A parsimonious approach to estimate flood depths — also in urban areas — for satellite-based flood maps

Floods are among the most disruptive and widespread natural disaster that kill and displace millions of people every year. To cope with the impacts associated with ongoing floods, it is fundamental to acquire a rapid, accurate and comprehensive overview of inundated areas. Imageries sensed by satellites are becoming indispensable for this purpose, especially those acquired in the radar frequencies (SAR), as they can detect floods during the night and with cloudy skies. However, SAR-based remote sensing has serious limitations when it comes to flood mapping in urban and/or vegetated areas (because of low sensitivity or over-detection issues). Furthermore, satellite-based flood delineation does not provide any information on flood depths, which are critical for emergency response planning and for post-event impact evaluation. This contribution introduces a new framework to estimate water depth and to augment flood mapping where satellites cannot sense floodwater. As input, the method simply requires flood delineation (including the areas excluded from mapping because of the aforementioned limitations) and land surface topography. Although the framework is designed to be coupled with the recently release Global Flood Monitoring system of the Copernicus Emergency Management Service, its range of applicability is wide, provided that the basic input needs are met. The approach is especially suited to enhance flood mapping in systematic large-scale applications that require minimum supervision.