A climate-change attribution retrospective of some impactful weather extremes of 2021

Abstract

Abstract. The IPCC report AR6 indicates a general consensus that anthropogenic climate change is modifying frequency and intensity of class of extreme events such as cold spells, heatwaves, storms or floods. A different point of view is to investigate whether a specific extreme event would have been possible in the absence of climate change, or whether climate change may have affected its specific characteristics. Here, we address this question by performing an attribution of some major extreme events that occurred in 2021 over Europe and North America: the winter storm Filomena, the French Spring cold spell, the Westphalia Floods, the Mediterranean summer heatwave, the hurricane Ida, the Po Valley tornadoes outbreak, the medicane Apollo and the late autumn Scandinavian cold-spell. We focus on the role of the atmospheric circulation associated with the events and its likelihood in present (factual world) and past climate conditions (counterfactual world) – defined using the ERA5 dataset 1950 to present. We use an analogs-based methodology whose aim is to find the most similar sea-level pressure patterns to the target events in the factual and counterfactual worlds and compute significant shifts in probability, persistence, predictability and seasonality of the patterns. We also diagnose whether in the present climate the analogs of the studied events lead to warmer/cooler or dryer/wetter conditions than in the past. We find that most of the events are significantly modified in present climate with respect to the past, because of changes in position, persistence and seasonality of cyclonic/anticyclonic patterns. Two of the events, storm Filomena and Medicane Apollo, appears to be a black swan of the atmospheric circulation, with analogs of bad quality. Our approach, complementary to the statistical methods already available in the community, warns that the role of the atmospheric circulation should be taken into account when performing attribution studies.