Investigating the Peak Flood Estimation Methods in a Regional Analysis Approach for Low-Data Catchments

The methods of regional flood frequency analysis is one of the most reliable and applicable technique for the regions with insufficient recorded data. In this study three techniques of flood analysis were used to evaluate the priority and importance to estimate the peak flow for different return periods. The hybrid, multiple regression and L-moments are the three approaches applied for Yazd-Ardakan and Isfahan-Sirjan plains in central Iran. At the first step, the most effective physical and hydrological parameters were identified by factor analysis technique. Then the coefficients of the power formula of Qp=aXbYcZd were estimated regarding the hybrid and multiple-regression methods. The results of this analysis were compared to the L-moments approach using RRMSE and MAE statistics. The findings showed better performance of the multiple regression and hybrid rather than L-moments methodat short return periods (2, 5 and 10-year). For longer return periods (25 and 50-year) the L-moments approach represented better-fit to the observed flood data. In general, due to variability of the parameters such as, climatic factors, data length and type, affecting peak flood, each one of the regional flood frequency analyses may have better performance for the case studies. Therefore, these techniques should be evaluated for each hydrological homogeneous region.