raster layer to generate a visual map of geographic elements.

[0019] Furthermore, the data collection module arranges real-time monitoring points based on the visual map of geographic elements. The process of collecting hydrological and meteorological data includes:

[0020] Obtain the comprehensive risk index of each grid in the geographic element visual map and the geographic elements (such as shallows, deep pools, water confluences, etc.), preset the maximum coverage of the real-time monitoring points corresponding to different comprehensive risk indices, and obtain the maximum coverage of the real-time monitoring points of each grid based on the comprehensive risk index; the maximum coverage of the real-time monitoring points in the high comprehensive risk index area is smaller than the maximum coverage of the real-time monitoring points in the medium and low comprehensive risk index areas;

[0021] According to the maximum coverage of the real-time monitoring points of the grid and the coverage area of ​​the grille Get the number of real-time monitoring points of the grid , , Indicates rounding up, preset adjustment coefficients corresponding to different geographical elements , adjust the number of real-time monitoring points according to the adjustment coefficient corresponding to the geographical elements of the grid, and obtain the adjusted number of real-time monitoring points , ;

[0022] The point distribution of the real-time monitoring points of the grid is obtained according to the adjusted number of real-time monitoring points. The real-time monitoring points are used to collect hydrological and meteorological data, mark the collection time, and set the collection cycle.

[0023] Furthermore, the process of obtaining the point distribution of the real-time monitoring points of the grid according to the adjusted number of real-time monitoring points includes:

[0024] Initialize the position of the real-time monitoring point in the grid and set the coordinate set of the initial real-time monitoring point ;

[0025] Construct a polygonal monitoring area for each initial real-time monitoring point. , define its polygonal monitoring area as all The set of points whose distance is less than the distance to other initial real-time monitoring points :

[0026] ;

[0027] in, Initial real-time monitoring point Maximum coverage, for point To the initial real-time monitoring point The Euclidean distance of : ;

[0028] Optimize the initial real-time monitoring point positions to minimize the variance of each polygonal monitoring area, and iteratively adjust the node positions to make the coverage more uniform. The objective function is: ;

[0029] in, For the The area of ​​the polygonal monitoring area, is the average area.

[0030] Furthermore, the hydrometeorological forecast module constructs a hydrometeorological forecast model, and the process of outputting the initial hydrometeorological forecast data for the current collection period includes:

[0031] A hydrometeorological prediction model with a visual graph of geographic elements is constructed based on deep learning. Meteorological data and hydrological data from several historical collection periods of the target area are used as training sets and test sets. The training sets are input into the hydrometeorological prediction model for training until the loss function training is stable. The model parameters are saved and the hydrometeorological prediction model is tested on the test set until it meets the preset requirements. The hydrometeorological prediction model is then output.

[0032] The initial hydrological and meteorological forecast data of each grid in the current collection period are output based on the trained hydrological and meteorological forecast model.

[0033] Furthermore, the rapid and accurate detection module combines the hydrological and meteorological data collected at each real-time monitoring point to perform ultra-short-term difference detection on the initial hydrological and meteorological forecast data. The process of hourly adjusting the initial hydrological and meteorological data based on the ultra-short-term difference detection results includes:

[0034] Divide the collection period into a number of identical collection sub-periods, and set estimated threshold intervals for various types of indicators corresponding to the hydrological and meteorological data within each collection sub-period based on the initial hydrological and meteorological forecast data and a preset error upper limit. The various types of indicators corresponding to the hydrological and meteorological data include hydrological data indicators (including water quantity indicators, water quality indicators, groundwater indicators, etc.) and meteorological data indicators (including basic meteorological elements, precipitation and evaporation indicators, radiation and sunshine indicators, etc.);

[0035] Extract the hydrological and meteorological data collected by each real-time monitoring point within the collection sub-period at the end timestamp of the collection sub-period, extract the numerical time series of each type of indicator corresponding to the hydrological and meteorological data, compare the numerical time series of each type of indicator with the corresponding estimated threshold interval, and obtain the cumulative time that each type of indicator is not within the corresponding estimated threshold interval;

[0036] The cumulative time corresponding to each type of indicator at each real-time monitoring point within the collection sub-cycle is compared with the preset cumulative time threshold. If the cumulative time corresponding to a certain type of indicator at a certain real-time monitoring point is greater than the cumulative time threshold, the real-time monitoring point will be marked as a key point, and the collection sub-cycle will be marked as the starting collection sub-cycle. The initial hydrological and meteorological forecast data of the key point will be adjusted hour by hour.

[0037] Furthermore, the process of hourly adjusting the initial hydrological and meteorological forecast data at key points includes:

[0038] Step 1: Input the hydrological and meteorological data of the initial collection sub-cycle into the hydrological and meteorological prediction model, and output the hydrological and meteorological prediction data of the next collection sub-cycle of the initial collection sub-cycle according to the hydrological and meteorological prediction model;

[0039] Step 2: Perform ultra-short-term difference detection on the hydrological and meteorological forecast data of the next collection sub-cycle, obtain the cumulative time in which each type of indicator in the hydrological and meteorological forecast data of the next collection cycle is not within the corresponding forecast threshold range, and if the cumulative time corresponding to a type of indicator is greater than the cumulative time threshold, use the initial hydrological and meteorological forecast data of the next collection cycle as the updated object, and update the initial hydrological and meteorological forecast data of the next collection cycle according to the hydrological and meteorological forecast data of the next collection cycle, and update the initial hydrological and meteorological forecast data of the next collection cycle to the hydrological and meteorological forecast data, and then execute step 3. If the cumulative time of orders corresponding to each type of indicator is less than or equal to the cumulative time threshold, execute step 4;

[0040] Step 3: Mark the next collection period as the target collection period, input the hydrometeorological forecast data of the target collection period into the hydrometeorological forecast model, output the hydrometeorological forecast data of the next collection sub-period of the target collection period according to the hydrometeorological forecast model, and then execute step 2;

[0041] Step 4: Using the initial hydrometeorological forecast data of the next collection period as the updated object, updating the initial hydrometeorological forecast data of the next collection period according to the hydrometeorological forecast data of the next collection period, and updating the initial hydrometeorological forecast data corresponding to the next collection period to the hydrometeorological forecast data;

[0042] Step 5: Determine whether the next acquisition sub-cycle in step 2 is the last acquisition sub-cycle. If so, force step 2 to end, and based on the hydrological and meteorological forecast data of the next acquisition sub-cycle in step 2, use the initial hydrological and meteorological forecast data of the next acquisition cycle in step 2 as the updated object, and update the initial hydrological and meteorological forecast data of the next acquisition cycle based on the hydrological and meteorological forecast data of the next acquisition cycle in step 2, and update the initial hydrological and meteorological forecast data corresponding to the next acquisition cycle in step 2 to the hydrological and meteorological forecast data.

[0043] Furthermore, the feature extraction module extracts features from the hydrological and meteorological forecast data after ultra-short-term difference detection. The process of obtaining multiple feature parameters includes:

[0044] Statistical analysis is performed on the numerical time series of various types of indicators corresponding to the hydrological and meteorological forecast data of each grid after ultra-short-term difference detection (including initial hydrological and meteorological forecast data that have not been adjusted hourly and hydrological and meteorological forecast data that have been adjusted hourly) to obtain multiple feature parameters of each grid in the current acquisition period. The multiple feature parameters include precipitation feature parameters, terrain feature parameters and water level feature parameters. The precipitation feature parameters include cumulative rainfall, maximum rainfall intensity and rainfall variance. The terrain feature parameters include altitude and terrain undulation, basin area and shape, river network density and river longitudinal gradient. The water level feature parameters include real-time water level, average water level and water level variance.

[0045] Furthermore, the coupling analysis module constructs a coupling feature database, performs coupling analysis on multiple feature parameters of several historical acquisition periods, obtains coupling coefficients between different precipitation feature parameters and flood flow levels under different terrain feature parameters and water level feature parameters, and stores them in the coupling feature database. The process includes:

[0046] Obtain flood occurrence records for several historical collection periods of each grid, wherein the flood occurrence records include precipitation characteristic parameters, terrain characteristic parameters, water level characteristic parameters, and flood flow level, mark the terrain characteristic parameters and water level characteristic parameters as clustering parameters, cluster the flood occurrence records for several historical collection periods of each grid according to the clustering parameters, perform cosine similarity comparison on the terrain characteristic parameters and water level characteristic parameters in each flood occurrence record with the terrain characteristic parameters and water level characteristic parameters in other flood occurrence records, obtain cosine similarity between each flood occurrence record, preset a cosine similarity threshold, and if the cosine similarity between the flood occurrence records is less than the cosine similarity threshold, cluster the flood occurrence records to generate several cluster sets of flood occurrence records;

[0047] A coupling analysis was conducted on the precipitation characteristic parameters and flood flow levels in each flood occurrence record cluster set to obtain the coupling coefficients between different precipitation characteristic parameters and flood flow levels under different terrain characteristic parameters and water level characteristic parameters (i.e., the coupling coefficients between cumulative rainfall, maximum rainfall intensity, rainfall variance and flood flow levels);

[0048] A coupling characteristic database is constructed, and the coupling coefficients between different precipitation characteristic parameters and flood flow levels under different terrain characteristic parameters and water level characteristic parameters are stored in the coupling characteristic database.

[0049] Furthermore, a coupling analysis is performed on the precipitation characteristic parameters and flood flow levels in each flood occurrence record cluster set to obtain the coupling coefficients between different precipitation characteristic parameters and flood flow levels under different terrain characteristic parameters and water level characteristic parameters. The process includes:

[0050] Traverse each cluster set, for the kth cluster set (denoted as ), its topographical features and water level characteristics It has been determined by cosine similarity clustering (i.e., all flood occurrence records within the cluster set are considered to be of the same topography-water level conditions);

[0051] from Extract the dataset ,in for The number of records of internal torrent occurrences, For the The cumulative rainfall within the flood record, For the The maximum rainfall intensity during the flood record, For the The rainfall variance within each flood occurrence record;

[0052] right Perform multiple linear regression fitting to obtain the coupling coefficient 、 、 ;

[0053] Among them, the model for multiple linear regression fitting is:

[0054] ;

[0055] in, is the intercept term, is a random error term that obeys normal distribution. 、 、 Represented in terrain features and water level characteristics Under these conditions, the impact of cumulative rainfall, maximum rainfall intensity and rainfall variance on flood magnitude is significant;

[0056] Construct loss function:

[0057] ;

[0058] right 、 、 、 Find the partial derivative and set it to zero, and solve for the coefficient matrix:

[0059] ;

[0060] in, is the independent variable matrix, each row corresponds to a record, the form is ;

[0061] Flood level .

[0062] Furthermore, the coupling evaluation module performs coupling evaluation based on the coupling feature database, and the process of obtaining the flood flow level of each grid in the target area includes:

[0063] Input the terrain characteristic parameters and water level characteristic parameters of each grid in the current acquisition period into the coupling characteristic database for retrieval to obtain the coupling level between each precipitation characteristic parameter of each grid and the flood flow level;

[0064] The precipitation characteristic parameters of each grid are used as evaluation indicators. The indicator weights of the evaluation indicators are set according to the coupling level between the precipitation characteristic parameters of each grid and the flood flow level. The membership matrix of each grid for different flood flow levels is obtained through fuzzy comprehensive evaluation. The flood flow level of each grid is obtained based on the membership matrix and the indicator weights.

[0065] Furthermore, the process of obtaining the flood level of each grid according to the membership matrix and the index weight includes:

[0066] The fuzzy comprehensive evaluation matrix of the evaluation index is obtained by integrating the indicator weights and the membership matrix of the evaluation index through a formula. The membership of each grid for different flood flow levels is obtained according to the fuzzy comprehensive evaluation matrix. The flood flow level with the highest membership corresponding to each grid is screened out and the flood flow level with the highest membership corresponding to each grid is used as the flood flow level of each grid.

[0067] Wherein, the formula is:

[0068] ;

[0069] in, is the fuzzy comprehensive evaluation matrix of evaluation indicators, is the indicator weight of the evaluation index, is the membership matrix, " represents the multiplication of the elements at the corresponding positions of the weight matrix of the evaluation index and the membership matrix, It is the weighting parameter used to control the balance between the weight matrix and the membership matrix in the fuzzy comprehensive evaluation matrix of the evaluation index.

[0070] Furthermore, the process of the risk warning module constructing a risk warning visual graph based on the flood flow level of each grid includes:

[0071] Preset colors corresponding to different flood levels, apply graded color rendering to the geographic feature visual map based on the colors corresponding to the flood levels of each grid, and select continuous color bands (such as blue-yellow-red color system) to enhance the visual distinction of risk levels and generate a risk warning visual map;

[0072] Preset early warning response measures corresponding to different colors (such as blue-yellow-orange-red), and generate early warning response measures for each grid according to the color of each grid in the risk visualization diagram.

[0073] Compared with the prior art, the present invention has the following beneficial effects:

[0074] 1. Deep Integration and Precision Analysis of Multi-Source Data: The system integrates socioeconomic data and GIS geographic data through a regional characteristics analysis module. After data format preprocessing and weighted overlay, it generates a comprehensive risk index raster layer, overlaying a geographic element vector layer to form a visual map of geographic elements. Compared to traditional systems that rely solely on hydrological or meteorological data, this system can more comprehensively reflect regional characteristics. For example, when analyzing urban flood risk, it can combine socioeconomic data such as population density and building distribution to accurately locate high-risk areas, providing a solid foundation for subsequent monitoring point deployment and early warning.

[0075] 2. Intelligent monitoring point layout optimizes resource allocation: The data acquisition module scientifically calculates the number of real-time monitoring points based on the comprehensive risk index and geographic elements of the grid in the geographic element visual map, and optimizes the layout by adjusting the coefficients. This approach avoids blind deployment of monitoring points, ensuring full coverage in high-risk areas while reasonably reducing the number of points in low-risk areas, improving monitoring efficiency while reducing construction and operation and maintenance costs. For example, in mountainous areas with complex terrain, point density can be flexibly adjusted based on geographic factors such as slope and river networks to accurately capture hydrological and meteorological changes.

[0076] 3. Deep Learning Improves Forecast Model Accuracy: The hydrometeorological forecast module builds a model based on deep learning, utilizing extensive historical data for training and testing, enabling the model to learn complex patterns of hydrometeorological changes. Compared to traditional empirical models, it offers enhanced forecasting capabilities for complex scenarios such as extreme weather and unusual terrain. The output of initial hydrometeorological forecast data is more accurate, providing a reliable basis for subsequent warnings.

[0077] 4. Ultra-short-term difference detection enhances real-time performance: The rapid and accurate detection module subdivides the collection cycle and, by setting estimated threshold intervals, compares monitored data with estimated data in real time. If an anomaly is detected, the initial hydrological and meteorological data is promptly adjusted hourly. This mechanism significantly improves the system's response to sudden hydrological and meteorological changes. For example, during a sudden downpour, forecast deviations can be quickly corrected, making warning information more relevant to actual conditions and buying valuable time for emergency decision-making.

[0078] 5. Coupling Analysis Exploits Intrinsic Relationships to Improve the Scientificity of Early Warnings: The coupling analysis module clusters and couples historical multi-characteristic parameters to obtain the coupling coefficients between rainfall characteristic parameters and flood levels under different conditions, and constructs a coupling characteristic database. This process deeply explores the inherent connections between hydrological and meteorological elements, enabling early warning models to move away from relying on simple empirical correlations and instead rely on quantitative coupling relationships, improving the scientificity and accuracy of flood level forecasts.

[0079] 6. Fuzzy Comprehensive Evaluation for Accurate Risk Grading: The coupled evaluation module uses a fuzzy comprehensive evaluation method, combining the coupling level determined by the coupling coefficient to set indicator weights. This method calculates the membership matrix of each grid for different flood flow levels, thereby accurately determining flood flow levels. This method comprehensively considers multiple uncertainties, avoids the one-sidedness of single-indicator evaluation, and provides a more reasonable classification of flood risks, making warning results more valuable.

[0080] 7. Visualized early warning improves decision-making efficiency: The risk warning module generates a visual risk warning map by rendering geographic elements in graded colors. This intuitive visual display shows the corresponding risk level for each grid flood flow level. Compared to traditional text or data reports, this visual interface enables decision makers to quickly grasp the overall risk distribution, more efficiently formulate emergency plans and allocate resources, and improve overall emergency management capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 This is a schematic diagram of a hydrological and meteorological coupled forecasting and warning system according to an embodiment of the present application. DETAILED DESCRIPTION

[0082] The following is a clear and complete description of the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0083] like Figure 1 As shown, a hydrological and meteorological coupled forecasting and warning system includes a monitoring center, which is communicatively connected to a regional feature analysis module, a data acquisition module, a hydrological and meteorological prediction module, a fast and accurate detection module, a feature extraction module, a coupling analysis module, a coupling evaluation module, and a risk warning module;

[0084] The regional feature analysis module is used to perform feature analysis on the socio-economic data and GIS geographic data of the target area and generate a visual map of geographic elements;

[0085] The data acquisition module is used to arrange real-time monitoring points based on the visual map of geographic elements and collect hydrological and meteorological data;

[0086] The hydrometeorological forecast module is used to build a hydrometeorological forecast model and output the initial hydrometeorological forecast data for the current collection period;

[0087] The fast and accurate detection module is used to perform ultra-short-term difference detection on the initial hydrological and meteorological forecast data in combination with the hydrological and meteorological data collected from each real-time monitoring point, and to adjust the initial hydrological and meteorological data hour by hour based on the ultra-short-term difference detection results;

[0088] The feature extraction module is used to extract features from the hydrological and meteorological forecast data after ultra-short-time difference detection to obtain multiple feature parameters;

[0089] The coupling analysis module is used to build a coupling feature database, conduct coupling analysis on multiple feature parameters of several historical acquisition periods, obtain the coupling coefficients between different precipitation feature parameters and flood flow levels under different terrain feature parameters and water level feature parameters, and store them in the coupling feature database;

[0090] The coupling evaluation module is used to perform coupling evaluation based on the coupling feature database to obtain the flood flow level of each grid in the target area;

[0091] The risk warning module is used to construct a risk warning visual graph based on the flood flow level of each grid.

[0092] It should be further explained that, in the specific implementation process, the regional feature analysis module performs feature analysis on the socio-economic data and GIS geographic data of the target area, and the process of generating a visual map of geographic features includes:

[0093] Obtain socioeconomic data and GIS geographic data for the target area, pre-process the data format, convert all vector data (such as rivers and administrative divisions) into raster data, and unify the coordinate system (national geodetic coordinate system), projection (Gauss-Krüger projection), and resolution (100m×100m). Convert indicators of different dimensions to standardized values ​​in the interval [0,1]. Obtain grid data layers corresponding to each type of indicator in the socioeconomic and GIS geographic data (composed of the standardized values ​​corresponding to the indicators in different grids). The various types of indicators in the socioeconomic and GIS geographic data include: physical geographic indicators (precipitation frequency, river slope, elevation, river network density, soil permeability, vegetation cover, and land use type); socioeconomic indicators (population density, GDP density, infrastructure density, historical flood frequency, and drainage system integrity). Set indicator weights for each type of indicator (the indicator weights for different indicators are determined based on expert experience to reduce uncertainty in the fuzzy comprehensive evaluation process). Perform a weighted overlay on the grid data layers corresponding to all types of indicators to generate a comprehensive risk index raster layer.

[0094] It should be further explained that in the specific implementation process, The calculation process of weighted overlaying of grid data layers to generate a comprehensive risk index grid layer includes:

[0095] ;

[0096] in, is the comprehensive risk index, For the The indicator weight of the type indicator, For the The normalized value corresponding to the type indicator, is the total number of indicator types;

[0097] Based on GIS geographic data, several geographic element vector layers (including rivers, settlements, roads, etc.) of the target area are constructed, and several geographic element vector layers are superimposed on the comprehensive risk index raster layer to generate a visual map of geographic elements.

[0098] It should be further explained that, in the specific implementation process, the data collection module arranges real-time monitoring points based on the visual map of geographic elements, and the process of collecting hydrological and meteorological data includes:

[0099] Obtain the comprehensive risk index of each grid in the geographic element visual map and the geographic elements (such as shallows, deep pools, water confluences, etc.), preset the maximum coverage of the real-time monitoring points corresponding to different comprehensive risk indices, and obtain the maximum coverage of the real-time monitoring points of each grid based on the comprehensive risk index; the maximum coverage of the real-time monitoring points in the high comprehensive risk index area is smaller than the maximum coverage of the real-time monitoring points in the medium and low comprehensive risk index areas;

[0100] According to the maximum coverage of the real-time monitoring points of the grid and the coverage area of ​​the grille Get the number of real-time monitoring points of the grid , , Indicates rounding up, preset adjustment coefficients corresponding to different geographical elements , adjust the number of real-time monitoring points according to the adjustment coefficient corresponding to the geographical elements of the grid, and obtain the adjusted number of real-time monitoring points , ;

[0101] The point distribution of the real-time monitoring points of the grid is obtained according to the adjusted number of real-time monitoring points. The real-time monitoring points are used to collect hydrological and meteorological data, mark the collection time, and set the collection cycle.

[0102] It should be further explained that, in a specific implementation process, the process of obtaining the point distribution of the real-time monitoring points of the grid according to the adjusted number of real-time monitoring points includes:

[0103] Initialize the position of the real-time monitoring points within the grid and set the coordinate set P of the initial real-time monitoring points: ;

[0104] Construct a polygonal monitoring area for each initial real-time monitoring point. , define its polygon monitoring area as all The set of points whose distance is less than the distance to other initial real-time monitoring points :

[0105] ;

[0106] in, Initial real-time monitoring point The Euclidean distance of : ;

[0107] Optimize the initial real-time monitoring point positions to minimize the variance of each polygonal monitoring area, and iteratively adjust the node positions to make the coverage more uniform. The objective function is:

[0108] ;

[0109] in, For the The area of ​​the polygonal monitoring area, is the average area.

[0110] It should be further explained that, in the specific implementation process, the hydrometeorological forecast module constructs the hydrometeorological forecast model and outputs the initial hydrometeorological forecast data for the current collection period, including the following steps:

[0111] A hydrometeorological prediction model with a visual graph of geographic elements is constructed based on deep learning. Meteorological data and hydrological data from several historical collection periods of the target area are used as training sets and test sets. The training sets are input into the hydrometeorological prediction model for training until the loss function training is stable. The model parameters are saved and the hydrometeorological prediction model is tested on the test set until it meets the preset requirements. The hydrometeorological prediction model is then output.

[0112] The initial hydrological and meteorological forecast data of each grid in the current collection period are output based on the trained hydrological and meteorological forecast model.

[0113] Building a deep learning-based hydrometeorological prediction model is a complex process that involves multiple steps, including model selection, training, validation, and testing. The following is a detailed supplementary explanation of this process:

[0114] This paper selects a CNN-LSTM neural network, suitable for spatiotemporal analysis, as the deep learning architecture for the hydrological and meteorological forecasting model. CNN extracts spatial features, while LSTM captures temporal dependencies. After determining the model architecture, the mean squared error loss function is selected as the optimization objective. The prepared training set is then fed into the selected deep learning model to begin training. During training, weights are continuously updated using a backpropagation algorithm, gradually reducing the loss function until a steady state is reached. During this process, techniques such as early stopping are used to prevent overfitting. In addition to the basic training process, grid search is also used to fine-tune various model parameters, including the learning rate, batch size, and regularization coefficient.

[0115] After model training is complete and parameter adjustments are complete, a final evaluation is performed on the test set to obtain the model's evaluation results. These results include classification metrics such as accuracy, recall, and F1 score. Based on the evaluation results on the test set, it is determined whether the model meets the expected standards. If the requirements are met, the model parameters are saved and prepared for deployment. If not, it is necessary to return to a previous stage to re-examine issues such as data quality, model structure, or training strategy.

[0116] It should be further explained that, in the specific implementation process, the rapid and accurate detection module combines the hydrological and meteorological data collected by each real-time monitoring point to perform ultra-short-term difference detection on the initial hydrological and meteorological forecast data. The process of hourly adjustment of the initial hydrological and meteorological data based on the ultra-short-term difference detection results includes:

[0117] Divide the collection period into a number of identical collection sub-periods, and set estimated threshold intervals for various types of indicators corresponding to the hydrological and meteorological data within each collection sub-period based on the initial hydrological and meteorological forecast data and a preset error upper limit. The various types of indicators corresponding to the hydrological and meteorological data include hydrological data indicators (including water quantity indicators, water quality indicators, groundwater indicators, etc.) and meteorological data indicators (including basic meteorological elements, precipitation and evaporation indicators, radiation and sunshine indicators, etc.);

[0118] Extract the hydrological and meteorological data collected by each real-time monitoring point within the collection sub-period at the end timestamp of the collection sub-period, extract the numerical time series of each type of indicator corresponding to the hydrological and meteorological data, compare the numerical time series of each type of indicator with the corresponding estimated threshold interval, and obtain the cumulative time that each type of indicator is not within the corresponding estimated threshold interval;

[0119] The cumulative time corresponding to each type of indicator at each real-time monitoring point within the collection sub-cycle is compared with the preset cumulative time threshold. If the cumulative time corresponding to a certain type of indicator at a certain real-time monitoring point is greater than the cumulative time threshold, the real-time monitoring point will be marked as a key point, and the collection sub-cycle will be marked as the starting collection sub-cycle. The initial hydrological and meteorological forecast data of the key point will be adjusted hour by hour.

[0120] It should be further explained that, in the specific implementation process, the process of hourly adjustment of the initial hydrological and meteorological forecast data at key points includes:

[0121] Step 1: Input the hydrological and meteorological data of the initial collection sub-cycle into the hydrological and meteorological prediction model, and output the hydrological and meteorological prediction data of the next collection sub-cycle of the initial collection sub-cycle according to the hydrological and meteorological prediction model;

[0122] Step 2: Perform ultra-short-term difference detection on the hydrological and meteorological forecast data of the next collection sub-cycle, obtain the cumulative time in which each type of indicator in the hydrological and meteorological forecast data of the next collection cycle is not within the corresponding forecast threshold range, and if the cumulative time corresponding to a type of indicator is greater than the cumulative time threshold, use the initial hydrological and meteorological forecast data of the next collection cycle as the updated object, and update the initial hydrological and meteorological forecast data of the next collection cycle according to the hydrological and meteorological forecast data of the next collection cycle, and update the initial hydrological and meteorological forecast data of the next collection cycle to the hydrological and meteorological forecast data, and then execute step 3. If the cumulative time of orders corresponding to each type of indicator is less than or equal to the cumulative time threshold, execute step 4;

[0123] Step 3: Mark the next collection period as the target collection period, input the hydrometeorological forecast data of the target collection period into the hydrometeorological forecast model, output the hydrometeorological forecast data of the next collection sub-period of the target collection period according to the hydrometeorological forecast model, and then execute step 2;

[0124] Step 4: Using the initial hydrometeorological forecast data of the next collection period as the updated object, updating the initial hydrometeorological forecast data of the next collection period according to the hydrometeorological forecast data of the next collection period, and updating the initial hydrometeorological forecast data corresponding to the next collection period to the hydrometeorological forecast data;

[0125] Step 5: Determine whether the next acquisition sub-cycle in step 2 is the last acquisition sub-cycle. If so, force step 2 to end, and based on the hydrological and meteorological forecast data of the next acquisition sub-cycle in step 2, use the initial hydrological and meteorological forecast data of the next acquisition cycle in step 2 as the updated object, and update the initial hydrological and meteorological forecast data of the next acquisition cycle based on the hydrological and meteorological forecast data of the next acquisition cycle in step 2, and update the initial hydrological and meteorological forecast data corresponding to the next acquisition cycle in step 2 to the hydrological and meteorological forecast data.

[0126] The error correction process utilizes a closed-loop mechanism of "test-predict-retest-reupdate." For example, the estimated data for the next cycle generated in step 2 must undergo ultra-short-term variance testing again. If anomalies persist, iterative error correction continues (step 3). If the data meets the standards, the initial data is directly updated (step 4). These multiple rounds of verification mitigate the potential for random errors in a single test and ensure the robustness of the error correction results.

[0127] Even during the final collection sub-cycle, the initial forecast is still updated based on the latest estimated data, ensuring data integrity throughout the entire cycle. For example, if an anomaly is detected one hour before the end of the forecast, the data for that hour can still be calibrated to avoid overlooking end-of-cycle errors due to the end of the cycle. This eliminates process blind spots and ensures consistency and accuracy throughout the entire forecast cycle.

[0128] It should be further explained that, in the specific implementation process, the feature extraction module extracts features from the hydrological and meteorological forecast data after ultra-short-term difference detection, and the process of obtaining multiple feature parameters includes:

[0129] Statistical analysis is performed on the numerical time series of various types of indicators corresponding to the hydrological and meteorological forecast data of each grid after ultra-short-term difference detection (including initial hydrological and meteorological forecast data that have not been adjusted hourly and hydrological and meteorological forecast data that have been adjusted hourly) to obtain multiple feature parameters of each grid in the current acquisition period. The multiple feature parameters include precipitation feature parameters, terrain feature parameters and water level feature parameters. The precipitation feature parameters include cumulative rainfall, maximum rainfall intensity and rainfall variance. The terrain feature parameters include altitude and terrain undulation, basin area and shape, river network density and river longitudinal gradient. The water level feature parameters include real-time water level, average water level and water level variance.

[0130] It should be further explained that, in the specific implementation process, the coupling analysis module constructs a coupling feature database, performs coupling analysis on multiple feature parameters of several historical acquisition periods, obtains the coupling coefficients between different precipitation feature parameters and flood flow levels under different terrain feature parameters and water level feature parameters, and stores them in the coupling feature database. The process includes:

[0131] Obtain flood occurrence records for several historical collection periods of each grid, wherein the flood occurrence records include precipitation characteristic parameters, terrain characteristic parameters, water level characteristic parameters, and flood flow level, mark the terrain characteristic parameters and water level characteristic parameters as clustering parameters, cluster the flood occurrence records for several historical collection periods of each grid according to the clustering parameters, perform cosine similarity comparison on the terrain characteristic parameters and water level characteristic parameters in each flood occurrence record with the terrain characteristic parameters and water level characteristic parameters in other flood occurrence records, obtain cosine similarity between each flood occurrence record, preset a cosine similarity threshold, and if the cosine similarity between the flood occurrence records is less than the cosine similarity threshold, cluster the flood occurrence records to generate several cluster sets of flood occurrence records;

[0132] A coupling analysis was conducted on the precipitation characteristic parameters and flood flow levels in each flood occurrence record cluster set to obtain the coupling coefficients between different precipitation characteristic parameters and flood flow levels under different terrain characteristic parameters and water level characteristic parameters (i.e., the coupling coefficients between cumulative rainfall, maximum rainfall intensity, rainfall variance and flood flow levels);

[0133] A coupling characteristic database is constructed, and the coupling coefficients between different precipitation characteristic parameters and flood flow levels under different terrain characteristic parameters and water level characteristic parameters are stored in the coupling characteristic database.

[0134] It should be further explained that, in the specific implementation process, the coupling analysis of precipitation characteristic parameters and flood flow levels in each flood occurrence record cluster set is carried out to obtain the coupling coefficients between different precipitation characteristic parameters and flood flow levels under different terrain characteristic parameters and water level characteristic parameters. The process includes:

[0135] Traverse each cluster set, for the kth cluster set (denoted as ), its topographical features and water level characteristics It has been determined by cosine similarity clustering (i.e., all flood occurrence records within the cluster set are considered to be of the same topography-water level conditions);

[0136] from Extract the dataset ,in for The number of records of internal torrent occurrences, For the The cumulative rainfall within the flood record, For the The maximum rainfall intensity during the flood record, For the The rainfall variance within each flood occurrence record;

[0137] right Perform multiple linear regression fitting to obtain the coupling coefficient 、 、 ;

[0138] Among them, the model for multiple linear regression fitting is:

[0139] ;

[0140] in, is the intercept term, is a random error term that obeys normal distribution. 、 、 Represented in terrain features and water level characteristics Under these conditions, the impact of cumulative rainfall, maximum rainfall intensity and rainfall variance on flood magnitude is significant;

[0141] Construct the loss function:

[0142] ;

[0143] right 、 、 、 Find the partial derivative and set it to zero, and solve for the coefficient matrix:

[0144] ;

[0145] in, is the independent variable matrix, each row corresponds to a record, the form is ;

[0146] Flood level .

[0147] It should be further explained that, in the specific implementation process, the coupling evaluation module performs coupling evaluation based on the coupling feature database, and the process of obtaining the flood flow level of each grid in the target area includes:

[0148] Input the terrain characteristic parameters and water level characteristic parameters of each grid in the current acquisition period into the coupling characteristic database for retrieval to obtain the coupling level between each precipitation characteristic parameter of each grid and the flood flow level;

[0149] The precipitation characteristic parameters of each grid are used as evaluation indicators. The indicator weights of the evaluation indicators are set according to the coupling level between the precipitation characteristic parameters of each grid and the flood flow level. The membership matrix of each grid for different flood flow levels is obtained through fuzzy comprehensive evaluation. The flood flow level of each grid is obtained based on the membership matrix and the indicator weights.

[0150] It should be further explained that, in the specific implementation process, the process of obtaining the flood flow level of each grid according to the membership matrix and the indicator weight includes:

[0151] The fuzzy comprehensive evaluation matrix of the evaluation index is obtained by integrating the indicator weights and the membership matrix of the evaluation index through a formula. The membership of each grid for different flood flow levels is obtained according to the fuzzy comprehensive evaluation matrix. The flood flow level with the highest membership corresponding to each grid is screened out and the flood flow level with the highest membership corresponding to each grid is used as the flood flow level of each grid.

[0152] Wherein, the formula is:

[0153] ;

[0154] in, is the fuzzy comprehensive evaluation matrix of evaluation indicators, is the indicator weight of the evaluation index, is the membership matrix, " represents the multiplication of the elements at the corresponding positions of the weight matrix of the evaluation index and the membership matrix, It is the weighting parameter used to control the balance between the weight matrix and the membership matrix in the fuzzy comprehensive evaluation matrix of the evaluation index.

[0155] It should be further explained that, in the specific implementation process, the risk warning module constructs a risk warning visual graph based on the flood flow level of each grid, including the following steps:

[0156] Preset colors corresponding to different flood levels, apply graded color rendering to the geographic feature visual map based on the colors corresponding to the flood levels of each grid, and select continuous color bands (such as blue-yellow-red color system) to enhance the visual distinction of risk levels and generate a risk warning visual map;

[0157] Preset early warning response measures corresponding to different colors (such as blue-yellow-orange-red), and generate early warning response measures for each grid according to the color of each grid in the risk visualization diagram.

[0158] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A hydrological and meteorological coupled forecasting and early warning system, characterized in that: The monitoring center includes a regional feature analysis module, a data acquisition module, a hydrological and meteorological prediction module, a fast and accurate detection module, a feature extraction module, a coupling analysis module, a coupling evaluation module and a risk warning module. The regional feature analysis module is used to perform feature analysis on the socio-economic data and GIS geographic data of the target area and generate a visual map of geographic elements; The data acquisition module is used to arrange real-time monitoring points based on the visual map of geographic elements and collect hydrological and meteorological data; The hydrometeorological forecast module is used to build a hydrometeorological forecast model and output the initial hydrometeorological forecast data for the current collection period; The fast and accurate detection module is used to perform ultra-short-term difference detection on the initial hydrological and meteorological forecast data in combination with the hydrological and meteorological data collected at each real-time monitoring point. The module then adjusts the initial hydrological and meteorological data hour by hour based on the ultra-short-term difference detection results, including: The collection period is divided into several identical collection sub-periods. Based on the initial hydrological and meteorological forecast data and the preset error limit, the estimated threshold range of each type of indicator corresponding to the hydrological and meteorological data in each collection sub-period is set; Extract the hydrological and meteorological data collected by each real-time monitoring point within the collection sub-period at the end timestamp of the collection sub-period, extract the numerical time series of each type of indicator corresponding to the hydrological and meteorological data, compare the numerical time series of each type of indicator with the corresponding estimated threshold interval, and obtain the cumulative time that each type of indicator is not within the corresponding estimated threshold interval; Compare the cumulative time corresponding to each type of indicator of each real-time monitoring point within the collection sub-cycle with the preset cumulative time threshold. If the cumulative time corresponding to a certain type of indicator of a certain real-time monitoring point is greater than the cumulative time threshold, mark the real-time monitoring point as a key point, mark the collection sub-cycle as the starting collection sub-cycle, and execute the following steps: Step 1: Input the hydrological and meteorological data of the initial collection sub-cycle into the hydrological and meteorological prediction model, and output the hydrological and meteorological prediction data of the next collection sub-cycle of the initial collection sub-cycle according to the hydrological and meteorological prediction model; Step 2: Perform ultra-short-term difference detection on the hydrological and meteorological forecast data of the next acquisition sub-period, and obtain the cumulative time during which each type of indicator in the hydrological and meteorological forecast data of the next acquisition period is not within the corresponding estimation threshold range. If the cumulative time corresponding to any type of indicator is greater than the cumulative time threshold, the initial hydrological and meteorological forecast data of the next acquisition period is updated to the hydrological and meteorological forecast data, and then step 3 is executed. If the cumulative time corresponding to each type of indicator is less than or equal to the cumulative time threshold, step 4 is executed. Step 3: Mark the next collection period as the target collection period, input the hydrometeorological forecast data of the target collection period into the hydrometeorological forecast model, output the hydrometeorological forecast data of the next collection sub-period of the target collection period according to the hydrometeorological forecast model, and then execute step 2; Step 4: Update the initial hydrological and meteorological forecast data corresponding to the next collection period to the hydrological and meteorological forecast data; Step 5: Determine whether the next acquisition sub-cycle in step 2 is the last acquisition sub-cycle. If so, forcibly terminate step 2 and update the initial hydrological and meteorological forecast data corresponding to the next acquisition sub-cycle in step 2 to the hydrological and meteorological forecast data based on the hydrological and meteorological forecast data of the next acquisition sub-cycle in step 2; The feature extraction module is used to extract features from the hydrological and meteorological forecast data after ultra-short-time difference detection to obtain multiple feature parameters; The coupling analysis module is used to build a coupling feature database, conduct coupling analysis on multiple feature parameters of several historical acquisition periods, obtain the coupling coefficients between different precipitation feature parameters and flood flow levels under different terrain feature parameters and water level feature parameters, and store them in the coupling feature database; The coupling evaluation module is used to perform coupling evaluation based on the coupling feature database to obtain the flood flow level of each grid in the target area; The risk warning module is used to construct a risk warning visual graph based on the flood flow level of each grid.

2. A hydrological and meteorological coupled forecasting and early warning system according to claim 1, characterized in that: The regional feature analysis module analyzes the characteristics of the target area's socioeconomic data and GIS geographic data. The process of generating a visual map of geographic features includes: Obtain socioeconomic data and GIS geographic data of the target area, pre-process the socioeconomic data and GIS geographic data, obtain grid data layers corresponding to various indicators in the socioeconomic data and GIS geographic data, set indicator weights for various indicators, perform weighted overlay on grid data layers corresponding to all types of indicators, and generate a comprehensive risk index grid layer; Based on GIS geographic data, several geographic element vector layers of the target area are constructed, and several geographic element vector layers are superimposed on the comprehensive risk index raster layer to generate a visual map of geographic elements.

3. The hydrological and meteorological coupled forecasting and early warning system according to claim 2, characterized in that: The data collection module arranges real-time monitoring points based on a visual map of geographic elements. The process of collecting hydrological and meteorological data includes: Obtain the comprehensive risk index of each grid in the geographic element visual map and the geographic elements, and obtain the maximum coverage of the real-time monitoring points of each grid based on the comprehensive risk index; The number of real-time monitoring points of the grid is obtained according to the maximum coverage range of the real-time monitoring points of the grid and the coverage area of ​​the grid, and the adjustment coefficients corresponding to different geographical elements are preset. The number of real-time monitoring points is adjusted according to the adjustment coefficients corresponding to the geographical elements of the grid to obtain the adjusted number of real-time monitoring points; The point distribution of the real-time monitoring points of the grid is obtained according to the adjusted number of real-time monitoring points. The real-time monitoring points are used to collect hydrological and meteorological data, mark the collection time, and set the collection cycle.

4. The hydrological and meteorological coupled forecasting and early warning system according to claim 3, characterized in that: The hydrometeorological forecast module builds a hydrometeorological forecast model and outputs the initial hydrometeorological forecast data for the current collection period. The process includes: A hydrometeorological prediction model with a visual graph of geographic elements is constructed based on deep learning. Meteorological data and hydrological data from several historical collection periods of the target area are used as training sets and test sets. The training sets are input into the hydrometeorological prediction model for training until the loss function training is stable. The model parameters are saved and the hydrometeorological prediction model is tested on the test set until it meets the preset requirements. The hydrometeorological prediction model is then output. The initial hydrological and meteorological forecast data of each grid in the current collection period are output based on the trained hydrological and meteorological forecast model.

5. The hydrological and meteorological coupled forecasting and early warning system according to claim 4, characterized in that: The feature extraction module extracts features from the hydrological and meteorological forecast data after ultra-short-term difference detection. The process of obtaining multiple feature parameters includes: Statistical analysis is performed on the numerical time series of various types of indicators corresponding to the hydrological and meteorological forecast data of each grid after ultra-short-term difference detection to obtain the multi-feature parameters of each grid in the current acquisition period. The multi-feature parameters include precipitation characteristic parameters, terrain characteristic parameters, and water level characteristic parameters.

6. The hydrological and meteorological coupled forecasting and early warning system according to claim 5, characterized in that: The coupling analysis module builds a coupling feature database, performs coupling analysis on multiple feature parameters of several historical acquisition periods, obtains the coupling coefficients between different precipitation feature parameters and flood flow levels under different terrain feature parameters and water level feature parameters, and stores them in the coupling feature database. The process includes: Obtaining flood occurrence records for several historical collection periods of each grid, the flood occurrence records including precipitation characteristic parameters, terrain characteristic parameters, water level characteristic parameters, and flood flow level, marking the terrain characteristic parameters and water level characteristic parameters as clustering parameters, clustering the flood occurrence records for several historical collection periods of each grid according to the clustering parameters, and generating several cluster sets of flood occurrence records; The coupling analysis of precipitation characteristic parameters and flood flow levels in each flood occurrence record cluster set was carried out to obtain the coupling coefficients between different precipitation characteristic parameters and flood flow levels under different terrain characteristic parameters and water level characteristic parameters. A coupling characteristic database is constructed, and the coupling coefficients between different precipitation characteristic parameters and flood flow levels under different terrain characteristic parameters and water level characteristic parameters are stored in the coupling characteristic database.

7. The hydrological and meteorological coupled forecasting and early warning system according to claim 6, characterized in that: The coupling evaluation module performs coupling evaluation based on the coupling feature database. The process of obtaining the flood flow level of each grid in the target area includes: Input the terrain characteristic parameters and water level characteristic parameters of each grid in the current acquisition period into the coupling characteristic database for retrieval to obtain the coupling level between each precipitation characteristic parameter of each grid and the flood flow level; The precipitation characteristic parameters of each grid are used as evaluation indicators. The indicator weights of the evaluation indicators are set according to the coupling level between the precipitation characteristic parameters of each grid and the flood flow level. The membership matrix of each grid for different flood flow levels is obtained through fuzzy comprehensive evaluation. The flood flow level of each grid is obtained based on the membership matrix and the indicator weights.

8. The hydrological and meteorological coupled forecasting and early warning system according to claim 7, characterized in that: The process of constructing a risk warning visualization graph based on the flood flow levels of each grid by the risk warning module includes: The colors corresponding to different flood flow levels are preset, and graded color rendering is applied to the geographic element visual map based on the colors corresponding to the flood flow levels of each grid to generate a risk warning visual map.

Images