K₁ 峭度约束的地震频域反褶积方法研究

Due to the complex and changeable near-surface topography of Ordos Basin, and the seismic absorption effect of seismic wave, the main frequency and vertical resolution of the collected seismic data are low. Frequency domain deconvolution is a key technique to improve the resolution of seismic data in Ordos Basin. Recently, kurtosis is the most commonly used non-Gaussian metric for frequency domain deconvolution. However, the traditional definition of kurtosis is difficult to accurately measure the non-Gaussian property of the actual reflection coefficient series due to its peak and long tail distribution. Different from traditional kurtosis, K\ kurtosis can make full use of the peak information of probability density to indicate the non-Gaussian of random variables. Therefore, a seismic deconvolution method based on Ki kurtosis measure is proposed in this paper, which can judge the non-Gaussiality of reflection coefficient sequence more accurately. In addition, two-parameter generalized Beta mother wavelet, defined in the frequency domain, is introduced, which can make the proposed method better match different seismic wavelet. Finally, a numerical example of synthetic seismic data and 2D seismic data of Ordos Basin is used to demonstrate that the proposed method can effectively improve the longitudinal resolution of low-quality seismic data, especially for several adjacent small reflection coefficients. Furthermore, the well-to-tie results verify that the proposed method increases the dominant frequency of the original seismic data from 25 Hz to 35 Hz, and effectively identifies multiple gas-bearing thin reservoirs, which is benefit for subsequent optimization of well placement.