Evaluating Model Sensitivity of Various Data Input Resolutions for Large Scale Flood Inundation Modeling for Hurricane Helene, <scp>NC</scp>

ABSTRACT This study applies a statistically sound methodology for contrasting real‐valued datasets against simulated results generated from the US Army Engineer Research and Development Center's Coastal and Hydraulics Laboratory's first‐order flood inundation hydrologic modeling suite, AutoRoute, to determine the impact of spatial resolution on flood inundation extent, velocity, depth, and top‐width estimates. Nine AutoRoute simulations were conducted for Buncombe County, NC, for flooding events resulting from Hurricane Helene in September 2024. Each AutoRoute simulation was run with a unique combination of spatial resolutions for two of its primary input products: digital elevation model (DEM) and land cover (LC). The velocity, depth, and top‐width (VDT) outputs were tested for distributional nonoverlap by calculating Cliff's δ estimate, the Spatial Bhattacharyya Coefficient (SBC), and polygonal overlap with NASA‐JPL's Observational Products for End‐Users from Remote Sensing Analysis observed flood extents. The results confirm DEM products more heavily influence VDT output than LC, and lower input product resolution causes lower output velocity values and higher output top‐width values, potentially resulting in velocity underestimation and top‐width overestimation. This methodology offers an efficient and systematic evaluation of real‐valued outputs and can be applied to additional areas of interest, flood extents, flow scenarios, and input products of varying spatial resolutions.