Evaluation of Benguela Niño/Niña events in the CMIP6 historical simulations

Historical Coupled Model Intercomparison Project Phase 6 (CMIP6) model outputs are analyzed to evaluate models' ability in simulating the seasonal and interannual variability in the South-Eastern Atlantic Ocean. The study focuses on SST in February-March-April, the main season of occurrence of the extreme interannual warm and cold Benguela Niño-Niña events. 

In the Angola-Benguela-Area (ABA), the CMIP6 SST ensemble-mean resembles the seasonality of the observations but with a strong bias. Unlike the seasonal cycle, the SST interannual variability in CMIP6 ensemble-mean is underestimated and occurs 3-4 months after the peak of maximum variability in the observations (June/July). Among the model ensemble, 2 groups of models emerge: a group featuring a maximum interannual variability in the right location (ABA) but delayed by about 2/3 months compared to the observation (~60% of the models); a group featuring the maximum variability in the right season (FMA) but in a location shifted southward in the South-Benguela (~30% of models). For the first group, results suggest that the time-shift of the peak in the SST variability is induced by the time-shift of the equatorial zonal wind stress. For the second group results show that the latitudinal-shift of the peak in SST variability is controlled by intense coastal wind activity in the south Benguela rather than by model bias and a southward shift in the position of Angola-Benguela-Frontal Zone. 

Finally, we examined the models’ ability in simulating extreme interannual Benguela Niño-Niña events. Very few individual models correctly simulate the phenology of the Benguela events, including the modulation of the equatorial zonal and coastal winds that drive development in the preceding months. Interestingly most of the best models have in common a fairly good representation of the South-Atlantic High-pressure system.