The Role of Stochastic Forcing in Ensemble Forecasts of the 1997/98 El Nino

The impact of stochastic intraseasonal variability on the onset of the 1997/98 El Nino was examined using a large ensemble of forecasts starting on 1 December 1996, produced using the Australian Bureau of Mete- orology Predictive Ocean Atmosphere Model for Australia (POAMA) seasonal forecast coupled model. This coupled model has a reasonable simulation of El Nino and the Madden-Julian oscillation, so it provides an ideal framework for investigating the interaction between the MJO and El Nino. The experiment was designed so that the ensemble spread was simply a result of internal stochastic variability that is generated during the forecast. For the initial conditions used here, all forecasts led to warm El Nino-type conditions with the amplitude of the warming varying from 0.58 to 2.78C in the Nino-3.4 region. All forecasts developed an MJO event during the first 4 months, indicating that perhaps the background state favored MJO development. However, the details of the MJOs that developed during December 1996- March 1997 had a significant impact on the subsequent strength of the El Nino event. In particular, the forecasts with the initial MJOs that extended farther into the central Pacific, on average, led to a stronger El Nino, with the westerly winds in the western Pacific associated with the MJO leading the development of SST and thermocline anomalies in the central and eastern Pacific. These results imply a limit to the accuracy with which the strength of El Nino can be predicted because the details of individual MJO events matter. To represent realistic uncertainty, coupled models should be able to represent the MJO, including its propa- gation into the central Pacific so that forecasts produce sufficient ensemble spread.