Artificial intelligence pathways from weather to climate

2026-10-07

AI Summary

This study addresses the reliability limitations of AI-based weather forecasting when extended to climate scales, particularly under external forcing interventions and out-of-distribution extreme scenarios. To this end, it proposes a minimal requirements framework for AI climate modeling. Methodologically, two criteria are established: independent variation of external forcings and out-of-distribution robustness. By integrating autoregressive emulators, physics-AI hybrid models, and diverse downscaling techniques alongside long-step and target-variable strategies, the approach substantially reduces computational costs. The resulting framework achieves low-cost, high-accuracy predictions across multiple temporal scales, supports cross-cycle localized hazard assessments, and significantly shortens solution times. Ultimately, this work provides a reliable pathway for extending AI modeling from weather to climate applications.

Abstract

Deep learning has made rapid advances in weather forecasting: autoregressive models trained on atmospheric reanalyses now rival dynamical models across nowcasting, medium-range, and subseasonal-to-seasonal lead times, producing well-calibrated ensemble forecasts at reduced cost. We review these advances and consider their extension to climate horizons, where the challenge shifts from initial-condition skill to producing reliable statistical responses under altered forcings. AI-powered climate prediction systems must produce credible forced responses to drivers (e.g., greenhouse gases, land-use change) typically outside the observed record. We propose two minimum requirements for AI in climate modeling: (i) external forcing agents must enter explicitly enough to support interventions in which they vary independently; and (ii) robustness must be stress-tested in out-of-distribution regimes, including extremes and counterfactual trajectories. Using leading AI autoregressive emulators and hybrid physics-AI models, we identify development and coupling challenges. Comparing the reported throughput of these models with that of GPU-ported dynamical models highlights how AI can reduce time-to-solution by advancing only the target variables at the required resolution and using longer time steps, rather than integrating a full high-frequency, multivariate state. Diverse AI downscaling strategies can partially substitute for explicit fine-scale resolution, paving the way toward inexpensive local hazard assessment across prediction horizons.

Problem

• climate prediction
• deep learning
• external forcing
• out-of-distribution robustness
• autoregressive models

Innovation

• deep learning
• autoregressive models
• climate modeling
• out-of-distribution robustness
• AI downscaling