M. Sc students of water resources engineering, Tarbiat Modares University.
For determining homogeneous regions and estimation of maximum flood in various return periods, cluster method and regional frequency analysis using linear moments are used in areas without hydrometric stations. Results showed that using Cluster analysis in the under study region was detected suitable to 5 homogeneous regions with k-means clustering method in Urmia lake basin. Discordance test has conducted and two stations such as Sahzab and Bokan were found as discordant stations and frequency analysis has implicated at-site. With elimination of discordant stations, all regions was found homogeneous using heterogeneity test. In this research, maximum annual flood fitted to eleven distribution functions. Using goodness-of-fit the Generalized logistic distribution were selected for three regions of A, B and C, Log normal 3 parameters for D, Generalized extreme value for E and for Sahzab and Bokan stations Log normal 2 and 3 parameters were chosen respectively. Besides, Generalized extreme value distribution was found appropriate for all of the regions.