Characteristics of predictable MJOs

Abstract:

The Madden-Julian Oscillation (MJO) has been considered as a major potential source of global climate predictability on subseasonal time scales. Current operational forecasting systems are able to predict the MJO up to 3-4 weeks with the mean of ensembles, while the skill is still below the theoretical estimates of the predictability (6-7 weeks). It is well accepted that MJO prediction skill in operational systems is distinctly better when the MJO is well organized with having strong amplitude at the beginning of the forecast compared to when it starts with weak or nonexistent MJOs. However, while initially strong MJOs have better skill than weak/non MJOs, the ‘initial amplitude-skill’ relationship is not linear. It is not every initially strong MJO that has high skill, and not every weak/non MJO that has low skill. What physical mechanism drives some MJOs to be more predictable than others? Using dynamical MJO reforecasts, this study will show that the skill of highly predictable MJOs depends on the background environment that influence on MJO evolution. The favorable conditions for highly predictable MJOs (starting with moderate amplitude) support for stronger ocean-atmosphere coupling, and for sufficient heat and low-tropospheric moisture supply for better MJO propagation, particularly for the MJO crossing over the Maritime Continent. Understanding the characteristics of predictable MJOs may provide insights into the overall predictability of MJO, thus advance the prediction of 3-4 weeks towards its theoretical limit.