A method and system for tropical cyclone disaster loss assessment

Abstract

The application relates to a tropical cyclone disaster loss assessment method and system, and relates to the technical field of natural disaster risk assessment. The method is executed by a computer device and comprises the following steps: acquiring historical cyclone paths, wind field parameters and social and economic statistical data; performing uniform spatial resolution and coordinate reference alignment processing on population and economic spatial distribution data; calculating a hazard intensity index P; determining a cyclone influence range, extracting population exposure PE and economic exposure G; selecting a population and economic vulnerability index set, constructing a population vulnerability index PV and an economic vulnerability index EV; adopting Poisson regression to construct a death assessment model with P, PE and PV as independent variables; and adopting gamma regression to construct an economic loss assessment model with P, G and EV as independent variables. The application realizes double-dimensional and comparable collaborative attribution analysis of historical tropical cyclone event economic loss and population death, and outputs the relative contribution degrees of hazard, exposure and vulnerability to the two types of losses.

Description

Technical Field

[0001] This application relates to the field of natural disaster risk assessment technology, and in particular to a method and system for assessing losses from tropical cyclone disasters. Background Technology

[0002] Tropical cyclones (such as typhoons and hurricanes) are among the most destructive natural disasters globally. The extreme winds, torrential rains, and storm surges they bring pose a continuous and increasingly severe threat to the lives and property of people and the socio-economic development of coastal areas. Scientifically assessing the driving mechanisms of tropical cyclone-induced damage and accurately quantifying the contribution of each risk factor are core prerequisites for formulating and implementing effective disaster prevention and mitigation strategies and enhancing climate resilience.

[0003] Currently, research on the assessment and attribution of losses from tropical cyclone disasters has evolved theoretically from early single-factor examinations to systematic analyses that comprehensively consider multi-dimensional driving factors such as the hazard of causative factors, the exposure of affected bodies, and socioeconomic vulnerability. At the methodological level, two main analytical frameworks have emerged: one is to achieve comparable, integrated loss attribution analysis, clarifying the differences in the dominant driving factors between mortality and economic losses; the other is to attempt comprehensive assessment methods that integrate the three dimensions of hazard, exposure, and vulnerability. However, existing research largely focuses on loss simulation and prediction for future scenarios, or only conducts statistical analyses on a single dimension such as the number of deaths in historical disasters.

[0004] While existing technologies have established the theoretical direction of multi-factor analysis, significant limitations and gaps remain in practical applications. First, most methods focus only on a single dimension of loss, such as predicting future economic losses or retrospectively analyzing historical mortality. There is a lack of technical solutions that can provide a coordinated, parallel, and comparable two-dimensional attribution analysis of economic losses and mortality caused by historical tropical cyclone events within a unified methodological framework. In particular, there is a lack of technical solutions that, based on grid exposure data aligned to a unified spatiotemporal benchmark, determine the impact range using parameterized wind field model thresholds, simultaneously extract population and economic exposure levels within that range, and construct mortality and economic loss assessment models in parallel under the same hazard index. This makes it difficult to reveal the differentiated impact mechanisms of disasters on different types of disaster-affected entities. Second, existing comprehensive assessment studies invest heavily in future scenario simulations, while research on systematically attributing historical tropical cyclone disasters at global or regional scales and quantifying the contribution of various risk factors to actual losses is significantly insufficient. This results in a lack of solid, reproducible quantitative evidence for risk perception and strategy formulation based on historical disaster patterns. Finally, many studies still rely on the "hazard-exposure" two-factor model, neglecting the independent and crucial moderating role of socioeconomic vulnerability in the ultimate disaster loss. This prevents a comprehensive and profound understanding of the true causes of disaster losses. These shortcomings collectively result in the inability, on a global scale, to conduct a unified, two-dimensional, comprehensive, and comparable systematic attribution analysis of historical tropical cyclone disasters.

[0005] Therefore, there is an urgent need for a comprehensive risk assessment method and system that can overcome the above-mentioned shortcomings, integrate a complete framework of hazard, exposure and vulnerability, and be able to conduct a collaborative assessment of the economic losses and human mortality of historical disasters. Summary of the Invention

[0006] The purpose of this application is to provide a method and system for assessing tropical cyclone disaster losses, in order to solve the problems existing in the prior art.

[0007] To achieve the above objectives, the technical solution adopted in this application is as follows: This application provides a method for assessing tropical cyclone disaster losses, executed by computer equipment, including the following steps: S1. Acquire historical tropical cyclone path data, historical population mortality data, historical economic loss data, population spatial distribution data, economic spatial distribution data, and socio-economic statistical indicators for the target area, and perform resampling and alignment processing on the population spatial distribution data and economic spatial distribution data to unify spatial resolution and coordinate benchmark. S2. Based on historical tropical cyclone track data, calculate the hazard intensity index P, which characterizes the intensity of tropical cyclones; S3. Based on the historical tropical cyclone path data and combined with the wind field model, determine the influence range of the tropical cyclone. The influence range is the area where the wind speed output by the wind field model is greater than or equal to a preset threshold. Within the influence range, population spatial distribution data and economic spatial distribution data are superimposed, and population exposure PE and economic exposure G are extracted respectively. Population exposure PE is the total population within the influence range, and economic exposure G is the total economic output within the influence range. S4. From the socio-economic statistical data, select the indicator sets related to population vulnerability and the indicator sets related to economic vulnerability respectively, use the objective weighting method to assign weights to the indicators in each indicator set, and construct the population vulnerability index PV and the economic vulnerability index EV based on the weighted synthesis method respectively. S5. A first generalized linear model is constructed using Poisson regression with a log-connect function. The model is fitted with the hazard intensity index P, population exposure PE, and population vulnerability index PV as independent variables and historical population mortality data as dependent variables to obtain the mortality assessment model. The regression coefficients of the hazard intensity index P, population exposure PE, and population vulnerability index PV in the mortality assessment model are then output. A second generalized linear model was constructed using gamma regression with a log-connect function. The model was fitted with the hazard intensity index P, economic exposure G, and economic vulnerability index EV as independent variables and historical economic loss data as dependent variables to obtain an economic loss assessment model. The regression coefficients of the hazard intensity index P, economic exposure G, and economic vulnerability index EV in the economic loss assessment model were output for comparable co-attribution analysis with the mortality assessment model.

[0008] Furthermore, in S3, determining the extent of influence of a tropical cyclone includes: Based on the Holland wind field model and historical tropical cyclone track data, a continuous wind field is constructed within the center of a tropical cyclone. The area with a continuous wind speed of 17 m / s or greater at a height of 10 m for 1 minute is defined as the area of ​​influence.

[0009] Furthermore, in S2, the energy dissipation index PDI of the target tropical cyclone over its life cycle is calculated as the hazard intensity index P. The formula for calculating the Power Dissipation Index (PDI) is as follows: Among them, V MS τ is the maximum sustained wind speed; τ is the tropical cyclone life cycle.

[0010] Furthermore, in S4, the objective weighting method employs the CRITIC calculation method, the calculation steps of which include: S41. Standardize the vulnerability indicators and calculate their variability and conflict. The calculation formula is as follows: in, To address the variability and conflict among indicator variables, r ij The correlation coefficient between vulnerability indicators; S42. Calculate the weighting coefficients for each vulnerability indicator. The calculation formula is as follows: in, Weights for various indicators of economic vulnerability σ represents the weights of various indicators of population vulnerability. j This is expressed as the standard deviation of the indicator; S43. Construct the population vulnerability index PV and the economic vulnerability index EV, and calculate them using the following formulas: in, A standardized indicator of economic vulnerability; This represents the population vulnerability index after standardization.

[0011] Furthermore, before constructing the vulnerability index, the selected vulnerability indicators are standardized: For positive indicators, the following standardization formula is used: For negative indicators, the following standardization formula is used: in, A standardized index of positive indicators; A standardized index for negative indicators; The original value of the indicator; The maximum value of the indicator; The minimum value of the indicator. and The value range is [0, 1].

[0012] Furthermore, in S5, the calculation formula for the mortality assessment model is as follows: The calculation formula for the economic loss assessment model is as follows: in, This represents the expected number of deaths. The expected value of economic loss; It is the expected number of deaths The natural logarithm; It is the expected value of economic loss. The natural logarithm; For constant terms; , and These represent the hazard intensity, exposure level, and vulnerability coefficient, respectively; β represents the coefficient of each influencing factor. It is in exponential form, representing the multiple by which the dependent variable changes when the independent variable increases by 1 unit.

[0013] This application also provides a tropical cyclone disaster loss assessment system for implementing the tropical cyclone disaster loss assessment method described above. The system includes: a processor, a memory, and a computer program stored in the memory and executed by the processor. The data acquisition module is used to execute S1; The hazard intensity assessment module is used to perform S2; The exposure extraction module is used to execute S3; Vulnerability index building module, used to execute S4; The two-dimensional loss assessment module, used to perform S5, includes: Mortality assessment unit, used to build mortality assessment models; The economic loss assessment unit is used to construct an economic loss assessment model.

[0014] Furthermore, the exposure extraction module can construct a continuous wind field within the center of the tropical cyclone based on the historical tropical cyclone path data and the Holland wind field model, delineate the area of ​​influence, and calculate the total affected population and total economic output within the area of ​​influence, which are respectively used as the population exposure PE and the economic exposure G.

[0015] Furthermore, the hazard intensity assessment module can calculate the energy dissipation index (PDI) of a target tropical cyclone over its life cycle, which is used as the hazard intensity index (P).

[0016] Furthermore, the vulnerability index construction module also includes a data standardization unit, which is used to standardize the selected vulnerability indicators to distinguish between positive and negative values.

[0017] The beneficial effects of the technical solution provided in this application include at least the following: (1) This application realizes a two-dimensional synergistic attribution analysis of economic loss and population mortality by constructing a mortality assessment model based on Poisson regression and an economic loss assessment model based on gamma regression in parallel. It realizes a synergistic and comparable attribution analysis of two different types of losses in historical tropical cyclone events. In the model verification of the embodiment, the R2 (pseudo-R2) of the PG-EV model and the P-PE-PV model is 0.54, which is significantly higher than the comparative model when MP or PP is used as a risk indicator. This enables decision-makers to compare the differentiated mechanisms of the impact of disaster-causing factors, exposure and vulnerability on life safety and economic development on the same benchmark.

[0018] (2) This application integrates the three elements of hazard intensity, exposure and vulnerability into a loss assessment model, and delineates the scope of influence through the threshold range of the wind field model and performs area-weighted spatial overlay statistics on the population and economic grids, thereby realizing the quantitative simulation of the complete link of disaster-causing factors-disaster-resistance capacity, and improving the systematicness and physical completeness of loss assessment.

[0019] (3) This application quantifies the actual contribution of each driving factor to the incurred loss by fitting a statistical model, and sets the regression coefficients as follows: The model is transformed into an impact multiple for contribution quantification. With historical disaster loss data as the dependent variable, the established model is based on objective historical data and provides a quantitative basis and scientific decision support tool for disaster risk cognition, disaster prevention and mitigation strategy formulation and evaluation based on historical experience. Attached Figure Description

[0020] The accompanying drawings are provided to further illustrate the present application and form part of the specification. They are used together with the embodiments of the present application to explain the application and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart illustrating the evaluation method in one embodiment of this application; Figure 2 This is a schematic diagram of the structural framework of the system in one embodiment of this application.

[0021] Explanation of key figure labels: 10. Data Acquisition Module; 20. Hazard Intensity Assessment Module; 30. Exposure Extraction Module; 31. Wind Field Simulation and Range Delineation Unit; 32. Spatial Overlay Statistical Unit; 40. Vulnerability Index Construction Module; 41. Data Standardization Unit; 42. Weight Calculation and Index Synthesis Unit; 50. Two-Dimensional Loss Assessment Module; 51. Mortality Assessment Unit; 52. Economic Loss Assessment Unit. Detailed Implementation

[0022] 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.

[0023] In this specification, identical components are represented by the same reference numerals. The terms "front," "rear," "left," "right," "upper," and "lower" used in the following description refer to directions in the accompanying drawings, while the terms "bottom surface," "top surface," "inner," and "outer" refer to directions towards or away from a specific component. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this specification, "multiple" means two or more.

[0024] Terminology Explanation: 1. Tropical cyclone: ​​A weather system that forms over tropical or subtropical oceans with low central pressure, strong convection, and persistent cyclonic circulation. It is a major hazardous weather system causing extreme weather events such as strong winds, torrential rains, and storm surges. It is also the core hazard-causing factor described in this technical solution. Its intensity, track, and extent are key input variables for assessing disaster losses (economic losses and population mortality).

[0025] 2. Hazard: Refers to the inherent physical characteristics, intensity, and probability of occurrence of a natural or man-made disaster-causing factor that may cause adverse effects. It is an objectively existing possibility of disaster, independent of the disaster-bearing entity (such as population or assets). For tropical cyclones, their hazard is mainly determined by the cyclone's intensity, size, duration, and accompanying extreme weather (such as strong winds, torrential rain, and storm surges).

[0026] 3. Exposure: The degree to which humans, assets, economic activities, or ecosystems are within the potential impact range of a natural disaster. It is one of the core elements of disaster risk assessment. The level of exposure directly affects the scale of potential losses caused by a disaster; the higher the exposure, the greater the potential loss. Exposure in this application mainly includes population exposure and economic exposure.

[0027] Population exposure uses the total population size or density within the affected area as a quantitative indicator, which is then input into a Poisson regression model to assess the risk of population mortality.

[0028] Economic exposure uses the total amount or density of GDP in the affected region as a quantitative indicator, which is input into the gamma generalized linear model to assess direct economic losses.

[0029] 4. Vulnerability: refers to the inherent physical, social, economic, or environmental properties of a disaster-bearing entity (social system, economic sector, community, or natural environment) in response to specific disaster-causing factors, making it susceptible to damage, lacking the ability to cope, and difficult to recover.

[0030] Population vulnerability: Assessing population dynamics is a crucial foundation for understanding the health and well-being of its residents. This application primarily focuses on mortality assessment, typically quantifying it using proxy variables (such as aging population ratio, number of hospital beds, and building structure type) related to the affected area's population structure, healthcare conditions, housing quality, and early warning and emergency response capabilities.

[0031] Economic vulnerability: Assessing the degree to which an economic system is vulnerable to disasters, thus providing a key basis for evaluating its economic condition. In this application, it is mainly used for economic loss assessment, quantified by proxy variables (such as GDP per capita, the proportion of investment in disaster prevention projects) such as the development level, industrial structure, economic resilience, or infrastructure quality of the affected area.

[0032] 5. Holland Wind Field Model: The wind field model is a semi-empirical analytical model widely used to simulate the near-surface wind field distribution characteristics of tropical cyclones (typhoons or hurricanes). This model is based on gradient wind balance theory and constructs a symmetrical wind field around the center of the tropical cyclone using a simplified pressure profile formula. This application primarily uses this model to determine the influence range of tropical cyclones, thereby extracting exposure.

[0033] 6. Poisson Regression: Poisson regression is an important form of generalized linear model, specifically used for modeling and analyzing count data or the number of events. Its core assumption is that the dependent variable follows a Poisson distribution, describing the probability distribution of the number of times an event occurs within a fixed time or space. This model uses a logarithmic link function to link the linear combination of independent variables (explanatory variables) with the expected value of the dependent variable (count result), ensuring that the predicted value is non-negative. In the comprehensive assessment method and system provided in this application, Poisson regression is specifically used to construct a population mortality risk assessment model.

[0034] 7. Gamma Generalized Linear Model (Gamma GLM, log link): Used for modeling continuous, non-negative, and significantly right-skewed direct economic loss data. The Gamma Generalized Linear Model is an important type within the generalized linear model framework, specifically designed for modeling and analyzing continuous positive response variables that follow a Gamma distribution. In the integrated assessment method and system provided in this application, the Gamma Generalized Linear Model is used to construct a sub-model for direct economic loss risk assessment.

[0035] See Figure 1 A method for assessing losses from tropical cyclone disasters, executed by computer equipment, includes the following steps: S1. Acquire historical tropical cyclone path data, historical population mortality data, historical economic loss data, population spatial distribution data, economic spatial distribution data, and socio-economic statistical indicators for the target area, and perform resampling and alignment processing on the population spatial distribution data and economic spatial distribution data to unify spatial resolution and coordinate benchmark. S2. Based on historical tropical cyclone track data, calculate the hazard intensity index P, which characterizes the intensity of tropical cyclones; S3. Based on the historical tropical cyclone path data and combined with the wind field model, determine the influence range of the tropical cyclone. The influence range is the area where the wind speed output by the wind field model is greater than or equal to a preset threshold. Within the influence range, population spatial distribution data and economic spatial distribution data are superimposed, and population exposure PE and economic exposure G are extracted respectively. Population exposure PE is the total population within the influence range, and economic exposure G is the total economic output within the influence range. S4. From the socio-economic statistical data, select the indicator sets related to population vulnerability and the indicator sets related to economic vulnerability respectively, use the objective weighting method to assign weights to the indicators in each indicator set, and construct the population vulnerability index PV and the economic vulnerability index EV based on the weighted synthesis method respectively. S5. A first generalized linear model is constructed using Poisson regression with a log-connect function. The model is fitted with the hazard intensity index P, population exposure PE, and population vulnerability index PV as independent variables and historical population mortality data as dependent variables to obtain the mortality assessment model. The regression coefficients of the hazard intensity index P, population exposure PE, and population vulnerability index PV in the mortality assessment model are then output. A second generalized linear model was constructed using gamma regression with a log-connect function. The model was fitted with the hazard intensity index P, economic exposure G, and economic vulnerability index EV as independent variables and historical economic loss data as dependent variables to obtain an economic loss assessment model. The regression coefficients of the hazard intensity index P, economic exposure G, and economic vulnerability index EV in the economic loss assessment model were output for comparable co-attribution analysis with the mortality assessment model.

[0036] In this embodiment, step S1 is the data acquisition step: This step prepares the basic data for subsequent analysis. In practice, it is necessary to obtain multi-source data of the target geographic area, such as a country, province or specific river basin within a selected historical period, such as 1990-2019, as well as historical population mortality data and historical direct economic loss data corresponding to tropical cyclone events. The loss data is correlated with the tropical cyclone path data based on name matching, time window constraints and spatial consistency. These data include historical tropical cyclone track data, typically sourced from meteorological agencies, and should at least include the center location, central pressure, maximum wind speed, and their time-varying sequence for each cyclone. Furthermore, this historical tropical cyclone track data should include the parameters required for wind field models; high spatial resolution population grid data to reflect population distribution; high spatial resolution GDP grid data or asset value grid data to reflect economic wealth distribution; and regional socioeconomic statistical data, typically sourced from statistical departments or international institutions such as the World Bank. This data should cover multi-dimensional indicators that may affect disaster response and recovery capabilities, including social, economic, and demographic structures. The population grid data and GDP grid data should undergo coordinate projection unification and spatial resampling to ensure they share the same spatial resolution and coordinate reference as the wind field's influence area.

[0037] Step S2 is the hazard assessment step: The goal of this step is to quantify the physical destructive potential of each historical tropical cyclone. In practice, based on the path data obtained from S1, an index that comprehensively characterizes the overall intensity of a single cyclone is calculated as the hazard intensity index P. The calculation of this index aims to integrate key information such as wind speed and duration during the cyclone's life cycle. A corresponding P value can be calculated for each historical cyclone.

[0038] Step S3 is the exposure extraction step: The goal of this step is to quantify the scale of vulnerable bodies exposed to cyclone hazards. First, based on the wind field model parameters and historical path data obtained in S1, a parametric wind field model is used to simulate the spatial distribution of near-surface wind speeds for each cyclone. Then, areas with a sustained wind speed of 17 m / s or greater per minute at a height of 10 m are designated as hazard zones, determining the wind field impact range for each cyclone. Finally, within this impact range, population grid data and GDP grid data obtained in S1 are overlaid, and area-weighted spatial statistics are used to calculate the total population and GDP falling within or partially falling within this impact range. These are then used as the population exposure (PE) and economic exposure (G) corresponding to that cyclone, respectively.

[0039] Step S4 is the vulnerability index construction step: This step aims to quantify the vulnerability and coping capabilities of disaster-bearing entities. First, based on domain knowledge, the socioeconomic statistical data obtained from S1 are used to select indicator sets for assessing societal sensitivity to casualties and economic system sensitivity to property damage. Then, an objective weighting method is used to calculate weights for each indicator in each indicator set to eliminate subjective bias. Finally, the standardized values ​​of each indicator are linearly weighted and summed with their corresponding weights to construct the population vulnerability index (PV) and the economic vulnerability index (EV). A fixed set of PV and EV values ​​is obtained for each region.

[0040] Step S5 is the construction step of the two-dimensional loss evaluation model: The core of this step is to establish a quantitative relationship model between losses and driving factors using historical observation data. The data obtained in the preceding steps are organized into a sample set: each sample corresponds to a historical tropical cyclone event, including its hazard intensity index P, population exposure PE in the affected area, population vulnerability index PV in the affected area, and historical mortality data actually caused by that cyclone. A first generalized linear model is constructed using Poisson regression, with the natural logarithms of P, PE, and PV as independent variables and historical mortality as the dependent variable, to obtain the mortality assessment model. The regression coefficients of each variable in the mortality assessment model are then output. The impact factor of the form. Meanwhile, another sample set includes the P value for each cyclone, the economic exposure G of the affected area, the economic vulnerability index EV of the affected area, and historical direct economic losses actually caused by the cyclone. A second generalized linear model is constructed using gamma regression, with the natural logarithms of P, G, and EV as independent variables and historical economic losses as the dependent variable. This results in an economic loss assessment model, and the regression coefficients of each independent variable in the economic loss assessment model are output. The impact factor of form. These two models statistically quantify the independent contribution of each driving factor to the two types of losses, thus forming comparable co-attribution results for mortality and economic losses.

[0041] The above method achieves a two-dimensional synergistic attribution analysis of economic losses and population mortality by constructing a mortality assessment model based on Poisson regression and an economic loss assessment model based on gamma regression in parallel. It also achieves synergistic and comparable attribution analysis of two different types of losses from historical tropical cyclone events, enabling policymakers to compare the differentiated mechanisms of the impact of disaster-causing factors, exposure, and vulnerability on life safety and economic development on the same benchmark.

[0042] This method integrates the three elements of hazard intensity, exposure, and vulnerability into a loss assessment model, placing them within a unified mathematical framework. This enables quantitative simulation of the entire chain of "hazard-affected body-resistance capacity," effectively overcoming the one-sidedness of single-element assessment, improving the systematicness and physical completeness of loss assessment, and providing a technical foundation for panoramic diagnosis of disaster risks.

[0043] This method quantifies the actual contribution of each driving factor to the losses that have occurred by fitting a statistical model. Using historical disaster loss data as the dependent variable, the established model is based on objective historical data. The analysis process is standardized, transparent, and reproducible, providing a quantitative basis and scientific decision support tool for disaster risk cognition, disaster prevention and mitigation strategy formulation and evaluation based on historical experience.

[0044] In S3, determining the impact range of a tropical cyclone includes: Based on the Holland wind field model and historical tropical cyclone track data, a continuous wind field is constructed within the center of a tropical cyclone. The area with a continuous wind speed of 17 m / s or greater at a height of 10 m for 1 minute is defined as the area of ​​influence.

[0045] In this embodiment, to determine the influence range of a tropical cyclone, a continuous wind field near the center of the tropical cyclone is constructed based on the tropical cyclone path data. For each historical tropical cyclone, each time point in its path data is used as input points, and calculations are performed based on the Holland wind field model. For each time point on the path, the model can calculate the gradient wind speed at any radius outward from the cyclone center. By traversing all time points and spatially superimposing and fusing the circular or elliptical wind fields calculated at each time point according to the cyclone's movement path, and taking the spatial union of the wind speed threshold regions at each time point, a continuous spatial near-surface wind speed distribution field covering the entire life cycle and movement trajectory of the cyclone is finally constructed to reflect the complete structure of wind speed changes with distance from the cyclone center. The area with a continuous wind speed of greater than or equal to 17 m / s per minute at a height of 10 m is defined as the range affected by the tropical cyclone. This model considers factors such as air pressure, tropical cyclone intensity, latitude, and movement speed, and has the advantages of being comprehensive and objective. The specific formula is as follows: Where Vm (in m / s) is the sustained wind speed of a tropical cyclone at a height of 10m for 1 minute, as simulated by the Holland model; p is the air density; e is the base of the natural logarithm; b is a key parameter that varies with air pressure, latitude, and cyclone movement speed; ∆p (in hPa) is the pressure difference between the lowest air pressure at the cyclone center and the external ambient air pressure; bs depends on higher powers of ∆p.

[0046] Furthermore, in step S2, the analysis of the hazard intensity of tropical cyclones uses the Power Dissipation Index (PDI). This index is obtained by integrating the cube of the wind speed over time to obtain the total energy dissipation and potential damage potential of each tropical cyclone throughout its entire life cycle. Preliminary data processing includes path extraction and time conversion. This method extracts the path data of tropical cyclones intersecting with the Holland wind field and calculates the PDI index. The path data is 6-hour data, converted to seconds during calculation. The specific formula for the PDI index is as follows: In the formula, τ is the maximum sustained wind speed, in m / s; τ is the tropical cyclone life cycle, in seconds; PDI index is in m. 3 / s 2 .in, It is based on the maximum sustained wind speed recorded every 6 hours in the tropical cyclone track data.

[0047] Specifically, in step S4, the selected indicators are divided into economic vulnerability indicators and population vulnerability indicators. Economic vulnerability indicators aim to assess the extent to which an economic system is affected by disasters, providing a key perspective for understanding its economic situation. Simultaneously, population vulnerability indicators can assess the population dynamics of a country or region from multiple dimensions, serving as an important foundation for understanding the health and well-being of its residents. This dual approach ensures the comprehensiveness of vulnerability assessment (as shown in Table 1 below), with relevant data sourced from the World Bank (https: / / data.worldbank.org.cn / ). The objective weighting method in this approach uses the Criteria Importance Through Inter-criteria Correlation (CRITIC) method to calculate the weights of each vulnerability indicator and construct a vulnerability index. The CRITIC method determines weights based on the statistical characteristics of the data itself, eliminating subjective bias and ensuring the objectivity and scientific nature of weight allocation. It can effectively handle multi-indicator, multi-dimensional data, is suitable for complex comprehensive assessments, and is widely used in multi-attribute decision analysis. The CRITIC method determines weights based on the variability of indicators and the conflict between indicators, primarily measured by standard deviation. The specific calculation steps are as follows: First, the vulnerability indicators are standardized to eliminate dimensional differences. Then, the variability and conflict of the vulnerability indicators are calculated using the following formula: In the formula, To address the variability and conflict among indicator variables, The correlation coefficient between vulnerability indicators.

[0048] Finally, the weighting coefficients for each vulnerability indicator are calculated using the following formula: in, Weights for various indicators of economic vulnerability Let σj represent the weights of the various indicators of population vulnerability, and let σj represent the standard deviation of the indicators. S43. Construct the population vulnerability index PV and the economic vulnerability index EV, and calculate them using the following formulas: in, A standardized indicator of economic vulnerability; This represents the standardized population vulnerability index. Table 1: Vulnerability Indicators and Weights: In addition, the data on the intensity, exposure level, and vulnerability index of tropical cyclone hazard factors have all been standardized. The vulnerability index is divided into positive and negative indices; a higher value for a positive index indicates greater vulnerability, and a higher value for a negative index indicates less vulnerability. Standardized values ​​are scaled to between 0 and 1 using the following formula; the selected vulnerability indices are standardized before constructing the vulnerability index: For positive indicators, the following standardization formula is used: For negative indicators, the following standardization formula is used: in, A standardized index of positive indicators; A standardized index for negative indicators; The original value of the indicator; The maximum value of the indicator; The minimum value of the indicator. and The value range is [0, 1].

[0049] In step S5, this method employs Poisson regression and the Gamma Generalized Linear Model (GLM) to construct global-scale models for assessing tropical cyclone-related deaths and direct economic losses, respectively. Poisson regression is suitable for disaster death analysis because it effectively models count results. In contrast, direct economic losses are continuous, non-negative, and exhibit a significantly right-skewed distribution. The Gamma Generalized Linear Model better fits this distribution characteristic and, under a logarithmic link function, can robustly identify the relative contributions of hazard-causing factors, exposure, and vulnerability to economic losses.

[0050] For assessing the number of deaths, a first generalized linear model, i.e., the mortality assessment model, was constructed with P-PE-PV as independent variables. To assess the direct economic losses caused by tropical cyclones, a second generalized linear model, i.e., the economic loss assessment model, was constructed with PG-EV as independent variables. The conceptual formulas for both models are as follows: The calculation formula for the mortality assessment model is as follows: The calculation formula for the economic loss assessment model is as follows: in, This represents the expected number of deaths. The expected value of economic loss; It is the expected number of deaths The natural logarithm; It is the expected value of economic loss. The natural logarithm; For constant terms; , and These represent the hazard intensity, exposure level, and vulnerability coefficient, respectively; β represents the coefficient of each influencing factor. It is in exponential form, representing the multiple by which the dependent variable changes when the independent variable increases by 1 unit.

[0051] Model Validation 1. Data Extraction This method extracts the maximum daily precipitation (MP) and cumulative precipitation (PP) associated with tropical cyclones as indicators to assess the uncertainty of tropical cyclone loss models. The precipitation extraction process is as follows: First, the specific typhoon name is determined based on the latitude, longitude, and formation time recorded in the tropical cyclone track data. Since the EM-DAT database only provides disaster loss information at the country-event level and lacks unique cyclone identifiers, EM-DAT loss records are associated with tropical cyclone tracks based on three criteria: name matching, time window constraints, and spatial / country consistency. This identifies 7,923 tracks corresponding to the loss events. Second, a 500-kilometer buffer zone is constructed for each selected tropical cyclone track to define the spatial extent of the cyclone's impact. Finally, precipitation data within the intersection of the buffer zone and the tropical cyclone wind field, corresponding to the time period of the cyclone track, are extracted as tropical cyclone-related precipitation indicators.

[0052] 2. Required materials 2.1 Tropical Cyclone Track Data Global tropical cyclone trajectory data from 1990 to 2019 were used as the basis for simulating tropical cyclone wind fields and determining their impact range. This data was sourced from the International Best Tracks Archive for Climate Management (IBTrACS), which compiles tropical cyclone trajectory information from regional specialized meteorological centers, other international organizations, and independent contributors. It is recognized by the World Meteorological Organization's Tropical Cyclone Programme as an official archive and distribution resource (Peduzzi et al., 2012). (IBTrACS website is...) (https: / / www.ncei.noaa.gov / products / international-best-track-archive). This dataset provides estimates of the cyclone center location and intensity for each tropical cyclone every 6 hours throughout its entire lifespan; this example uses its fourth edition data.

[0053] 2.2 Precipitation Data This data comes from the Climate Prediction Center of the National Oceanic and Atmospheric Administration (https: / / psl.noaa.gov / data / gridded / data.cpc.globalprecip.html). This example primarily uses daily precipitation data from 1990 to 2019, with a spatial resolution of 0.5°×0.5°. This dataset covers global precipitation and has high spatial resolution, making it suitable for analyzing regional and global precipitation patterns.

[0054] 2.3 Tropical Cyclone Loss Data The tropical cyclone loss data used covers direct economic losses and fatalities caused by tropical cyclones globally between 1990 and 2019. This data comes from the Emergency Data Database (EM-DAT) (https: / / public.emdat.be / data), a free and publicly available disaster database managed by the National Centre for Epidemiology and Disaster Research (CERD) at KU Leuven, Belgium. The EM-DAT database integrates data from multiple sources, including major international organizations, insurance companies, and news media, making it one of the most widely used databases in the field of international disaster management and research.

[0055] 2.4 Population and GDP Data The annual global population data from 1990 to 2019 is derived from the GlobPOP dataset (Liu et al., 2024), with a spatial resolution of 30 arcseconds. This dataset integrates five existing population data products—Global Human Settlement Layer Population (GHS-POP), Global Rural and Urban Mapping Project (GRUMP), Grid World Population Version 4 (GPWv4), LandScan population data, and WorldPop—using cluster analysis and statistical learning methods to form a new global continuous gridded population dataset (GlobPOP) (https: / / zenodo.org / records / 10088105). The annual GDP data is based on the Purchasing Power Parity (PPP) per capita GDP gridded dataset constructed by Kummu et al. (2025). Through innovative extrapolation and downscaling methods, they generated a high-resolution global dataset with a spatial resolution of 30 arcseconds (https: / / zenodo.org / records / 16741980).

[0056] 3 Model Effects To verify the robustness of the model, this embodiment uses tropical cyclone precipitation (PP) and maximum daily precipitation (MP) as alternative indicators of the PDI index to assess their impact on mortality and economic losses (see Table 2 below). In the economic loss assessment, the pseudo-R² of the PG-EV model is 0.54, significantly higher than that of the MP-G-EV model (0.41) and the PP-G-EV model (0.49). In the mortality assessment, the pseudo-R² of the P-PE-PV model is also 0.54, significantly higher than that of the MP-PE-PV model (0.13) and the PP-PE-PV model (0.11). These results indicate that the constructed regression model for economic losses and mortality has high reliability. Furthermore, log(MP) and log(PP) replace log(P) in the above formulas to use process precipitation and maximum daily precipitation as hazard indicators for model validation.

[0057] Table 2 Comparison of validation results for each model Example 2 See Figure 2 A tropical cyclone disaster loss assessment system includes a processor, a memory, and a computer program stored in the memory and executed by the processor to implement the aforementioned tropical cyclone disaster loss assessment method. The system includes: a data acquisition module 10 for executing S1; a hazard intensity assessment module 20 for executing S2; an exposure extraction module 30 for executing S3; a vulnerability index construction module 40 for executing S4; and a two-dimensional loss assessment module 50 for executing S5, which includes: a mortality assessment unit 51 for constructing a mortality assessment model; and an economic loss assessment unit 52 for constructing an economic loss assessment model.

[0058] Specifically, the exposure extraction module 30 can construct a continuous wind field within the center of a tropical cyclone based on historical tropical cyclone path data and the Holland wind field model, delineate the area of ​​influence, and calculate the total population and economic output affected within the area of ​​influence. The calculation includes weighting and summarizing the grid cells that intersect with the area of ​​influence according to the proportion of the intersection area, which are respectively used as the population exposure PE and the economic exposure G.

[0059] In addition, the hazard intensity assessment module 20 can calculate the energy dissipation index PDI of the target tropical cyclone during its life cycle, as the hazard intensity index P.

[0060] In addition, the vulnerability index construction module 40 also includes a data standardization unit 41, which is used to perform standardization processing to distinguish between positive and negative values ​​for the selected vulnerability indicators.

[0061] In this embodiment, the data acquisition module 10 is the system's data input interface, configured to establish connections with various internal and external data sources or receive uploaded data files for executing step S1. Specifically, this module includes a data adapter and a preprocessing subunit, capable of automatically acquiring and formatting the historical tropical cyclone best path dataset (containing parameters required for wind field construction), high-resolution population spatial distribution raster data, economic (GDP) spatial distribution raster data, and a regional-level time-series database of socioeconomic statistical indicators for the target area. This module provides standardized, unified spatiotemporal benchmark input data for all downstream analyses.

[0062] The hazard intensity assessment module 20 receives path data from the data acquisition module 10 to execute step S2, which is an energy dissipation index calculation unit programmed to implement a specific physical algorithm: it reads hourly or 6-hourly path records for the entire life cycle of a single tropical cyclone and extracts the maximum wind speed near the center, V. Ms Given time t, the integral formula is calculated using a numerical integration algorithm. The calculation result, namely the energy dissipation index (PDI), is output and stored as the hazard intensity index (P) of the cyclone. This module can process all historical cyclone events in batches and generate corresponding P-value sequences.

[0063] The exposure extraction module 30 is the system's spatial analysis engine, used to execute step S3. It contains two core sub-units: The wind field simulation and extent delineation unit 31 integrates the Holland wind field model algorithm, receives path data and model parameters of a cyclone, and dynamically simulates the continuous spatial wind speed field near the ground surface (10 meters high) within the cyclone's lifespan. Subsequently, it applies a predefined wind speed threshold (≥17 m / s) to automatically extract all grids with wind speeds exceeding this threshold from the continuous wind field, generating one or more spatial polygons representing the extent of the disaster impact.

[0064] Spatial overlay statistical unit 32 performs precise spatial overlay analysis with the influence range polygon obtained in the previous step and the population raster and economic (GDP) raster provided by the data acquisition module 10. It automatically calculates and outputs the total population and total GDP of all raster cells that fall completely or partially within the influence polygon. These two values ​​are defined as the population exposure PE and economic exposure G of the cyclone, respectively.

[0065] The vulnerability index construction module 40 is the system's index synthesis processor, used to execute step S4. Its structure includes: Data standardization unit 41 first preprocesses the original vulnerability indicators selected from socioeconomic statistical indicators. It automatically distinguishes the positive and negative aspects of the indicators according to preset rules (i.e., the larger the value, the higher the vulnerability, which is positive, and vice versa), and applies the corresponding standardization algorithm (such as range standardization) to normalize all indicator values ​​to between 0 and 1 in order to eliminate the influence of dimensions.

[0066] The weight calculation and index synthesis unit 42 receives standardized indicator data. It incorporates an objective weighting algorithm (such as the CRITIC method) to automatically calculate the weights of each indicator based on the inherent variability and conflict of the indicator data. This unit performs a weighted linear summation of the standardized indicators according to the weights, outputting the population vulnerability index (PV) and the economic vulnerability index (EV) respectively. Each assessment region (such as a country or province) corresponds to a fixed set of PV and EV values.

[0067] The two-dimensional loss assessment module 50 is the core of the system's modeling and attribution process, used to execute step S5. It internally consists of two independent modeling units operating in parallel: Mortality Assessment Unit 51: This unit is configured to run a Poisson regression modeling program, taking all relevant data of historical events (P, PE, PV, historical number of deaths) as input, automatically fitting the model, generating and storing a mortality assessment model that quantifies the contribution of P, PE, and PV to the risk of death in the population.

[0068] Economic Loss Assessment Unit 52: This unit is configured to run a gamma regression modeling program, taking all relevant data of historical events (P, G, EV, historical economic losses) as input, automatically fitting the model, generating and storing an economic loss assessment model that quantifies the contribution of P, G, and EV to economic losses.

[0069] In the embodiments disclosed in this application, the terms "installation," "connection," "linking," and "fixing" should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral connection; "linking" can be a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in the embodiments disclosed in this application according to the specific circumstances.

[0070] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for assessing losses from tropical cyclone disasters, characterized in that, Performed by a computer device, including the following steps: S1. Acquire historical tropical cyclone path data, historical population mortality data, historical economic loss data, population spatial distribution data, economic spatial distribution data, and socio-economic statistical indicators for the target area, and perform resampling and alignment processing on the population spatial distribution data and the economic spatial distribution data to unify spatial resolution and coordinate reference. S2. Based on the historical tropical cyclone track data, calculate the hazard intensity index P, which characterizes the intensity of tropical cyclones; S3. Based on the historical tropical cyclone path data and combined with the wind field model, determine the influence range of the tropical cyclone, wherein the influence range is the area where the wind speed output by the wind field model is greater than or equal to a preset threshold. Within the influence range, overlay the population spatial distribution data and the economic spatial distribution data, and extract the population exposure PE and economic exposure G respectively. The population exposure PE is the total population within the influence range, and the economic exposure G is the total economic output within the influence range. S4. From the aforementioned socio-economic statistical indicator data, select the indicator sets related to population vulnerability and the indicator sets related to economic vulnerability respectively, assign weights to the indicators in each indicator set using the objective weighting method, and construct the population vulnerability index PV and the economic vulnerability index EV based on the weighted synthesis method respectively. S5. A first generalized linear model is constructed using Poisson regression with a log-connect function. The model is fitted with the hazard intensity index P, the population exposure PE, and the population vulnerability index PV as independent variables and historical population mortality data as dependent variables to obtain a mortality assessment model. The regression coefficients of the hazard intensity index P, the population exposure PE, and the population vulnerability index PV are output based on the mortality assessment model. A second generalized linear model is constructed using gamma regression with a log-connect function. The model is fitted with the hazard intensity index P, the economic exposure G, and the economic vulnerability index EV as independent variables and historical economic loss data as dependent variables to obtain an economic loss assessment model. Based on the economic loss assessment model, the regression coefficients of the hazard intensity index P, the economic exposure G, and the economic vulnerability index EV are output for comparable co-attribution analysis with the mortality assessment model.

2. The method for assessing tropical cyclone disaster losses according to claim 1, characterized in that, In S3, determining the influence range of the tropical cyclone includes: Based on the Holland wind field model and the historical tropical cyclone track data, a continuous wind field within the center of the tropical cyclone is constructed, and the area with a continuous wind speed of 17 m / s or greater at a height of 10 m for 1 minute is defined as the influence range.

3. The method for assessing tropical cyclone disaster losses according to claim 2, characterized in that, In S2, the energy dissipation index PDI of the target tropical cyclone during its life cycle is calculated as the hazard intensity index P. The formula for calculating the energy dissipation index (PDI) is as follows: Among them, V MS τ is the maximum sustained wind speed; τ is the tropical cyclone life cycle.

4. The method for assessing tropical cyclone disaster losses according to claim 1, characterized in that, In step S4, the objective weighting method employs the CRITIC calculation method, the calculation steps of which include: S41. Standardize the vulnerability indicators and calculate their variability and conflict. The calculation formula is as follows: in, To address the variability and conflict among indicator variables, r ij The correlation coefficient between vulnerability indicators; S42. Calculate the weighting coefficients for each vulnerability indicator. The calculation formula is as follows: Where is the weight of each indicator of economic vulnerability, is the weight of each indicator of population vulnerability, and σj represents the standard deviation of the indicator; S43. Construct the population vulnerability index PV and the economic vulnerability index EV, and calculate them using the following formulas: in, A standardized indicator of economic vulnerability; This represents the population vulnerability index after standardization.

5. The method for assessing tropical cyclone disaster losses according to claim 4, characterized in that, Before constructing the vulnerability index, the selected vulnerability indicators are standardized: For positive indicators, the following standardization formula is used: For negative indicators, the following standardization formula is used: in, A standardized index of positive indicators; A standardized index representing a negative indicator; The original value of the indicator; The maximum value of the indicator; The minimum value of the indicator. and The value range is [0, 1].

6. The method for assessing tropical cyclone disaster losses according to claim 1, characterized in that, In S5, the calculation formula for the mortality assessment model is: The calculation formula for the economic loss assessment model is as follows: in, This represents the expected number of deaths. The expected value of economic loss; It is the expected number of deaths The natural logarithm; It is the expected value of economic loss. The natural logarithm; For constant terms; , and These represent the hazard intensity, exposure level, and vulnerability coefficient, respectively; β represents the coefficient of each influencing factor. It is in exponential form, representing the multiple by which the dependent variable changes when the independent variable increases by 1 unit.

7. A tropical cyclone disaster loss assessment system, characterized in that, The system includes a processor, a memory, and a computer program stored in the memory and executed by the processor, for implementing the tropical cyclone disaster loss assessment method as described in any one of claims 1-6, the system comprising: The data acquisition module is used to execute S1; A hazard intensity assessment module is used to perform S2; Exposure extraction module, used to perform S3; A vulnerability index construction module is used to execute S4; A two-dimensional loss assessment module, used to perform S5, includes: A death assessment unit, used to construct the death assessment model; An economic loss assessment unit is used to construct the economic loss assessment model.

8. The tropical cyclone disaster loss assessment system according to claim 7, characterized in that, The exposure extraction module can construct a continuous wind field within the center of a tropical cyclone based on the historical tropical cyclone path data and the Holland wind field model, delineate the influence range, and calculate the total population and economic output affected within the influence range. The calculation includes weighting and summing the grid cells intersecting with the influence range according to the proportion of the intersection area, which are respectively used as the population exposure PE and the economic exposure G.

9. The tropical cyclone disaster loss assessment system according to claim 7, characterized in that, The hazard intensity assessment module can calculate the energy dissipation index (PDI) of the target tropical cyclone during its life cycle, which is used as the hazard intensity index P.

10. The tropical cyclone disaster loss assessment system according to claim 7, characterized in that, The vulnerability index construction module also includes a data standardization unit, which is used to perform standardization processing to distinguish between positive and negative values ​​for the selected vulnerability indicators.

Images