Simulation of Tropical Climate with a Linear Primitive Equation Model

Abstract

The tropical climate simulated with a new global atmosphere model is presented. The model is purposely designed for climate studies and is still under development. It is designed to bridge the gap between very efficient but simple models of the tropical atmosphere and sophisticated but inefficient general circulation models (GCMs). In this paper the authors examine the sensitivity of the model's climate to specific formulations of convection, boundary-layer physics, and radiation. The model uses the Betts?Miller convection scheme and a parameterization of the planetary boundary layer (PBL) that combines similarity theory for computation of surface fluxes with a simple scheme for diagnosing PBL depth. Radiative cooling is specified and land surface processes are bypassed by relaxing modeled low-level values to observed quantities. Orography is ignored. The model contains six vertical layers and has a horizontal resolution of about 3° ? 5.625°. The authors compare the climate simulated with two different versions of the Betts?Miller convection scheme. More realistic simulations of rainfall are obtained with the later version, which includes the effects of convective downdrafts. These, by cooling and drying the PBL, act to restrict the areas of convection while strengthening the intertropical convergence zone. The sensitivity to choice of PBL physics is less, and quite similar results were obtained when the PBL scheme was replaced with constant exchange coefficients and PBL depth. In contrast, the amount of precipitation varied strongly with the prescribed radiative cooling. The important role that shallow convection and cloud-radiation interactions play in the spatial organization of deep convection is demonstrated, by default, in an experiment using clear-sky radiative transfer. The modeled climate, as judged qualitatively by its simulation of quantities of importance to air?sea interaction and climate, such as the low-level wind field and precipitation, is in many ways comparable to that achieved by much more complex GCMs. Indeed the rainfall simulation appears better than obtained by many models that use other convection parameterizations. This adds to the accumulating evidence that the Betts?Miller scheme is a quite reliable scheme, at least for simulation of convection in the current climate. A major model flaw is a very poor Asian summer monsoon, which is attributed to lack of orography in the model. It is demonstrated, by inclusion of a specified monsoonal forcing, that this also has an effect, though modest, on the simulation of the trade winds over the Pacific. The results suggest there is hope for development of models of intermediate complexity that achieve a degree of realism exceeding the simple models that have often been used in El Niño studies while retaining much of their efficiency.