Tropical Intraseasonal Variability in Version 3 of the GFDL Atmosphere Model 4 5 6

24 25 Tropical intraseasonal variability is examined in version 3 of the Geophysical Fluid Dynamics 26 Laboratory Atmosphere Model (AM3). Compared to its predecessor AM2, AM3 uses a new treatment 27 of deep and shallow cumulus convection and mesoscale clouds. The AM3 cumulus parameterization is 28 a mass flux-based scheme but also, unlike that in AM2, incorporates subgrid-scale vertical velocities; 29 these play a key role in cumulus microphysical processes. The AM3 convection scheme allows multi-30 phase water substance produced in deep cumuli to be transported directly into mesoscale clouds, which 31 strongly influence large-scale moisture and radiation fields. 32 We examine four AM3 simulations, using a control model and three versions with different mod-33 ifications to the deep convection scheme. In the control AM3, using a convective closure based on 34 CAPE relaxation, both the MJO and Kelvin waves are weak compared to those in observations. By 35 modifying the convective closure and trigger assumptions to inhibit deep cumuli, AM3 produces reason-36 able intraseasonal variability but a degraded mean state. MJO-like disturbances in the modified AM3 37 propagate eastward at roughly the observed speed in the Indian Ocean but up to twice the observed 38 speed in the West Pacific. Distinct differences in intraseasonal convective organization and propagation 39 exist among the modified AM3 versions. Differences in vertical diabatic heating profiles associated 40 with the MJO are also found. The two AM3 versions with the strongest intraseasonal signals have a 41 more prominent " bottom-heavy " heating profile leading the disturbance center and " top-heavy " heating 42 profile following the disturbance. The more realistic heating structures are associated with an improved 43 depiction of moisture convergence and intraseasonal convective organization in AM3. 45 In the tropical atmosphere, variability on 20-100 day time scales (hereafter, " intraseasonal ") is 46 dominated by the Madden-Julian Oscillation (MJO; Madden and Julian 1971). Kelvin and equatorial 47 Rossby waves modify tropical precipitation and wind distributions on shorter timescales. In the Indo-48 Pacific region, these disturbances typically involve a coupling between moist convection and the large-49 scale flow field. General circulation models (GCMs) use parameterizations to represent the bulk effects 50 of these convective clouds on grid-scale heat, moisture, and momentum budgets. GCM simulations are 51 often strongly sensitive to the choice of convective parameterization and parameter variations within a 52 given parameterization scheme (e.g., Maloney and Hartmann 2001). The challenges of accurately …