Climate variability and change within the discharge time series: a statistical approach

Abstract

The paper contains some results of long-time series analysis of discharges with respect to climate variability and change. The appropriate statistical computations based on data supplied by the Global Runoff Data Center in Koblenz. The computations have been carried out for the case of annual and of monthly (of each month) time series. The verification of the assumed hypotheses has been conducted for 5% significance level. The hypothesis that the mean value and the variance are stationary and ergodic (Kruskal-Wallis test) have to be rejected, respectively: (a) from 10.2% to 21.6% of cases and from 1.7% to 7.4% of cases for monthly discharges; and (b) in 23.3% and 3.4% of cases for annual discharges. Whereas, in case of the Mann-Kendall test trends emerge (a) in the mean value from 25% of cases to 42.6% of cases, and in the variance from 10.2% of cases to 19.3% of cases for monthly discharges; (b) in 42% and 9.1% of cases there occur trends in the mean value and the variance, respectively, for annual discharges. Moreover, results of the tests are presented separately for the time series being independently and dependently distributed in time.