AbstractChanges in precipitation and flow regimes may cause more catastrophic floods in the future. This study examines the trends in a subbasin of Bartin Stream in the Western Black Sea Basin, Turkey, and questions the influence of non-stationarity by performing flood frequency analyses on one stationary and nine non-stationary models by incorporating the covariates of time and annual precipitation as a linear function of their location and scale parameters. In this context, variabilities in annual precipitation, (AP), mean annual flow (MAF), and maximum daily flow (MDF), of five precipitation gauging stations and four streamflow gauging stations were examined using the Mann-Kendall and modified Mann-Kendall tests, and Sen’s slope estimator. The breakpoints were identified in the series by employing homogeneity tests, and the relative changes before and after the shifts were analyzed. The modified Mann-Kendall showed a significantly increasing trend at Bartin precipitation station, with an increase of 0.69  mm/year, and significant decreasing trends at Cide and Ulus precipitation stations, with a rate of change 7.10  mm/year and 3.26  mm/year, respectively, at the 5% significance level. Moreover, significant downward trends were detected at Kocairmak station (MAF varied 0.33  m3/s/year and MDF varied 4.55  m3/s/year); at Kocanaz station (MAF varied 0.04  m3/s/year); and at Bayiryuzu station (MDF varied 1.49  m3/s/year). The greatest relative changes in MAF and MDF were found at Kocairmak station, which can be considered as an indicator of non-stationarity. Thus, a non-stationary frequency analysis of the MDF series at Kocairmak station was performed, and the best-fitting model was found to use the AP series as a covariate, the same as other streamflow stations in the basin. In all stations, the non-stationary design flood quantiles were estimated to be greater than the stationary ones by up to 30%, which should be taken into account, particularly for designing flood mitigation facilities.

Source link

Leave a Reply

Your email address will not be published.