Preliminary simulations have been performed using the MM5 v.3 regional model aimed at testing initiation and growth hypotheses of the Madden-Julian Oscillation (MJO). These model runs start 1 June 1990 and last for two years with a tropical domain extending from the western edge of the Indian Ocean to the Dateline. Model long term means of wind, outgoing longwave radiation, and moist static energy show that MM5 reproduces well most aspects of the tropical climate. The primary difference is that MM5 produces more active convection than the NCEP/NCAR Reanalysis. This leads to a stronger thermally direct circulation and slightly higher wind speeds.

To isolate the MJO signal the MM5 output, reanalysis data, and NCEP OLR data are filtered using a 30-70 day band pass Lanczos filter and then compared. The results show that the model captures the dominant features of the MJO. Convection begins in the western Indian Ocean and develops into large convective clusters which propagate eastward. The convection weakens over the Maritime continent and strengthens again over the West Pacific. Also, the model clearly reproduces the observed strong first baroclinic structure of the MJO. To establish the importance of signals propagating into the model domain for the development of the MJO, another model run has been performed such that the model's boundary conditions are filtered with a 30-70 day band pass filter to remove the MJO signal in the boundary forcings. The resulting model output shows little to no MJO signal. Subsequent runs test various hypotheses for growth and initiation of the MJO.