1.3
The spectrum of middle atmospheric motions predicted by three-dimensional general circulation models
John N. Koshyk, Univ. of Toronto, Toronto, ON, Canada; and K. Hamilton
Spatial kinetic energy spectra simulated by several different middle-atmosphere general circulation models are examined. The spectra are computed as a function of the total horizontal wavenumber, n. The horizontal and vertical resolution among the models varies but all have upper boundaries corresponding to heights at or above the mesospause.
Tropospheric kinetic energy spectra show the familiar ~ -3 power law with wavenumber for horizontal length scales between about 5000 and 500 km and are dominated by the rotational part of the flow. A high-resolution version of the GFDL SKYHI model, the only model in the study that resolves spatial scales smaller than 500 km, also successfully simulates the observed -5/3 mesoscale spectral regime. In the mesoscale, the model results show that the rotational and divergent parts of the flow provide approximately equal contributions to the total kinetic energy spectrum.
The kinetic energy spectra in the middle atmosphere are significantly different from tropospheric spectra. The range of wavenumbers for which rotational and divergent parts of the flow are comparable increases with height, owing to the rapid growth of the divergent component in the vertical. The divergent component is understood to be associated with naturally resolved gravity waves in the models. In the mesosphere, rotational and divergent components are comparable for all but planetary and sub-planetary scales.
The kinetic energy budget of an extremely high resolution version of the GFDL SKYHI general circulation model is analyzed in detail. It is shown that vertical gravity-wave fluxes play a dominant role in the kinetic energy budget of the middle atmosphere. Parameterized vertical and horizontal diffusion processes are also considered and the formulation of horizontal subgrid-scale processes in middle atmosphere general circulation models is critically re-evaluated.
Session 1, Modeling and data assimilation
Monday, 10 January 2000, 8:55 AM-12:00 PM
Previous paper Next paper