Digital twin emergency assistance method and system for estuary and coastal disaster risk

Abstract

The present application relates to the field of wisdom ocean technology and public safety emergency management, and provides a digital twin emergency assistance method and system for estuary and coastal disaster risk, which comprises: obtaining terrain elevation and time-varying tidal level data to calculate relative water depth; based on the ergonomics escape constraint parameter, the evacuation path connectivity is evaluated to generate a dynamic risk matrix to represent the hidden risk area; in response to the position selection instruction, a Lagrange particle initial matrix file is generated and input into a back-end hydrodynamic model to solve and output trajectory data; through a standardized data interface, the data is transmitted to a three-dimensional visualization scene, the coordinates are converted, and a dynamic flooding risk map and a drift trajectory are rendered. The present application realizes the accurate quantitative identification of hidden risk areas and the rapid deduction and visualization of search and rescue trajectories through a decoupled architecture of calculation and rendering, effectively overcoming the defects that the traditional two-dimensional flooding analysis cannot represent the three-dimensional space dynamic evolution logic and the existing system has a long search and rescue response period.

Description

Technical Field

[0001] This invention relates to the fields of smart ocean technology and public safety emergency management, and in particular to a digital twin emergency assistance method and system for estuarine and coastal disaster risks. Background Technology

[0002] River estuaries and beaches with large tidal ranges typically feature complex topography, large tidal ranges, and numerous tidal channels. During high tide, the rapidly rising tide can easily cut off people's escape routes through hidden tidal channels, creating a "cut-off effect" and putting beachgoers at risk of being trapped or even drowning.

[0003] Currently, the management and assessment of this type of coastal drowning risk mainly relies on static physical warnings and traditional hydrodynamic numerical simulations. Traditional numerical simulation methods typically use hydrodynamic models to calculate the flow field and water level, and display the inundation results through two-dimensional color block maps or contour maps. However, existing methods have the following drawbacks: First, existing two-dimensional maps and static warnings are insufficient to accurately represent the dynamically evolving flooding risk over time. Traditional flooding analysis focuses on outputting the physical water depth at a specific moment, without considering personnel mobility constraints, and cannot dynamically assess the connectivity of evacuation routes in three-dimensional space. Due to the complex spatial topological relationship between tidal channels and sand ridges, the tides exhibit a filling sequence of lower water levels followed by higher water levels during high tide. Existing two-dimensional maps cannot intuitively display this dynamic evolution logic in three-dimensional space, resulting in the inability to effectively identify hidden risk areas where the water depth is currently shallow but evacuation routes are about to be blocked.

[0004] Secondly, existing hydrodynamic simulation and visualization systems struggle to balance computational accuracy and interactive efficiency. While professional hydrodynamic models offer high physical calculation accuracy, their visualization is poor and computation is time-consuming. Conventional 3D visualization systems lack physical mechanism support and can only provide simple animation demonstrations. Furthermore, traditional search and rescue trajectory prediction processes are largely sequential, involving significant human interaction and data transfer at each stage, from receiving the alarm, manually preparing model input parameters, running the model, to manual post-processing, drawing, and result feedback. This tightly coupled computation and rendering system architecture, heavily reliant on human intervention, results in excessively long search and rescue response cycles, failing to meet the need for rapid generation of water trajectories and search and rescue decision support in the event of a sudden disaster. Summary of the Invention

[0005] Therefore, it is necessary to provide a digital twin emergency assistance method and system for estuarine and coastal disaster risks, addressing the problems of poor intuitiveness in coastal disaster risk perception and low efficiency in search and rescue decision-making and response in existing solutions, so as to achieve accurate quantitative identification of hidden risks and rapid simulation and visualization of search and rescue trajectories.

[0006] This invention provides a digital twin emergency response method for estuarine and coastal disaster risks, comprising: Obtain topographic elevation data and time-varying tidal level data for the target beach; Based on the terrain elevation data and time-varying tide data, the relative water depth of the discrete spatiotemporal grid within a preset time step is calculated. Ergonomic escape constraint parameters are obtained. Based on the relative water depth and ergonomic escape constraint parameters, the connectivity of the safe evacuation path is dynamically evaluated to generate a dynamic risk matrix. The ergonomic escape constraint parameters include a lethal water level threshold and differentiated mobility parameters. The dynamic risk matrix is ​​used to characterize hidden risk areas. The hidden risk areas satisfy the following conditions: the relative water depth at the current moment is less than the lethal water level threshold, and within the evacuation time window calculated based on the differentiated mobility parameters, the relative water depth on the necessary evacuation path is greater than or equal to the lethal water level threshold. In response to a location selection command for a 3D visualization scene, the geographic coordinates of the selected location are obtained, and an initial Lagrange particle matrix file is generated based on the geographic coordinates. The initial matrix file of the Lagrange particles is input into the backend hydrodynamic model for solution, and the output is a trajectory data file containing the time series coordinates of multiple Lagrange particles; the backend hydrodynamic model runs independently of the three-dimensional visualization scene, and the two interact with each other through a standardized data interface; The dynamic risk matrix and the trajectory data file are transmitted to the 3D visualization scene through the standardized data interface; the geographic coordinates in the dynamic risk matrix are converted into local spatial coordinates of the 3D visualization scene and color-mapped rendering is performed to generate a dynamic flooding risk map; the trajectory data file is parsed, and the particle system is called to render a dynamic 3D drift trajectory based on the time series coordinates of the Lagrange particles.

[0007] In one embodiment, calculating the relative water depth within a preset time step using the topographic elevation data and the time-varying tidal data includes: The terrain elevation data is divided into multiple two-dimensional spatial grids, and the ground elevation value of each two-dimensional spatial grid is obtained; Based on the time-varying tide data, the instantaneous tide value corresponding to each preset time step is calculated by interpolation; The difference between the instantaneous tide level and the corresponding ground elevation value is used to obtain the relative water depth of each two-dimensional spatial grid within the preset time step.

[0008] In one embodiment, the dynamic assessment of the connectivity of the safe evacuation path based on the relative water depth and the ergonomic escape constraint parameters, generating a dynamic risk matrix, includes: Based on the differentiated mobility parameters, the evacuation speed for different target groups is determined; Based on the evacuation speed and the relative water depth of the discrete spatiotemporal grid, the passage impedance of each of the two-dimensional spatial grids is calculated; A time-varying topology graph is constructed using the two-dimensional spatial grid as nodes and the passage impedance between adjacent grids as edge weights. Based on the time-varying topology map, the shortest path algorithm is used to calculate the optimal evacuation path and corresponding evacuation time from each of the two-dimensional spatial grids to the preset safe area.

[0009] In one embodiment, generating the dynamic risk matrix further includes the step of identifying the hidden risk areas, wherein identifying the hidden risk areas includes: For each node on the optimal evacuation path, extract its relative water depth prediction value in the future time series; Determine whether there is a target time within the future time series such that the predicted relative water depth at the target time is greater than or equal to the lethal water level threshold. If it exists, calculate the time difference between the current time and the target time, as the available evacuation time; If the available evacuation time is less than the evacuation time corresponding to the optimal evacuation path, then the two-dimensional spatial grid where the current node is located is marked as the hidden risk area.

[0010] In one embodiment, the step of obtaining the geographic coordinates of the selected location in response to a location selection command for a 3D visualization scene, and generating an initial Lagrange particle matrix file based on the geographic coordinates, includes: Parse the location selection command and extract the latitude and longitude coordinates and release timestamp of the selected location; Based on the preset particle diffusion radius and particle number, multiple initial particle positions are generated within the neighborhood of the latitude and longitude coordinates. The initial particle position, the release timestamp, and the preset particle physical properties are encapsulated into structured matrix data to generate the initial Lagrange particle matrix file.

[0011] In one embodiment, the step of inputting the initial Lagrange particle matrix file into the backend hydrodynamic model for solution, and outputting a trajectory data file containing multiple Lagrange particle time series coordinates, includes: The back-end hydrodynamic model solves the flow field distribution based on the two-dimensional shallow water equation to obtain the velocity field and water level field at each preset time step. For each particle in the initial Lagrange particle matrix file, bilinear interpolation is performed in the velocity field according to its position to obtain the instantaneous velocity. The Runge-Kutta method is used to integrate the instantaneous flow velocity over time to update the spatial coordinates of the particles; The spatial coordinates of the particles are recorded at preset output time intervals to generate the trajectory data file.

[0012] In one embodiment, the step of converting the geographic coordinates in the dynamic risk matrix into local spatial coordinates of the 3D visualization scene and performing color mapping rendering to generate a dynamic flooding risk map includes: Using preset coordinate transformation parameters, the geographic coordinates in the dynamic risk matrix are converted into local 3D coordinates of the 3D visualization scene; Based on the risk level or relative water depth value of each discrete spatiotemporal grid in the dynamic risk matrix, a preset color mapping table is queried to determine the corresponding color attribute value. The local 3D coordinates and color attribute values ​​are used to construct vertex data, and the dynamic flooding risk map is generated by rendering primitives in the 3D visualization scene.

[0013] In one embodiment, parsing the trajectory data file and calling the particle system to render a dynamic three-dimensional drift trajectory based on the time-series coordinates of the Lagrange particles includes: The trajectory data file is parsed to extract the three-dimensional spatial coordinates of each Lagrange particle at different timestamps; Initialize the particle emitter in the three-dimensional visualization scene, using the initial coordinates of the Lagrange particles as the emission position; In each rendering frame, the position attributes of each particle in the particle emitter are updated by matching the timestamp with the current system time; The dynamic 3D drift trajectory is generated by connecting the positions of the same particle in consecutive rendering frames using a trailing rendering component.

[0014] The present invention also provides a digital twin emergency support system for estuarine and coastal disaster risks, including a back-end computing server and a front-end interactive terminal, wherein the back-end computing server and the front-end interactive terminal are connected to communicate through a standardized data interface; The front-end interactive terminal is used to display a three-dimensional visualization scene, obtain the geographic coordinates of the selected location in response to the location selection command of the three-dimensional visualization scene, generate an initial Lagrange particle matrix file based on the geographic coordinates, and send the initial Lagrange particle matrix file to the back-end computing server. The backend computing server is used to acquire topographic elevation data and time-varying tidal level data of the target beach; calculate the relative water depth of the discrete spatiotemporal grid within a preset time step based on the topographic elevation data and the time-varying tidal level data; acquire ergonomic escape constraint parameters, including a lethal water level threshold and differentiated mobility parameters; dynamically evaluate the connectivity of the safe evacuation path based on the relative water depth and the ergonomic escape constraint parameters, and generate a dynamic risk matrix; wherein, the dynamic risk matrix is ​​used to characterize hidden risk areas, and the hidden risk areas satisfy: the relative water depth at the current moment is less than the lethal water level threshold, and within the evacuation time window calculated based on the differentiated mobility parameters, the relative water depth on the necessary evacuation path is greater than or equal to the lethal water level threshold; input the initial Lagrange particle matrix file into the backend hydrodynamic model for solution, and output a trajectory data file containing multiple Lagrange particle time series coordinates; send the dynamic risk matrix and the trajectory data file to the frontend interactive terminal through the standardized data interface; The front-end interactive terminal is also used to convert the geographic coordinates in the dynamic risk matrix into local spatial coordinates of the three-dimensional visualization scene and perform color mapping rendering to generate a dynamic flooding risk map; parse the trajectory data file and call the particle system to render a dynamic three-dimensional drift trajectory based on the time series coordinates of the Lagrange particles.

[0015] In one embodiment, the standardized data interface uses the WebSocket protocol or the gRPC protocol; During the calculation process, the backend computing server periodically sends progress status information to the frontend interactive terminal through the standardized data interface; After receiving the dynamic risk matrix and the trajectory data file, the front-end interactive terminal returns a confirmation of receipt to the back-end computing server through the standardized data interface.

[0016] The aforementioned digital twin emergency response method and system for estuarine and coastal disaster risks calculates the relative water depth in a discrete spatiotemporal grid by acquiring topographic elevation data and time-varying tidal data of the target beach. It then dynamically assesses the connectivity of safe evacuation paths by combining ergonomic escape constraint parameters, including lethal water level thresholds and differentiated mobility parameters. This quantifies the dynamic filling sequence of tides in complex terrain, thereby identifying hidden risk areas where the relative water depth is currently below the lethal water level threshold but the evacuation route is about to be blocked within the evacuation time window. This overcomes the limitation of traditional two-dimensional inundation analysis, which cannot characterize the dynamic evolution logic in three-dimensional space. Simultaneously, it automatically generates lattices in response to location selection commands for a three-dimensional visualization scene. The initial matrix file of Langge particles is generated, and the trajectory data file output by the independently running backend hydrodynamic model and the dynamic risk matrix are transmitted to the 3D visualization scene through a standardized data interface. This decouples physical calculation and front-end rendering and automates the search and rescue simulation process, reducing manual intervention and data transfer links in the traditional serial working mode. While ensuring the accuracy of hydrodynamic model calculation, the 3D visualization scene is used for coordinate transformation, color mapping and particle system rendering to generate dynamic inundation risk maps and dynamic 3D drift trajectories. This overcomes the defect of the existing system's tight coupling of calculation and rendering, which leads to excessively long search and rescue response cycles, and improves the intuitiveness of coastal disaster risk perception and the response efficiency of search and rescue decisions. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention 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 some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a digital twin emergency response method for estuarine and coastal disaster risks according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the system architecture and data flow according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the data flow architecture of the digital twin system according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the end-to-end automated search and rescue decision-making workflow according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the Niagara-driven search and rescue trajectory visualization process according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the real-time tide level and water body synchronization process based on the MQTT protocol in an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Unless otherwise defined, all technical and scientific terms used in this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used in this specification includes any and all combinations of one or more of the associated listed items.

[0021] The following is combined with Figures 1-6 This invention describes a digital twin emergency response method and system for estuarine and coastal disaster risks.

[0022] like Figure 1 , Figure 2 and Figure 3 As shown, in one embodiment, a digital twin emergency response method for estuarine and coastal disaster risks adopts an architecture that decouples computation and visualization, separating high-precision physical numerical simulation from high-fidelity real-time 3D rendering, and achieving collaborative operation through a standardized data interface. Specifically, it includes the following steps: Step S110: Obtain the topographic elevation data and time-varying tidal level data of the target beach.

[0023] Topographic elevation data typically originates from digital elevation models (DEMs) generated by UAV RTK surveys or lidar scanning, accurately reflecting micro-topographic features such as tidal channels and sand ridges; time-varying tidal level data can be obtained through hydrological station measurements or tidal forecasting models. These two types of data constitute the spatiotemporal benchmark for all subsequent risk assessments and trajectory extrapolations.

[0024] Step S120: Based on topographic elevation data and time-varying tidal data, calculate the relative water depth of the discrete spatiotemporal grid within a preset time step.

[0025] This step involves the digital reconstruction of the geometric inundation process. By coupling dynamically changing tidal boundary conditions with static, high-precision topographic data, the system can quantify the water depth distribution at any given moment on a discrete spatiotemporal grid. This calculation method not only reflects the instantaneous inundation state but also captures the tidal filling sequence of complex terrain, from low to high, providing a mathematical basis for identifying potential cutoff effects.

[0026] Step S130: Obtain ergonomic escape constraint parameters. Based on the relative water depth and ergonomic escape constraint parameters, dynamically evaluate the connectivity of the safe evacuation path and generate a dynamic risk matrix. The ergonomic escape constraint parameters include the lethal water level threshold and differentiated mobility parameters. The dynamic risk matrix is ​​used to characterize hidden risk areas, which satisfy the following conditions: the relative water depth at the current moment is less than the lethal water level threshold, and within the evacuation time window calculated based on the differentiated mobility parameters, the relative water depth on the necessary evacuation path is greater than or equal to the lethal water level threshold.

[0027] Unlike traditional methods that rely solely on static water depth to assess risk, this step introduces a human-centered safety assessment dimension. The lethal water level threshold defines the critical point for personnel walking stability, while differentiated mobility parameters distinguish between mobility in dry and flooded environments. By incorporating these constraints into path connectivity analysis, the system can identify areas that appear safe initially but whose escape routes will be blocked by rising tides within the required evacuation window. This mechanism for identifying hidden risk areas effectively overcomes the limitation of traditional two-dimensional inundation analysis in representing the dynamic evolution logic within three-dimensional space. It enables non-experts to perceive the true risk of hidden risk areas, transforming hydrological data into early warning information that aligns with human behavioral characteristics.

[0028] In step S140, in response to the location selection instruction for the 3D visualization scene, the geographic coordinates of the selected location are obtained, and an initial Lagrange particle matrix file is generated based on the geographic coordinates.

[0029] In practice, when a user clicks on any location in the 3D visualization scene, the system automatically parses the latitude and longitude coordinates and release timestamp of that point, and automatically generates a structured initial particle matrix in the neighborhood based on the preset diffusion radius and particle number. This process is executed automatically by the program, replacing the manual editing of configuration files and preparation of model inputs in the traditional search and rescue process, reducing operational complexity and shortening response lead time.

[0030] In step S150, the initial matrix file of the Lagrange particles is input into the backend hydrodynamic model for solution, and the output is a trajectory data file containing the time series coordinates of multiple Lagrange particles; the backend hydrodynamic model runs independently of the three-dimensional visualization scene, and the two interact with each other through a standardized data interface.

[0031] The backend hydrodynamic model focuses on solving the fluid motion equations, ensuring the physical accuracy of trajectory derivation; while the 3D visualization scene focuses on graphics rendering and human-computer interaction. Both operate through standardized structured data files (such as .csv or .dat formats) or communication protocols, providing unidirectional data flow. This decoupled architecture allows for parallel processing of physical calculations and visual representation, avoiding the time wasted in traditional serial workflows that require waiting for model computations to complete before visualization preparation can begin. Compared to the traditional method's response time of approximately 95 minutes, this embodiment, through an automated parallel workflow, can reduce the response time to approximately 35 minutes, significantly improving response efficiency.

[0032] Step S160: The dynamic risk matrix and trajectory data file are transmitted to the 3D visualization scene through a standardized data interface; the geographic coordinates in the dynamic risk matrix are converted into local spatial coordinates of the 3D visualization scene and color-mapped rendering is performed to generate a dynamic flooding risk map; the trajectory data file is parsed and the particle system is called to render the dynamic 3D drift trajectory based on the time series coordinates of the Lagrange particles.

[0033] Front-end rendering is not simply a visual animation demonstration, but rather a deterministic transformation controlled by back-end physical data. The conversion from geographic coordinates to local spatial coordinates aligns the virtual scene with real-world geographical locations down to the centimeter level; color mapping overlays risk levels onto the 3D terrain; and particle systems reconstruct dynamic flow lines with a time dimension based on time-series coordinates. This data-driven rendering approach maintains the rigor of scientific computation while leveraging the high-fidelity expressiveness of game engines to provide intuitive and reliable situational awareness support for emergency decision-making.

[0034] In one embodiment, based on the above embodiments, the specific algorithm implementation process for generating the dynamic risk matrix and identifying hidden risks is further detailed. This process converts hydrological data into safety assessment indicators that conform to human behavioral characteristics, distinguishing this invention from traditional technical solutions that rely solely on water depth for risk warning.

[0035] Specifically, based on topographic elevation data and time-varying tidal data, the relative water depth of a discrete spatiotemporal grid within a preset time step is calculated. This includes: dividing the topographic elevation data into multiple two-dimensional spatial grids and obtaining the ground elevation value for each two-dimensional spatial grid; interpolating and calculating the instantaneous tidal value corresponding to each preset time step based on the time-varying tidal data; and subtracting the instantaneous tidal value from the corresponding ground elevation value to obtain the relative water depth of each two-dimensional spatial grid within the preset time step. (Refer to...) Figure 2 In the backend computing layer, module A's processing flow involves the system first loading a high-precision digital elevation model (DEM) and discretizing it into regular or irregular two-dimensional spatial grid cells. For each grid cell... Read its static ground elevation value Meanwhile, for time-varying tidal level data from discontinuous observations, linear interpolation or spline interpolation algorithms are used to calculate the value relative to the model's time step. Strictly synchronized instantaneous tide values Subsequently, through the formula The relative water depth is calculated grid by grid. It should be understood that although this embodiment uses a two-dimensional planar grid as an example, in other implementations, a three-dimensional voxel grid or an unstructured triangular grid can also be used, as long as it can characterize the coupling relationship between topography and water level over time. This step, while outputting the instantaneous inundation depth, constructs a four-dimensional water depth dataset containing the time dimension, capturing the "low-then-high" filling sequence of tides in complex micro-topography, providing a mathematical basis for subsequent identification of the retreat cutoff effect.

[0036] Furthermore, based on relative water depth and ergonomic escape constraint parameters, the connectivity of safe evacuation routes is dynamically assessed, generating a dynamic risk matrix. This includes: determining the evacuation speed of different target groups based on differentiated mobility parameters; calculating the passage impedance of each two-dimensional spatial grid based on the evacuation speed and the relative water depth of the discrete spatiotemporal grid; constructing a time-varying topology graph using the two-dimensional spatial grid as nodes and the passage impedance between adjacent grids as edge weights; and using the time-varying topology graph, employing a shortest path algorithm to calculate the optimal evacuation path from each two-dimensional spatial grid to the preset safe area and the corresponding evacuation time. In this process, the system introduces the concept of a Score of Hazard (SOH) model, transforming the physical field into a behavioral field. The differentiated mobility parameters are not fixed constants but rather a set of configurable variables based on the application scenario. For example, in a science education scenario targeting ordinary tourists, the walking speed on dry land is set to 1.0 m / s, while the walking speed in wading water is reduced to 0.8 m / s. In a search and rescue scenario targeting fully armed rescue personnel, these parameters can be adjusted accordingly to 1.2 m / s and 0.6 m / s. The calculation of passage impedance comprehensively considers the nonlinear attenuation effect of water depth on movement speed; for example, when the relative water depth exceeds ankle height, the impedance coefficient increases exponentially. The time-varying topology map abstracts the continuous geographic space into a discrete graph theory model, where nodes represent grid cells and edge weights are dynamically updated with the time step. Using shortest path algorithms such as Dijkstra's or A*, the system can calculate in real time the minimum time cost required to reach the nearest safe high ground from any current location. This processing method maps hydrodynamic parameters to graph theory weights, enabling the computer to quantify the actual degree of obstruction to escape by terrain and water flow, thereby generating a risk matrix with behavioral guidance.

[0037] Furthermore, generating the dynamic risk matrix also includes the step of identifying hidden risk areas. Identifying hidden risk areas includes: for each node on the optimal evacuation path, extracting its relative water depth prediction value in the future time series; determining whether there exists a target time in the future time series such that the relative water depth prediction value at the target time is greater than or equal to the lethal water level threshold; if so, calculating the time difference between the current time and the target time as the available evacuation time; if the available evacuation time is less than the evacuation time corresponding to the optimal evacuation path, then marking the two-dimensional spatial grid where the current node is located as a hidden risk area. Lethal water level threshold. This is a key ergonomic safety boundary parameter, preferably set to 0.4m (approximately the height of an adult's knee) in this embodiment. Studies have shown that flowing water exceeding this depth significantly impairs human walking stability, leading to a dramatically increased risk of falls or being swept away. It should be understood that... It can also be adjusted according to the characteristics of the target audience; for example, it can be set to 0.2m for children and 0.6m for vehicles. The core of the algorithm lies in "time window comparison": even at the current moment... Relative water depth of a certain grid (i.e., currently appears safe), but if it is a node on its optimal retreat path, predicting future moments... The water depth will exceed ,and If the time required for personnel to evacuate from their current location to a safe zone is less than the theoretical time, the grid is identified as a hidden risk area. This means that even if there is no water or the water is shallow at present, the escape route will be submerged by the rising tide before personnel arrive. Through this forward-looking spatiotemporal coupling analysis, the system can accurately identify areas that appear safe on traditional static maps but are actually traps, and present them intuitively in a 3D scene in the form of red heat maps, effectively overcoming the shortcomings of traditional two-dimensional inundation analysis in representing the dynamic evolution logic in three-dimensional space. This reduces the difficulty for non-professionals to understand the risks of complex tidal flat environments and provides a scientific and reliable risk warning basis for emergency decision-making and public education.

[0038] In one embodiment, based on the foregoing embodiments, the specific implementation mechanism of Lagrange particle tracking and backend hydrodynamic calculation is further detailed. This mechanism forms the physical computational basis for search and rescue decision support functions, ensuring the scientific validity and realism of drift trajectory extrapolation.

[0039] Specifically, in response to a location selection command for a 3D visualization scene, the system obtains the geographic coordinates of the selected location and generates an initial Lagrange particle matrix file based on these coordinates. This includes: parsing the location selection command and extracting the latitude and longitude coordinates and release timestamp of the selected location; generating multiple initial particle positions within the neighborhood of the latitude and longitude coordinates based on preset particle diffusion radius and particle quantity; and encapsulating the initial particle positions, release timestamps, and preset particle physical properties into structured matrix data to generate the initial Lagrange particle matrix file. (See reference...) Figure 4 The end-to-end automated search and rescue decision-making workflow involves the system's underlying interactive module instantly capturing the screen pixel coordinates corresponding to the user's selection of a suspected water-fall point in a 3D visualization scene via mouse click or touch. This data, combined with the current camera pose parameters and geographic reference system, is used to calculate the precise WGS84 latitude and longitude coordinates and the current simulated timestamp. Considering that actual water-fall point locations often have observational errors and that real water-falling objects have a certain spatial distribution range above the water surface, this embodiment does not directly treat a single click point as the sole calculation object. Instead, it uses the click point as the center and, based on a preset particle diffusion radius (e.g., adjustable from 5 to 20 meters) and particle count (e.g., 6400), automatically generates an N×N Lagrange particle matrix in the neighborhood using a Gaussian or uniform grid distribution algorithm. This approach transforms the uncertainty of a single point into a simulation of a probability density field, enabling subsequent extrapolation results to cover the potential search area. The generated particle matrix data is serialized into a structured file (e.g., .dat format) conforming to the input specifications of the backend hydrodynamic model. This file contains not only the initial three-dimensional coordinates of each particle but also physical property parameters such as release timestamp, particle density, and drag coefficient. This automated generation logic replaces the traditional process of manually editing model input files, directly converting interactive operations into boundary conditions for physical calculations, thus shortening the preparation time for emergency response.

[0040] Furthermore, the initial Lagrange particle matrix file is input into the backend hydrodynamic model for solution, outputting a trajectory data file containing the time-series coordinates of multiple Lagrange particles. This includes: the backend hydrodynamic model solving for the flow field distribution based on two-dimensional shallow water equations to obtain the velocity field and water level field at each preset time step; for each particle in the initial Lagrange particle matrix file, bilinear interpolation is performed in the velocity field according to its position to obtain the instantaneous velocity; the Runge-Kutta method is used to integrate the instantaneous velocity over time to update the spatial coordinates of the particles; and the spatial coordinates of the particles are recorded at preset output time intervals to generate the trajectory data file. The backend hydrodynamic model runs independently of the frontend rendering engine, and its main function is to provide a flow field environment with physically conserved properties. In this embodiment, the back-end hydrodynamic model preferably adopts the Ecom-si (Estuarine and Coastal Ocean Model - Semi-Implicit) model. This model is based on the Reynolds-averaged Navier-Stokes equations or the two-dimensional shallow water equations, and comprehensively considers factors such as tidal drive, wind stress, bottom friction, and Coriolis force to solve for the depth-averaged flow velocity at each grid node in the computational domain at different time steps. and water level It should be understood that although this embodiment uses the Ecom-si model as an example, other mature hydrodynamic solvers such as FVCOM, Delft3D, or MIKE can also be used in other implementations, as long as they can provide spatiotemporally continuous flow field data.

[0041] After obtaining the background flow field, the system independently tracks the motion of each Lagrange particle. The particle's trajectory is not a simple animated path, but strictly follows the following Lagrange particle motion control equations: , In the formula, Let be the position vector of the particle. For time, The deterministic velocity field is the interpolated velocity at the particle's current position, calculated from the mean flow field obtained by the back-end hydrodynamic model. This is the random perturbation velocity term, used to characterize the turbulent fluctuation effects at the sub-grid scale. Since the grid resolution of hydrodynamic models is typically larger than the particle scale, the average flow field alone cannot reflect the diffusion effect of small-scale eddies on floating objects. Therefore, a random walk model is introduced to simulate this turbulent diffusion. Random perturbation term. They typically follow a normal distribution with a mean of zero and a variance that is related to the turbulent diffusion coefficient and the time step. After introducing this physical mechanism, the simulated particle swarm exhibits a diffusion pattern that conforms to natural laws over time, rather than mechanically translating along streamlines, thus ensuring the scientific credibility of the search and rescue range prediction.

[0042] At the numerical solution level, since the particle position is continuously changing while the flow field data is discretely gridded, the system employs a bilinear interpolation algorithm to calculate the instantaneous flow velocity at the particle's location by weighting the flow velocity values ​​from the surrounding four grid nodes based on the grid cell where the particle is located and its relative position. Subsequently, the fourth-order Runge-Kutta method was used to integrate the aforementioned motion control equations over time. Compared to the simple Euler method, the fourth-order Runge-Kutta method performs four slope evaluations and weighted averages within each time step, effectively suppressing numerical dissipation and phase errors, and ensuring the accuracy of long-term trajectory extrapolation in complex tidal fields. The system records the spatial coordinates of all particles at preset output time intervals (e.g., every 30 or 60 seconds). The data includes the corresponding timestamps, which are then serialized and written to the trajectory data file. This file serves as a standardized data carrier connecting physical calculations and visual representation, preserving complete spatiotemporal evolution information while shielding the complex numerical solution details from the backend. This allows the front-end 3D visualization scene to focus on high-fidelity rendering, achieving separate operation of scientific computing and interactive rendering.

[0043] In one embodiment, based on the foregoing embodiments, the specific implementation method of 3D visualization rendering and data mapping is further described in detail. This method converts the physical data calculated on the backend into visual signals that can be presented on the frontend. It is a connecting link in the decoupled architecture of computation and visualization, so that the visual performance is controlled by scientific data rather than artistic presets.

[0044] Specifically, the geographic coordinates in the dynamic risk matrix are converted into local spatial coordinates of the 3D visualization scene and rendered using color mapping to generate a dynamic inundation risk map. This includes: using preset coordinate transformation parameters to convert the geographic coordinates in the dynamic risk matrix into local 3D coordinates of the 3D visualization scene; querying a preset color mapping table to determine the corresponding color attribute values ​​based on the risk level or relative water depth value of each discrete spatiotemporal grid in the dynamic risk matrix; constructing vertex data from the local 3D coordinates and color attribute values, and rendering the dynamic inundation risk map in the 3D visualization scene by drawing primitives. Since the dynamic risk matrix output by the backend hydrodynamic model is usually based on the WGS84 geographic coordinate system or the UTM projected coordinate system, while the 3D visualization engine internally uses a local Cartesian coordinate system in meters, high-precision spatial registration is necessary. In this embodiment, an affine transformation matrix between the real geographic coordinates and the origin of the virtual scene is established by integrating a geographic reference plugin (such as Cesium for Unreal) or a custom projection transformation algorithm. This transformation process includes not only translation and rotation but also Earth curvature correction to ensure that the fitting error between the risk patch and the terrain model is controlled within centimeters in large-scale estuary and coastal scenes. It should be understood that although this embodiment uses the local coordinate transformation commonly used in game engines as an example, in other embodiments, if the 3D visualization scene itself is built based on the global geographic coordinate system, the transformation step can be simplified to a direct coordinate projection transformation.

[0045] After spatial registration, the system maps risk values ​​to visual attributes. The color map is not an arbitrary palette, but rather a semantic encoding rule defined by ergonomics and safety standards. For example, areas with a relative water depth less than 0.2m are designated green (safe), 0.2m to 0.4m are yellow (warning), and areas greater than 0.4m and in a concealed risk state are red (high-risk blocking). The system traverses each grid node in the dynamic risk matrix, using its risk level or water depth value as an index key to retrieve the corresponding RGBA color value from a pre-defined texture lookup table (LUT) or gradient function. Subsequently, the transformed local 3D coordinates and color attribute values ​​are assembled into a GPU-recognizable vertex buffer object (VBO), and a dynamic flooding risk map is rendered by overlaying triangular strips or sprite primitives onto the 3D terrain surface.

[0046] This data-driven rendering mechanism allows every color change to be traced back to specific physical calculation results, avoiding the problem of rendering effects being disconnected from actual data in traditional visualization, and providing decision-makers with a reliable situational awareness base map.

[0047] The system supports a large-scale overlay of risk heat maps from a third-person top-down perspective, classifying risks into three levels: safe, warning, and obstruction. It also accelerates the simulation of the dynamic process of the tide cutting off the escape route through a timeline. At the same time, it introduces a first-person immersive perspective, simulating blind spots through multiple sets of continuous perspectives, restoring the counterintuitive scenario of the escape route being quietly cut off by the rising tide, and revealing hidden risks that traditional two-dimensional maps cannot represent.

[0048] Furthermore, the trajectory data file is parsed, and the particle system is invoked to render dynamic 3D drift trajectories based on the time-series coordinates of the Lagrange particles. This includes: parsing the trajectory data file to extract the 3D spatial coordinates of each Lagrange particle at different timestamps; initializing the particle emitter in the 3D visualization scene, using the initial coordinates of the Lagrange particles as the emission positions; updating the position attributes of each particle in the particle emitter in each rendering frame by matching the timestamp with the current system time; and using a trailing rendering component to connect the positions of the same particle in consecutive rendering frames to generate dynamic 3D drift trajectories. (Refer to...) Figure 5 The Niagara-driven search and rescue trajectory visualization workflow typically outputs trajectory data files as discrete-time coordinate sequences (e.g., 1 point per second), while the front-end rendering frame rate is usually high (e.g., 60 FPS). To address the mismatch between data frequency and rendering frequency, the system executes timestamp matching logic for each rendering frame: first, it obtains the simulation time corresponding to the current rendering frame; then, it searches for two adjacent sampling points in the trajectory data sequence and calculates the precise position of the particle at that moment through linear interpolation; alternatively, it employs a nearest neighbor search strategy, directly reusing the coordinates of the sampling point with the smallest time difference. This synchronization strategy ensures the smoothness and physical realism of particle motion, avoiding trajectory jumps or backtracking caused by frame rate fluctuations.

[0049] At the visual level, simple point-like particles cannot intuitively express the spatiotemporal continuity of the "drift path" concept. Therefore, this embodiment calls a trail rendering component (such as Ribbon Renderer or Trail Renderer) in the particle system to generate dynamic 3D drift trajectories. This component maintains a historical position buffer for each particle in memory, and during rendering, it sequentially connects the current frame position with the historical positions of the previous N frames to construct a continuous strip-shaped geometry or line primitive. The length, width, and transparency decay curve of the trail can all be configured to characterize the time span or confidence level of the trajectory. For example, setting the trail length to 30 frames and applying a gradient material from opaque at the head to fully transparent at the tail allows the observer to clearly see the current drift direction and intuitively perceive the past movement trend through the trail length. It should be understood that although this embodiment uses Ribbon rendering as an example, in other implementations, Line Renderer, Shader Trail, or screen-space post-processing can also be used to achieve trajectory visualization, as long as its essence is based on the time-series reconstruction of backend physical data. In addition, to support search and rescue decision-making, the system can also generate thermal layers in real time based on the density distribution of the particle swarm, aggregating a large number of discrete trajectories into high-probability search areas, so as to provide front-line rescuers with directly operable search range information.

[0050] In one embodiment, the present invention provides a digital twin emergency response system for estuarine and coastal disaster risks. The system includes a backend computing server (see reference...). Figure 2 Backend computing layer) and frontend interactive terminal (refer to Figure 2 The front-end visualization layer), the back-end computing server, and the front-end interactive terminal communicate through standardized data interfaces (see...). Figure 2 The data interface layer facilitates communication connections. Specifically, the backend computing server, serving as the core of physical computing, is typically deployed on high-performance workstations or cloud computing clusters. It integrates the SOH disaster scoring algorithm module and the Ecom-si hydrodynamic model solver, responsible for high-computational-power-consuming tasks such as terrain data processing, relative water depth calculation, dynamic risk matrix generation, and Lagrange particle trajectory solving. The frontend interactive terminal is deployed on graphics workstations or high-performance PCs, running a 3D visualization application built on Unreal Engine 5 (UE5), incorporating the Niagara particle system, Cesium georeferenced plugin, and FluidFlux2 water rendering module, focusing on high-fidelity scene rendering and human-computer interaction response. This hardware-level physical separation and functional specialization design allows scientific computing and graphics rendering to be processed in parallel, avoiding performance bottlenecks caused by a single device simultaneously bearing dual loads, providing hardware support for the system to achieve a 35-minute rapid response time.

[0051] The front-end interactive terminal is used to display 3D visualization scenes. Responding to location selection commands within the 3D visualization scene, it obtains the geographic coordinates of the selected location, generates an initial Lagrange particle matrix file based on these coordinates, and sends the initial Lagrange particle matrix file to the back-end computing server. In specific implementation, the front-end integrates the Cesium for Unreal plugin to load Gaofen-2 satellite imagery and high-precision DEM terrain data, constructing a geographically accurate digital twin base. When a user clicks on a suspected water-fall point in the 3D scene, the front-end blueprint script automatically captures the screen pixel coordinates and converts them into WGS84 latitude and longitude. Combined with preset diffusion parameters, it generates a structured initial particle matrix file (e.g., .dat format) and instantly triggers the back-end computing task through a standardized data interface. This process is fully automated, eliminating the tedious steps of manually recording coordinates and editing model input files in traditional workflows.

[0052] The backend computing server acquires topographic elevation data and time-varying tidal level data of the target beach; based on the topographic elevation data and time-varying tidal level data, it calculates the relative water depth of the discrete spatiotemporal grid within a preset time step; it acquires ergonomic escape constraint parameters, including the lethal water level threshold and differentiated mobility parameters; based on the relative water depth and ergonomic escape constraint parameters, it dynamically evaluates the connectivity of the safe evacuation path and generates a dynamic risk matrix; the dynamic risk matrix is ​​used to characterize hidden risk areas, which satisfy the following conditions: the relative water depth at the current moment is less than the lethal water level threshold, and within the evacuation time window calculated based on the differentiated mobility parameters, the relative water depth on the necessary evacuation path is greater than or equal to the lethal water level threshold; the initial Lagrange particle matrix file is input into the backend hydrodynamic model for solution, and the output is a trajectory data file containing the time series coordinates of multiple Lagrange particles; the dynamic risk matrix and trajectory data file are sent to the front-end interactive terminal through a standardized data interface. Upon receiving a computation request triggered by the frontend, the backend automatically invokes the SOH module to perform parallel or serial computation with the hydrodynamic model. In addition to the core algorithm, the backend integrates multi-source data fusion interfaces. For example, it accesses the Wind Weather API via timed polling or message subscription to obtain real-time meteorological data such as wind direction, wind speed, and rainfall rate, and uses this data as boundary conditions to input into the Ecom-si model to improve the accuracy of flow field simulation. Simultaneously, the meteorological parameters are forwarded to the frontend to drive plugins such as Ultra Dynamic Sky to achieve synchronized virtual-real environment rendering. This automated multi-source data fusion mechanism ensures that the digital twin scene is geometrically aligned with the real world while maintaining dynamic consistency in environmental conditions, providing context-aware support for emergency decision-making.

[0053] The front-end interactive terminal is also used to convert the geographic coordinates in the dynamic risk matrix into local spatial coordinates of the 3D visualization scene and perform color mapping rendering to generate a dynamic flooding risk map; it parses the trajectory data file and calls the particle system to render dynamic 3D drift trajectories based on the time-series coordinates of Lagrange particles. After receiving the data file returned from the back end, the front end uses a custom data parser to convert the numerical matrix into vertex data and particle attributes that can be recognized by the GPU, driving the Niagara system to reconstruct dynamic streamlines that conform to physical laws in real time, and superimposing the risk level with semantic colors on the 3D terrain, completing the deterministic mapping from scientific data to visual signals.

[0054] Furthermore, the standardized data interface adopts the WebSocket or gRPC protocol; during the calculation process, the backend computing server periodically sends progress status information to the frontend interactive terminal through the standardized data interface; after receiving the dynamic risk matrix and trajectory data file, the frontend interactive terminal returns a confirmation of receipt instruction to the backend computing server through the standardized data interface. (Refer to...) Figure 6 The communication process, although illustrated in the diagram, uses the MQTT protocol. However, in this embodiment, to meet the requirements of large file transfer and low-latency bidirectional communication, the WebSocket full-duplex protocol or the gRPC high-performance remote procedure call protocol is preferred. Specifically, when the backend starts a time-consuming hydrodynamic calculation task, it periodically (e.g., once per second) sends progress information to the frontend via this standardized interface in the form of heartbeat packets or message pushes. This information includes the current number of calculation steps, the estimated remaining time, and the task status. The frontend updates the progress bar or status prompts on the user interface in real time based on this information, avoiding the user's misjudgment that the system is frozen due to a lack of feedback for a long time. After the frontend has fully received the dynamic risk matrix (.csv) and trajectory data file (.dat) and completed local verification, it immediately returns an acknowledgment instruction (ACK) to the backend via the same interface. Only after receiving this instruction does the backend release the temporary cache or mark the task as completed. This communication mechanism, which includes status feedback and handshake confirmation, effectively solves common problems in traditional one-way file transfer such as packet loss, version inconsistency, and status asynchrony. It ensures the reliability and timeliness of data transmission in high-concurrency emergency scenarios, thereby supporting a rapid response capability of only about 35 minutes from receiving the alarm to generating a visualized search and rescue plan. This represents a significant efficiency improvement compared to the traditional 95-minute serial process.

[0055] In one specific embodiment, the digital twin emergency response assistance method and system for estuarine and coastal disaster risks, based on the aforementioned embodiments, further illustrates its typical application scenarios in emergency response assistance and public education regarding estuarine and coastal disaster risks. Through practical application verification, this invention can effectively address the shortcomings of traditional technologies in terms of search and rescue response timeliness and public awareness of risk.

[0056] In emergency search and rescue applications, this invention significantly shortens decision-making response time by reconstructing business processes. Traditional search and rescue support processes typically employ a sequential workflow, involving multiple stages such as accident detection, police notification of experts, expert waiting for responses, manual model input preparation, hydrodynamic model execution, manual post-processing drawing, and feedback to the police, averaging approximately 95 minutes. Cross-departmental communication and coordination, manual data conversion, and post-processing visualization consume a significant amount of non-computational time, easily leading to missed opportunities during critical rescue windows. In contrast, this invention employs an automated parallel architecture that decouples computation and visualization, simplifying the process into four core stages: accident detection, direct interaction by police in a 3D visualization scene, automatic backend hydrodynamic model calculation, and real-time 3D visualization rendering. Frontend interaction and command parsing take only about 3 minutes, backend hydrodynamic model calculation maintains a physical computation baseline of about 30 minutes, and result feedback and real-time rendering take only about 1 minute, compressing the overall response time to approximately 35 minutes, representing an efficiency improvement of approximately 63% compared to traditional methods. This efficiency improvement does not stem from simplifying the physical calculation process or sacrificing accuracy, but rather from eliminating the bottleneck of human intervention through standardized data interfaces, achieving a minute-level closed loop from receiving the alarm to generating a visualized search and rescue plan.

[0057] Specifically, in actual search and rescue operations, commanders do not need professional numerical simulation knowledge. They only need to click on the suspected drowning area in the 3D visualization scene based on on-site intelligence, and the system can automatically generate the initial Lagrange particle matrix and trigger the backend calculation. The dynamic 3D drift trajectory output by the system can intuitively show the drift and diffusion range of the drowning person over time and the high-probability search area. The dynamic 3D drift trajectory is distinguished and rendered in the 3D visualization scene using particle streamlines of different colors (such as pink and blue trajectories). In practice, different colored trajectories can be used to represent the displacement path of the drowning person in different time windows, or to intuitively distinguish search areas of different probability levels (for example, a high-density pink spiral trajectory represents the core search area with the highest probability, and a blue trajectory represents the diffusion and evolution area with a lower probability). Combined with the underlying satellite imagery, this multi-dimensional color feature mapping can help commanders quickly locate key inspection areas in complex sea conditions, improving the accuracy of search and rescue deployment. In a real-world search and rescue operation, based on the predicted path generated by the system, frontline rescue forces successfully located the target within the predicted area, and the actual location of the target closely matched the high-density particle aggregation area predicted by the system. This practical feedback demonstrates that the Lagrange particle tracking algorithm and the backend hydrodynamic model used in this invention are scientifically sound and reliable in complex tidal field environments, and can be used to support search and rescue decision-making.

[0058] In public science education applications, this invention utilizes the high-fidelity rendering capabilities and immersive interactive features of 3D visualization scenes to reduce the difficulty for non-professionals to understand hidden hazards along estuaries and coastlines. During science popularization activities such as World Water Day, the system is deployed on public display platforms, allowing the public to experience a virtual beach from a first-person perspective using controllers or touch devices. Unlike traditional static display boards or 2D diagrams, users can observe firsthand in a realistic 3D environment the "lower then higher" filling sequence of hidden tidal channels during high tide, as well as the dynamic process of escape routes being quietly cut off.

[0059] 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.

[0060] The above-described embodiments are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A digital twin emergency response method for estuarine and coastal disaster risks, characterized in that, include: Obtain topographic elevation data and time-varying tidal level data for the target beach; Based on the terrain elevation data and time-varying tide data, the relative water depth of the discrete spatiotemporal grid within a preset time step is calculated. Obtain ergonomic escape constraint parameters, and based on the relative water depth and ergonomic escape constraint parameters, dynamically assess the connectivity of the safe evacuation path to generate a dynamic risk matrix; The ergonomic escape constraint parameters include lethal water level thresholds and differentiated mobility parameters; The dynamic risk matrix is ​​used to characterize hidden risk areas, which satisfy the following conditions: the relative water depth at the current moment is less than the lethal water level threshold, and within the evacuation time window calculated based on the differentiated mobility parameters, the relative water depth on the necessary evacuation path is greater than or equal to the lethal water level threshold. In response to a location selection command for a 3D visualization scene, the geographic coordinates of the selected location are obtained, and an initial Lagrange particle matrix file is generated based on the geographic coordinates. The initial matrix file of the Lagrange particles is input into the backend hydrodynamic model for solution, and the output is a trajectory data file containing the time series coordinates of multiple Lagrange particles; the backend hydrodynamic model runs independently of the three-dimensional visualization scene, and the two interact with each other through a standardized data interface; The dynamic risk matrix and the trajectory data file are transmitted to the 3D visualization scene through the standardized data interface; the geographic coordinates in the dynamic risk matrix are converted into local spatial coordinates of the 3D visualization scene and color-mapped rendering is performed to generate a dynamic flooding risk map; the trajectory data file is parsed, and the particle system is called to render a dynamic 3D drift trajectory based on the time series coordinates of the Lagrange particles.

2. The digital twin emergency response method for estuarine and coastal disaster risks according to claim 1, characterized in that, The step of calculating the relative water depth within a preset time step using the topographic elevation data and the time-varying tidal data includes: The terrain elevation data is divided into multiple two-dimensional spatial grids, and the ground elevation value of each two-dimensional spatial grid is obtained; Based on the time-varying tide data, the instantaneous tide value corresponding to each preset time step is calculated by interpolation; The difference between the instantaneous tide level and the corresponding ground elevation value is used to obtain the relative water depth of each two-dimensional spatial grid within the preset time step.

3. The digital twin emergency response method for estuarine and coastal disaster risks according to claim 2, characterized in that, The connectivity of the safe evacuation path is dynamically assessed based on the relative water depth and the ergonomic escape constraint parameters to generate a dynamic risk matrix, including: Based on the differentiated mobility parameters, the evacuation speed for different target groups is determined; Based on the evacuation speed and the relative water depth of the discrete spatiotemporal grid, the passage impedance of each of the two-dimensional spatial grids is calculated; A time-varying topology graph is constructed using the two-dimensional spatial grid as nodes and the passage impedance between adjacent grids as edge weights. Based on the time-varying topology map, the shortest path algorithm is used to calculate the optimal evacuation path and corresponding evacuation time from each of the two-dimensional spatial grids to the preset safe area.

4. The digital twin emergency response method for estuarine and coastal disaster risks according to claim 3, characterized in that, The generation of the dynamic risk matrix further includes the step of identifying the hidden risk areas, wherein identifying the hidden risk areas includes: For each node on the optimal evacuation path, extract its relative water depth prediction value in the future time series; Determine whether there is a target time within the future time series such that the predicted relative water depth at the target time is greater than or equal to the lethal water level threshold. If it exists, calculate the time difference between the current time and the target time, as the available evacuation time; If the available evacuation time is less than the evacuation time corresponding to the optimal evacuation path, then the two-dimensional spatial grid where the current node is located is marked as the hidden risk area.

5. The digital twin emergency response method for estuarine and coastal disaster risks according to claim 1, characterized in that, The step of responding to a location selection command for a 3D visualization scene, obtaining the geographic coordinates of the selected location, and generating an initial Lagrange particle matrix file based on the geographic coordinates includes: Parse the location selection command and extract the latitude and longitude coordinates and release timestamp of the selected location; Based on the preset particle diffusion radius and particle number, multiple initial particle positions are generated within the neighborhood of the latitude and longitude coordinates. The initial particle position, the release timestamp, and the preset particle physical properties are encapsulated into structured matrix data to generate the initial Lagrange particle matrix file.

6. The digital twin emergency response method for estuarine and coastal disaster risks according to claim 5, characterized in that, The process involves inputting the initial Lagrange particle matrix file into the backend hydrodynamic model for solution, and outputting a trajectory data file containing multiple Lagrange particle time series coordinates, including: The back-end hydrodynamic model solves the flow field distribution based on the two-dimensional shallow water equation to obtain the velocity field and water level field at each preset time step. For each particle in the initial Lagrange particle matrix file, bilinear interpolation is performed in the velocity field according to its position to obtain the instantaneous velocity. The Runge-Kutta method is used to integrate the instantaneous flow velocity over time to update the spatial coordinates of the particles; The spatial coordinates of the particles are recorded at preset output time intervals to generate the trajectory data file.

7. The digital twin emergency response assistance method for estuarine and coastal disaster risks according to claim 1, characterized in that, The step of converting the geographic coordinates in the dynamic risk matrix into local spatial coordinates of the 3D visualization scene and performing color mapping rendering to generate a dynamic flooding risk map includes: Using preset coordinate transformation parameters, the geographic coordinates in the dynamic risk matrix are converted into local 3D coordinates of the 3D visualization scene; Based on the risk level or relative water depth value of each discrete spatiotemporal grid in the dynamic risk matrix, a preset color mapping table is queried to determine the corresponding color attribute value. The local 3D coordinates and color attribute values ​​are used to construct vertex data, and the dynamic flooding risk map is generated by rendering primitives in the 3D visualization scene.

8. The digital twin emergency response method for estuarine and coastal disaster risks according to claim 1, characterized in that, The process of parsing the trajectory data file and calling the particle system to render a dynamic three-dimensional drift trajectory based on the time-series coordinates of the Lagrange particles includes: The trajectory data file is parsed to extract the three-dimensional spatial coordinates of each Lagrange particle at different timestamps; Initialize the particle emitter in the three-dimensional visualization scene, using the initial coordinates of the Lagrange particles as the emission position; In each rendering frame, the position attributes of each particle in the particle emitter are updated by matching the timestamp with the current system time; The dynamic 3D drift trajectory is generated by connecting the positions of the same particle in consecutive rendering frames using a trailing rendering component.

9. A digital twin emergency support system for estuarine and coastal disaster risks, characterized in that, It includes a back-end computing server and a front-end interactive terminal, which are connected to each other through a standardized data interface. The front-end interactive terminal is used to display a three-dimensional visualization scene, obtain the geographic coordinates of the selected location in response to the location selection command of the three-dimensional visualization scene, generate an initial Lagrange particle matrix file based on the geographic coordinates, and send the initial Lagrange particle matrix file to the back-end computing server. The backend computing server is used to acquire topographic elevation data and time-varying tidal level data of the target beach; calculate the relative water depth of the discrete spatiotemporal grid within a preset time step based on the topographic elevation data and the time-varying tidal level data; acquire ergonomic escape constraint parameters, including a lethal water level threshold and differentiated mobility parameters; dynamically evaluate the connectivity of the safe evacuation path based on the relative water depth and the ergonomic escape constraint parameters, and generate a dynamic risk matrix; wherein, the dynamic risk matrix is ​​used to characterize hidden risk areas, and the hidden risk areas satisfy: the relative water depth at the current moment is less than the lethal water level threshold, and within the evacuation time window calculated based on the differentiated mobility parameters, the relative water depth on the necessary evacuation path is greater than or equal to the lethal water level threshold; input the initial Lagrange particle matrix file into the backend hydrodynamic model for solution, and output a trajectory data file containing multiple Lagrange particle time series coordinates; send the dynamic risk matrix and the trajectory data file to the frontend interactive terminal through the standardized data interface; The front-end interactive terminal is also used to convert the geographic coordinates in the dynamic risk matrix into local spatial coordinates of the three-dimensional visualization scene and perform color mapping rendering to generate a dynamic flooding risk map; parse the trajectory data file and call the particle system to render a dynamic three-dimensional drift trajectory based on the time series coordinates of the Lagrange particles.

10. The digital twin emergency response system for estuarine and coastal disaster risks according to claim 9, characterized in that, The standardized data interface adopts the WebSocket protocol or the gRPC protocol; During the calculation process, the backend computing server periodically sends progress status information to the frontend interactive terminal through the standardized data interface; After receiving the dynamic risk matrix and the trajectory data file, the front-end interactive terminal returns a confirmation of receipt instruction to the back-end computing server through the standardized data interface.

Images