Browsing by Author "Bartlett, Edward"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    In search of chaotic elements in the Bay of Fundy water level measurements
    Bartlett, Edward
    The Bay of Fundy has extraordinary water level variations. These variations make accurate prediction using conventional harmonic approaches difficult. A possible approach to improving the predictions is to apply non-linear methods. Since errors in the prediction using harmonic methods appear random, the possibility of this being a system exhibiting chaotic behaviour rather than a stochastic behaviour is investigated. Errors in the prediction are dependent upon the chosen method of analysis in the firs tplace, here a least squares spectral analysis is used. The residual time series from the least squares analysis is then subjected to a search for the largest Lyapunov exponent using two algorithms. Existence of a positive Lyapunov exponent is generally accepted as being proof of chaotic behaviour and this existence is investigated. To further enhance the investigation the Correlation Dimension is estimated. The results are presented and conclusions draw. With the data sample used no clear indication of chaotic behaviour was identified.
  • Thumbnail Image
    Item
    In search of chaotic elements in the Bay of Fundy water level measurements
    Bartlett, Edward
    The Bay of Fundy has extraordinary water level variations. These variations make accurate prediction using conventional harmonic approaches difficult. A possible approach to improving the predictions is to apply non-linear methods. Since errors in the prediction using harmonic methods appear random, the possibility of this being a system exhibiting chaotic behaviour rather than a stochastic behaviour is investigated. Errors in the prediction are dependent upon the chosen method of analysis in the firs tplace, here a least squares spectral analysis is used. The residual time series from the least squares analysis is then subjected to a search for the largest Lyapunov exponent using two algorithms. Existence of a positive Lyapunov exponent is generally accepted as being proof of chaotic behaviour and this existence is investigated. To further enhance the investigation the Correlation Dimension is estimated. The results are presented and conclusions draw. With the data sample used no clear indication of chaotic behaviour was identified.