Understanding spring forecast El Niño false alarms in the North American Multi-Model Ensemble

Abstract

El Niño is responsible for the largest part of the seasonal-to-interannual climate variability, so forecasting El Niño events correctly is important. However, forecasting El Niño events during boreal spring remains challenging. The dynamical seasonal forecast models of the North American Multi-Model Ensemble are over-confident for high confidence (>75% ensemble member agreement) El Niño forecasts. In general, confident El Niño forecasts have a warming tendency in equatorial SSTs in the month prior to the forecast initialization and positive equatorial heat content anomalies during the first month of the forecast. However, confident forecasts often fail when negative SST anomalies were present in the subtropical north eastern Pacific. We find that the models’ equatorial SST anomalies persist too long and that the precipitation response along the warm pool edge to these anomalies is too deterministic. Therefore, the forecast models are too reliant on coupled equatorial processes resulting in excessively deterministic forecasts.