Fitting of GPA, GLO and GEV Distributions with Trimmed L-moments (1,0).

Statistical moments have been used to characterize probability distributions and samples of observed data. This work briefly describes the theory of L-moments and trimmed L-moments (1,0), which can reduce the influence of the lowest value in a sample, in order to improve the fit and obtain more reliable extreme predictions. Recent equations found in the statistical literature that estimate the location, scale and shape of the probability distribution functions are cited, which are often used to analyze the frequencies of extreme hydrological data. These include the General Extreme Values (GEV), Generalized Logistic (GLO) and Generalized Pareto (GPA) equations. These three distributions were fitted with the methods of L-moments and trimmed L-moments (1,0) to 21 annual maximum flow registries from Hydrological Region No. 10 (Sinaloa). The quality of each fit was evaluated based on the standard error. The analysis of the results indicate that the GPA distribution provides the smallest fitting errors for 13 registries using the trimmed L-moments (1,0) and for the rest of the registries using L-moments. The conclusions suggest that the three probabilistic models studied can be applied with the trimmed L-moments (1,0) as an advanced version of the L-moments procedure which is universally used.