Evaluation of six distributions of probability at maximum temperature series in Iguatu - CE, Brazil

Abstract

This study had as objective to evaluate the distribution of probability that better adapts to daily series of maximum temperature in Iguatu city, Ceará, located in the Northeastern Brazil, with climate, according to Köppen, of the type BSw’h. Were used the distribution of probability functions Gumbel I, Gama, Log-normal, Normal, Weibull and Beta. The studied series included a period of 45 years (1961 to 2005), obtained from the 3rd District of the Instituto Nacional de Meteorologia (INMET). The adherence of the distributions was tested by the tests Chi-square and Lilliefors both to the level of significance of 5%. The distribution that best fit the data series of maximum temperature for Iguatu - CE was the Normal distribution and the worse was the distribution of Gumbel type I.