Accounting for uncertainty in real-time flood inundation mapping using HAND model: Iowa case study

This study examines the uncertainties associated with a rapid flood inundation mapping approach based on height above the nearest drainage (HAND) model's parameters, inputs, reference, and computing procedure. The effects of changing the drainage threshold, water depth, and horizontal resolution were studied in two Iowan regions of around 100 km2. To assess the uncertainty from rasterized inundation extent maps used as the reference, a regional and community-level evaluation with three distinct water depth estimates was performed for a five-county region in Iowa. An inner zone approach was utilized to demonstrate the uncertainties around the computational boundary of regions. The results reveal that the HAND model's performance depends on the research area's morphological characteristics, even though they are not directly input into the calculation. A good parameter set for one site does not ensure good performance for another. Moreover, the findings suggest that the DEM's resolution should be carefully chosen, as overgeneralization reduces the reliability of forecasts. Finally, the HAND model relies on drainage accumulation to build the simulation network; therefore, the calculation around the region border can introduce uncertainty. The findings of this study provide guidelines for effective model setup, adjustments, and performance improvement for non-rating-curve-based HAND.