In this paper we propose to use the Extended Geometric Exponential distribution (EGE) as an alternative model to the commonly used distributions such as Gamma, Weibull, Lognormal among others, for modeling rainfall data. The distribution EGE has been little explored in the literature although it can be applied in many research fields such as biology, demography, and reliability of electronic products and can also be applied to analyze meteorological phenomenon. The EGE distribution has been proposed by Adamidis and collaborators in 2005 and one of its peculiarities is that the hazard function can be increasing or decreasing. Another important characteristic is the easiness to obtain different probability levels which do not demand numerical approaches. Several distributions such as Gamma, Weibull and Lognormal were used to fit the rainfall data measured at Presidente Prudente city. The results showed EGE being the best fit for the data according to the Akaike information criterion, the Kolmogorov-Smirnov test and the Chi-square test. The estimators of the EGE distribution parameters were obtained by the maximum likelihood approach and thus allowing the estimation of monthly rainfall for different probability levels.
Extended Geometric Exponential Distribution; rainfall; maximum likelihood; probability levels
