Regional Flood Frequency Analysis of Northern Iran

Abstract

The combination of the L-moment approach and multiple regression offers an attractive solution to provide flood estimation at ungauged sites within the Gorganrood and Ghare-sou river basins in the north of Iran. This research has two main goals including regionalization by cluster analysis and regional estimation of flood quantile at the site of interest. After data analysis regarding climatic and hydrologic data series, hierarchical approach was carried out to identify homogeneous regions. The homogeneity test was done by H-Statistic, a testing method based on L-moments. The results showed that a subdivision of selected watersheds into homogenous groups is necessary; therefore, two homogenous regions were formed. In the present study, five three-parameter distributions were fitted to the homogeneous regions and the best-fit one was identified using the L-moments approach. The results of the goodness-of-fit analysis for the two regions introduced the Generalized Pareto (GPA) distribution for both regions as acceptably close fits to the regional average L-moments. Besides, multiple regression was applied to diagnose the effective independent parameters on discharge value. The results reported percent of permeable formations, average annual precipitation, and stream slope as the most effective variables.