Urban flood risk evaluation using social media data and Bayesian network approach: A spatial-temporal dynamic analysis in Wuhan city, China

Abstract

Under the dual impacts of urbanization and climate change, urban flooding has become one of the major threats to urban development, while the spatial-temporal evolution of urban flood risk remains unclear. In this study, we delineated the inundation areas of flooding in Wuhan from 2012–2023 using social media data and assessed urban flood risk through a Bayesian network based approach. Furthermore, future risks were evaluated considering various climate and land use scenarios. The results show that flood events are mainly concentrated in the central part of the city. Road density and impervious surfaces are identified as the primary factors influencing flooding. The proportion of high-risk areas will increase by an average of 3.13 % (from 14.66 % to 17.79 %) in 2100 under the SSP5–8.5 and SSP2–4.5 scenarios compared to the historical period. Land use alterations have a greater influence in the southern region. In contrast, variations in precipitation have a more substantial impact in the northern region and the influence of land use change is more significant compared to precipitation change. This study indicates that the Bayesian network model can effectively depict the complex process and probabilistic characteristics of urban floods. Integrated with future climate and land use scenarios, this approach would provide a scientific basis for the refined management of flood risk under climate change, thereby enhancing urban resilience.