Abstract
A simple coupled ocean–atmosphere model, similar to that of Zebiak and Cane, is used to examine the dynamic and thermodynamic processes associated with El Niño/Southern Oscillation (ENSO). The model is run for 300 years. The interannual variability which results is regular, with a period of either 3 or 4 years, quantized by the annual cycle. The amplitude (∼1.5 m s−1 wind and 2°C SST anomalies), period and structure of the interannual variability compare well with observations. The model warm event is initiated in the spring prior to the event peak, and is well described as an instability of the coupled system. During instability growth, the sea surface temperature (SST) anomaly is primarily generated by vertical upwelling processes. The SST anomaly can be approximately described by the expression ∂T/∂t = KTh − α*T, where T is the SST anomaly, t time, h the upper layer thickness (pycnocline) perturbation and α* an effective damping time which includes heat loss to the atmosphere. KT parameterizes vertical upwelling and mixed layer processes.
Oceanic wave dynamics determines the fate of the growing instability. The warming of the SST produces westerly wind anomalies in the equator central Pacific, forcing equatorially trapped Rossby waves that propagate freely to the western boundary. These waves reflect at the western boundary, sending upwelling equatorial Kelvin waves back to the central basin. These cooling Kelvin waves act to terminate instability growth and rapidly plunge the coupled system into a cold regime. The western boundary reflection is necessary for event termination. The system returns from a cold regime via reduced heat flux to the atmosphere and, to a lesser extent, by wave induced processes like that which lead to the warm event termination. The interannual variability is not produced by vacillation between two equilibrium states: a cold and a warm state. The growth rate to either the cold or warm state is too slow for the system to achieve equilibrium, even for a basin the size of the Pacific. The model results indicate that shortly after the initial set of gravest mode Rossby reflections on the western boundary, the instability growth is already being substantially moderated by the equatorial wave processes in the ocean. Thus the system is oscillatory around a single basic state.
Of the Rossby waves produced in the central Pacific by the warm event, only the two gravest mode symmetric modes are important in the reflection process, which produce the Kelvin waves that terminate the warm event. In nature, the actual western boundary for the equatorial Pacific wave guide is very ambiguous. Calculations indicate, however, that efficient reflection of the gravest symmetric Rossby waves from a more realistic boundary than the meridional wall in the model is possible. Finally, if the model is indeed simulating the correct processes controlling ENSO events, the nature of the instability mechanism that leads to growth and the wave-induced termination of the model warm event suggests that, for realistic instability growth rates for the coupled equatorial ocean-atmosphere system, interannual variability analogous to ENSO should not be possible in equatorial basins significantly smaller than the Pacific.
