The experiments suggest that CRI is the most important of the three processes for the development of MJO events in this model. The simulated MJO tends to weaken and accelerate when CRI is disabled. In the control model, the positive feedback from radiative heating is strong (dimensionless ratio between daily radiative and latent heating >0.2) when anomalous precipitation is small (< 14 mm day-1), consistent with a major role for this process in developing incipient disturbances. When WISHE is turned off, the MJO is strengthened, but only when CRI remains turned on. This is because anomalous surface winds have a phase relationship with precipitation that is different from that observed, such that anomalous surface fluxes negatively feed back to the MJO-related convection anomalies when surface latent heat flux interacts with surface wind in this model. When FWC is switched off, the simulated MJO shows no systematic change compared to the control, although the phase difference between convection and moisture convergence in the boundary layer is reduced as expected.
We propose that it would be informative if similar mechanism-denial experiments were conducted using other models.