信息工程专业英语Unit 9 Wavelet Analysis.pptx

Unit 9 Wavelet Analysis9.1 Text9.2 Reading Materials9.1 TextAn Introduction to Wavelet Analysis　Many time series in geophysics exhibit non-stationarity in their statistics. While the series may contain dominant periodic signals, these signals can vary in both amplitude and frequency over long periods of time. 　An example would be sea surface temperatures in the equatorial Pacific Ocean (Fig 9.1). The dominant “mode” of variability is the EL Ni?o-Southern Oscillation (ENSO), shown by the high-frequency “spikes” on a time scale of 2-7 years. Superimposed on this signal are much longer interdecadal fluctuations. The interdecadal fluctuations have the effect of modulating the amplitude and frequency of occurrence of El Ni?o events. Fig 9.1 Sea surface temperatures in the equatorial Pacific Ocean　　Sea surface temperatures averaged over the NINO3 region in the eastern Pacific (5°S-5°N, 90°W-150°W). Blue curve is low-pass filtered (12 months) SST. Yellow background curve is running 15-year variance, plotted at mid-point of 15-year period. Curve has been reproduced upside-down to show “envelope” of variance. 　　Ideally, one would like to separate the shorter period oscillations from the longer. [Note: the word “oscillation” is used here to indicate any repeating fluctuation in the time series, regardless of whether the fluctuation repeats on a regular basis or not.] In our example the El Ni?o is definitely an aperiodic (irregular) signal.　　The simplest method for analyzing non-stationarity of a time series would be to compute statistics such as the mean and variance for different time periods and see if they are significantly different. In Fig 9.1, we have also plotted the running 15-year variance, as a measure of total power inherent in the signal versus time. One can see that ENSO had more variance during 1880-1920 and also since 1950, with a relatively quiet period during 1920-1950. 　While the running variance tells us what the overall strength of the signal was at c