A Global Flood Inundation Database (GloFID)

Recent advancements in cloud-based computing and machine learning algorithms have ushered in a new era for water resource planning, flood preparedness, and management through high-resolution surface water maps and real-time monitoring of water body dynamics. These elements are crucial for informed decision-making, yet they remain unavailable across many regions worldwide, creating obstacles for effective water resource management.

Nonetheless, the synergy of cutting-edge technologies has provided a solution. The marriage of cloud-based computing and machine learning has enabled the efficient processing of extensive satellite data, granting unprecedented access to near-real-time surface water information. This triumph has transcended the limitations of existing databases that grapple with cloudy or extreme weather conditions and offer lower resolutions. The advent of Synthetic Aperture Radar (SAR)-based satellites has emerged as a game-changer, offering unparalleled spatial and temporal resolutions in generating surface water products.

In view of these revolutionary strides, the current study centers on the formulation of a global flood inundation database. This endeavor involves the meticulous creation of precise surface water maps and continuous monitoring of water body dynamics. Our approach involves harnessing the potential of the Sentinel-1 Synthetic Aperture Radar (SAR) satellite data repository, accessible through Google Earth Engine. The algorithm we have developed is fine-tuned to extract surface water extent from Sentinel-1 SAR data spanning from 2015 to the present. This technique allows us to generate surface water extent updates approximately every 6 days at the equator, with an impressive global spatial resolution of 10 meters.

Through a predominant reliance on the Sentinel-1 SAR dataset, we have successfully erected a software framework that engenders a global flood inundation database, encompassing river stretches exceeding 50 meters in width. This achievement has transcended the constraints of existing data sources, culminating in a comprehensive database rich in flood inundation insights. To address potential gaps in our flood inundation database, our algorithm seamlessly integrates 10-meter Sentinel-2 satellite data, thereby refining the temporal resolution of Sentinel-1 SAR data whenever cloud-free data is accessible.

Our flood inundation database stands in harmony with the SWOT Satellite River Database (SWORD), covering a substantial 214,000 river stretches. Users are presented with two distinct choices for delineating their area of interest. The Global Flood Inundation Database App presents an intuitive graphical interface that empowers users to select their specific region of interest. Alternatively, users can opt for SWORD river segments. Both alternatives expedite the generation of the inundation database, contingent on the user-specified time frame, predominantly spanning from 2015 to the present.