Predictability of Precipitation and Intraseasonal Variability: Insights From ECMWF Model Skill Over Brazil

Abstract

This study assesses the capabilities and limitations of the state-of-the-art European Centre for Medium-Range Weather Forecasts (ECMWF) subseasonal to seasonal (S2S) model in forecasting precipitation and a regional intraseasonal oscillation index over Brazil. Distinct from previous studies, we employ a regional index rather than a global one, enabling a more focused analysis of the complex intraseasonal variability. Weekly accumulated precipitation forecasts are evaluated against satellite-derived precipitation data for selected regions within Brazil. Our findings indicate that the ECMWF model demonstrates enhanced forecast skill for up to 4 weeks along the northern coast of Northeast Brazil, where tropical–tropical teleconnections linked to the Madden–Julian Oscillation significantly improve predictability. However, forecast accuracy decays after 2 weeks in subtropical and extratropical areas, such as the South Atlantic Convergence Zone and Southern Brazil, primarily due to challenges in capturing synoptic-scale systems and tropical–extratropical interactions. Additionally, the ECMWF model shows strong predictability for the regional intraseasonal oscillation index up to 10 days, offering valuable insights for planning and decision-making in the face of extreme weather events. This regional index achieves a level of accuracy not possible with a global index, which is less effective at capturing the intraseasonal signals specific to South America and their impacts on extreme weather.