Evaluation of interannual variability simulation over South America using a dynamic downscaling approach

The potential predictability of seasonal-interannual climate variability has been an increasing concern in the past years. In this study, the dynamical downscaling technique is used to provide a general assessment of potential predictability in South America (SA) with focus on the interannual climate variability. Comparison between the summer (December to February) precipitation in 1988 and 1997 show a large difference in the seasonal mean over subtropical SA. 1997 was anomalously wet while 1988 was anomalously dry. The same difference is also found in the winter (July to August) between 1995 and 1997 with the former (latter) being dryer (wetter) than normal over southern Brazil.

Five 3-month ensemble hindcasts, differing in the initial conditions for the atmosphere, for each year were carried out with an 80-km-resolution version of the National Center for Environmental Prediction (NCEP) regional ETA model nested in the NCEP atmospheric general circulation model (AGCM) to investigate the models predictability. AGCM simulations were initialized with NCEP reanalysis. Observed time-varying daily observed SST was used in all simulations. Initial sea ice and snow cover for both models were taken from observed daily and weekly datasets, respectively.

In general, results indicate that the nested model was able to improve AGCM simulations and capture most of the interannual anomaly signal in both seasons. AGCM failed to simulated excessive rainfall during summer of 1997 over subtropical SA, which was well simulated by the nested model. Both models simulated the dry conditions in the Amazon River Basin during the summer of 1988.

The nested model results were also superior to AGCM's for the winter season. Eta model captured the anomalous wet conditions over southern Brazil as well as the dry conditions along the northeastern coast of the continent, both not present in the AGCM outputs. Analysis of low- and high-level wind indicated that precipitation in the nested model responded to both continental and regional scale circulation anomalies, such as intensification of SA low-level jet and Walker cell shift, probably associated with El-Nino.