Rapid inundation mapping using the US National Water Model, satellite observations, and a convolutional neural network

Convolution neural networks (CNN) are suitable for rapid modeling of surface water dynamics for large-scale inundation mapping.• We deploy a CNN for continuous flood mapping across all of California during the devastating 2023 atmospheric river (AR) events.• Inundation extent across Sacramento is more accurately predicted with CNN than the Height Above Nearest Drainage (HAND).