A framework for a probabilistic Height Above Nearest Drainage

Flooding is the number one cause of natural disaster fatalities worldwide. The Height Above the Nearest Drainage (HAND) is widely used in large-scale flood inundation modeling. While the high efficiency and computationally friendly of HAND, it results in a large error. We first evaluate the error and, compare it with a benchmark model which is a 2D hydraulic solver. We proposed a new alternative named distance to drainage to measure the distance of all cells to the drainage cells. Then we calculate the distribution of the HAND error given the distance to drainage. We find that the error follows the Laplacian distribution with different scale parameters depending on the distance to drainage. Finally, based on the fitted distribution, we construct two rasters that represent the upper and lower bound of the potential flood extent and depth under a certain confidence interval.