Item

Abstract

Climate analogues are a comprehensive approach for learning how to deal with the expected climatic future from current examples. Existing literature on climate analogues struggles with 2 methodological challenges: how to deal with the unavoidable uncertainty of climate projections and how to define reasonable lower limits of similarity for climate analogues. Here, we suggest a new method to identify robust climate analogues (RCAs) based on a clustering approach in climate space where each spatial grid element is represented by 3 points: its current climate, and a lower and upper bound of the climate projections. If the upper and lower bound of the projections for such a grid element share the same cluster and, additionally, this cluster contains current climate points, then the grid elements related to the latter are defined as RCAs. This definition divides the map of the investigated region into areas with RCAs and uncharted areas where, under the current uncertainty range of climate projections, such an attribution is not justified. An exemplary application of the algorithm for Europe shows that RCAs can be identified for 37% of the land area and that the new method allows selection of socioeconomically reasonable RCAs from climatologically equivalent (given the current uncertainty) grid elements.