What is Extreme Events Attribution?

Extreme Event Attribution (EEA) is an emerging field in climate sciences, at the intersection between statistics, atmospheric dynamics and human sciences. One of the goals is to describe whether and how the probability of an event, like a heatwave, depends on climate change [Stott et al. 2015].

Methodological advances were made thanks to the ERC A2C2 and ANR DADA projects. ESTIMR made a significant contribution to the FP7 Space EUCLEIA project, and led the Extremoscope project at a national level. Currently attribution activities are maintained through the ERA4CS EUPHEME project and the “National convention on climate services”. Recently, the ESTIMR team is involved in the XAIDA project where sixteen European research institutes are joining forces with climate risk practitioners to better assess and predict the influence of climate change on extreme weather using novel artificial intelligence methods. This new EU-funded 4-year project, called “XAIDA”, started in September 2021.

Estimating changes of low probabilities has required the development of statistical tools based on extreme value statistics [Naveau et al. 2018]. To express event attributions within a mathematical context, we reframed the computation of rare events probabilities with the elements of causality theory [Hannart et al. 2016a,b]. New statistical approaches to model climate change in EEA were proposed and studied with our colleagues from Météo-France [Ribes et al. 2016].

The physical interpretation of EEA is based on the separation of contributions due to temperature increase (a.k.a. thermodynamical) and to properties of the atmospheric circulation leading to the event (a.k.a. dynamical). We have developed complementary statistical methods based on analogues of circulation to estimate those contributions [Vautard et al. 2016; Yiou et al. 2017]. Those methods are based on the idea that one can “fix” the atmospheric circulation by using analogues, and compute a response of climate change that is conditional to climate change. Trends of the dynamical contributions in extreme heatwaves have been investigated [Jézéquel et al. 2018a].

We investigated how the circulation analogues can be adapted for event attribution [Jézéquel et al. 2017]. This helps optimizing the parameters of the computations to the event to be investigated.

Some of those methodologies require the analysis of many large ensembles of model simulations and reanalyses. We were involved in the analyses of the weather@home simulations. We provided an assessment of dynamical and thermodynamical contributions for the wet January 2014 [Schaller et al. 2016]. We have exploited the Euro-CORDEX simulations to attribute the precipitation events in the Cévennes region [Luu Nhat et al., 2018].

Most of those tools are mature to be used for a rapid event attribution, i.e. a few days after the event occurs. We have contributed to several issues of the Bull. Amer. Meteorol. Soc. since 2013 [Yiou and Cattiaux 2013, 2014; Vautard et al. 2015, 2017; Jézéquel et al. 2017]. We have designed event summaries for the general public.

The tools for EEA have been assembled in a Web Processing Service [Hempelmann et al. 2017], which also allows for a quick visualization of the computations. Documentation and training to this tool is also provided.

EEA is now acknowledged as a climate service (see associated document) and is at the center of a few European and national projects.

We organized two schools (Cargèse, 2015; Aussois, 2017) on the mathematics and statistics of EEA. Members of the ESTIMR team are often contacted by the media to comment recent extremes, and put them in a climate change perspective.

The motivation and perception of EEA in the climate community and in the decision maker spheres were investigated [Jézéquel et al. 2018b]. This sociological analysis helps framing the motivation behind EEA.