Determination of the most suitable probability distrubiton model for mean and maximum flows of Eastern and Western Black Sea river basins

This study explains the current approaches used to define the optimal probability distribution model of the annual maximum and annual average flows of the Eastern and Western Black Sea streams. For this purpose, ten distribution functions commonly used in the literature were applied to the flow (annual maximum and annual average) series of 18 flow monitoring stations located in the East and West Black Sea basins. Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), Kolmogorov-Smirnov (KS), Cramer-Von Mises (CvM), Anderson-Darling (AD) compliance tests were used to determine the most appropriate distribution models among alternative distributions. As a result of conducted tests, the most appropriate distribution types were determined for flow monitoring stations in East and West Black Sea basins as follows. For annual maximum flow, Log-Logistics for stations 2202 and 1334; Generalized Extreme Value for stations 2218, 2232 and 1332; Gumbel for station 2233; Weibull for station 2238; Three-parameter Lognormal for stations 2245 and 1338; Lognormal for stations 2247 and 1339; and Pearson Type-3 for stations 2247 and 1307; and Gamma for station 1314 were determined. As for the average annual flows, Lognormal for stations 2202, 2232 and 1339; Logistics for stations; 2215, 2218, 2238, 2247, 1302,1319 and 1334; Gamma for stations 2233 and 1340; Three-parameter Lognormal for station 2245; and Weibull for stations 1307,1332 and 1338 were determined. Then, using the most appropriate distribution models obtained, the flow values corresponding to the 10, 20, 50, 100 and 200 year return intervals as well as the lower and upper limits in the 95% confidence interval were calculated. These flow values, calculated for different repetition times, are very important for the design, planning and management of water structures in the basins in question.