Multivariate synthetic indexes for evaluating local climatic changes

In this work the application of two novel synthetic multivariate indexes for characterizing and interpreting the correlation structure of climatic data measured in 40 sites of Basilicata (Southern Italy) from 2001 to 2011 is presented. These indexes, in a recursive application of multivariate procedure, based on PCA and clustering, allow exploiting the information content of the descriptors (meteorological variables) and objects (sites), as well as to compare the behavior of the different variables in the correlation structures. Therefore, they enable assigning standardized weights to descriptors and objects, in order to identify the role of each variable in all the investigated period. The temporal analysis has allowed the identification of the behavior of each station in the classification structure, defining a group of stable sites or sites for which the variation of the variables (centroids) does not stray far from the average on the entire observation period (Acerenza, Aliano, Ferrandina, San Giorgio Lucano, Senise and Stigliano) and a group of extreme sites or sites in which the variation of the variables is very different from the average over the entire period of observation (Irsina, Viaggianello and Castel Saraceno).