Author(s): CARVALHO, J. R. P. de; ASSAD, E. D.; OLIVEIRA, A. F. de; PINTO, H. S.

Summary: The aim of this study was to model, based on the overall distribution of extreme values, the probability of occurrence of a particular level of annual maximum daily rainfall in three Brazilian regions (Midwest, Southeast and South) and study their behavior over the past 71 years. The parameters of the general distribution of extreme values were estimated by the maximum likelihood method. The Mann-Kendall test showed that there is a positive trend in the annual maximum daily rainfall data series. The non-stationarity was rejected by the augmented Dickey-Fuller test supporting the use of the density function of extreme value distribution to describe the values of the occurrence of annual maximum daily rainfall. The Kolmogorov?Smirnov/Lilliefors goodness-of-fit test showed the good fit of the studied variable to the probability distribution function. The Midwest region has a return period of more frequent annual maximum daily rainfall below 300 mm in comparison with other regions. There is a clear change in the behavior of this extreme event in the Southern region. According to the literature, in past decades annual maximum daily rainfall of 248 mm has been estimated for a return period of 100 years for the state of Santa Catarina-South region, while the results found with the current series, annual maximum daily rainfall of 250 mm was estimated for a return period of 10 years. Extreme annual maximum daily rainfalls for return periods smaller were also found in other regions.

Publication year: 2014

Types of publication: Journal article

Unit: Embrapa Digital Agriculture

Keywords: Distribuição de valores extremos, Tendências do clima