Abstract
A new global atmosphere model purpose designed for climate studies is introduced. The model is solved in terms of the normal modes of the linearized primitive equations on a sphere, which allows use of long time steps without introducing computational instability or phase errors of the linear wave components. The model is tested by attempting to simulate the tropical intraseasonal oscillation using an idealized sea surface temperature distribution. Simple treatments of radiation and boundary-layer processes are used together with the much more complete Betts–Miller convection scheme. The Betts–Miller scheme maintains the atmosphere in a state of near neutrality to reversible saturated ascent. It is found that for different values of the surface evaporation time scale, either the evaporation-wind feedback mechanism postulated by Neelin et al. and Emmanuel or low-level convergence of moisture can create eastward propagating deep convective modes. In general, both mechanisms seem important, but it is the latter mechanism that provides phase speeds more in line with observations. Moisture convergence in this model works to erode the low-level equivalent potential temperature inversion that is ubiquitous in nonconvecting regions, thus triggering convection. In contrast to CISK models, changes in boudary-layer equivalent potential temperature are essential in this model to create propagating modes.
The primary deficiency of the model is the tendency of the model to favor horizontal scales of convective disturbances that are much smaller than the zonal wavenumber one or two disturbances observed. This is related to the absence in the model of any pulsation of convection on an intraseasonal time scale over the warmest water regions that has been observed in satellite OLR data. Possible reasons for these differences are discussed.
