On the Regional Temperature Series Evolution in the South-Eastern Part of Romania

Abstract

In the context of reported climate variations in different regions of the world, this work investigates the evolution of the temperature series in the Dobrogea region, Romania, using the maximum, average, and minimum annual temperature series from 1965 to 2005. The Mann–Kendall test and Sen’s slope emphasized increasing trends of nine (out of ten) minimum temperature series (nine of them at significance levels less than or equal to 0.05, and two at 0.1), three average temperature series (at a significance level of 0.1), and five maximum temperature series (at significance levels less than or equal to 0.05). The selection of the representative series at the regional scale, called the ‘Regional series’, was performed using two algorithms proposed by the author that are easy to employ, even by individuals without deep knowledge in the field. The first (called MPPM) was initially introduced for evaluating the ‘Regional precipitation series’, and the second is a version of MPPM based on clustering the data series. Comparisons with the series from the ROCADA database were performed to prove the algorithms’ performances. The best results were obtained by running the second algorithm with two clusters, for the minimum and maximum temperature series, and with three clusters, for the average temperature series. In comparison with the initial data series, the average MAEs were, respectively, 1.39, 0.37, and 0.84 for the minimum, average, and maximum series, and the corresponding average MSEs were, respectively, 1.49, 0.41, and 0.93. Comparison of the ‘Regional series’ with the series from ROCADA led to a decrease in the modeling errors, with the best ones corresponding to the average ‘Regional series’—MAE = 0.36 and average MSE = 0.25.