FloodForecaster: A Domain-Adaptive Geometry-Informed Neural Operator Framework for Rapid Flood Forecasting

This paper presents FloodForecaster, a time-dependent geometry-informed neural operator framework for rapid flood forecasting over irregular domains. It combines graph neural operators for complex terrain, Fourier neural operators for long-range dynamics, and a gradient-reversal-based domain adaptation strategy that transfers effectively to new river segments while preserving source-domain knowledge and remaining data-efficient.

Recommended citation: Taghizadeh, M., Zandsalimi, Z., Nabian, M. A., Goodall, J. L., & Alemazkoor, N. (2025). "FloodForecaster: A Domain-Adaptive Geometry-Informed Neural Operator Framework for Rapid Flood Forecasting." Journal of Hydrology.