Title: A novel framework for the spatiotemporal assessment of urban flood vulnerability
Abstract: Considering the pressing Urban Flood (UF) challenges faced by developing countries, the assessment of Urban Flood Vulnerability (UFV) assumes paramount importance in comprehending the detrimental impacts of UF on urban environments, thereby facilitating effective UF management. Addressing a notable gap in previous UFV research, which predominantly focused on spatial characteristics while overlooking temporal dynamics, we introduce a novel framework for spatiotemporal UFV assessment, by presenting a multiplicative equation to elucidate the interaction between Urban Flood Susceptibility (UFS) and Vulnerable Entities (VE). Employing this framework in the Great Bay Area (GBA), by considering the composition of the UFV database, we leveraged Machine Learnings and the Urban Vitality Index for quantifying the UFS and VE, respectively. SHapley Additive exPlanations was employed to quantify the local contributions of various factors to UFV. We found (1) a discernible upward trajectory in UFV across the GBA from 2012 to 2020, coupled with an expansion in spatial scope, indicative of persistent challenges in the future, and (2) the impervious surface percentage emerges as the primary contributor to UFV, exhibiting spatial heterogeneity reflective of regional environments. We believe this study has the potential to significantly contribute to addressing UF challenges and fostering the development of sustainable cities.
Keywords: Urban Flood Vulnerability；Machine learning；spatiotemporal distribution；SHapley Additive exPlanations
