Since its discovery by Madden and Julian (1971), a comprehensive theory of the ~30-60-day Madden-Julian oscillation (MJO) has proven elusive. The MJO is characterized by an envelope of eastward propagating deep convective organization (cloud clusters) which appears to originate over the Indian Ocean and dissipates near the dateline. The perturbations in large-scale circulations, nevertheless, continue to propagate globally in the upper troposphere (Yanai et al., 2000). Numerical simulations of MJO with models with various degrees of complexity have difficulty in producing the right phase speed and periodicity. Recently, Grabowski (2003) performed an idealized simulation by implementing a two-dimensional cloud resolving model into a global model, suggesting that MJO-like structures with a reasonable propagation speed may be generated. As similar work progresses along the same lines, there is a practical need for statistical depictions of MJO from observations in order to validate the model results.

In this work, we examine climatological spatial and temporal characteristics of the Madden-Julian oscillation. Statistical features of the thermodynamic and dynamic fields as well as the coupling between convection and large-scale disturbances associated with MJO are obtained through a lagged regression method. The pronounced irregularity of MJO occurrence and periodicity is investigated by using a wavelet transform.