Rapid Flood Inundation Forecast Using Fourier Neural Operator
This work addresses the lack of real-time flood forecast tools for emergency planning, though it is incremental as it builds on existing methods like FNO and U-Net.
The authors tackled the problem of real-time flood inundation forecasting by developing a hybrid process-based and data-driven approach using the Fourier neural operator (FNO) as a surrogate model, which outperformed a baseline U-Net model and maintained high predictability up to 3-hour lead times with strong generalization to new sites.
Flood inundation forecast provides critical information for emergency planning before and during flood events. Real time flood inundation forecast tools are still lacking. High-resolution hydrodynamic modeling has become more accessible in recent years, however, predicting flood extents at the street and building levels in real-time is still computationally demanding. Here we present a hybrid process-based and data-driven machine learning (ML) approach for flood extent and inundation depth prediction. We used the Fourier neural operator (FNO), a highly efficient ML method, for surrogate modeling. The FNO model is demonstrated over an urban area in Houston (Texas, U.S.) by training using simulated water depths (in 15-min intervals) from six historical storm events and then tested over two holdout events. Results show FNO outperforms the baseline U-Net model. It maintains high predictability at all lead times tested (up to 3 hrs) and performs well when applying to new sites, suggesting strong generalization skill.