Climate Analogs Analysis over Europe: Accounting for dependencies between variables

Abstract

Increasing the awareness of society about climate change by using a simplified way for the explanation of its impacts might be one of the key elements to adaptation and mitigation of its possible effects. This study investigates climate analogs, which allow a comparison between current and future climate conditions. The grid-based calculation of analogs over the selected European domain was carried out using a newly proposed distance between multivariate distributions, the Wasserstein Distance (WD), never been used so far for climate analog calculations. By working on the whole multivariate distributions, WD allows us to account for dependencies between the variables of interest. Its features are compared with the Euclidean Distance which is currently the most used method. Multi-model climate analogs analysis is achieved between the reference period 1981-2010 and three future periods 2011-2040, 2041-2070, and 2071-2100, respectively by using 15 different datasets in total obtained from five different CMIP6 climate models and three different emission scenarios. The multi-model analysis also enables the comparison of models from a climate analogs perspective. The model comparison results show that consistency between models decreases as the time approaches the end of the century or when scenarios worsen. Our analysis suggests that Europe’s future analogs are located today south of our regions, except for the Balkans that need to look east to find their analogs. In addition, towards the end of the century, the similarity between future and current climatic conditions will gradually decrease and the spatial distance between each reference grid and its best analog location will increase. This means that the warmer the climate, the more difficult it will be to find an analog and therefore the more difficult it will be for us to think about adaptation.