Estimating ensemble likelihoods for the Sentinel-1 based Global Flood Monitoring product of the Copernicus Emergency Management Service

• DOI: 10.36227/techrxiv.22688101
• Publication Date: 2023-04-28T19:41:50Z
• Authors: Christian Krullik...; Candace Chow; Marc Wieland; Sandro Martinis; Bernhard Bauer-Ma...; Florian Roth; Patrick Matgen; Marco Chini; Yu Li; Peter Salamon

<p>The Global Flood Monitoring (GFM) system of the Copernicus Emergency Management Service (CEMS) addresses challenges and impacts that are caused by flooding. The GFM provides global, near-real time flood extent masks for each newly acquired Sentinel-1 Interferometric Wide Swath Synthetic Aperture Radar (SAR) image, as well information from whole archive 2015 on. is an ensemble product based on a combination three independently developed mapping algorithms individually derive data. Each algorithm also classification uncertainty aggregated into likelihood mean individual likelihoods. As detection with different methods, value range input likelihoods must be harmonized to low [0] high [100] likelihood. evaluated two test sites in Myanmar Somalia, showcasing performance during actual event area challenging conditions SAR-based detection. use case demonstrates robustness if detections step disagree how communicated end-user. Somalia setting where misclassifications likely, process mitigates false can interpreted such results adequate caution. <br> </p>