Document Type


Journal/Book Title/Conference

Water Resources Research



Publication Date


First Page


Last Page



Recent floods from intense storms in the southern United States and the unusually active 2017 Atlantic hurricane season have highlighted the need for real‐time flood inundation mapping using high‐resolution topography. High‐resolution topographic data derived from lidar technology reveal unprecedented topographic details and are increasingly available, providing extremely valuable information for improving inundation mapping accuracy. The enrichment of terrain details from these data sets, however, also brings challenges to the application of many classic approaches designed for lower‐resolution data. Advanced methods need to be developed to better use lidar‐derived terrain data for inundation mapping. We present a new workflow, GeoFlood, for flood inundation mapping using high‐resolution terrain inputs that is simple and computationally efficient, thus serving the needs of emergency responders to rapidly identify possibly flooded locations. First, GeoNet, a method for automatic channel network extraction from high‐resolution topographic data, is modified to produce a low‐density, high‐fidelity river network. Then, a Height Above Nearest Drainage (HAND) raster is computed to quantify the elevation difference between each land surface cell and the stream bed cell to which it drains, using the network extracted from high‐resolution terrain data. This HAND raster is then used to compute reach‐average channel hydraulic parameters and synthetic stage‐discharge rating curves. Inundation maps are generated from the HAND raster by obtaining a water depth for a given flood discharge from the synthetic rating curve. We evaluate our approach by applying it in the Onion Creek Watershed in Central Texas, comparing the inundation extent results to Federal Emergency Management Agency 100‐yr floodplains obtained with detailed local hydraulic studies. We show that the inundation extent produced by GeoFlood overlaps with 60%~90% of the Federal Emergency Management Agency floodplain coverage demonstrating that it is able to capture the general inundation patterns and shows significant potential for informing real‐time flood disaster preparedness and response.


This article was corrected on 15 FEB 2019.