An Alternative Parameter Estimation Method in Frequency Analysis

Frequency analysis of extreme hydrologic events such as floods, storms, and droughts provides important information on planning, design, and management of water resources systems, and this information is helpful in avoiding negative economic and social consequences. An important step of the frequency analysis is to estimate the appropriate distribution s parameters. This paper shows the application of an alternative parameter-estimation method, RSM (Robust Statistics Method), which calculates the robust measure of central tendency M (median), statistical dispersion IQR (interquartile range) and quartile coefficient of skewness QCs; and uses these robust statistics by the estimation of the parameters of various distribution functions. Six probability distributions (Normal, 2- and 3- parameter lognormal, Gamma, Gumbel and generalized extreme value GEV), which are commonly used in hydrological frequency analysis were discussed within the study. The advantage of using robust statistics like median and interquartile range in parameter estimation is to ensure the resistance to the effect of a change in value or presence of outlying observations. Numerical analyses as part of this research were carried out on the annual maximum 24h rainfall intensities of rainfall gages in the Aegean Region (Turkey). Eventually, the quantile estimations calculated with the parameters of Robust Statistics Method were compared with the results of conventional methods like Maximum Likelihood, Method of Moments, and Probability-Weighted Moments.