Application of multivariate methods for the characterization of meteorological drought periods in Venezuela

Abstract

Multivariate analysis consists in determining if there are simpler ways to represent a complex set of data, besides exploring if the observations are concentrated in groups and if there is interdependence between the elements. These types of techniques have been widely used to analyze climatological data. That is why the objective of this research was to characterize the occurrence of drought periods by means of time series of the Standard Precipitation Index (SPI) for seven agricultural locations in Venezuela, by using two multivariate methods. The conditions of deficit or excess of precipitation in the agricultural localities were quantified through the SPI with a monthly scale of the period 1980-2014. The combination of two multivariate methods was used for the analysis: Principal Coordinate Analysis of the data matrices using Euclidean distance and Cluster Analysis. Two or three groups of years of SPI were described in the seven locations. In the case of the locations that resulted with three groups (CENIAP, El Cují and Yaritagua), these were categorized in wet years, intermediate years and years with significant water deficit. Meanwhile, in the case of the locations that resulted with two groups (Turén, Quíbor, Mucuchíes and Bramón), they were classified into wet years and years associated with the water deficit. The application of these multivariate methods made it possible to identify the dominant monthly spatial patterns of the SPI on the studied locations, besides being related to the occurrence of local droughts of great importance from the agricultural point of view.