This study presents a flood prediction framework that uses multiscale operational models representing meteorological, coastal, hydrologic, and hydraulic stormwater components. A dual-drainage (street-sewer) model forms the framework core and receives inputs from coastal and meteorological models. The framework was tested in a flood-prone area within the City of Hoboken, New Jersey. Hoboken's aging combined sewer system is regularly overwhelmed by rainfall, storm surge, or their combination resulting in nuisance flooding. The utility of the framework was demonstrated by retrospectively forecasting (72-hr horizon) two contrasting extreme flood events, a rainfall dominated event (Hurricane Irene) and a surge dominated event (Hurricane Sandy). The simulations showed that overland flow from storm surge and low-lying topography were major factor in surcharging the sewer system resulting in flooding. This modelling approach captures multiscale interactions and demonstrates the importance of holistically representing short-term stressors in urban-coastal systems. The framework can be run in ensemble mode to account for uncertainty in atmospheric forcing and aid decision making but this option was not explored in this study. Despite the limitations in expanding the dual-drainage domain to the entire city of Hoboken, the study offers interesting perspectives on leveraging existing models and predicting system response to different stressors.