Limitations on El Niño predictability
Martin Puy (UTIG)
Fri, January 26, 2018, 10:30am - 11:30am
Host: Pedro Di Nezio
Abstract: El Niño, the recurrent warming pattern of the Pacific Ocean, affects weather patterns worldwide, causing droughts, flooding and long-lasting heat waves. Predicting these events enough in advance is therefore of crucial importance. El Niño forecasts have improved over the past three decades, yet our ability to predict these events is less than six months. In my seminar, I will discuss the factors that could explain this limited El Niño predictability. We focus on two recent events which exhibited very different evolutions and revealed limitations in our ability to predict El Niño. Based on long-standing theories, the scientific community expected that the event of 2014 was going to become a record-breaking El Niño. This event was heralded by the occurrence of strong westerly wind event (WWEs), random atmospheric fluctuations, which are thought to favor the initiation of El Niño events. Operational forecasts pointed indeed toward a strong El Niño, but instead a borderline event occurred. In contrast, a similar WWE led to one of the strongest El Niño record in 2015, with a magnitude similar to as the super El Niño of 1997. The reason for such large error in the prediction of El Niño's magnitude remains unknown. Motivated by this issue I explore the role of WWEs in the evolution of El Niño. First, we characterize the WWEs in the observations and show that their occurrence is associated with propagating disturbances in the tropical atmosphere that could be predicted a few weeks in advance. We also show that WWEs have an increased probability of occurrence when the equatorial Pacific is warmer than usual, such as during El Niño events. This relationship can reinforce the development of an El Niño. Then I performed forecasts with a model that simulates realistic WWEs. The model shows that even with favorable conditions for a strong El Niño to occur, a weak event, such as in 2014, can happen by random chance. Additional forecasts, where WWEs are hindered, confirm that the different outcome between the 2014 and 2015 El Niño are mainly caused by WWEs. These results show that, at lead times longer than six months, the prediction of strong El Niño may be limited by the unpredictability of WWEs.
