Abstract：The final goal of seismic data deconvolution is to eliminate influence resulting from seismic wavelet to recover the reflection coefficient series of formation. Spectrum modeling method is advanced, which involves the following essential paint: With the consideration of space and time variations of wavelet, improved spectrum modeling processing is taken to have both desirable deconvolution operator and modeled amplitude spectrum to finally achieve zero-phase deconvolution of seismic data in frequency domain. In view of the fact that reflection coefficient is not always random white noise, and with the use of existing borehole data in the prospect, spectrum fitting is done to obtain the reflection coefficient spectrum of formations that correspond to sedimentary characteristic in the prospect. Spectral "color compensation"to processing result in the first step is achieved to improve the resolution of poststack seismic data. Modeling and real processing results prove the method right and practical.
