Computation of extreme heat waves in climate models using a large deviation algorithm

Abstract

Studying extreme events and how they evolve in a changing climate is one of the most important current scientific challenges. Starting from complex climate models, a key difficulty is to be able to run long enough simulations to observe those extremely rare events. In physics, chemistry, and biology, rare event algorithms have recently been developed to compute probabilities of events that cannot be observed in direct numerical simulations. Here we propose such an algorithm, specifically designed for extreme heat or cold waves, based on statistical physics. This approach gives an improvement of more than two orders of magnitude in the sampling efficiency. We describe the dynamics of events that would not be observed otherwise. We show that European extreme heat waves are related to a global teleconnection pattern involving North America and Asia. This tool opens up a wide range of possible studies to quantitatively assess the impact of climate change.

Computation of Extreme Values of Time Averaged Observables in Climate Models with Large Deviation Techniques

“…Stochastic weather generators provide to some extent a hybrid statistical-dynamical approach [33,2]. A new approach entirely based on the dynamics of numerical models was proposed in [26], where we have introduced the use of rare event algorithms to improve the sampling efficiency of climate models. These techniques allow to increase the number of extreme events observed for a given computational cost, by generating trajectories that are real solutions of the equations of the model, without additional statistical assumptions.…”

“…With the exception of a few works on multifractal modeling of rainfall [31], the use of large deviation theory to study climate extremes has not been considered until very recently. In [26] we used a rare event algorithm developed for the computation of Donsker-Varadhan type large deviation functions, and applied it to study rare heat waves, although for durations shorter than what necessary to be in the large deviation limit. [13] recently performed a comparison of extreme value theory and large deviation theory based approaches to study time and space averages of climatic observables in an idealized general circulation model.…”

“…Given the typical limitations in the amount of available data in these applications, it is likely that it will be rather difficult to go substantially (if at all) beyond the Gaussian regime given by the central limit theorem. In this case, one could use rare event algorithms dedicated to compute Donsker-Varadhan type large deviation functions in numerical models that have been developed in recent years [16,21], and very recently applied to study heat waves in a climate model [26].…”

Maximizing Simulated Tropical Cyclone Intensity With Action Minimization

“…In addition to Hoffman et al (2006a) and Hoffman et al (2006b), several previous studies have explicitly pointed to the potential impact of rare event simulation and analysis tools on geophysical applications: a similar path finding approach was considered in the context of rogue ocean waves by Dematteis et al (2018). Methods designed to generate random samples of rare events were considered in the contexts of rare transitions of an ocean current model in Weare (2009) and Vanden-Eijnden and Weare (2013) and extreme heat waves in Ragone et al (2018).…”

An Interdisciplinary Approach to the Study of Extreme Weather Events: Large-Scale Atmospheric Controls and Insights from Dynamical Systems Theory and Statistical Mechanics

“…These tools provide a means of obtaining large samples of extreme events and the associated large-scale atmospheric patterns, something that is impossible to recover from observational datasets and very costly to generate from long climate model simulations (Ragone et al 2017). They also provide a framework to diagnose changes in the atmospheric dynamics-and the associated extremes-in both past and future climates by using mathematically robust atmospheric indicators whose definition is independent of the variable, geographical domain, or season chosen.…”