Tropical Precipitation Variability and Convectively Coupled Equatorial Waves in Reanalyses and TRMM Observations

Tropical convection often arises from equatorial waves. By releasing latent heat, tropical moist convection also generates a broad spectrum of waves, having impacts on the middle atmospheric climate as well as on the tropospheric weather variability. Due to the complex interactions of convection with dynamical, thermodynamical, and cloud microphysical processes, the tropical precipitation system is not fully understood. Also, the relationships between moist convection and waves are not well captured in current weather prediction and climate models. One deficiency of convectively coupled equatorial waves (CCEWs) can entail the lack of precipitation variability in simulations. In this study, we evaluate 3-year (2005-2007) tropical precipitation in reanalyses including ERA-interim, MERRA, NCEP-NCAR, and NCEP-DOE in terms of mean, variability, diurnal cycle, and CCEWs. The TRMM 3B42 3-hourly product was used as a validation dataset. To study variability associated with CCEW activities, we performed a space-time spectral analysis on precipitation at 15S-15N. Besides the common positive bias of mean tropical precipitation in four reanlayses, a spectral analysis revealed that the reanalyses have over-reddened spectra with persistent rainfall. The prominent spectral peaks in the precipitation spectrum correspond to the dispersion curves of equatorial waves, solutions of the shallow water equations on the equatorial beta plane. The low frequency scale CCEWs such as Kelvin, Mixed-Rossby gravity, and Rossby waves are properly reproduced in ERA-interim and MERRA. In all the reanalyses, however, the CCEW signals do not exist in a high frequency region with a period shorter than 1 day. We also examine the influence of different wave modes on regional precipitation, by applying spectral filters for the inverse Fourier transform. The time frame of 36 days with 6-day overlap for each spectral analysis allows us to investigate the monthly evolution of precipitation variability related to equatorial wave modes. It appears that the latest reanalyses such as ERA-interim, MERRA, and NCEP-DOE generally better represent CCEWs and variability over NCEP-NCAR. However, all the reanalyses still need to be improved in resolving CCEWs and variability especially on the high frequency scale.