Role of cloud microphysics in improved simulation of the Asian monsoon quasi-biweekly mode (QBM)

A major sub-seasonal variability of the tropics and sub-tropics, the quasi-biweekly mode (QBM), is known to have significant influence on the seasonal mean of the south Asian monsoon rainfall. A coupled Atmosphere–Ocean General Circulation Model (AOGCM) being essential for seasonal prediction, the ability of the AOGCMs in simulating the space–time characteristics with fidelity is critical for successful seasonal prediction of the south Asian monsoon in particular and seasonal prediction in the tropics in general. However, strength and weaknesses in simulating the QBM by AOGCMs have remained poorly investigated so far. Here, we examine the simulation of the QBM in AOGCM and show that improvement of parameterizations of both convection and microphysics is required to improve the simulation of the QBM. While the standard version of the model overestimates the variance of QBM and simulates a smaller scale Rossby waves (n = 1), the modified version of the model where the simple Arakawa–Schubert (SAS) convection parameterization is combined with a new improved microphysics parameterization (MCMv.1) proposed by us, simulates a more realistic space–time characteristics of the QBM. In yet another version of the model, we combine the new SAS with the new improved microphysics parameterization. Interestingly, this version of the model also simulates the space–time structure poorly with poor westward propagation and fragmented organization, but it simulates a reasonable variance. These results indicate that a synergy among the convective parameterization and microphysics parameterizations is critical in simulating the QBM in particular and equatorial waves in general. We show that most the biases in simulating the QBM may be related to the biases of the model in simulating the stratiform fraction of precipitation. While the simulation of the space–time characteristics of QBM is better simulated in the MCMv.1, the convective coupling is still too strong as compared to observations, an area for future improvement of the model.