REFLEX—A novel method for the rapid estimation of flood extent

Abstract

Reliable flood hazard mapping is crucial for enhancing flood preparedness, especially in poorly gauged and vulnerable areas. Yet, current flood models are either highly detailed and computationally intensive or oversimplified, failing to capture important flow dynamics. Here, we present the Rapid Estimation of FLood EXtent (REFLEX), a hydro-geomorphological model for flood hazard mapping at multiple scales that improves upon the existing Height Above the Nearest Drainage model by constraining flood extent using physically-coherent flood volumes. Input volume is distributed from channel to hillslope cells by using an iterative optimization based on the Manning's equation, bypassing the need for accurate rating curves. The model improves the accuracy of inundation extent in flat areas by accounting for backwater effects, and in coastal zones by extending the calculation boundaries beyond the direct watershed area. REFLEX was tested over five areas with different climatic conditions, including flash-flood prone catchments in Italy and large floodplains in the Zambezi river basin. Results denote skillful representation of flooded areas in the Italian catchments, with critical success index (CSI) up to 0.62. In the African cases, REFLEX produced the best estimates of the flood extent and CSI comparable with two state of the art global inundation models. REFLEX is a competitive flood modelling tool for large scale and high resolution applications thanks to its limited input requirements and computing times two orders of magnitude shorter than equivalent hydraulic simulations.