EVIDENCE FOR THE DELAYED OSCILLATOR MECHANISM FOR ENSO - THE OBSERVEDOCEANIC KELVIN MODE IN THE FAR WESTERN PACIFIC

Observed surface winds from 1961 through September 1992 are used to fo rce a reduced gravity shallow water ocean model. Results from the hind cast of ocean variability are found to be consistent with the results Li and Clarke presented in a previous study of sea level and zonal win d variability in the tropical Pacific. In this note the apparent discr epancies are reconciled between ''delayed oscillator'' theory and calc ulated lag correlations between the observationally based records of w estern boundary Kelvin mode amplitude (eta(w)) and zonal wind forcing. Evidence for sensitivity in these ''delayed oscillator'' lag correlat ions is presented from a variety of sources, including: the hindcast d ata, output from the standard version of the Zebiak and Cane coupled o cean-atmosphere model, and through three cases with idealized time ser ies of ENSO variability. The authors demonstrate that the low lag corr elations for eta(w) leading the zonal wind forcing by 1 to 1 1/2 years is not inconsistent with the hypothesized role of western boundary re flections as the ultimate termination mechanism for ENSO anomalies. Th e lack of regularity in the system studied guarantees a degraded corre lation for eta(w) leading the zonal wind by 1 to 1 1/2 years. This lac k of regularity is not contained in, nor explained by, delayed oscilla tor theory. A robust feature in all of the records examined in this wo rk is the existence of upwelling Kelvin signals in the far western equ atorial Pacific Ocean due to the developing warm (ENSO) events. The am plitude of the observed Kelvin signals in the western Pacific is suffi cient to terminate the developing ENSO events via the delayed oscillat or mechanism.