基于时频谱模拟的时变混合相位子波提取

为解决地层衰减作用造成传统子波提取方法提取结果不准确的问题,提出一种基于时频谱模拟的时变混合相位子波提取方法.该方法在对非平稳地震记录进行改进的广义S变换的基础上,首次引入时频滤波处理,然后采用谱模拟方法估计每一时刻的子波振幅谱,在子波相位时不变假设下采用高阶累积量的双谱法重构子波的混合相位谱,最后将振幅谱和相位谱结合实现时变混合相位子波的提取.与常规分段提取方法相比,该方法不需要子波分段平稳的假设,当相邻层段子波属性存在较大变化时也能准确估计出时变子波.理论分析和仿真实验结果表明,该方法有效可行.

Time-variant mixed-phase seismic wavelet estimation based on spectral modeling in the time-frequency domain

Seismic wavelet estimated by conventional methods is inaccurate due to earth attenuation. To solve this problem, we propose in this paper time-variant mixed-phase seismic wavelet estimation method based on spectral modeling in the time-frequency domain. First non-stationary seismic record is transformed into the time-frequency domain with the improved generalized S-transformation, and processed with time-frequency filtering. And then amplitude spectrum of time-variant wavelet is estimated using spectral modeling method. With the assumption of time-invariant wavelet phase, mixed phase is estimated by the bi-spectrum. Finally the time-variant mixed-phase seismic wavelet is estimated with the joint use of amplitude spectrum and phase spectrum. Compared with conventional sectional methods for time-variant wavelet estimation, this method can accurately estimate time-variant seismic wavelet without sectional stationary wavelet even in the case of big wavelet change in adjacent strata. Theoretical analysis and synthetic example results show that the proposed method is effective and feasible.