多震源多分量混采数据全波场重构与联合成像研究

In this study, we propose the maximum distributed acquisition and its application to multi-wave and multi-component seismic wavefield data obtained by multiple active and passive sources from the view of the deep metallic deposit resources and the reflection seismic exploration in hard rock area. The high-accuracy imaging of deep geological structure and mineral resources is realized by forming the broadband seismic records based on the theory of simultaneous acquisition and wavefield reconstruction. The reconstructed seismic records of active and simultaneous source can be obtained by iterative denoising or blind source separation, and the passive seismic records can be obtained by the optimal interferometry to reconstruct reliable multi-component data of virtual source. On the basis of data assimilation, the multi-source, broadband and multi-component joint seismic data are formed by the optimal matching of active and passive sources, which have the time-space consistency and the physical consistency. The reconstructed multi-component data of active and passive sources can also be used in series connection full waveform inversion, and the high-accuracy results can provide accurate parameters for migration imaging. The low frequency of joint data is mainly passive signals, which can ensure the convergence of the global minimum solution of full waveform inversion. The low frequency reconstruction method based on compressed sensing theory can verify the rationality of signals in non-connection region of frequency band of active and passive source seismic data. This project will also develop the study of interferometry virtual directional illumination and direct migration method of multi-source and multi-component blended data.

从深地金属矿资源及硬岩地区反射波地震勘探出发，本课题提出最大化采集并应用多个主动和被动震源产生的多分量地震波场信息，通过混合采集理论和波场重构方法形成宽频多分量地震记录，实现深部地质构造和矿产资源的高精度成像。主动源混采重构的地震记录可通过迭代去噪或稀疏反演方法获取，被动源多分量地震记录可通过优化干涉法或者盲源分离重构。应用数据同化原理，主动源和被动源波场经优化匹配处理后可以形成具有时空一致性和物理一致性的宽频带多震源多分量联合地震记录或称为全波场记录。联合数据低频端以被动源信号为主，可以保证全波形反演的结果向全局最小值收敛。混采重建的主动和被动源多分量数据用于串联全波形反演可以得到可靠的高精度偏移参数模型。基于压缩感知的低频重建方法可以保证主被动源地震数据频带过渡区信号匹配的合理性。课题还将开展多源多分量地震干涉虚拟定向照明、多源多分量混采数据直接偏移等方法研究。

在复杂金属矿或硬岩等地区开展地震勘探时，仅靠常规主动源勘探手段往往难以获得地下构造的高精度成像结果，主要原因在于主动源地震勘探固有的照明局限性和有限带宽的数据特点。被动源地震信号广泛存在于自然环境中，但由于其信号弱且规律性难寻，在常规地震勘探中，往往将其作为噪声而去除。然而，被动震源真实来自于地下，被动源波场携带了大量地下构造的反射响应，且其具有照明角度多样和低频丰富等特点。本项目提出最大化采集并应用多个主动和被动源产生的多分量地震波场信息，通过混合采集理论和波场重构方法形成宽频多分量地震记录，以实现深部地质构造和矿产资源的高精度成像。为在全波形反演中充分利用被动源地震波场的低频优势，以克服缺失低频主动源地震数据全波形反演的跳周问题，本项目提出了主动源和被动源的联合全波形反演方法，并在声波和弹性波介质对其相关方法理论进行了研究。针对被动源信号弱、震源位置和子波未知等问题，本项目提出联合地震干涉法与不依赖震源方法进行主、被动源的联合全波形反演，并最终在接近于实际的复杂条件下获得了高精度的纵、横波速度成像。本项目针对多震源混合采集波场的高保真分离方法开展了研究，提出了基于三边滤波和基于K-SVD字典的混采数据分离方法。为精确定位被动源位置，本项目结合逆时传播与走时定位思想，提出了基于逆向走时成像与基于筛选模型的微震定位方法。另外，本项目还针对克服常规主动源全波形反演的跳周问题，提出了一系列稳健的全波形反演方法策略。本项目研究对于在复杂采集条件下重构宽频多分量地震记录以及获得高精度地下速度结构成像具有重要的科学意义和良好的应用前景。