The associations between El Nino-Southern Oscillation and tropical South American climate in a regional climate model

High-resolution regional climate models (RCMs), run over a limited domain, are increasingly used to simulate seasonal to interannual climate variability over South America and to assess the spatiotemporal impact of future climate change under a variety of emission scenarios. These models often give a better spatiotemporal representation of climate at a regional scale; however, they are subject to errors introduced by the driving global models. Here we analyze two different simulations with the Hadley Centre Regional Climate Modeling System Providing Regional Climate for Impact Studies (PRECIS) model over tropical South America. The two simulations cover the same 30 year period (1961-1990) but were forced with different lateral boundary conditions. The first simulation was forced with the Hadley Centre Atmospheric Model version 3 (baseline), and the second was forced with European Centre for Medium-Range Weather Forecasts Re-analysis (ERA) data. Our results indicate that the ERA-forced simulation more accurately portrays seasonal temperature and precipitation, consistent with previous studies. Empirical orthogonal function and spatial regression analyses further indicate that the ERA-forced simulation more realistically simulates the El Nino-Southern Oscillation related fingerprint on interannual climate variability over South America during austral summer. The two gridded observational data sets used for model validation display large differences, which highlight significant uncertainties and errors in observational data sets over this region. In some instances the observational data quality is rivaled or even surpassed by the ERA-forced RCM results.