Multi-Model Ensemble Prediction during Central and Eastern Pacific El Niño Events

Thursday, 18 December 2014
Johnna Infanti, University of Miami, Miami, FL, United States and Ben P Kirtman, University of Miami - RSMAS, Miami, FL, United States
Recent research has shown that there are events where SSTA warming peaks in the eastern Pacific (EP El Niño) and where warming peaks in the central Pacific (CP El Niño), and CP events seem to be more prevalent in recent years. From a prediction standpoint, the ability of climate models to predict the differences in events may be important, as differences lead to distinctive climate impacts over regions such as North America. Research using the North American Multi-Model Ensemble (NMME) showed a tendency for excess eastern Pacific warming during CP events. Additionally, observed and modeled CP events tend to be weaker than EP events. One might assume that the comparatively weaker CP events would lead to weaker teleconnections and less predictability in regions such as North America. However, large scale, expected impacts are captured in NMME predictions for both types of events. Using ensemble agreement on the sign of the anomaly as a proxy for prediction skill in NMME, the strength of the event in the eastern or central Pacific leads to larger ensemble agreement in some North American regions. Precipitation ensemble agreement in the southeast US demonstrates a strong connection to eastern Pacific warming, and strong warm events lead to larger ensemble agreement. Precipitation ensemble agreement in the northwest US demonstrates a weak connection to central Pacific warming. Other regions, such as Northwestern US 2-meter temperature, do not show any change in ensemble agreement with strength of event. While the strength of event is only important for some North American regions, the ability of model predictions to correctly capture the position of warming for teleconnections is nevertheless of concern. Anomalously high SSTA in the eastern Pacific during CP events could lead to excessive forecast agreement in regions such as the Southeast, where larger NINO3 anomalies lead to larger ensemble agreement.