Análise de frequência de secas utilizando análise de agrupamento e distribuições de probabilidades.

Abstract

The objective of this study was to adjust probability density functions to data of severity and duration of droughts in three sub-regions of the semiarid region of Brazil (SAB). To accomplish this task, the totals monthly precipitated of 320 rain gauges were used in the period 1984-2014. These data were kindly provided by the National Water Agency (ANA). Then, was used the method "RUN" to get the severity and duration of droughts based on totals monthly precipitated. Then we used the Cluster Analysis technique to regionalize the severity and duration of droughts. The next step was to get the adjustment of data to Probability Distribution Function (PDF) for each sub-region. The GEV and Logistics were the PDFs that best fitted to the severity of droughts. On the other hands, the GEV models, Weibull three parameters and Gamma with two parameters were adjusted better to the duration of droughts in the sub homogeneous-regions 1, 2 and 3 respectively. Lastly, the severity and duration of droughts were adjusted to normal bivariate distribution. This adjustment was not successful in any subregion of the SAB studied.