No Epoch Like the Present: Robust Climate Emulation Requires Out-of-Distribution Generalisation

2026-05-21

AI Summary

This study addresses the limited robustness of current climate emulators when confronted with out-of-distribution (OOD) data induced by climate change, a challenge inadequately captured by standard evaluation protocols. To better assess future generalization, the work innovatively leverages seasonal variation as a zero-cost proxy for real-world distribution shifts and introduces a novel OOD evaluation framework. It further proposes a physics-informed model decomposition strategy to enhance compositional generalization. Experiments reveal that state-of-the-art hybrid machine learning climate emulators suffer significant performance degradation under seasonal distribution shifts. In contrast, the proposed physics-driven decomposition achieves substantially improved OOD robustness with only minor sacrifices in in-domain accuracy, offering a new paradigm for reliable climate simulation under future climatic conditions.

Abstract

Climate emulation is an out-of-distribution (OOD) projection task. This is precisely the challenge where modern Machine Learning (ML) methods are most prone to failure. Consequently, while current ML emulators trained on present climate achieve high in-distribution performance, their future reliability under the inevitable distribution shifts of a changing climate remains a critical, poorly understood blind spot. Addressing this challenge requires a fundamental shift in how we understand, evaluate, and design climate emulators. In this work, we first confirm that climate change drives a statistically significant and progressively growing shift in atmospheric state distributions, rendering standard evaluation protocols insufficient. We empirically establish that seasonal variation serves as an effective proxy for these long-term climate shifts, providing access to $\textit{real-world}$ distribution shifts without recourse to heuristics like synthetic perturbations. Motivated by this link, we introduce a novel evaluation framework that leverages seasonal shifts as a rigorous, zero-overhead testbed for emulator robustness. Our systematic characterisation confirms that current state-of-the-art hybrid-ML emulators degrade significantly under these realistic shifts. Finally, we chart a path forward by identifying compositional generalisation, the ability to form novel combinations from observed elementary components, as a principled route towards robust climate emulation. We demonstrate that physically motivated decompositions substantially improve OOD performance with only modest trade-offs against in-distribution performance, providing an avenue towards ML-driven climate emulators robust to an unknown future.

Problem

• out-of-distribution generalisation
• climate emulation
• distribution shift
• robustness
• climate change

Innovation

• out-of-distribution generalization
• climate emulation
• compositional generalization
• seasonal shift proxy
• robust machine learning