Persistence of Rainfall Time Series: Kırşehir Case Study

Abstract

This study examines the persistence and long-term correlation of monthly and seasonal rainfall time series of Kırşehir for the period of 1960-2019, with widely used Hurst exponent and Detrended Fluctuation Analysis (DFA) analyses. Both Hurst exponent and DFA analyses could be used to detect the long-term memory and correlation that can be assessed as a reference of predictability. To support the analyses results Augmented Dickey Fuller and Mann-Kendall tests also applied to time series. Within various rainfall series, evidence of persistence and long-term correlation was identified. According to H exponent values of simple R/S and corrected R/S methods, 10 out of 12 months and winter, spring (only simple R/S), summer (only corrected R/S) and autumn season and according to DFA scaling exponent values 4 out of 12 months and winter and autumn seasons exhibit long term correlation. On the other hand, when the H exponent and DFA scaling exponent values compared only four monthly and two seasonal rainfall series concluded to be consistent with both H exponent and DFA scaling exponent results.