Uso de modelos probabilísticos para estimar a precipitação máxima anual na bacia hidrográfica do rio Govuro em Vilankulo – Moçambique

Abstract

The planning and management of water resources requires a thorough knowledge of the hydrological events and consequently the probabilistic studies of their prediction for the good use in the design of hydraulic structures. The probability distributions make it possible to estimate the extreme events of the annual maximum precipitation associated to other parameters of occurrence. Thus, the aim of this study was to analyze the performance of four probability distribution models (Gamma, Gumbel, Weibull and 3 parameter Weibull), whose parameters were adjusted with the Maximum Likelihood (ML) applied to a historical rainfall record of 29 years in the Rio Govuro watershed and the goodness-of-fit criteria played a role by using Anderson-Darling adherence test at a significance level of 5%. As results, it was verified that the Gamma’s method was the most susceptible to describe the annual maximum precipitation distribution in the case study.