Multi-Decadal Convection Permitting Dynamical Downscaling of Current and Future Climates over South America

This paper reports on two sets of 22-year convection-permitting regional climate simulations conducted with the Weather Research and Forecasting model over the entirety of South America at 4-km grid spacing. The first simulation downscales the fifth-generation European Centre for Medium-Range Weather Forecasts atmospheric reanalysis (ERA5) to reproduce the present-day climate conditions over the period 2000-2021. The second simulation projects late-century climate change using the Pseudo-Global Warming (PGW) approach, in which the meteorological reanalysis fields are perturbed using monthly climate change signals from the Community Earth System Model Large Ensemble (LENS2), corresponding to ~3°C of global warming under the CMIP6 SSP3-7.0 scenario during 2060-2080. Results show that the historical simulation successfully captures multiscale climatological characteristics of precipitation and near-surface air temperature, including their interannual cycle down to diurnal variabilities, as validated against multiple satellite and reanalysis datasets, and demonstrates clear added value compared to the original ERA5 data, particularly at daily to sub-daily temporal and mesoscale to local spatial scales. The PGW simulation projects significant continental-scale warming, especially over the Andes, broadly consistent with LENS2 projections. Precipitation changes are spatially and seasonally complex, with most regions undergoing either increases or slight decreases, while the Amazon Basin exhibits predominant warm season drying. Additionally, extreme precipitation intensifies across the continent, and snowfall and snowpack decline over the Andes. These simulations provide a first-of-its-kind long-term high-resolution dataset for the climate science and policymaking communities, enabling detailed studies of atmospheric, land-surface, and hydrological processes in South America and their changes under a warming climate.