Author(s): Seth Bryant, University of Wisconsin Madison (First Author, Presenting Author) Saurabh Kaushik, University of Wisconsin Madison Jonathan Sullivan, University of Wisconsin Madison Beth Tellman, University of Arizona
Here we present a free and open, nationwide database of flood inundation footprints for roughly 10,000 U.S. storms. We first screen NOAA storm records, gather satellite images from six radar- and optical-based sensors and feed this data into a deep-learning model that flags flooded locations, even through clouds. This provides “maximum flood” footprints for each event, giving cities, researchers, and planners a sharper picture of real-world flooding so they can test models, study compounding risks, allocate disaster response resources, and design better protections.
