![[2602.22293] Global River Forecasting with a Topology-Informed AI Foundation Model](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.22293] Global River Forecasting with a Topology-Informed AI Foundation Model
Summary
The paper presents GraphRiverCast (GRC), a topology-informed AI model designed for global river forecasting, enabling robust hydrodynamic simulation without historical data reliance.
Why It Matters
This research addresses the challenge of data scarcity in hydrology by introducing a model that enhances predictive accuracy and efficiency in river systems. It bridges the gap between global hydrodynamic knowledge and local realities, which is crucial for effective water resource management and environmental monitoring.
Key Takeaways
- GraphRiverCast (GRC) simulates river hydrodynamics using topology-informed AI.
- GRC operates effectively in 'ColdStart' mode, predicting without historical data.
- The model outperforms traditional physics-based and locally-trained AI methods.
- Topological encoding is essential for accurate flow dynamics reconstruction.
- GRC establishes a new paradigm for integrating global and local hydrological knowledge.
Computer Science > Machine Learning arXiv:2602.22293 (cs) [Submitted on 25 Feb 2026] Title:Global River Forecasting with a Topology-Informed AI Foundation Model Authors:Hancheng Ren, Gang Zhao, Shuo Wang, Louise Slater, Dai Yamazaki, Shu Liu, Jingfang Fan, Shibo Cui, Ziming Yu, Shengyu Kang, Depeng Zuo, Dingzhi Peng, Zongxue Xu, Bo Pang View a PDF of the paper titled Global River Forecasting with a Topology-Informed AI Foundation Model, by Hancheng Ren and 13 other authors View PDF HTML (experimental) Abstract:River systems operate as inherently interconnected continuous networks, meaning river hydrodynamic simulation ought to be a systemic process. However, widespread hydrology data scarcity often restricts data-driven forecasting to isolated predictions. To achieve systemic simulation and reduce reliance on river observations, we present GraphRiverCast (GRC), a topology-informed AI foundation model designed to simulate multivariate river hydrodynamics in global river systems. GRC is capable of operating in a "ColdStart" mode, generating predictions without relying on historical river states for initialization. In 7-day global pseudo-hindcasts, GRC-ColdStart functions as a robust standalone simulator, achieving a Nash-Sutcliffe Efficiency (NSE) of approximately 0.82 without exhibiting the significant error accumulation typical of autoregressive paradigms. Ablation studies reveal that topological encoding serves as indispensable structural information in the absence of his...
