Regional analysis is an appropriate method for predicting design flood; based on statistical data and local information about the hydrological catchment .The suitability of this method is due to its ability to model (simulate accurately) the relationship between flood and physiographic specification of basin. Regional analysis uses local statistical data which was obtained from a limited number of locations and then extending these data to represent the whole of the region. However, an assessment should be made about data accuracy and bias as well as the appropriateness and homogeneity of the chosen area in terms of representative characteristics of the whole area. Initially in this research, data of maximum flood discharge in 7 selected stations were re-built and evaluated with homogeneous test. Then, statistical distributions were drawn by using momentum method. By employing the graphic and statistical tests the best distribution was selected for each station. Then separate the region into several homogeneous regions. To reach this objective cluster analysis was used. Based on linear regression analysis, relationship between estimated discharge and physiographic specifications were suggested. Finally, error analysis method was used to compare these models together in each homogeneous region and in the whole of the region.
