Frequency analysis of river drought by common and advanced statistical distributions, case study: western rivers of Urmia Lake

Estimating return period of hydrological processes such as flood flow, maximum discharge, drought and etc, is related directly to selecting a suitable probability distribution function. With selecting a proper distribution function, estimated return period will be more close to actual data and error will be reduced. So, it should be tried to select the best probability distribution function. In this study, using daily discharge of western rivers of Urmia Lake, as well as the annual mean method, the data of the drought volume of these rivers were extracted. Several distributions from each of the continuous distributions, such as the continuous distribution of generalized maximums and wakeby, the continuous non-zero Erlang distribution, the continuous Johnson SB distribution, and the continuous boundary of Normal distribution for fitting the data of the dry volume of the western rivers Lake Urmia was used in the 38-year statistical period. Anderson-Darling and Kolmogorov-Smirnov tests used to compare results of each distribution. Also correlations between the data obtained from the sample data and statistical models computed. The results of the frequency distribution of the data of the three rivers showed that among the distributions of the continuous group, the advanced statistical distributions of Wakeby and Johnson SB had the best distribution and also had better performance than the conventional statistical methods.