We construct an MJO detection method by modifying Realtime Multivariate MJO (RMM) index (Wheeler and Hendon 2004) to identify dates under atmospheric states commonly associated with the MJO in the satellite observed Outgoing Longwave Radiation (OLR), and daily mean zonal winds in the reanalysis data. Events are detected during those potential MJO dates by incorporating continuity to the RMM time sequence, and requiring linear relationship between the phase speed on the RMM phase space, and propagation speed of active convection calculated by daily tracking of OLR minimum.
To characterize fast and slow MJOs, propagation speed of the convective activity is calculated from tracking daily minimum OLR of each detected MJO event. Analysis on the fastest and the slowest 10 MJO events revealed that they occur under different oceanic states in which the slower events are associated with condition of higher SST in the Western Pacific than the Indian Ocean. Extending the analysis of propagation speed to rest of the events revealed that the relationship between zonal SST difference and the propagation speed is a relationship that appears to be broadly shared by all of the MJO events detected in this analysis. Furthermore, the analysis on the evolution pattern of the SST field during a lifetime of an MJO seemed to indicate that slow and fast MJO both act to reduce the zonal SST difference anomaly to a similar level. The results of this research suggest that reduction of zonal SST difference to be one of the essential characteristics of the MJO, and therefore the propagation speed of the MJO is largely regulated by the zonal SST difference between the Western Pacific and the Indian Ocean.
Following this analysis, reproducibility of MJO in the AMIP type simulation on Non-hydrostatic Icosahedral Atmospheric Model (NICAM-AMIP) is investigated. Through this analysis, it is found that the simulated MJO tend to be slower than reality and, are biased in the season of their occurrence to late winter. However, similar relationship between the MJO propagation speed and zonal SST difference is reproduced for the simulated boreal-winter MJO. The cause for the slowness and the seasonal bias of the simulated MJO will be discussed from the altered circulation pattern in the NICAM-AMIP simulation due to the warm bias over land surfaces.