The El Niño Southern Oscillation is a recurring climate event in the tropical Pacific Ocean with a period of between 2 and 7 years. It is a coupled phenomenon of sea surface temperatures and zonal wind anomalies in the equatorial Pacific. ENSO has a far-reaching effect and leads to extensive rainfall and floods or anomalous droughts in certain regions of the globe, thus affecting local agriculture. It also has a substantial impact on the marine ecosystems and the sustained exploitation of marine resources in some coastal zones, especially the Pacific South American coast. Such events like floods, drought and temporary extinction of the marine life could lead to devastation in various regions of the world. Therefore, ENSO prediction is very important so that preparations could be made on time and communities could be protected from its consequences. Good forecasting services could save human lives and millions of dollars from being spent on storm/flood, drought or famine damage.
Desislava Petrova, a PhD student, and her supervisor Dr Xavier Rodo from the Unit of Climate Dynamics and Impact in IC3, in collaboration with Professor Siem Jan Koopman from VU Amsterdam have developed an El Niño prediction model based on unobserved components time series, using a state space approach. In this approach the real state of the dynamical system is directly accounted for in the model through the use of explanatory covariates. The prediction system that has been developed is based on several predictor variables - subsurface ocean temperature data in the western equatorial Pacific, a high latitude Southern Pacific Ocean temperature difference tracer – the RossBell dipole, and horizontal wind stress data in the western to central equatorial Pacific region. Based on theoretical study of the onset stage of El Niño different predictors are used for the different stages of development of an EN event. Within the framework of this approach the lag times at which every explanatory covariate is incorporated in the forecasting scheme is estimated by the Kalman Filter. Thus, at every lag only one or a combination a few of the above-mentioned variables is used depending on the dynamics of the system and on the estimation results.
In May SSTs were above average in the eastern equatorial Pacific and in some parts of the central equatorial Pacific, but the overall state is still defined as ENSO-neutral by the Climate Prediction Center. It is important to note that the oceanic heat budget close to the Equator has been the highest since historical records began this past March, with substantial subsurface temperature anomalies (Fig.2 a and b). This fact along with low level westerlies and enhanced convection over the central equatorial Pacific are all compatible with the developing stage of an El Niño event (Fig. 2 c). The significant subsurface anomalies have already reached the surface in the eastern equatorial Pacific (east of 125 W), further supporting the prognosis for an El Niño in the summer and over the rest of the year (Fig.3). Provided that in our prediction system we use subsurface temperatures in the western tropical Pacific and horizontal wind stress in the central tropical Pacific, it is consistent that our forecasts indicate a high chance of EN evolving over the summer, autumn and winter of 2014. The probability of EN occurring by winter is close to 85% (an increase of about 25% in comparison with the earlier March forecasts). Although the uncertainty of how strong the event might be is still high, according to the forecast it will be a mild to moderate event with some chance of a major El Niño as can be seen looking at the upper confidence intervals in the two forecasts (Fig.1).