时频联合域贝叶斯地震反演方法

时间域和频率域地震反演分别在不同域中考虑地震响应与模型合成地震记录的匹配程度进而搜索数据空间对应的最优模型参数。从贝叶斯估计理论出发,综合考虑时间域和频率域地震反演在地层分辨率和抗噪性方面特有的性质,联合贝叶斯稀疏反演框架、时间域反演理论及频率域反演方法,提出了基于贝叶斯框架的时频联合域地震反演方法,改善了频率域反演对噪声非常敏感的问题。模型测试表明,时频联合域稀疏反演在地震资料信噪比高的情况下保持了频率域反演分辨率高的优良特性,同时在信噪比低的情况下延续了时间域反演抗噪能力强的优势,有效地提高了地震资料分辨率和压制随机噪声之间的平衡。理论模型测试和实际资料试处理结果验证了该方法的稳定性和可靠性。

Seismic inversion in different domain is utilized to get reflectivity information by considering the matching condition between actual seismic response and synthetic seismogram and searching for the optimal model parameters.We considered the difference of resolution and noise sensitivity for time domain inversion and frequency domain inversion.Combining the theory of time domain inversion and frequency domain inversion,we present a new robust inversion in joint time-frequency domain based on Bayesian scheme,which improves the characteristic of frequency domain inversion oversensitive to noise.The model tests show high resolution of frequency domain inversion for high S/N seismic data and anti-noise of time domain inversion for low S/N seismic data.It balances between enhancing resolution of inversion results and suppressing noise.Finally,applications on theoretical model and trial test of field seismic data verified the stability and reliability of our method.

国家自然科学基金-石油化工基金联合重点项目(U1562215)、国家重点基础研究发展规划(973计划)项目(2013CB228604)、山东省自然科学基金(2014BSE28009)及中国石化地球物理重点实验室开放基金(33550006-15-FW2099-0027)联合资助。