A waterlogging risk dynamic early warning method and system based on space-time aging response

Abstract

The application relates to the technical field of meteorological data processing and risk early warning, in particular to an inner submergence risk dynamic early warning method and system based on space-time time-effect response. The application scheme constructs a first time-effect response zone of an inner submergence point and a second time-effect response zone corresponding to the inner submergence point based on a time-effect response mechanism, forms double time-effect response zones, and then according to cloud cluster information and a preset early warning and prediction processing mechanism, the current cloud cluster is tracked and predicted and / or the inner submergence risk is early warned in a targeted manner. The application further organically integrates "spatial hierarchical monitoring and time-effect condition triggering", constructs a dynamic adaptive time-effect response zone and an early warning and prediction processing mechanism for the inner submergence water point, can guarantee that radar extrapolation information has relatively stable prediction accuracy, and can guarantee that the response emergency time of thunderstorm clouds from entering the time-effect response zone to reaching the inner submergence water point.

Description

Technical Field

[0001] This application relates to the field of meteorological data processing and risk early warning technology, specifically to a dynamic early warning method and system for urban flooding risk based on spatiotemporal response. Background Technology

[0002] In recent years, against the backdrop of global warming, heavy precipitation events have shown an increasing frequency and intensity, leading to a continuous exacerbation of the frequency, intensity, and losses caused by urban flooding disasters in various regions. Urban flooding meteorological risk early warning, based on precipitation forecasts from meteorological departments and combined with urban or regional underlying surface conditions, provides advance warning of potential urban flooding risks and their risk levels. Accurate, timely, and reliable urban flooding meteorological risk early warning helps to shift disaster prevention and emergency response models from passive response to proactive prevention.

[0003] However, existing urban flooding warning schemes or models suffer from a large number of false alarms. The main reason is the random uncertainty of rain clouds. Their extrapolated paths and movement speeds may vary greatly, leading to spatial misalignment between the predicted rainfall area and the actual rainfall area, thus rendering the warning ineffective or false.

[0004] In urban flooding warnings, a rainfall warning zone is typically established centered on the flooded area. A warning is triggered when a cloud mass enters this zone. Traditional urban flooding warning zones are divided using three main methods: the fixed radius method (setting a predetermined radius or distance around the flooded area), the administrative boundary method (setting a warning zone within a predetermined administrative region where the flooded area is located), and the isochronous method (drawing isochronous boundaries based on a predetermined cloud mass movement speed). However, these methods essentially create fixed, static warning zones, which suffer from two major drawbacks: First, they fail to establish a balance between "forecast reliability" and "early warning lead time." The warning zone or boundary lacks scientific basis based on radar extrapolation error statistics and the response time of the flooded area. If the zone is too large, the cloud mass will be far away when it enters, leading to high extrapolation uncertainty and potentially false alarms. If the zone is too small, the warning may be triggered too late, resulting in insufficient emergency preparedness and a failure to balance accuracy and timeliness in the timing of the warning. Secondly, static boundaries cannot adapt to the real-time dynamic changes in the intensity and speed of thunderstorm clouds. The actual movement of clouds is often accompanied by nonlinear changes such as acceleration, deceleration, enhancement, and weakening, which can easily lead to warning failure and errors. Summary of the Invention

[0005] Based on this, this application proposes a dynamic early warning method and system for urban flooding risk based on spatiotemporal and timely response, aiming to improve the stability and accuracy of urban flooding early warning and significantly reduce false alarms and misreports.

[0006] This application provides a dynamic early warning method for urban flooding risk based on spatiotemporal response, the method comprising: Construct the first time-sensitive response zone for the waterlogging point based on the surface hydrological characteristics of the waterlogging point; Based on historical cloud cluster data, a second time-sensitive response zone corresponding to the flood-prone area is constructed, and the corresponding time-sensitive response time is determined; Extract the target cloud clusters of heavy precipitation based on the current radar mosaic, and monitor the position and motion vector of the current target cloud clusters of heavy precipitation; If the current cloud cluster enters or approaches the first-response zone, the flood risk warning and prediction will be triggered. If the target cloud cluster enters or approaches the second time-response zone corresponding to the flooding point, the tracking and monitoring process will be initiated, and a flood risk warning and prediction will be made based on the time-response time and the movement vector of the target cloud cluster.

[0007] Preferably, in one embodiment, constructing a first time-response zone for a flood-prone area includes: Collect high-resolution DEM data of the area surrounding the flood-prone area; Flow direction analysis and calculation of waterlogging points based on unidirectional flow direction algorithm; Extract surface runoff paths, classify runoff paths into different levels, and determine the catchment areas of waterlogging points; Based on DEM topographic data, the upstream surface runoff path is determined, and the spatial range and boundary of the surface water flow that eventually flows into the flood-prone area are extracted to obtain the first time-response zone of the flood-prone area.

[0008] Preferably, in one embodiment, constructing a second time-response zone corresponding to the flooding point includes: Collect historical heavy rainfall events at waterlogged areas and track the movement trajectory of historical heavy rainfall echoes that have passed through these areas; Based on the movement trajectory of historical heavy precipitation echoes, calculate the radial velocity and azimuth of the corresponding historical heavy precipitation cloud cluster relative to the waterlogging point; Centered on the flood-prone area, the surrounding area is divided into multiple directional zones. The maximum radial velocity of historical heavy precipitation cloud clusters in various regions was statistically analyzed to determine the warning radius and vertex coordinates of each region, and a second time-limited response zone corresponding to the flooding point was constructed. The relationship between false alarm rate and extrapolation lead time is generated. The false alarm rate is used as the control target to determine the reliable extrapolation response time, and the lead time response time T corresponding to the second lead time zone is determined. rad .

[0009] In one embodiment, the method further includes: Calculate the relative distance from each tracking point to the flooded point, and obtain the radial velocity of each tracking point relative to the flooded point based on the time interval between two adjacent tracking points; Determine the azimuth angle of each tracking point relative to the flooded point based on the coordinates of each tracking point; Select preset effective tracking points, and calculate the radial velocity and azimuth of historical target heavy precipitation cloud clusters relative to the waterlogging points based on the effective tracking points.

[0010] Preferably, in one embodiment, the method further includes: Based on the classification of heavy precipitation cloud clusters by azimuth region, the historical number and maximum radial velocity v of heavy precipitation cloud clusters in each azimuth region are statistically analyzed. max,d ; Based on the preset target time t, determine the warning radius R corresponding to each location area. d : , Determine the vertex coordinates corresponding to each location area based on the warning radius of each location area; The second time-sensitive response zone corresponding to the flooding point is constructed based on the vertices corresponding to each directional zone.

[0011] In one embodiment, the method further includes: Receive the latest radar composite reflectivity mosaic at preset time intervals; Independent cloud clusters were segmented based on the connected component labeling method, and target cloud clusters with heavy precipitation whose combined reflectance was greater than a preset threshold were extracted. The target cloud cluster is identified and converted into a grid mask, and spatial intersection analysis is performed with the second time-response area and / or the first time-response area; If there is overlap, it is determined that the target cloud cluster has entered or is approaching the corresponding second-time response zone and / or first-time response zone.

[0012] Preferably, in one embodiment, initiating the tracking and monitoring process includes: Increase tracking frequency to extract and / or monitor target cloud clusters; Real-time updates calculate the target cloud's direction of movement, speed of movement, and the time T it takes for the target cloud to reach the flooded area. arrival If the current T arrival ≤T rad If so, an early warning prediction will be triggered.

[0013] Preferably, the early warning judgment includes: Based on the motion vector of the current target cloud cluster, generate a future precipitation estimate for the catchment area of ​​the flood-prone point within a preset time. Based on the flow relationship between each catchment area and the flood-prone point, the total flow rate and water depth entering the flood-prone point are calculated. The meteorological risk level of waterlogging at the waterlogging point is generated based on the water depth, and an early warning is issued.

[0014] A second aspect of this application provides a dynamic early warning system for urban flooding risk based on spatiotemporal response, the system comprising: The first response zone construction unit is used to construct the first time-effect response zone of the waterlogging point based on the surface hydrological characteristics of the waterlogging point. The second response zone construction unit is used to construct the second time-sensitive response zone corresponding to the flooding point based on historical cloud cluster data, and to determine the corresponding response time parameters. The cloud extraction unit is used to extract heavy precipitation clouds based on the current radar mosaic and to monitor the position and motion vector of the current heavy precipitation target cloud. The early warning and prediction unit is used to trigger the activation of waterlogging risk early warning and prediction if the current cloud cluster enters or approaches the first time-response zone; if the target cloud cluster enters or approaches the second time-response zone corresponding to the waterlogging point, it tracks and monitors the target cloud cluster and determines whether to issue a waterlogging risk early warning and prediction based on the time-response time and the movement vector of the target cloud cluster.

[0015] A third aspect of this application provides a computer-readable storage medium storing a computer program, characterized in that, when the computer program is executed by a processor, the processor performs the steps of the above-described method.

[0016] A fourth aspect of this application provides a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that, when the computer program is executed by the processor, the processor performs the steps of the above-described method.

[0017] The urban flooding meteorological risk early warning scheme provided in this application constructs a first-time-response zone and a second-time-response zone corresponding to the urban flooding point based on a time-response mechanism, forming a dual-time-response zone. Then, based on cloud cluster information and a pre-set early warning and prediction processing mechanism, it can specifically track and predict current cloud clusters and / or issue early warnings of urban flooding risks. This scheme organically integrates "spatial layered monitoring and time-sensitive triggering" to construct a dynamic adaptive time-response zone and early warning and prediction processing mechanism for urban flooding points. This ensures that radar extrapolation information has relatively stable forecast accuracy after thunderstorm clouds move into the time-response zone, and also guarantees the emergency response time from when thunderstorm clouds enter the time-response zone to when they reach the urban flooding point.

[0018] Furthermore, the first time-response zone constructed in this application is delineated based on high-precision topographic and flow direction analysis of the flood-prone areas, forming an "instant response zone." Within this zone, once heavy rainfall occurs, the risk of flooding will rapidly increase with virtually no uncertainty from extrapolated paths, thus minimizing response delays. Simultaneously, the second time-response zone is an asymmetric dynamic response zone driven by azimuth based on historical cloud data. Its anchored early warning window ensures sufficient monitoring lead time while conserving computational resources. By establishing a linkage mechanism with time-response timeliness, false alarms caused by distant, unreliable cloud clusters are effectively filtered out. The asymmetric design by azimuth allows the shape of the response zone to dynamically change with the direction and properties of the cloud clusters, overcoming the drawback of ignoring the physical laws of cloud movement in fixed-area applications. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] in: Figure 1 This is a flowchart of a dynamic early warning method for urban flooding risk based on spatiotemporal response in one embodiment; Figure 2 This is a schematic diagram of the first-time response zone for a flood-prone area constructed in one embodiment; Figure 3 This is a schematic diagram of the second time-response zone for an urban flooding point constructed in one embodiment; Figure 4 This is a graph showing the changes of CSI, POD, and FAR with extrapolation time, generated in one embodiment of this application. Figure 5 This is a structural block diagram of a computer device in one embodiment; Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0022] The terms "comprising," "including," and "having," and any variations thereof, in the specification, claims, and accompanying drawings of this application, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such processes, methods, products, or apparatus. Terms such as "first" and "second," and other relational terms, in the claims, specification, and accompanying drawings of this application, are used merely to distinguish one entity / operation / object from another, and do not necessarily require or imply any such actual relationship or order between these entities / operations / objects.

[0023] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase at various points in the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0024] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0025] In one implementation, such as Figure 1 The diagram shows a flowchart of a dynamic early warning method for urban flooding risk based on spatiotemporal response according to this application. The method includes: S10. Construct the first time-effect response zone for the waterlogging point based on the surface hydrological characteristics of the waterlogging point.

[0026] Specifically, the first time-response zone for waterlogging points constructed in this application is a "spatial area" requiring immediate response to waterlogging warnings and predictions. In one embodiment, to construct this first time-response zone reasonably and accurately, this application refers to an upstream catchment area delineated based on high-precision topographic and flow direction analysis, where all surface runoff naturally converges towards the waterlogging hazard point. Rainfall within this area directly converges towards the waterlogging hazard point; once heavy rainfall occurs within this area, the risk of waterlogging will rapidly increase with almost no uncertainty related to extrapolation paths. Preferably, the aforementioned first time-response zone for waterlogging points includes: S101. Collect high-resolution DEM data of the area surrounding the flood-prone area.

[0027] Specifically, in order to accurately obtain the micro-topographic features and surface runoff information around the flood-prone areas, this application proposes a solution based on a high-resolution digital elevation model (DEM) of lidar to obtain refined surface elevation data around the flood-prone areas. The DEM data is then preprocessed to fill depressions, eliminating local depression artifacts in the topographic data and ensuring the continuity and rationality of the water flow direction.

[0028] S102. Perform flow direction analysis and calculation based on the preset single flow direction algorithm.

[0029] Preferably, the proposed solution employs the D8 algorithm (Deterministic Eight-node algorithm, also known as the maximum slope method) for flow direction analysis. For each raster cell in the DEM, the maximum average slope criterion is used to determine the flow direction. It is assumed that rainwater falling on each raster cell will flow towards the downstream cell with the steepest slope among its eight neighboring cells. The slope is calculated using the maximum average slope S. max :

[0030] Where h0 is the elevation of the center pixel, h i Let dᵢ be the elevation of the i-th neighboring pixel, and dᵢ be the distance between the center pixel and the i-th neighboring pixel. The diagonal neighborhood is taken as... orthogonal neighborhood , The raster resolution is used. The flow direction code calculated by the D8 algorithm uses eight values: 1, 2, 4, 8, 16, 32, 64, and 128, which correspond to the eight flow directions: east, southeast, south, southwest, west, northwest, north, and northeast, respectively. This facilitates computer storage and subsequent convergence and accumulation calculations.

[0031] S103. Extract surface runoff paths, classify runoff paths by level, and determine the catchment areas of waterlogging points.

[0032] Specifically, based on flow direction analysis and confluence results, continuous surface runoff paths are extracted, and a regional vector grid network is constructed. Simultaneously, to further describe the topological structure and hierarchical relationship of the runoff paths, this application adopts the Strahler classification method to classify the runoff paths. The basic rule is: an initial arc segment without tributaries is defined as level 1; the level of the new arc segment increases by one level if and only if two arc segments of the same level merge into a new arc segment; if a lower-level arc segment merges into a higher-level arc segment, the level of the higher-level arc segment remains unchanged. Let S... i If the Strahler level is for arc segment i, then:

[0033] in, The maximum level among all upstream tributaries flowing directly into arc segment i is defined as follows: when the maximum level appears twice or more among all tributaries flowing into arc segment i... =1, otherwise =0.

[0034] Furthermore, based on runoff path extraction and Strahler classification, this application's scheme discretizes the area surrounding the flood-prone point into several spatially non-overlapping catchment units, and performs catchment zoning, specifically including: First, the river network links are encoded using graph theory partitioning: the confluence points and river initiation points in the runoff network are used as partitioning nodes, and the runoff arc between any two adjacent partitioning nodes is defined as an independent river network link and assigned a unique number, thus discretizing the continuous river network into basic units that can be processed individually. Second, based on the flow direction grid, the downstream endpoint of each river network link is traced along the flow direction, and the geometric center of the endpoint grid is automatically extracted as the outlet (drainage point). Finally, using all outlet points as input, a reverse catchment area backtracking algorithm is executed: starting from each outlet grid, the algorithm searches upstream level by level according to the pointer-back relationship of the flow direction grid, classifying all grid units that eventually flow into the outlet into the same sub-basin, with the boundaries of adjacent sub-basins naturally corresponding to topographic watersheds (elevation ridges). After completing the above analysis, a raster-type sub-basin encoded map is output. After raster-to-vector conversion, a planar vector catchment area with complete topological relationships is finally formed.

[0035] S104. Based on DEM topographic data, determine the upstream surface runoff path and extract the spatial range and boundary of the surface water flow that eventually flows into the waterlogging point to obtain the first time response zone of the waterlogging point.

[0036] Specifically, this application's scheme uses the flood-prone area as the dumping point. Based on the DEM topography and water flow direction, it determines the topographic watershed by tracing the confluence path from the source, and draws all upstream surface runoff paths naturally flowing towards the flood-prone area. Further, based on runoff path analysis and the results of using the flood-prone area as the lowest outflow point in the region, it extracts the spatial range and boundary of the flood-prone area's catchment area by combining the water flow direction. This catchment area is the regional range of the first-time response zone corresponding to the flood-prone area. Through the above steps, it outputs spatial polygon vector data corresponding to each flood-prone area, which can be converted into a spatial index (such as a grid mask) for real-time use by the early warning system. Figure 2 The diagram shown is a schematic diagram of the first time-effect response zone of a waterlogging point constructed based on the topographic information of a waterlogging point in one embodiment of this application. In the diagram, the path paths 1-6 are runoff paths of different levels formed based on step S103.

[0037] The above-mentioned scheme in this application is based on the first time-sensitive response zone delineated by high-precision topographic and flow direction analysis. It takes into account the real-time response mechanism and the influence of topographic runoff, which can effectively avoid the uncertainty of extrapolation path and avoid the false alarms and missed alarms caused by the neglect of runoff direction in the traditional warning zone method.

[0038] S11. Construct the second time-sensitive response zone corresponding to the waterlogging point based on historical cloud cluster data, and determine the corresponding time-sensitive response time.

[0039] In one embodiment, the second time-response zone constructed in this application is an asymmetric spatiotemporal monitoring area based on historical heavy rainfall cloud cluster statistics, anchored to a preset target time as the early warning window, and dynamically expanding and contracting with the direction of the cloud cluster. The design basis for constructing the second time-response zone in this application includes: first, anchoring the time-response to a preset target time to ensure that the time-response duration is within the radar extrapolation confidence limit; second, dividing the area into multiple azimuth sectors, considering the different impacts of spatial terrain on cloud cluster movement and dissipation in different directions, and conducting independent statistics. Preferably, the second time-response zone constructed in this application corresponding to the flooding point specifically includes: S111. Collect historical heavy rainfall events at waterlogged areas and track the movement trajectory of historical heavy rainfall echoes passing through the waterlogged areas.

[0040] Preferably, considering that the number of waterlogging records corresponding to potential waterlogging points may be relatively small and difficult to support statistical modeling, this application uses short-duration heavy rainfall cases as effective substitute samples. The two are closely related physically and have sufficient sample size. Centered on each waterlogging point, hourly rainfall observation data from the nearest meteorological station over the past three years are collected to screen all historical short-duration heavy rainfall events: the start time of the event is defined as the moment when hourly rainfall increases from 0 to ≥0.1 mm, and the end time is defined as the moment when hourly rainfall drops back to 0 mm. During the event, at least one hour of rainfall must be ≥20 mm. For each selected short-duration heavy rainfall event, real-time radar combined reflectivity factor data are collected from the three hours before its start to the end. Furthermore, the movement trajectory of historical heavy rainfall echoes passing through the waterlogging points is tracked, specifically including: S1111, Construct a time series of radar reflectivity grayscale images.

[0041] This application's scheme is based on the LK optical flow method to trace texture features in images. It amplifies texture features through echo normalization, thus facilitating the calculation of optical flow vectors. In this scheme, echo data is first mapped to the 0-255 range, resulting in an image time series composed of radar combined reflectivity grayscale images for each time interval.

[0042] In the formula, x is the combined reflectance of the grid points, x min and x maxThese are the minimum and maximum thresholds for the normalized echo, such as -30 and 70 respectively.

[0043] S1112, Bidirectional trajectory tracking based on LK optical flow.

[0044] Specifically, the tracking starting point of this application is set as the radar grid point closest to the water accumulation point, and the tracking start time is the time when the combined reflectivity of this grid point is greater than a preset value (e.g., 35 dBz). Furthermore, bidirectional trajectory tracking based on the pyramid LK optical flow method is performed for each process and each start time to obtain the continuous movement trajectory of the heavy precipitation echo before and after passing the water accumulation point. The specific process is as follows: Step 1: For each starting time, take the radar grid point closest to the water accumulation point as the initial tracking point, and use this starting time as the anchor point to track time by time in both the future and past directions.

[0045] Step 2: During future direction tracking, starting from the current tracking time and the current tracking point, normalized grayscale images from two adjacent time intervals are taken. Using the current tracking point as the center, the displacement (i.e., optical flow vector) of the center point of the window between two frames is calculated according to the LK optical flow equation within a preset local window. The calculation of the optical flow vector between two frames is a technique known to those skilled in the art and will not be elaborated upon here. The current tracking point position is added to this optical flow vector to obtain the tracking point position for the next time interval. If the point passes the validity check, the tracking point for that time interval is recorded, and this serves as a new starting point to continue recursively tracking in the future direction.

[0046] Step 3: Based on a method similar to future direction tracking, past direction tracking (reversing the directions of two adjacent time steps) is performed. The displacement from the current time step to an earlier time step is calculated to obtain the tracking point position of the earlier time step, and this process continues to be recursively applied to the past direction. The entire tracking process adopts a pyramid strategy, first estimating the displacement at a low-resolution layer, and then passing it layer by layer to a high-resolution layer, finally obtaining the sub-pixel displacement at the original resolution.

[0047] Step 4: During bidirectional tracking, the validity of each new tracking point is assessed at each time step. Tracking in that direction is stopped if any of the following occurs: 1) the optical flow equation cannot be solved; 2) the new tracking point exceeds the radar grid range; 3) the combined reflectivity of the grid point containing the new tracking point and the grid point containing the water accumulation point is simultaneously missing. After both the future and past directions have terminated, the tracking points recorded in the two directions are merged in chronological order to obtain the complete trajectory of the strong echo corresponding to that initial time step. Through bidirectional tracking, each initial time step ultimately yields a sequence of tracking points centered on the initial time step and extending towards both ends of the time axis. Each tracking point is saved in sub-pixel coordinates to preserve the floating-point accuracy of the optical flow estimation.

[0048] Since the above-mentioned scheme of this application performs the bidirectional tracking independently for each starting time, if the water accumulation point reaches 35dBZ or more multiple times during the same heavy rainfall process, multiple echo movement trajectories starting from different starting times can be obtained. This allows for a more comprehensive depiction of the movement process of the heavy precipitation cloud before and after passing the water accumulation point, and thus enables more accurate tracking of the movement trajectory of the heavy precipitation cloud, obtaining the echo tracking point sequence corresponding to each starting time.

[0049] S112. Based on the movement trajectory of historical heavy precipitation echoes, calculate the radial velocity and azimuth of the corresponding historical heavy precipitation cloud cluster relative to the waterlogging point.

[0050] Based on the strong precipitation echo trajectory obtained from the above steps, the radial velocity and azimuth of each tracking point, each strong precipitation echo trajectory, and each strong precipitation cloud cluster are calculated sequentially to quantitatively describe the speed and direction of movement of historical strong precipitation clouds within a preset target time t before they reach the flooding point. Preferably, in one embodiment of this application, the preset target time t is set to 2 hours (according to statistics, this duration is within the radar extrapolation confidence limit). The specific value can be determined or selected according to the actual situation. Calculating the radial velocity and azimuth of the corresponding historical strong precipitation cloud cluster relative to the flooding point specifically includes: S1121. Calculate the relative distance from each tracking point to the flooded point, and obtain the radial velocity of each tracking point relative to the flooded point based on the time interval between two adjacent tracking points.

[0051] Specifically, each tracking point is obtained based on a preset time interval Δt (e.g., 6 minutes), and the latitude and longitude coordinates of each tracking point are acquired. i lat i The latitude and longitude coordinates of the waterlogged point are (lon) w lat w The spherical distance r from tracking point i to the flooded point is calculated using the Haversine formula. i :

[0052] Where R≈6371km, , .

[0053] The radial velocity of each tracking point relative to the flooded point is then obtained based on the time interval between two adjacent tracking points:

[0054] Among them, v i >0 indicates that the tracking point is close to the flooded area, v i <0 indicates that the tracking point is far from the flooded area.

[0055] S1122. Determine the azimuth angle of each tracking point relative to the flooded point based on the coordinates of each tracking point.

[0056] The azimuth angle of the tracking point in this application represents the direction from the water accumulation point to the tracking point, with true north as 0° and increasing clockwise. Specifically, the intermediate parameter x is first calculated based on the coordinates of each tracking point. i y i :

[0057]

[0058] Then, the azimuth angle θ of each tracking point relative to the flooded point is calculated and determined. i for:

[0059] θ i After calculation, the directions are further divided into N, NNE, NE, ENE, E, ESE, SE, SSE, S, SSW, SW, WSW, W, WNW, NW, NNW.

[0060] S1123. Select a preset effective tracking point, and calculate the radial velocity and azimuth of the historical target heavy precipitation cloud cluster relative to the waterlogging point based on the effective tracking point.

[0061] Specifically, for each tracking trajectory, tracking points are selected at the starting time and at a preset number of times n (e.g., 20) prior to it. These selected tracking points are then filtered, retaining only those whose radial velocities meet the specified requirements. The effective points are used to analyze the effective radial velocity and effective azimuth angle of the strong precipitation echo before it reaches the water accumulation point at a preset target time (e.g., 2 hours). The median or mean of the radial velocity and the circular mean of the azimuth angle of the effective points are used as the azimuth angle of the tracking trajectory. The circular mean θ of the azimuth angle of each tracking trajectory is... traj The calculation is as follows:

[0062] Further, based on the cloud cluster code, all heavy precipitation echo tracking trajectories within the same cloud cluster are summarized, and the arithmetic mean of the average radial velocity of each trajectory is taken as the average radial velocity of the historical target heavy precipitation cloud cluster. The circular mean of the average azimuth angle of each trajectory is taken as the average radial velocity of the historical target heavy precipitation cloud cluster.

[0063] S113. Taking the flood-prone area as the center, the area surrounding the flood-prone area is divided into multiple directional zones.

[0064] Specifically, to fully utilize the influence of spatial topography in different directions on the movement and dissipation of cloud clusters, this application's scheme takes the flood-prone area as the center and performs a 360° global scan of the area surrounding the flood-prone area, dividing it into multiple directional zones. For example, dividing the 360° circumference of the global scan around the flood-prone area into 16 equal sectors (each sector being 22.5° wide) yields 16 directional zones, with each zone corresponding to an angle α. d =d×22.5°, d=0,1,...,15.

[0065] S114. Statistically determine the maximum radial velocity of historical heavy precipitation cloud clusters in each region, identify the warning radius and vertex coordinates of each region, and construct the second time-limited response zone corresponding to the flooding point.

[0066] Specifically, this application's scheme obtains the warning radius for each direction by statistically analyzing the maximum radial velocity of all heavy precipitation cloud clusters in each azimuth region and extrapolating it to a preset target time (e.g., 2 hours). Then, it generates an irregular second-term response zone centered on the water accumulation point. This includes: S1141. Classify heavy precipitation cloud clusters based on each azimuth region, and statistically analyze the historical number and maximum radial velocity v of heavy precipitation cloud clusters in each azimuth region. max,d .

[0067] Specifically, this application classifies heavy precipitation clouds according to different azimuth regions, and counts the number of clouds and the maximum radial velocity v in each azimuth. max,d For azimuth regions with zero cloud clusters, the system fills the azimuth regions using a ring-shaped neighborhood: if both the left and right neighboring azimuth regions contain clouds, the average of the maximum radial velocities of the two neighboring regions is taken as the maximum radial velocity for that azimuth; if only one neighboring region contains clouds, that velocity is used or interpolated. After filling, a complete sequence of maximum radial velocities for 16 azimuth regions is formed to ensure the continuity of the outer spatial ring.

[0068] S1142. Based on the preset target time t (e.g., 2 hours), determine the warning radius R corresponding to each location area. d : .

[0069] S1143, Based on the warning radius R of each location area d Determine the vertex coordinates corresponding to each directional region.

[0070] Specifically, based on the warning radius in each direction, the latitude and longitude coordinates of the outer vertices of each direction area are calculated using the spherical coordinate formula.

[0071]

[0072]

[0073] S1144. Construct the second time-sensitive response zone corresponding to the flooding point based on the vertices corresponding to each directional zone.

[0074] Specifically, in one implementation, the vertices can be connected sequentially according to their orientation, and the first point can be appended to the end to form a closed polygon, thus obtaining an irregular second time-response zone surrounding the flooded point. For example... Figure 3 The diagram shown is a schematic diagram of the second time-limited response zone corresponding to a certain waterlogging point constructed based on the above scheme in one embodiment of this application.

[0075] S115. Generate the relationship between the false alarm rate and the extrapolated response time, and determine the extrapolated response time T based on the false alarm rate as the control objective. rad .

[0076] Preferably, the proposed solution uses the extrapolated response time of the cloud cluster as the response time parameter for the corresponding second-time response zone. The extrapolated response time (i.e., the effective radar forecast time) is the maximum lead time during which the radar extrapolated echo forecast product can maintain reliable forecast accuracy. The purpose of this extrapolated response time is to filter out false alarms caused by excessively large long-distance extrapolation errors, and to monitor whether the estimated time of the cloud cluster arriving at the flood-prone area is lower than the effective radar forecast time. If so, the early warning judgment process is triggered. Specifically, this includes the following steps: S1151. Collect data and generate extrapolated echo fields with various prediction periods.

[0077] Specifically, this application's approach uses the aforementioned historical short-duration heavy precipitation events as historical event samples, collecting real-time radar combined reflectivity factor data and 2-hour extrapolated forecast data for each time point within these historical events. The forecast data has a temporal resolution of 6 minutes and a spatial resolution of 1 km, covering the surrounding adjacent areas. Based on the extrapolated forecast data, a lead time (i.e., extrapolation lead time) is generated. τ Extrapolated echo field of ϵ(6,12,18,24,30,36,42,48,54,60,66,72,78,84,90,96,102,108,114,120).

[0078] S1152, Calculate evaluation indicators and thresholds.

[0079] Specifically, this application selects areas of heavy precipitation with a combined reflectance ≥35dBz as the radar extrapolation evaluation targets (this reflectance value corresponds to a rainfall intensity of approximately 20mm / h under typical ZR relationships, which is the key rainfall intensity for causing waterlogging). Furthermore, this application's scheme uses the Critical Success Index (CSI), Probability of Detection (POD), and False Alarm Ratio (FAR) as quantitative evaluation indicators.

[0080] Because radar extrapolation often involves small spatial location errors (e.g., 1-3 km), strict grid-to-grid matching may lead to misjudgments, resulting in excessively low CSI values ​​that fail to accurately reflect the effective coverage of the extrapolation. Therefore, to improve accuracy, this application employs a neighborhood method to introduce spatial neighborhood tolerance: fuzzy matching is performed within a circular neighborhood of a preset radius (e.g., 2-3 km). For a grid point i in the extrapolated field with a reflectivity ≥35 dBz, a circular area of ​​3 km in radius is used. Within the grid, if any actual grid point satisfies a reflectivity ≥ 35 dBz, then that extrapolated grid point is considered a hit grid point. Furthermore, the computational quantitative evaluation indicators for this application's scheme are as follows:

[0081]

[0082]

[0083] In the formula: Represents the extrapolated set of strong echo grid points; , representing the set of grid points for real-time strong echoes; , representing the set of grid points in the neighborhood where there are real-time strong echoes; G represents the set of grid points in the neighborhood that have extrapolated strong echoes; G is the total number of spatial grid points in the study area participating in the evaluation. y represents the extrapolated radar composite reflectivity at grid point i; i This represents the actual radar composite reflectivity at grid point i. For intersection; For set difference, It is an empty set.

[0084] S1153. The relationship between the generated CSI, POD, and FAR and the extrapolation time.

[0085] Specifically, for each forecast period τ, extrapolated forecast fields and actual scene mosaics are extracted from all historical samples to form a set of comparison sample pairs, with a total of N sample pairs. τThe three domain evaluation indicators for each sample pair are calculated. Then, for each forecast period τ, the values ​​of the three evaluation indicators for all samples are summed, and the median is calculated to obtain the result. , , The 25th percentile and 75th percentile can also be calculated simultaneously. Then, with the forecast period τ as the x-axis and each quantitative evaluation indicator as the y-axis, the change curves of the three types of evaluation indicators are plotted respectively, such as... Figure 4 The figure shown is a graph illustrating the changes in CSI, POD, and FAR as extrapolated over time, generated based on the above scheme in one embodiment of this application.

[0086] S1154. Determine the extrapolated response time with the false alarm rate as the control objective. And determine the response time T corresponding to the second time zone. rad .

[0087] Specifically, this application's scheme uses the aforementioned generated curves of CSI, POD, and FAR changing with extrapolated lead time, and takes controlling the false alarm rate of urban flooding warnings as the control objective (e.g., using...). To control the target, constraints can be applied using CSI and POD to obtain a definite extrapolated response time. .like Figure 4 The control target can be obtained from the curves showing the changes in CSI, POD, and FAR with extrapolated aging time. The corresponding extrapolation response time (extrapolation timeout) is within 54 minutes, and within this extrapolation timeout, , This indicates that the extrapolation has reached an acceptable level in terms of its ability to capture areas of heavy precipitation, its ability to control false alarms, and its overall forecast accuracy. Therefore, this value is determined as the reliable radar extrapolation response time (extrapolation lead time).

[0088] Furthermore, this application also introduces emergency preparedness time as a constraint, stipulating that the effective advance warning time of radar should at least meet the emergency preparedness time T corresponding to the flood-prone area. resp (Determined by the specific emergency preparedness time required for each flood-prone area), therefore, this application determines the response time T corresponding to the second time limit zone. rad satisfy: .

[0089] The above-mentioned scheme in this application generates curves of CSI, POD, and FAR as extrapolation lead time, and then obtains a reliable forecast period based on the false alarm rate target as the extrapolation response time. Furthermore, it uses the emergency preparedness time corresponding to the waterlogging point as a constraint to determine the lead time response time corresponding to the second lead time zone. This can effectively reduce the false alarm rate caused by long-distance extrapolation error, enable the early warning judgment process to dynamically adapt to the extrapolation characteristics of different weather types, and improve the accuracy of early warning.

[0090] S12. Extract the target cloud cluster of heavy precipitation based on the current radar mosaic, and monitor the position and motion vector of the target cloud cluster of heavy precipitation.

[0091] Preferably, the solution in this application is based on a two-dimensional radar combined reflectivity mosaic and uses a lightweight threshold rule to identify each independent cloud cluster in real time, without the need for complex three-dimensional scanning or machine learning models. Further, the method includes: S121. Receive the latest radar composite reflectivity mosaic at preset time intervals.

[0092] Specifically, the system in this application connects to the meteorological radar data source at a pre-set time interval (e.g., every 6 minutes) to obtain the radar composite reflectivity mosaic of the latest observation time, and completes the timed reception, time-series storage and preprocessing of real-time radar echo data, providing continuous time-series observational basic data for subsequent cloud cluster identification, tracking and short-term rainfall extrapolation forecast.

[0093] S122. Segment independent cloud clusters based on connected component labeling method, and extract target cloud clusters of heavy precipitation with combined reflectance greater than a preset threshold.

[0094] Specifically, this application determines whether the radar composite reflectivity is greater than an effective precipitation threshold (e.g., 15 dBZ). Data with reflectivity greater than or equal to this threshold is considered effective precipitation, and the effective reflectivity field data is filtered out. Then, cloud cluster grid points are marked using 8-neighbor or 4-neighbor connected domains, and spatially connected independent thunderstorm cloud clusters are extracted. Fragmented clouds with areas smaller than a preset area value are further filtered out, and finally, one or more independent heavy precipitation target cloud clusters are obtained.

[0095] S123. Monitor the location and motion vector of the current heavy precipitation target cloud cluster.

[0096] Specifically, for the extracted target cloud clusters of heavy precipitation, tracking and monitoring are carried out based on a preset time interval Δt (e.g., 6 minutes), and the latitude and longitude coordinates of the target cloud clusters at each monitoring time point are obtained. The radial velocity vector of the current target cloud cluster relative to the flood point is calculated using a method similar to step S1121.

[0097] S13. If the current cloud cluster enters or approaches the first time-sensitive response zone, the flood risk warning and prediction will be triggered.

[0098] Preferably, in one embodiment of this application, the target cloud cluster is identified and converted into a grid mask, and spatial intersection analysis is performed with the second and / or first time-response areas. If there is overlap, it is determined that the target cloud cluster has entered or is approaching the corresponding second and / or first time-response areas. At any time, if it is determined that the current target cloud cluster has entered or is approaching the first time-response area, the early warning prediction process is triggered immediately without any conditions, and a zero-delay response is performed.

[0099] S14. If the target cloud cluster enters or approaches the second time-response zone corresponding to the waterlogging point, the tracking and monitoring process is initiated, and the waterlogging risk warning and prediction are determined based on the time-response time and the movement vector of the target cloud cluster.

[0100] Preferably, in one embodiment of this application, if the grid mask corresponding to the target cloud cluster does not overlap with either the second or first time-response area, it is determined that the current heavy precipitation cloud cluster is outside the warning space, and the system does not perform any warning-related processing, only maintaining conventional radar tracking. If it overlaps with the second time-response area, it is determined that the target cloud cluster has entered or is approaching the corresponding second time-response area, and the system initiates a monitoring process, specifically including: S141. Increase tracking frequency to extract and / or monitor target cloud clusters.

[0101] Specifically, if the corresponding target cloud cluster enters the corresponding second-time response zone, the system initiates a monitoring process: increasing the tracking frequency (e.g., to every 2-4 minutes) to extract and / or monitor the target cloud cluster, i.e., monitoring and extracting heavy rainfall target clouds at a higher frequency. Furthermore, using preset algorithms, such as the pyramid Lucas-Kanade sparse optical flow method, the system updates, calculates, and monitors in real time the movement direction, speed, radial velocity relative to the water accumulation point, and azimuth of the heavy rainfall target cloud cluster.

[0102] S142. Real-time updates of the target cloud's movement direction, speed, and the time T it takes for the target cloud to reach the flooded area. arrival If the current T arrival ≤T rad If so, an early warning prediction will be triggered.

[0103] Specifically, in the aforementioned monitoring process, and based on the aforementioned updated calculation information, the time T for the target cloud cluster to arrive at the flood-prone area is updated in real time. arrival If the current T arrival ≤T rad If the cloud cluster is within the radar's reliable forecast window and there is sufficient time for emergency preparedness, the proposed solution will trigger the early warning and prediction process.

[0104] The proposed solution constructs a first-time-response zone and a second-time-response zone corresponding to the flood-prone area based on a time-response mechanism, forming a dual-time-response zone. Then, based on cloud cluster information and a pre-set early warning and prediction mechanism, it can specifically track and predict current cloud clusters and / or issue early warnings of flood risk. This solution organically integrates "spatial layered monitoring and time-based triggering" to construct a dynamic adaptive time-response zone and early warning and prediction mechanism for flood-prone areas. This ensures both relatively stable forecast accuracy of radar extrapolation information after thunderstorm clouds move into the time-response zone and guarantees sufficient emergency response time from the entry of thunderstorm clouds into the time-response zone to their arrival at the flood-prone area.

[0105] Furthermore, in one embodiment, the early warning judgment of the above-described solution of this application includes: S20. Based on the motion vector of the current target cloud cluster, generate a precipitation estimate for the catchment area of ​​the flood-prone point within a preset time period.

[0106] Specifically, in one embodiment, the present application first uses the surface runoff path corresponding to the first time-response zone as the lowest outflow point in the region, taking the waterlogging point as the lowest outflow point, and combines the water flow direction to obtain the catchment area corresponding to the waterlogging point. Then, based on the fine motion vector field of the echo cells of the cloud cluster, extrapolation is performed to obtain the predicted radar echo spatial field for each preset forecast period (e.g., 6, 12, 18, 24, 30... min). By establishing the combined reflectivity factor-hourly rainfall intensity (ZR) relationship, the extrapolated predicted radar echoes are inverted to obtain the quantitative precipitation forecast (QPF) corresponding to the preset time period in the future, thereby generating the estimated spatial distribution of precipitation in each area of ​​the catchment area within the preset time period in the future.

[0107] Preferably, when generating a rainfall estimate for the catchment area of ​​the flood-prone point within a preset time period, the present application scheme can also dynamically correct the historical radar QPE relationship in real time based on the rainfall observation data of minute-level densified automatic meteorological stations around the flood-prone point, thereby generating a high-resolution fused rainfall field that is closer to the actual ground conditions.

[0108] S21. Based on the flow relationship between each catchment area and the flood point, the total flow rate and water depth entering the flood point are calculated.

[0109] Specifically, this application's scheme establishes the runoff relationship between each catchment area and the corresponding flood-prone point based on the topographic and hydrological analysis results and / or historical statistical data of the flood-prone points. Based on the aforementioned quantitative precipitation estimation, surface precipitation is calculated for each catchment area, and then the runoff from all catchment areas flowing into the flood-prone point is superimposed to obtain the total runoff volume flowing into the flood-prone point. Furthermore, a statistical relationship model of "runoff volume - water depth" is established using historical waterlogging records (including rainfall events and corresponding water depths), or a locally pre-set hydrodynamic model, to obtain the corresponding water depth.

[0110] S22. Generate the meteorological risk level of waterlogging points based on the water depth, and make early warnings and predictions.

[0111] Specifically, based on the calculated water depth and referring to the water resources department's classification standards for urban flooding disasters, the predicted water depth is converted into an urban flooding meteorological risk level, such as: 0-5cm is no risk, 5-15cm is a level four risk, etc.

[0112] Furthermore, to balance the timeliness and stability of forecasts and avoid false alarms caused by fluctuations in single extrapolation, this application's scheme uses the following methods for early warning and dynamic dissemination: Initial release: When the forecast shows that the risk level reaches the forecast level for three consecutive time intervals (i.e., three consecutive 6-minute intervals), the warning information is officially released. Warning escalation: When the risk level in two consecutive time-sharing forecasts shows an upward trend, warning escalation information will be issued; Warning lifted: When five consecutive time-sharing forecasts show no risk, the warning will be automatically lifted.

[0113] A second aspect of this application provides an urban flooding meteorological risk early warning system based on spatiotemporal response, the system comprising: The first response zone construction unit is used to construct the first time-effect response zone of the waterlogging point based on the surface hydrological characteristics of the waterlogging point. The second response zone construction unit is used to construct the second time-sensitive response zone corresponding to the flooding point based on historical cloud cluster data, and to determine the corresponding response time parameters. The cloud extraction unit is used to extract heavy precipitation clouds based on the current radar mosaic and to monitor the position and motion vector of the current heavy precipitation target cloud. The early warning and prediction unit is used to trigger the activation of urban flood risk early warning and prediction if the target cloud enters or approaches the first time-response zone; if the target cloud enters or approaches the second time-response zone corresponding to the flood point, it tracks and monitors the target cloud and determines whether to issue an urban flood risk early warning and prediction based on the time-response time and the movement vector of the target cloud.

[0114] In one embodiment, such as Figure 5 As shown, this application also provides a computer device, which includes a processor, a memory, and a network interface connected via a system bus. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs. When the computer programs are executed by the processor, the processor performs the steps of the above-described method.

[0115] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0116] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0117] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A dynamic early warning method for urban flooding risk based on spatiotemporal response, characterized in that, The method includes: Construct the first time-sensitive response zone for the waterlogging point based on the surface hydrological characteristics of the waterlogging point; Based on historical cloud cluster data, a second time-sensitive response zone corresponding to the flood-prone area is constructed, and the corresponding time-sensitive response time is determined; Extract the target cloud clusters of heavy precipitation based on the current radar mosaic, and monitor the position and motion vector of the current target cloud clusters of heavy precipitation; If the target cloud cluster enters or approaches the first-response zone, the flood risk warning and prediction will be triggered. If the target cloud cluster enters or approaches the second time-response zone corresponding to the flooding point, the tracking and monitoring process will be initiated, and a flood risk warning and prediction will be made based on the time-response time and the movement vector of the target cloud cluster.

2. The method according to claim 1, characterized in that, Establish the first-response zone for areas prone to flooding, including: Collect high-resolution DEM data of the area surrounding the flood-prone area; Flow direction analysis and calculation of waterlogging points based on unidirectional flow direction algorithm; Extract surface runoff paths, classify runoff paths into different levels, and determine the catchment areas of waterlogging points; Based on DEM topographic data, the upstream surface runoff path is determined, and the spatial range and boundary of the surface water flow that eventually flows into the flood-prone area are extracted to obtain the first time-response zone of the flood-prone area.

3. The method according to claim 1, characterized in that, Construct a second time-sensitive response zone corresponding to the flood-prone area, including: Collect historical heavy rainfall events at waterlogged areas and track the movement trajectory of historical heavy rainfall echoes that have passed through these areas; Based on the movement trajectory of historical heavy precipitation echoes, calculate the radial velocity and azimuth of the corresponding historical heavy precipitation cloud cluster relative to the waterlogging point; Centered on the flood-prone area, the surrounding area is divided into multiple directional zones. The maximum radial velocity of historical heavy precipitation cloud clusters in various regions was statistically analyzed to determine the warning radius and vertex coordinates of each region, and a second time-limited response zone corresponding to the flooding point was constructed. The relationship between false alarm rate and extrapolation lead time is generated. The false alarm rate is used as the control target to determine the reliable extrapolation response time, and the lead time response time T corresponding to the second lead time zone is determined. rad .

4. The method according to claim 3, characterized in that, The method further includes: Calculate the relative distance from each tracking point to the flooded point, and obtain the radial velocity of each tracking point relative to the flooded point based on the time interval between two adjacent tracking points; Determine the azimuth angle of each tracking point relative to the flooded point based on the coordinates of each tracking point; Select preset effective tracking points, and calculate the radial velocity and azimuth of historical target heavy precipitation cloud clusters relative to the waterlogging points based on the effective tracking points.

5. The method according to claim 3, characterized in that, The method further includes: Based on the classification of heavy precipitation cloud clusters by azimuth region, the historical number and maximum radial velocity v of heavy precipitation cloud clusters in each azimuth region are statistically analyzed. max,d ; Based on the preset target time t, determine the warning radius R corresponding to each location area. d : ， Determine the vertex coordinates corresponding to each location area based on the warning radius of each location area; The second time-sensitive response zone corresponding to the flooding point is constructed based on the vertices corresponding to each directional zone.

6. The method according to claim 3, characterized in that, The method further includes: Receive the latest radar composite reflectivity mosaic at preset time intervals; Independent cloud clusters were segmented based on the connected component labeling method, and target cloud clusters with heavy precipitation whose combined reflectance was greater than a preset threshold were extracted. The target cloud cluster is identified and converted into a grid mask, and spatial intersection analysis is performed with the second time-response area and / or the first time-response area; If there is overlap, it is determined that the target cloud cluster has entered or is approaching the corresponding second-time response zone and / or first-time response zone.

7. The method according to claim 6, characterized in that, The tracking and monitoring process includes: Increase tracking frequency to extract and / or monitor target cloud clusters; Real-time updates calculate the target cloud's direction of movement, speed of movement, and the time T it takes for the target cloud to reach the flooded area. arrival If the current T arrival ≤T rad If so, an early warning prediction will be triggered.

8. The method according to any one of claims 1-7, characterized in that, The early warning judgment includes: Based on the motion vector of the target cloud cluster, the precipitation estimate of each catchment area of ​​the waterlogging point is generated within a preset time in the future. Based on the flow relationship between each catchment area and the flood-prone point, calculate the total flow rate and water depth of the flood-prone point; The meteorological risk level of waterlogging at the waterlogging point is generated based on the water depth, and an early warning is issued.

9. A dynamic early warning system for urban flooding risk based on spatiotemporal response, characterized in that, The system includes: The first response zone construction unit is used to construct the first time-effect response zone of the waterlogging point based on the surface hydrological characteristics of the waterlogging point. The second response zone construction unit is used to construct the second time-sensitive response zone corresponding to the flooding point based on historical cloud cluster data, and to determine the corresponding response time parameters. The cloud extraction unit is used to extract heavy precipitation clouds based on the current radar mosaic and to monitor the position and motion vector of the current heavy precipitation target cloud. The early warning and prediction unit is used to trigger the activation of urban flood risk early warning and prediction if the target cloud enters or approaches the first time-response zone; if the target cloud enters or approaches the second time-response zone corresponding to the flood point, it tracks and monitors the target cloud and determines whether to issue an urban flood risk early warning and prediction based on the time-response time and the movement vector of the target cloud.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the computer program is executed by the processor, it causes the processor to perform the steps of the method as described in any one of claims 1 to 8.

Images