Hurricane Ida struck the U.S. East Coast in August 2021, driving the Schuylkill River in Philadelphia to a record discharge nearly 100 times its average flow. Ida exposes the growing challenge of predicting urban flooding arising from coupled rainfall–runoff and river–tide–landscape interactions that coarse models cannot resolve. Here, we address this gap with a street-resolving flood model that integrates LiDAR-derived terrain, bathymetric surveys, and land-use-based surface friction across Philadelphia’s watershed to reproduce Ida’s flood. We show that soil saturation, impervious surfaces, and fragmented infrastructure amplify pluvial flooding, increasing exposure in both low- and high-income communities. Scenario simulations reveal a flood tipping point: for river return periods exceeding 100 years, the inundated area grows logarithmically, with an additional 2–7% increase in flooding when peak discharge coincides with high tide and up to 15% under projected sea-level rise by 2100. As extreme rainfall intensifies and return periods shorten, this tipping point will be crossed more often, demanding integrated forecasting and adaptive planning in vulnerable, low-lying, rapidly urbanizing regions.
