Multiresolution Fourier transform and its application to analysis of CO2 fluctuations over alert

Difficulties in using conventional Fourier spectral analysis to explore the temporal variation of CO2 short-term fluctuations have indicated the need for some new statistical techniques. It is suggested here that the Multiresolution Fourier Transform (MFT) can be used for interpreting such fluctuations. The technique is adapted from wavelet theory in signal processing. To demonstrate the usefulness of this technique, some examples of analytic fluctuation are first examined. The examples suggest that this technique might be useful for analysis of fluctuations that exhibit locality in both frequency and time. MFT is then applied to study the CO2 fluctuations measured at Alert from 1988 to 1995. The results show that the CO2 concentration over Alert has two dominant regimes of quasi-periodic short-term fluctuation: 1)20-50 day intraseasonal fluctuation and 2) 6-14 day synoptic fluctuation. Both the amplitude and frequency of these fluctuations are strongly modulated by the seasonal cycle and the largest amplitude appears during winter time. In addition, these short-term fluctuations have significant interannual variability, especially for the synoptic fluctuation.