An idealized study of African easterly waves. II: A nonlinear view

Abstract

AbstractThe nonlinear behaviour of an easterly‐wave normal mode has been examined. Significant differences from normal‐mode character become apparent after saturation of the instability. The easterly wave is initially dominated by barotropic energy conversions (CK) but later grows mainly through baroclinic energy conversions (CE). The eddy kinetic energy of the wave peaks and then decays with both CK and CE contributing to this decay. The energy conversions associated with the nonlinear growth of the wave have more in common with those found in the GATE study over west Africa and GCM integrations than do the previous linear studies of African easterly waves.The life cycle is described using EP flux diagnostics and Rossby wave propagation ideas. After the linear growth period, characterized by divergent, mostly horizontal, EP flux vectors and barotropic growth, the life cycle becomes dominated mostly by vertical EP fluxes associated with vertical wave propagation from the jet region both above and below the wave, and by baroclinic growth. This results in increased wave amplitudes in the upper troposphere and at low levels. This mechanism for increased amplitudes at upper levels may help to explain the observed winds in the waves at these levels—a feature not found in linear studies.The effect of including simple parametrizations of different diabatic processes is also examined. The main effect of including boundary‐layer fluxes is merely to damp the large surface amplitudes which occur in the nonlinear part of the life cycle. Including a restoration of the basic‐state results in stronger growth and larger amplitudes at all levels.Two life cycles with different latent heating parametrizations are also examined. One has a CISK scheme driven by low‐level ascent, and the other a scheme based on the conservation of a prescribed value of the equivalent potential temperature. In each case CE becomes comparable to CK but does not dominate as in the basic life cycle. This is due to significant changes in the synoptic structure of the wave when latent heating is included: in particular the vertical velocity field changes dramatically. Indeed, with the CISK scheme it compares well with GATE observations, with a deep ascent region present throughout the troposphere ahead of the main cyclonic region. The maximum is also larger than in the dry case, although still somewhat less than observed.Two life cycles were also examined with both CISK and boundary‐layer fluxes included; the second also including zonal flow restoration. In the case of both life cycles CE dominates over CK in the nonlinear part of the life cycle, unlike the case with CISK alone. The structures found are similar to those with CISK alone except that the surface amplitude is considerably weakened by the boundary layer. However, the magnitude of the ascent is increased with the boundary layer and increased further with the restoration, making it comparable with observed values.