Articles | Volume 3, issue 1
18 Apr 2017
 | 18 Apr 2017

A statistical framework for conditional extreme event attribution

Pascal Yiou, Aglaé Jézéquel, Philippe Naveau, Frederike E. L. Otto, Robert Vautard, and Mathieu Vrac

Abstract. The goal of the attribution of individual events is to estimate whether and to what extent the probability of an extreme climate event evolves when external conditions (e.g., due to anthropogenic forcings) change. Many types of climate extremes are linked to the variability of the large-scale atmospheric circulation. It is hence essential to decipher the roles of atmospheric variability and increasing mean temperature in the change of probabilities of extremes. It is also crucial to define a background state (or counterfactual) to which recent observations are compared. In this paper we present a statistical framework to determine the dynamical (linked to the atmospheric circulation) and thermodynamical (linked to slow forcings) contributions to the probability of extreme climate events. We illustrate this methodology on a record precipitation event that hit southern United Kingdom in January 2014. We compare possibilities for the creation of two states (or worlds) in which probability change is determined. These two worlds are defined in a large ensemble of atmospheric model simulations (Weather@Home factual and counterfactual simulations) and separate periods (new: 1951–2014, and old: 1900–1950) in reanalyses and observations. We discuss how the atmospheric circulation conditioning can affect the interpretation of extreme event attribution. We eventually show the qualitative coherence of results between the choice of worlds (factual/counterfactual vs. new/old).

Short summary
The attribution of classes of extreme events, such as heavy precipitation or heatwaves, relies on the estimate of small probabilities (with and without climate change). Such events are connected to the large-scale atmospheric circulation. This paper links such probabilities with properties of the atmospheric circulation by using a Bayesian decomposition. We test this decomposition on a case of extreme precipitation in the UK, in January 2014.