Flood risk assessment of urban metro system using random forest algorithm and triangular fuzzy number based analytical hierarchy process approach

The metro system is an essential component of urban transportation; however, it is vulnerable to flooding under extreme rainfall. This study proposes a metro flood risk assessment approach by combining urban flood inundation model, random forest algorithm (RF), and triangular fuzzy number based analytic hierarchy process (TFNAHP). The evaluation indicators are selected based on the theory of hazard, exposure, and vulnerability, which are quantified through urban flood inundation model and geographic information system (GIS) technology. To solve the subjectivity of determining indicator weights, the data-driven RF algorithm is applied to identify indicator importance and create fuzzy judgment matrices of AHP. Moreover, the triangular fuzzy number is integrated with AHP to handle uncertain evaluation information. Taking the metro system in Zhengzhou, China as a case study, the flood risk is assessed for both metro line buffer zones and metro stations. The high-risk areas comprise more than 30 % of the study area. The assessment result is consistent with the historical flood condition. Furthermore, the proposed approach is more effective and suitable for metro flood risk assessment than the traditional AHP method. This study offers a new way for metro flood risk assessment and provides support for flood prevention in metro systems.