NONLINEAR TIME-SERIES ANALYSIS OF THE IONOSPHERIC MEASUREMENTS

The applications of the tools of nonlinear time-series analysis to mea surements obtained by digital ionosonde ''PARUS,'' created at IZMIRAN, is considered. Small-scale structure of the electron density Ne at F- region of ionosphere is examined by exploring observables of reflected high-frequency radio wave. Time series corresponding to given transmi tter frequency f and virtual height h and conforming to usual, not per turbed, day ionogram are examined. Using the standard correlation algo rithm, we have determined that for an embedding dimension n=15, comput ed correlation dimension D2(n) reaches its saturation, D2 congruent to 4. Reliability of the estimation of the correlation dimension was con firmed by applying to our sets of measurements J. Theiler's modificati on of standard algorithm. The test of time difference, smoothing proce dure, calculation of the largest Liapounov exponent, and analysis of t he surrogate data were employed in our study of data sets. Although ou r sets of measurements contain a significant ''random noise'', nonline ar time-series analysis clear up, that the dynamics of unperturbed ion osphere, generating this time series, exhibits low-dimensional behavio r.