Some problems on the global wavelet spectrum

Abstract

In order to test the validity of the global wavelet spectrum — a new period analysis method based on wavelet analysis, we carried out some simple experiments. In our experiments we used idealized time series and real Nino 3 sea surface temperature (SST) for testing purposes. First we combined different signals which have the same power but different periods into some new time series. Then we calculated the global wavelet spectra and Fourier power spectra for the testing time series. The testing results revealed that on some occasions the global wavelet spectrum tends to amplify the relative power of longer periods. By making comparisons with the results obtained by the traditional Fourier power spectrum, we demonstrated that on an occasion when the global wavelet spectrum does not work the Fourier power spectrum can be used to achieve the right results. Hence it is recommended that when making period analysis with the global wavelet spectrum one needs to do further tests to confirm their results.