WSSM: Geographic-enhanced hierarchical state-space model for global station weather forecast

2025-01-20

AI Summary

To address insufficient prediction accuracy for extreme weather events across global meteorological stations, this paper proposes Geo-HSSM—a geography-enhanced hierarchical state-space model. Methodologically, it introduces geographic coordinate embedding into the state-space architecture to explicitly model absolute spatiotemporal locations; designs a hierarchical dynamic mechanism jointly capturing seasonal, diurnal, and extreme-event timescales; and constructs a multi-scale time-frequency feature pyramid to overcome the long-range modeling limitations of standard Mamba on meteorological sequences. Evaluated on the Weather-5K subset, Geo-HSSM achieves state-of-the-art performance: overall forecasting error is reduced by 12.7%, and F1-scores for extreme precipitation and heatwave events improve by 19.3%. These advances significantly enhance the predictability of fine-grained, high-impact weather phenomena.

Technology Category

• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Data Mining & Knowledge Management: Mining of Spatial, Temporal or Spatio-Temporal Data

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Global Station Weather Forecasting (GSWF), a prominent meteorological research area, is pivotal in providing timely localized weather predictions. Despite the progress existing models have made in the overall accuracy of the GSWF, executing high-precision extreme event prediction still presents a substantial challenge. The recent emergence of state-space models, with their ability to efficiently capture continuous-time dynamics and latent states, offer potential solutions. However, early investigations indicated that Mamba underperforms in the context of GSWF, suggesting further adaptation and optimization. To tackle this problem, in this paper, we introduce Weather State-space Model (WSSM), a novel Mamba-based approach tailored for GSWF. Geographical knowledge is integrated in addition to the widely-used positional encoding to represent the absolute special-temporal position. The multi-scale time-frequency features are synthesized from coarse to fine to model the seasonal to extreme weather dynamic. Our method effectively improves the overall prediction accuracy and addresses the challenge of forecasting extreme weather events. The state-of-the-art results obtained on the Weather-5K subset underscore the efficacy of the WSSM

Problem

• Global Meteorological Stations
• Extreme Weather Prediction
• Accuracy Improvement

Innovation

• WSSM
• Geographic Information Integration
• Extreme Weather Prediction