Eta Model simulation of the South American Climate

Future climate projections suitable for local impact and adaptation studies require the use of regional climate models (RCMs) to downscale global climate models (GCMs). In the process of preparing the Eta RCM to produce projections using the CMIP6 GCMs, the Eta model has been integrated to produce a climate run. The objective of this work is to evaluate the Eta RCM simulation of the South American climate driven by ERA5 reanalysis. The model is set up at 20-km horizontal resolution with 38 vertical layers, covering the entire South American continent and parts of the adjacent oceans. The 30-year integration starts at 0000 UTC on January 1st, 1984, and ends on the same date in 2015. The climatology is assessed in four trimesters corresponding to the austral seasons: DJF (summer), MAM (autumn), JJA (winter), and SON (spring). The maximum precipitation is simulated in DJF over the Northern South America (NSA), extending into the South America Monsoon Region (SAM) and Southeast South America (SES), corresponding to the position of the South Atlantic Convergence Zone (SACZ). The minimum precipitation (< 1mm/day) is correctly simulated in JJA, mainly over SAM. Although the major features of the precipitation seasonal variation in the continent are reproduced by the simulation, there is a large overestimation over the Northwestern South America (NWS), a lack of precipitation band in the Intertropical Convergence Zone near the northern coast over the equatorial Atlantic Ocean, and an underestimation over SAM and SES. The seasonal variation of the 2-m temperature is well simulated over the continent; however, there is an overestimation in the region around Paraguay, Bolivia, northern Argentina, and western Brazil, and an underestimation in the eastern part of Brazil and southern Argentina. Climate extreme indices are calculated on a seasonal basis. The model correctly simulates the mean patterns of consecutive dry days (CDD), heavy precipitation days (R30mm), and total precipitation from very wet days (R95p). The CDD and R30mm trends are generally well simulated across seasons, but the positive R95p trend observed is not captured by the simulations. Similarly, the model correctly simulates the mean patterns of temperature extremes, including the warmest day (TXx) and warmest night (TNx). While the simulation shows an increasing trend in TXx across much of SAM in all seasons, the observations show greater spatial and seasonal variability in areas with a positive trend in TXx. The warm spell duration index (WSDI) shows a positive trend across the entire continent in both observations and simulations. The evaluation of extreme drought and rainy years in three major river basins will be presented.