Multiscale Oscillations of the Global Climate System as Revealed by Wavelet Transform of Observational Data Time Series

Summary  The surface air temperature time series of both hemispheres and the North Atlantic European area as well as the Southern Oscillation (SO) index time series were analysed using a wavelet transform technique. The values of the so-called singularity exponents of these series were estimated and compared with such estimations for some surrogate time series artificially created from the observed temperature series. It was concluded that the climate dynamics on interannual and interdecadal scales may be considered as a kind of classical Brownian motion although its consideration as a flicker-noise is also possible. The extracted temperature variations were shown to be closely coupled with the SO process. The wavelet-transformed SO series reveals itself as a whole self-similar “tree” the main branches of which are the appearances of the strongest El-Ninos of 1898 – 1899, 1941 – 1942, and 1982 – 1983. Similar “trees” can be seen in the wavelet-transformed temperature series. Thus, the extracted temperature variations were shown to be closely coupled with the SO process, and a decomposition of the current global climate dynamics into three climatic epochs (of about 40-year long) seems to be appropriate. Received May 4, 1998 Revised April 25, 1999