Abstract
This study investigates the major factors for the realistic simulation of convectively coupled Kelvin waves (CCKWs) using the National Centers for Environmental Prediction (NCEP) Climate Forecast System (CFS) models. CFS simulations employing relaxed Arakawa–Schubert (RAS; hereafter CTRL) and simplified Arakawa–Schubert (SAS) cumulus parameterization schemes show that the former generates the observed Kelvin wave signature more realistically than the latter does. For example, the space–time spectral signal, eastward propagation, and tilted (and second baroclinic mode) vertical structures in convection, temperature, moisture, and circulation anomalies associated with CCKWs in CTRL are more comparable to observations than in the SAS simulation. CTRL and observations demonstrate the characteristic evolution and vertical heating profile associated with CCKWs similar to those seen in mesoscale convective systems in the tropics: shallow convection, followed by deep convection and then stratiform cloudiness, and resulting in a top-heavy diabatic heating profile. Five additional experiments demonstrate that the effects of convective downdrafts, subgrid-scale convective rain evaporation, and large-scale rain evaporation on CCKWs are assessed to be insignificant in CTRL, possibly due to a more humid environment than observation. However, the Kelvin wave signals are reduced by ~40% when shallow convection is disabled. More importantly, the removal of convective detrainment at the cloud top results in the greatest reduction in Kelvin wave activity (by more than 70%). Therefore, the preconditioning of the atmosphere by shallow convection and detrainment of water vapor and condensate from convective updrafts to the environment and subsequent stratiform heating (grid-scale condensational heating)/precipitation processes are the two most crucial factors for the successful simulation of CCKWs.
