DIMENSION ANALYSIS OF EL NINO/SOUTHERN OSCILLATION TIME-SERIES

Even when the behaviour of a system is very irregular only a small fraction of the phase space might be active. The (fractal) dimension provides a measure of complexity to quantify the number of dominant degrees of freedom. We performed a dimension analysis for different time series which are connected with the ENSO-phenomenon. The algorithm was tested using the output of a low-order model showing deterministic chaos. Moreover, the dimension analysis was applied to measured time series and to stochastic processes with the same Fourier-coefficients. In contrast to earlier investigations, no clear evidence of low dimensionality of the underlying dynamics has been found.