Anticipating AMOC transitions via deep learning

2025-09-08

AI Summary

Conventional early-warning indicators based on critical slowing down (CSD) fail to reliably detect abrupt transitions of the Atlantic Meridional Overturning Circulation (AMOC) near bifurcation, rate-induced, and noise-induced tipping points. Method: We propose a deep learning–driven early-warning framework that bypasses reliance on CSD. Specifically, we design a convolutional neural network (CNN) trajectory classifier trained on high-dimensional time-series ensembles generated from a calibrated AMOC box model. The model directly learns high-order nonlinear precursory signatures to estimate, in real time, the probability of transition from a single trajectory. Results: The method demonstrates robust predictive performance across all three tipping mechanisms, substantially improving forecast reliability under stochastic forcing. This work establishes a novel paradigm for assessing predictability of critical transitions in Earth system science and provides a generalizable technical pathway applicable to other complex dynamical systems.

Technology Category

• Machine Learning: Deep Learning Theory
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Key components of the Earth system can undergo abrupt and potentially irreversible transitions when the magnitude or rate of external forcing exceeds critical thresholds. In this study, we use the example of the Atlantic Meridional Overturning Circulation (AMOC) to demonstrate the challenges associated with anticipating such transitions when the system is susceptible to bifurcation-induced, rate-induced, and noise-induced tipping. Using a calibrated AMOC box model, we conduct large ensemble simulations and show that transition behavior is inherently probabilistic: under identical freshwater forcing scenarios, some ensemble members exhibit transitions while others do not. In this stochastic regime, traditional early warning indicators based on critical slowing down are unreliable in predicting impending transitions. To address this limitation, we develop a convolutional neural network (CNN)-based approach that identifies higher-order statistical differences between transitioning and non-transitioning trajectories within the ensemble realizations. This method enables the real-time prediction of transition probabilities for individual trajectories prior to the onset of tipping. Our results show that the CNN-based indicator provides effective early warnings in a system where transitions can be induced by bifurcations, critical forcing rates, and noise. These findings underscore the potential in identifying safe operating spaces and early warning indicators for abrupt transitions of Earth system components under uncertainty.

Problem

• Predicting AMOC tipping points using deep learning
• Overcoming unreliability of traditional early warning indicators
• Estimating transition probabilities under bifurcation, rate, and noise-induced tipping

Innovation

• CNN-based deep learning for AMOC transition prediction
• Identifies higher-order statistical differences in ensemble trajectories
• Enables real-time probabilistic forecasting before tipping onset