A hybrid statistical-dynamical method to translate past extreme temperature days into the future climate

Abstract

This study presents a novel hybrid statistical-dynamical method intended to translate past observed weather events into the future climate, and applies it to warm and cold extreme temperature days over western Europe. The method estimates the temperature anomalies that would result if an observed event of the 1940–2023 period, defined in terms of atmospheric circulation, were to occur at the end of the 21st century, under new climatological conditions. In practice, constructed analogues of observed extreme events are built using data from regional climate projections. Three regional climate projections under the RCP8.5 emissions pathway are used in order to assess the role of model uncertainties in this context. The same approach is also used beforehand to assess the role of large-scale circulation in the observed extreme temperature days, and the ability of regional climate models to capture it is evaluated. The study finds significant variability in the role of atmospheric dynamics in extreme temperature days, contributing 35–80 % of the temperature anomaly for warm days and 20–90 % for cold days, with other factors such as land-atmosphere interactions playing an amplifying role. Regional climate models generally capture the dynamical part of temperature anomalies quite correctly. Not surprisingly extreme temperature days become more intense in the future climate, but a large inter-event spread exists. Some of the events could become much warmer, while others would not change much. Moreover, this intensification varies widely between regional climate models, and not necessarily in line with the average warming.