Seasonal evolution of the modes of interannual variability of the South American monsoon

Abstract

This work focuses on the interannual variability of precipitation during the December-January-February (DJF) season and the seasonal evolution of the modes of variability of precipitation from September to May over the South American continent. For this, we use Empirical Orthogonal Function (EOF) and Seasonal reliant EOF (S-EOF) analysis as well as area averaged precipitation anomalies over Eastern Brazil to characterize precipitation variability over the region. Using EOF analysis we find that the dominant mode of DJF variability of interannual and longer time scales is characterized by strong precipitation anomalies over Eastern Brazil and anomalies of opposite sing over South East South America (SESA) and the western part of the Amazon Basin. This mode is referred to as the Continental tripole mode in this study. This dominant DJF mode is a robust feature across all datasets examined and shows little correlation to SST anomalies. The second DJF mode is characterized by a meridional dipole of precipitation anomalies forced by ENSO SST anomalies. During SON the first two leading modes of precipitation variability are related to Pacific SST anomalies. The first SON EOF mode is related to SST anomalies over the central Pacific and is characterized by a dipole of precipitation between SESA and the South Atlantic Convergence Zone (SACZ). The second SON EOF mode is related to eastern Pacific SSST anomalies and is characterized by a meridional dipole of precipitation over South America. During MAM, the dominant EOF mode is the meridional dipole mode forced by ENSO SST anomalies. Due to the weak relationship between the DJF ontinental tripole mode (EOF 1) of variability and contemporary SST anomalies, Grimm and Zilli (2009) proposed that precipitation anomalies during the previous SON season may, via local surface-atmosphere interactions, be a contributing factor to the development of this DJF mode. This hypothesis is based on the significant correlation between the time series of the SON EOF 1 and the DJF EOF 1. This type of relationship is only weakly or not at all reproduced by the datasets used in this study.