LAUSR

Abstract The objective on this study is to fully statistically determine and describe the rainfall regime characteristics of the very long precipitation records of 124-year period (1892–2015) in Thessaloniki and moreover to apply extreme value statistical methodologies to extreme daily rainfall information (period, 1931–2015), in order to define the extreme precipitation probability distributions. To meet these objectives, the observed annual, monthly and daily rainfall measurements obtained at the Meteorological Station of the Department of Meteorology and Climatology of the Aristotle University of Thessaloniki are used. The use of both Mann-Kendall test and moving average methods are used to determine the profile of the examined very long-period rainfall regime in Thessaloniki, on annual, seasonal, and monthly basis. Taking into consideration the climatic change effect upon precipitation regimes, the extreme daily rainfall measurements are statistically analyzed with extreme value statistical methodologies, defining thus the extreme precipitation probability distribution functions. The Taylor diagram analysis, through combinations of appropriate correlation coefficients, standard deviations, and root-mean-square differences, are constructed to statistically quantify the degrees of similarity between reference points (mean annual precipitation) and results on seasonal and agricultural (agro-) seasonal basis. All the above tests are applied to detect possible changes in precipitation characteristics, over the study region, throughout the extreme long period of 124 years. The resulted probability distributions might potentially contribute to the prediction of extreme rainfall events, and be used for future climatic change projections. Moreover, based upon the resulted-adopted probability distributions, emerging from this extreme long study period, return periods for different extreme values of precipitation observed are calculated.