Hyper-Resolution Large Scale Flood Inundation Modeling: Development of AutoRAPID Model

Abstract:

Streamflow and the resultant flood inundation are defining elements in large scale flood analyses. High-fidelity predictive capabilities of flood inundation risk requires hydrologic and hydrodynamic modeling at hyper-resolution (<100 m) scales. Using spatiotemporal data from climate models as the driver, we couple a continental scale river routing model known as Routing Application for Parallel ComputatIon of Discharge (RAPID) with a regional scale flood delineation model called AutoRoute to estimate flood extents. We demonstrate how the coupled tool, referred to as AutoRAPID, can quickly and efficiently simulate flood extents using a high resolution dataset (~10 m) at the regional scale (> 100,000 km²). The AutoRAPID framework is implemented over 230,000 km² in the Midwestern United States (between latitude 38°N and 44°N, and longitude 86°W to 91°W, approximately 8% of the Mississippi River Basin) using a 10 m DEM. We generate the flood inundation map over the entire area for a June 2008 flood event. The model is compared with observed data at five select locations: Spencer, IN; Newberry, IN; Gays Mills, WI; Ft. Atkinson, WI, and Janesville, WI. We show that the model results are generally satisfactory with observed flow and flood inundation data and suggest that the AutoRAPID model can be considered for several potential applications, such as: forecast flow and flood inundation information; generating flood recurrence maps using high resolution vector river data; and for emergency management applications to protect/evacuate large areas when time is limited and data are sparse.