Zonal flow vacillation with very long time scales is observed in a 3070-day simple GCM simulation with zonally symmetric forcing. The long lasting zonal wind anomalies suggest that zonal flow vacillation is self-maintained. Wave-mean flow interactions are investigated by composite analysis and transform Eulerian momentum budget analysis. Nonlinear life-cycle simulations are conducted to demonstrate that each extreme phase of the zonal flow vacillation is a quasi stable state and is self-maintained by the embedded synoptic eddies.

The firm EOF mode of zonal-mean wind shows an out of phase relation between anomalies at 60°S and at 40°S with a barotropic structure. This structure is similar to the dominant vacillation pattern observed in the Southern Hemisphere. The composite jet stream in the high (low) index phase of zonal flow vacillation shifts poleward (equatorward) from the time-mean location and becomes broader (narrower) and weaker (stronger). Composite eddies in the high index Phase tilt NW-SE and show mostly equatorward propagation, while eddies in the low index phase have “banana” shapes and propagate both equatorward and poleward. Transformed Eulerian momentum budget analyses show that the differences of wave propagation between two extreme phases result in the anomalous eddy forcing needed to maintain zonal wind anomalies against frictional damping.

Budget analyses also indicate that eddy momentum flux convergence is the major positive forcing in both the extreme and transition phases. Eddy baroclinic forcing exerts weak damping on the wind anomalies in the upper troposphere but acts together with residual circulation forcing to counteract frictional damping near the surface. The major balance during the index cycle is between eddy barotropic forcing and residual circulation forcing in the upper troposphere and between residual circulation forcing and frictional damping in the lower troposphere. Further comparisons of eddy forcing from various time-scale eddies show that the anomalous eddy forcing is primarily provided by synoptic time scales. Two nonlinear life-cycle simulations, started separately from the composite zonal flows of the two extreme phases and small-amplitude wavenumber 6 perturbations, display the intensification of initial wind anomalies by the growing eddies. A dual-jet stream structure appears in the life-cycle simulation started from the high index composite, and a more intense single jet stream structure evolves from the low index initial state.

It is noticed that maximum wind anomalies are established earlier at higher latitudes than at lower latitudes. This suggests that the mechanisms triggering transitions from one self-maintained phase to the other operate at higher latitudes. It is suspected that barotropic instability/stability is a possible triggering mechanism for transition from one state to another.

The effect of large-scale mountains on atmospheric variability is studied in a series of GCM experiments in which a single mountain is varied in height from 0 to 4 km. High-frequency (τ < 7 days) and low-frequency (τ > 30 days) variability are largest in the jet exit region, while the intermediate-frequency (7 < τ < 30 days) variability has its maximum upstream of the mountain where it exhibits enhanced equatorward propagation. High and intermediate frequencies change from zonal wave trains to localized wave packets as orographic forcing is increased, but they retain their characteristic scale and frequency. The dominant pattern of low-frequency is variability changes from a zonally symmetric oscillation, for which transient eddy-zonal flow interaction is the dominant mechanism, to a more localized oscillation of the jet downstream of the mountain. The transient eddy forcing still plays a significant role in maintaining the variations of this more localized jet, however.

The total amount of wave energy remains almost constant as the mountain height is increased, but the distribution of wave energy shifts from transient to stationary and from high frequencies to low frequencies. Low-frequency variability shows a step function response to orographic forcing in that it shows no response to a 1-km mountain, increases substantially in response to a 2-km mountain, and then shows little further increase as the mountain is raised to 3 and 4 km. This behavior suggests that the mechanism that generates the additional low-frequency variability in the mountain-forced experiments becomes effective after the zonal asymmetry reaches a critical value and then does not respond much to further increases in asymmetry.