Identification of extreme temperature and precipitation patterns in Spain based on multiscale analysis of time series

Abstract

Climate change is a matter of global interest. Great part of the research focuses on extreme climate events. This study investigates the patterns of change in extreme climate events in Spain, from 1951 to 2021. A total of 16,156 multivariate time series, based on monthly extreme data of maximum and minimum temperatures as well as precipitation, corresponding to 5 × 5-km2 small areas, were analysed. The analysis procedure used reduces the dimensionality of the time series, clustering them and identifying changing patterns for each cluster. To carry out the research, a new multiscale analysis methodology is proposed, based on the financial field (Shi et al. in Phys A Stat Mech Appl, 567:71932008, 2021) and adapted to the climatic field: (1) considering the asymmetry of the precipitation series, (2) extending to the multidimensional case to obtain a better representation of the climatic reality, and (3) adapting the definition of distance between univariate series to multivariate series. The clustering results from applying the novel methodology, divide Spain into twelve zones. Among the results, it is worth highlighting that: extreme maximum temperatures have grown more in the cluster that covers Eastern Andalusia and the interior of Southern Murcia; there are three clusters corresponding mainly to Southern Spain and Middle Mountain that have significantly increased the duration of their droughts. Climate series show cyclical patterns and trends over multiple time scales, making the new methodology very useful. Furthermore, this approach allows detecting possible differences between the evolution of microclimates of small neighbouring territorial areas.