Comparison of RCM and GCM projections of boreal summer precipitation over Africa

Abstract

AbstractTo provide input data for the potential future climate change impact assessment at regional levels, dynamic downscaling of global climate models (GCMs) climate using regional climate models (RCMs) is frequently utilized. It is important to understand how the climate change signal is modified in the RCM driven by GCM so that the simulated results can be accurately interpreted. This study compares the performance of a RCM and four driving GCMs in simulating precipitation over Africa and investigate how dynamically downscaled present and future climate from the regional model RegCM4.1 driven with four GCMs differ from those of the GCMs. In general, RegCM4.1 has a lower summer (June–August) precipitation bias than the driving GCMs, increasing our confidence in their future projections. Despite uncertainty in future projections, RegCM4.1 shows decreased precipitation over Sahel regardless of which GCM is used to drive the RCM. RegCM4.1 shows an increased interannual variability of summer precipitation over Gulf of Guinea and Central Africa as compared to GCMs in both present‐day climate and projected future climate scenarios. The mean annual cycle of precipitation differs between RCM and the driving GCM, in particular during the West African Monsoon (WAM). The WAM onset, peak, and retreat phases are more clearly defined in the RegCM4.1 than the four driving GCMs, where the WAM rainband shifts more southward. This paper highlights the uncertainties in dynamical downscaling for climate prediction, which are due to the inconsistencies in the physical packages between RCM and GCMs.