The Role of Convection and Radiation in the Equatorial Kelvin Wave

We aim to investigate the zonally asymmetric features of the Madden Julian Oscillation (MJO). First, we perform a composite analysis of circulation, moisture, convection (OLR and rainfall) of the MJO over Central Indian Ocean. The analysis shows the equatorial zonal overturning circulation of the MJO consisting of the first baroclinic mode with a broad scale (60^oE-170^oE), and the second baroclinic mode with a 15 longitudes westward shift relative to the first baroclinic mode. The corresponding composite cloud types and radiative heating rates of CloudSat data show zonally asymmetric distribution from shallow cloud and LW cooling anomaly in the suppressed descending branch of the first baroclinic circulation, through deep cloud and LW warming in the MJO convection core, to mid-level stratiform cloud with distinct vertical heating-cooling contrast in the western branch of circulation.

We then perform four aqua-planet experiments in a global GCM with prescribed sea surface temperature varying only in latitude similar to the current climate. The experiment with default full physics is control (CNTL). The other two differ from CNTL, one with No Shallow Cumulus parameterization (NSC) and one with No Cloud Radiative Feedback in LW (NCRF). Eastward-moving disturbances are prevalent in all experiments with convection organized as super cloud clusters (SCCs) that are identified as convectively coupled equatorial Kelvin waves by their space-time spectra. The structure of the simulated SCCs is similar to that of the observed MJO, but its zonal scale is smaller (70-80^o), and the phase speed varies from 8-15 ms^-1 in CNTL (Control experiment), and 12-18 ms^-1 in NSC and NCRF. In CNTL, relative to the fast-moving SCCs (phase speed 12-15 ms^-1), the slower-moving SCCs (8-12 ms^-1) are more organized with broader shallow convective mixing featuring stronger shallow convective heating and moistening. The effect of shallow convection is further examined by comparing NSC with CNTL. Without convection mixing by shallow cumulus, moisture is trapped in the planetary boundary layer (PBL). The weaker moisture and stronger entrainment in lower troposphere together result in weaker shallow convective heating and slower phase speed of the SCC. On the other hands, the differences between CNTL and NCRF indicate that LW warming in the deep and mid-level stratiform cloud region enhances moisture and buoyancy in the upper troposphere that leads to organized deep overturning circulation, inactive shallow convection and faster SCC phase speed. But the cloud-radiative forcing does not affect the zonal scale of SCC. In an additional warm pool experiment, slow-moving disturbances similar to the MJO emerge.