Using Co-Behavior Analysis to Interrogate the Performance of CMIP5 GCMs over Southern Africa.

As established in earlier research, analysis of the combined roles (co-behavior) of multiple climate processes provides useful insights into the drivers of regional climate variability, especially for regions with no singular large-scale circulation control. Here, we extend the previous study in order to examine the performance of eight models from phase 5 of the Coupled Model Intercomparison Project (CMIP5) in representing co-behavior influence on surface expressions over southern Africa. We find that although models broadly simulate observed precipitation responses over southern Africa, they fail to produce statistically strong response signals for an important drought pattern (El Niño co-behaving with positive Antarctic Oscillation during summer) for the region. We also demonstrate that the models show statistically strong temperature response signals to co-behavior that agree well with observed responses over the region. The multimodel ensemble mean although consistent with observations shows a larger spread. By elucidating the performance of models in representing observed co-behavior of climate processes, we are able to evaluate models while establishing important information for understanding of climate variability. [ABSTRACT FROM AUTHOR]