Analysis of drought frequency using clustering technique and probabilities functions

Abstract The objective of this work was to adjust probabilities density function to drought indexes: Severity and Duration in three sub-regions of Brazilian semi-arid (BSA). For that, were used data of total monthlies precipited of 320 rain stations for period from 1984 to 2014, kindly provided by the Agência Nacional de Águas (ANA). Were used the “Run Method” and total monthlies precipited for obtain the monthlies indexes that characterize the drought: Severity (mm) and Duration (months). We then used the Cluster Analysis technique to regionalize theses characteristic drought indexes. The following step was to adjust probabilities density functions to drought indexes: Severity and Duration in three sub-regions of in the Brazilian semi-arid (BSA). The Gamma bi-parametric, Generalized Extreme Value (GEV) and Logistics functions were better adjusted to the indices of drought Severity (mm) and the GEV, Weibull tri-parametric and bi-parametric models, were better adjusted to drought indices Duration, to the homogeneous sub-regions 1, 2 and 3, respectively.