Urban flooding is a frequent problem affecting cities all over the world. The problem is more significant now that the climate is changing and urbanization trends are increasing. Various, physical hydrological models such as the Environmental Protection Agency Storm Water Management Model (EPA SWMM), MIKE URBAN-II and others, have been developed to simulate flooding events in cities. However, they require high accuracy mapping and a simulation of the underground storm drainage system. Sadly, this capability is usually not available for older or larger so-called megacities. Other hydrological model types are classified in the semi-physical category, like Cellular Automata (CA), require the incorporation of very fine resolution data. These types of data, in turn, demand massive computer power and time for analysis. Furthermore, available forecasting systems provide a way to determine total rainfall during extreme events, but they do not tell us what areas will be flooded. This work introduces an urban flooding tool that couples a rainfall-runoff model with a flood map database to expedite the alert process and estimate flooded areas. A 0.30-m Lidar Digital Elevation Model (DEM) of the study area (in this case Manhattan, New York City) is divided into 140 sub-basins. Several flood maps for each sub-basin are generated and organized into a database. For any forecasted extreme rainfall event, the rainfall-runoff model predicts the expected runoff volume at different times during the storm interval. The system rapidly searches for the corresponding flood map that delineates the expected flood area. The sensitivity analysis of parameters in the model show that the effect of storm inlet flow head is approximately linear while the effects of the threshold infiltration rate, the number of storm inlets, and the storm inlet flow reduction factor are non-linear. The reduction factor variation is found to exhibit a high non-linearity variation, hence requiring further detailed investigation.