Neural General Circulation Models for Weather and Climate

TL;DR

This paper introduces NeuralGCM, a neural network-based general circulation model that combines differentiable atmospheric dynamics with machine learning, achieving competitive weather and climate forecasts with significant computational efficiency.

Contribution

It presents the first GCM integrating a differentiable solver with ML components, enabling accurate deterministic and ensemble weather and climate predictions.

Findings

NeuralGCM performs comparably to state-of-the-art ML and physics-based models for short-term forecasts.

It accurately tracks climate metrics like global mean temperature over decades.

The model captures emergent phenomena such as tropical cyclone trajectories.

Abstract

General circulation models (GCMs) are the foundation of weather and climate prediction. GCMs are physics-based simulators which combine a numerical solver for large-scale dynamics with tuned representations for small-scale processes such as cloud formation. Recently, machine learning (ML) models trained on reanalysis data achieved comparable or better skill than GCMs for deterministic weather forecasting. However, these models have not demonstrated improved ensemble forecasts, or shown sufficient stability for long-term weather and climate simulations. Here we present the first GCM that combines a differentiable solver for atmospheric dynamics with ML components, and show that it can generate forecasts of deterministic weather, ensemble weather and climate on par with the best ML and physics-based methods. NeuralGCM is competitive with ML models for 1-10 day forecasts, and with the European Centre for Medium-Range Weather Forecasts ensemble prediction for 1-15 day forecasts. With prescribed sea surface temperature, NeuralGCM can accurately track climate metrics such as global mean temperature for multiple decades, and climate forecasts with 140 km resolution exhibit emergent phenomena such as realistic frequency and trajectories of tropical cyclones. For both weather and climate, our approach offers orders of magnitude computational savings over conventional GCMs. Our results show that end-to-end deep learning is compatible with tasks performed by conventional GCMs, and can enhance the large-scale physical simulations that are essential for understanding and predicting the Earth system.