This paper presents a composite climatology of the intraseasonal variability over the Indian Ocean and the west Tropical Pacific using OLR, TRMM, and NCEP-NCAR and ECMWF reanalysis for both summer and winter. The composites are constructed based on filtered data between 25 and 80 days using wavelets. The horizontal and vertical structure of the variability is examined. A description of the state of the atmosphere previous the triggering of an MJO is presented, since it is believed to be a defining factor of the characteristics of the strong convection.

Based on the composites, some of the stronger MJO cases for summer and winter are selected, and the potential predictability of those cases is assessed using an empirical scheme, which uses a wavelet banding technique and linear regression, and a numerical model run in coupled mode. The predictability of the start, growth and propagation of the convection associated to an MJO is analyzed separately. To account for the spatial structure, the empirical technique is used in two ways: One way is to forecast OLR variability in different regions along the Equator, and the other is to forecast the state of the principal oscillation patterns. In general, the empirical technique performs better than the numerical model, and while the quality of the results is satisfactory in all the mentioned stages, the forecasting of the propagation of convection particularly good.