Luo et al. (2025) GIS-integrated flood risk assessment for metro systems based on bayesian cosine maximization method: a case study in Beijing

Identification

• Journal: Scientific Reports
• Year: 2025
• Date: 2025-12-14
• Authors: Aizhong Luo, Xingyu Yang, Tao Li, Bo Liu, Zihan Wang, Jiajun Shu
• DOI: 10.1038/s41598-025-32871-5

Research Groups

• School of Civil Engineering, Guizhou University of Engineering Science, Bijie, China
• School of Mechanics and Civil Engineering, China University of Mining and Technology-Beijing, Beijing, China
• Beijing Jingtou Urban Utility Tunnel Investment Co., Ltd., Beijing, China

Short Summary

This study develops a GIS-integrated Bayesian cosine maximization method (BCMM) for objective and spatially accurate flood risk assessment in metro systems, applying it to Beijing's metro. It identifies central urban areas and Tongzhou District as extremely high-risk zones, with over 50% of stations on key lines facing high or extremely high flood risks, which significantly increases under a 200-year extreme rainfall scenario.

Objective

• To develop and apply a comprehensive flood risk assessment model that integrates the Bayesian cosine maximization method (BCMM) with geographic information system (GIS) technology to enhance the objectivity and spatial accuracy of flood risk assessment for metro systems.
• To systematically assess the flood risk of the Beijing metro system under normal and extreme rainfall conditions using a 13-indicator evaluation system covering hazard, exposure, and vulnerability.
• To validate the effectiveness and superiority of the proposed BCMM by comparing its weight determination and risk identification results with the traditional fuzzy analytic hierarchy process (FAHP).

Study Configuration

• Spatial Scale: Beijing metro system, focusing on central urban areas and Tongzhou District. Data processed at a 30 meter x 30 meter grid resolution. Metro station buffer zones of 500 meters were established for analysis.
• Temporal Scale: Assessment under current conditions and a simulated once-in-200-year extreme rainfall scenario (annual exceedance probability of 0.5%, 24-hour duration, total daily rainfall of 329 millimeters). Data derived from long-term monitoring records.

Methodology and Data

• Models used:
  • Bayesian Cosine Maximization Method (BCMM) for weight optimization and pairwise comparison matrix correction.
  • Geographic Information System (GIS) for spatial quantification, analysis, and visualization (ArcGIS software, version 10.8).
  • Fuzzy Analytic Hierarchy Process (FAHP) and Analytic Hierarchy Process (AHP) for comparative analysis of weight determination.
  • Chicago rainfall method for designing the 200-year return period extreme rainfall scenario.
• Data sources:
  • Topography: Digital Elevation Model (DEM) data for Beijing (30 m x 30 m resolution).
  • Rainfall: Maximum daily rainfall, annual average rainfall, and number of days with rainfall exceeding 50 mm, from the Beijing Meteorological Bureau and Geographic System Science Database.
  • Socioeconomic factors: Population size, Gross Domestic Product (GDP), and road density, initially at district/county level, downscaled to 30 m x 30 m grid resolution.
  • Metro system: Metro station density, metro line density, and daily metro passenger flow, from the official Beijing Metro website.
  • Land use: Land use data from the Geographic System Science Database.
  • Hydrology: River network density from OpenStreetMap.
  • Validation: Historical flood event data (e.g., Jinanqiao Station in 2021, Taoranting Station in 2012).

Main Results

• The BCMM effectively mitigates subjective bias in expert judgments and improves weight calculation consistency and accuracy compared to AHP and FAHP, achieving a minimum violation criterion of 0 and significantly lower Euclidean distance values (e.g., a 92.3% reduction compared to FAHP at the integrated hazard-exposure-vulnerability level).
• Beijing's central urban areas (including Shijingshan, Dongcheng, and Xicheng Districts) and Tongzhou District are identified as extremely high-risk flood areas, accounting for 4.89% (775.20 square kilometers) of the total area. High-risk areas cover 22.98% (3,643.68 square kilometers).
• Within the 500-meter metro buffer zone, 12.45% (75.01 square kilometers) is classified as extremely high-risk, and 21.97% (132.45 square kilometers) is classified as high-risk.
• More than 50% of stations on metro Lines 1, 2, 5, 6, and 7 are exposed to high or extremely high flood risks, primarily located in low-lying, densely populated central urban areas.
• Under a simulated once-in-200-year extreme rainfall scenario (329 mm total daily rainfall), the overall flood risk level for the metro system increases significantly, with the cumulative proportion of high- and extremely high-risk areas reaching 92.3%.
• The BCMM demonstrates low sensitivity to weight perturbations (maximum area variation below 5% for ±10% weight change), indicating strong computational stability and reliability.

Contributions

• Proposes a novel GIS-integrated flood risk assessment model for metro systems based on the Bayesian cosine maximization method (BCMM), which enhances objectivity and spatial accuracy by mitigating subjective expert bias and optimizing weight allocation.
• Establishes a comprehensive 13-indicator evaluation system covering hazard, exposure, and vulnerability dimensions, spatially quantified using multisource data on a GIS platform.
• Provides a systematic comparison with traditional fuzzy analytic hierarchy process (FAHP), demonstrating the superior performance of BCMM in weight consistency and accuracy, particularly in aligning with historical flood-prone locations.
• Offers a robust framework for identifying and prioritizing flood-prone areas within metro networks, especially under extreme rainfall scenarios, providing a scientific basis for urban metro site selection, flood control standard formulation, and disaster prevention spatial layout optimization.

Funding

• Major Science and Technology Project of Xinjiang Uygur Autonomous Region (Grant NO: 2024A01003)
• National Natural Science Foundation of China (Grant No. 51508556)
• Key Support Project of the National Natural Science Foundation of China Joint Fund (Grant No. U24B2039)
• Guizhou Provincial Basic Research Program (Natural Science)-ZK (2024) General 600

Citation

@article{Luo2025GISintegrated,
  author = {Luo, Aizhong and Yang, Xingyu and Li, Tao and Liu, Bo and Wang, Zihan and Shu, Jiajun},
  title = {GIS-integrated flood risk assessment for metro systems based on bayesian cosine maximization method: a case study in Beijing},
  journal = {Scientific Reports},
  year = {2025},
  doi = {10.1038/s41598-025-32871-5},
  url = {https://doi.org/10.1038/s41598-025-32871-5}
}