Sensitivity of MJO simulations to diabatic heating profiles

The difficulty for global atmospheric models to reproduce the Madden-Julian oscillation (MJO) is a long-lasting problem. In an attempt to understand this difficulty, simple numerical experiments are conducted using a global climate model. This model, in its full paramterization package (control run), is capable of producing the gross features of the MJO, namely, its planetary-scale, intraseasonal, eastward slow propagation. When latent heating profiles in the model are artificially modified, the characteristics of the simulated MJO changed drastically. Intraseasonal perturbations are dominated by stationary component over the Indian and western Pacific Oceans when heating profiles are top heavy (maximum in the upper troposphere). In contrast, when diabatic heating is bottom heavy (maximum in the lower troposphere), planetary-scale, intraseasonal, eastward propagating perturbations are reproduced with a phase speed similar to that of the MJO. The difference appears to come from surface and low-level moisture convergence, which is much stronger and more coherent in space when the heating profile is bottom heavy than when it is top heavy. These sensitivity experiments, along with other theoretical, numerical, and observational results, have led to a hypothesis that the difficulty for global models to produce the MJO partially is rooted in a lack of sufficient diabatic heating in the lower troposphere, presumably from shallow convection.