基于高阶统计的时变子波估计及其应用

地震子波估计是地震资料高分辨处理、解释的基础,快速准确地估计地震子波具有重要意义。笔者利用三谱估计子波的振幅谱,根据峰度最大值准则提取子波的相位,规避了相位的卷绕问题,并讨论了该准则的适用条件。比较了不同窗函数的特性,最终选定Hamming窗函数滑动求取时变子波。利用变频雷克子波模型讨论了时变子波提取的有效性。应用到实际资料处理中,与不同主频的雷克子波标定的地震记录进行对比分析。结果表明:该方法使得子波振幅谱估计更加稳定,相位谱估计更易于实现,与实际情况更加符合。

The application of time-varying wavelet estimation based on high-order cumulants

Wavelet estimation is the foundation of high-resolution seismic data processing and interpretation. It is of great significance to estimate the wavelet rapidly and exactly. This paper gets the amplitude spectrum from the trispetrum and phase from the seismic data based on the principle of kurtosis maximum, which avoids phase wrapping properly, and discusses the application condition of this principle. The property of the different window functions is also compared, and Hamming is moved to calculate the time-varying wave- lets. The ticker wavlet model is used which is frequency-changing to illustrate its effectiveness. In the real data processing, the method is comparatively analyzed for the well-seismic calibrated data based on different frequency ticker wavlets. The results show that the esti- mated wavlet amplitude spectrum is more stable, and the phase spectrum can be realized easily and is more consistent with the true wavelet.