复杂衰减和噪声干扰下的时变地震子波提取与反褶积方法研究

时变地震子波提取和反褶积是影响地震资料处理和解释精度的关键技术。现有研究无法根据目的层的衰减特征准确估计时变地震子波，同时难以在复杂噪声干扰下反演反射系数序列。本项目针对不同地层介质、地质构造的影响，根据目的层地震波的衰减规律，在品质因子Q为常数的线性衰减模型下研发一种基于经验模态分解和地震波动方程的时变子波提取方法，在Q随频率变化的非线性衰减模型下开发一种基于变分模态分解和递推最小二乘的时变子波提取方法，从而在叠后非平稳地震资料中提取能够真实反映地层衰减特性的时变地震子波。建立高斯和非高斯噪声共同干扰下非平稳稀疏反褶积的约束条件和目标函数，利用所估计的时变子波在复杂地震资料中反演反射系数序列。构建迭代反馈策略，通过多次迭代逐步提高时变子波和反射系数序列的提取精度，最终实现基于非平稳褶积模型的地震反问题的准确求解，为有效提高地震资料分辨率提供系统的解决方案。

Time-varying seismic wavelet extraction and deconvolution are critical techniques affecting the accuracy of seismic data processing and interpretation. Existing studies cannot accurately estimate time-varying seismic wavelets according to the attenuation characteristics of the target layers, and it is difficult to invert the reflection coefficient sequence under the interference of complex noise. In this project, aiming at the influence of different stratigraphic medium and geological structure on the attenuation law of seismic wave, a time-varying wavelet extraction method based on Empirical Mode Decomposition and seismic wave equation is proposed under the linear attenuation model of quality factor Q independent of frequency, and a time-varying wavelet extraction method combining Variational Mode Decomposition and recursive least squares is developed under the nonlinear attenuation model of Q varying with frequency, so that time-varying seismic wavelets which can truly reflect the attenuation characteristic of strata can be extracted from post-stack non-stationary seismic data. The constraint conditions and objective functions of nonstationary sparse deconvolution with Gaussian and non-Gaussian noises are established, then the estimated time-varying wavelets are used to invert the reflection coefficient sequence in complex seismic data. An iterative feedback strategy is constructed, and the extraction precision of time-varying wavelet and reflection coefficient sequences is gradually improved by multiple iterations. Finally the accurate solution of seismic inverse problem based on non-stationary convolution model is realized, which provides a systematic solution for effectively improving the resolution of seismic data.