Highway meteorological disaster risk assessment method based on hazard-bearing body census evaluation system

Abstract

The application discloses a highway meteorological disaster risk assessment method based on a disaster-bearing body census evaluation system and belongs to the field of highway disaster risk assessment. In view of the problems that the existing early warning system lacks a rainfall risk assessment for highway disasters and does not combine the census evaluation, the application provides the following scheme: obtaining historical disaster damage data, rainfall grids and road network vectors, extracting daily cumulative rainfall according to the disaster damage date and calculating effective rainfall, statistically analyzing disaster damage frequency in each interval after normalization, fitting a probability density function to obtain a disaster probability; dividing the road network into grids, obtaining an inherent risk classification index according to the vulnerability evaluation of the census system; determining the weight by using the analytic hierarchy process and multiplying the probability to obtain a comprehensive risk; generating a risk density surface by kernel density estimation, and outputting a risk zoning map by the natural breakpoint method. The application is used for dynamic risk assessment and early warning management of rainfall-induced slope disasters in a highway network.

Description

Technical Field

[0001] This invention relates to the field of highway disaster risk assessment. More specifically, this invention relates to a method for assessing highway meteorological disaster risk based on a disaster-bearing body survey and evaluation system. Background Technology

[0002] Highway weather warnings have always been a challenging problem. Frequent heavy rainfall triggers highway slope disasters, posing a severe challenge to the daily disaster prevention, mitigation, and safe operation of the highway network. Furthermore, current warning systems primarily use rainfall volume as the standard for assessing disaster risk and determining warning levels, lacking a rainfall-based warning and risk assessment system specifically for highway disasters. On the one hand, existing research lacks sufficient relevance to the national highway disaster-bearing body survey and related industry operations, and a targeted vulnerability assessment system for highway disaster-bearing bodies, conforming to transportation industry management standards, has not yet been established. On the other hand, traditional methods cannot fully characterize the probabilistic relationship between effective rainfall and actual highway disaster losses, severely limiting the accuracy of regional highway slope disaster risk assessment driven by meteorological data. Moreover, existing assessment methods at the provincial level and above rarely combine the aforementioned rainfall-disaster probability relationship with standardized highway disaster-bearing body survey data, resulting in a mismatch between model outputs and the daily risk assessment, inspection, and emergency response workflows of highway management agencies. This disconnect is a key obstacle to integrating the functions of intelligent transportation systems with actual highway operation workflows. Therefore, establishing a targeted and practically applicable meteorological risk assessment method for highway slope disasters has become an urgent need for highway management. Summary of the Invention

[0003] One object of the present invention is to solve at least the above-mentioned problems and to provide at least the advantages that will be described later.

[0004] To achieve these objectives and other advantages of the present invention, a method for assessing highway meteorological disaster risk based on a disaster-bearing body survey and evaluation system is provided, comprising the following steps: Historical highway disaster data, historical precipitation raster data, and highway network vector data of the target area are acquired, and the historical precipitation raster data is preprocessed to obtain daily cumulative precipitation raster data. Based on the historical highway disaster data, the disaster occurrence date for each disaster site is determined. Based on this date, the daily cumulative precipitation sequence corresponding to each disaster site is extracted from the daily cumulative precipitation raster data. The effective precipitation for each disaster site is calculated using the effective precipitation formula. The effective precipitation is then subjected to a normalization transformation. The disaster occurrence frequency density of the historical highway disaster data within each effective precipitation interval is statistically analyzed. A probability density function for precipitation-induced highway disasters is fitted, and the probability of disaster occurrence corresponding to different effective precipitation levels is calculated based on the cumulative distribution function of the probability density function. Based on the highway network vector data, the highway network of the target area is divided into several grid units, and a static vulnerability assessment is performed on the effective grid units containing highways based on the disaster-bearing body survey and evaluation system to obtain the inherent risk classification index of each effective grid unit. Based on the inherent risk classification index of each effective grid unit, the risk weight of each effective grid unit is determined by the analytic hierarchy process (AHP). The probability of disaster occurrence is multiplied by the corresponding risk weight to obtain the comprehensive risk score of each effective grid unit. Based on the comprehensive risk score of each effective grid cell, a continuous risk density surface of the target area is generated using the kernel density estimation method, and the risk level is classified using the natural breakpoint method. Based on the highway network vector data, the visualized risk zoning results of the highway network area are output.

[0005] Preferably, the historical precipitation raster data is preprocessed, specifically including: extracting hourly precipitation data for the target date from the NetCDF dataset and summing them to obtain the daily cumulative precipitation; based on the latitude and longitude coordinates of each disaster point in the historical highway disaster data, using bilinear interpolation to calculate the precise precipitation value at the location of each disaster point, and generating the daily cumulative precipitation sequence corresponding to each disaster point.

[0006] Preferably, the effective rainfall formula in step S2 is: R c =R0+αR1+α 2 R2+…+α n R n Among them, R c The effective precipitation is R0, which is the daily precipitation on the day the disaster occurs. n Let α be the daily precipitation on the nth day before the disaster, α be the rainfall attenuation coefficient, and n be the number of days of rainfall in the preceding period.

[0007] Preferably, the effective precipitation is normalized by using the Yeo-Johnson adaptive transformation, and the optimal transformation parameter λ is solved by maximum likelihood estimation to obtain the normalized precipitation sequence.

[0008] Preferably, the frequency density of historical highway disaster data within each effective precipitation interval is statistically analyzed. Specifically, the effective precipitation sequence is divided into several equal intervals, and the ratio of the number of historical highway disaster events falling within each interval to the interval width is calculated as the frequency density of disaster occurrence corresponding to that interval. The least squares method is used to perform curve fitting on the frequency density to obtain a probability density function that follows a normal distribution.

[0009] Preferably, a static vulnerability assessment is conducted based on the disaster-bearing body survey and evaluation system. Specifically, this includes: calculating the risk assessment index for disaster-damaged points within each effective grid unit based on five indicators specified in the "Technical Guidelines for Comprehensive Risk Assessment of Highway Disaster-Bearing Bodies," namely, the frequency of disaster occurrence, the severity of historical disasters, the status of disaster response, the degree of disaster development, and the importance of the highway; determining the risk level of each disaster-damaged point based on the risk assessment index; and calculating the inherent risk classification index of each effective grid unit by weighting the risk scores corresponding to different risk levels and the number of disaster-damaged points at each level.

[0010] Preferably, the risk levels include Level 1, Level 2, Level 3, and Level 4, with corresponding risk scores of 100, 40, 15, and 6 points, respectively; the inherent risk grading index (QI) is calculated using the following formula: QI=G1F1+G2F2+G3F3+G4F4 Among them, G1, G2, G3, and G4 are the risk scores of disaster points from level one to level four, respectively, and F1, F2, F3, and F4 are the quantity indices of disaster points from level one to level four within the grid unit.

[0011] Preferably, the highway network of the target area is divided into several grid units based on the highway network vector data. Specifically, this includes: generating a regular grid covering the target area with a preset spatial resolution; identifying grid units containing highway segments as effective grid units based on the highway network vector data; and ensuring that adjacent effective grid units do not overlap. The area covered by the effective grid unit constitutes the highway network area for outputting the visualized risk zoning.

[0012] Preferably, the risk weight of each effective grid cell is determined using the analytic hierarchy process (AHP), specifically including: classifying the effective grid cells into inherent risk levels based on their inherent risk grading indices; assigning weights to grid cells with different inherent risk levels using the AHP, and verifying the scientific validity of the weight allocation through a consistency check; the formula for calculating the comprehensive risk score S is: S=P×W Where P is the probability of a road disaster induced by precipitation, and W is the risk weight calculated by the analytic hierarchy process.

[0013] Preferably, a continuous risk density surface for the target area is generated using kernel density estimation, and risk levels are classified using the natural breakpoint method. Based on the highway network vector data, a visualized risk zoning result for the highway network area is output. Specifically, the continuous risk density surface covering the entire target area is generated using a weighted kernel density estimation method with the comprehensive risk score of each effective grid unit as the weight. The continuous risk density surface is divided into four risk levels and one risk-free zone using the natural breakpoint method. Based on the highway network vector data, the risk zoning result is spatially clipped, retaining only the area covered by the highway network. Different colors are used to mark the areas of each level, realizing daily dynamic zoning and visualization of highway disaster damage risk.

[0014] The present invention has at least the following beneficial effects: The highway meteorological disaster risk assessment method constructed in this invention couples the probability density function characterizing the relationship between effective precipitation and highway disaster damage with the disaster-bearing body assessment system in the "Technical Guidelines for Comprehensive Risk Assessment of Highway Disaster-Bearing Bodies". This achieves the organic integration of quantitative calculation of the probability of precipitation-induced highway disasters and static vulnerability assessment of highway disaster-bearing bodies, filling the gap in the current highway meteorological early warning system that lacks precipitation early warning and risk assessment methods for highway disasters.

[0015] This invention combines historical highway disaster data with precipitation raster data, and uses the effective rainfall formula and Yeo-Johnson adaptive transform to establish a probability density function for precipitation-induced disasters. It can accurately characterize the probability of highway disasters corresponding to different levels of effective precipitation, effectively solving the technical problem that traditional methods cannot fully characterize the probabilistic relationship between precipitation and actual highway disaster losses, and significantly improving the accuracy of regional highway slope disaster risk assessment driven by meteorological data.

[0016] This invention expands the evaluation scope of unit road sections in the "Technical Guidelines for Comprehensive Risk Assessment of Highway Disaster-Bearing Bodies" to the grid scale. Based on highway network vector data, the target area is divided into effective grid units, realizing the efficient integration of historical disaster data at the provincial and above scales with highway disaster-bearing body vulnerability assessment. This significantly reduces the computational complexity of the model and ensures that the model output results are closely linked to the daily risk assessment, inspection, and emergency response workflow of highway management agencies. This effectively solves the problems of disconnect between scientific research and practical engineering applications and high barriers to implementation.

[0017] This invention employs the Analytic Hierarchy Process (AHP) to assign risk weights to highway grids with different inherent risk levels, incorporating the practical experience of frontline staff into the model and eliminating subjective biases from manual scoring through consistency checks. This ensures both the scientific rigor and objectivity of the assessment results while fully reflecting the actual needs of the project. Furthermore, by multiplying the probability of precipitation-induced disasters by the risk weights, a reasonable coupling of "hazard" and "vulnerability" is achieved, avoiding logical biases caused by simply multiplying the two directly.

[0018] This invention uses the weighted kernel density estimation (WKDE) method to generate a continuous risk density surface covering the entire target area, combines it with the natural discontinuity method to classify risk levels, and outputs a visualized risk zoning map of the highway network area based on highway network vector data. Different colors are used to mark the areas of each level, realizing the continuous spatial expression of discrete grid scoring and the daily dynamic zoning of highway disaster damage risk. It provides intuitive visualization materials for the accurate location of high-risk zones or road sections, and can directly serve the daily risk assessment, inspection and emergency response work of highway management agencies.

[0019] Other advantages, objectives and features of the present invention will become apparent in part from the following description, and in part from those skilled in the art through study and practice of the invention. Attached Figure Description

[0020] Figure 1 This is a flowchart of the highway meteorological disaster risk assessment method based on the disaster-bearing body survey and evaluation system described in this invention; Figure 2 This is a schematic diagram illustrating the inherent risk classification of rainfall-induced highway disasters according to an embodiment of the present invention. Detailed Implementation

[0021] The present invention will now be described in further detail with reference to the accompanying drawings, so that those skilled in the art can implement it based on the description.

[0022] This invention addresses the limitations of existing rainfall-induced highway disaster risk assessment models applied at the provincial level. Following the classic paradigm of "disaster × vulnerability = comprehensive risk," it constructs a fully automated dynamic comprehensive risk assessment system for highway networks. This method combines the probability density function characterizing the relationship between effective precipitation and highway disaster damage with a highway disaster-bearing body survey and assessment method. It can quantify the impact of multi-day precipitation on highway damage, directly calculate risk scores and levels for regional road networks and road sections, and issue real-time, dynamic early warning information.

[0023] The method flow of the present invention is as follows: Figure 1 As shown below, the specific implementation methods of each step in the highway meteorological disaster risk assessment method based on the disaster-bearing body survey and evaluation system described in this invention will be described in detail.

[0024] S1. Obtain historical highway disaster data, historical precipitation raster data, and highway network vector data for the target area, and preprocess the historical precipitation raster data to obtain daily cumulative precipitation raster data.

[0025] First, historical highway damage data, historical precipitation raster data, and highway network vector data (shp files) for the target area are acquired. The historical highway damage data is used to determine the dates of damage and to calculate the frequency of damage; the historical precipitation raster data comes from the NetCDF dataset and provides a data foundation for subsequent precipitation calculations; and the highway network vector data is used to define the assessment scope and for visualization output.

[0026] Preprocessing historical precipitation raster data is fundamental for accurate precipitation calculation. To eliminate the spatial scale mismatch between gridded precipitation data and highway disaster sites, this invention employs bilinear interpolation (also known as a regular grid interpolator) to calculate the precise precipitation values ​​at the latitude and longitude locations of each disaster site. The specific implementation process is as follows: First, hourly precipitation data for the target date is extracted from the NetCDF dataset and summed to obtain the daily cumulative precipitation. Second, using bilinear interpolation, based on the latitude and longitude coordinates of each disaster site in the historical highway disaster data, a weighted average is performed using four neighboring data points to obtain the interpolation result for the target point. This calculates the precise precipitation values ​​at the location of each disaster site, generating a daily cumulative precipitation sequence for each disaster site. This interpolation method effectively eliminates the spatial scale mismatch between gridded precipitation data and highway disaster sites, laying a data foundation for accurate calculation of effective precipitation.

[0027] S2. Determine the disaster occurrence date for each disaster point based on the historical highway disaster data. Extract the daily cumulative precipitation sequence corresponding to each disaster point from the daily cumulative precipitation raster data based on the disaster occurrence date. Calculate the effective precipitation for each disaster point using the effective precipitation formula. Perform a normalization transformation on the effective precipitation, statistically analyze the disaster occurrence frequency density of the historical highway disaster data within each effective precipitation interval, fit the probability density function of precipitation-induced highway disasters, and calculate the disaster occurrence probability corresponding to different effective precipitation levels based on the cumulative distribution function of the probability density function.

[0028] This step is the core of the invention in quantifying the "danger" of precipitation-induced disasters. This step first calculates the effective precipitation at each disaster site, and then establishes a quantitative relationship between precipitation and the probability of disaster occurrence.

[0029] 1) Calculation of effective precipitation Based on historical highway disaster data, the disaster occurrence date for each disaster site is determined. Then, the corresponding daily cumulative precipitation sequence for each disaster site is extracted from the pre-processed daily cumulative precipitation data based on this date. It should be noted that the contribution of previous rainfall to slope instability gradually weakens over time; therefore, the daily precipitation cannot be simply used as the causative factor. Instead, an effective rainfall formula is needed to quantify the combined cumulative effect of the decaying previous effective rainfall and daily rainfall on highway slope stability. This formula is widely used in agricultural irrigation management, hydrology, and water resources assessment, and its expression is as follows: R c =R0+αR1+α 2 R2+…+α n R n Among them, R c The effective precipitation is R0, which is the daily precipitation on the day the disaster occurs. n Let α be the daily precipitation on the nth day before the disaster, α be the rainfall attenuation coefficient (also known as the attenuation coefficient), which is related to factors such as evaporation intensity, surface vegetation, and soil and rock properties, and n be the number of days of rainfall in the preceding period. By combining the characteristics of standardized highway construction across the country and historical disaster patterns, unique model parameters can be determined for each region. Substituting the daily cumulative precipitation sequence for each disaster-affected point into the above formula, the effective precipitation for each disaster-affected point can be calculated.

[0030] 2) Normalization transformation of effective precipitation Some climate diagnostic methods and prediction models are built on the assumption that climate variables follow a normal distribution. However, effective precipitation data often does not conform to this distribution, and directly using the raw data for probability fitting will produce significant bias. Therefore, this invention employs the Yeo-Johnson adaptive transform to normalize effective precipitation. The Yeo-Johnson transform is an improved version of the Box-Cox transform, and its advantage lies in its ability to handle data sequences containing zero and negative values, making it more suitable for the actual distribution characteristics of precipitation data. The expression for this transform is: In the above transformation, the parameters l The optimal transformation shape parameter, whose magnitude directly determines the direction and strength of the transformation's correction to the skewness of the original data. Unlike fixed-form logarithmic or square-root transformations, the transformation in this invention... l The maximum likelihood estimation (MLE) method adaptively solves the problem from the historical effective precipitation series of each target area without requiring prior human specification.

[0031] In the application scenario of highway slope disaster early warning in this application, l The value of has both clear physical and statistical meanings. Specifically, when lWhen the value approaches 1, the transformation is approximately linear, indicating that the effective precipitation data in this region is close to a normal distribution, and the triggering effect of precipitation on slope instability varies approximately proportionally with the amount of precipitation; when l As the transform approaches 0, it approximates the logarithmic transform ln( R c +1), at this point the effective precipitation sequence exhibits a typical right-skewed distribution characteristic—that is, a large number of low-intensity precipitation events coexist with a small number of extreme rainstorm events. Logarithmic transformation can effectively compress the scale of extreme high values, allowing the model to focus on the fine differences in data distribution within the medium and low precipitation range.

[0032] In real-world highway disaster scenarios, effective precipitation sequences typically exhibit a significant right-skewed distribution, meaning that a large number of historical disasters are concentrated in the moderate rainfall intensity range, while extreme heavy rainfall samples are relatively sparse. Therefore, the effective precipitation sequence obtained by MLE (Mean Estimation Expiration) is... l The values ​​are mostly located on the smaller side of the range between 0 and 1. Taking a certain region as an example, this region has distinct dry and wet seasons, with frequent single-point rainstorms in summer, and the raw data of effective precipitation often shows a long-tailed, right-dragging distribution. Through the Yeo-Johnson adaptive transform, the MLE algorithm automatically outputs an optimal value based on the dispersion of the samples in the dataset. l Values, for example l The value ≈0.3 asymmetrically stretches the resolution of the low-to-medium precipitation range, which covers the trigger thresholds of most catastrophic events, while moderately compressing the weight of the extreme high precipitation range to avoid rare extreme values ​​dominating the fitting results. This data-driven adaptive adjustment mechanism ensures that the transformed normalized precipitation sequence... x To more accurately follow the assumption of a normal distribution, thereby ensuring that subsequent probability density functions based on the normal distribution are... P = f ( x It can accurately depict the nonlinear physical relationship between precipitation and the trigger threshold for highway slope instability, avoiding probability estimation bias caused by data distribution deviating from normality.

[0033] Furthermore, due to significant differences in climate characteristics (such as average annual precipitation and seasonal concentration of rainfall), surface vegetation cover, and soil and rock properties across different provinces and regions, the skewness and kurtosis of effective precipitation sequences vary from place to place. In this invention... l The values ​​are independently fitted using maximum likelihood estimation based on historical disaster damage data from various regions. The model automatically adapts to the optimal parameters as the target region changes, requiring no human intervention or empirical parameter tuning. This ensures the universality and adaptability of the method used in different highway climate zones across the country. In other words, lThe solution is entirely data-driven—the MLE algorithm searches for the transformation parameters that maximize the probability of occurrence of the data within a given effective precipitation sequence. This process outputs a deterministic optimal value, rather than an empirical fixed constant. The resulting... l The value is the most suitable normalization transformation parameter for the region, ensuring the statistical rationality of the subsequent disaster probability density function fitting.

[0034] 3) Fitting the probability density function of disaster occurrence To quantify the probabilistic relationship between effective precipitation and highway disaster damage, this invention divides the effective precipitation sequence into several equal intervals and statistically analyzes the frequency density of highway disasters corresponding to each interval. Specifically, the maximum value of the effective precipitation sequence is taken, and this sequence is divided into 30 equal intervals. The ratio of the number of historical highway disaster events falling within each interval to the interval width is calculated as the frequency density of disaster occurrence for that interval. The least squares method is used to fit the above frequency density to a curve, obtaining a probability density function that follows a normal distribution. in, x This represents the normalized effective precipitation after Yeo-Johnson transformation. m and s These are the location parameter (expected value) and scale parameter (standard deviation) of the normal distribution, respectively. In the highway slope disaster early warning application scenario of this invention, these two parameters are no longer merely abstract statistical variables, but regional characteristic indicators with clear engineering physical significance.

[0035] Specifically, m This characterizes the concentrated locations of highway slope damage events in the region along the dimension of precipitation. From a physical mechanism perspective, when the normalized value corresponding to effective precipitation... x Approaching m This indicates that current precipitation conditions are approaching the "critical trigger threshold" for slope instability in the area—at which point the increased water content of the slope's soil and rock leads to a decrease in matrix suction and shear strength, and the probability of disaster reaches its peak; while x Deviation m The farther away, the more exponentially the probability of disaster triggering decreases, whether the precipitation is too low (not reaching the critical water content) or too high (exceeding the main distribution range of historical disaster samples). This characteristic precisely reflects the "threshold effect" of rainfall-induced slope instability: there is a dominant precipitation range that is most likely to trigger disasters, rather than the risk necessarily increasing with higher precipitation.

[0036] parameter sThis is used to control the broadening of the probability density function curve, which in this application directly reflects the differences in the sensitivity of highway slopes to precipitation response and the dispersion of the trigger threshold. If s Smaller values ​​result in a narrow and high density curve, indicating that the geological structure, soil type, and topographic conditions of the highway slopes in the area are relatively uniform, and the response of each road section to precipitation tends to be consistent, with disaster events concentrated under specific precipitation intensities; if s Larger values ​​result in a flatter density curve, indicating a complex and diverse geological environment in the area (e.g., spanning multiple geomorphic units). The stability of slopes along different road sections varies significantly in its cumulative effect on prior effective rainfall, leading to disasters occurring dispersedly across a wider range of rainfall intensities. Therefore, s The numerical value is essentially a quantitative characterization of the regional geological homogeneity and the consistency of disaster triggering mechanisms.

[0037] Through the above m and s Through engineering interpretation, the normal distribution probability density function established in this invention is not only a statistical fitting result, but also builds an interpretive bridge between precipitation data and the physical instability mechanism of highway slopes, so that the model output can directly serve the disaster assessment and actual decision-making of front-line staff.

[0038] S3. Based on the highway network vector data, the highway network of the target area is divided into several grid units, and the effective grid units containing highways are statically vulnerable based on the disaster-bearing body survey and evaluation system to obtain the inherent risk classification index of each effective grid unit.

[0039] This step is the core component of quantifying the "vulnerability" of highway disaster-bearing bodies in this invention. Based on the relevant methods in the official and trial versions of the "Technical Guidelines for Comprehensive Risk Assessment of Highway Disaster-Bearing Bodies" (hereinafter referred to as the "Guidelines"), this step quantifies geological, engineering construction, and other factors contributing to highway disasters to form a disaster-bearing body risk index.

[0040] 1) Mesh cell generation To match the spatial resolution of precipitation data, this invention divides the highway network of the target area into regular grid cells based on highway network vector data (SHP files), with no overlap between adjacent grid cells. Specifically, a regular grid covering the target area is generated with a preset spatial resolution (e.g., 0.1° × 0.1°), and grid cells containing highway segments are identified as valid grid cells based on the highway network vector data. Taking a certain province as an example, this study sets up 2248 valid grid cells, completely covering all expressways, national highways, and provincial trunk roads in the province. The area covered by these valid grid cells constitutes the highway network area for subsequent output of visualized risk zoning, realizing the expansion of the assessment scale from the unit road segment to the grid scale, significantly reducing the computational complexity of the model while ensuring assessment accuracy.

[0041] 2) Risk level assessment of disaster-damaged sites The Guidelines outline a risk assessment system for highway disaster sites based on five indicators: disaster frequency (A), historical disaster severity (B), disaster response status (C), disaster development level (D), and highway importance (E). A Risk Assessment Index (CRI) is calculated based on these indicators, and risks are categorized according to the CRI values. This system not only specifies the weight of each factor but also establishes differentiated weight allocation schemes for different types of highway disaster sites. The weight coefficients for each disaster type are shown in Table 1.

[0042] Table 1 Weighting coefficients of risk assessment indicators for disaster-prone areas Using the coefficients in Table 1, the CRI score of each road disaster damage point is calculated by weighted summation. Then, the points are classified according to the grading standards in Table 2 to determine the risk level of each disaster point.

[0043] Table 2 Risk Level Assessment of Disaster-Damaged Sites 3) Calculation of inherent risk grading index of grid To adapt to the grid-scale assessment of this invention, the evaluation scope of unit road sections in the "Guidelines" has been expanded to include the number of disaster-prone points within a single grid unit. This optimization achieves the integration of historical disaster data at the provincial and higher levels with the vulnerability assessment of highway disaster-bearing bodies. The score of the number of risk points index for each level is the number of risk points of the corresponding level in the grid unit. The scoring values ​​of the disaster-prone point level index are shown in Table 3.

[0044] Table 3 Risk Level Indicator Scores Highway grid inherent risk classification index ( QI Calculate using the following formula: in, G 1. G 2. G 3. G 4 represents the risk scores for disaster sites ranging from level one to level four. F 1. F 2. F 3. F 4 represents the number of disaster points ranging from level one to level four within a grid cell. According to... QI The numerical values ​​are used to classify the inherent risk of highway grids into four levels, with specific thresholds shown in Table 4. This classification only covers effective grid unit areas containing highways and represents the inherent static vulnerability of highway disaster-bearing bodies, which does not change with rainfall conditions.

[0045] Table 4 Comprehensive Risk Rating Standards for Disaster-Damaged Sites S4. Based on the inherent risk classification index of each effective grid unit, the risk weight of each effective grid unit is determined by the analytic hierarchy process (AHP). The probability of disaster occurrence is multiplied by the corresponding risk weight to obtain the comprehensive risk score of each effective grid unit.

[0046] This step is a crucial link in coupling "hazard" and "vulnerability" in this invention. This invention adopts the classic paradigm of "hazard × vulnerability = comprehensive risk," where the inherent risk index of the grid ( QI Multiplying the probability of rainfall-induced road disasters directly by the overall risk score reveals a logical flaw: regardless of the probability of rainfall causing disasters, the overall risk score will always be zero (due to the low-risk grid). QI The value is ≤6 but not zero; in reality, it is not always zero, but direct multiplication cannot reflect the differentiated risk response of different inherent risk level grids under the same precipitation conditions. This is obviously inconsistent with engineering practice. Therefore, in order to fully incorporate the practical experience of highway operation and management personnel, this invention adopts the Analytic Hierarchy Process (AHP) to assign weights to highway grids with different inherent risk levels according to the pre-labeled grid risk levels.

[0047] In practice, the first step is to determine the inherent risk grading index of each effective grid unit ( QI The inherent risk level of effective grid cells is classified into high risk, relatively high risk, medium risk, and low risk. Then, the analytic hierarchy process (AHP) is used to assign weights to grid cells of different inherent risk levels. Finally, a consistency check is performed to verify the scientific validity of the weight allocation and eliminate subjective bias from manual scoring. The formula for calculating the comprehensive risk score of highway grid cells is as follows: S = P × W in,S For comprehensive grid risk scoring, P The probability of highway disasters caused by precipitation. W The risk weights are calculated using the analytic hierarchy process (AHP). This fusion method utilizes both objective probabilities obtained by fitting historical data and industry technical standards and empirical knowledge, achieving an organic connection between the model's quantitative output and actual engineering needs.

[0048] S5. Based on the comprehensive risk score of each effective grid unit, the kernel density estimation method is used to generate a continuous risk density surface of the target area, and the risk level is classified by the natural breakpoint method. Based on the highway network vector data, the visualized risk zoning result of the highway network area is output.

[0049] This step is the final output of this invention. To achieve a continuous spatial representation of discrete grid risk scores and objectively characterize the spatial variation and diffusion characteristics of highway disaster damage risk, this invention employs the weighted kernel density estimation (WKDE) method. Using the comprehensive risk score of each effective grid cell as the weight, a continuous risk density surface covering the entire target area is generated. Kernel density estimation, as a non-parametric spatial smoothing method, can effectively avoid the boundary abruptness problem caused by gridded data, making the risk assessment results more consistent with the natural law of continuous and gradual spatial variation of disaster risk.

[0050] Combining the natural breakpoint method, the daily dynamic comprehensive risk is divided into 5 levels: Level I (red, highest risk), Level II (orange, relatively high risk), Level III (yellow, medium risk), Level IV (blue, low risk), and the risk-free zone. The natural breakpoint method classifies levels by identifying natural boundaries in the data distribution, ensuring that levels are as similar as possible within each level while exhibiting significant differences between levels, thus guaranteeing the objectivity and rationality of the classification results.

[0051] Finally, the risk zoning results were spatially cropped based on the highway network vector data, retaining only the areas covered by the highway network. Different colors were used to label areas of different levels, achieving daily dynamic zoning and visualization of highway disaster damage risk. This classification realizes daily dynamic zoning of highway disaster damage risk across the province, providing intuitive visualization materials for the accurate location of high-risk zones or road sections, and can directly serve the daily risk assessment, inspection, and emergency response work of highway management agencies.

[0052] It should be noted that the above step numbers are only used to distinguish different operational stages and do not constitute a limitation on the execution order of each step. Among them, the precipitation-induced highway disaster probability modeling described in step S2 and the static vulnerability assessment of disaster-bearing bodies described in step S3 are independent of each other. They can be executed in parallel or in any order. Subsequent steps S4 and S5 perform fusion calculations and risk zoning based on the output results of the two.

[0053] The following uses a certain province as an example to illustrate the specific implementation of the present invention.

[0054] Data Acquisition and Preprocessing: Historical highway disaster data, historical precipitation raster data (NetCDF format), and highway network vector data (shp file) for a certain province were acquired. Hourly precipitation data for the target date were extracted from the NetCDF dataset and summed to obtain the daily cumulative precipitation. The accurate precipitation value at the latitude and longitude of each disaster point was calculated using bilinear interpolation.

[0055] Precipitation risk probability calculation: Based on historical highway disaster data, the disaster occurrence dates for each disaster site are determined, and the corresponding daily cumulative precipitation sequences are extracted. The effective precipitation formula (with an attenuation coefficient α determined to be 0.75 based on the evaporation intensity and soil properties of a specific region, and the number of preceding rainfall days n being 15 days) is used to calculate the effective precipitation at each disaster site. A Yeo-Johnson adaptive transform is applied to the effective precipitation, and the optimal λ value is obtained through maximum likelihood estimation to obtain a normalized precipitation sequence. The sequence is divided into 30 equal intervals, and the disaster occurrence frequency density corresponding to each interval is statistically analyzed. The probability density function of a normal distribution is obtained by fitting the sequence using the least squares method, and the probability of disaster occurrence corresponding to different precipitation levels is calculated using the cumulative distribution function.

[0056] Static Vulnerability Assessment: The highway network of a certain province was divided into 0.1°×0.1° grid units, and 2248 effective grid units were identified based on highway network vector data. According to the "Technical Guidelines for Comprehensive Risk Assessment of Highway Disaster-Bearing Bodies," the CRI scores and risk levels of disaster-prone points within each grid unit were assessed, and the inherent risk grading index (QI) of each grid unit was calculated. Based on the QI values, the 2248 grids were divided into four levels: 32 grids were high-risk areas, 34 were relatively high-risk areas, 64 were medium-risk areas, and the remaining 2118 were low-risk areas. A schematic diagram of the inherent risk grading of rainfall-induced highway disasters is shown below. Figure 2 As shown.

[0057] Comprehensive Risk Assessment and Visualization: The Analytic Hierarchy Process (AHP) is used to assign weights to grid cells with different inherent risk levels and pass a consistency test (consistency ratio CR < 0.1). The probability of precipitation-induced disasters is multiplied by the corresponding risk weight to obtain the comprehensive risk score for each grid cell. Using the comprehensive risk score as weight, a weighted kernel density estimation method is used to generate a continuous risk density surface covering the entire province. Combined with the natural discontinuity method, the daily dynamic comprehensive risk is divided into four risk levels and one risk-free zone. Based on highway network vector data, a visualized risk zoning map of the province's highway network area is output, realizing daily dynamic zoning and early warning of highway disaster damage risk.

[0058] This invention couples the probability of precipitation-induced disasters with relevant attributes of highways bearing the risk, creating a fully automated dynamic comprehensive risk assessment system for highway networks. The application of this method can quantify the impact of multi-day precipitation on highway damage, directly calculate risk scores and levels for regional road networks and road sections, and issue real-time, dynamic early warning information, providing scientific decision support for highway management agencies' disaster prevention and mitigation efforts.

[0059] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. A method for assessing highway meteorological disaster risk based on a disaster-bearing body survey and evaluation system, characterized in that, Includes the following steps: Historical highway disaster data, historical precipitation raster data, and highway network vector data of the target area are acquired, and the historical precipitation raster data is preprocessed to obtain daily cumulative precipitation raster data. Based on the historical highway disaster data, the disaster occurrence date for each disaster site is determined. Based on this date, the daily cumulative precipitation sequence corresponding to each disaster site is extracted from the daily cumulative precipitation raster data. The effective precipitation for each disaster site is calculated using the effective precipitation formula. The effective precipitation is then subjected to a normalization transformation. The disaster occurrence frequency density of the historical highway disaster data within each effective precipitation interval is statistically analyzed. A probability density function for precipitation-induced highway disasters is fitted, and the probability of disaster occurrence corresponding to different effective precipitation levels is calculated based on the cumulative distribution function of the probability density function. Based on the highway network vector data, the highway network of the target area is divided into several grid units, and a static vulnerability assessment is performed on the effective grid units containing highways based on the disaster-bearing body survey and evaluation system to obtain the inherent risk classification index of each effective grid unit. The process of dividing the highway network of the target area into several grid units based on the highway network vector data specifically includes: generating a regular grid covering the target area with a preset spatial resolution; identifying grid units containing highway segments as valid grid units based on the highway network vector data; and ensuring that adjacent valid grid units do not overlap. The aforementioned assessment of the static vulnerability of effective grid units containing highways, based on the disaster-bearing body survey and evaluation system, yields the inherent risk grading index for each effective grid unit. Specifically, this includes: calculating the risk assessment index for disaster-damaged points within each effective grid unit based on five indicators specified in the "Technical Guidelines for Comprehensive Risk Assessment of Highway Disaster-Bearing Bodies," namely, frequency of disaster occurrence, historical disaster severity, disaster response status, disaster development degree, and highway importance; determining the risk level of each disaster-damaged point based on the risk assessment index, whereby risk levels are categorized into Level 1, Level 2, Level 3, and Level 4, with corresponding risk scores of 100, 40, 15, and 6 points, respectively; and calculating the inherent risk grading index QI for each effective grid unit using the following formula: QI = G1F1 + G2F2 + G3F3 + G4F4. Among them, G1, G2, G3, and G4 are the risk scores of disaster points from level 1 to level 4, respectively, and F1, F2, F3, and F4 are the quantity indices of disaster points from level 1 to level 4 within the grid unit. Based on the QI, the effective grid units are divided into high, relatively high, medium, and low inherent risk levels. Based on the inherent risk level of each effective grid unit, the analytic hierarchy process (AHP) is used to assign weights to grid units with different inherent risk levels, and the weight allocation is verified through a consistency check to obtain the risk weight W of each effective grid unit. The probability of disaster occurrence P is multiplied by the corresponding risk weight W to obtain the comprehensive risk score S of each effective grid unit: S = P × W. Using the comprehensive risk score of each effective grid unit as a weight, a continuous risk density surface covering the entire target area is generated using a weighted kernel density estimation method. The continuous risk density surface is divided into four risk levels and one risk-free zone using the natural breakpoint method. Based on the highway network vector data, the risk zoning results are spatially cropped, retaining only the highway network coverage area. Different colors are used to mark the areas of each level, realizing daily dynamic zoning and visualization of highway disaster damage risk.

2. The method according to claim 1, characterized in that, The historical precipitation raster data is preprocessed, specifically including: extracting hourly precipitation data for the target date from the NetCDF dataset and summing them to obtain the daily cumulative precipitation; based on the latitude and longitude coordinates of each disaster point in the historical highway disaster data, using bilinear interpolation to calculate the precise precipitation value at the location of each disaster point, and generating the daily cumulative precipitation sequence corresponding to each disaster point.

3. The method according to claim 1, characterized in that, The effective rainfall formula is: R c =R0+αR1+α 2 R2+…+a n R n Among them, R c The effective precipitation is R0, which is the daily precipitation on the day the disaster occurs. n Let α be the daily precipitation on the nth day before the disaster, α be the rainfall attenuation coefficient, and n be the number of days of rainfall in the preceding period.

4. The method according to claim 1, characterized in that, The effective precipitation is normalized by using the Yeo-Johnson adaptive transformation, and the optimal transformation parameter λ is obtained by solving the maximum likelihood estimation to obtain the normalized precipitation sequence.

5. The method according to claim 1, characterized in that, The frequency density of road damage occurrences within each effective precipitation interval is statistically analyzed. Specifically, the effective precipitation sequence is divided into several equal intervals, and the ratio of the number of historical road damage events falling within each interval to the interval width is calculated as the frequency density of road damage occurrences corresponding to that interval. The least squares method is used to fit the frequency density to obtain a probability density function that follows a normal distribution.

Images