Disentangling Forced and Internal Climate Variability in Single Realizations using Dynamic Mode Decomposition with Control

2026-07-22 · Source: stat.ML updates on arXiv.org · Field: Science & Research — Mathematics & Computational Sciences, Environmental Science & Earth Systems · Depth: Expert, quick

Summary

PullbackDMDc, a novel method introduced in arXiv:2607.18298, decomposes a single climate realization into its forced and internal components. This approach addresses a core challenge in climate science by disentangling the forced climate response from internal variability within a single observed record. Grounded in non-autonomous dynamical systems theory and Dynamic Mode Decomposition with Control (DMDc), PullbackDMDc incorporates pullback attractor estimation, treating external forcing as a dynamical driver within a linear stochastic system. Unlike Linear Inverse Models (LIMs) which ignore forcing, or linear regression which omits climate system dynamics, PullbackDMDc provides a physically interpretable picture of underlying dynamics. The method was applied to near-surface air temperature and sea-level pressure from reanalysis and four Earth System Model (ESM) large ensembles. Results show PullbackDMDc estimates the forced response with skill matching or exceeding established baselines and identifies optimal forcing predictors. Its internal variability components reveal ESMs qualitatively capture interannual and decadal modes, though with systematic differences.

Key takeaway

For research scientists analyzing single climate realizations or evaluating Earth System Models, PullbackDMDc offers a powerful new tool. You can use this method to accurately disentangle forced climate responses from internal variability, improving climate projection and attribution. Integrating PullbackDMDc into your analysis workflow will provide a physically interpretable picture of underlying dynamics and help identify optimal forcing predictors, enhancing model evaluation and understanding of climate system behavior.

Key insights

PullbackDMDc disentangles forced and internal climate variability from single observations using dynamic system theory.

Principles

• External forcing acts as a dynamical driver.
• Pullback attractor theory aids climate decomposition.
• Climate system dynamics are crucial for accuracy.

Method

PullbackDMDc applies DMDc and pullback attractor estimation to decompose single climate realizations into spatial modes, identifying forced and internal components by treating external forcing as a dynamical driver.

In practice

• Evaluate Earth System Models (ESMs) effectively.
• Estimate forced climate responses accurately.
• Identify optimal forcing predictors.

Topics

• Dynamic Mode Decomposition
• Climate Variability
• Earth System Models
• Climate Forcing
• Pullback Attractors
• Machine Learning

Best for: AI Scientist, Research Scientist

Related on AIssential

• PyDMD: A Python Package for Dynamic Mode Decomposition (DMD) — PyDMD enables data-driven modeling of complex systems by decomposing time-varying data into dominant spatio-temporal modes.
• Learning the coupled dynamics of global climate modes — UniCM unifies global climate mode forecasting by learning coupled ocean-atmosphere dynamics and inter-mode interactions, revealing emergent predictability.
• Uncertainty quantification via conformal prediction in data assimilation — Conformal Prediction offers a promising machine learning approach for uncertainty quantification in data assimilation, complementing traditional ensemble methods.
• LFNO: Bridging Laplace and Fourier via Transient-Steady Decomposition — LFNO unifies Laplace and Fourier operators to model dynamical systems across transient and steady-state regimes.
• The Granularity Paradox: How Temporal Disaggregation Inflates In-Sample Fit and Compounds Out-of-Sample Error — Finer temporal disaggregation improves in-sample fit but degrades out-of-sample accuracy due to recursive error compounding.
• REGAIN: REconciliation GAIN-driven Auxiliary Direction Learning — REGAIN optimizes downstream forecast reconciliation by learning auxiliary directions that reduce unresolved target uncertainty.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by stat.ML updates on arXiv.org.