Aplicación de métodos multivariados para la caracterización de periodos de sequía meteorológica en Venezuela

Multivariate analysis consists of determining if there are simpler ways of representing a complex data set, as well as exploring if the observations are concentrated in groups and if there is an interdependence between the elements. These types of techniques have been widely used to analyze climatological data. For this reason, the objective of this investigation was to characterize the occurrence of drought periods using time series of the Normalized Precipitation Index (SPI) for seven agricultural localities in Venezuela, using two multivariate methods. Through the SPI, the deficit or excess precipitation conditions in the agricultural localities were quantified on a monthly scale for the period 1980-2014. For the analysis, the combination of two multivariate methods was used: the Principal Coordinate Analysis of the data matrices using Euclidean distance and the Cluster Analysis. In the seven localities, two or three groups of years of RLS were described. In the case of the localities that resulted with three groups (CENIAP, El Cuji and Yaritagua) these were categorized into wet years, intermediate years and years with significant water deficits. Whereas, in the case of the localities that resulted with two groups (Turen, Quibor, Mucuchies and Bramon) they were classified as wet years and years associated with water deficit. The application of these multivariate methods allowed identifying the monthly spatial patterns of the SPI over the studied localities, in addition to being related to the occurrence of local droughts of great importance from an agricultural point of view.