A method for identifying key flood control weak points of a city and a basin based on space-time array position analysis

Abstract

The application discloses a kind of based on confirmation method of city, basin key flood prevention weak point of space-time array position analysis, comprising the following steps: S1: constructing flood source-sink relationship database, flood data of target area are collected to establish standardized data set;S2: the research area is divided into multiple spatial units, and the space-time response relationship model of regional unit and flood source is constructed;S3: based on high-fidelity hydrodynamic model and historical flood data, dimensionless disaster response index is constructed;S4: according to the disaster response similarity index of each unit and flood source obtained in S3, the spatial distribution graph with weak level coding is generated.The application has the advantages compared with prior art: provide a kind of based on confirmation method of city, basin key flood prevention weak point of space-time array position analysis, which is convenient to operate and use, realize the quick, accurate, automatic identification of flood prevention weak point under different flood scenarios.

Description

Technical Field

[0001] This invention relates to the field of urban flood control and watershed risk early warning technology, specifically to a method for identifying key flood control weaknesses in cities and watersheds based on spatiotemporal position analysis. Background Technology

[0002] With the acceleration of global climate change and urbanization, extreme rainstorms and floods are becoming more frequent, and the risks of urban flooding and watershed-wide flooding are increasing. Scientifically and accurately identifying the weaknesses in flood control systems is a crucial prerequisite for effective flood prevention and disaster reduction, optimized engineering layout, and emergency management.

[0003] Traditional methods often employ high-fidelity two-dimensional hydrodynamic models to simulate flood processes and identify risk areas. While these models can accurately depict the details of water flow, they are computationally complex, time-consuming, and have stringent requirements for data input (such as high-precision topography and hydrological boundary conditions) and hardware computing power. They are difficult to apply efficiently to rapid assessment under multiple scenarios (such as different return periods and different precipitation combinations), and it is also difficult to achieve quantitative analysis of the dynamic response relationship between flood sources and inundated areas.

[0004] Therefore, there is an urgent need for a new analytical method that can quickly, accurately, and comprehensively assess the risk of multi-source floods and adapt to different design scenarios to overcome the shortcomings of existing technologies. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to overcome the above-mentioned technical defects and provide a method for identifying key flood control weak points in cities and watersheds based on spatiotemporal position analysis, which is easy to operate and use and enables rapid, accurate and automated identification of flood control weak points under different flood scenarios.

[0006] To address the aforementioned technical problems, the present invention provides the following technical solution: a method for identifying key flood control weaknesses in cities and watersheds based on spatiotemporal array analysis, comprising the following steps: S1: Construct a flood source-sink relationship database, collect multi-source flood data of the target area, including historical flood event data, real-time monitoring data and remote sensing inversion data, and standardize the multi-source heterogeneous data through artificial intelligence-based data cleaning algorithms and outlier detection models to form a spatiotemporally consistent flood source-sink dataset; S2: Using topographic elevation, slope, and hydrological distance from the flood source as feature vectors, principal component analysis (PCA) is used for feature dimensionality reduction and interpretability extraction. An improved K-nearest neighbor (K-NN) algorithm is used to optimize the number of clusters based on the silhouette coefficient, dividing the study area into multiple spatial response units and constructing a spatiotemporal response mapping model between the units and the flood source. S3: Based on a high-fidelity hydrodynamic model, design flood hydrograph parameters from different flood sources are input to simulate the temporal changes in inundation depth and inundation area of ​​each spatial unit under different scenarios. Combining historical flood process data, the dynamic time warping (DTW) algorithm is used to calculate the morphological similarity between the inundation sequence of each unit and the flood source process sequence, and the Spearman rank correlation coefficient is used to quantify the consistency of their changing trends. The maximum inundation depth, inundation area, and process time of each data set are normalized by range processing to construct a dimensionless disaster response index. S4: Based on the disaster response similarity index between each unit and the flood source obtained in S3, an unsupervised clustering algorithm or the adaptive interquartile range (IQR) method is used to automatically determine the threshold of weak points. Units with values ​​higher than the threshold are identified as key flood control weak points, and a spatial distribution map with weak point level codes is generated and visualized.

[0007] Preferably, the flood data in S1 includes historical flood inundation data, flood source data, and geographic information data, and outlier processing, data standardization, and spatiotemporal matching are performed based on the acquired data information; The historical flood inundation data sources include field observation data, remote sensing interpretation data, and high-fidelity hydrodynamic model simulation data.

[0008] Preferably, the historical flood inundation data includes time-series information on the inundation range, inundation depth, and duration; Flood source data includes river water level, flow rate, peak flow rate, and time of occurrence, as well as rainfall, rainfall intensity, rainfall duration, and spatial distribution. Geographic information data includes land use types, river network density and distribution, and terrain slope.

[0009] Preferably, in step S2, a machine learning algorithm is used to establish the nonlinear response relationship between the flood source and the inundation zone; The influence weights of river floods were determined using spatiotemporal lag correlation analysis. Spatial correlations are extracted by combining precipitation-type floods with topographic features.

[0010] Preferably, the scenario parameters in S3 include flood intensity, duration, and spatial distribution characteristics.

[0011] Preferably, the machine learning algorithm includes one or more combinations of random forest and artificial neural network.

[0012] Preferably, S4 includes identifying key flood control weaknesses by using an unsupervised clustering algorithm or an adaptive threshold determination method based on the disaster response similarity index between each spatial unit and the flood source; The weakness level is divided based on the numerical range of the similarity index.

[0013] Preferably, the visualization output in S4 includes a spatial distribution map of key flood control weak points and corresponding weak point level codes.

[0014] The advantages of this invention compared to existing technologies are as follows: it introduces the systematic thinking of "spatiotemporal position" and "source-sink relationship", uses machine learning models (such as improved K-NN, unsupervised clustering, etc.) to replace some of the complex hydrodynamic model calculations, reduces the computational cost through the spatiotemporal response relationship model, can quickly respond to a variety of design scenarios, and combines multi-source data for training and verification, thereby improving the accuracy and reliability of vulnerability identification and enabling predictive risk assessment. Attached Figure Description

[0015] Figure 1 This is a flowchart of a method for identifying key flood control weaknesses in cities and watersheds based on spatiotemporal position analysis. Detailed Implementation

[0016] The present invention will now be described in further detail with reference to the accompanying drawings.

[0017] Combined with appendix Figure 1 As shown, a method for identifying key flood control weaknesses in cities and watersheds based on spatiotemporal position analysis includes the following steps: S1: Construct a flood source-sink relationship database, collect multi-source flood data of the target area, including historical flood event data, real-time monitoring data and remote sensing inversion data, and standardize the multi-source heterogeneous data through artificial intelligence-based data cleaning algorithms and outlier detection models to form a spatiotemporally consistent flood source-sink dataset; S2: Using topographic elevation, slope, and hydrological distance from the flood source as feature vectors, principal component analysis (PCA) is used for feature dimensionality reduction and interpretability extraction. An improved K-nearest neighbor (K-NN) algorithm is used to optimize the number of clusters based on the silhouette coefficient, dividing the study area into multiple spatial response units and constructing a spatiotemporal response mapping model between the units and the flood source. S3: Based on a high-fidelity hydrodynamic model, design flood hydrograph parameters from different flood sources are input to simulate the temporal changes in inundation depth and inundation area of ​​each spatial unit under different scenarios. Combining historical flood process data, the dynamic time warping (DTW) algorithm is used to calculate the morphological similarity between the inundation sequence of each unit and the flood source process sequence, and the Spearman rank correlation coefficient is used to quantify the consistency of their changing trends. The maximum inundation depth, inundation area, and process time of each data set are normalized by range processing to construct a dimensionless disaster response index. S4: Based on the disaster response similarity index between each unit and the flood source obtained in S3, an unsupervised clustering algorithm or the adaptive interquartile range (IQR) method is used to automatically determine the threshold of weak points. Units with values ​​higher than the threshold are identified as key flood control weak points, and a spatial distribution map with weak point level codes is generated and visualized.

[0018] When used, the data in S1 includes: Historical flood inundation data: derived from field observations, remote sensing image interpretation, and high-fidelity hydrodynamic model simulations, this time-series data includes information such as inundation range, inundation depth, duration, and receding time. Outlier processing, data standardization, and spatiotemporal matching are performed using AI-based data cleaning algorithms and outlier detection models to ensure data consistency and comparability.

[0019] Flood source data includes river hydrological data: water level, flow rate, peak flow rate and occurrence time, and precipitation data: rainfall amount, rainfall intensity, precipitation duration and spatial distribution.

[0020] Geographic information data includes land use types, river network density and distribution, topographic slope, and distribution of drainage facilities.

[0021] Outlier handling, data standardization, and spatiotemporal matching were performed on the above data to ensure data consistency and comparability; In S2, topographic elevation, slope, and hydrological distance from the flood source are used as feature vectors. Principal component analysis (PCA) is used for feature dimensionality reduction and interpretability extraction. An improved K-nearest neighbor (K-NN) algorithm is used to optimize the number of clusters based on the silhouette coefficient, dividing the study area into multiple spatial response units. For river floods, spatiotemporal lag correlation analysis is used to determine the lag time and influence weight of the water level / discharge process lines of different upstream sections and the inundation situation of this unit. For precipitation-type floods, the correlation between the unit's topographic slope, confluence path, impervious area ratio, and other topographic features is extracted and the spatial distribution of precipitation is extracted. Machine learning algorithms, including one or more combinations of random forest, artificial neural network, support vector machine, or gradient boosting tree, are used to establish a nonlinear mapping relationship between flood source features and confluence response, and to construct a spatiotemporal response mapping model between the unit and the flood source. The S3 design scenario parameters include flood intensity, duration, and spatiotemporal distribution patterns with different return periods. Based on a high-fidelity hydrodynamic model, design flood process parameters from different flood sources are input to simulate the temporal changes in inundation depth and inundation area of ​​each spatial unit under different scenarios. Combining historical flood process data, the dynamic time warping (DTW) algorithm is used to calculate the morphological similarity between the inundation sequence of each unit and the flood source process sequence, and the Spearman rank correlation coefficient is used to quantify the consistency of their changing trends. The maximum inundation depth, inundation area, and process time of each data set are normalized by range to construct a dimensionless disaster response index. Based on the disaster response similarity index between each unit and the flood source obtained from S3, unsupervised clustering algorithms (such as K-means) or adaptive interquartile range (IQR) are used to automatically determine the threshold of weak points, and units with values ​​higher than the threshold are identified as key flood control weak points. At the same time, based on three key indicators, namely the proportion of inundated area, the maximum water depth, and the time of water receding, the comprehensive vulnerability index of each spatial unit is calculated by weighted superposition and other methods. The comprehensive vulnerability indices of all units are sorted, and a threshold is set. Units with index values ​​higher than the threshold are identified as key flood control weak points, and a spatial distribution map with a weakness level code is generated and visualized.

[0022] The contents not described in detail in this specification are existing technologies known to those skilled in the art.

[0023] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can include direct contact between the first and second features, or contact between the first and second features through another feature between them. Furthermore, "above," "over," and "on top" of the second feature includes the first feature being directly above or diagonally above the second feature, or simply indicates that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature includes the first feature being directly below or diagonally below the second feature, or simply indicates that the first feature is at a lower horizontal level than the second feature. The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A method for identifying key flood control weaknesses in cities and watersheds based on spatiotemporal position analysis, characterized in that: Includes the following steps: S1: Construct a flood source-sink relationship database, collect multi-source flood data of the target area, including historical flood event data, real-time monitoring data and remote sensing inversion data, and standardize the multi-source heterogeneous data through artificial intelligence-based data cleaning algorithms and outlier detection models to form a spatiotemporally consistent flood source-sink dataset; S2: Using topographic elevation, slope, and hydrological distance from the flood source as feature vectors, principal component analysis is used for feature dimensionality reduction and interpretability extraction. An improved K-nearest neighbor algorithm is used to optimize the number of clusters based on the silhouette coefficient, dividing the study area into multiple spatial response units and constructing a spatiotemporal response mapping model between the units and the flood source. S3: Based on a high-fidelity hydrodynamic model, design flood hydrograph parameters from different flood sources are input to simulate the temporal changes in inundation depth and inundation area of ​​each spatial unit under different scenarios. Combining historical flood process data, the dynamic time warping algorithm is used to calculate the morphological similarity between the inundation sequence of each unit and the flood source process sequence, and the Spearman rank correlation coefficient is used to quantify the consistency of their changing trends. The maximum inundation depth, inundation area, and process time of each data set are normalized by range to construct a dimensionless disaster response index. S4: Based on the disaster response similarity index between each unit and the flood source obtained in S3, an unsupervised clustering algorithm or an adaptive interquartile range method is used to automatically determine the threshold of weak points. Units with values ​​higher than the threshold are identified as key flood control weak points, and a spatial distribution map with weak point level codes is generated and visualized.

2. The method for identifying key flood control weaknesses in cities and watersheds based on spatiotemporal array analysis according to claim 1, characterized in that: In S1, the flood data includes flood inundation data in flood reservoirs, flood source data, and geographic information data; the AI-based data cleaning and outlier detection algorithm is used to perform outlier processing, data standardization, and spatiotemporal matching on the acquired multi-source heterogeneous data. The flood inundation data sources include field observation data, remote sensing interpretation data, and high-fidelity hydrodynamic model simulation data.

3. The method for identifying key flood control weaknesses in cities and watersheds based on spatiotemporal array analysis according to claim 2, characterized in that: The flood inundation data includes time-series information on the inundation range, inundation depth, and duration. Flood source data includes river level, flow rate, peak flow rate, and occurrence time; rainfall amount, rainfall intensity, rainfall duration and spatial distribution; and storm surge height, duration and affected area. Geographic information data includes elevation, slope, land use type, river network density and distribution.

4. The method for identifying key flood control weaknesses in cities and watersheds based on spatiotemporal array analysis according to claim 1, characterized in that: In S2, a machine learning algorithm is used to establish the nonlinear response relationship between the flood source and the inundation zone; The influence weights of river floods were determined using spatiotemporal lag correlation analysis. Spatial correlations are extracted by combining precipitation-type floods with topographic features.

5. The method for identifying key flood control weaknesses in cities and watersheds based on spatiotemporal array analysis according to claim 1, characterized in that: The scenario parameters in S3 include flood intensity, duration, and spatial distribution characteristics.

6. The method for identifying key flood control weaknesses in cities and watersheds based on spatiotemporal array analysis according to claim 4, characterized in that: The machine learning algorithm includes one or more combinations of random forest and artificial neural network.

7. The method for identifying key flood control weaknesses in cities and watersheds based on spatiotemporal array analysis according to claim 1, characterized in that: Based on the disaster response similarity index between each spatial unit and the flood source, an unsupervised clustering algorithm or an adaptive threshold determination method is used to identify key flood control weaknesses. The weakness level is divided based on the numerical range of the similarity index.

8. The method for identifying key flood control weaknesses in cities and watersheds based on spatiotemporal array analysis according to claim 1, characterized in that: The visualization output in S4 includes a spatial distribution map of key flood control weak points and corresponding weak point level codes.

Images