Partitioned flood control early warning method and system fusing multi-source data

Abstract

The application provides a kind of partition flood control early warning method and system of fusion multi-source data, it is related to flood control early warning technical field, the method includes: obtaining the multi-source monitoring data set in target river basin;The multi-source monitoring data set is preprocessed, and the standardized data set is obtained;Based on the standardized data set, the flood evolution characteristic sub-model of each partition in target river basin is respectively constructed;Lightweight physical-empirical hybrid model is used to the output result of the flood evolution characteristic sub-model of each partition is coupled calculation, and the coupling calculation result is obtained;Based on machine learning algorithm analysis historical flood case data, generate the dynamic early warning threshold that is adapted to the flood evolution characteristic sub-model of each partition;Based on the coupling calculation result and the dynamic early warning threshold, generate partition early warning information.The application can fuse multi-source data, be aimed at partition characteristic, dynamically adjust threshold and efficiently calculate.

Description

Technical Field

[0001] This invention relates to the field of flood warning technology, and in particular to a method and system for regional flood warning that integrates multi-source data. Background Technology

[0002] Floods are one of the major natural disasters threatening the lives and property of people in a river basin. Accurate and timely flood warnings are crucial to reducing disaster losses. Some traditional flood warning methods rely heavily on monitoring data from fixed stations, such as rain gauges and water level gauges, resulting in limited data coverage and insufficient timeliness. In addition, the flood evolution patterns vary significantly across different geographical units within a river basin (such as mountainous areas, alluvial fans, and plains), and other flood warning methods often use uniform parameters for calculation, making it difficult to reflect the characteristics of each region.

[0003] Furthermore, in traditional methods, warning thresholds are mostly set based on historical extreme values ​​or experience, and cannot be dynamically adjusted according to real-time hydrological conditions and meteorological changes, which easily leads to false alarms or missed alarms. Summary of the Invention

[0004] To overcome the problems existing in related technologies, this invention provides a method and system for zonal flood prevention early warning that integrates multi-source data, targets zonal characteristics, dynamically adjusts thresholds, and performs efficient calculations, thereby improving the accuracy and timeliness of early warning.

[0005] According to a first aspect of the present invention, a method for regional flood prevention and early warning that integrates multi-source data is provided, comprising:

[0006] Acquire a multi-source monitoring dataset within the target watershed, wherein the multi-source monitoring dataset includes rain gauge monitoring data, snow depth monitoring equipment data, flow meter data, water level gauge data, BeiDou positioning mobile monitoring equipment data, and satellite remote sensing data;

[0007] The multi-source monitoring dataset is preprocessed to obtain a standardized dataset;

[0008] Based on the standardized dataset, flood evolution feature sub-models for each zone within the target watershed are constructed; wherein, each zone includes mountainous areas, alluvial fans, and plains;

[0009] A lightweight physics-empirical hybrid model was used to couple the output results of the flood evolution characteristic sub-models of each zone to obtain the coupled calculation results;

[0010] Based on machine learning algorithms, analyze historical flood case data to generate dynamic early warning thresholds for flood evolution feature sub-models adapted to each zone;

[0011] Based on the coupling calculation results and the dynamic early warning threshold, zonal early warning information is generated.

[0012] In some exemplary embodiments of the present invention, based on the foregoing scheme and the standardized dataset, constructing flood evolution feature sub-models for each sub-region within the target watershed includes:

[0013] The unique geographical parameters, hydrological features, and underlying surface attributes of each region are selected from the standardized dataset to establish a regional feature vector library.

[0014] Based on the regional feature vector library and historical flood process data of each region, the initial sub-model of each region is trained and a dynamic correction module is embedded to obtain the flood evolution feature sub-model of each region.

[0015] In some exemplary embodiments of the present invention, based on the foregoing scheme, a lightweight physical-empirical hybrid model is used to couple the output results of the flood evolution characteristic sub-models of each partition, and the resulting coupled calculation results include:

[0016] Based on the fundamental laws of watershed hydrodynamics, a water exchange equation between different intervals is constructed.

[0017] The output results of each sub-model are used as the boundary input of the water exchange equation between the sub-regions. The hydraulic connection relationship between adjacent sub-regions is solved by simplifying the Saint-Venant equations, and the preliminary coupled flood propagation process is obtained.

[0018] By introducing empirical coefficients extracted from historical flood processes and combining them with machine learning methods, the bias of the initially coupled flood propagation process is corrected, and the corrected zonal flood process lines are output.

[0019] For the output results of each sub-model at different time and spatial scales, a spatiotemporal interpolation algorithm is used to unify the scale, generate consistent spatiotemporal distribution data of flood evolution across the entire watershed, and output the consistent spatiotemporal distribution data of flood evolution across the entire watershed as the coupled calculation result;

[0020] Specifically, the Monte Carlo simulation method is used to perform propagation analysis on the uncertainty in the coupled calculation process, and output the confidence interval and probability distribution of the coupled calculation results to quantify the uncertainty in the coupling process.

[0021] In some exemplary embodiments of the present invention, based on the foregoing scheme, the dynamic early warning threshold for generating a flood evolution feature sub-model adapted to each zone by analyzing historical flood case data using machine learning algorithms includes:

[0022] Key influencing factors for each region were extracted from historical flood case data, and a region feature matrix was constructed.

[0023] Based on the actual degree of damage to each zone in historical flood events, and in accordance with flood control standards and specifications, a corresponding warning level label is marked for each historical case;

[0024] The model is trained by using gradient boosting tree, random forest or deep learning model, with the partition feature matrix as input and the warning level label as output.

[0025] A real-time update module is embedded in the trained model, taking current hydrological monitoring data, short-term meteorological forecast data and underlying surface change information as dynamic inputs, and the model outputs the specific warning thresholds for each zone under the current situation.

[0026] By combining historical extreme events and expert experience, the dynamic early warning thresholds generated by the model are verified to obtain dynamic early warning thresholds that are adapted to the flood evolution characteristic sub-models of each zone.

[0027] In some exemplary embodiments of the present invention, based on the foregoing scheme, the zoning early warning information includes one or more of the following: early warning zone identifier, early warning level, expected risk duration, expected flood arrival time, impact range, and risk avoidance recommendations.

[0028] In some exemplary embodiments of the present invention, based on the foregoing scheme, and based on the coupling calculation result and the dynamic early warning threshold, generating zonal early warning information includes:

[0029] When the coupling calculation result is greater than 50% or more of the dynamic early warning threshold, early warning information including early warning level, scope of impact, expected duration of risk and risk avoidance suggestions is generated.

[0030] When the coupling calculation result is greater than 20% of the dynamic early warning threshold, or when the coupling calculation result is less than 20% of the dynamic early warning threshold, an early warning message is generated that includes the early warning zone identifier, early warning level, expected flood arrival time, impact range, and evacuation advice; when the coupling calculation result is less than 20% of the dynamic early warning threshold, an early warning message is generated that includes the early warning zone identifier, early warning level, and impact range.

[0031] In some exemplary embodiments of the present invention, based on the foregoing scheme, after generating zonal early warning information based on the coupled calculation results and the dynamic early warning threshold, the zonal flood prevention early warning method that integrates multi-source data further includes:

[0032] Distribute zone alert information through at least two of the following methods: SMS, broadcast, app push, and display screen.

[0033] According to a second aspect of the present invention, a regional flood prevention and early warning system that integrates multi-source data is provided, comprising:

[0034] The data acquisition module is used to acquire multi-source monitoring datasets within the target watershed, wherein the multi-source monitoring datasets include rain gauge monitoring data, snow depth monitoring equipment data, flow meter data, water level gauge data, BeiDou positioning mobile monitoring equipment data, and satellite remote sensing data;

[0035] The data preprocessing module is used to preprocess the multi-source monitoring dataset to obtain a standardized dataset;

[0036] The sub-model building module is used to construct flood evolution feature sub-models for each partition within the target watershed based on the standardized dataset; wherein, each partition includes mountainous areas, alluvial fans, and plains;

[0037] The coupling calculation module is used to perform coupling calculations on the output results of the flood evolution characteristic sub-model of each partition using a lightweight physical-empirical hybrid model, and obtain the coupling calculation results.

[0038] The dynamic threshold generation module is used to analyze historical flood case data based on machine learning algorithms and generate dynamic early warning thresholds that are adapted to the flood evolution feature sub-models of each zone.

[0039] The early warning information generation module is used to generate zoned early warning information based on the coupled calculation results and the dynamic early warning threshold.

[0040] According to a third aspect of the present invention, an electronic device is provided, comprising: a processor; and a memory storing computer-readable instructions that, when executed by the processor, implement the method of the first aspect.

[0041] According to a fourth aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method of the first aspect.

[0042] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0043] In this embodiment of the invention, on the one hand, considering the differences in flood evolution patterns among mountainous areas, alluvial fans, and plains within the watershed, flood evolution characteristic sub-models for each region are customized based on the unique geographical parameters, hydrological characteristics, and underlying surface attributes of each region. This avoids simulation biases caused by insufficient adaptability in traditional unified models and can more accurately depict the flood generation and propagation processes in different regions. On the other hand, through a lightweight physics-empirical hybrid model, water exchange relationships between regions are first constructed based on hydrodynamic laws. Then, combined with empirical corrections and spatiotemporal scale unification, the output results of each region's sub-models are efficiently coupled to form coherent flood evolution data across the entire region. This coupling method preserves the accuracy of the regional models, solves the problem of independent regional models, and significantly reduces computational complexity, meeting the efficiency requirements of real-time early warning.

[0044] Secondly, based on the comparison between the partition coupling results and the dynamic early warning thresholds adapted to each partition, this invention clarifies the early warning level, impact range, risk time, and targeted risk avoidance suggestions for each partition. It can directly match the emergency response needs of different regions, allowing management departments and affected people to quickly obtain key information and improve the accuracy and efficiency of disaster response.

[0045] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0046] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the specification, serve to explain the principles of the invention.

[0047] Figure 1 A schematic diagram of the system architecture of an exemplary application environment for a partitioned flood prevention early warning method and system that integrates multi-source data, to which embodiments of the present invention can be applied, is shown;

[0048] Figure 2 The schematic diagram illustrates a flow chart of a partitioned flood prevention and early warning method that integrates multi-source data according to some embodiments of the present invention;

[0049] Figure 3 A schematic diagram of a partitioned flood warning system that integrates multi-source data according to some embodiments of the present invention is shown.

[0050] Figure 4 The schematic diagram illustrates the structure of a computer system of an electronic device according to some embodiments of the present invention;

[0051] Figure 5 A schematic diagram of a computer-readable storage medium according to some embodiments of the present invention is shown. Detailed Implementation

[0052] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention as detailed in the appended claims.

[0053] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The singular forms “a,” “the,” and “the” used in this invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0054] It should be understood that although the terms first, second, third, etc., may be used in this invention to describe various information, this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first information may also be referred to as second information without departing from the scope of this invention, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0055] Figure 1 The diagram illustrates a system architecture of an exemplary application environment for a multi-source data-integrated flood early warning method and apparatus that can be applied according to embodiments of the present invention.

[0056] like Figure 1 As shown, system architecture 100 may include one or more terminal devices such as desktop computer 101, portable computer 102, and smartphone 103, network 104, and server 105. Network 104 is used as a medium to provide a communication link between the terminal devices and server 105. Network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables. Terminal devices may be various electronic devices with data processing capabilities, which have a display screen for displaying early warning information of various sections within the target watershed to the user, including but not limited to the aforementioned desktop computer, portable computer, and smartphone. It should be understood that... Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, there can be any number of terminal devices, networks, and servers. For example, server 105 could be a sub-server cluster composed of multiple sub-servers.

[0057] The multi-source data fusion-based regional flood prevention and early warning method provided in this embodiment of the invention can generally be executed by a terminal device, and correspondingly, the multi-source data fusion-based regional flood prevention and early warning device is generally installed in the terminal device. However, those skilled in the art will readily understand that the multi-source data fusion-based regional flood prevention and early warning method provided in this embodiment of the invention can also be executed by a server 105, and correspondingly, the multi-source data fusion-based regional flood prevention and early warning device can also be installed in the server 105. This exemplary embodiment does not impose any special limitations on this.

[0058] Furthermore, it should be understood that the multi-source data fusion-based zonal flood prevention and early warning method of this invention can be configured as a software module. In some implementation scenarios, the multi-source data fusion-based zonal flood prevention and early warning scheme of this invention can be deployed independently to display early warning information for each zone within different target watersheds, or to display early warning information for each zone within a target watershed individually. In other implementation scenarios, the multi-source data fusion-based zonal flood prevention and early warning scheme of this invention can be deployed within other software as a functional module, such as in underground pipeline analysis software. This invention does not impose any particular restrictions on the application of the multi-source data fusion-based zonal flood prevention and early warning method.

[0059] The embodiments of the present invention will now be described in detail.

[0060] like Figure 2 As shown, Figure 2 This is a flowchart illustrating a partitioned flood prevention and early warning method that integrates multi-source data according to an exemplary embodiment of the present invention, including the following steps:

[0061] S210: Obtain a multi-source monitoring dataset within the target watershed, wherein the multi-source monitoring dataset includes rain gauge monitoring data, snow depth monitoring equipment data, flow meter data, water level gauge data, BeiDou positioning mobile monitoring equipment data, and satellite remote sensing data;

[0062] S220: Preprocess the multi-source monitoring dataset to obtain a standardized dataset;

[0063] S230: Based on the standardized dataset, construct flood evolution characteristic sub-models for each zone within the target watershed; wherein, each zone includes mountainous areas, alluvial fans, and plains;

[0064] S240: A lightweight physical-empirical hybrid model is used to couple the output results of the flood evolution characteristic sub-models of each zone to obtain the coupled calculation results;

[0065] S250: Based on machine learning algorithms, analyze historical flood case data to generate dynamic early warning thresholds that adapt to the flood evolution feature sub-models of each zone;

[0066] S260: Based on the coupling calculation results and the dynamic early warning threshold, generate zonal early warning information.

[0067] In this embodiment of the invention, on the one hand, considering the differences in flood evolution patterns among mountainous areas, alluvial fans, and plains within the watershed, flood evolution characteristic sub-models for each region are customized based on the unique geographical parameters, hydrological characteristics, and underlying surface attributes of each region. This avoids simulation biases caused by insufficient adaptability in traditional unified models and can more accurately depict the flood generation and propagation processes in different regions. On the other hand, through a lightweight physics-empirical hybrid model, water exchange relationships between regions are first constructed based on hydrodynamic laws. Then, combined with empirical corrections and spatiotemporal scale unification, the output results of each region's sub-models are efficiently coupled to form coherent flood evolution data across the entire region. This coupling method preserves the accuracy of the regional models, solves the problem of independent regional models, and significantly reduces computational complexity, meeting the efficiency requirements of real-time early warning.

[0068] Secondly, based on the comparison between the partition coupling results and the dynamic early warning thresholds adapted to each partition, this invention clarifies the early warning level, impact range, risk time, and targeted risk avoidance suggestions for each partition. It can directly match the emergency response needs of different regions, allowing management departments and affected people to quickly obtain key information and improve the accuracy and efficiency of disaster response.

[0069] In S210, a multi-source monitoring dataset within the target watershed is acquired, wherein the multi-source monitoring dataset includes rain gauge monitoring data, snow depth monitoring equipment data, flow meter data, water level gauge data, BeiDou positioning mobile monitoring equipment data, and satellite remote sensing data.

[0070] A target watershed refers to an independent geographical unit with a complete hydrological cycle system that is designated as the core research or monitoring object in specific research and application scenarios such as hydrology, water resources management, ecological protection, or disaster prevention. Its boundaries are usually determined by topography, such as mountains, plains, and valleys. Internally, it encompasses the complete hydrological process of precipitation, runoff generation, confluence, and discharge, including main streams, tributaries, reservoirs, lakes, groundwater systems, and corresponding land areas.

[0071] The delineation of target watersheds needs to be combined with specific application requirements. For example, if used for flash flood early warning, the target watershed might be a small or medium-sized river prone to flash floods and its catchment area; if used for water resource allocation, the target watershed might be the catchment area of ​​a reservoir; if used for ecological protection, the target watershed might be the watershed where a rare aquatic organism's habitat is located. Its core characteristics are a clearly defined spatial scope and complete hydrological functions, serving as the basic spatial carrier for conducting multi-source monitoring, data integration, and subsequent analysis and applications.

[0072] In this embodiment of the invention, based on the topography, hydrological units and climate zones of the target watershed, GIS spatial analysis technology is first used to divide the watershed into several monitoring grids to ensure that each grid covers at least one type of core monitoring indicator. The core monitoring indicators include rain gauge monitoring data, snow depth monitoring equipment data, flow meter data, water level gauge data, Beidou positioning mobile monitoring equipment data and satellite remote sensing data, so as to avoid data collection blind spots.

[0073] Rain gauge monitoring data can be obtained by installing rain gauges in rain gauge stations. For example, in some implementations, fixed tipping bucket rain gauges can be installed, and rain gauges can be densely deployed in key areas (such as flash flood-prone areas and upstream of reservoirs) based on the rainfall distribution patterns in the watershed (such as rainy areas and dry areas) and terrain features (such as open areas and areas avoiding obstruction by buildings / trees) to ensure coverage without blind spots.

[0074] The rain gauge automatically measures rainfall by the number of times the bucket flips. The sampling frequency can be set to 10-30 minutes / time, and it records the timestamp, rainfall and device status simultaneously. It also uses 4G / 5G wireless transmission or LoRa low-power transmission technology to upload the real-time collected data to the data center of the target watershed.

[0075] Snow depth monitoring data can be obtained by deploying ultrasonic snow depth monitors within the snow-covered areas of the target watershed. For example, in some implementations, for areas within the watershed with an altitude ≥2500m or continuous snow cover in winter, the ultrasonic snow depth monitors are installed on stable supports 1.5-2m above the ground to avoid the impact of ground freeze-thaw on measurement accuracy. The distance from the ground / snow surface to the sensor is measured using the ultrasonic ranging principle to calculate the snow depth, and the snow surface temperature is collected simultaneously. The sampling frequency can be set to 1-2 hours / time. Finally, LoRa low-power transmission (adapted to high-altitude environments with weak signals) is used to upload the data to the data center in real time, ensuring data continuity during snowmelt and snowfall periods.

[0076] Flow meter data can be obtained by deploying fixed Doppler ultrasonic flow meters at key cross-sections of the target watershed. In some implementations, representative cross-sections of the main stream and major tributaries of the watershed can be selected, such as the watershed outlet, reservoir inlet / outlet, or town water intake. The fixed Doppler ultrasonic flow meters are installed at the midline of the cross-section, i.e., the location where the water flow is most turbulent and the flow velocity is most representative, ensuring that the sensor is completely submerged in water and avoiding areas of siltation. The flow velocity is measured by the Doppler effect, and the instantaneous flow rate is calculated by combining it with the cross-sectional area. The average flow velocity of the cross-section is recorded simultaneously, and the sampling frequency can be set to 5-10 minutes / time. Finally, the data is transmitted to the data acquisition terminal at the cross-section using a wired cable, and then the terminal uploads it to the watershed data center via 4G / 5G, avoiding the influence of wireless signal obstruction by the surrounding terrain, thereby ensuring the stability of data transmission.

[0077] Water level data can be obtained by deploying fixed submersible hydrostatic level gauges in water bodies such as rivers and reservoirs within the target watershed. In some implementations, the level gauges can be submerged at a fixed depth on stable bedrock or concrete platforms on the left / right bank of the river cross-section, using the Yellow Sea elevation or local benchmark elevation as a reference. They are preferentially deployed on the same side of the flow meter to facilitate spatiotemporal matching of flow and water level data. The water level elevation is calculated by measuring the hydrostatic pressure of the water body, with a sampling frequency set to 1-5 minutes / time. High-frequency acquisition is used to capture rapid rises and falls in water level, such as during flood season. The data is transmitted to the terminal via wired connection along with the flow data and then uploaded to the data center to achieve synchronous correlation between water level and flow data.

[0078] Data from BeiDou-based mobile monitoring equipment can be acquired via mobile sampling in target watersheds using drones or by personnel carrying mobile monitoring terminals integrated with BeiDou dual-mode positioning. For example, in some implementations, the mobile terminal integrates a BeiDou satellite navigation system + Global Positioning System dual-mode positioning module and a portable hydrological sensor. Portable hydrological sensors, such as handheld rain gauges and small current meters, have an IP68 waterproof rating to withstand rain and wading operations and have at least 16GB of offline storage capacity. For areas with insufficient coverage by fixed equipment, such as remote tributaries or temporary engineering areas, 2-3 mobile monitoring routes are planned monthly to ensure coverage of all monitoring grids without fixed equipment. The monitoring team stops at the center point and river edge of each monitoring grid along the route, using BeiDou positioning to obtain the latitude, longitude, and altitude of the sampling points. Instantaneous rainfall (during rainfall) and near-surface humidity are collected using integrated sensors, and land cover type (e.g., farmland, forest land) is manually recorded. After daily monitoring, the terminal data is uploaded to the data center. If offline data exists, it is re-transmitted via USB data cable to ensure no data loss.

[0079] Satellite remote sensing data is obtained by filtering and downloading remote sensing imagery from various types of satellites, followed by preprocessing to acquire relevant information about the target watershed. In some implementations, different satellite data can be selected according to monitoring needs, such as optical remote sensing (Landsat-9, Sentinel-2, etc., for obtaining land cover and vegetation cover), microwave remote sensing (Sentinel-1, etc., for obtaining snow cover area and water body boundaries during cloudy and rainy weather), and hydrological-specific remote sensing (GRACE-FO, etc., for obtaining changes in groundwater storage). Imagery data covering the target watershed and with low / no cloud cover is automatically downloaded or manually filtered through the official platform using API interfaces. Preliminary preprocessing is performed using methods such as radiometric calibration, atmospheric correction, geometric correction, and watershed cropping to form a remote sensing dataset.

[0080] This invention integrates multiple types of data, including rain gauges, snow depth equipment, BeiDou mobile monitoring, and satellite remote sensing, covering all monitoring scenarios from ground-based fixed and mobile patrols to space-based remote sensing. It can effectively fill the monitoring gaps in remote areas and complex terrains. At the same time, data preprocessing achieves format unification and quality optimization, avoiding calculation deviations caused by data chaos or anomalies, and providing more comprehensive and reliable data source support for subsequent modeling and analysis.

[0081] S220: Preprocess the multi-source monitoring dataset to obtain a standardized dataset;

[0082] Preprocessing of multi-source monitoring datasets includes, but is not limited to, outlier removal, missing value completion, and format standardization.

[0083] Outlier removal requires a three-step process: preliminary identification, secondary verification, and final removal, tailored to the characteristics of different data and considering the geographical environment and physical patterns of the target watershed, such as the mountainous areas, alluvial fans, and plains of Xinjiang. This process avoids the accidental deletion of valid data. Preliminary identification can utilize Python's Pandas library or MATLAB's statistical analysis tools to perform statistical calculations on single-type numerical data (such as hourly rainfall data from a rain gauge station), obtaining the mean μ and standard deviation σ. The normal data range is defined as [μ-3σ, μ+3σ]. Data exceeding this range are marked as suspected outliers. For example, for a mountainous rain gauge station's July hourly rainfall data with μ=8mm and σ=5mm, the normal range is [-7mm, 23mm]. Data with hourly rainfall >23mm or <3σ are marked as suspected outliers. Data with a value of 0 mm (no negative rainfall) is marked as a suspected outlier.

[0084] The secondary verification process can combine historical extreme values ​​and physical patterns of the target watershed to establish neighborhood thresholds for various data types, for example, by consulting watershed hydrological yearbooks or historical data from local water resources departments. The secondary verification process can involve comparing initially identified suspected outliers with the neighborhood thresholds. If a value exceeds both the 3σ range and the neighborhood threshold (e.g., hourly rainfall = 55 mm, exceeding both [μ-3σ, μ+3σ] and the 50 mm neighborhood threshold for mountainous areas), it is marked as a confirmed outlier. In some implementations, the neighborhood threshold may include:

[0085] In mountainous areas of Xinjiang, the maximum hourly rainfall typically does not exceed 50 mm, and in plain areas, the maximum daily snow depth increase does not exceed 15 cm. In alluvial fan areas, the maximum range of flow meters is 500 m³ / s, and flow data exceeding this range can be directly identified as abnormal.

[0086] Finally, the confirmed outliers are removed.

[0087] The operation process for other data can be similar to that for rainfall data, the difference being the 3σ range and the neighborhood threshold. Those skilled in the art can calculate and set these according to the actual situation, and this invention does not impose any specific limitations.

[0088] Missing value completion can be handled by classifying the missing scenarios. In this process, the missing situations can be classified first, for example, by the duration of the missing data: short-term ≤1 hour, medium-term 1-24 hours, long-term >24 hours, etc.; or by the data type: continuous numerical values, discrete locations, image pixels, etc. Then, an appropriate method can be selected in combination with the data application scenario to ensure that the deviation rate between the completed data and the actual situation is less than or equal to a preset threshold.

[0089] The preset threshold can be 3% to 8%, and can be set according to the actual situation. This invention does not impose any restrictions.

[0090] In some implementations, data of the same type from the same region can be used first to complete the missing target data. If data of the same type from the same region is insufficient, detection data under similar conditions from the same historical period can be combined. A random forest regression model can be built using Python's Scikit-learn library, and the detection data under similar conditions from the same historical period can be input for prediction and completion. After completion, once the device resumes data acquisition (e.g., BeiDou devices re-upload data), the completed values ​​are compared with the actual values, and the deviation rate is calculated. If the deviation rate is >8%, the model parameters are adjusted (e.g., increasing the number of decision trees in the random forest model) or the reference data source is changed, and completion is repeated to ensure the reliability of the completed data.

[0091] Standardizing the format requires addressing the differences in format, dimension, and spatiotemporal variations among data from different devices. In some implementations, for numerical data conversion, the TXT format data from rain gauges and the Excel format data from flow meters are uniformly converted to CSV format. In Python, the to_csv() function from the Pandas library is used, setting the following fields: device number (e.g., YL-001, where YL represents a rain gauge), monitoring time (UTC+8, format "YYYY-MM-DD HH:MM:SS"), data type (e.g., "hourly rainfall"), value (e.g., "12.5"), unit (e.g., "mm"), and data quality identifier (e.g., "valid", "complete", "after anomaly removal").

[0092] For spatial data conversion, the JSON format location data of Beidou positioning devices is converted to GeoJSON format. In QGIS software, the JSON data is imported, the coordinate system is set to WGS84 (EPSG:4326), the spatial feature type (point feature) is defined, and attribute fields are added: device number (BD-008), monitoring time, location coordinates (latitude and longitude), and data quality identifier. The GeoJSON file is then output.

[0093] For the conversion of remote sensing image data, remote sensing images from different satellites (such as Landsat-8 and Sentinel-2) are uniformly preserved in TIFF format. In ENVI software, radiometric calibration and atmospheric correction are performed, and the resolution is unified (10m for terrain data and 30m for land cover data). Metadata is added: shooting time, sensor type, resolution, and correction method to ensure that the image data can be directly used for subsequent terrain analysis.

[0094] Next, the converted data needs to be standardized in terms of both time dimension and physical quantity units. For time dimension standardization, rain gauge data (5 minutes / time), water level gauge data (10 minutes / time), and satellite remote sensing data (1 day / time) can be standardized to 1 hour / time using high-frequency downsampling and low-frequency upsampling methods. For physical quantity unit standardization, rainfall data can be standardized to millimeters, snow depth data to centimeters, flow rate data to meters cubed per second, water level data to meters, location coordinates to degrees (6 decimal places), runoff velocity to meters per second, and remote sensing image resolution to meters. All units should be uniformly labeled in the CSV format data to avoid unit confusion during calculations.

[0095] S230: Based on the standardized dataset, construct flood evolution characteristic sub-models for each zone within the target watershed; wherein, each zone includes mountainous areas, alluvial fans, and plains;

[0096] In some exemplary embodiments of the present invention, based on the foregoing scheme and the standardized dataset, constructing flood evolution feature sub-models for each sub-region within the target watershed includes:

[0097] From the standardized dataset, unique geographical parameters (such as topographic slope, river curvature, and drainage area), hydrological characteristics (such as historical flood peak flow and confluence time), and underlying surface attributes (such as vegetation coverage and soil permeability coefficient) for each region are selected to establish a regional feature vector library.

[0098] Based on the regional feature vector library and historical flood process data of each region, the initial sub-model of each region is trained and a dynamic correction module is embedded to obtain the flood evolution feature sub-model of each region.

[0099] Here, the established partition feature vector library can first be aligned to a unified index to form partition feature vectors. :

[0100]

[0101] in, A static feature vector representing geographic parameters. The time-varying feature vector representing the hydrological characteristics driving historical processes. This represents the feature vector of the underlying surface attributes. For partition identifier The time indicates the mountainous area. The time indicates an alluvial fan. The time indicates a plain. For unit ID, For time steps.

[0102] The vector library is stored as a columnar database or a binary file, with fields including: partition identifier, cell ID, time index, feature name, feature value, and quality marker (missing / imputation flag).

[0103] The characteristic sub-model of mountain flood evolution is composed of a graph neural network based on river network / slope topology and differentiable runoff generating units (such as Green-Ampt or SCS-CN approximation), and the loss adopts weighted Huber (giving additional weights to peak time and peak quantity).

[0104] For each node, the output at each time step:

[0105]

[0106]

[0107] in, In time step Inside, unit Soil infiltration rate; In time step Inside, unit Rainfall intensity / amount; For unit A set of soil properties; for At any given time, the soil moisture content in the surface layer or the effective infiltration layer; To comprehensively consider the impact of soil matrix suction on the propagation of the infiltration front, approximately The order of m varies depending on the soil. This represents the upper limit of the water conductivity of soil under saturated conditions; sandy soil has a higher limit, while clay soil has a lower limit. Generally, it is... m / s; Indicates at time step Inside, unit The abortion; In time step Inside, unit The amount of snow melt, To simplify evaporation or ignore the terms.

[0108] Loss function of the feature sub-model of flood evolution in mountainous areas for:

[0109]

[0110] in, Indicates the basic regression loss and , This represents the difference between the predicted flow rate and the observed flow rate. express, Indicates predicted flow. Indicates the observed flow rate; Indicates peak quantity loss and , Indicates the observed peak value and , Indicates the predicted peak value and ; Indicates the loss at that time and , Indicates the estimated peak time. Indicates the time of the observed peak; Indicates physical consistency regularity and ; Indicates the mass conservation constraint and , , and These represent different weighting coefficients. In time step Internal control volume Actual / estimated total water volume change For local losses, To control volume, The step size for discrete-time integration; Indicates L2 regularization term, This represents the set of parameters that participated in training and were regularized. and These represent different loss weights.

[0111] The flood evolution characteristic sub-model of an alluvial fan can be designed to include a dual-topological diffusion map and retention-infiltration units. The principal edges of the dual-topological diffusion map are the fan ribs, oriented from the inlet along the maximum slope or fan rib line to the fan edge to simulate the main flood discharge path. The set is denoted as... A bidirectional edge is established between the horizontal edge and its 4 / 8 neighbors to simulate the diffusion and overflow of the flow. Let set [set name missing]. Thus, the resulting bitopological graph is: .

[0112] The outflow function of the storage-infiltration unit is:

[0113]

[0114] in, Representation unit The storage-discharge outflow coefficient, Representation unit outflow power and and Related to roughness coefficient, roughness resistance, and surface roughness type. Indicates at time step Inside, unit The water depth, Representation unit The micro-topography affects the water depth.

[0115] Infiltration function and characteristic sub-model of mountain flood evolution at time step Inside, unit The soil has the same amount of water infiltration.

[0116] After horizontal swapping:

[0117]

[0118] in, Indicates at time step From unit Pointing unit flux; Indicates at time step The proportionality coefficient that converts potential energy difference into flux; Represents a smooth non-negative activation function. Indicates time step Time unit The water depth, Indicates time step Time unit The water depth, Representation unit ground elevation, Representation unit Ground elevation.

[0119] The loss function for the alluvial fan flood evolution characteristic sub-model is:

[0120]

[0121] in, , and These represent the weight coefficients corresponding to different loss functions. This represents the flow fitting loss. Indicates the water depth fitting loss and , For the set of nodes that have been observed, This indicates the number of time steps used to calculate the loss. Indicates at time step Inside, unit The predicted water depth Indicates at time step Inside, unit The observed water depth; Indicates physical consistency regularity and ; Indicates the mass conservation constraint and , , and These represent different weighting coefficients. In time step Internal control volume Actual / estimated total water volume change In time step Inside, unit Local losses, To control volume, The step size for discrete-time integration; Indicates L2 regularization term, This represents the set of parameters that participate in training and are regularized.

[0122] The characteristic sub-model of plain flood evolution can be achieved using a lightweight U-Net network model. The lightweight U-Net network model reduces the number of channels per layer of the original U-Net network model to 1 / 2–1 / 4 of that of the standard U-Net.

[0123] In some implementations, the loss function of the plain flood evolution characteristic sub-model is:

[0124]

[0125] in, Indicates the water depth fitting loss and , For the set of nodes that have been observed, This indicates the number of time steps used to calculate the loss. Indicates at time step Inside, unit The predicted water depth Indicates at time step Inside, unit The observed water depth; The weights for the water depth fitting loss; Indicates intersection, union, and ratio. Indicates the intersection-union ratio weights. Indicates the mass conservation constraint and , The weighting coefficients representing the mass conservation constraints. In time step Internal control volume Actual / estimated total water volume change In time step Inside, unit Local losses, To control volume, The step size for discrete-time integration; Indicates L2 regularization term, This represents the set of parameters that participate in training and are regularized.

[0126] This invention customizes and constructs flood evolution characteristic sub-models based on the unique geographical conditions, hydrological characteristics, and underlying surface properties of each region. This makes the model structure and parameter settings more closely match the actual hydrological process of the region (e.g., focusing on the simulation of runoff velocity in mountainous areas and focusing on the characterization of flood retention effect in plains). This significantly reduces simulation deviations caused by one-size-fits-all modeling, making the restoration and prediction of flood evolution process more consistent with the actual situation.

[0127] S240: A lightweight physical-empirical hybrid model is used to couple the output results of the flood evolution characteristic sub-models of each zone to obtain the coupled calculation results;

[0128] In some exemplary embodiments of the present invention, based on the foregoing scheme, a lightweight physical-empirical hybrid model is used to couple the output results of the flood evolution characteristic sub-models of each partition, and the resulting coupled calculation results include:

[0129] Based on the fundamental laws of watershed hydrodynamics, a water exchange equation between different water zones is constructed.

[0130] The interval water exchange equation includes the control volume at the boundary of each interval. The above establishes the conservation of mass and the approximation of momentum; and for the boundaries of mountainous areas and alluvial fans, the boundaries of alluvial fans and plains, and the boundaries of plains and downstream outlets, the head difference driven or open channel formulas are used for approximation.

[0131] Among them, the conservation of mass and the approximation of momentum are:

[0132]

[0133] The formula for head difference-driven or open channel is approximated as follows:

[0134]

[0135] in, Indicates control volume The water storage capacity at the next time step, Indicates control volume The water storage at the current time step, Indicates at time step Inflow control volume The sum of the total flow Indicates at time step Outflow control volume The sum of the total flow Indicates at time step Area source supply converted to control volume Volumetric flow rate, Indicates at time step Converted to control volume Volumetric flow rate, This represents the step size for discrete-time integration. Indicates at time step From unit Pointing unit flux, Representation unit Pointing unit The effective transmission coefficient, Representation unit The water depth, Representation unit ground elevation, Representation unit The water depth, Representation unit Ground elevation.

[0136] The output results of each sub-model are used as the boundary input of the water exchange equation between the sub-regions. By simplifying the Saint-Venant equations, the hydraulic connection relationship between adjacent sub-regions is solved, and the preliminary coupled flood propagation process (such as the propagation speed of flood waves at the boundary of the sub-regions and the water level connection value) is obtained.

[0137] The simplified Saint-Venant equations are as follows:

[0138]

[0139] in, For time steps, For instantaneous water passage, For vertical spatial coordinates, For water passage section, For hydraulic terms related to water level, For bed surface slope, For roughness loss and , For hydraulic radius, It is the acceleration due to gravity. For lateral inflow, The effective velocity of the lateral incoming water in the mainstream direction.

[0140] The hydraulic connection between adjacent zones is solved by simplifying the Saint-Venant equations. Specifically, the boundary quantities (usually flow rates) given by the mountain flood evolution characteristic sub-model, alluvial fan flood evolution characteristic model, and plain flood evolution characteristic model at the zone boundaries are obtained. Water level / depth and possible lateral inflow Using the simplified Saint-Venant equations as boundary conditions, a one-dimensional hydraulic calculation is performed near the boundary to solve for the hydraulic state and exchange flux that should be consistent on both sides of the boundary at the same time, thus obtaining the preliminary coupled flood propagation process.

[0141] Empirical coefficients extracted from historical flood events are introduced and combined with machine learning methods to correct deviations in the initially coupled flood propagation process. The focus is on correcting physical model errors caused by complex underlying surfaces such as urban building complexes and wetlands, outputting corrected zonal flood hydrographs (including peak flow, peak time, and flood duration). Here, the empirical coefficients include, but are not limited to, equivalent roughness correction coefficients, diffusion / retention correction coefficients, and boundary energy loss coefficients, with spatially defined effective ranges based on the boundary perimeter (e.g., 200–1000 m upstream and downstream). Machine learning methods can be conventional models such as neural network models (e.g., MLP, RNN, TCN, Transformer, etc.), tree models (GBDT, XGBoost, LightGBM, CatBoost, etc.), etc., and are not specifically limited in this invention. The correction target is the residual between the initial coupled output and the observations. ), or directly output the scaling factor of the correction amount ( ).

[0142] For the output results of each sub-model at different time scales (such as minute-level real-time monitoring and hour-level forecasting) and spatial scales (such as river cross-sections and sub-basin units), spatiotemporal interpolation algorithms (such as Kriging interpolation and linear interpolation) are used to unify the scale, generating consistent spatiotemporal distribution data of flood evolution across the entire basin (such as the inundation range and water depth distribution at different times), and outputting the consistent spatiotemporal distribution data of flood evolution across the entire basin as the coupled calculation result.

[0143] Specifically, the Monte Carlo simulation method is used to perform propagation analysis on the uncertainty in the coupled calculation process, and output the confidence interval and probability distribution of the coupled calculation results to quantify the uncertainty in the coupling process.

[0144] This invention employs a lightweight hybrid model combining physical mechanisms and empirical corrections. By simplifying hydrodynamic equations, it quickly solves the hydraulic correlations between different regions. At the same time, it combines historical experience and machine learning to correct biases, significantly reducing the time required for coupled calculations while ensuring computational accuracy. Furthermore, by unifying spatiotemporal scale processing, it generates coherent flood evolution data across the entire region, providing a complete and intuitive computational basis for early warning judgments.

[0145] S250: Based on machine learning algorithms, analyze historical flood case data to generate dynamic early warning thresholds that adapt to the flood evolution feature sub-models of each zone;

[0146] In some exemplary embodiments of the present invention, based on the foregoing scheme, the dynamic early warning threshold for generating a flood evolution feature sub-model adapted to each zone by analyzing historical flood case data using machine learning algorithms includes:

[0147] Key influencing factors for each region were extracted from historical flood case data, including hydrological parameters (such as historical peak flow, flood hydrograph morphology, and confluence time), geographical features (such as topographic slope, river flow capacity, and levee elevation), meteorological conditions (such as rainfall, rainfall intensity, and rainfall duration), and socioeconomic data (such as population density and asset distribution), to construct a region feature matrix.

[0148] Based on the actual disaster severity of each zone in historical flood events (such as inundation loss rate and hazard level), and in conjunction with flood control standards and specifications, each historical case is labeled with a corresponding warning level label (such as blue, yellow, orange, and red warnings).

[0149] The model is trained by using gradient boosting trees (such as GBDT), random forests, or deep learning models (such as LSTM), with the partition feature matrix as input and the warning level label as output.

[0150] A real-time update module is embedded in the trained model, taking current hydrological monitoring data (such as real-time water level and flow rate), short-term meteorological forecast data, and underlying surface change information (such as land use change and water conservancy project scheduling information) as dynamic inputs. The model outputs the specific warning threshold for each zone under the current situation (such as the critical water level value and flow rate value corresponding to a red warning in a certain zone).

[0151] By combining historical extreme events and expert experience, the dynamic early warning thresholds generated by the model are validated to obtain dynamic early warning thresholds adapted to the flood evolution characteristic sub-models of each zone. By setting upper and lower limits for the thresholds and deviation correction coefficients, it is ensured that the thresholds not only conform to the data analysis results of the machine learning model but also meet the safety redundancy requirements of actual flood control work.

[0152] In some implementations, the dynamic warning threshold can be directly output by a trained model: using the partition feature matrix as input, the model is trained to obtain the threshold for different warning levels. Critical water level / flow rate prediction function Calculate the current context vector. The model generates dynamic early warning thresholds The final threshold is obtained by overlaying a safety margin and a deviation correction. .in and Limit updates are performed based on historical extreme events, expert knowledge, and recent observations to ensure operational security, redundancy, and robustness. Among these, Indicates partition unit The model generates dynamic early warning thresholds. Representation unit Correction deviation term, Indicates partition unit Safety margin coefficient.

[0153] This invention analyzes historical cases using machine learning, incorporates current hydrological monitoring data, short-term weather forecasts, and changes in the underlying surface (such as water conservancy project scheduling and land use adjustments) into threshold calculations, and generates dynamic early warning thresholds that are adapted to the real-time scenarios of each zone. At the same time, it verifies the thresholds with historical extreme values ​​and expert experience to ensure their rationality, making early warning judgments more in line with the current actual situation and effectively reducing early warning deviations caused by fixed thresholds.

[0154] S260: Based on the coupling calculation results and the dynamic early warning threshold, generate zonal early warning information.

[0155] In some implementations, the zoning warning information may include one or a combination of the following: warning zoning identifier, warning level, expected duration of risk, expected time of flood arrival, scope of impact, and evacuation recommendations.

[0156] The existence of warning zone identifiers not only solves the resource waste caused by traditional large-scale vague warnings and effectively improves the resource allocation efficiency of target zones, but also directly maps to specific villages and road sections on the map, providing accurate spatial coordinates for subsequent area lockdowns and personnel evacuations, reducing emergency delays caused by ambiguous area positioning. Warning levels provide a tiered technical basis for emergency response, avoiding over- or under-response due to unclear risk levels. In scenarios with simultaneous warnings in multiple areas, the system can automatically prioritize actions based on warning levels, such as prioritizing the push of information for red warning zones to the command terminal, ensuring that emergency instructions for high-risk areas are implemented first, improving overall emergency decision-making efficiency. The expected duration of the risk, the expected arrival time of the flood, and the expected impact range provide technical support for risk avoidance from both temporal and geographical dimensions. Risk avoidance recommendations not only improve public emergency response capabilities but also standardize emergency actions, avoiding confusion and improving the overall emergency coordination efficiency of communities and villages.

[0157] In some implementations, the warning zone identifier can be an administrative unit identifier (e.g., based on existing administrative divisions, directly associated with mature management units such as streets, townships, villages, and communities, facilitating rapid connection at the grassroots level), a geographic network identifier (e.g., dividing the area into standardized grids, such as 1km×1km or 500m×500m, identified by latitude and longitude or grid coding), or a key facility identifier (e.g., marking the warning range around high-risk targets such as bridges, tunnels, low-lying road sections, and large communities as the core).

[0158] In some implementations, the warning levels may include Level 1, Level 2, Level 3, etc., or in other implementations, Level 4, Level 5, etc. Warning indicators can be distinguished by color, sound, etc. For example, when the warning level for a certain area is Level 1, the warning indicator may be red, and the warning sound may be sharp and last for a long time.

[0159] In some implementations, the expected duration of the risk can be expressed in hours or days. The expected arrival time of the flood can be expressed in hours or minutes. The scope of impact can be specified down to specific areas, key locations, etc. Evacuation recommendations can include recommendations for ordinary residents, recommendations for frontline emergency personnel, and recommendations for specific positions (such as transportation departments or water conservancy departments). For example, in some implementations, the evacuation recommendation for the transportation department is: close XX Road and XX Road (flood-prone sections) before 00:00 on day X, set up 'No Entry' signs at intersections, and guide vehicles to detour via XX Road and XX Road.

[0160] In some exemplary embodiments of the present invention, based on the foregoing scheme, and based on the coupling calculation result and the dynamic early warning threshold, generating zonal early warning information includes:

[0161] When the coupling calculation result is greater than 50% or more of the dynamic early warning threshold, early warning information including early warning level, scope of impact, expected duration of risk and risk avoidance suggestions is generated.

[0162] When the coupling calculation result is greater than 20% of the dynamic early warning threshold, or when the coupling calculation result is less than 20% of the dynamic early warning threshold, an early warning message is generated that includes the early warning zone identifier, early warning level, expected flood arrival time, impact range, and evacuation advice; when the coupling calculation result is less than 20% of the dynamic early warning threshold, an early warning message is generated that includes the early warning zone identifier, early warning level, and impact range.

[0163] This invention generates customized early warning information based on the comparison between the partition coupling results and dynamic thresholds. This information includes partition identifiers, warning levels, risk time, impact range, and risk avoidance suggestions. It can directly match the response priorities of different regions (such as focusing on personnel evacuation routes in mountainous areas and focusing on dike patrol priorities in plains). Subsequently, combined with multi-channel dissemination, it can ensure that the early warning information quickly reaches the target population and management departments, providing accurate guidance for disaster emergency response and improving response efficiency.

[0164] In some exemplary embodiments of the present invention, based on the foregoing scheme, after generating zonal early warning information based on the coupled calculation results and the dynamic early warning threshold, the zonal flood prevention early warning method that integrates multi-source data further includes:

[0165] Distribute zone alert information through at least two of the following methods: SMS, broadcast, app push, and display screen.

[0166] According to a second aspect of the present invention, a regional flood prevention and early warning system that integrates multi-source data is also provided, with reference to... Figure 3 As shown, the zonal flood control early warning system that integrates multi-source data includes:

[0167] The data acquisition module 310 is used to acquire a multi-source monitoring dataset within the target watershed, wherein the multi-source monitoring dataset includes rain gauge monitoring data, snow depth monitoring equipment data, flow meter data, water level gauge data, Beidou positioning mobile monitoring equipment data, and satellite remote sensing data;

[0168] Data preprocessing module 320 is used to preprocess the multi-source monitoring dataset to obtain a standardized dataset;

[0169] The sub-model construction module 330 is used to construct flood evolution characteristic sub-models for each partition within the target watershed based on the standardized dataset; wherein, each partition includes mountainous areas, alluvial fans, and plains;

[0170] The coupling calculation module 340 is used to perform coupling calculations on the output results of the flood evolution characteristic sub-model of each partition using a lightweight physical-empirical hybrid model, and obtain the coupling calculation results.

[0171] The dynamic threshold generation module 350 is used to analyze historical flood case data based on machine learning algorithms and generate dynamic early warning thresholds that are adapted to the flood evolution feature sub-model of each zone.

[0172] The early warning information generation module 360 ​​is used to generate zoned early warning information based on the coupled calculation results and the dynamic early warning threshold.

[0173] It should be noted that although several modules of a zoned flood control early warning system integrating multi-source data have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of the present invention, the features and functions of two or more modules or sub-modules described above can be embodied in a single module or unit. Conversely, the features and functions of a single module described above can be further divided and embodied by multiple modules or sub-modules.

[0174] Furthermore, in an exemplary embodiment of the present invention, an electronic device capable of implementing the above-described method for partitioned flood prevention and early warning by fusing multi-source data is also provided.

[0175] Those skilled in the art will understand that various aspects of the present invention can be implemented as systems, methods, or program products. Therefore, various aspects of the present invention can be specifically implemented as entirely hardware embodiments, entirely software embodiments (including firmware, microcode, etc.), or embodiments combining hardware and software aspects, collectively referred to herein as “circuit,” “module,” or “system.”

[0176] The following reference Figure 4 To describe an electronic device 400 according to such an embodiment of the present invention. Figure 4 The electronic device 400 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of the present invention.

[0177] like Figure 4 As shown, the electronic device 400 is manifested in the form of a general-purpose computing device. The components of the electronic device 400 may include, but are not limited to: at least one processing unit 410, at least one storage unit 420, a bus 430 connecting different system components (including storage unit 420 and processing unit 410), and a display unit 440.

[0178] The storage unit stores program code that can be executed by the processing unit 410, causing the processing unit 410 to perform the steps described in the "Exemplary Method" section above, based on various exemplary embodiments of the present invention. For example, the processing unit 410 can perform actions such as... Figure 2 S210: Obtain a multi-source monitoring dataset within the target watershed, wherein the multi-source monitoring dataset includes rain gauge monitoring data, snow depth monitoring equipment data, flow meter data, water level gauge data, BeiDou positioning mobile monitoring equipment data, and satellite remote sensing data; S220: Preprocess the multi-source monitoring dataset to obtain a standardized dataset; S230: Based on the standardized dataset, construct flood evolution characteristic sub-models for each zone within the target watershed; wherein each zone includes mountainous areas, alluvial fans, and plains; S240: Use a lightweight physics-empirical hybrid model to couple the output results of the flood evolution characteristic sub-models for each zone to obtain coupled calculation results; S250: Analyze historical flood case data based on machine learning algorithms to generate dynamic early warning thresholds adapted to the flood evolution characteristic sub-models for each zone; S260: Based on the coupled calculation results and the dynamic early warning thresholds, generate zoned early warning information.

[0179] Storage unit 420 may include readable media in the form of volatile storage units, such as random access memory (RAM) 421 and / or cache memory 422, and may further include read-only memory (ROM) 423.

[0180] Storage unit 420 may also include a program / utility 424 having a set (at least one) of program modules 425, including but not limited to: an operating system, one or more application programs, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment.

[0181] Bus 430 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0182] Electronic device 400 can also communicate with one or more external devices 470 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 400, and / or with any device that enables electronic device 400 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 450. Furthermore, electronic device 400 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 460. As shown, network adapter 460 communicates with other modules of electronic device 400 via bus 430. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 400, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0183] Through the description of the above embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions of the embodiments of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, portable hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, terminal device, or network device, etc.) to execute the methods according to the embodiments of the present invention.

[0184] In exemplary embodiments of the present invention, a computer-readable storage medium is also provided, on which a program product capable of implementing the methods described above is stored. In some possible embodiments, various aspects of the present invention may also be implemented as a program product comprising program code, which, when the program product is run on a terminal device, causes the terminal device to perform the steps of the various exemplary embodiments of the present invention described in the "Exemplary Methods" section above.

[0185] refer to Figure 5 As shown, a program product 500 for implementing the above-described method for merging multi-source data for regional flood warning according to an embodiment of the present invention is described. It may employ a portable compact disc read-only memory (CD-ROM) and include program code, and may run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited thereto. In the present invention, the readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0186] The program product may employ any combination of one or more readable storage media. Readable storage media may be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0187] Program code for performing the operations of this invention can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0188] Furthermore, the above figures are merely illustrative of the processes included in the method according to exemplary embodiments of the present invention, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Additionally, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0189] Through the description of the above embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions of the embodiments of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, portable hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, touch terminal, or network device, etc.) to execute the methods according to the embodiments of the present invention.

[0190] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. The invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the claims.

[0191] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for regional flood prevention and early warning that integrates multi-source data, characterized in that, include: Obtain multi-source monitoring datasets for each zone within the target watershed, wherein the multi-source monitoring datasets include rain gauge monitoring data, snow depth monitoring equipment data, flow meter data, water level gauge data, BeiDou positioning mobile monitoring equipment data, and satellite remote sensing data; The multi-source monitoring dataset is preprocessed to obtain a standardized dataset; Based on the standardized dataset, flood evolution feature sub-models for each zone within the target watershed are constructed; wherein, each zone includes mountainous areas, alluvial fans, and plains; A lightweight physics-empirical hybrid model was used to couple the output results of the flood evolution characteristic sub-models of each zone to obtain the coupled calculation results; Based on machine learning algorithms, analyze historical flood case data to generate dynamic early warning thresholds for flood evolution feature sub-models adapted to each zone; Based on the coupling calculation results and the dynamic early warning threshold, zonal early warning information is generated; Based on the standardized dataset, flood evolution feature sub-models for each zone within the target watershed are constructed, including: The unique geographical parameters, hydrological features, and underlying surface attributes of each region are selected from the standardized dataset to establish a regional feature vector library. Based on the regional feature vector library and historical flood process data of each region, the initial sub-model of each region is trained and a dynamic correction module is embedded to obtain the flood evolution feature sub-model of each region. A lightweight physics-empirical hybrid model was used to couple the output results of the flood evolution characteristic sub-models for each zone, and the resulting coupled calculations included: Based on the fundamental laws of watershed hydrodynamics, a water exchange equation between different intervals is constructed. The output results of each sub-model are used as the boundary input of the water exchange equation between the sub-regions. The hydraulic connection relationship between adjacent sub-regions is solved by simplifying the Saint-Venant equations, and the preliminary coupled flood propagation process is obtained. By introducing empirical coefficients extracted from historical flood processes and combining them with machine learning methods, the bias of the initially coupled flood propagation process is corrected, and the corrected zonal flood process lines are output. For the output results of each sub-model at different time and spatial scales, a spatiotemporal interpolation algorithm is used to unify the scale, generate consistent spatiotemporal distribution data of flood evolution across the entire watershed, and output the consistent spatiotemporal distribution data of flood evolution across the entire watershed as the coupled calculation result; Among them, the uncertainty in the coupling calculation process is analyzed by Monte Carlo simulation, and the confidence interval and probability distribution of the coupling calculation results are output to quantify the uncertainty in the coupling process; Based on machine learning algorithms, historical flood case data is analyzed to generate dynamic early warning thresholds for flood evolution characteristic sub-models adapted to each region, including: Key influencing factors for each region were extracted from historical flood case data, and a region feature matrix was constructed. Based on the actual degree of damage to each zone in historical flood events, and in accordance with flood control standards and specifications, a corresponding warning level label is marked for each historical case; The model is trained by using gradient boosting tree, random forest or deep learning model, with the partition feature matrix as input and the warning level label as output. A real-time update module is embedded in the trained model, taking current hydrological monitoring data, short-term meteorological forecast data and underlying surface change information as dynamic inputs, and the model outputs the specific warning thresholds for each zone under the current situation. By combining historical extreme events and expert experience, the dynamic early warning thresholds generated by the model are verified to obtain dynamic early warning thresholds that are adapted to the flood evolution characteristic sub-models of each zone.

2. The method for regional flood prevention and early warning by fusing multi-source data according to claim 1, characterized in that, The zoning early warning information includes one or more of the following: early warning zone identifier, early warning level, expected duration of risk, expected time of flood arrival, scope of impact, and risk avoidance recommendations.

3. The method for regional flood prevention and early warning by fusing multi-source data according to claim 1, characterized in that, Based on the coupling calculation results and the dynamic early warning threshold, the generated zonal early warning information includes: When the coupling calculation result is greater than 50% or more of the dynamic early warning threshold, early warning information including early warning level, scope of impact, expected duration of risk and risk avoidance suggestions is generated. When the coupling calculation result is greater than 20% of the dynamic early warning threshold, or when the coupling calculation result is less than 20% of the dynamic early warning threshold, an early warning message is generated that includes the early warning zone identifier, early warning level, expected flood arrival time, impact range, and evacuation advice; when the coupling calculation result is less than 20% of the dynamic early warning threshold, an early warning message is generated that includes the early warning zone identifier, early warning level, and impact range.

4. The method for regional flood prevention and early warning by fusing multi-source data according to claim 1, characterized in that, After generating zonal early warning information based on the coupled calculation results and the dynamic early warning threshold, the zonal flood prevention early warning method that integrates multi-source data further includes: Distribute zone alert information through at least two of the following methods: SMS, broadcast, app push, and display screen.

5. A regional flood control early warning system integrating multi-source data, characterized in that, include: The data acquisition module is used to acquire multi-source monitoring datasets within the target watershed, wherein the multi-source monitoring datasets include rain gauge monitoring data, snow depth monitoring equipment data, flow meter data, water level gauge data, BeiDou positioning mobile monitoring equipment data, and satellite remote sensing data; The data preprocessing module is used to preprocess the multi-source monitoring dataset to obtain a standardized dataset; The sub-model construction module is used to construct flood evolution characteristic sub-models for each partition within the target watershed based on the standardized dataset; wherein, each partition includes mountainous areas, alluvial fans, and plains; The coupling calculation module is used to perform coupling calculations on the output results of the flood evolution characteristic sub-models of each partition using a lightweight physical-empirical hybrid model, and obtain the coupling calculation results. The dynamic threshold generation module is used to analyze historical flood case data based on machine learning algorithms and generate dynamic early warning thresholds that are adapted to the flood evolution feature sub-models of each zone. The early warning information generation module is used to generate zoned early warning information based on the coupled calculation results and the dynamic early warning threshold; Based on the standardized dataset, flood evolution feature sub-models for each zone within the target watershed are constructed, including: The unique geographical parameters, hydrological features, and underlying surface attributes of each region are selected from the standardized dataset to establish a regional feature vector library. Based on the regional feature vector library and historical flood process data of each region, the initial sub-model of each region is trained and a dynamic correction module is embedded to obtain the flood evolution feature sub-model of each region. A lightweight physics-empirical hybrid model was used to couple the output results of the flood evolution characteristic sub-models for each zone, and the resulting coupled calculations included: Based on the fundamental laws of watershed hydrodynamics, a water exchange equation between different intervals is constructed. The output results of each sub-model are used as the boundary input of the water exchange equation between the sub-regions. The hydraulic connection relationship between adjacent sub-regions is solved by simplifying the Saint-Venant equations, and the preliminary coupled flood propagation process is obtained. By introducing empirical coefficients extracted from historical flood processes and combining them with machine learning methods, the bias of the initially coupled flood propagation process is corrected, and the corrected zonal flood process lines are output. For the output results of each sub-model at different time and spatial scales, a spatiotemporal interpolation algorithm is used to unify the scale, generate consistent spatiotemporal distribution data of flood evolution across the entire watershed, and output the consistent spatiotemporal distribution data of flood evolution across the entire watershed as the coupled calculation result; Among them, the uncertainty in the coupling calculation process is analyzed by Monte Carlo simulation, and the confidence interval and probability distribution of the coupling calculation results are output to quantify the uncertainty in the coupling process; Based on machine learning algorithms, historical flood case data is analyzed to generate dynamic early warning thresholds for flood evolution characteristic sub-models adapted to each region, including: Key influencing factors for each region were extracted from historical flood case data, and a region feature matrix was constructed. Based on the actual degree of damage to each zone in historical flood events, and in accordance with flood control standards and specifications, a corresponding warning level label is marked for each historical case; The model is trained by using gradient boosting tree, random forest or deep learning model, with the partition feature matrix as input and the warning level label as output. A real-time update module is embedded in the trained model, taking current hydrological monitoring data, short-term meteorological forecast data and underlying surface change information as dynamic inputs, and the model outputs the specific warning thresholds for each zone under the current situation. By combining historical extreme events and expert experience, the dynamic early warning thresholds generated by the model are verified to obtain dynamic early warning thresholds that are adapted to the flood evolution characteristic sub-models of each zone.

6. An electronic device, characterized in that, include: processor; as well as A memory storing computer-readable instructions that, when executed by the processor, implement the method as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1 to 4.

Images