Multi-source information fused flood disaster risk intelligent early warning method

Abstract

The invention discloses a flood disaster risk intelligent early warning method fusing multi-source information, and relates to the technical field of flood disaster prediction.The method comprises the steps that multi-source data of meteorology, hydrology, terrain, remote sensing, ground monitoring terminals and the like are obtained, and a unified data set is constructed through time synchronization, space resampling, abnormal value elimination and missing data filling; constructing multi-dimensional spatio-temporal features based on factors such as rainfall process, topographic features, surface coverage, drainage capacity, river network structure and soil humidity; inputting the spatial-temporal characteristics into a multi-source information fusion model, and predicting a flood occurrence probability in a future time period; flood risk indexes are generated according to the prediction probability, real-time monitoring and historical samples, and regional risks are divided into different grades; early warning information is automatically generated and issued according to the risk level; and the early warning result is fed back and corrected through field monitoring and patrol data, so that self-adaptive updating of early warning parameters and a model structure is realized.

Description

Technical Field

[0001] This invention relates to the field of flood disaster prediction technology, and more specifically, to an intelligent early warning method for flood disaster risks that integrates multi-source information. Background Technology

[0002] Floods are among the most common and destructive natural disasters globally, significantly impacting urban infrastructure, agricultural production, and human lives. With climate change and the increasing frequency of extreme weather events, short-duration heavy rainfall and localized torrential downpours are exacerbating urban flooding, flash floods, and regional waterlogging. Traditional flood control systems rely primarily on observation systems comprised of limited monitoring points such as rain gauges and water level stations. These systems cannot comprehensively depict rapidly changing rainfall processes and small-scale runoff responses, resulting in widespread delays in early warning and insufficient coverage.

[0003] Current flood warning technologies typically rely on a single data source for judgment. For example, empirical warning methods based on rainfall thresholds depend on preset rainfall levels to trigger warnings, but they struggle to consider the combined effects of factors such as topography, soil moisture, and drainage capacity, easily leading to false alarms or missed warnings. Warnings based on water level thresholds depend on the distribution of hydrological stations, but the monitoring density in many cities and small and medium-sized river areas is insufficient, making it impossible to achieve spatially continuous risk assessment. While water body identification technology based on remote sensing imagery has advantages in monitoring water surface changes, it is limited by timeliness and weather conditions, making it difficult to use for real-time warnings. Overall, existing warning methods generally suffer from problems such as a single information source, untimely data updates, insufficient spatial resolution, and models that cannot cover surface response mechanisms, making it difficult to meet the current demand for short-term, refined, and highly reliable flood warnings.

[0004] With the development of IoT, big data, and AI technologies, multi-source data fusion has become an important direction for improving flood early warning capabilities. Data from radar rainfall, remote sensing imagery, digital elevation models, ground sensors, and drone patrols can reflect regional hydrological response characteristics from different perspectives. Effective integration of multi-source data, and the use of data-driven models combined with hydrogeomorphological characteristics to construct a flood risk prediction mechanism, will significantly improve the timeliness and accuracy of early warnings. However, at present, there is still a lack of a systematic approach encompassing multi-source data preprocessing, spatiotemporal feature construction, multi-model fusion and extrapolation, risk index generation, and dynamic correction. This is especially true in rapid response scenarios involving urban flooding and floods in small and medium-sized rivers, where a general technical framework capable of providing high-precision risk predictions on a minute-by-minute scale is needed.

[0005] Therefore, there is an urgent need for an intelligent early warning method for flood disaster risks that integrates information from multiple sources to solve these problems. Summary of the Invention

[0006] The purpose of this invention is to solve the technical problems mentioned in the background section and to provide an intelligent early warning method for flood disaster risks that integrates multi-source information.

[0007] The above-mentioned objective of the present invention is achieved through the following technical solution:

[0008] A method for intelligent early warning of flood disaster risks that integrates multi-source information includes the following steps:

[0009] S1. Acquire multi-source data for the target area. The multi-source basic data includes meteorological data, hydrological data, topographic data, remote sensing data, and ground monitoring terminal data.

[0010] S2. Perform time synchronization, spatial resampling, outlier removal, missing data imputation, and data format conversion on the multi-source basic data to construct a unified dataset that can be used for risk analysis;

[0011] S3. Based on rainfall processes, topographic features, land cover types, regional drainage capacity, river network structure, and soil moisture conditions, a multi-dimensional spatiotemporal feature set reflecting regional runoff generation, confluence, water accumulation, and river response capabilities is constructed.

[0012] S4. Input the multi-dimensional spatiotemporal feature set into the multi-source information fusion model. The multi-source information fusion model predicts the probability of regional flood disasters within the target time window based on rainfall evolution trends, surface runoff response, river network water level propagation relationships, and remote sensing observation changes.

[0013] S5. Based on predicted probabilities, historical event samples, and real-time monitoring of changes, generate a flood risk index and divide the target area into different risk levels according to the preset risk classification standards.

[0014] S6. Automatically generate early warning content based on risk level, and release flood warning information through mobile terminals, management terminals or broadcast systems, and provide spatial distribution maps of key risk points, possible impact range and suggested defense measures;

[0015] S7. Based on data from on-site monitoring equipment, video surveillance, and user reports, the early warning results are fed back and verified, and the early warning parameters, threshold ranges, and model structure are dynamically updated.

[0016] As a preferred technical solution of the present invention, the meteorological data includes, but is not limited to, radar echo reflectivity, minute-level short-term rainfall forecast, hourly grid rainfall prediction, satellite cloud imagery, measured rain gauge data, and regional meteorological reanalysis data.

[0017] As a preferred technical solution of the present invention, the hydrological data includes river water level, water surface velocity, levee stress state, dam regulation conditions, reservoir inflow and outflow, and observation records of historical flood events in the basin.

[0018] As a preferred technical solution of the present invention, the terrain data includes digital elevation model, slope and aspect, surface roughness, distribution of depressions, drainage zone boundaries, tributary confluence points and watershed morphological characteristics.

[0019] As a preferred technical solution of the present invention, the remote sensing data includes optical remote sensing images, radar remote sensing images, multispectral remote sensing images, surface humidity inversion products, water body boundary identification results, and the distribution of the proportion of impervious surfaces in urban built-up areas.

[0020] As a preferred technical solution of the present invention, the ground monitoring terminal includes a water accumulation sensor, a low-power water level gauge, a water quality monitoring node, a road monitoring camera, a mobile drone patrol terminal, and flood information reported by the public.

[0021] As a preferred technical solution of the present invention, constructing the spatiotemporal feature set includes the following:

[0022] (1) Construct indicators of rainfall intensity change, rainfall center movement speed, cumulative rainfall and short-term suddenness based on rainfall sequence;

[0023] (2) Based on topography and land use, the catchment area, runoff obstruction degree, regional drainage capacity and surface storage potential are generated;

[0024] (3) Extract water level response velocity, tributary replenishment intensity, flood peak propagation path and sensitive river section identification results based on river network structure;

[0025] (4) Based on remote sensing images, construct the water body boundary expansion rate, surface humidity change frequency, water accumulation probability distribution and abnormal surface reflection characteristics.

[0026] As a preferred technical solution of the present invention, the multi-source information fusion model includes a time prediction module, a spatial correlation module, and a risk assessment module, used for:

[0027] (1) Integrating rainfall evolution trends and meteorological dynamic characteristics to predict rainfall changes in future periods;

[0028] (2) Estimate the rate of water level rise, the direction of flood peak propagation, and the accumulation of water volume by combining the river network topology and topographic confluence relationship;

[0029] (3) Assess the overall risk probability of future floods based on historical samples and real-time monitoring information.

[0030] As a preferred technical solution of the present invention, the risk level classification includes at least four levels, namely low risk, relatively high risk, high risk and extremely high risk, and is marked with different colors or warning methods to guide different levels of disaster prevention and mitigation measures, including personnel evacuation, traffic control, drainage scheduling and emergency rescue deployment.

[0031] As a preferred technical solution of the present invention, the method further includes an adaptive update step for the early warning system. By comparing the model prediction results with the actual monitoring data, the source of prediction deviation is identified, and the data processing flow, risk classification rules, monitoring weight parameters and model structure are continuously optimized, so that the entire early warning system has the ability to learn and correct itself under long-term operating conditions.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] 1. The intelligent early warning method for flood disaster risks proposed in this invention, which integrates multi-source information, can overcome the limitations of traditional flood early warning methods, such as reliance on a single data source, delayed response, and coarse spatial characterization, and achieve comprehensive judgment across multiple dimensions and spatiotemporal scales. By integrating multi-source data such as meteorological, hydrological, topographic, remote sensing, and ground monitoring terminals, this invention constructs more detailed regional hydrological response characteristics, which can reflect key processes such as rainfall, runoff generation, confluence, river water levels, and surface water accumulation. This significantly enhances the early warning model's response capability to complex scenarios (such as short-term torrential rain, localized heavy rainfall, and water accumulation in urban hardened areas).

[0034] 2. This invention utilizes a multi-source information fusion model formed by introducing a time prediction module, a spatial correlation module, and a risk assessment module to achieve a chain-like logical deduction from rainfall evolution and surface runoff to river network response. This model combines physical mechanism constraints with data-driven capabilities, capturing key elements such as rainfall center movement, flood peak propagation paths, and the storage characteristics of depressions, making flood risk prediction more consistent with the inherent laws of hydrological and hydraulic processes. Simultaneously, by constructing dimensionless rainfall indicators, runoff potential indicators, drainage capacity indicators, river channel exceedance indicators, and waterlogging risk indicators, the consistency and stability of model calculations are guaranteed, reducing the impact of regional differences on early warning results.

[0035] 3. This invention further introduces a dynamic feedback correction mechanism. By comparing the predicted results with actual monitoring data, it continuously corrects the model weight parameters and risk classification thresholds, enabling the early warning system to have adaptive learning capabilities and achieve closed-loop operation of prediction, verification, and updating. This mechanism not only effectively reduces false alarms and missed alarms but also continuously optimizes the model as urban drainage system construction changes, underlying surface evolution, and extreme weather frequency changes, ensuring the invention maintains high confidence and stability in long-term operation. By intelligently generating flood risk levels and corresponding visualized early warning information, this invention can provide high-precision, real-time decision support for urban flood control scheduling, flood prevention in small and medium-sized rivers, and emergency command, demonstrating significant engineering application value and promising prospects for widespread application. Attached Figure Description

[0036] Figure 1 This is a flowchart of an intelligent early warning method for flood disaster risks that integrates multi-source information according to the present invention;

[0037] Figure 2 This is a logic diagram of an intelligent early warning method for flood disaster risks that integrates multi-source information, as proposed in this invention. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of this invention clearer, the following description is provided in conjunction with embodiments and appendices. Figures 1-2 The present invention will be further described in detail below. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0039] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments, but the present invention is not limited to these embodiments. Equivalent modifications made by those skilled in the art without departing from the principles of the present invention should fall within the protection scope of the present invention.

[0040] The following detailed description, with reference to specific embodiments, provides a further explanation of the intelligent early warning method for flood disaster risks that integrates multi-source information proposed in this invention. The method of this invention can be deployed in urban flood control command platforms, river basin flood control systems, or provincial / national natural disaster monitoring and early warning platforms, and achieves automated operation through software systems and databases.

[0041] In this embodiment of the invention, the method of the invention uses gridded spatial units as the basic analysis units, divides the target area into several regular grids, each grid unit is denoted as i, and uses discrete time steps for rolling calculation, each time step is denoted as t.

[0042] In actual operation, the method of the present invention may include the following steps, corresponding to steps S1 to S7 in the inventive content of the present invention.

[0043] For ease of explanation, an initialization step can be added before the method begins, for the purpose of illustrating this implementation: Before step S1, first determine the spatial range of the target area, read the basic geographic information data and historical flood event database of the area, and configure the warning time scale, for example, select 10 minutes or 1 hour as the time step.

[0044] In step S1, multi-source basic data of the target area are acquired. The multi-source basic data includes meteorological data, hydrological data, topographic data, remote sensing data, and ground monitoring terminal data.

[0045] Specifically, the meteorological data includes radar echo reflectance, minute-level short-term rainfall forecasts, hourly grid rainfall predictions, satellite cloud images, measured rain gauge data, and regional meteorological reanalysis data;

[0046] Hydrological data include river level, surface velocity, levee stress status, dam regulation conditions, reservoir inflow and outflow, and observation records of historical flood events in the basin.

[0047] Topographic data includes digital elevation model (DEM), slope and aspect, surface roughness, distribution of depressions, drainage zone boundaries, tributary confluence locations, and watershed morphological characteristics.

[0048] Remote sensing data includes optical remote sensing images, radar remote sensing images, multispectral remote sensing images, surface humidity inversion products, water body boundary identification results, and the distribution of the proportion of impervious surfaces in urban built-up areas.

[0049] Ground monitoring terminal data includes water depth data collected by water accumulation sensors, river water level data collected by low-power water level gauges, water quality monitoring node data, image recognition results from road surveillance cameras, on-site images collected by mobile drone patrol terminals, and flood information reported by the public via mobile devices.

[0050] In step S2, the aforementioned multi-source basic data undergoes unified preprocessing. Specifically, this includes: time synchronization of data from different clock sources, aligning radar, remote sensing, and ground monitoring data to a unified time step t; spatial resampling of data with different spatial resolutions to represent them on a unified grid cell i; outlier removal using statistical methods or physical constraints; completion of missing data using interpolation or historical similar event imputation methods; and normalization and data format conversion of data from different sources and units, ultimately constructing a unified dataset suitable for risk analysis.

[0051] To ensure dimensional consistency in subsequent formulas, this implementation method preferably converts key physical quantities into dimensionless indices. For example, the rainfall, soil moisture content, water level, and water depth of each grid i at time step t can be standardized to construct normalized basic features.

[0052] In step S3, based on rainfall process, topographic features, land cover type, regional drainage capacity, river network structure and soil moisture status, a multi-dimensional spatiotemporal feature set reflecting regional runoff generation, confluence, water accumulation and river response capacity is constructed.

[0053] In one specific embodiment, a normalized rainfall intensity index can be constructed first. Its definition is:

[0054] ;

[0055] in, Represents grid cells At time step The normalized index of rainfall intensity is a dimensionless number; Represents grid cells At time step The actual rainfall, in millimeters; This represents the minimum rainfall within a historical sample or a defined statistical interval, expressed in millimeters. This represents the maximum rainfall within a historical sample or a defined statistical interval, expressed in millimeters. Through the above normalization process, The value of is limited to between 0 and 1, and it is a dimensionless quantity.

[0056] When considering the combined effects of soil moisture and impermeable surfaces, a runoff potential index can be defined. ,For example:

[0057] ;

[0058] in, Represents grid cells At time step The potential for runoff generation is a dimensionless number. Represents grid cells At time step The normalized index of soil moisture, ranging from 0 to 1, is dimensionless; Represents grid cells The proportion of impermeable surfaces, such as the proportion of hardened urban surfaces, ranges from 0 to 1 and is dimensionless. and The weighting coefficients reflect the relative influence of soil moisture and impermeability on runoff generation; both are dimensionless parameters.

[0059] In terms of topography and drainage capacity, a comprehensive index of runoff collection and drainage capacity can be defined. ,For example:

[0060] ;

[0061] in, Represents grid cells The drainage and storage capacity index is a dimensionless number; the larger the value, the better the drainage conditions. Represents grid cells The normalized value of the catchment area reflects the influence of the catchment area on the runoff, and is dimensionless. Represents grid cells The normalized value of drainage facility capacity, such as the comprehensive index of stormwater pipe network density, cross-sectional capacity, etc., is dimensionless; Represents grid cells The normalized value of the depth or degree of closure of the depression; the larger the value, the easier it is for water to accumulate. Dimensionless. , , These are weighting coefficients, all of which are dimensionless parameters and can be obtained through historical event inversion or model training.

[0062] Regarding river response, river exceedance indicators can be constructed based on the relative relationship between water level and warning water level. ,For example:

[0063] ;

[0064] in, This indicates the water level exceeding the warning level corresponding to a river channel unit; it is dimensionless. Indicates the location of the river channel Time step Measured or predicted water level, in meters; Indicates the location of the river channel The safety control level or warning level at the location, in meters; Indicates the location of the river channel The maximum permissible water level or historical extreme water level at the location, in meters. At that time, it can be Cut off to 0 when The value can be truncated to 1 to ensure the stable range of the dimensionless variable.

[0065] Regarding water accumulation and surface response, surface water accumulation risk indicators can be constructed based on surface water accumulation sensor data and remote sensing water body information. ,For example:

[0066] ;

[0067] in, Represents grid cells At time step The surface water accumulation risk index is dimensionless. This represents the normalized value of the water depth collected by the water depth sensor, and is dimensionless. Normalized value representing the extent of water body boundary expansion or water coverage ratio extracted from remote sensing images; dimensionless. , These are weighting coefficients, all dimensionless parameters, used to balance the influence of ground sensor and remote sensing information. By constructing these multiple indicators, each grid cell can be formed. Each time step Multidimensional spatiotemporal feature vectors:

[0068] ;

[0069] in, Represents grid cells At time step The multidimensional feature vector contains multiple normalized dimensionless features; the items in the curly braces represent feature indicators such as rainfall, runoff, drainage capacity, river channel exceeding warning level, and surface water accumulation, all of which are dimensionless quantities.

[0070] In step S4, the aforementioned multi-dimensional spatiotemporal feature set is input into the multi-source information fusion model. The multi-source information fusion model may include a time prediction module, a spatial correlation module, and a risk assessment module. The time prediction module utilizes historical time series data. Predicting future time steps The key variables; the spatial correlation module considers the river network topology, confluence direction and mutual influence between adjacent grids; the risk assessment module calculates the probability of flood disasters based on time prediction results and spatial propagation relationships.

[0071] In one implementation, the risk probability can be expressed as:

[0072] ;

[0073] in, Represents grid cells At time step The probability of flooding occurs, with a range of values. , is a dimensionless number; This indicates mapping the result of a linear combination to... Standardized functions of an interval, such as logistic functions or other monotonic mapping functions, are themselves dimensionless operators; These are weighting coefficients, all of which are dimensionless parameters; All of these are the aforementioned dimensionless characteristic indicators; Representation and grid cells Corresponding historical flood event similarity indicators, such as normalization based on whether the region has experienced floods in historical events, the frequency of occurrence, and the severity, can be used, with a range of... , dimensionless.

[0074] In step S5, a flood risk index is generated based on predicted probabilities, historical event samples, and real-time monitoring changes. The target area is then divided into different risk levels according to a preset risk grading standard. In one specific implementation, the flood risk index can be... Defined as:

[0075] ;

[0076] in, Represents grid cells At time step The flood risk index typically ranges from [value range missing]. , is a dimensionless number; The aforementioned probability of flooding is dimensionless.

[0077] Subsequently, tiered thresholds can be set based on the risk index, for example:

[0078] when At that time, it was determined to be low risk;

[0079] when At that time, it was determined to be a high-risk situation;

[0080] when At that time, it was determined to be high risk;

[0081] when At that time, it was determined to be extremely high risk.

[0082] The above thresholds can be adjusted according to the flood control standards and management requirements of different regions without affecting the dimensionless nature of the index itself.

[0083] In step S6, early warning content is automatically generated based on the risk level. This content includes: a spatial distribution map of areas at each risk level, potentially affected roads and residential areas, locations of critical infrastructure, and corresponding recommended preventative measures. The system can disseminate flood warning information through various channels such as mobile apps, web management platforms, SMS, broadcasts, and government intranets. For example, it automatically recommends advance evacuation in extremely high-risk areas, implements traffic control and drainage facility inspections in high-risk areas, strengthens patrols in relatively high-risk areas, and maintains routine monitoring in low-risk areas.

[0084] In step S7, the early warning results are verified based on data from on-site monitoring equipment, video surveillance, and public reports. The system can record the risk index distribution and actual flooding situation at the time of each early warning issuance, and compare and analyze the deviation between the prediction and the actual situation. For this purpose, a prediction deviation index can be defined. ,For example:

[0085] ;

[0086] in, Indicates at time step The overall prediction deviation index is a dimensionless number. This represents the total number of grid cells participating in the statistics, and is a positive integer. Represents grid cells At time step The actual occurrence of floods can be recorded as 1 for occurrence and 0 for non-occurrence, and is a dimensionless number. Indicates based on risk probability or risk index The inferred flood occurrence judgment, for example, converts the probability into an estimate of 0 or 1 using a certain threshold, and is a dimensionless number. Based on the deviation index... The system can adjust the warning parameters, threshold ranges, and weight parameters of the multi-source information fusion model based on these changes. , , It allows for dynamic adjustments; and the weight parameters in feature construction can be updated when needed. This enables the early warning system to achieve adaptive optimization, allowing the system to learn and correct itself under long-term operating conditions.

[0087] As can be seen from the above implementation methods, the intelligent early warning method for flood disaster risk that integrates multi-source information proposed in this invention constructs a complete technology from multi-source data acquisition, unified preprocessing, multi-dimensional spatiotemporal feature construction, multi-source fusion modeling, risk index calculation, early warning release to feedback correction, which can provide a unified technical foundation for urban waterlogging early warning, flood early warning of small and medium-sized rivers, and regional flood disaster prevention and control.

[0088] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent early warning of flood disaster risks that integrates multi-source information, characterized in that, Includes the following steps: S1. Acquire multi-source data for the target area. The multi-source basic data includes meteorological data, hydrological data, topographic data, remote sensing data, and ground monitoring terminal data. S2. Perform time synchronization, spatial resampling, outlier removal, missing data imputation, and data format conversion on the multi-source basic data to construct a unified dataset that can be used for risk analysis; S3. Based on rainfall processes, topographic features, land cover types, regional drainage capacity, river network structure, and soil moisture conditions, a multi-dimensional spatiotemporal feature set reflecting regional runoff generation, confluence, water accumulation, and river response capabilities is constructed. S4. Input the multi-dimensional spatiotemporal feature set into the multi-source information fusion model. The multi-source information fusion model predicts the probability of regional flood disasters within the target time window based on rainfall evolution trends, surface runoff response, river network water level propagation relationships, and remote sensing observation changes. S5. Based on predicted probabilities, historical event samples, and real-time monitoring of changes, generate a flood risk index and divide the target area into different risk levels according to the preset risk classification standards. S6. Automatically generate early warning content based on risk level, and release flood warning information through mobile terminals, management terminals or broadcast systems, and provide spatial distribution maps of key risk points, possible impact range and suggested defense measures; S7. Based on data from on-site monitoring equipment, video surveillance, and user reports, the early warning results are fed back and verified, and the early warning parameters, threshold ranges, and model structure are dynamically updated.

2. The intelligent early warning method for flood disaster risk fusion based on multi-source information as described in claim 1, characterized in that, The meteorological data includes, but is not limited to, radar echo reflectivity, minute-level short-term rainfall forecasts, hourly grid rainfall predictions, satellite cloud images, measured rain gauge data, and regional meteorological reanalysis data.

3. The intelligent early warning method for flood disaster risk fusion based on multi-source information as described in claim 1, characterized in that, The hydrological data includes river water level, surface flow velocity, levee stress status, dam regulation conditions, reservoir inflow and outflow, and observation records of historical flood events within the basin.

4. The intelligent early warning method for flood disaster risk fusion based on multi-source information as described in claim 1, characterized in that, The topographic data includes digital elevation models, slope and aspect, surface roughness, distribution of depressions, drainage zone boundaries, tributary confluence locations, and watershed morphological characteristics.

5. The intelligent early warning method for flood disaster risk fusion based on multi-source information according to claim 1, characterized in that, The remote sensing data includes optical remote sensing images, radar remote sensing images, multispectral remote sensing images, surface humidity inversion products, water body boundary identification results, and the distribution of the proportion of impervious surfaces in urban built-up areas.

6. The intelligent early warning method for flood disaster risk fusion based on multi-source information according to claim 1, characterized in that, The ground monitoring terminal includes water accumulation sensors, low-power water level gauges, water quality monitoring nodes, road surveillance cameras, mobile drone patrol terminals, and flood information reported by the public.

7. The intelligent early warning method for flood disaster risk fusion based on multi-source information according to claim 1, characterized in that, Constructing the spatiotemporal feature set includes the following: (1) Construct indicators of rainfall intensity change, rainfall center movement speed, cumulative rainfall and short-term suddenness based on rainfall sequence; (2) Based on topography and land use, the catchment area, runoff obstruction degree, regional drainage capacity and surface storage potential are generated; (3) Extract water level response velocity, tributary replenishment intensity, flood peak propagation path and sensitive river section identification results based on river network structure; (4) Based on remote sensing images, construct the water body boundary expansion rate, surface humidity change frequency, water accumulation probability distribution and abnormal surface reflection characteristics.

8. The intelligent early warning method for flood disaster risk fusion based on multi-source information according to claim 1, characterized in that, The multi-source information fusion model includes a time prediction module, a spatial correlation module, and a risk assessment module, used for: (1) Integrating rainfall evolution trends and meteorological dynamic characteristics to predict rainfall changes in future periods; (2) Estimate the rate of water level rise, the direction of flood peak propagation, and the accumulation of water volume by combining the river network topology and topographic confluence relationship; (3) Assess the overall risk probability of future floods based on historical samples and real-time monitoring information.

9. The intelligent early warning method for flood disaster risk fusion based on multi-source information according to claim 1, characterized in that, The risk level classification includes at least four levels: low risk, relatively high risk, high risk, and extremely high risk. Different colors or warning methods are used to mark these levels to guide disaster prevention and mitigation measures, including personnel evacuation, traffic control, drainage scheduling, and emergency rescue deployment.

10. The intelligent early warning method for flood disaster risk fusion based on multi-source information according to claim 1, characterized in that, The method also includes an adaptive update step for the early warning system. By comparing the model prediction results with the actual monitoring data, the source of prediction deviation is identified, and the data processing flow, risk classification rules, monitoring weight parameters and model structure are continuously optimized, so that the entire early warning system has the ability to learn and correct itself under long-term operating conditions.

Images