Roles of deep and shallow convection and microphysics in the MJO simulated by the Model for Prediction Across Scales

Abstract

The November event of the Madden-Julian oscillation (MJO) during the Dynamics of North Atlantic Models (DYNAMO) field campaign was simulated using the global compressible nonhydrostatic Model for Prediction Across Scales with global coarse (60 and 15 km) and regional (the Indian Ocean) cloud-permitting (3 km) meshes. The purpose of this study is to compare roles of parameterized deep and shallow cumulus and microphysics in MJO simulations. Two cumulus schemes were used: Tiedtke and Grell-Freitas. The deep and shallow components of Tiedtke scheme can be turned on and off individually. The results reveal that microphysics alone (without cumulus parameterization) is able to produce strong signals of the MJO in precipitation with 3 km mesh and weak MJO signals with 15 km mesh. A shallow scheme (Tiedtke) along with microphysics strengthens the MJO signals but makes them less well organized on large scales. A deep cumulus scheme can either improve the large-scale organization of MJO precipitation produced by microphysics and shallow convection (Tiedtke) or impair them (Grell-Freitas). The deep scheme of Tiedtke cannot reproduce the MJO well without its shallow counterpart. The main role of shallow convection in the model is to transport moisture upward to the lower to middle troposphere. By doing so, it removes dry biases in the lower to middle troposphere, a distinct feature in simulations with weak or no MJO signals, and enhances total precipitation and diabatic heating produced by microphysics and deep cumulus schemes. Changing model grid spacing from 60 to 15 km makes a little difference in the model fidelity of reproducing the MJO. All roles of shallow convection in 15 km simulations with parameterized deep convection cannot be reproduced in 3 km simulations without parameterized deep convection. Results from this study suggest that we should pay more attention to the treatment of shallow convection and its connection to other parameterized processes for improving MJO simulations. In other words, a holistic approach should be taken that consider parameterization of shallow cumulus, microphysics, boundary layer, and deep cumulus as a whole for model improvement.Many studies of modeling comparisons have attempted to identity causes of model infidelity in simulating the MJO [Lin et al., 2006;Zhang et al., 2006;Kim et al., 2009]. None of them has led to any definitive answer. Recent studies diagnosed reproduction of the MJO in three types of global model runs: 20 year climate simulations [Jiang et al., 2015], 20 day hindcast [Klingaman et al., 2015a], and 2 day hindcast [Xavier et al., 2015]. Their finding is intriguing: There is no one-to-one correspondence between model success or failure in producing MJO statistics in 20 year runs and MJO hindcast of 20 and 2 days, there is no clear indication of what biases in the mean state might be responsible for model failure of reproducing the MJO, and there is PILON ET AL. DEEP AND SHALLOW CONVECTION IN THE MJO 10,575

Upscale Impact of Mesoscale Convective Systems and Its Parameterization in an Idealized GCM for an MJO Analog above the Equator

“…In particular, we focus on the cases with upscale impact of MCSs propagating upshear at a slow speed, modulated by the effects of both deep heating excess and vertical shear strength. It is worthwhile mentioning that the upscale impact of MCSs is missing from the current cumulus parameterizations implemented in GCMs (Moncrieff et al 2017), which is another important GCM deficiency besides the reduced congestus and stratiform convection (Pilon et al 2016;Cao and Zhang 2017). By adding this proposed parameterization to the deficient GCM in Fig.…”

Atmospheric Overturning across Multiple Scales of an MJO Event during the CINDY/DYNAMO Campaign

“…Wang et al 2015, hereinafter W15;W. Wang et al 2015;Xiang et al 2015;Hagos et al 2016;Janiga and Zhang 2016;Pilon et al 2016;Powell 2016). W15 simulated the October and November MJO events using a convection-permitting regional model with 9-km grid spacing.…”

Hindcasting the Madden‐Julian Oscillation With a New Parameterization of Surface Heat Fluxes

“…The simulation domain covers the tropical Indian Ocean and western Maritime Continent, from 208S to 208N and 488E to 1208E (Figure 2), similar to that in Wang et al (2015a). All simulations are run at 5 km horizontal grids without cumulus parameterization as cumulus parameterization could add more uncertainties to MJO Journal of Advances in Modeling Earth Systems 10.1002/2017MS001175 simulations (Hagos & Leung, 2011;Pilon et al, 2016;Zhang, 2005). The simulation has 42 vertical levels with 12 levels in the lowest 1 km and a nominal top at 20 hPa.…”