Abstract: According to tuning effect of thin bed reflection, spectral-decomposition technique images the thickness and discontinuity of thin bed in frequency domain. It's a good and effective tool for identification and characterization of thin bed. Usually, common spectraldecomposition methods include short-time window Fourier transform (STFT), wavelet transform (WT), maximum entropy(ME), etc. STFT method has the window problem, which limits the vertical resolution, and cannot adapt the thickness transverse change of layers. Normal WT methods use scale parameter, which cannot correspond directly with frequency with ambiguity geological means. NonorthogonalWT avoids upper limitations under the control of frequency parameter directly. It can specify frequency band and distribution density in order to satisfy the identification requirement of thin bed and seismic sequence analysis, and benefit to characterize thin beds

Key words: spectral decomposition, wavelet transform, tuning energy