Monitoring hurricane damage is essential to improving damage predictions in storm models and learning how to avoid or reduce future storm impacts. However, recording hurricane damage in the field can be dangerous and can take resources away from recovery activities. Remote damage assessment offers a safe and fast way to acquire building damage information after storms.
NOAA-funded researchers published a study that describes the use of Artificial Intelligence for faster, more accurate, large-scale assessments of building damage from hurricane-related wind and flooding. The new approach combines existing wind and flooding models with remote Interferometric Synthetic Aperture Radar (InSAR) imagery. InSAR remote sensing imagery penetrates clouds, unlike traditional optical satellite imagery, providing timely and complete imagery of a region regardless of weather conditions. Next, Artificial Intelligence rapidly analyzes all of the datasets to identify crucial information for storm response and recovery.
The team used field data collected from Hurricane Ian (2022) to compare their approach to traditional damage prediction methods and found that their approach was better at predicting building damage and that it reduced processing times by as much as 83.8 percent compared to traditional methods. The team’s approach also bypasses the need for onsite building damage data collection, offering a safer method for estimating building damage after a storm.
The project is led by the University of Florida and is funded through the NCCOS Effects of Sea Level Rise Program. Project partners include Johns Hopkins University, Florida Gulf Coast University, the South Florida Water Management District, the Rookery Bay National Estuarine Research Reserve, the U.S. Geological Survey, and Riada Engineering.
This work is authorized by the NOAA Authorization Act of 1992, Pub. L. 102-567 (Oct. 29, 1992); sec. 201(c), which directs appropriation for the NCCOS Competitive Research Program to augment and integrate existing NOAA programs, with a specific focus on improving predictions of coastal hazards and protecting human life and property.
Citation: Wang C., Y. Liu, X. Zhang, X. Li, V. Paramygin, P. Sheng, X. Zhao, S. Xu. 2024. Scalable and rapid building damage detection after hurricane Ian using causal Bayesian networks and InSAR imagery. Int. J. Disaster Risk Reduct., Article 104371. https://doi.org/10.1016/j.ijdrr.2024.104371