A multi-sensor approach for increased measurements of floods and their societal impacts from space

Extreme hydrological events are short-lived. Therefore, merging observations from multiple satellites is necessary to ensure that they are consistently observed. Here we evaluate potential improvements to flood detectability afforded by combining data collected by Landsat, Sentinel-2, and Sentinel-1 for the first time globally. The enhanced temporal sampling increased the number of floods with at least 1 useful image (image with ≤20% clouds) from 7% for single sensors to up to 66% for a potential multi-sensor product. As dramatic as the increased coverage is, the socioeconomic impacts are even more tangible. In the pre-Sentinel era, only 22% of the total population displaced by flood events could benefit from having high-resolution images, whereas a current multi sensor product would serve 75% of the total displaced population. Additionally, the merged data could provide consistent observations of flood types rarely seen by single sensors and enable unprecedented insights into governing mechanisms of these events.