In this context, it is crucial to assess the performance of GCMs to determine whether they may represent realistically future changes in East African hydroclimate. The recent period, when direct measurements are available, is the first natural test but it is complemented here with analyses covering the last millennium in order to study also multidecadal to centennial variability. Specifically, we consider forced transient experiments simulated by six GCMs as well as lake-sediment records reconstructing hydroclimate conditions of four East African lakes: Lake Challa, Lake Naivasha, Lake Malawi and Lake Masoko.
While the unimodal seasonal cycle of precipitation characterizing the region including the lakes Masoko and Malawi is fairly well represented by GCMs, the magnitude of the two rainy seasons of Lake Challa and Lake Naivasha is less well captured. The comparison between model results and proxy-based reconstructions display very different time development over the last millennium. Additionally, there is no common signal among model time series. This suggests that simulated hydroclimate fluctuations are mostly driven by internal variability rather by a common forcing until 1850. After that, half of the models used simulate a relatively clear response to forcing, but this response is different between the models.
Overall, the link between precipitations and tropical sea surface temperature (SSTs) over the last millennium is stronger and more robust for the rainfall over the Challe/Naivasha region than over the Masoko/Malawi region. At interannual time scale, Challa/Naivasha precipitations are positively correlated with the western Indian Ocean and negatively with the eastern part of the basin, while the influence of the Pacific Ocean seems minimal and unclear. Although most of times not significant, the dipole in the Indian Ocean is still present using time series smoothed to highlight centennial variability, but only in unforced simulations. This means that, at the centennial time scale, the effect of forcing can overwhelm natural variability in large scale teleconnections.