Tropical Cumulus Convection and Upward-Propagating Waves in Middle-Atmospheric GCMs

Abstract

It is recognized that the resolved tropical wave spectrum can vary considerably among general circulation models (GCMs) and that these differences can have an important impact on the simulated climate. A comprehensive comparison of low-latitude waves is presented for the December–January–February period using high-frequency data from nine GCMs participating in the GCM Reality Intercomparison Project for Stratospheric Processes and Their Role in Climate (GRIPS; SPARC). Quantitative measures of the wavenumber-frequency structure of resolved waves and their impacts on the zonal mean circulation are given. Space–time spectral analysis reveals that the wave spectrum throughout the middle atmosphere is linked to the variability of convective precipitation, which is determined by the parameterized convection. The variability of the precipitation spectrum differs by more than an order of magnitude among the models, with additional changes in the spectral distribution (especially the frequency). These differences can be explained primarily by the choice of different cumulus parameterizations: quasi-equilibrium mass-flux schemes tend to produce small variability, while the moist-convective adjustment scheme is the most active. Comparison with observational estimates of precipitation variability suggests that the model values are scattered around the observational estimates. Among the models, only those that produce the largest precipitation variability can reproduce the equatorial quasi-biennial oscillation (QBO). This implies that in the real atmosphere, the forcing from the waves, which are resolvable with the typical resolutions of present-day GCMs, is insufficient to drive the QBO. Parameterized cumulus convection also impacts the nonmigrating tides in the equatorial region. In most of the models, momentum transport by diurnal nonmigrating tides in the mesosphere is comparable to or larger than that by planetary-scale Kelvin waves, being more significant than has been thought. It is shown that the westerly accelerations in the equatorial semi-annual oscillation in the models examined are driven mainly by gravity waves with periods shorter than 3 days, with some contribution from parameterized gravity waves, and that the contribution from the wavenumber-1 Kelvin waves is negligible. These results provide a state-of-the-art assessment of the links between convective parameterizations and middle-atmospheric waves in present-day middle-atmosphere climate models.