Analysis of time series by means of empirical orthogonal functions

An attempt is made to apply the method of empirical orthogonal functions to stationary random time series. The basic assumption motivating the use of the method is that a time series in many real situations may be considered as composed of randomly distributed physical processes of a duration, which is limited due to dissipation, dispersion or diffusion. The method makes it possible to determine certain characteristic features of the processes but is so far incomplete in the sense that it does not allow for a determination of the corresponding energy spectrum.