Aplicabilidade da distribuição GEV ao estudo da precipitação máxima diária anual na região sul de Minas Gerais

Abstract

The rainfall is one of the climates elements with great influence on human activities, and the studies of extremes events are very important in the sizing of hydraulic constructions. The probability distribution, GEV (Generalize Extreme Value), has been applied to hydrology for estimate the events of maximum daily annual rainfall, associated with extreme frequency (high return time). The objective of this paper is to analyze the GEV distribution, adjusted using the Moments Method (MM), Maximum Verisimilitude (MMV) and Moments-L (ML). For this, was used historical data from 8 rainfall stations located at south of Minas Gerais state. The results has showed that the ML method provided the best adjustment for the GEV distribution, being the only one that provided statistic adequacy for the 8 historical data by using Fiberfill test. The maximum daily annual rainfall for the Poços de Caldas and Silvianópolis station, where presented critical situation, ranged between 167 and 297 mm, with the return time of 50 and 500 years, showing the potentiality for tragedies related with extreme events of rainfall in this areas. However, the lowest values occurred at Camanducaia and Monte Sião stations, ranging between 103 and 174 mm for the same return interval. Generally speaking, the GEV distribution adjusted using the MV, and mostly ML methods, are presented as an alternative for estimate maximum daily annual rainfall using different return times.