Enhancing Operational Flood Detection Solutions through an Integrated Use of Satellite Earth Observations and Numerical Models

Among natural disasters floods are the most common and widespread hazards worldwide (CRED and UNISDR, 2018). Thus, making communities more resilient to flood is a priority, particularly in large flood-prone areas located in emerging countries, because the effects of extreme events severely setback the development process (Wright, 2013). In this context, operational flood preparedness requires novel modeling approaches for a fast delineation of flooding in riverine environments. Starting from a review of advances in the flood modeling domain and a selection of the more suitable open toolsets available in the literature, a new method for the Rapid Estimation of FLood EXtent (REFLEX) at multiple scales (Arcorace et al., 2019) is proposed. The simplified hydraulic modeling adopted in this method consists of a hydro-geomorphological approach based on the Height Above the Nearest Drainage (HAND) model (Nobre et al., 2015). The hydraulic component of this method employs a simplified version of fluid mechanic equations for natural river channels. The input runoff volume is distributed from channel to hillslope cells of the DEM by using an iterative flood volume optimization based on Manning’s equation. The model also includes a GIS-based method to expand HAND contours across neighbor watersheds in flat areas, particularly useful in flood modeling expansion over coastal zones. REFLEX’s flood modeling has been applied in multiple case studies in both surveyed and ungauged river basins. The development and the implementation of the whole modeling chain have enabled a rapid estimation of flood extent over multiple basins at different scales. When possible, flood modeling results are compared with reference flood hazard maps or with detailed flood simulations. Despite the limitations of the method due to the employed simplified hydraulic modeling approach, obtained results are promising in terms of flood extent and water depth. Given the geomorphological nature of the method, it does not require initial and boundary conditions as it is in traditional 1D/2D hydraulic modeling. Therefore, its usage fits better in data-poor environments or large-scale flood modeling. An extensive employment of this slim method has been adopted by CIMA Research Foundation researchers for flood hazard mapping purposes over multiple African countries. As collateral research, multiple types of Earth observation (EO) data have been employed in the REFLEX modeling chain. Remotely sensed data from the satellites, in fact, are not only a source to obtain input digital terrain models but also to map flooded areas. Thus, in this work, different EO data exploitation methods are used for estimating water extent and surface height. Preliminary results by using Copernicus’s Sentinel-1 SAR and Sentinel-3 radar altimetry data highlighted their potential mainly for model calibration and validation. In conclusion, REFLEX combines the advantages of geomorphological models with the ones of traditional hydraulic modeling to ensure a simplified steady flow computation of flooding in open channels. This work highlights the pros and cons of the method and indicates the way forward for future research in the hydro-geomorphological domain.