Floodbase, a key provider of flood intelligence and data for disaster management and re/insurance sectors, has developed an advanced AI technique that integrates satellite data with inputs from the U.S. National Water Model (NWM). The peer-reviewed study highlights the superior accuracy of Floodbase's large-area flood maps for California's 2023 atmospheric river, outperforming those created by NOAA's National Water Center.

This innovative approach supports more robust multi-sensor fusion, enabling Floodbase to incorporate additional data sources and further enhance the accuracy of its flood mapping.

"Flooding is the most common and costly weather-related disaster in the U.S.," said Dr. Beth Tellman, Co-Founder and Chief Science Officer at Floodbase. "This research builds on the already incredible capabilities of the National Water Center to detect and predict flooding over large land areas, and will enable more accurate flood monitoring, more affordable flood insurance, and more efficient disaster response across the U.S."

Tellman continued, "Flooding costs the Federal government between $180 and $496 billion each year. This rapid inundation mapping technology will enable the most vulnerable communities and sectors to better understand and mitigate their flood risks."

The study, published in 'Geophysical Research Letters', details Floodbase's use of deep learning to combine satellite observations with NWM outputs for near-real-time, high-resolution flood mapping. The deep learning model operates at a continental scale, covering large areas like the contiguous United States (CONUS) on an hourly basis over 45 years. These improvements in accuracy significantly enhance insurers' ability to assess flood risk, allowing them to offer more affordable flood insurance policies. Government agencies can also use this data to make better-informed policy decisions.

"Amid increasing billion-dollar disasters, it is critical that new and enhanced technologies be developed and deployed to predict, mitigate, and monitor flood events," said Dr. Jonathan Frame, the study's primary investigator, who led the research at Floodbase and now serves as an Assistant Professor at the University of Alabama's Department of Geological Sciences.

"In the era of big data, we are excited to use the latest technology to build on the work done by the National Water Center. This fusion allows us to significantly improve flood predictability, providing more accurate and timely information that will benefit policymakers, insurers, community leaders, and the nation as a whole."

Research Report:Rapid Inundation Mapping Using the US National Water Model, Satellite Observations, and a Convolutional Neural Network

Related Links
Floodbase
Bringing Order To A World Of Disasters
When the Earth Quakes
A world of storm and tempest