The NCEP Eta model coupled with a biosphere model, SSiB, was set up over the South American continent to study the role of land/atmosphere interactions in the South American hidrometeorology and regional climate predictability.

NCEP Reanalysis and NCEP GCM output were used as initial and lateral boundary conditions for the simulations. Sea surface temperature, sea ice concentration and snow cover were updated daily during the simulation. Results were compared to GCM and Reanalysis data as well as to observation to evaluate the dynamic downscaling of the regional climate model in regional hydrometeorological study. To understand the predictability, a series of sensitivity studies has been designed to explore the role of variety of factors in water cycle simulations. These factors include domain size, horizontal resolution, different lateral boundary conditions (NCEP reanalysis and GCM outputs), sea surface temperature, and land conditions. The first part of this study is shown herein.

Preliminary results indicate the regional model was able to capture most of the features of precipitation distribution over South America when compared to the GCM results and Reanalysis data. Improvement was greater for the dry and transition seasons. Eta improvement was less significant in the wet season when the model underestimated precipitation over much of the Amazon Basin, ITCZ and southern Brazil.

The study shows the importance of a regional model and the proper land surface processes representation in the South American rainfall simulation and the role of more realistic boundary condition description in predicting its regional climate.