Methods and systems for dynamic simulation and prediction of typhoon disaster losses

Abstract

This application discloses a dynamic extrapolation and prediction method and system for typhoon disaster losses, relating to the field of typhoon disaster loss extrapolation technology. Based on a parallel hot-start strategy of Monte Carlo set path generation and state inheritance with physical constraints, this application significantly improves the computational efficiency of large-scale set simulations and solves the bottleneck of traditional methods that are difficult to balance accuracy and timeliness. By calculating the joint exceedance probability of multiple disaster factors and the disaster chain triggering index, it breaks through the limitation of isolated analysis of single disasters and can accurately characterize the compound risks of multiple disasters such as floods, landslides and strong winds and their chain transmission effects, providing a new paradigm for risk quantification under complex disaster scenarios. The constructed "risk navigator" realizes full-element dynamic visualization from disaster-causing factors to the failure consequences of disaster-bearing bodies on a spatiotemporal GIS platform, supporting emergency command personnel to quickly locate high-risk hotspots, identify the main disaster-causing factors and predict chain consequences.

Description

Technical Field

[0001] This application relates to the field of typhoon disaster loss projection technology, specifically to dynamic projection and prediction methods and systems for typhoon disaster losses. Background Technology

[0002] Typhoons are one of the most significant meteorological drivers of chain disasters, including torrential rains, floods, storm surges, and landslides. Their path and intensity evolution exhibit significant uncertainty, posing a major challenge to disaster risk assessment and emergency decision-making. Existing typhoon disaster risk assessment methods largely rely on deterministic forecast paths or static scenario simulations, failing to adequately characterize the spatial variability of the impact of path uncertainty on affected areas. Furthermore, traditional hydrological, hydrodynamic, and slope stability models typically operate independently, lacking the comprehensive simulation capability for multi-hazard coupling processes and the dynamic triggering mechanisms of disaster chains. Current decision support systems primarily rely on static risk maps for single hazards, lacking dynamic integration and visualization of the joint probabilities of multiple hazard factors, infrastructure network vulnerability, and chain consequences, making it difficult to support high-resolution, sophisticated emergency command.

[0003] Regarding the above-mentioned solutions, the inventors of this application have discovered that the above-mentioned technology has at least the following technical problems:

[0004] 1. Currently, the lack of dynamic smoothing constraints in Monte Carlo random perturbation processes, relying solely on statistical perturbations to generate typhoon ensemble paths, results in numerous physically unreasonable paths (such as discontinuous paths, abrupt changes in velocity direction, etc., which do not conform to atmospheric motion laws). The existence of such "non-physical paths" seriously interferes with the accuracy of risk assessment, causing subsequent disaster simulations to deviate from the actual typhoon evolution characteristics, and significantly reducing the authenticity and reliability of the assessment results.

[0005] 2. Currently, there is a lack of a state inheritance hot-start mechanism. All state variables (including long-memory variables such as deep soil water and groundwater, and short-memory variables such as surface runoff) are initialized using a uniform "cold start" method. In ensemble simulations, each path needs to repeatedly perform the complete physical process initialization calculation, resulting in a linear increase in computational cost with the number of paths, making it difficult to support large-scale, efficient simulations with N ≥ 200 paths. Furthermore, existing technologies do not differentiate between inheritable states and states that need to be reset, causing serious redundant calculations and disrupting the continuity of the physical process.

[0006] 3. Current risk assessment methods are mostly limited to single disaster factors, failing to construct joint exceedance probabilities of multiple disaster factors and disaster chain triggering indices, making it difficult to achieve joint diagnosis of compound disasters and chain risks. Traditional single-hazard assessment methods cannot identify high-risk areas under multi-hazard coupling, and also lack the ability to quantitatively analyze combinations of main disaster-causing factors. As a result, the identification of "risk hotspots" and the judgment of chain failure paths lack scientific basis, and the comprehensiveness and practicality of risk assessment are significantly insufficient. Summary of the Invention

[0007] In view of the above-mentioned technical deficiencies, the purpose of this application is to provide a method and system for dynamic simulation and prediction of typhoon disaster losses.

[0008] To solve the above-mentioned technical problems, this application adopts the following technical solution: In the first aspect, this application provides a method for dynamic simulation and prediction of typhoon disaster losses, which includes the following steps: S1, acquiring multi-source heterogeneous data and constructing a comprehensive geographic information database containing infrastructure network topology relationships.

[0009] S2. Based on a comprehensive geographic information database, generate a set of typhoon paths driven by physical constraints.

[0010] S3. Based on the typhoon ensemble path and the comprehensive geographic information database, generate dynamic precipitation field sequences for each typhoon path.

[0011] S4. Parallel simulation of the typhoon ensemble path is performed using a hydrological and hydrodynamic model to generate a disaster evolution result set.

[0012] S5. Based on the disaster evolution result set, generate a joint probability distribution map of disasters and a spatial distribution map of disaster chain triggering index.

[0013] S6. Generate a risk navigator based on the spatial distribution map of the disaster chain triggering index.

[0014] Preferably, the step of calculating the actual rainfall intensity of spatial grid points at future times based on the baseline rainfall rate and the topographic amplification effect, and generating a dynamic rainfall field sequence with respect to the typhoon path, includes: using a calculation formula... The result is the The first typhoon path Each spatial lattice point at time... Actual rainfall intensity ,in Represented as time The baseline rainfall rate, Represented as the first Each spatial lattice point at time... The terrain amplification effect.

[0015] Preferably, the method of using a hydrological and hydrodynamic model to perform parallel simulation of the typhoon ensemble path to generate a disaster evolution result set includes: S701, based on the physical characteristics of the target watershed and historical hydrological data, initializing a hydrological and hydrodynamic-slope stability coupled model to obtain the complete set of initial state variables of the model at the start of the simulation.

[0016] S702 defines an inheritable subset of state variables and a subset of state variables that need to be reset, based on the principle of independence between the physical response time scale of state variables and the typhoon path.

[0017] S703 deploys a parallel computing framework to allocate independent computing resources to each typhoon path in the typhoon ensemble path in order to initiate parallel simulations.

[0018] S704 performs a state inheritance hot start evolution operation for each typhoon path and outputs a disaster evolution result set for each typhoon path.

[0019] Preferably, the step of generating a joint probability distribution map of disasters and a spatial distribution map of disaster chain triggering index based on the disaster evolution result set includes: S801, performing multi-scenario result integration processing based on the disaster evolution result sets to obtain a spatiotemporally punctuated multi-disaster scenario dataset.

[0020] S802, based on the multi-hazard scenario dataset, perform exceedance probability statistical calculation on the preset single hazard intensity threshold to obtain the single hazard factor exceedance probability distribution field at each time point.

[0021] S803, based on the single disaster factor exceedance probability distribution field, perform joint probability calculation on exceedance threshold events of at least two disaster factors to obtain the multi-disaster factor joint exceedance probability distribution field at each time point.

[0022] S804, based on infrastructure network, disaster-bearing body data, single disaster factor exceedance probability distribution field and the multi-disaster factor joint exceedance probability distribution field, calculates the disaster chain triggering index at each time point through a comprehensive risk function.

[0023] S805, based on the joint transcendence probability distribution field of the multi-hazard factors and the disaster chain triggering index, generate and output the joint probability distribution map of the disaster and the spatial distribution map of the disaster chain triggering index.

[0024] Preferably, the step of generating a risk navigator based on the spatial distribution map of the disaster chain triggering index includes: S901, performing a hierarchical early warning threshold setting operation on a preset hierarchical rule base based on the spatial distribution map of the disaster chain triggering index to obtain a hierarchical early warning threshold set.

[0025] S902, based on the graded early warning threshold set and the spatial distribution map of the disaster chain triggering index, perform spatial clustering and extreme value identification operations on the disaster chain triggering index values ​​to obtain a risk intervention hotspot set.

[0026] S903, based on the aforementioned risk intervention hotspot set, disaster joint probability distribution map, and multi-hazard scenario dataset, perform correlation analysis on the main disaster-causing factor combinations and possible chain consequences to obtain an enhanced risk hotspot set.

[0027] S904. Based on the enhanced risk hotspot set, typhoon ensemble path, disaster joint probability distribution map, and disaster chain triggering index spatial distribution map, an integrated dynamic visualization rendering operation is performed on the spatiotemporal geographic information system platform to obtain a risk navigator.

[0028] In its second aspect, this application provides a system for dynamic simulation and prediction of typhoon disaster losses, comprising: preferably, a comprehensive geographic information database construction module for acquiring multi-source heterogeneous data and constructing a comprehensive geographic information database containing infrastructure network topology relationships.

[0029] The typhoon ensemble path generation module generates typhoon ensemble paths driven by physical constraints, based on a comprehensive geographic information database.

[0030] The dynamic precipitation field sequence generation module generates dynamic precipitation field sequences for each typhoon path based on the typhoon ensemble path and a comprehensive geographic information database.

[0031] The disaster evolution result set generation module uses a hydrological and hydrodynamic model to simulate the typhoon ensemble path in parallel in order to generate a disaster evolution result set.

[0032] The loss projection module generates a joint probability distribution map of disasters and a spatial distribution map of disaster chain triggering index based on the disaster evolution result set.

[0033] The risk navigator generation module generates a risk navigator based on the spatial distribution map of the disaster chain triggering index.

[0034] The beneficial effects of this application are as follows: 1. The dynamic simulation and prediction method and system for typhoon disaster losses provided in this application, based on the parallel hot-start strategy of Monte Carlo set path generation and state inheritance under physical constraints, significantly improves the computational efficiency of large-scale set simulation and solves the bottleneck of traditional methods that are difficult to balance accuracy and timeliness; by calculating the joint exceedance probability of multiple disaster factors and the disaster chain triggering index, it breaks through the limitation of isolated analysis of a single disaster type, and can accurately characterize the compound risks of multiple disasters such as floods, landslides and strong winds and their chain transmission effects, providing a new paradigm for risk quantification under complex disaster scenarios; the constructed "risk navigator" realizes full-element dynamic visualization from disaster-causing factors to the failure consequences of disaster-bearing bodies on the spatiotemporal GIS platform, supporting emergency command personnel to quickly locate high-risk hotspots, identify the main disaster-causing factors and predict chain consequences.

[0035] 2. This application introduces dynamic smoothing constraints into Monte Carlo random perturbations to generate physically reasonable typhoon ensemble paths (such as continuous paths and velocity direction changes conforming to atmospheric motion laws), rather than relying solely on statistical perturbations. This significantly improves the physical realism and representativeness of the ensemble paths, avoids the interference of "non-physical paths" on risk assessment in traditional random generation methods, and makes subsequent disaster simulations closer to the actual typhoon evolution characteristics, thereby enhancing the reliability of the assessment results.

[0036] 3. This application introduces a state inheritance hot-start mechanism, separating inheritable states (such as deep soil water and groundwater) from states requiring reset (such as surface runoff) for efficient parallel computation in ensemble simulations. While maintaining the continuity of physical processes, it significantly reduces the computational overhead of ensemble simulations, solving the problem of repetitive computation in large-path simulations using traditional "cold-start" methods, making large-scale simulations of N≥200 paths computationally feasible and efficient. It constructs a joint exceedance probability of multiple hazard factors (such as simultaneous inundation and strong winds) and a hazard chain triggering index (HCI), elevating single-hazard assessment to a joint diagnosis of compound hazards and chain risks. This overcomes the limitations of traditional single-hazard risk assessment, identifying high-risk areas and their main hazard-causing factor combinations under multi-hazard coupling, providing quantitative evidence for accurately identifying "risk hotspots" and chain failure paths, and significantly improving the comprehensiveness and practicality of risk assessment. Attached Figure Description

[0037] 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.

[0038] Figure 1This is a flowchart illustrating the steps involved in implementing the method described in this application.

[0039] Figure 2 This is a schematic diagram of the system structure connection of this application. Detailed Implementation

[0040] 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 skilled in the art without creative effort are within the scope of protection of this application.

[0041] Please see Figure 1 As shown, this application provides a method for dynamic simulation and prediction of typhoon disaster losses in the first aspect, including: S1, acquiring multi-source heterogeneous data and constructing a comprehensive geographic information database containing infrastructure network topology relationships.

[0042] In a specific example, the multi-source heterogeneous data includes real-time typhoon forecast information, high-precision geographic information data, historical disaster cases, vulnerability curves of disaster-bearing bodies, real-time monitoring data, and spatial distribution and topological relationship data of critical infrastructure.

[0043] In a specific example, the construction of a comprehensive geographic information database containing infrastructure network topology relationships includes: S301, based on a preset standardized rule base, performing format unification, unit conversion, and logical consistency verification on multi-source heterogeneous data to obtain a standardized dataset.

[0044] S302, based on the target geographic reference system and coordinate transformation model, the spatiotemporal attribute data in the standardized dataset are aligned with coordinate system one and spatial location to obtain an aligned dataset.

[0045] S303, based on preset quality control logic, checks and corrects the data accuracy, timeliness and outliers in the aligned dataset to obtain a valid dataset.

[0046] S304, based on a spatial database engine, integrates the effective dataset and constructs its network topology relationship based on the spatial distribution data of critical infrastructure to generate the comprehensive geographic information database containing the infrastructure network topology relationship.

[0047] It should be noted that the format unification, unit conversion, and logical consistency verification based on a preset standardized rule base involve processing data from various sources, structures, and formats to create a unified data paradigm directly usable by the model. Specifically, format unification standardizes all time information to the ISO 8601 format defined by the International Organization for Standardization, all spatial coordinates to the WGS84 latitude and longitude format defined by the World Geodetic System, and all text attributes to UTF-8 encoding. Unit conversion standardizes wind speed units in typhoon intensity data to meters per second (m / s), rainfall intensity to millimeters per hour (mm / h), and geographic elevation to meters (m). Logical consistency verification verifies the internal logical relationships within the data; for example, it checks whether the time series of typhoon path points strictly increases and whether the geographic coordinates in the spatial distribution data of infrastructure are within the geographic boundaries of the study area. This operation eliminates the interference of data heterogeneity on subsequent model calculations and is the foundation for data fusion.

[0048] It should be noted that the standardized dataset is a collection of data obtained after standardization processing. This eliminates inconsistencies in format, units, and basic logic of the original data, forming a logically consistent intermediate data product with a unified spatiotemporal reference and unit of measurement. The purpose of this dataset is to provide standardized input for subsequent accurate spatial registration and high-quality data integration.

[0049] It should be noted that the coordinate system alignment process based on the target geographic reference system and coordinate transformation model aims to unify all data within the same geospatial framework, ensuring that typhoon paths, geographic features, and infrastructure locations can be accurately overlaid and analyzed spatially. The specific process is as follows: First, the target geographic reference system is determined (e.g., using the CGCS2000 National Geodetic Coordinate System); then, for vector data from different coordinate systems or projections (such as administrative boundaries and road networks), a seven-parameter or four-parameter coordinate transformation model is applied to transform them to the target coordinate system; for raster data (such as digital elevation models), resampling and projection transformation are performed to align its pixels with the grid in the target coordinate system. Spatial alignment ensures that the spatial positions of geographic entities such as points, lines, and areas are strictly matched across different data layers.

[0050] It should be noted that the alignment dataset is a collection of data that has been registered in the spatial dimension. It is a dataset in which all geospatial data elements have an absolute or relative position in a unified and accurate coordinate reference system. The purpose of this dataset is to ensure that the calculation of typhoon impact range, disaster field simulation (such as inundation analysis), and disaster-bearing body exposure assessment can be carried out at accurate geographical locations. It is a fundamental prerequisite for the correctness of spatial analysis.

[0051] It should be noted that the aforementioned inspection and correction based on preset quality control logic aims to identify and address errors, missing information, or unreasonable information in the data, thereby improving data reliability. Specifically, this includes: data accuracy checks, such as verifying whether the resolution of high-precision geographic information data meets modeling requirements; data timeliness checks, such as confirming that the delay between the timestamp of real-time monitoring data and the current time is within an acceptable threshold; and outlier checks and corrections, such as identifying and removing significant outliers in the typhoon forecast error ellipse parameters (e.g., abnormally large major axis radius), or imputing missing vulnerability curve parameters of disaster-bearing bodies based on historical statistical data. For critical infrastructure network topologies, the integrity of their connectivity needs to be checked to avoid isolated nodes or broken edges.

[0052] It should be noted that the effective dataset is a high-reliability data set that has undergone quality control and meets the requirements of accuracy, timeliness, and rationality. Noise and errors have been filtered out, improving the overall reliability and usability of the data. This dataset serves as "clean" raw material for building a high-quality geographic information database, directly determining the reliability of subsequent disaster simulation and loss projection results.

[0053] It should be noted that the integration based on a spatial database engine and the construction of network topology relationships involves consolidating scattered, different types of "effective datasets" into a unified data management structure and establishing a connection relationship model between critical infrastructures (such as power grids and transportation networks). Specifically, a spatial database engine (such as PostgreSQL, an extension of PostGIS) is used to store vector data (points, lines, and polygons), raster data, attribute tables, and time-series data in a unified database, and spatial indexes are established to accelerate queries. Constructing network topology relationships refers to creating a network graph model based on the spatial distribution data of critical infrastructures (such as substation locations, transmission lines, and road intersections). Nodes represent infrastructure entities (such as substations and transportation hubs), edges represent physical connections or functional dependencies between entities (such as transmission lines and road segments), and data structures such as adjacency matrices or association matrices are used to store the connectivity, directionality, and weights (such as transmission capacity and road capacity) between nodes.

[0054] It should be noted that the comprehensive geographic information database containing infrastructure network topology is the final output of the entire preprocessing workflow. It is a structured data warehouse integrating meteorological, geographical, disaster-bearing body, historical cases, real-time monitoring, and infrastructure network topology data, possessing a unified spatiotemporal benchmark and high-quality standards. It serves as a digital twin data foundation for supporting dynamic simulation and chain-based risk analysis of the entire typhoon disaster process.

[0055] S2. Based on a comprehensive geographic information database, generate a set of typhoon paths driven by physical constraints.

[0056] In a specific example, the step of generating a typhoon ensemble path driven by physical constraints based on a comprehensive geographic information database includes: S401, extracting the principal axis and constructing the covariance matrix based on the forecast error ellipse parameters in the comprehensive geographic information database to obtain the covariance matrix.

[0057] S402, based on the covariance matrix, perform random perturbation sampling on the preset initial center position of the typhoon path to obtain a set containing N random offset positions.

[0058] S403, based on the dynamic smoothing constraint algorithm, performs temporal smoothing and terrain constraint correction on each position sequence in the random offset position set to generate N typhoon set paths.

[0059] It should be noted that principal axis extraction and covariance matrix construction refer to the process of resolving the semi-major axis length, semi-minor axis length, and rotation angle of the principal axis relative to true north from the parameters of the forecast error ellipse, and then calculating a 2x2 covariance matrix based on these geometric parameters. This covariance matrix mathematically defines the forecast uncertainty of the typhoon center location in the east-west and north-south directions, as well as the correlation between the two. Furthermore, the major and minor axes of the ellipse are geometric characteristic parameters of the forecast error ellipse. The length of the major axis represents the direction and degree of maximum uncertainty in the typhoon location forecast, while the length of the minor axis represents the degree of uncertainty in the vertical direction. Together, they quantify the possible distribution range of the typhoon's future location.

[0060] It should be noted that the covariance matrix is ​​a symmetric positive definite matrix, and its expression is usually as follows: ,in and These represent the forecast error variances in the east-west and north-south directions, respectively. This is represented as the correlation coefficient between the errors in the two directions. This matrix is ​​the core parameter for subsequent multidimensional normal random sampling, ensuring that the generated random disturbances are consistent with the forecast uncertainty in statistical characteristics.

[0061] It should be noted that random perturbation sampling is performed using fused Monte Carlo methods. Based on the covariance matrix, the initial center position of the typhoon path (usually the starting point of the deterministic forecast path or the position at the previous moment) is sampled N times using a multidimensional normal distribution. Each sampling generates a two-dimensional random offset vector, which is then superimposed on the initial center position to obtain a random perturbation position that considers forecast uncertainty. This process is repeated N times to generate N random offset positions, forming a set representing the N discretized probabilities of the typhoon's future possible center position. Furthermore, the set of random offset positions is a set of N two-dimensional spatial points, each representing a positional deviation from the starting point or key node of a possible typhoon path. This set forms the basis for generating diverse paths, and its distribution is controlled by the covariance matrix, closely surrounding the initial center position and conforming to the shape and direction of an error ellipse.

[0062] It should be noted that the dynamic smoothing constraint algorithm comprises two core steps: temporal smoothing and terrain constraint correction. Temporal smoothing applies physical constraints (such as typhoon movement inertia and maximum feasible wind speed) to correct location sequences generated by random perturbations that may exhibit abrupt changes or discontinuities. Terrain constraint correction utilizes terrain elevation data from a comprehensive geographic information database to adjust the positions of pathpoints that may move to unreasonably high-altitude areas (such as mountains).

[0063] Furthermore, one approach to time-series smoothing is to apply a combination of Gaussian filtering and moving average filtering. First, Gaussian filtering is applied to the position sequence to reduce high-frequency fluctuations; its mathematical expression is: in, Represented as original Position coordinates at that moment It is the smoothed position. It is the radius of the time window. yes and Spatial distance between them This is a parameter that controls the smoothing intensity. Subsequently, the moving speed is calculated based on the smoothed position sequence, and a moving average filter is applied to the speed sequence to ensure that the speed changes are gradual and consistent with the dynamic characteristics of typhoon movement. The moving average formula is: in, yes The instantaneous speed of the typhoon at any given moment. Represented as The typhoon speed after time-smoothing Represented as the moving window index, , Represented as the moving average window size, Indicates in The speed of the typhoon at any given moment.

[0064] Furthermore, this operation is essentially a moving window averaging. For each time point in the velocity time series, the above process is repeated, generating a new velocity series that is smoother overall and has lower volatility. This effectively filters out short-term random fluctuations or noise that may exist in the original velocity, forcing velocity changes to be more continuous and gradual, conforming to the physical constraint that velocity changes during typhoon movement should not be abrupt due to atmospheric dynamic inertia. The size of the moving average window is a key parameter, and its value needs to be reasonably set according to the duration of the typhoon data and the frequency of noise to be filtered out: the larger the value of the moving average window, the stronger the smoothing effect, but the slower the response to changes in velocity trends; the smaller the value of the moving average window, the weaker the smoothing effect, but the faster it reflects the true changes in velocity. This calculation ensures that the input (original velocity series) and the output (smoothed velocity series) have consistent velocity dimensions, and there are no dimensional contradictions.

[0065] Furthermore, the terrain constraint correction process is as follows: for the smoothed path points... Query the terrain elevation of its corresponding location. .like ( This is expressed as the preset feasible altitude threshold for typhoons, in meters. This parameter is set based on physical constraints maintained by the typhoon structure (such as boundary layer friction and orographic lifting effects). (For example, based on physical constraints maintained by the typhoon structure). Then, based on terrain gradient information, the location of this point is corrected by an offset along the downhill direction. Offset Calculation and This is proportional to the corrected path point, ensuring it lies within the physically reachable geographical area of ​​the typhoon. Further adjustments are made using the following formula: At time Typhoon center coordinates after terrain constraint correction , which are the final output pathpoints that satisfy the terrain physical constraints; where The unit vector (i.e., a vector of length 1) representing the normalized correction direction is determined by the terrain gradient and usually points in the direction of the steepest descent (downhill direction). It is a dimensionless directional parameter. , This is the proportionality coefficient. Indicates the calculation of elevation Exceeding the threshold The value of the part that does not exceed the limit is 0. This ensures that the correction is triggered only when the waypoint is at an "excessively high" altitude.

[0066] It should be noted that the N physically reasonable typhoon paths are the final output. Each typhoon path is a time series, including the typhoon's center position (longitude and latitude), movement speed, direction of movement, and intensity indicators (such as minimum central pressure and maximum wind speed) at specified time intervals within a future set time period. These paths mathematically satisfy the statistical distribution of forecast errors and physically satisfy the basic dynamic laws of typhoon movement and terrain constraints, providing high-quality and reliable input for subsequent multi-scenario disaster simulations.

[0067] This application introduces dynamic smoothing constraints into Monte Carlo random perturbations to generate physically reasonable typhoon ensemble paths (such as continuous paths and velocity direction changes conforming to atmospheric motion laws), rather than relying solely on statistical perturbations. This significantly improves the physical realism and representativeness of the ensemble paths, avoids the interference of "non-physical paths" on risk assessment in traditional random generation methods, and makes subsequent disaster simulations closer to the actual typhoon evolution characteristics, thereby enhancing the reliability of the assessment results.

[0068] S3. Based on the typhoon ensemble path and the comprehensive geographic information database, generate dynamic precipitation field sequences for each typhoon path.

[0069] In a specific example, the step of generating dynamic rainfall field sequences for each typhoon path based on the typhoon ensemble path and a comprehensive geographic information database includes: S501, calculating the radius of influence of the typhoon at future times based on the typhoon ensemble path and its intensity sequence.

[0070] S502, Based on the terrain data in the comprehensive geographic information database, extract the terrain elevation and terrain slope of each spatial grid point within the area defined by the radius of the typhoon's influence range.

[0071] S503, based on the minimum air pressure and maximum wind speed at the typhoon center in the intensity sequence, a parameterized typhoon rainfall model is used to calculate the baseline rainfall rate in the core area of ​​the typhoon.

[0072] S504 calculates the terrain amplification effect based on the Euclidean distance between spatial grid points and the typhoon center, topographic elevation, and topographic slope.

[0073] S505, based on the baseline rainfall rate and the terrain amplification effect, calculate the actual rainfall intensity of spatial grid points at future times, and generate a dynamic rainfall field sequence with the typhoon path.

[0074] It should be noted that calculating the radius of the typhoon's influence area is a crucial step in determining the spatial distribution boundary of typhoon rainfall. This operation is based on the relationship between typhoon intensity (such as maximum wind speed) and wind field structure. The process of calculating the radius of the typhoon's influence area is as follows: First, obtain the maximum wind speed of the typhoon at time t, in meters per second; then, multiply the maximum wind speed by the power of B (the exponent B is a dimensionless exponent fitted based on historical typhoon observation data) by an empirical coefficient A (also fitted based on historical data, with dimensions of kilometers per second to the power of B), thus obtaining the radius of the typhoon's influence area in kilometers. In this formula, the radius of the typhoon's maximum wind speed exhibits a power-law relationship with its maximum wind speed, meaning that the higher the wind speed, the wider the area affected by strong winds and accompanying heavy rainfall may be, but the growth is non-linear. Calculating this radius is to efficiently and accurately select the spatial grid points for which rainfall intensity needs to be calculated in subsequent steps, avoiding unnecessary calculations for areas far from the typhoon and unaffected. Furthermore, the radius of the influence area is a scalar value that varies with time, representing the radius of the circular area that the typhoon's core heavy rainfall may cover at time t. This value serves as a threshold for spatial filtering, used to define the "path impact area," which is the core area for rainfall calculation.

[0075] It should be noted that extracting terrain elevation and slope is to obtain the underlying physical parameters required for rainfall calculation. The specific process is as follows: First, based on the radius of the typhoon's influence range and the typhoon's center position C(t), a circular area is determined with the typhoon's center as the center and a radius equal to the radius of the typhoon's influence range multiplied by a safety factor (the safety factor is usually ≥1). Then, spatial queries are performed on the terrain data within this circular area to obtain the elevation values ​​of each regular grid point, i.e., the terrain elevation, in meters. Next, based on the elevation data within this area, the terrain slope at each grid point is calculated using a spatial difference algorithm (such as the third-order inverse distance squared weighted difference method), and its value is the elevation change rate, in percentage or degrees. Further, based on the digital elevation model data, the process of calculating the terrain slope at each grid point using the third-order inverse distance squared weighted difference method is as follows: For the target grid point to be calculated, a 7×7 pixel neighborhood window centered on that point is first determined, covering the central grid point and a total of 49 elevation data points around it. Subsequently, the elevation values ​​of all neighboring grid points within the window are extracted, and a weight is calculated for each neighboring grid point. This weight is the reciprocal of the square of the Euclidean distance from the grid point to the central grid point; that is, the closer the neighboring point, the greater its weight, reflecting the principle that nearby points contribute more to the slope calculation. Next, the east-west slope component is calculated: using a weighted averaging method, the elevation values ​​of neighboring points within a certain angular tolerance range on the east and west sides, along with their corresponding weights, are used to calculate the weighted average elevation on the east and west sides respectively. Then, the weighted average elevation on the east side is subtracted from the weighted average elevation on the west side, and then divided by twice the product of the average number of pixels and the spatial resolution to obtain the east-west elevation change rate. Similarly, the north-south slope component is calculated: the weighted average elevation on the north and south sides is calculated using a weighted averaging method, and the weighted average elevation on the north side is subtracted from the weighted average elevation on the south side, and then divided by twice the product of the average number of pixels and the spatial resolution to obtain the north-south elevation change rate. Finally, the total slope value is synthesized: add the square of the elevation change rate in the east-west direction to the square of the elevation change rate in the north-south direction, and then take the square root to obtain the slope gradient amplitude; if expressed as a percentage, multiply the amplitude by 100%, which is the change in vertical height when advancing 100 units horizontally.

[0076] It is important to note that calculating the baseline rainfall rate using a parametric typhoon rainfall model is the core step. This operation estimates the rainfall intensity that might occur in the typhoon's core region (such as near the eyewall) under ideally flat underlying surface conditions, based on the typhoon's own dynamic and thermodynamic structural parameters (central minimum pressure and maximum wind speed). The process of calculating the baseline rainfall rate is as follows: First, calculate the difference between the ambient pressure and the typhoon's central minimum pressure; then, multiply this difference by the near-center maximum wind speed; finally, multiply the product by an empirical coefficient for water vapor condensation efficiency to obtain the baseline rainfall rate. Furthermore, the baseline rainfall rate is a time-varying scalar, representing the fundamental and characteristic rainfall intensity that the typhoon system itself can produce in its core region at time t, without considering topographical influences. It serves as the benchmark for subsequent calculations of the actual rainfall intensity at various points in space.

[0077] It should be noted that calculating the topographic amplification effect is crucial for reflecting the spatial heterogeneity of rainfall. This operation comprehensively considers the positional relationship of grid points relative to the typhoon, topographic elevation, and topographic slope. The process of calculating the topographic amplification effect is as follows: first, the exponential decay factor of the distance between the grid point and the typhoon center is calculated; then, the linear amplification factor based on topographic elevation and slope is calculated; finally, the two factors are multiplied to obtain the comprehensive topographic amplification effect coefficient. Furthermore, the topographic amplification effect is a dimensionless coefficient that quantifies the modulating effect of topography and spatial location relative to the baseline rainfall rate of the typhoon core. It transforms the uniform baseline rainfall rate into a potential rainfall field with spatially differentiated structures.

[0078] In a specific example, the calculation of the actual rainfall intensity at future time points for spatial grid points based on the baseline rainfall rate and the topographic amplification effect, and the generation of a dynamic rainfall field sequence with respect to the typhoon path, includes: using a calculation formula... The result is the The first typhoon path Each spatial lattice point at time... Actual rainfall intensity ,in Represented as time The baseline rainfall rate, Represented as the first Each spatial lattice point at time... The terrain amplification effect.

[0079] It should be noted that calculating the actual rainfall intensity is a comprehensive integration step. This operation multiplies the baseline rainfall rate, representing the typhoon's own intensity, by the topographic amplification effect, representing local modulation, to obtain the actual rainfall intensity. This ensures dimensional consistency: the baseline rainfall rate (mm / h) multiplied by the dimensionless amplification coefficient still yields the rainfall intensity (mm / h). Furthermore, the actual rainfall intensity is spatiotemporal field data, quantifying the specific rainfall intensity experienced by various spatial locations at different times in the future under the typhoon's path scenario. It serves as a crucial bridge connecting the typhoon's path and the subsequent hydrological response.

[0080] It should be noted that generating a dynamic rainfall field sequence is a wrapper and loop of a complete rainfall simulation process for a typhoon ensemble path. Specifically, it involves generating a dynamic rainfall field sequence for each discrete moment within the future time window covered by the path. (At intervals of Δt), the actual rainfall intensity is repeatedly calculated to obtain the rainfall intensity field of all grid points within the entire affected area at that moment. Arranging the rainfall intensity fields at all moments in chronological order creates a spatiotemporally continuous, three-dimensional (longitude, latitude, time) data cube, which is the dynamic rainfall field sequence. This operation ensures strict synchronization and binding between the rainfall field and the typhoon path in time and space. Furthermore, the dynamic rainfall field sequence is the final output, fully depicting the spatially continuously changing rainfall scenario corresponding to a possible typhoon path over a period of time.

[0081] S4. Parallel simulation of the typhoon ensemble path is performed using a hydrological and hydrodynamic model to generate a disaster evolution result set.

[0082] In a specific instance, the method of using a hydrological and hydrodynamic model to simulate the typhoon ensemble path in parallel to generate a disaster evolution result set includes: S701, based on the physical characteristics of the target watershed and historical hydrological data, initializing a hydrological and hydrodynamic-slope stability coupled model to obtain the complete set of initial state variables of the model at the start of the simulation.

[0083] S702 defines an inheritable subset of state variables and a subset of state variables that need to be reset, based on the principle of independence between the physical response time scale of state variables and the typhoon path.

[0084] S703 deploys a parallel computing framework to allocate independent computing resources to each typhoon path in the typhoon ensemble path in order to initiate parallel simulations.

[0085] S704 performs a state inheritance hot start evolution operation for each typhoon path and outputs a disaster evolution result set for each typhoon path.

[0086] It should be noted that initializing the hydrological-hydrodynamic-slope stability coupled model is the process of constructing a numerical model of the physical mechanisms that can simulate rainfall runoff formation, surface water infiltration and confluence, river hydrodynamic evolution, and slope stability changes within the watershed. This operation is based on geographic information data such as the target area's digital elevation model, soil type distribution, land use, and river cross-sectional morphology, as well as historical hydrological observation data. It is accomplished by solving a series of governing equations (such as the Saint-Venant equations, the Richards equations, and the slope safety factor formula) and setting their initial conditions. Its purpose is to establish a digital twin that can realistically reflect the watershed's hydrological cycle and geomechanical processes, providing a computational foundation for subsequent disaster simulations under different typhoon rainfall scenarios.

[0087] It should be noted that the complete set of initial state variables is the set of values ​​for all internal variables of the coupled model at the start of the simulation. It characterizes the initial hydrological and geological conditions of the watershed before the arrival of the typhoon, such as the volumetric water content of each soil layer, the depth of the groundwater level, the initial water level and flow rate at each cross-section of the river channel, and the stress state of each unit of the slope. The complete set of initial state variables serves as the common starting benchmark for the entire ensemble simulation, and its accuracy directly affects the reliability of the simulation results for all paths.

[0088] It should be noted that the inheritable subset of state variables and the subset of state variables to be reset, defined based on the principle of independence between physical response time scales and paths, are the core preprocessing steps of the state inheritance hot start method. (The inheritable subset of state variables includes variables that respond slowly to short-term rainfall changes, while the subset of state variables to be reset includes variables that are highly sensitive to the current rainfall event and are independent of each other under different typhoon path scenarios.) The response times of different physical processes in the watershed vary greatly. The changes of some state variables (such as deep soil moisture content, groundwater level, and main channel baseflow) are mainly affected by long-term climate and previous rainfall. Their response to short-term heavy rainfall in a single typhoon event is slow, and they do not change much during a typhoon process simulation. Moreover, their values ​​are highly correlated under different but similar typhoon path scenarios. Therefore, they can be inherited from the final state of the preceding path. Other state variables (such as surface saturation depth, surface runoff, and instantaneous water level and flow rate at river cross-sections) respond directly and rapidly to the intensity and spatial distribution of rainfall at the current moment. Their values ​​are highly dependent on the specific rainfall sequence of the current path and are independent of each other between different paths. Therefore, they need to be reset at the beginning of each new path to eliminate the "memory" of rainfall from the previous path and ensure the independence of each typhoon path simulation.

[0089] It should be noted that the inheritable subset of state variables is a collection of state variables whose elements are slowly varying state quantities that lag in response to short-term forcing (such as typhoon rainfall lasting several hours). For example, soil volumetric water content (depth > 50 cm) characterizes the amount of water stored in the soil, and its changes are controlled by long-term processes of deep infiltration and evapotranspiration; groundwater level reflects the pressure head of shallow aquifers, and its fluctuation period is much longer than that of a typhoon; the baseflow of the main river channel is the stable runoff mainly formed by groundwater recharge. Inheriting the values ​​of these variables essentially transfers the slowly varying physical information accumulated in the previous path simulation, reflecting the watershed's early wet state, to subsequent paths. This avoids simulating each typhoon path from the exact same "dry" or "wet" initial conditions, thus more realistically reflecting the dynamic evolution of the watershed state under the continuous influence of typhoons.

[0090] It should be noted that the subset of state variables to be reset is a set of state variables whose elements are transient state quantities that are highly sensitive to instantaneous forcing. For example, the meaning of surface saturation depth is the depth at which the soil below the surface reaches saturation, which directly depends on the current rainfall intensity and duration; surface runoff is the surface water flow generated under the infiltration or saturation runoff generation mechanism, which is entirely driven by the current rainfall event; the water level and flow rate of the river section are the result of hydrodynamic wave propagation, which are directly related to the current rainfall input and upstream water inflow. Resetting these variables to background values ​​(such as the river baseflow level corresponding to the stable groundwater level) or zero values ​​(such as no surface runoff) is to ensure that the simulation of each new path starts from a "calm" instantaneous state unaffected by the current path's rainfall, so that the disaster evolution results (such as flood peaks) of each typhoon path are determined only by the rainfall sequence of that typhoon path, ensuring the statistical independence and comparability of the results of different scenarios in the ensemble simulation.

[0091] It should be noted that deploying a parallel computing framework provides the computing environment support for simultaneous or concurrent simulation of various typhoon paths. This operation utilizes high-performance computing clusters or cloud computing resources to create an independent model instance process or thread for each typhoon path and allocate corresponding computing cores and memory resources. The aim is to overcome the bottleneck of excessively long simulation times in traditional serial simulations (calculating one path at a time). Through parallel processing, the simulation time, which originally required linear accumulation, is significantly compressed, making it possible to complete detailed physical simulations of hundreds of paths within the emergency decision-making time window, thereby enabling rapid and comprehensive risk scenario set analysis.

[0092] It should be noted that for each typhoon path, a state inheritance hot-start evolution operation is performed: For the first typhoon path (i=1), the complete set of initial state variables is used as its simulated complete initial state. For subsequent typhoon paths (i>1, i is an integer and 2≤i≤N), based on the state inheritance hot-start method, its initial state is constructed, specifically: from the final state simulated by the previous typhoon path, the values ​​corresponding to the subset of inheritable state variables are extracted and retained; at the same time, the values ​​corresponding to the subset of state variables that need to be reset are set to preset background values ​​or zero values. Subsequently, the coupled model is driven to perform evolution calculations using the dynamic rainfall field sequence generated by S3 and strictly bound to path Pi as external forced input, outputting the disaster evolution result set under the typhoon path, and thus outputting the disaster evolution result set under each typhoon path.

[0093] Furthermore, constructing the initial state is a specific implementation of the state inheritance hot-start method. This operation is a data synthesis process, and its synthesis mathematical formula is as follows: in Function representation from the complete set of states Extract the subset of inheritable state variables and the corresponding variable values; The function indicates that a subset of state variables needs to be reset. All variables are set to preset default values ​​(background value or zero value); This represents the union of two subsets, which are then recombined into a complete new initial state. It preserves the cumulative effect of slow-changing processes while eliminating the instantaneous dependence of fast-changing processes on the previous path, providing a reasonable and logically independent starting point for simulating the current typhoon path.

[0094] It should be noted that the core simulation calculation for each typhoon path is the evolution of the driving model using dynamic rainfall field sequences as input. This operation inputs the constructed initial state and the spatiotemporally high-resolution rainfall forcing data specific to the typhoon path into the coupled model, and solves the governing equations step by step through numerical integration methods (such as finite difference and finite volume methods), advancing the model state from the initial time to the forecast end time. During this evolution process, the hydrological processes (runoff generation and confluence), hydrodynamic processes (flood evolution), and slope stability processes (safety factor calculation) within the model interact, dynamically generating the spatiotemporal distribution field of disaster indicators such as inundation depth, flow velocity, and landslide risk index.

[0095] It should be noted that the disaster evolution result set is a collection of all disaster-related data output after the simulation of a single typhoon path is completed. It is a multi-dimensional spatiotemporal data field that includes specific indicators such as inundation depth (meters), flow velocity (meters / second), and landslide risk index (dimensionless) at different times for each grid point or assessment unit during the simulation period.

[0096] S5. Based on the disaster evolution result set, generate a joint probability distribution map of disasters and a spatial distribution map of disaster chain triggering index.

[0097] In a specific instance, the step of generating a joint probability distribution map of disasters and a spatial distribution map of disaster chain triggering index based on the disaster evolution result set includes: S801, performing multi-scenario result integration processing based on the disaster evolution result sets to obtain a spatiotemporally punctuated multi-disaster scenario dataset.

[0098] S802, based on the multi-hazard scenario dataset, perform exceedance probability statistical calculation on the preset single hazard intensity threshold to obtain the single hazard factor exceedance probability distribution field at each time point.

[0099] S803, based on the single disaster factor exceedance probability distribution field, perform joint probability calculation on exceedance threshold events of at least two disaster factors to obtain the multi-disaster factor joint exceedance probability distribution field at each time point.

[0100] S804, based on infrastructure network, disaster-bearing body data, single disaster factor exceedance probability distribution field and the multi-disaster factor joint exceedance probability distribution field, calculates the disaster chain triggering index at each time point through a comprehensive risk function.

[0101] S805, based on the joint transcendence probability distribution field of the multi-hazard factors and the disaster chain triggering index, generate and output the joint probability distribution map of the disaster and the spatial distribution map of the disaster chain triggering index.

[0102] It should be noted that the multi-scenario result integration processing is a structured aggregation operation performed on the disaster evolution result sets output by the S4 parallel simulation. This operation aligns and reorganizes the spatiotemporal field data such as inundation depth, flow velocity, and landslide risk index contained in each result set according to a unified spatial grid and time series, forming a three-dimensional (scenario, spatial, and temporal) data cube. This cube represents a complete set of various possible disaster consequences caused by typhoon uncertainties and serves as the data foundation for subsequent probability statistics and risk diagnosis.

[0103] It should be noted that the exceedance probability distribution field of the single disaster factor is a spatiotemporal field data that quantifies the probability that the intensity of a certain disaster will exceed a preset critical value at a specific time and geographical location in the future. For example, if b disaster scenarios are traversed and the number d scenarios in which the inundation depth exceeds a preset upper water depth threshold at that spatiotemporal point is counted, then the exceedance probability is d / b. This value ranges from 0 to 1, and the higher the value, the greater the probability that an inundation disaster exceeding the threshold intensity will occur at that point at that time. The upper water depth threshold is a preset disaster intensity threshold, such as the critical inundation depth set according to the disaster resistance capacity of the disaster-bearing body (such as buildings and roads), and the unit is meters (m). Similarly, the exceedance probability distribution fields of other single disaster factors such as wind speed and landslide risk index can be calculated.

[0104] It should be noted that the joint exceedance probability distribution field of multiple disaster factors is also a spatiotemporal field data. Its core innovation lies in characterizing the probability that multiple disasters will simultaneously exceed their respective thresholds, which is used to assess the compound risk of concurrent or coupled disasters. For example, the calculation process of the joint exceedance probability of flooding and strong winds occurring simultaneously is as follows: simultaneously traverse b disaster scenarios, and count the number K of scenarios where the flood depth exceeds the upper water depth threshold and the wind speed exceeds the wind speed threshold at that spatiotemporal point. Then the joint exceedance probability is K / b. Quantifying the probability of compound disaster scenarios such as "wind and rain" is crucial for assessing the chain failure risk of critical infrastructure (such as transmission towers simultaneously affected by flooding and strong winds).

[0105] It should be noted that the specific process for calculating the disaster chain triggering index at each time point using a comprehensive risk function is as follows: Based on infrastructure network and disaster-bearing body data, the disaster-bearing body exposure factor and network topology vulnerability factor at each time point are extracted and quantified to obtain a standardized exposure vulnerability factor set; based on the single disaster factor transcendence probability distribution field and the multi-disaster factor joint transcendence probability distribution field, relevant disaster probability factors at each time point are extracted to obtain a standardized disaster probability factor set; for each factor in the standardized disaster probability factor set and the standardized exposure vulnerability factor set, a linear weighted fusion calculation based on preset weight coefficients is performed to obtain the disaster chain triggering index at each time point.

[0106] It should be noted that the disaster chain triggering index calculated through the comprehensive risk function is a comprehensive quantitative indicator for dynamically assessing regional chain disaster risks. This index is not a simple summation, but a multi-dimensional function that comprehensively considers the probability (hazard) of disaster-causing factors, the exposure of disaster-bearing bodies (such as population and asset density), and system vulnerability (such as infrastructure network topology characteristics). The calculation process of the disaster chain triggering index is as follows: The comprehensive calculation function receives multiple sets of input parameters, which are divided into three categories. The first category is the probability of exceeding a single disaster factor, including but not limited to the probability of occurrence of single disaster events such as flooding, strong winds, and landslides caused by typhoons. The second category is the probability of multiple disasters exceeding a combined threshold, specifically referring to the probability of occurrence of a composite event where two or more disasters (such as flooding and strong winds) simultaneously exceed their respective thresholds. The third category is disaster-bearing body-related factors, which are further divided into two categories: disaster-bearing body exposure factors, such as population size or economic output per unit area; and network topology vulnerability factors, such as betweenness centrality describing the importance of power network nodes, or connectivity describing the tightness of transportation network connections. The comprehensive calculation function weights and fuses all these input parameters to finally output a value for quantifying the potential for triggering regional chain disasters at a specific geographical location and a specific future time, namely the disaster chain triggering index.

[0107] It should be noted that the extraction and quantification of the disaster-bearing body exposure factor and the network topology vulnerability factor specifically involves: retrieving spatial data such as population distribution, land use, and asset value from a comprehensive geographic information database, and calculating the standardized exposure value (e.g., normalized population density) for each grid point. Simultaneously, based on the topological relationship data of infrastructure networks (e.g., power grids, transportation networks), calculating the topological attributes (e.g., node degree, betweenness centrality, clustering coefficient, etc.) of network nodes within or near each grid point, and standardizing them to obtain the vulnerability increment characterizing the point's vulnerability due to network dependence or hub status. All these factors are processed into standardized values ​​that are consistent with or dimensionless in terms of the probability factor's dimensions.

[0108] This application presents a state inheritance hot-start mechanism that separates inheritable states (such as deep soil water and groundwater) from states requiring reset (such as surface runoff), enabling efficient parallel computation in ensemble simulations. While maintaining the continuity of physical processes, it significantly reduces the computational overhead of ensemble simulations, solving the problem of repetitive computation in large-path simulations using traditional "cold-start" methods. This makes large-scale simulations of N≥200 paths computationally feasible and efficient. It constructs a joint exceedance probability of multiple hazard factors (such as simultaneous inundation and strong winds) and a hazard chain triggering index (HCI), elevating single-hazard assessment to a joint diagnosis of compound hazards and chain risks. This overcomes the limitations of traditional single-hazard risk assessment, identifying high-risk areas and their main hazard-causing factor combinations under multi-hazard coupling, providing quantitative evidence for accurately identifying "risk hotspots" and chain failure paths, and significantly improving the comprehensiveness and practicality of risk assessment.

[0109] S6. Generate a risk navigator based on the spatial distribution map of the disaster chain triggering index.

[0110] In a specific example, the step of generating a risk navigator based on the spatial distribution map of the disaster chain triggering index includes: S901, performing a hierarchical early warning threshold setting operation on a preset hierarchical rule base based on the spatial distribution map of the disaster chain triggering index to obtain a hierarchical early warning threshold set.

[0111] S902, based on the graded early warning threshold set and the spatial distribution map of the disaster chain triggering index, perform spatial clustering and extreme value identification operations on the disaster chain triggering index values ​​to obtain a risk intervention hotspot set.

[0112] S903, based on the aforementioned risk intervention hotspot set, disaster joint probability distribution map, and multi-hazard scenario dataset, perform correlation analysis on the main disaster-causing factor combinations and possible chain consequences to obtain an enhanced risk hotspot set.

[0113] S904. Based on the enhanced risk hotspot set, typhoon ensemble path, disaster joint probability distribution map, and disaster chain triggering index spatial distribution map, an integrated dynamic visualization rendering operation is performed on the spatiotemporal geographic information system platform to obtain a risk navigator.

[0114] It should be noted that the tiered early warning threshold set is a set of multiple critical values ​​obtained by dividing the disaster chain triggering index range based on a preset tiered rule base. The tiered rule base is a standard comprehensively formulated based on historical disaster data, emergency response levels, and the vulnerability of disaster-bearing entities. For example, the disaster chain triggering index is divided into four intervals: [0, 0.3), [0.3, 0.6), [0.6, 0.8), and [0.8, 1.0], corresponding to four early warning levels: blue (attention), yellow (early warning), orange (risk), and red (high risk), respectively. The resulting tiered early warning threshold set is {0.3, 0.6, 0.8}. This operation quantifies the continuous disaster chain triggering index risk into discrete, easily understood early warning levels, providing a quantitative criterion for risk spatial differentiation.

[0115] It should be noted that performing spatial clustering and extreme value identification on the disaster chain triggering index values ​​refers to identifying and extracting continuous spatial regions that meet the highest warning level (e.g., disaster chain triggering index ≥ 0.8) in the spatial distribution map of the disaster chain triggering index based on the aforementioned graded early warning threshold set. These spatially adjacent high-value regions are then aggregated into one or more independent "hotspot" regions using a spatial clustering algorithm. Furthermore, the significance of spatial clustering and extreme value identification is to automatically locate risk clusters, avoid misjudging isolated high-value points, and ensure that the identified hotspot regions have significant spatial scale and continuity, meeting the regional requirements for emergency resource allocation and key deployment.

[0116] It should be noted that the risk intervention hotspot set is a collection of one or more spatial regions (polygons) identified by the above operations and their corresponding risk levels (such as Level 1 risk). Each hotspot region is a concentrated spatial manifestation of the potential to trigger a disaster chain and is a geospatial target that emergency command needs to prioritize and intervene in.

[0117] It should be noted that the correlation analysis operation performed on the main disaster-causing factor combinations and their potential chain consequences is as follows: For each hotspot area in the risk intervention hotspot concentration, firstly, based on the disaster joint probability distribution map, analyze which disaster factors (such as flooding and strong winds) in that area have a significantly higher joint exceedance probability than other areas, thereby determining the main disaster-causing factor combinations for that hotspot area (e.g., "flooding and strong winds caused by heavy rainfall"). Then, based on the multi-hazard scenario dataset, retrieve which critical infrastructure (such as substations and transportation hubs) functional failures or performance degradation events were actually triggered by the main disaster-causing factor combinations in the area during the simulated disaster scenarios, and analyze how these primary failure events, through the network topology relationships constructed in step S1, trigger subsequent chain reaction consequences (e.g., "substation failure" leads to "regional power outage," and "regional power outage" in turn affects "water pumping station operation"). This correlation analysis operation aims to concretize the abstract high-risk numerical values ​​(disaster chain triggering index) into clear and understandable disaster chain development scenarios and potential impacts.

[0118] It should be noted that the enhanced risk hotspot set is an upgraded dataset based on the original risk intervention hotspot set, which adds a description of the corresponding main disaster-causing factor combination and a description of the possible chain consequences scenario for each hotspot area. It expands the static risk "point" information into dynamic risk "scenario" information that includes disaster causes and evolution paths, providing decision-makers with a complete cognitive chain of "what the risk is, why it will occur, and what consequences it may cause".

[0119] It should be noted that performing integrated dynamic visualization rendering on a spatiotemporal geographic information system platform refers to overlaying the enhanced risk hotspot set (labeled with risk levels, dominant factors, and chain consequences), typhoon ensemble paths (displayed as multiple possible paths), joint disaster probability distribution map (in the form of probability isosurfaces or color patches), and disaster chain triggering index spatial distribution map (rendered with hierarchical colors) onto an interactive geographic information system visualization interface, according to a unified time base and spatial coordinate system. Furthermore, the integrated dynamic visualization rendering operation supports time-series playback, dynamically displaying the movement of typhoon paths, the evolution of the disaster probability field, and the emergence and changes of risk hotspots, thus realizing the transformation from data to intuitive understanding.

[0120] It should be noted that the Risk Navigator is the final output product of the aforementioned integrated dynamic visualization operation. It is a comprehensive decision support interface that integrates multi-source prediction and diagnostic results, and has spatiotemporal dynamic display and interactive analysis functions. As a "digital sand table" serving emergency command, it enables decision-makers to intuitively, quickly, and comprehensively grasp the dynamic risk pattern, key risk areas, and their causes and potential consequences throughout the entire process of typhoon disasters, thereby supporting accurate risk assessment and emergency resource allocation.

[0121] Please see Figure 2 As shown, in the second aspect, this application provides a system for dynamic simulation and prediction of typhoon disaster losses.

[0122] The system 100 of the dynamic extrapolation and prediction method for typhoon disaster losses described in this invention can be installed in an electronic device. Depending on the functions implemented, the system 100 may include a comprehensive geographic information database construction module 101, a typhoon ensemble path generation module 102, a dynamic rainfall field sequence generation module 103, a disaster evolution result set generation module 104, and a loss extrapolation module 105. The module described in this invention can also be called a unit, referring to a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, stored in the memory of the electronic device.

[0123] In this embodiment, the functions of each module / unit are as follows: The integrated geographic information database construction module is used to acquire multi-source heterogeneous data and construct an integrated geographic information database containing infrastructure network topology relationships.

[0124] The typhoon ensemble path generation module generates typhoon ensemble paths driven by physical constraints, based on a comprehensive geographic information database.

[0125] The dynamic precipitation field sequence generation module generates dynamic precipitation field sequences for each typhoon path based on the typhoon ensemble path and a comprehensive geographic information database.

[0126] The disaster evolution result set generation module uses a hydrological and hydrodynamic model to simulate the typhoon ensemble path in parallel in order to generate a disaster evolution result set.

[0127] The loss projection module generates a joint probability distribution map of disasters and a spatial distribution map of disaster chain triggering index based on the disaster evolution result set.

[0128] The risk navigator generation module generates a risk navigator based on the spatial distribution map of the disaster chain triggering index.

[0129] The dynamic simulation and prediction method and system for typhoon disaster losses provided in this application, based on a parallel hot-start strategy of Monte Carlo ensemble path generation and state inheritance under physical constraints, significantly improves the computational efficiency of large-scale ensemble simulation and solves the bottleneck of traditional methods that struggle to balance accuracy and timeliness. By calculating the joint exceedance probability of multiple disaster factors and the disaster chain triggering index, it breaks through the limitations of isolated analysis of single disasters and can accurately characterize the compound risks of multiple disasters such as floods, landslides, and strong winds and their chain transmission effects, providing a new paradigm for risk quantification under complex disaster scenarios. The constructed "risk navigator" realizes full-element dynamic visualization from disaster-causing factors to the failure consequences of disaster-bearing bodies on a spatiotemporal GIS platform, supporting emergency command personnel to quickly locate high-risk hotspots, identify the main disaster-causing factors, and predict chain consequences.

[0130] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0131] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0132] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0133] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0134] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0135] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for dynamic extrapolation and prediction of typhoon disaster losses, characterized in that, include: S1. Acquire multi-source heterogeneous data and construct a comprehensive geographic information database containing infrastructure network topology relationships; S2. Based on a comprehensive geographic information database, generate a set of typhoon paths driven by physical constraints; S3. Based on the typhoon ensemble path and the comprehensive geographic information database, generate dynamic precipitation field sequences for each typhoon path; S4. Parallel simulation of the typhoon ensemble path is performed using a hydrological and hydrodynamic model to generate a disaster evolution result set; S5. Based on the disaster evolution result set, generate a joint probability distribution map of disasters and a spatial distribution map of disaster chain triggering index; S6. Generate a risk navigator based on the spatial distribution map of the disaster chain triggering index.

2. The method for dynamic extrapolation and prediction of typhoon disaster losses according to claim 1, characterized in that, The multi-source heterogeneous data includes real-time typhoon forecast information, high-precision geographic information data, historical disaster cases, vulnerability curves of disaster-bearing bodies, real-time monitoring data, and spatial distribution and topological relationship data of critical infrastructure.

3. The method for dynamic extrapolation and prediction of typhoon disaster losses according to claim 1, characterized in that, The construction of a comprehensive geographic information database containing infrastructure network topology relationships includes: S301, based on a pre-defined standardized rule base, performs format unification, unit conversion, and logical consistency verification on multi-source heterogeneous data to obtain a standardized dataset; S302, Based on the target geographic reference system and coordinate transformation model, the spatiotemporal attribute data in the standardized dataset are aligned with coordinate system one and spatial location to obtain an aligned dataset; S303, based on preset quality control logic, check and correct the data accuracy, timeliness and outliers in the aligned dataset to obtain a valid dataset; S304, based on a spatial database engine, integrates the effective dataset and constructs its network topology relationship based on the spatial distribution data of critical infrastructure to generate the comprehensive geographic information database containing the infrastructure network topology relationship.

4. The method for dynamic extrapolation and prediction of typhoon disaster losses according to claim 1, characterized in that, The generation of typhoon ensemble paths based on a comprehensive geographic information database and driven by physical constraints includes: S401, based on the forecast error ellipse parameters in the comprehensive geographic information database, performs principal axis extraction and covariance matrix construction to obtain the covariance matrix; S402, Based on the covariance matrix, perform random perturbation sampling on the preset initial center position of the typhoon path to obtain a set containing N random offset positions; S403, based on the dynamic smoothing constraint algorithm, performs temporal smoothing and terrain constraint correction on each position sequence in the random offset position set to generate N typhoon set paths.

5. The method for dynamic extrapolation and prediction of typhoon disaster losses according to claim 1, characterized in that, The dynamic precipitation field sequence for each typhoon path is generated based on the typhoon ensemble path and a comprehensive geographic information database, including: S501, Based on the typhoon collection path and its intensity sequence, calculate the radius of influence of the typhoon at each future moment; S502, Based on the terrain data in the comprehensive geographic information database, extract the terrain elevation and terrain slope of each spatial grid point within the area defined by the radius of the typhoon's influence range; S503, Based on the minimum air pressure and maximum wind speed at the center of the typhoon in the intensity sequence, a parameterized typhoon rainfall model is used to calculate the baseline rainfall rate in the core area of ​​the typhoon; S504 calculates the terrain amplification effect based on the Euclidean distance between the spatial grid point and the typhoon center, the terrain elevation, and the terrain slope. S505, based on the baseline rainfall rate and the terrain amplification effect, calculate the actual rainfall intensity of spatial grid points at future times, and generate a dynamic rainfall field sequence with the typhoon path.

6. The method for dynamic extrapolation and prediction of typhoon disaster losses according to claim 5, characterized in that, The process of calculating the actual rainfall intensity at future time points for spatial grid points based on the baseline rainfall rate and the topographic amplification effect, and generating a dynamic rainfall field sequence with respect to the typhoon path, includes: Through calculation formula The result is the The first typhoon path Each spatial lattice point at time... Actual rainfall intensity ,in Represented as time The baseline rainfall rate, Represented as the first Each spatial lattice point at time... The terrain amplification effect.

7. The method for dynamic extrapolation and prediction of typhoon disaster losses according to claim 1, characterized in that, The method employs a hydrological and hydrodynamic model to perform parallel simulations of the typhoon ensemble path, generating a disaster evolution result set, including: S701. Based on the physical characteristics and hydrological historical data of the target watershed, a hydrodynamic-slope stability coupled model is initialized to obtain the complete set of initial state variables of the model at the start of the simulation. S702, based on the principle of independence between the physical response time scale of state variables and the typhoon path, defines an inheritable subset of state variables and a subset of state variables that need to be reset; S703 deploys a parallel computing framework to allocate independent computing resources to each typhoon path in the typhoon ensemble path in order to initiate parallel simulations; S704 performs a state inheritance hot start evolution operation for each typhoon path and outputs a disaster evolution result set for each typhoon path.

8. The method for dynamic extrapolation and prediction of typhoon disaster losses according to claim 1, characterized in that, The generation of a joint probability distribution map of disasters and a spatial distribution map of disaster chain triggering indices based on the disaster evolution result set includes: S801, Based on the disaster evolution result sets, perform multi-scenario result integration processing to obtain a multi-disaster scenario dataset with time-space point-based processing; S802, Based on the multi-hazard scenario dataset, perform exceedance probability statistical calculation on the preset single hazard intensity threshold to obtain the single hazard factor exceedance probability distribution field at each time point; S803, based on the single disaster factor exceedance probability distribution field, perform joint probability calculation on exceedance threshold events of at least two disaster factors to obtain the multi-disaster factor joint exceedance probability distribution field at each time point; S804, based on infrastructure network, disaster-bearing body data, single disaster factor exceedance probability distribution field and the multi-disaster factor joint exceedance probability distribution field, calculates the disaster chain triggering index at each time point through a comprehensive risk function; S805, based on the joint transcendence probability distribution field of the multi-hazard factors and the disaster chain triggering index, generate and output the joint probability distribution map of the disaster and the spatial distribution map of the disaster chain triggering index.

9. The method for dynamic extrapolation and prediction of typhoon disaster losses according to claim 1, characterized in that, The risk navigator generated based on the spatial distribution map of the disaster chain triggering index includes: S901, based on the spatial distribution map of the disaster chain triggering index, performs a hierarchical early warning threshold setting operation on the preset hierarchical rule base to obtain a hierarchical early warning threshold set; S902, Based on the graded early warning threshold set and the spatial distribution map of the disaster chain triggering index, perform spatial clustering and extreme value identification operations on the disaster chain triggering index values ​​to obtain a risk intervention hotspot set; S903, Based on the aforementioned risk intervention hotspot set, disaster joint probability distribution map, and multi-hazard scenario dataset, perform correlation analysis on the main disaster-causing factor combinations and the possible chain consequences to obtain an enhanced risk hotspot set; S904. Based on the enhanced risk hotspot set, typhoon ensemble path, disaster joint probability distribution map, and disaster chain triggering index spatial distribution map, an integrated dynamic visualization rendering operation is performed on the spatiotemporal geographic information system platform to obtain a risk navigator.

10. A system for implementing the dynamic extrapolation and prediction method for typhoon disaster losses according to any one of claims 1-9, characterized in that, include: The integrated geographic information database construction module is used to acquire multi-source heterogeneous data and construct an integrated geographic information database containing infrastructure network topology relationships. The typhoon ensemble path generation module generates typhoon ensemble paths driven by physical constraints, based on a comprehensive geographic information database. The dynamic precipitation field sequence generation module generates dynamic precipitation field sequences for each typhoon path based on the typhoon ensemble path and a comprehensive geographic information database. The disaster evolution result set generation module uses a hydrological and hydrodynamic model to simulate the typhoon ensemble path in parallel in order to generate a disaster evolution result set. The loss projection module generates a joint probability distribution map of disasters and a spatial distribution map of disaster chain triggering index based on the disaster evolution result set. The risk navigator generation module generates a risk navigator based on the spatial distribution map of the disaster chain triggering index.

Images