The method of empirical mode decomposition （EMD） based Hilbert transformation is that separate the nonlinear time series into intrinsic mode functions （IMFs ） with different time scale by using EMD, then apply the Hilbert transformation to every IMF to get the Hilbert spectrum.By systematically comparing the advan tages and disadvantages between EMD and wavelet decomposition （WD）, between Hi lbert transformation and wavelet transformation in transacting nonlinear time s eries using ideal time series and 18O time series of Guliya i ce co re on the Tibetan Plateau, we present some improving measures to reduce the di sadvantages. The result of investigation show that the combination of EMD-based and WD-based methods may be more effectively in recognizing the main informatio n of the time series.