Quantifying Impact of Geographical Variables on Climate Change Patterns over Spain by Time Series Clustering

Abstract One of the greatest environmental threats worldwide arises from temperature and precipitation variations driven by climate change. Recent studies have increasingly focused on climatic regionalization, generating clusters to analysing climate change patterns. However, most of these studies have analysed the effect of climate change on the groups once they have been formed. In this context, the present study proposes a novel regionalization approach by including climatic change estimates into each meteorological time series as input for clusters generation. This innovative methodology allows us quantify the impact of geographical variables, such as distance to the sea, height, latitude and longitude, on climate change patterns within a territory. Partial Least Squares Discriminant Analysis (PLS-DA) was employed to qualitatively describe the clusters and assess the influence of geographical variables. Additionally, from the PLS-DA analysis, a new index that optimizes clustering by balancing the main metrics from the model was generated. This approach was applied to variations in temperature and precipitation of Peninsular Spain from 1951 to 2021, analysing 15,992 multivariate time series. The results reveal significant regional differences in the observed climate change patterns. Maximum temperatures have increased the most in mountainous regions and central areas of Spain, while the smallest increases occurred in Southern Spain. A decrease in precipitation was observed, with most pronounced reductions in southern and inland regions. Furthermore, there was a marked increase in consecutive dry days, particularly in the South. Trends in cold temperature extremes have diminished across most regions. These findings provide valuable information for future climate adaptation and mitigation strategies. The proposed methodology is flexible and scalable, making it suitable for application to large regions with high climatic variability.