AI and the World's Hard Problems · Earth & climate science · arXiv · 2026-10-21
Learning global river system hydrodynamics with a topology-informed foundation model
Accurate river prediction is essential for water, food, and energy security, yet extending it reliably from sparse in-situ gauges to the full river network remains an open challenge. GraphRiverCast (GRC) is a topology-informed river foundation model that exploits river connectivity and energetics, is trained on sparse gauges, and predicts daily multivariate hydrodynamics at reach level across the global river network without prior river conditions, outperforming strong physics-based and machine-learning models. Because it reconstructs river states from topology and forcing rather than relying on temporal recurrence, its skill barely decays with lead time even far beyond its training horizon, and it generalises to ungauged reaches and finer-resolution networks, bringing reliable hydrodynamic prediction to poorly monitored regions.