Can urban pluvial flooding be predicted by open spatial data and weather data?

Cities are increasingly prone to urban flooding due to heavier rainfall, denser populations, augmenting imperviousness, and infrastructure aging. Urban pluvial flooding causes damage to buildings and contents, and disturbs stormwater drainage, transportation, and electricity provision. Designing and implementing efficient adaptation measures requires proper understanding of the urban response to heavy rainfall. However, implemented stormwater drainage models lack flood impact data for calibration, which results in poor flood predictions. Moreover, such models only consider rainfall and hydraulic parameters, neglecting the role of other natural, built, and social conditions in flooding mechanisms. This paper explores the potential of open spatial datasets to explain the occurrence of citizen-reported flood incidents during a heavy rain event. After a dimensionality reduction, imperviousness and proximity to watershed outflow point were found to significantly explain up to half of the flooding incidents variability, proving the usefulness of the proposed approach for urban flood modelling and management.