Use of the UIUC 11-layer atmospheric general circulation model to simulate and understand the tropical intraseasonal oscillation

The AMIP simulation from 1979 to 1988 by the UIUC 7-layer tropospheric general circulation model (GCM), using observed sea surface temperatures, did not reproduce the well-organized, intraseasonal oscillations that are observed in the tropics. Because the simulations of the oscillations by many other GCMs also are deficient, we began an investigation of the possible cause and correction. For this investigation we raised the top of the 7-1ayer GCM from 200 mb to 50 mb and added 4 layers. We then focused our investigation with the resulting 11-layer GCM on the parameterization of cumulus convection. Three different convection parameterizations have been tested: (1) the UIUC GCM's original cumulus-convection parameterization, which includes a modified version of the Arakawa-Schubert (1974) penetrative-convection parameterization (AS) and a middle-level convection (MLC) parameterization; (2) the parameterization of Kuo (1974); and (3) the moist convective adjustment (MCA) parameterization of Manabe et al. (1965). For each parameterization, a relative-humidity criterion $\rm (RH\sb{c})$ for the convection to occur is used, as in many GCMs. Perpetual-March simulations with these convection parameterizations have been performed for different values of $\rm RH\sb{c}.$ It is found that the value of $\rm RH\sb{c}$ is very important in the simulation of the tropical intraseasonal oscillation (TIO) for all three parameterizations. Composite horizontal and vertical structures of the oscillations show that all of our simulations generally capture the basic features of the observed TIO, but the amplitude of the TIO increases as $\rm RH\sb{c}$ is increased. The period of the oscillation varies with the convection parameterization and $\rm RH\sb{c},$ but no clear dependence can be defined. The dependence of the amplitude of the simulated oscillation on $\rm RH\sb{c}$ is consistent with the results from previous simulations by different general circulation models.The diagnosis of our GCM simulations and our investigation using a one-dimensional model that is identical to the 11-layer UIUC GCM, except that all dynamical processes are ignored, have been carried out to explore explanations for the results based on existing theories. It is found that, among the mechanisms proposed in these theories, the nonlinear dependence of the condensational heating on the large-scale flow, and the phase lag between the heating and the moisture convergence, are responsible for the dependence of the simulated oscillation on $\rm RH\sb{c}.$ This finding is consistent with that of Itoh (1989) who showed that the dominant wavenumber of the equatorial eastward-propagating waves decreased as the magnitude of the critical convergence for convective heating was increased, and is also consistent with the results of cho et al. (1994) and Goswami and Rao (1994) who showed that a phase lag between the condensational heating and the low-level moisture convergence leads to the selection of the longest equatorial eastward-propagating waves. We have also found that the time-mean precipitation is confined to the maritime continents and western Pacific increasingly as $\rm RH\sb{c}$ is increased. This is consistent with the results from the study by Slingo et al. (1995) who found that GCMs which simulate the tropical intraseasonal oscillation with large amplitude also simulate large time-mean precipitation rates over the Indonesian continents and western Pacific. (Abstract shortened by UMI.)U of I OnlyETDs are only available to UIUC Users without author permissio