Deep Learning Surrogates for Emulating Stochastic Climate Tipping Dynamics

TL;DR

This paper develops a deep learning surrogate model using a modified Temporal Fusion Transformer to efficiently predict stochastic climate tipping points in ocean transport simulations, achieving high accuracy and speedup.

Contribution

It introduces a dynamics-informed TFT architecture tailored for emulating complex Earth system simulations with stochastic transition events.

Findings

Achieves 465x speedup over traditional simulations.

Accurately forecasts Atlantic and Pacific collapse timings.

Captures stochastic uncertainty in transition predictions.

Abstract

This work explores a dynamics-informed Temporal Fusion Transformer (TFT) as a data-driven surrogate for computationally intensive Earth system simulations. Focusing on multivariate time series describing global ocean transport, we demonstrate the surrogate's ability to forecast tip events across thousands of time steps. The data involve up to 21 non-stationary time series in addition to static covariates describing free parameters and initial conditions. Modifications to the architecture and objective function yield a surrogate that anticipates the timing of Atlantic and Pacific collapses to high fidelity and captures the stochastic uncertainty in transition timing across ensemble predictions. The learned surrogate achieves a 465x computational speedup over the numerical simulator while maintaining differentiability with respect to parameters and initial conditions.