Multi-Fidelity Graph Neural Networks for Efficient and Accurate Flood Hazard Mapping

This paper presents a multi-fidelity graph neural network framework for high-resolution flood hazard mapping on unstructured meshes. By combining coarse and fine simulation data, the approach targets one of the main bottlenecks in operational flood analysis: producing accurate large-domain inundation estimates without paying the full computational cost of repeated high-fidelity hydrodynamic simulations.

Recommended citation: Taghizadeh, M., Zandsalimi, Z., Shafiee-Jood, M., & Alemazkoor, N. (2025). "Multi-Fidelity Graph Neural Networks for Efficient and Accurate Flood Hazard Mapping." Environmental Modelling & Software.