The seasonal cycle of precipitation in tropical South America shares general characteristics with classical monsoon climates in other parts of the world and is therefore referred to as the South American Monsoon System (SAMS). Monsoon precipitation has a profound influence on important interests in South America, including agriculture and hydroelectric power production. These economic sectors could benefit from improved understanding of the monsoon onset, and how the pattern of onset changes from year to year. Moreover, understanding the mechanisms, predictability and variability of the SAMS can improve seasonal climate forecasting in North America since the two regions are linked via variations in the Hadley circulation, and share trends of interannual variability for example via the ENSO mode.

Previous studies of SAMS onset are not in general agreement regarding the timing, pattern and mechanisms of onset across South America, where different data and methods were used to characterize onset. In this paper, we build on these previous studies to study the spatial and interannual variability of SAMS onset using homogeneous criteria and a well-established, quality controlled precipitation dataset over all of tropical and subtropical South America for 1979-2006. To accomplish this goal, we used the Global Precipitation Climatology Project (GPCP) 2.5 degree pentad averaged dataset, which provides rainfall estimates from 1979 - present using an optimization of satellite and gauge data. The strength of the GPCP product is that it is uniform in time and space, covers all of South America, is tuned to the more physically direct precipitation retrievals of microwave satellites, is calibrated with surface rain gauges, and is subject to stringent quality control. We then evaluate mechanisms for the spatial and interannual variability of SAMS onset using detailed observations of the horizontal and vertical structure of convection from the Tropical Rainfall Measuring Mission (TRMM) satellite precipitation radar, microwave radiometer, and lightning imaging sensor. Through a better understanding of onset mechanisms, we will begin to assess the sources and limits of predictability of the SAMS using GCM simulations.