Intercomparison of two model climates simulated by a unified weather-climate model system (GRIST), part II: Madden–Julian oscillation

Realistically simulating the Madden–Julian oscillation (MJO) while maintaining a balanced global mean state is a great challenge for global weather and climate models. This study assesses the MJO simulations performed by a unified weather-climate model system (GRIST) under the Atmospheric Model Intercomparison Project (AMIP)-style experimental protocol. The AMIP-style simulations employ two distinct physics suites. One suite (PhysC) generates stronger MJO signals, while the other (PhysW) does not. Fine-tuning the cumulus parameterization scheme in PhysW by enhancing its entrainment/detrainment rates significantly improves the MJO simulations, yielding key characteristics—such as intensity, spectrum, and eastward propagation—more aligned with those produced by PhysC. These simulations underscore the importance of linking the entrainment rate parameterization to environmental relative humidity and maintaining a proper relationship between precipitation and column relative humidity for accurately simulating the MJO. The relationships between several key physical processes and enhanced MJO signals are revealed, including: (i) a higher proportion of large-scale rainfall; (ii) a higher sensitivity of precipitation to moisture, which lowers the convective adjustment time scale; (iii) the ability to reproduce observed wave dynamics and vertical tilted structures of MJO-related fields, and (iv) a zonal asymmetry in the moist static energy (MSE) tendency, promoting the eastward propagation of MJO. In particular, the precipitation-moisture relationship and the contributions of individual terms and interactions across different time scales to the asymmetric MSE tendency are analyzed in detail.