VeinCast: Physics-Guided Dynamic Field Graphs with Graph-Conditioned Fusion for Global Medium-Range Weather Forecasting
2026-08-10 • Machine Learning
Machine LearningArtificial Intelligence
AI summaryⓘ
The authors developed VeinCast, a new weather forecasting model that predicts many atmospheric variables together by using a graph-based approach combining known physical relationships and data-driven updates. Their method improves how different weather factors interact dynamically in local regions and better fuses this information into predictions. They tested VeinCast on a global weather dataset and found it performs well compared to other leading models up to two weeks ahead. The authors also showed that combining physics-based guidance with data learning helps improve forecast accuracy.
medium-range weather forecastingatmospheric fieldsphysics-guided modelingdynamic graphERA5 datasetgraph neural networksdata-driven forecastinglatent fusiontop-K residual edges
Authors
Zhisheng Chen, Jinhan Li, Yuxuan Li, Yuan Gao, Hao Wu, Zheng Lu, Jinlong Du, Kun Wang, Bo An
Abstract
Global medium-range weather forecasting requires modeling structured yet state-dependent interactions among heterogeneous atmospheric fields. Existing data-driven models largely learn these interactions implicitly, whereas equation-level physical constraints may inherit approximation and model-form biases. We present VeinCast, a physics-guided dynamic field graph and graph-conditioned fusion framework that jointly forecasts 69 surface and upper-air fields. Within each local window, its Physics-Guided Dynamic Field Graph combines predefined atmospheric relations with state-dependent Top-K residual edges and adapts Earth-window attention using the resulting graph context. Graph-Conditioned Latent Fusion further employs graph context and source-node centrality to guide field-to-latent aggregation, while bounded feedback preserves field-specific information. On the $1.5^\circ$ ERA5 benchmark, VeinCast demonstrates competitive forecasting performance across all 69 meteorological fields at lead times of up to 14 days, compared with representative global weather forecasting models including FuXi, Pangu-Weather, GraphCast, FengWu, and ARROW. Ablations confirm that the two modules provide complementary gains, demonstrating the effectiveness of relational-level physical guidance for data-driven weather forecasting.