Informal Seminar: Separating False Alarms from Successful Forecasts in NMME Confident Forecasts of El Niño Initialized in Boreal Spring

by Aaron Levy

Meeting Room 6

Meeting Room 6


El Niño is responsible for the largest part of the interannual climate variability, so forecasting El Niño events correctly is important.  Yet, challenges still remain in producing successful El Niño forecasts during the boreal spring time.  For instance, the dynamical seasonal forecast models produce confident forecasts predicting El Niño events at a rate more often than they occur.  There have been a few notable forecast busts under these conditions that have occurred in the last decade.  Here, we explore 30 years of hindcasts initialized from February through May across six models that are part of the North American Multi-Model Ensemble (NMME) to find commonalities in the forecast busts and their differences from the successful forecasts that had similar levels of forecast confidence.  We find that, in general, the models produce confident forecasts of El Niño for the upcoming winter with anomalously warm equatorial SSTs and recently warming SSTs, and positive equatorial heat content anomalies.  However, false alarm occurred when negative SST anomalies were present in the subtropical north eastern Pacific .  Further exploration shows that the dynamical forecasts models respond too strongly to anomalously warm SSTs along the warm pool edge, initiating a too strong Bjerknes response and positive feedback between SSTs, precipitation and wind stress. This leaves the dynamical forecasts vulnerable to an overly deterministic El Niño forecast without enough impact of subsequent sub-seasonal and synoptic scale forcing from both the tropics and the extra-tropics. These findings are further compared to the European suite of seasonal forecast models. The implications of these findings on El Niño forecasting and seasonal prediction models will also be discussed.

Organized by

Magdalena Alonso Balmaseda