Climate models have difficulty simulating a puzzling feature of the Madden-Julian Oscillation: the transition from shallow-to-deep convection during its initiation. Such transition is often preceded by moistening of the mid-troposphere as shallow cumulus clouds are more likely to grow deep if the surrounding air that would be entrained is already moist. Using observations collected during the 2011 AMIE/DYNAMO field campaign over the Indian Ocean and cloud system resolving model simulation, a team of scientists led by Department of Energy researchers at Pacific Northwest National Laboratory compared the roles of large-scale and convective-scale moistening processes. The shallow-to-deep convection transitions mark the initiation and eastward propagation of the MJO. Examining the large and convective spatial scales, the team tracked 4600 shallow convective cells through their lifetime and the moistening processes that determine whether they are going to transition to deep convection or not. The team found that the frequency of shallow-to-deep convection transitions is sensitive to midlevel moisture and large-scale uplift. The researchers found that the uplift, along with the decline of large-scale drying by equator-ward advection, causes the moisture buildup leading to the initiation of the MJO. Convection scale moisture variability and uplift, and large-scale zonal advection play secondary roles. Of the 4600 convective cells tracked, 2039 transitioned to deep convection and were associated with the initiation or eastward movement of the MJO. Before this study, the initiation of MJO was believed to be related to local convective scale moistening. Integrated use of data from AMIE/DYNAMO campaign and high-resolution modeling shows that large-scale uplift and variations in meridional advection are the primary factors and convective scale processes are secondary.