Comprehensive assessment method for flood risk of subway station

Abstract

The invention discloses a subway station flood risk comprehensive assessment method, which comprises the steps of 1, acquiring multi-source data, and preprocessing the multi-source data to ensure the spatial consistency, availability and accuracy of the data; 2, performing data processing on the multi-source data obtained in the step 1 to obtain a secondary index; performing standardization processing on the second-level indexes, calculating the weight Wj of each second-level index after standardization processing through an entropy weight method, and performing weighted summation on the second-level indexes according to the weights to obtain a first-level index; and step 3, on the basis of the first-level indexes obtained in the step 2, a flood risk value and a damage risk value are calculated in sequence, and then a flood risk comprehensive value is obtained. According to the method, from the perspective of subway stations for the first time, a multi-factor comprehensive assessment method is constructed by systematically fusing four types of indexes including flood frequency, subway station design, exposure and vulnerability, and a systematic and integrated risk assessment framework is formed.

Description

Technical Field

[0001] This invention relates to the field of urban flood risk assessment and infrastructure resilience planning, specifically a comprehensive flood risk assessment method for subway stations. Background Technology

[0002] With the acceleration of global climate change and urbanization, urban flooding and extreme rainfall events are becoming increasingly frequent. As a vital backbone of urban public transportation, subway systems face unprecedented risks and challenges in the face of floods. Subway stations are usually located in the core areas of transportation hubs, with complex underground structures and numerous entrances and exits, making them highly susceptible to water accumulation and backflow. Once flooding occurs, it not only affects traffic operations but may also cause casualties and economic losses.

[0003] In recent years, the academic community has made positive progress in urban flood risk assessment. The paper "Wang, Y., Zhang, Q., Lin, K., Liu, Z., Liang, Y., Liu, Y., Li, C., 2024. A novel framework for urban flood risk assessment: Multiple perspectives and causal analysis. Water Res. 256, 121591." proposes a multi-indicator assessment model for urban floods based on the IPCC risk framework, incorporating factors such as adaptability. The paper "Seemuangngam, A., Lin, H.-L., 2024. The impact of urbanization on urban flood risk of Nakhon Ratchasima, Thailand. Appl. Geogr. 162, 103152." constructs the Urban Flood Risk Triangular Index (UFRTI) and combines BivariateLISA and K-means clustering to achieve spatiotemporal assessment of urban floods. The paper "Tang, X., Huang, X., Tian, ​​J., Pan, S., Ding, X., Zhou, Q., Sun, C., 2024. A novel framework for thespatiotemporal assessment of urban flood vulnerability. Sustain. Cities Soc. 109, 105523." proposes a framework for assessing urban flood vulnerability based on machine learning and the urban vitality index, which can quantify urban floodability and vulnerable entities. The paper "Guan, X., Yu, F., Xu, H., Li, C., Guan, Y., 2024. Flood risk assessment of urban metro system using random forest algorithm and triangular fuzzy number based analytical hierarchy process approach. Sustain. Cities Soc. 109, 105546." develops a method for assessing metro flood risk by integrating an urban flood inundation model, random forest algorithm, and triangular fuzzy number hierarchical analysis (TFNAHP).The paper “Gong, Y.,Xu, X., Tian, ​​K., Li, Z., Wang, M., Li, J., 2024. Subway station flood risk management level analysis. J. Hydrol. 638, 131473.” assesses the flood management level of a subway system based on complex network theory, combined with passenger flow statistics and flood risk maps.

[0004] While the aforementioned studies have provided various technical approaches for urban flood risk assessment, certain limitations remain. On the one hand, existing methods mostly focus on single dimensions such as surface water accumulation, vulnerability, or emergency management, lacking a systematic, integrated, and multi-dimensional comprehensive assessment framework for point-like facilities like subway stations. On the other hand, existing research has not fully utilized the flood frequency information contained in remote sensing time-series data, nor has it organically combined multiple factors such as station design characteristics, exposure levels, and emergency response capabilities to form a comprehensive assessment model for important transportation nodes. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a comprehensive method for assessing flood risk at subway stations.

[0006] The technical solution of the present invention to solve the aforementioned technical problem is to provide a comprehensive assessment method for flood risk at subway stations, characterized in that the method includes the following steps: Step 1: Acquire multi-source data and preprocess it to ensure spatial consistency, availability, and accuracy of the data; Step 2: Process the multi-source data obtained in Step 1 to obtain secondary indicators; then standardize the secondary indicators, and finally calculate the weight W of each standardized secondary indicator using the entropy weight method. j Then, the secondary indicators are weighted and summed according to their weights to obtain the primary indicators; Step 3: Based on the primary indicators obtained in Step 2, calculate the inundation risk value and damage risk value in sequence, and then obtain the comprehensive flood risk value.

[0007] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) This invention is the first to take the perspective of subway stations, systematically integrate four categories of indicators: flood frequency, subway station design, exposure and vulnerability, and construct a multi-factor comprehensive assessment method. This breaks through the shortcomings of existing assessments that are limited to single or two-dimensional indicators, and forms a systematic and integrated risk assessment framework.

[0008] (2) This invention integrates a multi-source data system, comprehensively utilizes remote sensing images, subway station design information and socio-economic data, and constructs a four-dimensional index system of flood frequency, subway station design, exposure and vulnerability, so as to realize the quantitative, spatial and dynamic assessment of flood risk of subway stations.

[0009] (3) The present invention introduces subway station design elements (such as the number of entrances and exits, slope and elevation), which are more in line with the flood intrusion mechanism of subway stations and make up for the shortcomings of existing assessment methods that ignore the structural characteristics of subway stations.

[0010] (4) Based on the dynamic updating capability of remote sensing flood frequency, the present invention can update the assessment results in real time as remote sensing data accumulates, which significantly improves the timeliness and accuracy of the assessment compared with traditional simulation or static risk maps.

[0011] (5) The present invention uses the entropy weight method to achieve objective weighting, which effectively overcomes the subjectivity problem of expert weighting in the existing methods and ensures the fairness and scientific nature of the allocation of contribution of each indicator.

[0012] (6) The output of this invention is diverse, and can generate various results such as risk level maps, spatial distribution maps and risk classification tables, which are convenient for comprehensive application by government management departments, emergency response agencies and urban resilience planning departments.

[0013] (7) This invention is universal and scalable. It is not only applicable to flood risk assessment of subway stations, but can also be extended to flood risk management and disaster prevention planning of other key urban infrastructure such as airports and transportation hubs. Attached Figure Description

[0014] Figure 1 This is a general framework diagram of the present invention; Figure 2 This is a schematic diagram illustrating the extraction of various secondary indicators in this invention; Figure 3 This is a map showing the comprehensive flood risk distribution results of subway stations in Shenzhen in 2023, based on Embodiment 1 of the present invention. Detailed Implementation

[0015] This invention provides a comprehensive assessment method for flood risk at subway stations (hereinafter referred to as the method), characterized by the following steps: Step 1: Acquire multi-source data and preprocess it to ensure spatial consistency, availability and accuracy of the data, thereby providing a standardized data foundation for subsequent indicator calculation and risk assessment; Preferably, in step 1, the multi-source data includes remote sensing data, geospatial data, and socioeconomic data.

[0016] Preferably, in step 1, the remote sensing data includes SAR (Synthetic Aperture Radar) image data, DEM (Digital Elevation Model) data, and HAND (Height to Nearest Drainage Channel) data; SAR image data has a spatial resolution that can vary from 10 to 100 m depending on the data source. Preferably, the SAR image data used is Sentinel-1 data, which has a spatial resolution of 10 meters and a time scale of the assessment year. It is obtained from remote sensing data sharing platforms (such as Google Earth Engine and the European Space Agency Copernicus Open Data Center) and is used for flood extent identification and frequency statistics. DEM data is typically spatially resolved at 10-90m (preferably 30m), obtained from geospatial data agencies (such as the U.S. Geological Survey (USGS) and the National Geographic Information Public Service Platform), and is used to calculate topographic parameters such as slope and elevation. HAND data: Spatial resolution is typically 10~90m (preferably 30m), obtained from remote sensing data platforms (such as Google Earth Engine dedicated dataset), used for terrain correction of flood classification results.

[0017] Preferably, in step 1, the geospatial data includes POI (Point of Interest) data, building vector data, subway line vector data, and drainage pumping station location data; POI data includes the locations of subway stations, subway entrances / exits, hospitals, and fire stations; the time scale is the assessment year, obtained from a map service platform, and supports the calculation of indicators related to subway station design and emergency response capabilities. Building vector data is: the time scale is the year of assessment or the previous year, obtained from open source map platforms (such as OpenStreetMap) or urban planning data platforms, and is used to count the number and density of buildings; The vector data for subway lines is: the time scale is the year of evaluation, and it is obtained from map service platforms or urban rail transit operators. It is used to calculate the density of subway lines. Location data of drainage pumping stations: The time scale is the assessment year, obtained from the city government's open data platform, to support the calculation of relevant indicators of emergency drainage capacity.

[0018] Preferably, in step 1, the socioeconomic data includes population density data and nighttime light data; Population density data has a spatial resolution that varies from 100 to 1000 m depending on the data source. Preferably, the population density data is from the LandScan global population database, which has a spatial resolution of 1 km and a time scale of the assessment year, and is used to characterize the degree of population concentration in the region. Nighttime light data is used to characterize the intensity of regional economic activity. Its spatial resolution varies depending on the data source and can range from 100 to 1000 m. Preferably, the nighttime light data is the annual average nighttime light data of VIIRS (in this embodiment, the spatial resolution is 500 m, using VNL V2 or VNL V2A), which is obtained from the Google Earth Engine platform and covers the assessment year in terms of time scale.

[0019] Preferably, in step 1, the preprocessing includes sequentially performing coordinate system unification, format conversion, region cropping, and noise removal; A unified coordinate system means converting all data to the same geographic coordinate system (such as WGS84) to avoid spatial offset. Format conversion is the process of standardizing data formats to facilitate subsequent processing by software such as ArcGIS and QGIS. Preferably, SAR image data, DEM data, HAND data, population density data, and nighttime light data are raster data, while POI data, building vector data, subway line vector data, and drainage pumping station location data are vector data; raster data are uniformly in GeoTIFF format, and vector data are uniformly in Shapefile format.

[0020] Regional clipping involves cropping all data to the evaluation range based on the boundaries of the target evaluation area (such as a city or the city's core area) to reduce data redundancy. Noise removal involves applying median filtering to SAR images, with a filter radius set between 5 and 20 meters. Population density data and nighttime light data are smoothed using mean filtering to reduce data noise.

[0021] Step 2: Construct a flood risk assessment framework: Process the multi-source data obtained in Step 1 to obtain secondary indicators; then standardize the secondary indicators, and finally calculate the weight W of each standardized secondary indicator using the entropy weight method. j Then, the secondary indicators are weighted and summed according to their weights to obtain the primary indicators, thus forming a flood risk assessment framework. Preferably, in step 2, the flood risk assessment framework is a hierarchical system of primary indicators and secondary indicators, containing 4 parallel primary indicators, each primary indicator corresponding to several secondary indicators. The four primary indicators are flood frequency (FF), subway station design (HS), exposure (E), and vulnerability (V). Flood frequency (FF) has no secondary index; The design of subway stations includes three secondary indicators: subway entrance / exit elevation (ME), subway entrance / exit slope (MES), and the number of subway entrances / exits (NSE). Exposure E includes four secondary indicators: population density (PD), nighttime light intensity (NTL), building density (BDC), and metro line density (MLD). Vulnerability V includes three secondary indicators: distance to the nearest hospital (DTH), distance to the nearest fire station (DTFS), and distance to the nearest drainage pumping station (DTPS).

[0022] Preferably, in step 2, the data processing specifically includes: A flood classification model was established using a random forest model to classify all SAR image data, obtaining the flood inundation range in each time phase of the image. DEM data and HAND data were used to perform topographic correction on the flood inundation range. Pixel-by-pixel statistics were performed on all flood inundation ranges to calculate the flood frequency of each pixel (i.e., the proportion of times the pixel is classified as a flood out of the total number of images). Pixels with a flood frequency ≥70% were defined as permanent water bodies (such as rivers and lakes). After removing permanent water bodies from the flood frequency results, the spatial distribution data of flood frequency at a resolution of 10m (TIF format) for the evaluation area was obtained. The subway station locations in the POI data are fitted to the flood frequency spatial distribution data (TIF format), and a circular buffer with a radius of 300-800m is set for each subway station location; preferably, the buffer radius is 500m. The flood frequency of all 10m resolution pixels within the buffer is extracted, and the average value is calculated as the flood frequency FF value for each subway station location. The location points of subway entrances and exits in the POI data are coupled with the DEM data. The elevation value of each subway entrance and exit is extracted based on the DEM data. The average elevation value of all entrances and exits of the same subway station is taken to obtain the subway entrance and exit elevation ME. Using GIS tools, slope data is calculated based on DEM data, and the slope value of each subway entrance / exit is extracted. The average slope value of all entrances / exits of the same subway station is then taken to obtain the subway entrance / exit slope MES. By using the location points of subway entrances and exits in the POI data, the number of entrances and exits for each subway station is counted, and the number of subway entrances and exits NSE is obtained. The subway station locations in the POI data are fitted to the population density data, and a circular buffer with a radius of 300~800m (preferably 500m) is set for each subway station location; the population density data within the buffer range is extracted to obtain the population density PD. The subway station locations in the POI data are fitted to the nighttime light data. A circular buffer with a radius of 300-800m (preferably 500m) is set for each subway station location. The pixel values ​​of the nighttime light data within the buffer range are extracted, and the average value is taken as the nighttime light intensity NTL. The subway station locations in the POI data are fitted into the building vector data. A circular buffer with a radius of 300-800m (preferably 500m) is set for each subway station location. The number of buildings within the buffer area is counted and normalized by area to obtain the building number density (BDC). The subway station locations in the POI data are fitted to the subway line vector data. A circular buffer with a radius of 300~800m (preferably 500m) is set for each subway station location. The total length of the subway line within the buffer area is calculated and normalized by area to obtain the subway line density MLD. Based on the hospital and fire station locations in the POI data, the straight-line distance DTH between each subway station location and the nearest hospital and the straight-line distance DTFS between each subway station location and the nearest fire station were calculated using GIS tools. Based on the location data of drainage pumping stations, the straight-line distance (DTPS) between each subway station location and the nearest drainage pumping station is calculated.

[0023] Preferably, in step 2, the standardization process specifically includes: To eliminate dimensional differences, all secondary indicators were standardized using min-max standardization. Secondary indicators with positive correlation include the number of subway entrances / exits (NSE), population density (PD), nighttime light intensity (NTL), building density (BDC), subway line density (MLD), distance from the nearest hospital (DTH), distance from the nearest fire station (DTFS), and distance from the nearest drainage pumping station (DTPS). Their standardized formulas are as follows: (1) Secondary indicators with negative correlation include the subway entrance / exit elevation (ME) and subway entrance / exit slope (MES). The original data must first be inverted to ensure that all indicators have a consistent impact on flood risk. The standardized formula is as follows: (2) (3) In equations (1) to (3), x is the original value of the secondary indicator. rev x is the inverse value of an indicator with negative correlation. std The standardized value; x min x max These are the minimum and maximum values ​​of the original values ​​of the secondary indicators, x. rev,min x rev,max These are the minimum and maximum values ​​of the inverted values, respectively.

[0024] Preferably, in step 2, the entropy weight method is used to calculate the weights of all standardized secondary indicators. The weights are assigned based on the data dispersion to avoid interference from subjective experience and ensure the objectivity of the weights. The specific process is as follows: Let the evaluation objects be m subway stations, and the total number of secondary indicators be n, then the indicator matrix is ​​X=(x ij )m×n, where x ij Let j be the standardized value of the j-th secondary indicator for the i-th station; For each secondary indicator, calculate the proportion P of the indicator value of the i-th station to all values ​​of that indicator. ij The formula is as follows: (4) like Then let P ij =1 / m, to avoid the meaninglessness of logarithmic calculations; Based on P ij Calculate the entropy value E of the j-th secondary indicator. j The formula is as follows: (5) If P ij =0, then let ; Based on entropy E j Calculate the weight W of the j-th secondary indicator. j The formula is as follows: (6) In equation (6), (Weight normalization); Entropy value E j The smaller the value, the greater the dispersion of the indicator data, and the more significant its contribution to risk assessment. (Weight W) j The higher.

[0025] Preferably, in step 2, the weighted summation specifically involves: The secondary indicators, namely, the elevation ME of the subway entrance / exit, the slope MES of the subway entrance / exit, and the number NSE of the subway entrance / exit, are standardized according to formulas (1) to (3) to obtain MEstd, MESstd, and NSEstd. Combined with the weights WME, WMES, and WNSE determined by formulas (4) to (6), the calculation formula for the design HS of the subway station is as follows: (7) The secondary indicators, population density (PD), nighttime light intensity (NTL), building density (BDC), and subway line density (MLD), are standardized according to equations (1) to (3) to obtain PDstd, NTLstd, BDCstd, and MLDstd. Combined with the weights WPD, WNTL, WBDC, and WMLD determined by equations (4) to (6), the formula for calculating exposure E is as follows: (8) The straight-line distances (DTH) to the nearest hospital, (DTFS) to the nearest fire station, and (DTPS) to the nearest drainage pumping station for the secondary indicators are standardized according to equations (1)-(3) to obtain DTHstd, DTFSstd, and DTPSstd. Combined with the weights (WDTH), (WDTFS), and (WDTPS) determined by equations (4)-(6), the formula for calculating vulnerability V is as follows: (9) Step 3: Based on the primary indicators obtained in Step 2, calculate the inundation risk value and damage risk value in sequence, and then obtain the comprehensive flood risk value.

[0026] Preferably, in step 3, the inundation risk value M1 focuses on the core risk of whether a flood can inundate the station, considering only the interaction between the intensity of the disaster source and the resilience of the facilities. This is suitable for scenarios involving station inundation risk screening, and the calculation formula is as follows: M1 = FF × HS (10) In equation (10), FF is the flood frequency calculated in step 2, which represents the intensity of the disaster source; HS is the subway station design calculated in step 2, which represents the station's ability to resist floods; the product of the two directly reflects the possibility that the station will be flooded.

[0027] Preferably, in step 3, the damage risk value M2 introduces exposure E based on the inundation risk value M1 to further quantify the potential losses after flooding. This is suitable for scenarios involving potential loss assessment, and the calculation formula is as follows: M2 = FF × HS × E (11) In Equation (11), FF is the flood frequency calculated in step 2, which represents the intensity of the disaster source; HS is the subway station design calculated in step 2, which represents the station's ability to resist floods; E is the exposure calculated in step 2, which represents the density of population, economy and facilities around the station and the importance of the station; the product of the three reflects the potential loss risk caused by flooding.

[0028] Preferably, in step 3, the comprehensive flood risk value M3 introduces vulnerability V based on the damage risk value M2, characterizing the comprehensive risk of flood occurrence, inundation, loss, and response at the subway station. This is applicable to scenarios where comprehensive disaster prevention and mitigation plans are developed. The calculation formula is as follows: M3 = FF × HS × E × V (12) In Equation (12), FF is the flood frequency calculated in step 2, representing the intensity of the disaster source; HS is the subway station design calculated in step 2, representing the station's ability to resist floods; E is the exposure calculated in step 2, representing the density of population, economy and facilities around the station and the importance of the station; V is the vulnerability calculated in step 2, representing the emergency facility guarantee capacity around the station. The product of the four factors fully reflects the comprehensive flood risk of the subway station.

[0029] Example 1: This embodiment uses the Shenzhen Metro system in 2023 (including 15 lines and 393 operating stations, excluding "Airport Station" and "Airport North Station"—because airport runways are low-reflection areas and easily confused with water bodies in SAR images, affecting classification accuracy) as the evaluation object. The data parameters used strictly follow the research data system. In step 1, the remote sensing data consisted of 84 Sentinel-1 SARGRD images (covering 29 rainstorm events in 2023), with a spatial resolution of 10 meters and a transit time of 10:26 AM. The images used VV and VH dual bands and were preprocessed with a 10-meter radius median filter for noise reduction. The NASA SRTM digital elevation data (2000) and Global30mHAND data (2017) both had a spatial resolution of 30 meters and were based on slope data derived from the DEM (used for topographic correction of flood classification results).

[0030] Geospatial data: 2023 POI data includes location information for 1,365 subway entrances / exits, hospitals, and fire stations; 2023 subway line vector data (covering 15 lines); 2024 building vector data (from OpenStreetMap); 2023 Shenzhen drainage pumping station location data (from the Shenzhen official open data platform, https: / / opendata.sz.gov.cn / ).

[0031] Socioeconomic data: LandScan global population database (2023, 1km resolution) and nighttime light data (2023, 500m resolution) were used for exposure index calculation.

[0032] In step 2, the indicator calculation and weight parameters are as follows: The standardization of secondary indicators adopts the min-max method, in which the elevation ME of subway entrances and exits and the slope MES of subway entrances and exits are first inverted and then standardized. The entropy values ​​and weights of the secondary indicators calculated using the entropy weight method are as follows: Entropy value of subway entrance / exit elevation (ME): 0.991, weight 0.019; Entropy value of subway entrance / exit slope (MES): 0.995, weight 0.011; Entropy value of subway entrance / exit quantity (NSE): 0.964, weight 0.074; Entropy value of population density (PD): 0.938, weight 0.127; Entropy value of nighttime light intensity (NTL): 0.984, weight 0.033; Entropy value of building density (BDC): 0.884, weight 0.237; Entropy value of subway line density (MLD): 0.962, weight 0.078; Entropy value of straight-line distance to the nearest hospital (DTH): 0.963, weight 0.076; Entropy value of straight-line distance to the nearest fire station (DTFS): 0.956, weight 0.091; Entropy value of straight-line distance to the nearest drainage pumping station (DTPS): 0.930, weight 0.143.

[0033] A 500-meter radius buffer zone is constructed centered on the station (covering all entrances and exits; the maximum distance between Shenzhen Metro entrances and exits and the station is 487 meters).

[0034] In step 3, the parameters used in the flood risk assessment method are as follows: Accuracy metrics for the flood classification model: Overall accuracy (OA) 95.95%, User accuracy at flood points (UAF) 95%, User accuracy at non-flood points (UANF) 97.06%, Kappa coefficient 0.92; Results of three risk calculation methods: Under method M1, 9.4% of stations are at high / very high risk (concentrated in Pingshan, Guangming and along the Bao'an Airport line); under method M2, 11.1% of stations are at high / very high risk (concentrated in Luohu and Yantian); and under method M3, 9.4% of stations are at high / very high risk (concentrated in Longgang, Guangming and along the Bao'an Airport line). High-risk subway lines: Line 16 (29.2% of stations are high / very high risk), Line 14 (27.8%), Line 20 (25.0%), and Line 6 (22.6%). These lines are prone to disruption during heavy rain.

[0035] Figure 1 The document clearly demonstrates the correlation logic of the four primary indicators FF, HS, E, and V, as well as the progressive relationship of the three risk value calculation methods M1-M3, corresponding to the entire process of "data acquisition - indicator calculation - weight determination - risk assessment" in this invention.

[0036] Figure 2 The text presents the extraction methods (such as FF extraction from SAR images, ME extraction from DEM, MES extraction from POI data, NSE extraction from POI data, DTH extraction, etc.) and their spatial distribution characteristics in a visually intuitive way. Figure 3Based on the M3 method, the risk is divided into five levels: "extremely low, low, medium, high, and extremely high". High and extremely high risk stations are mainly distributed in Longgang Central District, Guangming District and along Bao'an Airport, which can provide a basis for the flood prevention priority classification of Shenzhen Metro.

[0037] Any aspects not covered in this invention are applicable to existing technologies.

Claims

1. A comprehensive assessment method for flood risk at subway stations, characterized in that, The method includes the following steps: Step 1: Acquire multi-source data and preprocess it to ensure spatial consistency, availability, and accuracy of the data; Step 2: Process the multi-source data obtained in Step 1 to obtain secondary indicators; then standardize the secondary indicators, and finally calculate the weight W of each standardized secondary indicator using the entropy weight method. j Then, the secondary indicators are weighted and summed according to their weights to obtain the primary indicators; Step 3: Based on the primary indicators obtained in Step 2, calculate the inundation risk value and damage risk value in sequence, and then obtain the comprehensive flood risk value.

2. The comprehensive flood risk assessment method for subway stations according to claim 1, characterized in that, In step 1, the multi-source data includes remote sensing data, geospatial data, and socioeconomic data.

3. The comprehensive flood risk assessment method for subway stations according to claim 2, characterized in that, In step 1, the remote sensing data includes SAR image data, DEM data, and HAND data; In step 1, the geospatial data includes POI data, building vector data, subway line vector data, and drainage pumping station location data; In step 1, the socioeconomic data includes population density data and nighttime light data.

4. The comprehensive flood risk assessment method for subway stations according to claim 3, characterized in that, In step 1, the SAR image data has a spatial resolution of 10-100m; preferably, the SAR image data uses Sentinel-1 data, which has a spatial resolution of 10 meters and a time scale of the assessment year, and is obtained from the remote sensing data sharing platform for flood extent identification and frequency statistics. DEM data consists of topographic parameters with a spatial resolution of 10-90m, obtained from geospatial data agencies, used to calculate slope and elevation. HAND data: spatial resolution of 10~90m, acquired from remote sensing data platform, used for terrain correction of flood classification results; In step 1, the POI data includes subway station locations, subway entrance / exit locations, hospital locations, and fire station locations; The timescale is the year of assessment, obtained from the map service platform, to support the calculation of indicators related to subway station design and emergency response capabilities; Building vector data is: the time scale is the year of assessment or the previous year, obtained from open source map platforms or urban planning data platforms, and used to count the number and density of buildings; The vector data for subway lines is: the time scale is the year of evaluation, and it is obtained from map service platforms or urban rail transit operators. It is used to calculate the density of subway lines. Location data of drainage pumping stations: The time scale is the assessment year, obtained from the city government's open data platform, to support the calculation of relevant indicators of emergency drainage capacity; In step 1, the population density data has a spatial resolution of 100~1000m; preferably, the population density data uses LandScan global population database data, which has a spatial resolution of 1km and a time scale of the assessment year, to characterize the degree of population concentration in the region. Nighttime light data represents the intensity of regional economic activity and has a spatial resolution of 100–1000 m. Preferably, the nighttime light data is the VIIRS annual average nighttime light data, acquired from the Google Earth Engine platform, with a time scale covering the assessment year.

5. The comprehensive flood risk assessment method for subway stations according to claim 1, characterized in that, In step 1, preprocessing includes sequentially performing coordinate system unification, format conversion, region cropping, and noise removal; A unified coordinate system means converting all data to the same geographic coordinate system to avoid spatial offset. Format conversion is the process of standardizing data formats to facilitate subsequent software processing. Preferably, SAR image data, DEM data and HAND data, population density data and nighttime light data are raster data, and POI data, building vector data, subway line vector data and drainage pumping station location data are vector data; raster data are uniformly in GeoTIFF format, and vector data are uniformly in Shapefile format. Region clipping involves cropping all data to the evaluation range based on the target evaluation region boundaries to reduce data redundancy. Noise removal involves: applying median filtering to SAR images with a filter radius of 5-20m; and applying mean filtering to population density data and nighttime light data to smooth them and reduce data noise.

6. The comprehensive flood risk assessment method for subway stations according to claim 1, characterized in that, In step 2, the four parallel primary indicators are flood frequency (FF), subway station design (HS), exposure (E), and vulnerability (V). Flood frequency (FF) has no secondary index; The design of subway stations includes three secondary indicators: subway entrance / exit elevation (ME), subway entrance / exit slope (MES), and the number of subway entrances / exits (NSE). Exposure E includes four secondary indicators: population density (PD), nighttime light intensity (NTL), building density (BDC), and metro line density (MLD). Vulnerability V includes three secondary indicators: distance to the nearest hospital (DTH), distance to the nearest fire station (DTFS), and distance to the nearest drainage pumping station (DTPS).

7. The comprehensive flood risk assessment method for subway stations according to claim 1, characterized in that, In step 2, the data processing specifically involves: A flood classification model was established using a random forest model to classify all SAR image data, obtaining the flood inundation range in each time phase of the image. DEM data and HAND data were used to perform topographic correction on the flood inundation range. Pixel-by-pixel statistics were performed on all flood inundation ranges to calculate the flood frequency of each pixel. Pixels with a flood frequency ≥70% were defined as permanent water bodies. After removing permanent water bodies from the flood frequency results, the spatial distribution data of flood frequency at a resolution of 10m for the evaluation area was obtained. The subway station locations in the POI data were fitted to the flood frequency spatial distribution data, and a circular buffer with a radius of 300~800m was set for each subway station location. Extract the flood frequency of all 10m resolution pixels in the buffer, calculate the average value, and use it as the flood frequency FF value for each subway station. The location points of subway entrances and exits in the POI data are coupled with the DEM data. The elevation value of each subway entrance and exit is extracted based on the DEM data. The average elevation value of all entrances and exits of the same subway station is taken to obtain the subway entrance and exit elevation ME. Using GIS tools, slope data is calculated based on DEM data, and the slope value of each subway entrance / exit is extracted. The average slope value of all entrances / exits of the same subway station is then taken to obtain the subway entrance / exit slope MES. By using the location points of subway entrances and exits in the POI data, the number of entrances and exits for each subway station is counted, and the number of subway entrances and exits NSE is obtained. The subway station locations in the POI data are fitted to the population density data, and a circular buffer with a radius of 300-800m is set for each subway station location; the population density data within the buffer range is extracted to obtain the population density PD. The subway station locations in the POI data were fitted to the nighttime light data, and a circular buffer with a radius of 300-800m was set for each subway station location. The pixel values ​​of the nighttime light data within the buffer range were extracted, and the average value was taken as the nighttime light intensity NTL. The subway station location points in the POI data are fitted into the building vector data, and a circular buffer with a radius of 300~800m is set for each subway station location point; The number of buildings within the buffer zone is counted and normalized by area to obtain the building density (BDC). The subway station location points in the POI data are fitted to the subway line vector data, and a circular buffer with a radius of 300~800m is set for each subway station location point. Calculate the total length of the metro lines within the buffer zone and normalize it by area to obtain the metro line density MLD; Based on the hospital and fire station locations in the POI data, the straight-line distance DTH between each subway station location and the nearest hospital and the straight-line distance DTFS between each subway station location and the nearest fire station were calculated using GIS tools. Based on the location data of drainage pumping stations, the straight-line distance (DTPS) between each subway station location and the nearest drainage pumping station is calculated.

8. The comprehensive flood risk assessment method for subway stations according to claim 1, characterized in that, In step 2, the standardization process specifically involves: Secondary indicators with positive correlation include the number of subway entrances / exits (NSE), population density (PD), nighttime light intensity (NTL), building density (BDC), subway line density (MLD), distance from the nearest hospital (DTH), distance from the nearest fire station (DTFS), and distance from the nearest drainage pumping station (DTPS). Their standardized formulas are as follows: （1） Secondary indicators with negative correlation include the subway entrance / exit elevation (ME) and subway entrance / exit slope (MES). The original data must first be inverted to ensure that all indicators have a consistent impact on flood risk. The standardized formula is as follows: （2） （3） In equations (1) to (3), x is the original value of the secondary indicator. rev x is the inverse value of an indicator with negative correlation. std The standardized value; x min x max These are the minimum and maximum values ​​of the original values ​​of the secondary indicators, x. rev,min x rev,max These are the minimum and maximum values ​​of the inverted value, respectively; In step 2, the entropy weight method specifically involves: Let the evaluation objects be m subway stations, and the total number of secondary indicators be n, then the indicator matrix is ​​X=(x ij )m×n, where x ij Let j be the standardized value of the j-th secondary indicator for the i-th station; For each secondary indicator, calculate the proportion P of the indicator value of the i-th station to all values ​​of that indicator. ij The formula is as follows: （4） like Then let P ij =1 / m, to avoid the meaninglessness of logarithmic calculations; Based on P ij Calculate the entropy value E of the j-th secondary indicator. j The formula is as follows: （5） If P ij =0, then let ; Based on entropy E j Calculate the weight W of the j-th secondary indicator. j The formula is as follows: （6） In equation (6), Entropy value E j The smaller the value, the greater the dispersion of the indicator data, and the more significant its contribution to risk assessment. The weight W... j The higher.

9. The comprehensive flood risk assessment method for subway stations according to claim 1, characterized in that, In step 2, the weighted summation is specifically performed as follows: The secondary indicators, namely, the elevation ME of the subway entrance / exit, the slope MES of the subway entrance / exit, and the number NSE of the subway entrance / exit, are standardized according to formulas (1) to (3) to obtain MEstd, MESstd, and NSEstd. Combined with the weights WME, WMES, and WNSE determined by formulas (4) to (6), the calculation formula for the design HS of the subway station is as follows: （7） The secondary indicators, population density (PD), nighttime light intensity (NTL), building density (BDC), and subway line density (MLD), are standardized according to equations (1) to (3) to obtain PDstd, NTLstd, BDCstd, and MLDstd. Combined with the weights WPD, WNTL, WBDC, and WMLD determined by equations (4) to (6), the formula for calculating exposure E is as follows: （8） The straight-line distances (DTH) to the nearest hospital, (DTFS) to the nearest fire station, and (DTPS) to the nearest drainage pumping station for the secondary indicators are standardized according to equations (1)-(3) to obtain DTHstd, DTFSstd, and DTPSstd. Combined with the weights (WDTH), (WDTFS), and (WDTPS) determined by equations (4)-(6), the formula for calculating vulnerability V is as follows: （9）。 10. The comprehensive flood risk assessment method for subway stations according to claim 1, characterized in that, In step 3, the inundation risk value M1 focuses on the core risk of whether a flood can inundate the station. It only considers the interaction between the intensity of the disaster source and the resilience of the facilities, and is suitable for scenarios involving station inundation risk screening. The calculation formula is as follows: M1 = FF × HS (10) In Equation (10), FF is the flood frequency calculated in step 2, which represents the intensity of the disaster source; HS is the subway station design calculated in step 2, which represents the station's ability to resist floods; the product of the two directly reflects the possibility that the station will be flooded. In step 3, the damage risk value M2 introduces exposure E based on the inundation risk value M1 to further quantify the potential losses after flooding. This is applicable to scenarios involving potential loss assessment, and the calculation formula is as follows: M2 = FF × HS × E (11) In Equation (11), FF is the flood frequency calculated in step 2, which represents the intensity of the disaster source; HS is the subway station design calculated in step 2, which represents the station's ability to resist floods; E is the exposure calculated in step 2, which represents the density of population, economy and facilities around the station and the importance of the station. The product of these three factors reflects the potential risk of loss caused by flooding. In step 3, the comprehensive flood risk value M3 introduces vulnerability V based on the damage risk value M2, representing the comprehensive risk of flood occurrence, inundation, loss, and response at subway stations. It is applicable to scenarios where comprehensive disaster prevention and mitigation plans are developed. The calculation formula is as follows: M3 = FF × HS × E × V (12) In Equation (12), FF is the flood frequency calculated in step 2, which represents the intensity of the disaster source; HS is the subway station design calculated in step 2, which represents the station's ability to resist floods; E is the exposure calculated in step 2, which represents the density of population, economy and facilities around the station and the importance of the station. V represents the vulnerability calculated in step 2, characterizing the emergency support capacity of the area surrounding the station. The product of these four factors fully reflects the comprehensive flood risk of the subway station.

Images