Zhouyayan Li, Felipe Quintero Duque, Trevor Grout, Bradford Bates, Ibrahim Demir

For flood inundation extent prediction, it is important to have a faster, more accurate, and input-parsimonious model during response and recovery efforts. Height Above Nearest Drainage (HAND) is a simplified conceptual model whose efficacy and utility have been demonstrated in previous studies. This study aims to provide a comprehensive assessment of prediction performance of the rating-curve-based HAND generated with the framework adopted at NOAA's National Water Center and the non-rating-curve-based HAND inundation maps created with a web-based flood inundation mapping system. The study presents an in-depth analysis on the performance of HAND with varying model configurations, conditions where the HAND fails to provide accurate predictions, and underlying mechanism and guideline to overcome these challenges. The study also includes analysis of the model performance with bathymetry-based measurements. The results show that in areas where the water depth indicated by the synthetic rating curve are relatively consistent with those in catchments, the non-rating-curve-based HAND can generate comparable inundation extent predictions with fewer inputs. Otherwise, the non-rating-curve-based HAND may result in significant underestimations due to a combination of factors. The underestimation can be reduced by using a multi-depth technique to calculate water depth. Furthermore, the results show that the optimal HAND threshold is a percentage ranging between 8% and 12% of the basin drainage area, rather than a specific number as reported in previous studies. In comparison to the single-depth approach, the results show that proposed multi-point water depth calculation approaches are more robust against the causes of underestimation. However, there are no notable differences in prediction performance between proposed multi-point approaches. Finally, bathymetry measurements cause underestimation by increasing HAND values for non-drainage pixels. As a result, they should be handled with caution, as underestimation is riskier than overestimation when it comes to flood preparedness.