SEARCH

SEARCH BY CITATION

A comprehensive statistical analysis of how an index of the Southern Oscillation changed from 1882 to 1995 was given by Trenberth and Hoar [1996], with a focus on the unusual nature of the 1990–1995 El Niño-Southern Oscillation (ENSO) warm event in the context of an observed trend for more El Niño and fewer La Niña events after the late 1970s. The conclusions of that study have been challenged by two studies which deal with only the part of our results pertaining to the length of runs of anomalies of one sign in the Southern Oscillation Index. They therefore neglect the essence of Trenberth and Hoar, which focussed on the magnitude of anomalies for certain periods and showed that anomalies during both the post-1976 and 1990-mid-1995 periods were highly unlikely given the previous record. With updated data through mid 1997, we have performed additional tests using a regression model with autoregressive-moving average (ARMA) errors that simultaneously estimates the appropriate ARMA model to fit the data and assesses the statistical significance of how unusual the two periods of interest are. The mean SOI for the post-1976 period is statistically different from the overall mean at <0.05% and so is the 1990-mid-1995 period. The recent evolution of ENSO, with a major new El Niño event underway in 1997, reinforces the evidence that the tendency for more El Niño and fewer La Niña events since the late 1970s is highly unusual and very unlikely to be accounted for solely by natural variability.