Real-time flood forecasting and urban risk assessment method and system for mountainous area drainage basin

Abstract

The invention discloses a real-time flood forecasting and urban risk assessment method and system for a mountainous area drainage basin. The method comprises the steps of collecting historical hydrological data of a drainage basin where a river city of a target mountainous area is located, and inputting the historical hydrological data into a distributed hydrological model for initial flood forecasting; acquiring real-time hydrological data, taking the real-time hydrological data as an input feature of the deep learning model, and performing error correction on the real-time hydrological data; using the real-time hydrological data after error correction to construct a two-dimensional hydrodynamic model, performing flood routing and flooding simulation, and outputting dynamic flood flooding information; and according to the dynamic flood inundation information, constructing a flood disaster dynamic risk assessment system, carrying out comprehensive risk calculation, and calculating a risk level by using real-time hydrological data to realize dynamic level assessment of the flood disaster risk. According to the method, the forecasting precision and real-time performance can be improved in flood early warning, flooding simulation can be carried out more finely, the risk assessment real-time performance is enhanced, and the method has the dynamic characteristic.

Description

Technical Field

[0001] This invention relates to the field of flood early warning and management technology, and in particular to a method and system for real-time flood forecasting and urban risk assessment in mountainous watersheds. Background Technology

[0002] Mountainous watersheds are prone to flash floods under heavy rainfall conditions due to their steep terrain, rapid flow, sparse vegetation, and relatively weak flood control facilities. Although the existing flood management system has been continuously developing in terms of model algorithms, monitoring methods, and information platforms, there are still obvious shortcomings in its application in small and medium-sized watersheds in mountainous areas.

[0003] First, traditional flood forecasting models are mostly based on empirical parameters or static watershed characteristics, which makes them poorly adaptable to complex terrain and nonlinear confluence conditions. This leads to significant deviations in key indicators such as peak arrival time, flood volume, and peak discharge. Especially in small and medium-sized watersheds in mountainous areas, flood events are characterized by rapid rises and falls and short response times. Existing real-time error correction methods struggle to fully capture these dramatic dynamic changes, resulting in delayed or distorted forecasts.

[0004] Secondly, the accuracy of flood inundation simulation is limited by the resolution of topographic data and the ability to characterize the underlying surface. Conventional topographic data (such as DEMs with a resolution of 10 meters or more) cannot reflect the subtle topographic features of buildings, roads, and riverbank slopes in mountainous towns, resulting in significant errors in the two-dimensional hydrodynamic simulation results in terms of inundation range, depth, and duration, which makes it difficult to meet the needs of refined disaster management.

[0005] Furthermore, existing flood disaster risk assessments mostly employ static or semi-static methods, treating risk as a result under long-term average conditions, neglecting the temporal evolution characteristics of flood processes and the dynamic changes in socio-economic activities. Such assessments fail to reflect the spatiotemporal evolution of hazard, exposure, and vulnerability as floods spread during a disaster, limiting the scientific rigor and timeliness of real-time early warning and emergency response.

[0006] Although deep learning models (such as LSTM, CNN, and Transformer) have shown strong nonlinear fitting capabilities in hydrological time series prediction in recent years, UAV remote sensing technology can acquire high-precision terrain data, and two-dimensional hydrodynamic models have high physical realism in flood process simulation, there is still a lack of a method to achieve dynamic risk quantification and graded early warning for the entire process of flood disasters in mountainous areas. Summary of the Invention

[0007] In view of the aforementioned existing problems, the present invention is proposed.

[0008] Therefore, the technical problem solved by the present invention is that the existing flood warning system has low forecast accuracy, imprecise inundation simulation, and static risk assessment.

[0009] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for real-time flood forecasting and urban risk assessment in mountainous watersheds, comprising: Historical hydrological data of the river basin where the target mountainous city is located is collected and the historical hydrological data is input into a distributed hydrological model for initial flood forecasting; Acquire real-time hydrological data, use the real-time hydrological data as input features of a deep learning model, and perform error correction on the real-time hydrological data; Using real-time hydrological data after error correction, a two-dimensional hydrodynamic model is constructed to simulate flood evolution and inundation, and output dynamic flood inundation information. Based on the dynamic flood inundation information, a dynamic risk assessment system for flood disasters is constructed, comprehensive risk calculation is performed, and risk levels are calculated using real-time hydrological data to achieve dynamic risk level assessment of flood disasters.

[0010] As a preferred embodiment of the real-time flood forecasting and urban risk assessment method for mountainous watersheds described in this invention, the distributed hydrological model includes: utilizing DEM data to obtain flow direction, cumulative flow, and slope data; constructing a hydrological model with spatial distribution characteristics as the core, dividing the watershed into many computational units; and simulating the spatial distribution of processes such as rainfall, infiltration, runoff, and confluence; the distributed hydrological model acquires real-time sliding runoff and precipitation data; and inputting the initial forecast results of the distributed hydrological model and the real-time sliding runoff and precipitation data into a deep learning model for real-time correction.

[0011] As a preferred embodiment of the real-time flood forecasting and urban risk assessment method for mountainous watersheds described in this invention, the acquired data are real-time sliding runoff observation data and sliding cumulative precipitation.

[0012] As a preferred embodiment of the real-time flood forecasting and urban risk assessment method for mountainous watersheds described in this invention, the deep learning model includes: using the minimization of validation set loss as the objective function, automatically determining the optimal hyperparameter combination by iteratively selecting flood warning data with the greatest expected improvement for experimentation.

[0013] As a preferred embodiment of the real-time flood forecasting and urban risk assessment method for mountainous watersheds described in this invention, the two-dimensional hydrodynamic model constructed by the flood evolution and inundation simulation includes: taking digital terrain data represented by grid cells as input, simulating with an HEC-RAS two-dimensional model, and simulating dynamically changing two-dimensional water flow as output, to simulate the flow, diffusion and confluence process of floods under complex terrain.

[0014] The HEC-RAS two-dimensional model is based on the two-dimensional Saint-Venant equation, which is expressed as follows: (1) (2) (3) Where t represents the flood flow time; Indicates the depth of the flood; and They represent direction and The unit flow rate in the direction; g represents the acceleration due to gravity; Indicates the Manning coefficient; Indicates the density of water; , and Indicates the components of the effective shear stress; The Coriolis parameter is represented; Equation (1) is the continuity equation, and Equations (2) and (3) are the momentum equations in the x-axis and y-axis directions, respectively. When the flow velocity is slow, the water surface slope is small, the flow regime changes slowly and a rapid calculation is required, and the diffusion wave mode is selected, the inertial terms in the momentum equations of equations (2) and (3) are ignored, and the resulting set of equations is a two-dimensional diffusion wave equation. The two-dimensional diffuse wave equation is derived from the two-dimensional Saint-Venant equation under a specific approximation. Therefore, the vector form of the momentum equations for both the two-dimensional Saint-Venant equation and the two-dimensional diffuse wave equation is: (4) (5) In the formula, Represents the velocity vector; Represents the horizontal eddy viscosity tensor; Represents the gradient operator; Represents a unit vector in the vertical direction; and These represent the bottom shear force and the wind surface stress vectors, respectively. Indicates water depth; Represents the Coriolis parameter; Indicates atmospheric pressure; Indicates the hydraulic radius; Indicates water surface elevation; Represents gravitational acceleration; Indicates the Manning coefficient; Indicates the density of water; The dynamic flood inundation information output by two-dimensional hydrodynamics includes water depth distribution, flow velocity distribution, water level height, and inundation range.

[0015] As a preferred embodiment of the real-time flood forecasting and urban risk assessment method for mountainous watersheds described in this invention, the dynamic risk assessment system is represented by three dimensions: hazard, residential exposure, and residential vulnerability.

[0016] The risk level is measured by weighting and summing the flood inundation depth, flow velocity, and flood duration to obtain a flood risk index, which reflects the degree of danger in different regions during the flood evolution process.

[0017] The residential exposure is measured by the total population and expressed as follows: (6) in, This represents the total population of the i-th building; This represents the elevation value of the top floor of the i-th building. This represents the ground floor elevation of the i-th building. Both elevation values ​​are extracted based on a high-precision DEM. This represents the single-story area of ​​the i-th building, calculated from the outlines of each building. Indicates the height of a single floor in a residential building; This indicates the per capita housing construction area for urban residents.

[0018] The vulnerability of the residential buildings is represented by a discrete function of flooding depth versus loss: (7) in, This indicates the proportion of the affected population corresponding to different flooding depths within the assessment unit; This indicates the average flooding depth within the assessment unit; This indicates the percentage of the population within the pre-defined study interval.

[0019] As a preferred embodiment of the real-time flood forecasting and urban risk assessment method for mountainous watersheds described in this invention, the risk calculation employs a projection pursuit method and a real-number encoding-accelerated genetic algorithm for dynamic weight allocation, extracts assessment data from three dimensions in the dynamic risk assessment, performs weighted summation, and conducts dynamic level assessment by referring to the comprehensive risk assessment level map.

[0020] Secondly, as a preferred embodiment of the real-time flood forecasting and urban risk assessment system for mountainous watersheds described in this invention, wherein: The data acquisition unit collects historical hydrological data of the river basin where the target mountainous city is located and inputs the historical hydrological data into the distributed hydrological model for initial flood forecasting. The analysis unit acquires real-time hydrological data and uses the real-time hydrological data as input features for a deep learning model to perform error correction on the real-time hydrological data. The simulation unit uses real-time hydrological data after error correction to construct a two-dimensional hydrodynamic model, simulate flood evolution and inundation, and output dynamic flood inundation information; The assessment unit constructs a dynamic risk assessment system for flood disasters based on the dynamic flood inundation information, performs comprehensive risk calculations, and calculates risk levels using real-time hydrological data to achieve dynamic risk level assessment of flood disasters.

[0021] Thirdly, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the method for real-time forecasting of floods in mountainous watersheds and urban risk assessment as described in the first aspect of the present invention.

[0022] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the method for real-time forecasting of floods in mountainous watersheds and urban risk assessment as described in the first aspect of the present invention.

[0023] The beneficial effects of this invention are as follows: High forecast accuracy and strong real-time performance. By introducing a deep learning model to correct the output results of the distributed hydrological model in real time, it effectively adapts to the characteristics of rapid rise and fall of floods in small and medium-sized watersheds in mountainous areas, significantly improving the accuracy and timeliness of flood forecasts. Detailed inundation simulation. Using UAVs to acquire high-resolution digital terrain models, combined with a two-dimensional hydrodynamic model, the flood inundation process can be simulated, accurately depicting complex terrain and urban underlying surface features, greatly improving the spatial resolution and reliability of flood inundation range and depth calculations. Dynamic risk assessment. Based on the integration of real-time corrected forecast results and detailed inundation simulation information, a dynamic risk assessment system for flood disasters is constructed from three dimensions: hazard, residential exposure, and residential vulnerability. This achieves the spatiotemporal dynamic quantification of risk as it evolves with the flood, better reflecting the actual development patterns of disasters. High method integration. This invention covers the entire process of flood disaster management—forecasting, simulation, and assessment—organically integrating hydrological models, remote sensing data, hydrodynamic simulation, and socio-economic analysis to form a comprehensive assessment method with high systematicity and synergy, improving the scientific and intelligent level of flood management. Strong practicality. This method has a clear technical approach, a wide range of data sources, and a model that can run in real time. It can provide real-time, accurate, and dynamic risk warning and decision support for flood control and disaster reduction in mountainous and riverside cities, and has good engineering application prospects and promotion value. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 The present invention provides an overall flowchart of a method for real-time flood forecasting in small and medium-sized watersheds in mountainous areas and dynamic risk assessment of disasters in riverside cities.

[0026] Figure 2 This is a flowchart of the dynamic risk assessment method for flood disasters in mountainous and riverside cities according to the present invention.

[0027] Figure 3 This is a flowchart of the real-time correction method based on sliding data and deep learning described in this invention.

[0028] Figure 4 This is a schematic diagram of the real-time rolling forecast and sliding runoff precipitation data described in this invention.

[0029] Figure 5 This is a flowchart of the high-precision digital terrain data acquisition and processing described in this invention.

[0030] Figure 6 This is a framework diagram for dynamic risk assessment of riverside cities in mountainous areas, as described in this invention.

[0031] Figure 7 This is a spatial distribution map of the comprehensive dynamic risk assessment results of this invention. Detailed Implementation

[0032] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0033] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0034] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0035] Example 1, referring to Figures 1-7As one embodiment of the present invention, this embodiment provides a method for real-time flood forecasting and urban risk assessment in mountainous watersheds, comprising the following steps: S1. Collect historical hydrological data of the river basin where the target mountain city is located and input the historical hydrological data into the distributed hydrological model for initial flood forecasting; obtain real-time hydrological data and use the real-time hydrological data as input features of the deep learning model to perform error correction on the real-time hydrological data.

[0036] The distributed hydrological model includes using DEM data for flow direction, cumulative flow, and slope data to construct a hydrological model centered on spatial distribution characteristics. This model divides the watershed into numerous computational units and simulates the spatial distribution of processes such as rainfall, infiltration, runoff, and confluence. The distributed hydrological model acquires real-time sliding runoff and precipitation data. Subsequently, a real-time error correction method based on sliding runoff and precipitation data and a deep learning model (LSTM, CNN, or Transformer) is introduced to significantly improve forecast accuracy and timeliness. A preferred approach is to input the hydrological model output, real-time sliding runoff data, and sliding cumulative precipitation data into the deep learning model for correction.

[0037] It's worth noting that deep learning model selection allows for the choice of one or more of the three different architectures: CNN, LSTM, and Transformer, depending on the needs. All models are built using the most basic structure. All deep learning models are built using the TensorFlow and Keras deep learning libraries, with the Adam optimizer selected for all models, Mean-Square Error (MSE) as the loss function, a batch size of 32, and 100 training epochs for all models. L2 regularization and early stopping are incorporated into all models to prevent overfitting. Hyperparameters are determined using Bayesian optimization from the Keras Tuner library, with minimizing the validation set loss as the objective function. The optimal hyperparameter combination is automatically determined through iterative experimentation, selecting the parameter combination that yields the greatest expected improvement.

[0038] Before using deep models, it is necessary to train deep learning models (such as LSTM, CNN, Transformer), using historical hydrological model simulation results, sliding runoff (past 12 hours), and sliding cumulative maximum precipitation (maximum values ​​of the past 1, 3, 6, 9, and 12 hours) as input features. Figure 3The measured runoff was used as the target value. The precipitation runoff duration observation data were divided into training, validation, and test sets at ratios of 50%, 25%, and 25%, respectively. The training set data was used to train the model, update its parameters, and allow it to learn data patterns. The validation set data was used for hyperparameter tuning and early stopping, helping to select the best model. The test set data was not visible throughout the training process and was used to finally evaluate the model's generalization ability, simulating its performance in real-world application scenarios. Twelve forecast correction times were set, ranging from 1 to 12 hours. Figure 4 The model is trained for each set forecast correction time until it reaches convergence.

[0039] Table 1 Data Collection List

[0040] S2. Using real-time hydrological data after error correction, construct a two-dimensional hydrodynamic model to simulate flood evolution and inundation, and output dynamic flood inundation information. Constructing a two-dimensional model: The two-dimensional hydrodynamic model includes taking digital terrain data represented by grid cells as input, simulating it with an HEC-RAS two-dimensional model, and outputting a dynamic two-dimensional water flow simulation to simulate the flow, diffusion, and confluence of floods in complex terrain.

[0041] The HEC-RAS two-dimensional model is based on the two-dimensional Saint-Venant equation, which is expressed as follows: (1) (2) (3) Where t represents the flood flow time; Indicates the depth of the flood; and They represent direction and The unit flow rate in the direction; g represents the acceleration due to gravity; Indicates the Manning coefficient; Indicates the density of water; , and Indicates the components of the effective shear stress; The Coriolis parameter is represented; Equation (1) is the continuity equation, and Equations (2) and (3) are the momentum equations in the x-axis and y-axis directions, respectively.

[0042] When the flow velocity is slow, the water surface slope is small, the flow regime changes slowly, and a rapid calculation is required, and the diffusion wave mode is selected, the inertial terms in the momentum equations of equations (2) and (3) are ignored, and the resulting set of equations is a two-dimensional diffusion wave equation.

[0043] The two-dimensional diffuse wave equation is derived from the two-dimensional Saint-Venant equation under a specific approximation. Therefore, the vector form of the momentum equations for both the two-dimensional Saint-Venant equation and the two-dimensional diffuse wave equation is: (4) (5) In the formula, Represents the velocity vector; Represents the horizontal eddy viscosity tensor; Represents the gradient operator; Represents a unit vector in the vertical direction; and These represent the bottom shear force and the wind surface stress vectors, respectively. Indicates water depth; Represents the Coriolis parameter; Indicates atmospheric pressure; Indicates the hydraulic radius; Indicates water surface elevation; Represents gravitational acceleration; Indicates the Manning coefficient; This indicates the density of water.

[0044] The dynamic flood inundation information output by two-dimensional hydrodynamics includes water depth distribution, flow velocity distribution, water level height, and inundation range.

[0045] Importing a high-resolution DTM (integrating underwater topography) into the constructed two-dimensional hydrodynamic model, the specific process is as follows: Figure 5 As shown in the figure, the boundary conditions are set: the upstream input is the corrected flood hydrograph, and the downstream is the water depth calculated based on the river slope. Then, a reasonable Manning roughness coefficient (river channel, floodplain, urban area) is set to run the model. By simulating the flood evolution process, the dynamic inundation range raster map, inundation depth raster map and inundation duration information are output.

[0046] S3. Based on the dynamic flood inundation information, construct a dynamic risk assessment system for flood disasters, perform comprehensive risk calculations, and use real-time hydrological data to calculate the risk level, thereby achieving dynamic risk level assessment of flood disasters.

[0047] The dynamic risk assessment system is represented by three dimensions: hazard, residential exposure, and residential vulnerability.

[0048] The risk level is measured by weighting and summing the flood inundation depth, flow velocity, and flood duration to obtain a flood risk index, which reflects the degree of danger in different regions during the flood evolution process.

[0049] The residential exposure is measured by the total population and expressed as follows: (6) in, This represents the total population of the i-th building; This represents the elevation value of the top floor of the i-th building. This represents the ground floor elevation of the i-th building. Both elevation values ​​are extracted based on a high-precision DEM. This represents the single-story area of ​​the i-th building, calculated from the outlines of each building. Indicates the height of a single floor in a residential building; This indicates the per capita housing construction area for urban residents.

[0050] It should be noted that the characteristics of buildings in riverside cities in mountainous areas of my country are that most of the buildings along the river are residential, and due to their early construction and limited development costs, the average building height is around 10 stories, but they are densely distributed. Excluding a few large buildings, such as hospitals and schools, the majority of buildings in riverside cities in mountainous areas consist of ground-floor shops along the street, with residential units above. Therefore, this method divides the building exposure analysis into residential units and ground-floor shops.

[0051] For mountainous and riverside cities, the impact of floods on housing is mainly manifested in restricting residents' movement. Therefore, in the exposure analysis of housing, in addition to obtaining the number and distribution of buildings, it is also necessary to obtain the area and floor height of each building, which is crucial for determining the number of people affected in the housing. This invention example is based on the Application Programming Interface (API) of the national vector map provided by the National Geographic Information Public Service Platform, TianDiTu. The TianDiTu tool plugin in QGIS software is used to initially extract the building outlines within the study area, and then the building outlines are corrected using high-resolution UAV orthophotos obtained in previous studies. Next, QGIS software is used to extract the elevation data of the center point of each building and the surrounding roads based on the centimeter-level DTM obtained by UAVs in previous studies. The difference between the two is taken as the building floor height, and finally, the housing exposure formula is used to calculate and output the results.

[0052] It should also be noted that in other alternative embodiments, exposure analysis may include aspects other than residential exposure: Merchant exposure analysis: For the exposure analysis of ground-floor businesses in buildings in mountainous riverside cities, the focus should be on obtaining the spatial distribution of various types of businesses. Based on statistics from the Fourth National Economic Census Bulletin of Dazhou City, this study primarily conducts risk assessments on businesses in the catering and retail categories. The method for obtaining their spatial distribution is still to call the API of the Geographic Information Public Service Platform to perform Point of Interest (POI) searches for catering and retail businesses by region, obtaining information such as latitude and longitude, name, detailed type, and specific address of all catering and retail businesses within the study area. Simultaneously, to ensure the accuracy of the search results, this invention example calls the APIs provided by Tianditu, Gaode Maps, and Baidu Maps for mutual verification and supplementation of the search results.

[0053] Road exposure analysis: Roads play a vital role in the socio-economic system. While the impact of flooding on buildings is usually localized, the damage to roads is often more widespread. The impact of road flooding on transportation systems, especially the restrictions on the passage of residents and vehicles, can even threaten the entire urban system. This study uses the same methodology as previous studies, combining Tianditu (a Chinese online map platform) and high-resolution orthophotos to acquire the road network. Since the buffer radius of the assessment unit is 30m, this study sets road assessment points (within the flooded area) at 60m intervals on the extracted road network to ensure consistency with the assessment unit.

[0054] The vulnerability of the residential buildings is represented by a discrete function of flooding depth versus loss: (7) in, This indicates the proportion of the affected population corresponding to different flooding depths within the assessment unit; This indicates the average flooding depth within the assessment unit; This indicates the percentage of the population within the pre-defined study interval.

[0055] It's important to understand that, considering that buildings in mountainous riverside cities are primarily composed of ground-floor shops and upper-floor residences, the impact of flooding on residences mainly affects residents' access, with little direct damage to the buildings themselves. Therefore, this study's vulnerability analysis of residences focuses on the affected population. By considering population structure (elderly, children, and adults) and the impact of different inundation depths on residents' access, combined with detailed statistical analysis of the spatial distribution of residences, the study quantifies the degree of impact of flooding on residents.

[0056] It should also be noted that in other alternative embodiments, vulnerability analysis may include aspects other than residential vulnerability: Road vulnerability analysis: Due to the high complexity of road network disruptions, this study only considers the varying degrees of impact of flood inundation depth on road traffic, using a discrete function representing the inundation depth-loss relationship, as shown in the following equation: (8) In the formula: This indicates that the road is flooded at different depths within the assessment unit. The corresponding road traffic loss levels are as follows: Level 0 indicates that road traffic is slightly affected, and pedestrians and vehicles can pass as appropriate; Level 1 indicates that pedestrian traffic is affected, and vehicles can pass as appropriate; Level 2 indicates that vehicle traffic is affected, and vehicles with certain wading capabilities can pass as appropriate; Level 3 indicates that the road is completely closed to traffic.

[0057] Indicator Normalization: Based on the flood disaster risk factor analysis (hazard, exposure, vulnerability analysis) in the previous steps, eight assessment factors were selected to construct a comprehensive dynamic risk assessment indicator system. These include the seven key dynamic risk assessment objects from the previous steps (inundation center depth, average depth, maximum depth, inundation duration, affected residential population, business revenue loss, and road damage), plus the elevation of the assessment unit's center point as a supplementary topographic risk factor. A high-dimensional data matrix composed of the assessment unit and the eight risk assessment factors was then normalized.

[0058] Weight Calculation and Comprehensive Evaluation: Inputs are fed into the projection pursuit model, and a mapping relationship between high-dimensional data and low-dimensional space is constructed through linear projection. A real-code accelerated genetic algorithm is used to solve the projection objective function. In this study, the initial population size N = 400, the crossover probability was set to 0.80, the number of excellent individuals was set to 20, and the acceleration was performed 10 times to obtain the optimal projection direction values ​​of the indicator system. The larger the optimal projection direction value, the greater its contribution to the comprehensive evaluation. The projection direction values ​​are proportionalized, that is, each optimal projection direction value is divided by their sum to obtain the weight of each indicator.

[0059] After determining the comprehensive assessment weights for the eight assessment factors, the assessment data from three dimensions in the dynamic risk assessment were extracted and weighted, and then summed. After the comprehensive assessment weights were determined, the risk assessment results for all types of risks throughout the entire flood process needed to be dynamically weighted and averaged. To ensure that the comprehensive risk assessment accurately covers all risk situations in the assessment process, and to make the comprehensive risk assessment results more intuitive, this study aggregated the assessment results for each assessment factor at all times and normalized them. Then, based on the assigned weights, the comprehensive risk assessment results were obtained and classified according to the comprehensive risk assessment level chart.

[0060] A comprehensive risk assessment level map is a map that visualizes the spatial distribution of flood risk assessment results. It divides the hazard values ​​of each area into several hazard levels based on the comprehensive risk assessment results, and uses different colors or shades on the map to represent the distribution of each level of area, intuitively reflecting the spatial differences in flood risk and the scope of high-risk areas.

[0061] Example 2, this example also provides a real-time flood forecasting and urban risk assessment system for mountainous watersheds, which includes: The data acquisition unit collects historical hydrological data of the river basin where the target mountainous city is located and inputs the historical hydrological data into the distributed hydrological model for initial flood forecasting.

[0062] The analysis unit acquires real-time hydrological data and uses this data as input features for a deep learning model to perform error correction on the real-time hydrological data.

[0063] The simulation unit uses real-time hydrological data after error correction to construct a two-dimensional hydrodynamic model, simulate flood evolution and inundation, and output dynamic flood inundation information.

[0064] The assessment unit constructs a dynamic risk assessment system for flood disasters based on the dynamic flood inundation information, performs comprehensive risk calculations, and calculates risk levels using real-time hydrological data to achieve dynamic risk level assessment of flood disasters.

[0065] This embodiment also provides a computer device applicable to a method and system for real-time flood forecasting in mountainous river basins and urban risk assessment, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the method for real-time flood forecasting in small and medium-sized river basins in mountainous areas and dynamic risk assessment of disasters in riverside cities as proposed in the above embodiment.

[0066] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0067] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the method and system for real-time flood forecasting and urban risk assessment in mountainous watersheds as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0068] In summary, this invention can be widely applied in fields such as water conservancy, emergency management, and urban planning, providing mountainous and riverside cities with full-process technical support covering real-time flood early warning, inundation simulation, and dynamic disaster risk assessment, significantly improving the scientific and refined level of flood prevention and disaster reduction, and having significant comprehensive social and economic benefits.

[0069] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A mountainous watershed flood real-time forecasting and urban risk assessment method, characterized in that, The application relates to a dynamic flood disaster risk assessment method and system. The application comprises the following steps: collecting historical hydrological data of a target mountainous area along a river basin and inputting the historical hydrological data into a distributed hydrological model to perform initial flood forecasting; acquiring real-time hydrological data, taking the real-time hydrological data as input features of a deep learning model, and performing error correction of the real-time hydrological data; utilizing the real-time hydrological data after error correction, constructing a two-dimensional hydrodynamic model, performing flood evolution and submergence simulation, and outputting dynamic flood submergence information; 2. The mountain basin flood real-time forecasting and city risk assessment method according to claim 1, characterized in that: according to the dynamic flood submergence information, constructing a flood disaster dynamic risk assessment system, performing comprehensive risk calculation, and calculating a risk grade by using real-time hydrological data, so that dynamic grade evaluation of flood disaster risk is realized.

3. The mountain basin flood real-time forecasting and city risk assessment method of claim 2, wherein: The distributed hydrological model comprises the following steps: utilizing DEM data to perform flow direction, cumulative flow and slope data, constructing a hydrological model taking spatial distribution characteristics as a core, dividing a basin into many calculation units, and simulating spatial distribution of processes such as rainfall, infiltration, runoff and confluence; the distributed hydrological model acquires real-time sliding runoff and precipitation data; and the initial prediction result of the distributed hydrological model and the real-time sliding runoff and precipitation data are input into a deep learning model for real-time correction.

4. The mountain basin flood real-time forecasting and city risk assessment method of claim 3, wherein: The real-time hydrological data comprises real-time sliding runoff observation data and sliding cumulative precipitation.

5. The mountain basin flood real-time forecasting and city risk assessment method of claim 4, wherein: The deep learning model comprises the following steps: taking minimization of a verification set loss as an objective function, selecting flood early warning data with maximum expected promotion through iteration, and automatically determining an optimal hyperparameter combination. The two-dimensional hydrodynamic model constructed by the flood evolution and submergence simulation comprises the following steps: taking digital terrain data represented by grid units as input, taking a HEC-RAS two-dimensional model to perform simulation, taking dynamic change of two-dimensional water flow simulation as output, simulating flow, diffusion and confluence processes of flood in complex terrain, and outputting dynamic flood submergence information. （1） （2） （3） where t represents the flood flow time; represents the flood depth; and respectively represent the unit flow in the x-axis direction and the unit flow in the y-axis direction; g represents the gravitational acceleration; represents the Manning coefficient; represents the water density; , and represent the components of the effective shear stress; represents the Coriolis parameter; equation (1) is a continuity equation, and equations (2) and (3) are momentum equations in the x-axis direction and the y-axis direction, respectively; The HEC-RAS two-dimensional model is based on a two-dimensional Saint-Venant equation, and the two-dimensional Saint-Venant equation is expressed as: When a diffusion wave mode is selected for slow flow, small water surface slope, slow change of flow state and rapid calculation, the inertia term in the momentum equation of formula (2) and formula (3) is ignored, and the obtained equation group is a two-dimensional diffusion wave equation; （4） （5） wherein denotes the velocity vector; denotes the horizontal eddy viscosity tensor; denotes the gradient operator; denotes the unit vector in the vertical direction; and denote the bottom shear and the wind surface stress vector, respectively; denotes the water depth; denotes the Coriolis parameter; denotes the atmospheric pressure; denotes the hydraulic radius; denotes the water surface elevation; denotes the gravitational acceleration; denotes the Manning coefficient; denotes the water density; The vector form of the momentum equation of the two-dimensional Saint-Venant equation and the two-dimensional diffusion wave equation is:

6. The mountain basin flood real-time forecasting and city risk assessment method of claim 5, wherein: The dynamic flood submergence information output by the two-dimensional hydrodynamic model comprises water depth distribution, flow velocity distribution, water level height and submergence range. The dynamic risk assessment system is represented by three dimensions of danger, residential exposure and residential vulnerability; The danger is obtained by weighting and summing flood submergence depth, flow velocity and flood duration to obtain a flood danger index, so as to reflect the danger degree of different regions in the flood evolution process; （6） wherein, represents the total population of the i-th building; represents the top floor elevation value of the i-th building, represents the bottom floor elevation value of the i-th building, both elevation values are extracted based on high-precision DEM; represents the single-story area of the i-th building, calculated from the outline of each building; represents the single-story floor height of a residential building; represents the per capita housing construction area of urban residents; The residential exposure is represented by total population quantity and is expressed as: （7） wherein, represents the proportion of the population affected when the different submersion depths within the evaluation unit correspond; represents the average submersion depth within the evaluation unit; represents the proportion of the population in the interval preset for the study.

7. The mountain basin flood real-time forecasting and city risk assessment method of claim 6, wherein: In the residential vulnerability, a submergence depth-loss discrete function is adopted to represent:

8. A mountainous watershed flood real-time forecasting and urban risk assessment system based on any one of claims 1-7, characterized in that, The projection pursuit method and the real number coding accelerated genetic algorithm are adopted to perform dynamic weight distribution during the risk calculation, evaluation data of the three dimensions in the dynamic risk assessment are extracted, weighted summation is performed, and a dynamic grade evaluation is performed by referring to a comprehensive risk assessment grade chart. The application comprises the following steps: The data acquisition unit collects historical hydrological data of a target mountainous river basin city and inputs the historical hydrological data into a distributed hydrological model to perform initial flood forecasting; The analysis unit acquires real-time hydrological data, uses the real-time hydrological data as input features of a deep learning model, and performs error correction of the real-time hydrological data; The simulation unit uses the error-corrected real-time hydrological data, constructs a two-dimensional hydrodynamic model, performs flood evolution and inundation simulation, and outputs dynamic flood inundation information; The evaluation unit constructs a flood disaster dynamic risk assessment system according to the dynamic flood inundation information, performs comprehensive risk calculation, calculates a risk level using real-time hydrological data, and realizes dynamic level evaluation of flood disaster risk. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that: The processor executes the computer program to realize the steps of the mountainous river basin flood real-time forecasting and city risk assessment method of any one of claims 1-7.

10. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to realize the steps of the mountainous river basin flood real-time forecasting and city risk assessment method of any one of claims 1-7.

Images