Statistical evaluation and analysis of methods used in regional flood analysis and homogeneity of watersheds in Iran

Regional flood frequency analysis is a powerful tool for estimating and analyzing flood flow in watersheds. In this research, different methods of regional flood analysis and hydrological homogeneity of catchments that has been done in the country has been investigated. Among the many methods for determining the homogeneity of sub-catchments, cluster analysis due to the ability of factor analysis to select the most important factors, simplicity of considering various factors, independence of effective factors, accuracy of separation of homogeneous groups, and advantage of using proper diagnostic functions, was the most appropriate method. Multivariate regression method, especially when the homogeneity of catchments have been accurately determined, has shown good performance in regional flood analysis. L-moment method, due to uniqueness and existence has good performance in estimating parameters and selecting appropriate statistical distributions. So that in the regional flood frequency analysis, the L-moment method performs better than other methods, particularly in the case of shortage and skewed data. The analysis of the intelligent models such as artificial neural networks (ANNs) and fuzzy logic, showed the high ability to establish nonlinear relationships between multiple input variables. The results of the Investigation of different factors used in regional analysis methods showed that Physiographic factors were highest (72.11%) followed by climatic factors (17.69%) and land cover parameters (7.48%). Also, the lowest contribution was related to hydrological factor with 2.72%. Among the physiographic parameters, the area factor with 30.19% of the contribution had the most influence on the regional flood analysis. Among the climatic factors, the highest application was related to the average annual rainfall factor with 73.08% contribution.