El Niño – Southern Oscillation (ENSO), the large-scale fluctuation of the equatorial atmosphere and ocean over the tropical Pacific region with widespread climatic consequences, is arguable the most predictable climate mode at seasonal time scales and provides the scientific basis for global seasonal climate predictions. With significant progress made during the last decades in our understanding of the complexity of ENSO, the development of the observing system, coupled general circulation models and data assimilation techniques for improved forecast initialisation schemes, current forecast models can provide effective predictions of ENSO warm and cold events 6-12 months ahead.
A crucial component of any seasonal forecast system is the set of retrospective forecasts, or hindcasts, from past years that are used to estimate future forecast skill and to calibrate the forecasts for biases. Hindcasts of seasonal predictions are usually produced over a period of around 20–30 years. With an average frequency of 4-5 years, there is only a very limited number of ENSO cases available in the operational reforecast records and sampling the wide spectrum of ENSO flavours is not possible. In the presence of considerable variations in the coupled ocean-atmosphere system, good skill in predicting the most recent ENSO events cannot guarantee that future events will have similar predictability.
In this talk, I will introduce new historical retrospective research forecasting datasets created with a version of the ECMWF forecasting model that cover all of the 20th Century and extends the forecast lead times to 2 years. The reforecasts show substantial decadal modulations of forecast skill. In particular, skill to predict ENSO is very high during recent decades, but it is markedly reduced during the 1930s–1950s. ENSO skill at the beginning of the century is, however, as high as for recent high-skill periods suggesting that the loss of skill in the mid-century period is not related to the lack of good observational data. Alternative hypotheses related to changes in the mean state & variability as well as persistence characteristics and the role of the observing system will be discussed.