Analysis of project flood flow calculation methods: Example of Uzundere pond

In this study, it is aimed to calculate the flood flow quantiles of Erzurum Uzundere Pond Project, which was planned by the 8 th Regional Directorate of the General Directorate of State Hydraulic Works, Erzurum in 2013 and the tender was made in 2015. For this purpose, the eleven widely used probability distributions, namely Weibull, Two-parameter Lognormal, Two-parameter Gamma, Logistic, Normal, Gumbel, Generalized Extreme Value, Log-Logistics, Pearson type-III, Three-parameter Lognormal, Three-Parameter Gamma, were fitted to annual instantaneous maximum flow observations from 7 gauge stations located in the region. The most suitable models were selected using Akaike Information Criteria, Bayesian Information Criteria, Kolmogorov-Smirnov, Cramér-von-Mises, and Anderson Darling tests. Project flood discharges were calculated using the Regional Flood Frequency Analysis method. Unlike previous studies, a large number of probability distribution models were considered to obtain more precise flood quantile values. The computed flood quantiles under various return periods were compared with the flow values calculated during the planning phase of the project. As a result, it has been seen that more accurate results can be achieved by increasing the number of probability distribution functions, data years and the number of goodness of fit tests/criteria in projects such as Uzundere Pond where the computation of design flow is crucial. Since the results have a direct effect on the dimensioning of the water structures, it has been observed that it will be possible to obtain more economical or more reliable designs in this way.