Streamflow trends in the Sakarya basin

Abstract

The detection and attribution of past trends, changes, and variability in hydroclimatic variables is critical for the understanding of potential future changes resulting from anthropogenic activities. Secular trends in monthly streamflow data are evaluated for the past 31 years for 11 stations in Sakarya basin, Turkey. Several non-parametric tests, which were used successfully in water quality trend analysis, were also applied to detect trends in streamflow over Sakarya basin in this study. These tests were developed because the assumptions of classical parametric methods (i.e., normality, linearity, independence) are usually not met by typical water quality data. Moreover seasonality in data compounds the analysis problem. These data idiosyncrasies are mostly common in streamflow data as well. The non-parametric methods are more flexible and in turn can handle the foregoing difficulties. They are the Spearman's Rho test, the Mann-Kendall test, the seasonal Mann-Kendall test, Sen's T test, and the Van Belle and Hughes test. The magnitudes of linear trends were computed by using Sen's estimator. The homogeneity of trend direction at multiple stations and, in different seasons, was also tested by the Van Belle and Hughes test. The results show that all stations except one generally have downward trends according to the same conclusion from all four tests at the 95% significance level. The Van Belle and Hughes homogeneity tests for seasonal trends indicated that all monthly trends are homogeneous for the remaining ten stations. Based on the same test's procedures, a global trend did not exist for the basin.