Spatiotemporal distribution patterns of urban waterlogging in Shanghai based on spatiotemporal clustering analysis

Under the combined influence of frequent rainstorm events and rapid urbanization, urban waterlogging has emerged as a critical challenge in contemporary urban development. Based on Shanghai waterlogging data from 2013 to 2023, this study investigates the seasonal variations, spatial distribution, and characteristics of typical rainstorm-induced waterlogging utilizing K-means clustering and probability density analysis. Results indicate significant seasonality, with peak occurrence in summer strongly synchronized with typhoon-related rainstorms. Road inundation constitutes the primary waterlogging type, exhibiting prominent spatial heterogeneity; notably, the central urban areas experience frequent inundation primarily due to limited drainage capacities and insufficient water retention spaces. Among the three waterlogging categories identified via K-means clustering, high-risk zones with deeper water depths are concentrated at the interface between Jiading Distract and central urban areas. Typical storm-related inundation events are characterized by rapid accumulation and recession ("fast in and fast out"), with durations typically not exceeding 10 hours. It is recommended that the risk early-warning lead time be controlled within 40 minutes. These findings provide critical insights to support the refinement of urban waterlogging management and enhance emergency response capabilities.