Society has always wondered on the responsibility of each climate extreme. Even though it is impossible to prove that anthropogenic climate change triggered a individual events, it is possible to analyze their frequency change, their physical properties and determine potential links with external forcings, anthropogenic or not. This is the ambition of extreme event attribution (EEA), which is a new field at the border between atmospheric sciences and statistics.
I will present the physical and mathematical challenges that EEA faces. The climate models and statistical methods will be briefly described. The general idea is to compare probability distributions of events in a factual world (the world where we live, with all the natural and anthropogenic forcings), and a counterfactual world (an ideal world without human activities).
This presentation will be illustrated by recent climate extremes (heatwaves, large-scale precipitation and storms) and will show some limits to those approaches.