数据驱动的控制束偏移方法

A data-driven controlled-beam migration method

射线束偏移虽是一类灵活、高效且精度较高的叠前深度偏移成像方法,可实现陡倾地层和多波至成像,但传统的高斯束偏移将所有反射能量沿等时面投影到地下空间,对τ-p域中的所有同相轴进行偏移成像,带来巨大计算量、偏移噪声及偏移假象。作为高斯束偏移的一种改进,控制束偏移在τ-p域选取优势能量同相轴做偏移,能显著提高计算效率,降低偏移噪声;然而控制束偏移方法大多只利用了检波点处的水平慢度信息。为此,本文提出一种充分利用炮点和检波点水平慢度信息的数据驱动控制束偏移方法。即首先分别对局部共检波点道集和共炮点道集进行倾斜叠加,再利用相干分析拾取炮点和检波点水平慢度信息,然后依据设计的偏移质量指示因子筛选拾取的同相轴,最后对所选取同相轴进行成像。与传统高斯束偏移方法相比,数据驱动的控制束偏移方法对偏移噪声和偏移假象的压制更彻底,对低信噪比数据成像具有显著优势。模型数据和实际数据测试验证了该方法的可行性、有效性和适用性。

The ray-based beam migration is a flexible,efficient and relatively accurate prestack depth imaging approach,which can resolve steep-dip structures and image multiple arrivals.The Gaussian beam migration projects all the reflection events into the subsurface along the traveltime isochrones,indicating that it will migrate all the events in theτ-p domain,which leads to huge calculations,and migration noise and artifacts.As an improvement for Gaussian beam migration,controlled beam migrations use only selected events.But most controlled beam migrations just use the instantaneous slowness information at receivers,neglecting the slowness information at sources.In this paper,in order to make full use of the slowness information both at sources and receivers,we propose a datadriven controlled beam migration.First local slant stacks of common-receiver gathers and commonshot gathers are carried out,the slowness information at sources and receivers are extracted with the coherence analysis.Then a migration quality control factor is designed to choose events.Finally,the selected events are imaged.Compared with the conventional Gaussian beam migration,the data-driven controlled beam migration can better suppress migration noise and artifacts,and achieve better imaging of low signal-to-noise ratio data.Model and field data tests prove the feasibility,validity,and applicability of the proposed method.