LAUSR

Abstract The state of Alagoas has its economic base directed to agriculture, and the knowledge of wet and dry periods is fundamental for the success of agricultural enterprises. The objective was to define the wet and dry periods in the state of Alagoas via multivariate analysis applied to the Standardized Precipitation Index (SPI). Monthly rainfall data for 20 weather stations from 1960 to 2016 were used, with fault filling via imputation. Moreover, Cluster analysis (CA) was applied to the time series (definition of homogeneous regions) and SPI-12 (annual) - (determination of dry and wet spell periods). The annual accumulated rainfall showed two distinct periods: 1960 to 1990 (P1) and 1990 to 2016 (P2), according to Pettitt test. Both periods alternate according to the phases of La Nina and Neutral (rainy years) and El Nino (dry years). In the monthly rainfall, only the Eastern part of Alagoas increased the rainy months during long time series. In tested connection methods, the average connection, also known as cophenetic correlation coefficient (CCC = 0.8760), represented homogeneous groups of rainfall, featuring two regions: one on the coast (G1) and another hinterland and arid (G2); and two non-homogeneous groups (NA) in the state. The annual SPI helped identify dry and wet periods, regardless of El Nino-Southern Oscillation (ENSO) categorization. The SPI-12 categories show high annual and decadal variability in the groups, except for the extremely dry and wet categories. There is greater variability in dry and wet periods near the coast than hinterland of the state.