Spectral Analysis of Caspian Level Variations

The problem of extracting the most common spectral variations within multidimensional time series of Caspian Sea level variation on the shore stations is considered. Accurate data of 15 stations with a sampling time interval of 6 hours over a time period of observations since the beginning of 1977 till the end of 1991 are available for this purpose. The Fourier-aggregated signal method allows extract stationary common harmonics which form 2 groups of tidal sea level variation at the vicinity of 12 and 24 days periods. Besides that a long-periodic common variation was extracted with the period approximately 12.85 years. The non-stationary collective effects within period range from 4 up to 8 days were detected by estimating canonical coherences of multiple time series within moving time window. Non-stationary common effects are associated with influence of wind.