Uncertainty of Design Parameters From Viewpoint of Extreme Statistics

Abstract

Statistical variability of a sample of extreme data has been examined by means of Monte Carlo simulations. The Fisher-Tippett types I and II, the Weibull, and the log-normal distributions were chosen for the examination. The sample size covered from 10 to 400, and 10,000 runs were carried out for each combination of sample size, censoring rate, and distribution function. Empirical formulas and tables are presented for the mean and the coefficient of variation of the standard deviation of a sample, the standard deviation of return value, and the confidence interval of return period.