Improvement of edge effect of the wavelet time-frequency spectrum: application to the length-of-day series

The edge effect in the wavelet time-frequency spectrum of a time series is treated. The time series is first extended on both ends by applying a non-l inear model, namely the leap-step time series analysis (LSTSA) model: prior to the wavelet transform. The results of a series of simulation experiment s and an application to the observed length-of-day (LOD) series demonstrate that the edge effect is effectively reduced this way. Thus, the applicatio n of the LSTSA model improves the wavelet time-frequency spectrum, especial ly enhancing the ability to detect the low-frequency signals.