An assessment of the ENSO forecast skill in the monsoon mission coupled forecast system

This study comprehensively assesses the predictive capabilities of the Monsoon Mission Coupled Forecasting System (MMCFS) output for El Niño-Southern Oscillation (ENSO) conditions and its transitions in the various phases. The MMCFS model is being run for the operational seasonal forecast at the India Meteorological Department (IMD), and understanding the model skill for different seasons and various leads is imperative for informed decision-making in various climate-sensitive sectors. This analysis reveals various aspects of the model’s skill for the forecast of Niño 3.4 index by quantifying deterministic and probabilistic verifications across seven leads during all the three-month running seasons throughout the year for the period 1991–2020. Analysis indicates predominant cold bias across the Niño 3.4 region. The results highlight that the root mean square error (RMSE) is always less than the observed standard deviation (SD) regardless of the season for lead 1. The lowest RMSE for the JJA season using lead 1 is about 0.35℃, while the observed SD for the same season is approximately 0.61. Reliability and receiver operating characteristic (ROC) curve suggest higher forecasting skills for shorter leads. This study also analyses the performance of the MMCFS model on the four types of ENSO transitions, such as from El Niño (La Niña) to Neutral ENSO and Neutral ENSO to El Niño (La Niña), to enhance the understanding and provide insights for improved seasonal forecasting. Lead 1 shows significant skill in predicting ENSO transitions compared to other leads. Results reveal that the MMCFS captures four transitions of ENSO with a 95% level of significance pattern correlation coefficient (PCC) values. The acquired insights in this study deepen our understanding of ENSO forecasting, which is pivotal in facilitating informed decision-making processes across diverse climate-sensitive sectors, vulnerable to the multifaceted impacts of ENSO-related phenomena.