Abstract

The adequate representation of the dominant intra-seasonal and synoptic scale variability in the tropics, characterized by the Madden-Julian oscillation (MJO) and convectively coupled waves is still problematic in current-operational general circulation models (GCMs). Here we present results usin... Show moreThe adequate representation of the dominant intra-seasonal and synoptic scale variability in the tropics, characterized by the Madden-Julian oscillation (MJO) and convectively coupled waves is still problematic in current-operational general circulation models (GCMs). Here we present results using the next generation NCAR GCM HOMME, as a dry dynamical core at coarse resolution of about 167 km, coupled to a simple multicloud parametrization. The coupling is performed through a judicious choice of heating vertical profiles for the three cloud types, congestus, deep, and stratiform, that characterize organized tropical convection. Important control parameters that affect the types of waves that emerge are the background vertical gradient of the moisture and the stratiform fraction in the multicloud parametrization that set the strength of large-scale moisture convergence and unsaturated downdrafts in the wake of deep convection, respectively. Three numerical simulations using different moisture gradients and different stratiform fractions are considered. The first experiment uses a large moisture gradient and a small stratiform fraction and provides an MJO-like example. It results in an intra-seasonal oscillation of zonal wavenumber two, moving eastward at a constant speed of roughly 5 m s⁝š. The second uses a weaker background moisture gradient and a large stratiform fraction and yields convectively coupled Rossby, Kelvin and two-day waves, embedded in and interacting with each other while the last one combines the small stratiform fraction and the weak background moisture gradient to yield a planetary scale (wavenumber one) second baroclinic Kelvin wave. While the two first experiments provide two benchmark examples that reproduce several key features of the observational record, the third one is more of a demonstration of a bad MJO model solution that exhibits very unrealistic features. Show less