A river flood routing method based on river adaptive segmentation and strong water balance constraint

Abstract

The application discloses a river overflow flood evolution rapid simulation method based on river self-adaptive segmentation and strong water quantity balance constraint, realizes rapid and fine simulation of river overflow flood in a basin through the following execution processes of basin basic data collection and processing, river overflow flood inundation rapid calculation model construction, river overflow flood rapid prediction and risk analysis, meets the requirements of 'four predictions' of basin flood control fine simulation simulation, real-time early warning and prediction, rapid scenario deduction and the like, and can also serve basin flood control planning and design.

Description

Technical Field

[0001] This invention relates to the field of watershed flood early warning and forecasting technology, and in particular to a rapid simulation method for the evolution of river overflow floods based on river adaptive segmentation and strong water balance constraints. Background Technology

[0002] River overflow floods are characterized by their sudden onset, high destructiveness, and wide-ranging impact. For flood disaster emergency response, rapidly generating dynamic flood inundation risk maps and issuing early warnings are crucial for minimizing casualties and disaster losses. Traditional river overflow flood simulations are based on hydrodynamic models, requiring the discretization and generalization of the terrain using grid cells during the calculation process. The number of grid cells is generally proportional to the simulation accuracy; more grid cells result in smaller generalization errors and higher calculation accuracy. However, as the number of grid cells increases, the computational complexity also rises, consuming significant time and computational resources, leading to insufficient timeliness and making it difficult to support real-time flood inundation simulation and prediction. Furthermore, hydrodynamic modeling relies heavily on extensive underlying surface data, such as large river cross-sections; basins lacking or without data cannot be modeled, further hindering the realization of flood inundation simulation and prediction.

[0003] Currently, scholars at home and abroad are committed to exploring lightweight and rapid methods. One is a data-driven deep learning model, which generates training data through a hydrodynamic model and then uses a deep learning model to construct a mapping relationship between different inflows and inundation results in the upper reaches of the river, thereby conducting rapid flood inundation analysis. This type of method has high computational efficiency, but it still faces problems such as difficulty in modeling, high training costs, and insufficient interpretability of physical mechanisms, especially when applied to large-scale watersheds. Another type is a simplified alternative model that improves computational efficiency by reducing model complexity, such as the Height Above Nearest Drainage (HAND) model and the GeoFlood model. The computational efficiency is significantly improved compared to the hydrodynamic model, and the physical meaning of the model is clear. The key technologies are: (1) reasonably dividing the length and number of river segments to perform segmented river modeling; (2) converting the segmented flow of the river into segmented water depth, and dynamically calculating the inundation situation by comparing the segmented water depth with the inundation threshold. However, current technology still has shortcomings. First, using specified segment lengths for river segment modeling introduces significant subjectivity. Second, converting river segment flow into river segment depth based on the water level-discharge relationship curve lacks water balance constraints, leading to inundation simulation results that are too high during rising water periods and too low during receding water periods. These problems prevent the accurate simulation of the dynamic evolution of river overflow floods and are insufficient to support the creation of detailed real-time dynamic risk maps. Summary of the Invention

[0004] The purpose of this invention is to provide a rapid simulation method for the evolution of river overflow floods based on river adaptive segmentation and strong water balance constraints, thereby solving the aforementioned problems existing in the prior art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A rapid simulation method for river overflow flood evolution based on river adaptive segmentation and strong water balance constraints includes,

[0007] Construction of a rapid calculation model for river overflow flooding: Based on processed watershed elevation data, a HAND matrix composed of HAND values ​​representing the relationship between watershed slope units and rivers is calculated. Clustering algorithms are used to perform cluster analysis on ROI values ​​representing the inundation capacity of river catchments, achieving adaptive river segmentation. The total river storage capacity for each time period is calculated based on the water balance principle, and the flood inundation volume of each river segment's catchment area is calculated based on the water distribution coefficient AC. River depth-volume relationship curves for each river segment's catchment area are constructed, transforming the flood inundation volume of each river segment for each time period into segmented river depths for each time period. By comparing the segmented river depths with the slope HAND values, the flood inundation depths of each segment's catchment area are dynamically calculated, thereby drawing a dynamic flood inundation map, ultimately realizing the construction of a rapid calculation model for river overflow flooding.

[0008] Preferably, the construction of a rapid calculation model for river overflow flooding includes the following:

[0009] S11. Watershed River Network Data Generation: Using the hydrological analysis toolbox of GIS software, the processed watershed elevation data is sequentially subjected to depression filling calculation, flow direction calculation, pixel cumulative flow calculation, river linking, river network classification, and raster river network vectorization operations to obtain watershed river network distribution data.

[0010] S12. HAND Matrix Calculation: Calculate the elevation difference between a slope grid point and its nearest neighbor confluence river grid point within the watershed, i.e., the HAND value, to obtain the watershed's HAND matrix.

[0011] S13. ROI-based adaptive river segmentation: Calculate the ROI value of the river's overflow inundation capacity grid by grid, and use the Bisecting K-means clustering algorithm to perform cluster analysis on the ROI values ​​in the order from upstream to downstream of the river. Divide the continuous river grid with similar inundation capacity into a segment to form a continuous river segment.

[0012] S14. Dynamic calculation of flood inundation volume in river catchment area based on strong water balance constraint: Calculate the flood volume per unit time in the river catchment area based on the principle of water balance, and construct the water distribution coefficient AC according to the topographic characteristics of the river catchment area. Use the water distribution coefficient AC to distribute the total river water volume to each river section and obtain the flood inundation volume of each river section to the outside of the river channel.

[0013] S15. Dynamic mapping of river overflow floods: Based on the constructed segmented river water depth-volume relationship curves, the flood inundation volume of each river segment is converted into segmented river water depth. When the segmented water depth is greater than the slope HAND value, the flood inundation depth of the river segment is the difference between the segmented water depth and the slope HAND value; otherwise, the flood inundation depth of the river segment is 0. In this way, the flood inundation depth of each river segment to the river catchment area is obtained, thereby drawing a dynamic flood inundation map and realizing the construction of a rapid calculation model for river overflow flood inundation.

[0014] Preferably, the formula for calculating the ROI value is as follows:

[0015]

[0016] Wherein, D0 is the constructed equidistant water depth sequence of the river; is the maximum possible water depth of the river channel; n is the number of water depth values ​​in the water depth sequence D0; V is the i-th water depth value in the water depth sequence D0; i For the river channel depth is The amount of water that overflows and floods the river channel; F is the area of ​​a single grid cell; N i For the river channel depth is The number of grid cells that overflow and submerge outside the river channel.

[0017] Preferably, in step S14,

[0018] The formula for calculating the flood volume per unit time period in a river catchment area is as follows:

[0019] W T =W T-1 +W T,I -W T,O (2)

[0020]

[0021] Among them, W T The flood volume at time T; W T-1 The flood volume at time T-1; W T,I W represents the river inflow at time T; T,OLet T be the river outflow at time T; Q(t) be the measured or predicted flow sequence data from the upstream hydrological station; n0 be the roughness value; A be the cross-sectional area of ​​the downstream channel; R be the hydraulic radius of the downstream channel; S be the gradient of the downstream section; and Δt be the time difference.

[0022] The formula for calculating the water distribution coefficient of each river section is as follows:

[0023]

[0024] Among them, IEF k D is the inundation extent index of the k-th river segment; D is the constructed equidistant water depth sequence of the k-th river segment. n The maximum possible water depth of the k-th river segment; m represents the number of water depth values ​​in the water depth sequence D; N i This indicates that the water depth in the river channel is D. i At time, m0 represents the number of slope grids submerged outside the river channel in the k-th river segment; m0 represents the number of river segments.

[0025] The formula for calculating the flood inundation volume outside the river channel in each section is as follows:

[0026] V k =W T ·AC k (7)

[0027] Among them, V k This represents the flood inundation volume of the catchment area in the k-th river segment.

[0028] Preferably, in step S15, the process of constructing the segmented water depth-water volume relationship curve is as follows:

[0029] Based on the equidistant water depth sequence D = [0,…D] for each river segment i-1 D i D i+1 …,D m The formula for calculating the flood inundation volume corresponding to each water depth is as follows:

[0030]

[0031] Among them, V i The river depth is D i At that time, the flood inundation volume of the catchment area of ​​that river section; n′ is the number of grid cells in the catchment area of ​​that river section; V i,j When the river depth is D i At that time, the flood inundation volume of the j-th grid cell within the catchment area of ​​that river section; H j F is the HAND value of the j-th grid cell; j Let be the area of ​​the j-th grid cell;

[0032] Based on the equidistant water depth sequence D=[0,…D i-1 D i D i+1 …,D m ] and the corresponding water volume sequence V=[0,…V i-1 V i V i+1 …,V m By observing the relationship between the depth and volume of the river section, the curve of the relationship between the depth and volume can be obtained.

[0033] Preferably, a step S11 and a step S12 are further included between them.

[0034] The calculated river network distribution data of the watershed was corrected using Tianditu remote sensing imagery to ensure that the river distribution is consistent with the actual location.

[0035] Preferably, after constructing the rapid calculation model for river overflow flooding, it also includes:

[0036] Rapid prediction and risk analysis of river overflow floods: Based on the constructed rapid analysis model of river overflow flood inundation, and by incorporating future hydrological forecast data, the development and evolution of river floods in the basin are predicted in real time, and the dynamic inundation range and water depth of river overflow floods in the basin are obtained; flood inundation risk indicators are established based on the dynamic inundation range and water depth of river overflow floods in the basin, and a dynamic risk map of floods in the basin is drawn based on the flood inundation risk indicators.

[0037] Preferably, rapid prediction and risk analysis of river overflow floods specifically includes the following:

[0038] S21. Watershed flood inundation prediction: Based on the upstream section flow forecast results, drive the rapid analysis model of river overflow flood inundation to predict the watershed flood inundation range and water depth data in real time.

[0039] S22. Construction of flood inundation risk indicators: Taking into full account the threat posed by flood inundation depth to the safety of people's lives and property and social and economic security, as well as the standards for compiling flood risk maps, flood inundation depths of 0.5m, 1m, 2m, and 3m are set as indicators for low, medium, high, and extremely high risks of flood inundation in the basin, respectively, for extreme flood events.

[0040] S23. Real-time flood risk prediction and risk analysis of key protected objects: Based on real-time inundation prediction data and combined with flood inundation risk indicators, draw a dynamic flood risk map; overlay geospatial information layers such as flood control projects, lifeline projects, and key protected objects on the dynamic flood risk map to draw a thematic risk map.

[0041] Preferably, prior to building a rapid calculation model for river overflow flooding, the following steps are also included:

[0042] Watershed basic data collection and processing: Collect and process high-resolution digital elevation data, river channel and levee data and hydrological data of the watershed.

[0043] Preferably, the collection and processing of basic watershed data specifically includes the following:

[0044] S01. Download meter-level digital elevation data (DEM) of the watershed using a map downloader, with a spatial resolution of 30m, 12.5m, or 5m; collect river embankment top elevation data of the watershed, and correct the river embankment top elevation data based on the digital elevation data to make the elevation data inside and outside the river and the embankment more consistent with the actual topography of the watershed.

[0045] S02. Collect measured flow data from river hydrological stations in the basin and standardize the time interval of the flow data to 1 hour.

[0046] The beneficial effects of this invention are as follows: 1. This invention constructs a river overflow inundation capacity index to characterize the inundation potential of a river catchment area, and uses cluster analysis to achieve adaptive segmented modeling of the river. Compared with the traditional manual segmentation method, the adaptive segmentation method based on the river overflow inundation capacity index provides a more scientific and convenient segmentation modeling method, which can eliminate subjective errors caused by human factors, ensure the objectivity and rationality of the segmentation results, and significantly improve the efficiency of river segmentation. In addition, the traditional method converts the segmented flow into the river depth based on the river level-discharge curve. This curve is a single relationship and cannot depict the actual non-single relationship caused by flood rise and fall. The water balance constraint is insufficient, resulting in large errors in the inundation simulation results and an inability to accurately simulate the dynamic evolution of floods. The water volume-depth curve and the dynamic calculation method of flood inundation water volume in the river catchment area based on strong water balance constraints proposed in this invention improve the accuracy of flood dynamic evolution simulation and can accurately simulate the full-time and spatial dynamic process of floods from rising to receding. 2. Unlike traditional hydrodynamic models, which are mainly used for small-scale, refined simulations, flood control planning, and design, the proposed rapid simulation method for river overflow flood evolution based on adaptive segmentation and strong water balance constraints can meet the needs of large-scale, refined simulations, flood control planning, and design, and can also be applied to real-time early warning and rapid projection of basin floods. While maintaining similar accuracy to hydrodynamic models, it improves computational efficiency to the second level, significantly enhancing computational timeliness. Furthermore, the proposed method only requires upstream cross-sectional flow data and DEM data for rapid and lightweight modeling. Compared to hydrodynamic models, its dependence on modeling data is significantly reduced, providing a novel solution for flood inundation and evolution simulation in areas with scarce or no data. Moreover, unlike deep learning-based inundation calculation methods, the proposed method does not require large-scale, costly training with numerous samples, and the model has clear physical meaning, making modeling more convenient and lightweight. Combined with flood inundation depth risk indicators, it can support relevant departments in publishing real-time dynamic flood risk maps, as well as thematic risk maps for basin flood control projects, lifeline projects, and key protected objects, providing technical support for emergency response to sudden flood disasters. Attached Figure Description

[0047] Figure 1 This is a flowchart of the simulation method in an embodiment of the present invention;

[0048] Figure 2 This is a schematic diagram of adaptive segmentation of the river in an embodiment of the present invention. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0050] Addressing the technical challenges of insufficient timeliness of current classical dynamic models, complex modeling of deep learning models, and the inability of simplified models to accurately simulate floods, this embodiment focuses on the pain points and difficulties faced by lightweight and rapid models. It fully leverages and integrates the advantages of methods from topography, hydrology, and big data technologies. This embodiment provides a rapid simulation method for river overflow flood evolution based on river adaptive segmentation and strong water balance constraints. This method enables rapid and precise simulation of river overflow floods in a watershed, meeting the "four pre-monitoring" requirements of watershed flood control: precise simulation, real-time early warning and forecasting, and rapid scenario extrapolation. It can also serve watershed flood control planning and design. Figure 1 As shown, the method of the present invention mainly includes the following three parts:

[0051] I. Collection and Processing of Basic Watershed Data

[0052] We collect and process high-resolution digital elevation data (DEM), river channel and levee data, and hydrological data of the watershed (study area) to provide a data foundation for constructing a rapid simulation method for the evolution of river overflow floods based on river adaptive segmentation and strong water balance constraints.

[0053] 1.1 Watershed Elevation Data Collection and Processing

[0054] Download meter-level digital elevation data (DEM) using map downloaders such as BIGEMAP, with spatial resolutions of 30m, 12.5m, and 5m. Collect river embankment crest elevation data and correct the river embankment crest elevation data based on the DEM data to make the elevation data inside and outside the river channel and the embankment more consistent with the actual topography of the watershed.

[0055] 1.2 Hydrological Data Collection and Processing

[0056] Collect measured flow data from river hydrological stations and standardize the time interval of the flow data to 1 hour.

[0057] II. Construction of a rapid calculation model for river overflow flooding

[0058] Based on the corrected basin elevation data from step 1.1, basin river network distribution data is generated, and the HeightAbove Nearest Drainage matrix (hereinafter referred to as the "HAND matrix") is calculated. The HAND matrix is ​​used to construct physical characteristic indicators representing the inundation capacity of river grid cells. A clustering algorithm is employed to perform cluster analysis on these indicators, thereby enabling adaptive river segmentation. Based on the water balance principle and Manning's formula, the total river storage capacity for each time period is calculated, and a water distribution coefficient is constructed according to the topographic characteristics of the river segment catchment area. The flood inundation volume of each river segment's catchment area is then calculated. A river depth-volume relationship curve is constructed for each river segment's catchment area, converting the flood inundation volume of each river segment (river segment) for each time period into the segment's river depth for each time period. Finally, by comparing the segment's river depth with the inundation threshold, the flood inundation depth outside the grid cells of each river segment is dynamically calculated, thus generating a dynamic flood inundation map.

[0059] 2.1 Generation of River Network Data in the Watershed

[0060] Using the hydrological analysis toolbox of GIS software, perform operations such as depression filling calculation, flow direction calculation, pixel cumulative flow calculation, river linking, river network classification, and raster river network vectorization on the corrected DEM data in step 1.1 to obtain the watershed river network distribution data.

[0061] Meanwhile, to improve the accuracy of river network data, Tianditu remote sensing imagery was used to correct the calculated river network data, ensuring that the river distribution is consistent with the actual location.

[0062] 2.2 HAND Matrix Calculation

[0063] The HAND value is the elevation difference between a slope grid point and its nearest confluence river grid point within a watershed, serving as a physical indicator of the relationship between slope units and rivers. By traversing and calculating the HAND value of each grid point in the DEM grid of the study area, a HAND matrix with the same dimensions as the DEM data can be obtained, where the HAND value of the river grid is 0. Based on the assumptions of vertical water pressure balance along the slope confluence path and the absence of lateral flow along the confluence path, the HAND matrix and river depth data can be used to identify the water supply inundation range and flood inundation depth. That is, areas where the slope grid HAND value is lower than the corresponding river grid water depth are inundated areas, and the corresponding flood inundation depth is the difference between the river grid water depth and the slope grid HAND value.

[0064] 2.3 River Adaptive Segmentation Based on ROI

[0065] Determining the water depth values ​​of the river grid in step 2.2 is crucial for calculating the inundation range and depth of floodwaters. Due to the lack of grid-by-grid water depth data for the river channel, the average water depth of segmented river sections is typically used when applying the HAND method for inundation calculations. However, the artificial division of river segments introduces subjectivity and uncertainty; the length of each segment affects the model's simulation accuracy. Excessively long segments lead to over-homogenization of the river channel topography, while excessively short segments result in over-amplification of local errors.

[0066] This invention employs a ROI-based adaptive river segmentation technique, which automatically segments the river (channel) according to terrain features, merging areas with similar terrain in the river's catchment area into one segment. This method ensures the efficiency and rationality of segmentation. The principle diagram of the method is shown below. Figure 2 As shown. The River Overflow Index (ROI) is used to characterize the inundation potential of a river's catchment area, and the calculation method is as follows:

[0067]

[0068] In the formula, D0 is the constructed equidistant water depth sequence of the river channel. The maximum possible water depth (m) of the river channel can be obtained through flood surveys or flood frequency calculations; n represents the number of water depth values ​​in the water depth sequence D0. V represents the i-th water depth value (m) in the water depth sequence D0; i This indicates that the water depth in the river channel is... At that time, the volume of water overflowing from the river channel and inundating the area (m3); F represents the area of ​​a single grid cell (m2); N i This indicates that the water depth in the river channel is... At that time, the number of grid cells that overflowed and submerged from the river channel.

[0069] The Region of Interest (ROI) reflects the overall inundation capacity of a river to overflow its channel under different water depth scenarios, with a maximum value of 1. The closer the ROI value is to 1, the stronger the river's inundation capacity.

[0070] The ROI value of the river is calculated grid by grid. The Bisecting K-means clustering algorithm is used to perform cluster analysis on the ROI values ​​in the order from upstream to downstream of the river. Continuous river grids with similar inundation capacity are divided into segments to form continuous river segments.

[0071] 2.4 Dynamic Calculation of Flood Inundation Volume in River Reaches Based on Strong Water Balance Constraints

[0072] Based on the principle of water balance, the flood volume within a river catchment area per unit time period can be calculated using the following formula:

[0073] W T =WT-1 +W T,I -W T,O (2)

[0074] In the formula, W T The flood volume (m³) at time T 3 );W T-1 The flood volume (m³) at time T-1 3 );W T,I The river inflow at time T (m³) 3 );W T,O The river outflow at time T (m³) 3 ).

[0075] W T,I The flow rate can be calculated based on the flow data from the upstream hydrological station, using the following formula:

[0076]

[0077] In the formula, Q(t) represents the measured or predicted flow sequence data (m³) from the hydrological station in the upper reaches of the river. 3 / s).

[0078] W T,O It can be calculated using Manning's formula:

[0079]

[0080] In the formula, n is the roughness value; A is the cross-sectional area of ​​the downstream channel; R is the hydraulic radius of the downstream channel cross-section; S is the gradient of the downstream section; and Δt is the time difference.

[0081] In Formula 4, A and R are calculated based on the water depth of the downstream river segment at time T-1. This water depth value is the flood volume (W) of that river segment at time T-1 calculated using Formula (2). T-1 The value was obtained by interpolating the water depth-water volume relationship curves of the segmented river channel.

[0082] The total flood volume of the river (W) was calculated. T After that, the total river volume is allocated to each river segment divided in step 2.3 using the allocation coefficient AC. The allocation coefficient AC for the k-th river segment is... k The calculation formula is as follows:

[0083]

[0084] In the formula, IEF k D is the inundation extent factor for the k-th river segment; D is the equidistant water depth sequence constructed for the k-th river segment (applicable to all river segments).n The maximum possible water depth (m) of the k-th river segment can be obtained through flood surveys or flood frequency calculations; m represents the number of water depth values ​​in the water depth sequence D; N i This indicates that the water depth in the river channel is D. i At time, m0 represents the number of slope grids that are submerged outside the river channel in the k-th river segment; m0 represents the number of river segments.

[0085] After calculating the water distribution coefficients for each segment of the river, the flood inundation volume V of the k-th segment outside the river channel is... k The calculation formula is as follows:

[0086] V k =W T ·AC k (7)

[0087] 2.5 Dynamic mapping of river overflow floods

[0088] To draw a flood inundation map for each river segment, the flood inundation volume of each river segment calculated in step 2.4 needs to be converted into segmented river channel depth. Then, by comparing the segmented river channel depth with the slope HAND value (i.e., the inundation threshold of each slope grid), the flood inundation depth of each river segment to the outer slope grid (catchment area) is dynamically calculated, thereby drawing a dynamic flood inundation map.

[0089] The segmented river depth-volume relationship curve is used to convert the flood inundation volume of the river catchment area into segmented river depth. The method for constructing the segmented water depth-volume relationship curve is as follows:

[0090] For each river segment, construct an equidistant water depth sequence D = [0,…D i-1 D i D i+1 …,D m Then calculate the flood inundation volume corresponding to each water depth, using the following formula:

[0091]

[0092] In the formula, V i The depth of the river channel is represented by D. i At that time, the flood inundation volume (m³) in the catchment area of ​​this river section 3 ); n′ represents the number of grid cells in the catchment area; V i,j This indicates that when the river depth is D i At that time, the flood inundation volume (m³) of the j-th grid cell within the catchment area of ​​that river section. 3 ); H j This represents the HAND value (m) of the j-th raster cell; F j Represents the area (m²) of the j-th grid cell.2 ).

[0093] Based on the equidistant water depth sequence D=[0,…D i-1 D i D i+1 …,D m ] and the corresponding water volume sequence V=[0,…V i-1 V i V i+1 …,V m By observing the relationship between the depth and volume of the river section, the curve of the relationship between the depth and volume can be obtained.

[0094] Using the above method, a water depth-volume relationship curve is constructed for each river segment. The amount of water overflowing into the river channel at time T is calculated using formulas (2) and (7). The water depth-volume relationship curve is then used to convert the average water depth of each river segment at time T. By iteratively calculating formulas (2) and (7), the average water depth of each river segment at each time step can be obtained. Then, by comparing the average water depth of each river segment with the corresponding slope HAND value (i.e., the slope grid inundation threshold) at each time step, the flood inundation depth of each river segment into the slope grid (catchment area) outside the river channel is dynamically calculated, thereby drawing a dynamic flood inundation map.

[0095] The formula for calculating the flood inundation depth is as follows:

[0096]

[0097] In the formula, Z i D represents the flood inundation depth (m) of grid cell i; H represents the river channel depth (m); i This represents the HAND value (m) of grid cell i.

[0098] III. Rapid Prediction and Risk Analysis of River Overflow Floods

[0099] Based on the constructed rapid analysis model for river overflow flooding, and by integrating future hydrological forecasts (upstream cross-sectional flow data), the development and evolution of river floods in the basin are predicted in real time, and the dynamic inundation range and water depth of river overflow floods are obtained. Risk indicators are established, and a dynamic risk map of the basin floods is drawn.

[0100] 3.1 Watershed Flood Inundation Prediction

[0101] Based on the upstream cross-sectional flow forecast results, the inundation rapid analysis model constructed in step 2 is driven to predict the inundation range and water depth data of the basin flood in real time.

[0102] 3.2 Construction of Flood Inundation Risk Indicators

[0103] Taking into full account the threat posed by flood inundation depth to the safety of people's lives and property as well as to social and economic security, and the standards for compiling flood risk maps, flood inundation depths of 0.5m, 1m, 2m, and 3m are set as indicators for low, medium, high, and extremely high risks of flood inundation in the basin, respectively, for extreme flood events.

[0104] 3.3 Real-time Flood Risk Prediction and Risk Analysis of Key Protected Targets

[0105] Based on real-time inundation prediction data and the aforementioned flood risk indicators, a dynamic flood risk map is generated. On this risk map, geospatial information layers representing flood control projects, lifeline projects, and key protected objects are overlaid to create thematic risk maps, providing technical support for emergency response to sudden flood disasters.

[0106] By adopting the above-disclosed technical solution of this invention, the following beneficial effects are obtained:

[0107] This invention provides a rapid simulation method for river overflow flooding based on adaptive segmentation and strong water balance constraints. The invention constructs a river overflow inundation capacity index to characterize the inundation potential of the river catchment area and employs cluster analysis to achieve adaptive segmentation modeling of the river. Compared to traditional manual segmentation methods, the adaptive segmentation method based on the river overflow inundation capacity index provides a more scientific and convenient segmentation modeling approach, eliminating subjective errors caused by human factors, ensuring the objectivity and rationality of the segmentation results, and significantly improving the efficiency of river segmentation. Furthermore, traditional methods convert segmented flow into river depth based on the river level-discharge curve. This curve represents a single relationship and cannot depict the non-single relationship that actually exists due to flood fluctuations. Insufficient water balance constraints lead to large errors in the inundation simulation results, making it impossible to accurately simulate the dynamic evolution of floods. The water volume-depth curve and the dynamic calculation method for flood inundation in the river catchment area based on strong water balance constraints proposed in this invention improve the accuracy of flood dynamic evolution simulation, enabling precise simulation of the entire spatiotemporal dynamic process of floods from rise to fall. Unlike traditional hydrodynamic models, which are primarily used for small-scale, refined simulations, flood control planning, and design, this invention proposes a rapid simulation method for river overflow flood evolution based on adaptive segmentation and strong water balance constraints. This method can meet the needs of large-scale, refined simulations, flood control planning, and design, and can also be applied to real-time early warning and rapid projection of basin floods. While maintaining similar accuracy to hydrodynamic models, it improves computational efficiency to the second level, significantly enhancing computational timeliness. Furthermore, the proposed method only requires upstream cross-sectional flow data and DEM data for rapid and lightweight modeling. Compared to hydrodynamic models, it significantly reduces reliance on modeling data, providing a novel solution for flood inundation and evolution simulation in areas with scarce or no data. Moreover, unlike deep learning-based inundation calculation methods, the proposed method does not require costly training with a large number of samples, and the model has clear physical meaning, making modeling more convenient and lightweight. Combined with flood inundation depth risk indicators, it can support relevant departments in publishing real-time dynamic flood risk maps, as well as thematic risk maps for basin flood control projects, lifeline projects, and key protected objects, providing technical support for emergency response to sudden flood disasters.

[0108] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A rapid simulation method for the evolution of river overflow floods based on adaptive segmentation and strong water balance constraints, characterized in that: include, Construction of a rapid calculation model for river overflow flood inundation: Based on the processed watershed elevation data, a HAND matrix composed of HAND values ​​representing the relationship between watershed slope units and rivers is calculated. Clustering algorithms are used to perform cluster analysis on ROI values ​​representing the inundation capacity of river catchment areas to achieve adaptive river segmentation. The total water storage of the river is calculated for each time period based on the principle of water balance, and the flood inundation volume of each river segment's catchment area is calculated based on the water distribution coefficient AC. The river channel water depth-water volume relationship curve of each river segment's catchment area is constructed, and the flood inundation volume of each river segment for each time period is converted into the segmented river channel water depth for each time period. By comparing the relationship between the segmented river channel water depth and the slope HAND value, the flood inundation depth of each segmented river segment's catchment area is dynamically calculated, thereby drawing a dynamic flood inundation map, and finally realizing the construction of a rapid calculation model for river overflow flood inundation. The formula for calculating the ROI value is as follows: （1） in, For constructing a sequence of equidistant river depths; The maximum possible water depth in the river channel; water depth sequence The number of water depth values ​​in the middle; water depth sequence The first in A water depth value; In order to be in the river channel with a water depth of The amount of water that overflows and floods the river channel at that time; The area of ​​a single grid cell; In order to be in the river channel with a water depth of The number of grid cells that overflow and submerge outside the river channel.

2. The rapid simulation method for river overflow flood evolution based on river adaptive segmentation and strong water balance constraints as described in claim 1, characterized in that: The construction of a rapid calculation model for river overflow flooding includes the following specific components. S11. Watershed River Network Data Generation: Using the hydrological analysis toolbox of GIS software, the processed watershed elevation data is sequentially subjected to depression filling calculation, flow direction calculation, pixel cumulative flow calculation, river linking, river network classification, and raster river network vectorization operations to obtain watershed river network distribution data. S12, HAND Matrix Calculation: Calculate the elevation difference between the slope grid point and its nearest neighbor confluence river grid point within the watershed, i.e., the HAND value, to obtain the watershed's HAND matrix; S13. ROI-based adaptive river segmentation: Calculate the ROI value of the river's overflow inundation capacity grid by grid, and use the Bisecting K-means clustering algorithm to perform cluster analysis on the ROI values ​​in the order from upstream to downstream of the river. Divide the continuous river grid with similar inundation capacity into a segment to form a continuous river segment. S14. Dynamic calculation of flood inundation volume in river catchment area based on strong water balance constraint: Calculate the flood volume per unit time in the river catchment area based on the principle of water balance, and construct the water distribution coefficient AC according to the topographic characteristics of the river catchment area. Use the water distribution coefficient AC to distribute the total river water volume to each river section and obtain the flood inundation volume of each river section to the outside of the river channel. S15. Dynamic mapping of river overflow floods: Based on the constructed segmented river water depth-volume relationship curves, the flood inundation volume of each river segment is converted into segmented river water depth. When the segmented water depth is greater than the slope HAND value, the flood inundation depth of the river segment is the difference between the segmented water depth and the slope HAND value; otherwise, the flood inundation depth of the river segment is 0. In this way, the flood inundation depth of each river segment to the river catchment area is obtained, thereby drawing a dynamic flood inundation map and realizing the construction of a rapid calculation model for river overflow flood inundation.

3. The rapid simulation method for river overflow flood evolution based on river adaptive segmentation and strong water balance constraints as described in claim 2, characterized in that: In step S14, The formula for calculating the flood volume per unit time period in a river catchment area is as follows: （2） （3） （4） in, for The sheer volume of time; for The sheer volume of time; for River inflow at any given moment; for The flow rate of the river at any given moment; This refers to measured or forecasted flow sequence data from hydrological stations in the upper reaches of the river. This is the roughness value; This refers to the cross-sectional area of ​​the downstream channel. The hydraulic radius of the downstream section of the river channel; This refers to the gradient of the downstream section of the river. For time difference; The formula for calculating the water distribution coefficient of each river section is as follows: （5） （6） in, For the first The inundation range index of each river section; For the construction of the first A sequence of equidistant water depths in each river segment For the first The maximum possible water depth of each section of the river; Represents water depth sequence The number of water depth values ​​in the middle; Indicates the depth of the river channel. At that time, the first The number of slope grids submerged outside the river channel in each river section; The number of river sections; The formula for calculating the flood inundation volume outside the river channel in each section is as follows: （7） in, For the first The amount of floodwater inundated in the catchment area of ​​each river section.

4. The rapid simulation method for river overflow flood evolution based on river adaptive segmentation and strong water balance constraints as described in claim 2, characterized in that: In step S15, the process of constructing the segmented water depth-water volume relationship curve is as follows: Based on the equidistant water depth sequence of each river segment The formula for calculating the flood inundation volume corresponding to each water depth is as follows: （8） （9） in, The depth of the river channel is At that time, the flood inundation volume in the catchment area of ​​that river section; This represents the number of grid cells within the catchment area of ​​this river section. When the river is deep At that time, the first in the catchment area of ​​that river section The flood inundation volume of each grid cell; For the first The HAND value of each grid cell; For the first The area of ​​each grid cell; Based on the equidistant water depth sequence of the river segment With corresponding water volume sequence By observing the relationship, the water depth-water volume relationship curve of the river section can be obtained.

5. The rapid simulation method for river overflow flood evolution based on river adaptive segmentation and strong water balance constraints according to claim 2, characterized in that: Between steps S11 and S12, there is also, The calculated river network distribution data of the watershed was corrected using Tianditu remote sensing imagery to ensure that the river distribution is consistent with the actual location.

6. The rapid simulation method for river overflow flood evolution based on river adaptive segmentation and strong water balance constraints according to any one of claims 1 to 5, characterized in that: Following the construction of a rapid calculation model for river overflow flooding, the following also includes: Rapid prediction and risk analysis of river overflow floods: Based on the constructed rapid analysis model of river overflow flood inundation, and by incorporating future hydrological forecast data, the development and evolution of river floods in the basin are predicted in real time, and the dynamic inundation range and water depth of river overflow floods in the basin are obtained; flood inundation risk indicators are established based on the dynamic inundation range and water depth of river overflow floods in the basin, and a dynamic risk map of floods in the basin is drawn based on the flood inundation risk indicators.

7. The rapid simulation method for river overflow flood evolution based on river adaptive segmentation and strong water balance constraints as described in claim 6, characterized in that: Rapid prediction and risk analysis of river overflow floods specifically includes the following: S21. Watershed flood inundation prediction: Based on the upstream section flow forecast results, drive the rapid analysis model of river overflow flood inundation to predict the watershed flood inundation range and water depth data in real time. S22. Construction of flood inundation risk indicators: Taking into full account the threat posed by flood inundation depth to the safety of people's lives and property and social and economic security, as well as the standards for compiling flood risk maps, flood inundation depths of 0.5m, 1m, 2m, and 3m are set as indicators for low, medium, high, and extremely high risks of flood inundation in the basin, respectively, for extreme flood events. S23. Real-time flood risk prediction and risk analysis of key protected objects: Based on real-time inundation prediction data and combined with flood inundation risk indicators, draw a dynamic flood risk map; overlay geospatial information layers such as flood control projects, lifeline projects, and key protected objects on the dynamic flood risk map to draw a thematic risk map.

8. The rapid simulation method for river overflow flood evolution based on river adaptive segmentation and strong water balance constraints according to any one of claims 1 to 6, characterized in that: Before building a rapid computational model for river overflow and flood inundation, the following also included: Watershed basic data collection and processing: Collect and process high-resolution digital elevation data, river channel and levee data and hydrological data of the watershed.

9. The rapid simulation method for river overflow flood evolution based on river adaptive segmentation and strong water balance constraints as described in claim 8, characterized in that: The collection and processing of basic watershed data specifically includes the following: S01. Download meter-level digital elevation data (DEM) of the watershed using a map downloader, with a spatial resolution of 30m, 12.5m, or 5m; collect river embankment top elevation data of the watershed, and correct the river embankment top elevation data based on the digital elevation data to make the elevation data inside and outside the river and the embankment more consistent with the actual topography of the watershed. S02. Collect measured flow data from river hydrological stations in the basin and standardize the time interval of the flow data to 1 hour.

Images