Chaotic characteristics of the Southern Oscillation Index time series

The monthly time series of the Southern Oscillation Index (SOI) is analysed to examine its chaotic characteristics. Three schemes, moving average, low-pass filter and nonlinear smoothing, were used to reduce noise and entrance chaotic properties. Autocorrelation and spectral characteristics, as well as three chaos-oriented properties - phase space trajectory, the largest Lyapunov exponent and correlation dimension - were then examined. No significant signs of chaotic behaviour were found for either the noise-reduced SOI time series or the raw one. Although it contains long-term periodicity, the SOI time series is considered to be stochastic rather than chaotic from the viewpoint of dynamical systems theory. (C) 1998 Elsevier Science B.V.