Sub‐monthly variability of the South American summer precipitation under El Niño and La Niña backgrounds during the 1998–2012 period

ABSTRACT The sub‐monthly intra‐seasonal 8–24‐day period (SIS8‐24) timescale variability of the summer (1 November to 31 March) rainfall over South America (SA) under distinct inter‐annual (IA) backgrounds was analysed using the Tropical Rainfall Measure Mission (TRMM) based daily total precipitation data for the 1998–2012 period. The IA backgrounds refer to the El Niño (EN) and La Niña (LN) years. First, the summer daily precipitation anomaly fields were subject to the empirical orthogonal function (EOF) analysis. Using the first and the second principal component time series filtered at the SIS8‐24 timescale, positive and negative events were selected. The first SIS8‐24 mode features a precipitation anomaly dipole with centres over southeastern SA (SESA) and central and eastern tropical SA both extending southeastwards into the adjacent Atlantic Ocean. The second mode features a precipitation anomaly pattern similar to that previously documented for the oceanic South Atlantic convergence zone (SACZ). The SIS8‐24 precipitation anomaly patterns for the positive (negative) events show differences in the anomaly intensities between EN and LN years, but with almost the same locations of the anomaly centres. These differences result from the variability inter‐SIS8‐24 events, as indicated by the distinct paths of the SIS8‐24 Rossby wave train patterns in the subtropics, although they are regionally locked over tropical SA. The relation between the SIS8‐24 and IA variabilities might occur through variations in the Rossby wave train patterns. The Rossby wave trains of both timescales, depending on their phases, reinforce or weaken the rainfall anomalies over SA, in such a way that the rainfall anomalies over SA show similar patterns but with distinct magnitudes for EN and LN composites. The South American precipitation responses to the combined SIS8‐24 and IA variability timescales stratified according to the IA backgrounds have not been studied before and might be useful for operational forecasting services.