Long equatorial wave reflection in the Pacific Ocean from TOPEX/POSEIDON data during the 1992-1998 period

The TOPEX/POSEIDON sea level data have been shown to observe the sea surfac e height variability accurately over the globe and, more particularly, in t he tropical Pacific, core region of the El Nino/Southern Oscillation. Vario us studies have already provided descriptions of long equatorial wave propa gation as well as estimates of their reflection at both the eastern and wes tern boundaries using TOPEX/POSEIDON data over short time periods. In this present work, we are first interested in computing long equatorial wave amp litudes over a much longer period (October 1992-May 1998) focusing on the K elvin and first three largest Rossby waves. Secondly, we are examining qual itatively potential theories of ENSO in the light of observations of the st rong 1997-1998 El Nino event. At the eastern boundary, Kelvin waves are obs erved to reflect into first (third) mode Rossby waves with a reflection eff iciency of 75% (65%) of that of an infinite meridional wall. In the western Pacific Ocean, the reflection of the first Rossby wave does explain more t han 90% of the Kelvin wave variance. The second Rossby wave is found not to be correlated to the Kelvin wave amplitude, and the third Rossby wave is h ighly correlated to the Kelvin signal although its contribution is weak. Th e investigation during the 1997-1998 event of two theories potentially impo rtant for ENSO (the delayed action oscillator mechanism and the recent theo ry of Picaut et al., 1997) led us to the following conclusions. The westerl y wind anomalies observed in the western Pacific in December 1996 and March 1997 forced downwelling Kelvin waves which advected the eastern edge of th e warm pool eastward and deepened the thermocline in the east Pacific. Both zonal advection and vertical processes are suggested to act constructively during the onset of positive anomalies in the central and eastern Pacific. From May to September 1997, downwelling Kelvin wave wind-forced in the cen tral Pacific and reflected downwelling Rossby waves (reinforced by local ea sterly wind anomalies) acted against each other, and the 28 degrees C isoth erm was found to move slightly westward. In October 1997, strong westerly w ind anomalies forced a strong downwelling Kelvin wave potentially responsib le for the strong warming in the east Pacific. The 28 degrees C isotherm re ached the eastern boundary. At that time, the equatorial sea surface temper atures are zonally homogeneous. The strong downwelling Rossby waves reflect ed at the boundary could not act to terminate the warm event through zonal advection as suggested by Picaut et al. However the reflected upwelling Kel vin waves coming from the western boundary weakened the downwelling Kelvin signal in the central and east Pacific leading to a decrease of the sea sur face temperature anomalies. Following this decrease, westerly wind anomalie s in the central Pacific weakened, and upwelling Kelvin waves reinforced by easterly wind anomalies in the western Pacific propagated toward the easte rn Pacific. As a conclusion, both theories must be considered simultaneousl y to understand the variability observed during the 1997-1998 El Nino event . However observations strongly suggest the delayed action oscillator mecha nism to be the major process at work during the weakening of the warm 1997- 1998 ENSO.