In the Pacific equatorial band the zonal displacement of the eastern edge of the warm pool, subject to insignificant seasonal variations, is dominated by strong interannual variations almost in phase with the Southern Oscillation Index. Over the 1971--1973 El Ni¿o-La Ni¿a period, Gill (1983) suggested that such displacement was caused solely by horizontal advection by zonal current anomaly. Basin-wide Geosat-derived zonal surface current anomalies are available during the 1986--1989 period, and they agree quite well with observed equatorial currents in the western and central Pacific. Within the 4 ¿N--4 ¿S equatorial band the cumulative contribution of first-baroclinic Kelvin and first-symmetric (m=1) Rossby modes to surface zonal equatorial current anomalies account for most of the Geosat-derived current variability. From these Geosat-derived currents it is demonstrated that the eastward (westward) displacement of the eastern edge of the warm pool was primarily due to horizontal advection by zonal current anomalies during the 1986--1987 El Ni¿o (1988--1989 La Ni¿a). As a corollary, the El Ni¿o (La Ni¿a) warm (cold) sea surface temperature (SST) anomaly in the central western Pacific was the result of anomalous zonal advection. The 1986--1989 displacement of the warm pool appeared as a low-frequency movement resulting from high-frequency forcing, i.e., a succession of local wind forcing and its remote Kelvin and m=1 Rossby wave responses in the equatorial waveguide.

In the central western Pacific a m=1 downwelling Rossby wave, issued from the eastern Pacific, shifted the displacement from eastward to westward and therefore seems to have been the cause for El Ni¿o to turn to La Ni¿a in boreal fall 1987. In the eastern Pacific the simultaneous occurrence of free-propagating downwelling (upwelling) Kelvin waves with the seasonal SST warming (cooling) appeared necessary for the development of local El Ni¿o (La Ni¿a) SST anomaly. Over most of the equatorial Pacific, slow migration of SST anomalies clearly showed up, distinct during El Ni¿o and La Ni¿a. These zonal migrations appeared to result from the delayed association of two phenomena: the zonal advection in the central western Pacific and the arrival of downwelling (upwelling) Kelvin waves into the eastern Pacific, synchronous with the seasonal SST warming (cooling). Albeit different in terms of equatorial wave action, as compared to the delayed action oscillator theory, our results suggest that zonal advection and wave dynamics were both important for the generation and migration of El Ni¿o-Southern Oscillation SST anomalies in 1986--1989. Finally, our observational results about warm pool displacements, equatorial wave sequence, the shift from El Ni¿o to La Ni¿a, and the slow SST migrations are presented ina schematic form involving sequential ocean-atmosphere coupling which appears continuously over the November 1986 to February 1989 period. ¿ American Geophysical Union 1995