Comparing Regional Analysis methods for estimation of the peak flow in some watersheds of Isfahan- Sirjan and Yazd-Ardakan Basins

The main problems in flood frequency analysis are limited number of gauging stations and recorded data, together with the inaccurate at-site estimations in the study area. These problems have caused increasing application of regional methods. Regional analysis seems to be a useful method for estimating peak flow at area with no data or low-recorded length. Regional flood frequency analysis relies on physical and climatic characteristic of basins and applies statistical method to study flow records. The methods of regional analysis are numerous that the selection of each one of them in any study area depends on data length, climatic factors, data type and expected return periods. In this study four techniques of regional analysis were used to evaluate the priority and importance to estimate the peak flow for different return periods. The Hybrid, Multiple regression, L-moments and Canonical Correlation Analysis are the four approaches applied for some watersheds of Isfahan–Sirjan and Yazd-Ardakan Basins. A number of 16 stations were selected and their data were analyzed to find out peak flow. The results of this analysis were compared to the Hybrid and Multiple regression approaches using RRMSE and MAE statistics.The results showed better performance of the CCA method rather than the other methods in all return periods. After CCA, Multiple regression methods were selected to estimate the peak flow (Model 2,3). Therefore, CCA method can be adopted as regional flood frequency method for the study area.