This study investigates the performance of the Unified Forecast System (UFS) global coupled model to forecast the relationship between the Madden-Julian Oscillation (MJO) and El Niño- Southern Oscillation (ENSO) through the calculation of the MaKE and MaKI indices, developed in the ocean-atmosphere energetic framework. The MaKE index describes the covariability of MJO, Kelvin wave activity, and ENSO. The MaKI index characterizes the influence of MJO and Kelvin wave activity on ENSO through the wind power. The indices are calculated through the MJO components of meridional and zonal wind stress, the Kelvin components of meridional and zonal ocean surface currents, and sea surface temperature daily anomalies. UFS reforecasts of two prototypes for April 2011-March 2018 are used. Each reforecast has 35-day forecast leads, which poses challenges for isolating the MJO component of wind stress. To address this limitation, a novel filtering approach using a convolutional neural network (CNN) artificial intelligence model was implemented to capture the intraseasonal variability of the wind stress anomalies. Prototypes 6 and 8 are selected due to the exclusion/inclusion of the near sea surface temperature model (NSSTM).
The results show that the UFS prototypes can predict the El Niño event in 2015-2016, and appropriately predict the MJO-ENSO interaction for the event. The deterministic forecast skill of the model predictions of MJO, ENSO, and the MJO-ENSO relationship compared to the observations are analyzed using pattern correlation. The NSSTM shows improved wind stress forecasting for high frequencies on daily timescales, but no effect on the intraseasonal variability, and thus no impact on the MJO-ENSO relationship.
