Comparison efficiency of moments, L-moments and maximum likelihood methods in regional flood analysis

It is necessary to determine the appropriate probability distribution for fitting the flood data. There are several methods for estimating the parameters of probability distribution functions. In the present study, L-moments approach is used for flood frequency analysis in Gavkhouni Basin, Abarkouh- Sirjan Basin and Yazd- Ardakan Basin. Some stations were selected and their data were analyzed to find out peak discharge. Then, Hosking and Wallis homogeneity criterion were used to identify the homogeneous region. Examining the homogeneity criterion closely shows the discordant region. Eliminating and transferring some selected stations, leads to determining a homogeneous region of flood frequency. Parameters of the regional frequency distribution were evaluated using L-moment, moments, and maximum likelihood methods. The results of this analysis were compared using RMSE, R, and RRMSE statistics. The goodness of fit tests such as Chi-square, Kolmogorov-Smirnov were applied for checking the adequacy of fitting of probability distributions to the recorded data. The study reveals that the Pearson Type-3 distribution (using LMM) is better suited amongst the ten distributions used in estimating the peak flood. In general, the L-moment method estimated the peak flood for return periods of 10, 25, 50, and 100-year, and moments method at return periods of 2 and 5-year.