LH Moment: The Usage To Model General Logistic Distribution

Estimation of floods of large return periods is often required in engineering structure design. This studyintroduces a method based on LH-moments (LMT) to use generalized logistic (GL) distribution for estimating theupper-part of distribution. Application of the GL distribution using both LMOM and LMT parameter estimatesfrom two actual data are demonstrated. The performance of the LMT method and the LMOM method areassessed by evaluating the bias and mean square error (MSE) of quantile estimates through Monte Carlosimulations with the presence of outlier. Monte Carlo simulation results show that LMT lead to reduced bias and mean square error (MSE) for F = 0.90 and poorer estimation for F = 0.999 compared to L-moments.For the estimation of x(F = 0.9), LMT are in many cases significantly more efficient than LMOM only forsample sizes, n<20.