Estimating nonlinear stability from time series data

TL;DR

This paper presents a method to estimate basin stability, a measure of nonlinear stability in multi-stable systems, directly from time series data without requiring explicit models, demonstrated on climate tipping models.

Contribution

It introduces a novel approach for estimating basin stability from observational time series data, applicable even with strong perturbations and limited data.

Findings

Method successfully applied to climate tipping models

Estimates basin stability without explicit dynamical models

Discusses applicability to real observational data

Abstract

Basin stability (BS) is a measure of nonlinear stability in multi-stable dynamical systems. BS has previously been estimated using Monte-Carlo simulations, which requires the explicit knowledge of a dynamical model. We discuss the requirements for estimating BS from time series data in the presence of strong perturbations, and illustrate our approach for two simple models of climate tipping elements: the Amazon rain forest and the thermohaline ocean circulation. We discuss the applicability of our method to observational data as constrained by the relevant time scales of total observation time, typical return time of perturbations and internal convergence time scale of the system of interest and other factors.