How does the cumulus parameterization scheme influence the simulation of MJO propagation and structure?

In this study, eleven different cumulus parameterization schemes (CPSs) in Weather Research and Forecasting (WRF) model version 4.0 were used to test the sensitivity of Madden–Julian oscillation (MJO) eastward propagation to different CPSs and to explore the influences of different CPSs on MJO eastward propagation. The propagation and structures of the MJO-filtered outgoing longwave radiation (OLR) for the MJO active phase composites during the December–January–February (DJF) period of 2007–2016 over the Indian Ocean (IO) were diagnosed in this study. The results show that the simulated MJO propagation is sensitive to the selection of CPSs. And CPS5 (Tiedtke scheme) and CPS8 (New Tiedtke scheme) are selected for the good models composite while the CPS4 (Grell 3D ensemble scheme) and CPS9 (Grell-Devenyi ensemble scheme) are selected for the poor models composite. The good models composite provides similar MJO structures and MJO eastward propagation patterns as those observed, while poor models composite fails. The MJO propagation is highly correlated with boundary layer moisture convergence (BLMC) propagation pattern correlation coefficient (PCC), the U850 asymmetry index, the convective instability index and the new index reconstructed by the three main contributing factors, and BLMC propagation plays the most important role in simulating MJO propagation. Besides, different CPSs may result in different interactions between convective heating and moisture feedback, thus simulating different structures. The results support the trio‑interaction theory and provide a reference for the influence of CPS on the MJO eastward propagation.