Normalized Difference Flood Index for rapid flood mapping: Taking advantage of EO big data

Climate change projections foresee an increasing number of intense precipitation events with consequent flash and riverine floods. An accurate and rapid mapping of these phenomena is a key component of effective emergency management and disaster risk reduction plans. Earth Observation big data such as the ones acquired by the Copernicus programme, are providing unprecedented opportunities to detect changes and assess economic impacts in case of disasters. This paper presents an innovative flood mapping technique based on an index which is computed using multi temporal statistics of Synthetic Aperture Radar images. The index compares a large amount of reference scenes to those acquired during the investigated flood and allows an easy categorization of "flooded" areas; either areas solely temporarily covered by water or areas with mixed water and vegetation. The method has been developed specifically to exploit Sentinel-1 data but can be applied to any other sensor. It has been tested for the 2010 flood of Veneto (Italy) and the floods of 2015 in Malawi and Uganda. Extensive qualitative analysis and cross-comparison with other state-of-the art methods, proved the proposed approach highly reliable and particularly effective, allowing a precise, simple and fast flood mapping. Compared to the maps produced for emergency management for the event analyzed, we obtained an overall agreement of 96.7% for Malawi and an average of 96.5% for Veneto for the 5 maps presented.