Rainfall in uncoupled and coupled versions of the Met Office Unified Model over Central Africa: Investigation of processes during the September–November rainy season.

Although climate models are important for making projections of future climate, little attention has been devoted to model simulation of the complex climate of Central Africa (CA). This study investigates rainfall biases through processes in three versions of the Met Office Unified Model over CA with both coupled and atmosphere‐only formulations for each version. The study shows that the models depict a wet (dry) bias over the eastern (coastal western) CA in the September–November season with the wet (dry) bias stronger in coupled (atmosphere‐only) models. Here, we explore potential regional to large‐scale causes of these biases. Results reveal that the overestimation of the simulated sinking branch of the Atlantic‐Congo zonal overturning cell is associated with a strong near‐surface temperature and pressure gradient between CA and the eastern Atlantic Ocean. This leads to strong low‐level westerlies (LLWs) which dry the coastal western CA and strengthen the intensity of the Congo basin cell. Over eastern CA, the wet bias is partially due to the misrepresentation in the strength of both African easterly jet components that shift the mid‐tropospheric moisture flux convergence southward favouring more convection south of the equator. Furthermore, the overestimation in the simulated width and intensity of the Congo basin cell is associated with a strong low‐level moisture convergence over eastern CA which contributes to more precipitation. Remote contributions to the wet bias come from both the Atlantic and the Indian Oceans. The simulated Atlantic‐Congo zonal overturning circulation dries the coast through its overestimated lower branch (LLWs) which moves further into the continent and advects more moisture to the eastern CA. In the meantime, during positive Indian Ocean Dipole years, the advected moisture from the Indian Ocean to the CA region is overestimated in models, much more in coupled models and contributes to the eastern CA wet bias. [ABSTRACT FROM AUTHOR]