A new approach for deriving Flood hazard maps from SAR data and global hydrodynamic models

Abstract:

With the flood consequences likely to amplify because of the growing population and ongoing accumulation of assets in flood-prone areas, global flood hazard and risk maps are needed for improving flood preparedness at large scale. At the same time, with the rapidly growing archives of SAR images of floods, there is a high potential of making use of these images for global and regional flood management. In this framework, an original method that integrates global flood inundation modeling and microwave remote sensing is presented. It takes advantage of the combination of the time and space continuity of a global inundation model with the high spatial resolution of satellite observations. The availability of model simulations over a long time period offers opportunities for estimating flood non-exceedance probabilities in a robust way. These probabilities can be attributed to historical satellite observations. Time series of SAR-derived flood extent maps and associated non-exceedance probabilities can then be combined generate flood hazard maps with a spatial resolution equal to that of the satellite images, which is most of the time higher than that of a global inundation model. In principle, this can be done for any area of interest in the world, provided that a sufficient number of relevant remote sensing images are available.

As a test case we applied the method on the Severn River (UK) and the Zambezi River (Mozambique), where large archives of Envisat flood images can be exploited. The global ECMWF flood inundation model is considered for computing the statistics of extreme events. A comparison with flood hazard maps estimated with in situ measured discharge is carried out. The first results confirm the potentiality of the method. However, further developments on two aspects are required to improve the quality of the hazard map and to ensure the acceptability of the product by potential end user organizations. On the one hand, it is of paramount importance to improve the skill of the global inundation model for reliably estimating flood occurrence probabilities, which is critical since errors in the model are directly transferred to the resulting hazard map. On the other hand, the spatial and temporal sampling of the satellite observations needs to be improved in order to provide more detailed and more reliable maps.