Abstract Accurate simulations of convectively coupled equatorial waves (CCEWs) are key to properly forecasting rainfall and weather patterns within (and outside) the tropics. Many studies have shown that global numerical weather prediction (NWP) models usually do not accurately simulate CCEWs; however, it is unclear if this problem can be alleviated with a better representation of deep convection in the models. To this end, this study investigates the representation of multiple types of CCEWs in the Model for Prediction Across Scales-Atmosphere (MPAS-A). The simulated structure of CCEWs is analyzed from three MPAS-A aquaplanet experiments with horizontal cell spacing of 30, 15, and 3 km, respectively. Using a wave-phase composite technique, the simulated structure is compared against observed CCEWs as represented by satellite and reanalysis data. All aquaplanet experiments capture the overall structure of gravity wave–type equatorial waves (e.g., Kelvin waves and inertio-gravity waves). Those waves are more realistic in the 3-km experiment, particularly in terms of the vertical structure of temperature, water vapor, and wind anomalies associated with the waves. The main reason for this improvement is a more realistic diabatic heating profile; the experiment with resolved convection produces stronger heating (or weaker cooling) below the melting level during the convectively active phase of Kelvin and inertio-gravity waves. Intriguingly, the rainfall and lower-tropospheric structure associated with easterly waves show pronounced discrepancies between the aquaplanet experiments and reanalysis. Resolved deep convection primarily affects the intensity and propagation speeds of these waves.
Read full abstract