中文摘要：

高斯束偏移是一种高效、稳健的深度域成像方法,其不仅保持了射线类偏移的高效、灵活性,而且具有接近波动方程偏移的成像精度.匹配追踪是一种基于匹配寻优算法的信号稀疏分解技术,常用于地震信号的时频分析和去噪处理中.本文将建立在Ricker子波原子库的匹配追踪算法应用于常规的高斯束偏移中,并结合Kirchhoff偏移的单输入道成像方式,发展了一种成像精度更高且适用于低信噪比资料的叠前高斯束成像方法.该方法通过合理地控制匹配追踪稀疏分解的迭代次数,可以有效地去除地震信号中的随机干扰,提高成像结果的信噪比;此外,在偏移过程中,本文方法采用了Kirchhoff偏移的单输入道成像方式,解决了常规高斯束方法对浅层小尺度地质体成像不准的问题,提高浅部反射层的成像精度.两个典型的数值算例验证了本文方法的有效性和适应性.

英文摘要：

Gaussian beam migration is an efficient and robust imaging method in depth domain,which not only maintains the flexibility and efficiency of the migration method based on ray tracing but also has the imaging accuracy of the migration method based on solving wave equation numerically. Matching pursuit is a technique of signal sparse decomposition based on the algorithm of searching and matching,which is often applied to time-frequency analysis and denoising processing of seismic data. We apply the matching pursuit algorithm constructed on Ricker wavelet atom library to conventional Gaussian migration,and develop an accurate method of Gaussian beam migration in this paper, which is adaptable for low signal-to-noise ratio data. Our method can effectively remove the random noise of seismic signals by properly controlling iteration numbers of matching pursuit sparse decomposition,raising the signal to noise ratio of imaging results.Further, our method has solved the problem of conventional Gaussian beam migration which can＇t define the shallow small-scale geological constructions accurately,improving imaging quality of shallow reflectors. Two typical numerical examples verified the validity and adaptability of the proposed method.