Seasonal Prediction with Neural GCM and Simplified Boundary Forcings: Large-scale Atmospheric Variability and Tropical Cyclone Activity

2025-04-30

AI Summary

To address insufficient predictive skill for seasonal-scale atmospheric variability and Northern Hemisphere tropical cyclone (TC) activity, this study proposes a physics-informed Neural Global Circulation Model (NeuralGCM) hybrid framework. It couples a neural GCM with a simplified boundary forcing scheme—comprising climatological states plus sea surface temperature and sea ice concentration anomalies—while omitting complex coupled initialization to achieve lightweight, interpretable subseasonal-to-seasonal prediction. Innovatively, this work introduces, for the first time, a physically driven boundary simplification strategy into NeuralGCMs for seasonal forecasting, bridging weather- and climate-timescale prediction. For forecasts issued from July to November, the model achieves TC frequency skill over the North Atlantic and eastern North Pacific comparable to state-of-the-art dynamical models, accurately reproducing large-scale circulation patterns and TC climatological distributions. It significantly enhances TC predictability and demonstrates strong potential for operational implementation.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Neuro-Symbolic Learning
• Planning, Routing, and Scheduling: Model-Based Reasoning

Application Category

• User Modeling, Personalization and Recommendation: Psychology-informed user models and recommender systems
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Social Networks and Social Media: Influence propagation, information diffusion, and the prediction on networks

Abstract

Machine learning (ML) models are successful with weather forecasting and have shown progress in climate simulations, yet leveraging them for useful climate predictions needs exploration. Here we show this feasibility using NeuralGCM, a hybrid ML-physics atmospheric model, for seasonal predictions of large-scale atmospheric variability and Northern Hemisphere tropical cyclone (TC) activity. Inspired by physical model studies, we simplify boundary conditions, assuming sea surface temperature (SST) and sea ice follow their climatological cycle but persist anomalies present at initialization. With such forcings, NeuralGCM simulates realistic atmospheric circulation and TC climatology patterns. Furthermore, this configuration yields useful seasonal predictions (July-November) for the tropical atmosphere and various TC activity metrics. Notably, the prediction skill for TC frequency in the North Atlantic and East Pacific basins is comparable to existing physical models. These findings highlight the promise of leveraging ML models with physical insights to model TC risks and deliver seamless weather-climate predictions.

Problem

• Leveraging ML models for seasonal climate predictions
• Predicting large-scale atmospheric variability and tropical cyclones
• Simplifying boundary conditions for accurate TC activity forecasts

Innovation

• Hybrid ML-physics model NeuralGCM for predictions
• Simplified boundary conditions with SST anomalies
• Seasonal prediction skill comparable to physical models