Abstract

This study evaluates how elevation and hydrography inputs influence flood extent accuracy using ARC (Automated Rating Curve), a raster-based flood inundation model. It focuses on how spatial resolution and vertical accuracy, reflecting the satellite mission from which each dataset was derived, and bare-earth processing of the DEM affect the accuracy of predicted flood extents. The assessment compares four dataset combinations--FABDEM with GEOGLOWS River Forecasting System Version 1(RFS1), FABDEM with GEOGLOWS River Forecasting System Version 2(RFS2), SRTM with RFS1, and SRTM with RFS2--across five study locations: the Suncook River in New Hampshire, North Fork Kentucky River in Kentucky, Nimishillen Creek in Ohio, South Platte River in Colorado, and Flathead River in Montana. Each location was selected for analysis of flood model performance based on the various inputs chosen across different terrain conditions, with simulations conducted for four Return Periods: 2-year, 10-year, 50-year, and 100-year. Input streamflow data were sourced directly from reference flood maps developed by the U.S. Geological Survey (USGS) in collaboration with local agencies. These maps are based on recorded flood events and incorporate high-water marks, stream gauge data, and hydraulic model calibration to represent observed inundation extents. Elevation and hydrography datasets affect the credibility of flood inundation maps. We found that newer global datasets derived from satellite missions--offering better vertical accuracy, improved spatial resolution, and bare-earth DEM products--produced flood maps more closely resembling observed flood extents. Specifically, the FABDEM-RFS2 pairing, derived from the TanDEM-X mission, consistently outperformed the SRTM-RFS1 baseline. This improvement was quantified by higher F-statistics, indicating stronger agreement with reference flood maps. These results demonstrate that input data's vertical accuracy and hydrologic conditioning are important to model performance. Overall, the findings confirm that careful selection of elevation and hydrography inputs is important for credible flood modeling. The FABDEM-RFS2 combination, when used with low-complexity models such as ARC, provides a globally applicable framework for flood inundation mapping and can serve as a practical toolkit for countries seeking to generate flood maps anywhere in the world from anywhere in the world using freely available data.

Degree

College and Department

Ira A. Fulton College of Engineering; Civil and Environmental Engineering

Rights

https://lib.byu.edu/about/copyright/