Análisis de la predicción de precipitaciones mediante correlaciones canónicas en el NE de Argentina

Abstract

The forecast of the climate variability is of vital importance for the economy of an agricultural and livestock zone like Northeast of Argentina. Numerous research studies demonstrate that the extreme variation of precipitation in South America, like precipitation excess and droughts, are related to El Niño or La Niña events. Other works show the influence of the Atlantic temperatures on the variation of precipitation in some South American regions. 

The goal of the present work is the application and evaluation of the predictive capacity of a statistical method of long-term forecast of precipitation over northeastern Argentina, based on the sea surface temperature over south Pa- cific Ocean and the south Atlantic Ocean. The method consists of the canonical correlation analysis of empirical orthogonal functions of the fields of sea surface temperatures and precipitation. Then, an analysis of real possibilities of the statisti- cal prediction over the zone of interest is performed through the forecast of humid, normal and dry periods. 

The correlations show that the best results are obtained for spring and summer. In particular, the prediction of wet periods stands out when Pacific Ocean temperatures are used, due to their general link to the occurrence of an El Niño event in the previous months. Dry summers can be adequately forecast using the Atlantic Ocean temperatures. That is, the statistical prediction is useful only for certain seasons of the year. Nevertheless, for these seasons, the method can estimate the conditions with 5-7 months in advance.