a b s t r a c tDespite recent advances in supercomputing, current general circulation models (GCMs) have significant problems in representing the variability associated with organized tropical convection. Furthermore, due to high sensitivity of the simulations to the cloud radiation feedback, the tropical convection remains a major source of uncertainty in long-term weather and climate forecasts. In a series of recent studies, it has been shown, in paradigm two-baroclinic-mode systems and in aquaplanet GCMs, that a stochastic multicloud convective parameterization based on three cloud types (congestus, deep and stratiform) can be used to improve the variability and the dynamical structure of tropical convection, including intermittent coherent structures such as synoptic and mesoscale convective systems. Here, the stochastic multicloud model is modified with a parameterized cloud radiation feedback mechanism and atmosphere-ocean coupling. The radiative convective feedback mechanism is shown to increase the mean and variability of the Walker circulation. The corresponding intensification of the circulation is associated with propagating synoptic scale systems originating inside of the enhanced sea surface temperature area. In column simulations, the atmosphere ocean coupling introduces pronounced low frequency convective features on the time scale associated with the depth of the mixed ocean layer. However, in the presence of the gravity wave mixing of spatially extended simulations, these features are not as prominent. This highlights the deficiency of the column model approach at predicting the behavior of multiscale spatially extended systems. Overall, the study develops a systematic framework for incorporating parameterized radiative cloud feedback and ocean coupling which may be used to improve representation of intraseasonal and seasonal variability in GCMs.