Evaluation of ENSO Prediction Skill Changes since 2000 Based on Multimodel Hindcasts

In this study, forecast skill over four different periods of global climate change (1982–1999, 1984–1996, 2000–2018, and 2000–2014) is examined using the hindcasts five models in North American Multimodel Ensemble. The deterministic evaluation shows that forecasting skills Niño3.4 Niño3 indexes are much lower during 2000–2018 than 1982–1999, indicating previously reported decline continues through 2018. decreases most significant for target months from May to August, especially medium long lead times, showing forecasts suffer more effect spring predictability barrier (SPB) post-2000. Relationships between extratropical Pacific signal El Niño-Southern Oscillation (ENSO) weakened after 2000, contributing a reduction inherent ENSO, which may be connected with times. It great challenge predict ENSO memory local ocean itself because weakening intensity warm water volume (WWV) its relationship ENSO. These changes decrease autocorrelation coefficient persistence short months. Moreover, both indexes, tend further underestimate sea surface temperature anomalies (SSTAs) Niño developing year but overestimate them decaying year. For probabilistic forecast, post-2000 also generally pre-2000 tropical Pacific, particular, they decayed east 120° W 2000. Thus, advantages methods, such as dynamic modeling, statistical machine learning should integrated obtain best applicability deal current low phenomenon.