Process‐Based Analysis of the Added Value of Dynamical Downscaling Over Central Africa.

In this study, nine global climate models (GCMs) and corresponding downscaled runs by means of the regional climate model (RCM) RCA4 are used to investigate added value (AV) in precipitation and its some drivers over Central Africa (CA). By employing a process‐based analysis approach, we intercompare abilities of RCM to those of driving GCMs in representing the total atmospheric moisture flux convergence (TMFC), moisture transport, and African Easterly Jets (AEJs). Results indicate that simulations with highest AVs in the precipitation climatology also show improvements in the representation of the TMFC and AEJs. Degraded precipitation due to the downscaling is associated with deterioration of at least two of three analyzed mechanisms, and sometimes there is inconsistent AVs between precipitation and related drivers. This sustains that a realistic representation of the moisture transport and atmospheric circulation is of great importance for the correct simulation of present (and, consequently, future) precipitation over CA. Plain Language Summary: Although the downscaling is proved beneficial for the African Coordinated Regional‐climate Downscaling Experiment domain, meaningful climate uncertainties persist; the origins of many biases/deficiencies in RCM results often remain unexplained, in a context where there is a requirement to ensure the fitness for purpose of the models used to generate climate information for decision‐makers. The added value is a statistical metric frequently used to judge whether an RCM has improved the results of the driving GCM or no. Most of the previous studies which have investigated added value over Africa have used the performance‐based evaluation method and are focused only on the diagnostic variables precipitation and temperature and therefore did not investigate reasons behind the RCM's amelioration or degradation. By the means of a process‐based analysis, this work shows that added value from the downscaling must be interpreted with a certain caution over Central Africa, as improvements in the representation of precipitation climatology are not always associated with improvements in related processes. Key Points: Experiments which feature clear improvement in precipitation also show improved moisture fluxes and AEJsDownscaled GCMs showing deteriorated precipitation also show deterioration in at least two of the analyzed processesThere is doubtful added value in some runs, highlighted through the process‐based analysis [ABSTRACT FROM AUTHOR]