Validation of Urban Flood Inundation Models Applied Using Nationally Available Data Sets: Novel Analyses of Observed High Water Information

Abstract

This paper presents novel spatial analysis techniques to evaluate simulations of urban flood inundation from two hydrologic models applied using nationally available datasets. These techniques account for differences in model computational element size and discretization when comparing simulated flood depths to surveyed high-water marks. To complement direct evaluations of predicted depths at high-water marks, our techniques provide five additional metrics to assess modeled depths in areal sectors between the surveyed high-water mark and the stream channel. Our study also demonstrates a novel technique to evaluate flood predictions at damaged structures and crowd-sourced observations of flooded locations. The work in this study is part of a more complete evaluation of two hyper-resolution hydrologic models to generate street-level flood inundation predictions. We used the 679-km2 Sugar Creek watershed above USGS Gage 02146800, near Fort Mill, SC, which contains the Charlotte, NC, municipal area. We assessed model performance at 172 surveyed high-water marks, 373 locations of flooded structures, and nearly 2,000 crowd-sourced observations of flooded locations. Results indicate that the analyses techniques help distinguish model performance and identify model deficiencies.