Abstract: | Influenced by noise and non-random reflection coefficient, homomorphic deconvolution is hardly used in seismic data processing because of unstable phase of extracted wavelet. Aiming at the problem, we proposed calculating mixed phase wavelet by using combining average Cepstral with average autocorrelation Cepstral, which were computed by small windows and multi-traces data, to improve the stability of wavelet phase and the practicability of homomorphic deconvolution. During the actual data processing, some countermeasures were used to improve the wavelet stability. Firstly, seismic wavelet is changed into mini-phase wavelet by exponential decay on seismic data. Secondly, after energy positioning on the obtained wavelet, the energy concentration part of wavelet is chosen to carry window shaping, and the influence of wavelet side lobe is suppressed. Thirdly, the peak moveout between extracted wavelet and expected wavelet is adjusted. Finally, for 3-D post-stack data, the central trace deconvolution operator is calculated in adjacent multi-traces of INLINE and CDP directions. Theoretical model testing result proved that this method can be used in mixed-phase wavelet seismic data. Real data application results after deconvolution indicated that the calculated wavelet is stable and the resolution of seismic data is enhanced. |