The time series of geophysical data are chaotic and, on the other hand, extremely noisy. Thus, though there are a number of advanced methods of chaotic time series prediction, the improvement of geophysical data is crucial to succeed. Mainly, it is connected with low determinism of such time series. The improvement procedure, we are about to represent, does not directly enhance deterministic component, but regularity properties of a signal, and, therefore, causes the increase of a deterministic portion in data. The main advantage consists in the fact that the method preserves the initial structure (information) of a time series, while effectively reduces noises, even knowing nothing about its actual nature.To search for other articles by the author(s) go to: http://adsabs.harvard.edu/abstract_service.html