Rapid flooded area detection by applying Robust Satellite Techniques on different geostationary data

Facing the impact of floods by remote sensing data requires both satellite systems able to frequently observe soil conditions and robust processing algorithms capable to provide reliable information on surface changes. Only in this way an effective contribution to the management of flood risk can be carried out. In particular, geostationary satellites might give an useful support in timely detecting areas affected by floods. Their very high temporal resolution (presently up to 5 minutes) allows, in fact, to monitor in real time soil conditions and variations. Moreover, such a frequent observing capability maximises the chance of acquiring a cloud-free view of the land surface, the main limit when optical data are used for flood detection and monitoring from space. An advanced satellite data analysis methodology, named RST (Robust Satellite Techniques), has been recently implemented and successfully applied by using visible and near infrared Advanced Very High Resolution Radiometer (AVHRR) and Moderate Resolution Imaging Spectroradiometer (MODIS) data for providing reliable and accurate indication about flooded area presence and evolution in daytime. In this work, the RST technique has been further tested and assessed using two different sensors onboard geostationary satellites. In detail, the Berg river flood occurred in the South Africa’s Western Cape province in late July and early August 2007 was analyzed by means of Meteosat Second Generation (MSG) Spinning Enhanced Visible and InfraRed Imager (SEVIRI) data. Multi-functional Transport Satellite 1R (MTSAT-1R) data have been instead analyzed to assess their potential in tracking the effects of the huge tsunami which hit eastern coasts of the Honshu Island (Japan) as a consequence of the March 11, 2011 Tohoku earthquake. The analysis of the results, shown in this paper, highlights the potential of RST in furnishing reliable and updated information on flood presence and evolution also when implemented on satellite data at a coarse spatial resolution like the geostationary ones.