Anelastic attenuation in seismic data: modeling, measurement, and correction

Peng Cheng

Various aspects of the anelastic attenuation of seismic data are investigated in this thesis, including modeling, measurement, and correction using Gabor deconvolution.

For the modeling of seismic attenuation, a nonstationary convolution model, the reflectivity method, and a finite-difference technique for viscoelastic seismic modeling are described. For the implementation of the nonstationary convolution model, two approaches are proposed to incorporate the impulse response representing attenuation process accurately. The reflectivity method is implemented for stratified anelastic media, and the implementation is validated in terms of producing correct events and amplitudes, accurate incorporation of attenuation, and the flexibility to give total and partial response of the media. For the finite-difference method of viscoelastic seismic modeling, o both VSP data and reflection data.

For the estimation of seismic attenuation, three new methods are presented, including complex spectral-ratio method, interpretive spectral-ratio method, and match-filter method. The proposed methods have connections with the classic spectral-ratio method, the spectrum-modeling method and the match-technique method which are also described. The performances of these methods are evaluated in terms of robustness to noise and capacity of estimating from reflection data, using synthetic VSP data, real VSP data and synthetic 1D reflection data. The effects of spectral smoothing, frequency band, and stationary deconvolution on estimation are investigated. For the complex spectral ratio-method, the issue of inaccurate reference frequency is addressed and three approaches are proposed to deal with real VSP data. In addition, an approach to identify the localized low zone of reflection data is proposed and evaluated using synthetic 2D data and field 2D data. Among all the -estimation methods evaluated in this thesis, the match-filter method is significantly superior to other methods in terms of accuracy and robustness to noise when applied to both VSP data and reflection data.

As a nonstationary processing approach to compensate for seismic attenuation, Gabor deconvolution is investigated, and a practical way to correct the white-reflectivity assumption is presented. A definition of nonstationary phase rotation is proposed for the removal of phase rotation. The temporal color and spectral color of reflectivity are defined, and their influence on reflectivity estimation is analyzed in detail for Gabor deconvolution. The color correction method is applied to a field 2D line to restore the high frequency components and obtain a better well-tie.