Does the MJO Have a Westward Group Velocity?

Abstract

Current theoretical studies have a debate on whether the Madden–Julian oscillation (MJO) has a zero or westward group velocity. A recent analysis of the observed Hovmöller diagram of MJO signals suggested that the MJO has a significant westward group velocity. Here it is shown that the observed MJO has a negligibly small group velocity, which is manifested in two aspects. First, on the wavenumber–frequency spectra diagram the precipitation spectra indicate quasi independence of the MJO frequency on wavenumber, suggesting a nearly vanishing group velocity. Second, on the Hovmöller diagram of the regressed intraseasonal daily precipitation, the MJO group velocity is defined by the propagation of the wave envelopes of the precipitation and is shown to be negligibly small for the eastward propagating signals. The causes of the discrepancy between this study and the recent study mentioned above are the calculating method and the data filtering process. The group velocity in the recent study is calculated by the propagation of local convection extrema, which does not necessarily indicate the propagation of the wave envelopes. More importantly, the westward propagation of the local convection extrema is an artifact of the data filtering. The Hovmöller diagram in the recent study was constructed by using only the eastward propagating wavenumber-1–5 signals. This truncation of data onto the planetary scales of the eastward wavenumber domain fails to resolve the Maritime Continent “barrier effect,” causing significant artificial westward propagation of local convection extrema.