Flood-inundation modeling in an operational context: sensitivity to topographic resolution and Manning's n

Abstract

Abstract Streamflow forecasts from operational hydrologic models can be converted into forecasts of flood-inundation extent using either physically based hydraulic models or simpler terrain-based approaches. Two factors that influence simulated flood-inundation extent are spatial resolution of topographic data and in-channel and overland-flow roughness characterized by the Manning's n parameter. Here, AutoRoute, a raster-based flood-inundation model, was used to simulate two recent flood events in Florida (a forested floodplain) and Texas (an urban floodplain) using two different topographic resolutions and a range of Manning's n values. The AutoRoute-simulated flood-inundation extents were evaluated using observed extents from remotely sensed imagery. For comparison, the same flood events were also simulated using a one-dimensional Hydrologic Engineering Center River Analysis System (HEC-RAS) model. Results indicated that model performance was much improved with higher topographic resolution for the forested floodplain site and that the urban site was more sensitive to Manning's n. For the three different rivers analyzed, the fit for HEC-RAS was 5–10% higher than that for AutoRoute. Despite being only slightly less accurate than HEC-RAS in its simulation of flood extent, AutoRoute was much simpler to set up and required less computational time to run.