How well do coupled global climate models simulate convectively coupled equatorial waves?

Twenty WCRP CMIP3 coupled global climate models are analyzed for the existence and structure of convectively coupled equatorial waves. Forty years of 20th century data are used in the analysis, and are compared with observations based on 10 years of TRMM and NCEP Reanalysis data.

Only 25%, or five, of the 20 analyzed models contain reasonable representations of equatorial waves. These five models all contain one of only two convective parameterizations - Tiedtke/Nordeng (moisture convergence) or Pan and Randall (prognostic cumulus kinetic energy), while the remaining 15 models contain a variety of convective adjustment-based schemes. A model's convective parameterization appears to be the key factor in its ability to simulate convectively coupled equatorial waves. This same ability does not, however, translate to skill in simulating intraseasonal variability.

Details of key convective parameterization schemes will be compared to isolate the factors most important to model equatorial wave generation.