Understanding model diversity in future precipitation projections for South America

Abstract Precipitation patterns are expected to change in the future climate, affecting humans through a number of factors. Global climate models (GCM) our best tools for projecting large-scale changes but they cannot make reliable projections locally. To abate this problem, we have downscaled three GCMs with Weather Research and Forecasting (WRF) model 50 km horizontal resolution over South America, 10 central Chile, Peru southern Brazil. Historical simulations years 1996–2005 generally compare well precipitation observations reanalyses. Future Chile show reductions annual increases dry days at end-of-the-century high greenhouse gas emission scenario, regardless GCM boundary conditions used. However, Brazil more uncertain, that increasing can switch sign projections. Differences between global/regional (10 km) only mildly influenced by orography resolution, linked convection parameterization, reflected very different static energy flux divergence, vertical velocity layer height. Our findings imply using results directly from GCMs, even coarse-resolution (50 regional models, may give incorrect conclusions about regional-scale While modelling convection-permitting scales is computationally costly, scale-aware instead conventional scheme, better mirror fine-resolution