Elastic full waveform inversion based on visibility analysis and energy compensation for metallic deposit exploration
Elastic full waveform inversion based on visibility analysis and energy compensation for metallic deposit exploration
复制标题
DOI：
--
复制
发表时间：
2015
期刊：
影响因子：
--
通讯作者：
Sun Hong
中科院分区：
文献类型：
--
作者：
Sun Hong
复杂环境中弱、不连续的反射和散射特征，使常规金属矿纵波地震勘探难以建立适合成像的速度模型。多波多分量地震勘探技术在解决这一问题上具有优势。利用多分量和预处理的弹性全波形反演（EFWI）技术，可以有效地提高金属矿纵波地震勘探的速度分辨率，从而提高金属矿纵波地震勘探的精度。叠加地震资料可以重建高精度的多参数模型，用于地下深部复杂存款构造的成像，本文介绍了一种基于可见性分析和自适应能量补偿的EFWI方法，该方法可以更充分地利用与矢量地震场相关的先验信息对于特定的地震采集系统，只有一定范围内的震源和检波器信号才对目标体的反演和成像有影响，根据单炮-检波器对目标体成像的贡献，统计分析方法用于获得目标区域的整个照明强度，其被定义为单个源的可见度，为了在不增加计算和采集成本的情况下通过尽可能多地利用与目标区相关的地震信息来改善反演结果，在波场匹配过程中，可以根据几何可见度构造EFWI的目标函数，以增加与目标区相关的残差比例。本文提出了一种新的波场拟合方法，即利用地震波场的速度差来拟合目标函数，并将观测数据与计算数据进行拟合，能量较弱的波场即使速度差较大，对目标函数的拟合误差贡献也较小，几种情况下能量分布不均匀的现象可以用地震波场来表示本文提出的EFWI算法是用有限差分法实现的，它是一种基于有限差分法的EFWI算法，它的基本思想是利用弹性波场的双向照明强度作为加权因子，自适应地优化和平衡EFWI的梯度。在时间域中用差分法计算弹性波场，用LBFGS优化法修正初始模型，在时间域中用多参数梯度法计算弹性波场（密度、P波和S波速度）通过时域中的前向波场和后向波场之间的互相关来计算。首先，在Marmousi Ⅱ模型上验证了该方法的有效性，与常规EFWI方法相比，通过能见度分析和能量补偿，反演结果的精度得到了提高。在鲁宗盆地某金属模型上的模型试验结果也表明，该方法使反演更快地收敛到目标函数的全局最小值，并提供了准确的反演结果。多的数值结果表明，基于可见性分析的弹性全波形反演算法充分利用了目的层的有效信息，通过层间透射系数将弹性波场的照明强度引入EFWI，可以自适应地平衡波场的能量分布，并能有效地提高成像质量。两种算法均能有效地提高成像质量EFWI方法可以精确地重建金属矿床的深部多参数模型，包括密度、P波和S波速度等，为获得高分辨率的存款构造偏移成像提供了可能。
The weak and discontinuous reflections and scattering features in complex environments make it difficult to construct velocity models suitable for imaging in conventional P-wave seismic exploration for metallic deposits.Multi-wave and multi-component seismic exploration technology may have the advantages in effort to solve this problem.Elastic full waveform inversion(EFWI)exploiting the multi-component and pre-stack seismic data can reconstruct highly accurate multiparameter models for imaging complex deposit structure in the deep subsurface.The EFWI method based on visibility analysis and adaptive energy compensation is introduced in this paper which can utilize prior information more adequately relevant to vector seismic fields from the target.For a specific geometry in seismic acquisition,only signals from sources and receivers within a certain extent have effect on the inversion and imaging about the objective body.According to the contribution for a single shot-geophone pair to the imaging of the objective body,the statistical analysis method is used to obtain the whole illumination intensity of the target zone which is defined as the visibility of the single source-receiver pair.To improve the inversion result by utilizing the seismic information related to the target zone as much as possible without increasing computation and acquisition costs,we can construct the objective function of EFWI based on geometry visibility to increase the proportion of residuals related to the target zone during wavefield matching.FWI is a data-fitting procedure between observed and calculated data.Wavefields with weaker energy make less contribution to the misfit of objective function even though the velocity contrast is bigger.The uneven distribution of energy due to several cases can be expressed by seismic illumination.So bi-directional illumination intensity of the elastic wavefield is used as a weighting factor to adaptively optimize and balance the gradients of EFWI.Zoeppritz equations control the process of automatic compensation because they describe the energy distribution of seismic waves across the interface of impedance.The EFWI algorithm proposed in this paper is implemented by the finite-difference method to calculate the elastic wavefields in the time domain and LBFGS optimization method to update the initial models.The multi-parameter gradients(density,P-wave and S-wave velocity)are calculated by the cross-correlation between forward and backward wavefields in the time domain.First,the validity of our inversion method is verified on the MarmousiII model.The accuracy of inversion results with visibility analysis and energy compensation is improved compared with conventional EFWI.The model test results based on one metallic model of Luzong Basin also illustrate that our new method makes the inversion converge to global minimum of objective function faster and provides accurate and multi-parameter models which can be applied to highresolution seismic migration imaging of metallic deposits.The numerical results show that the elastic full waveform inversion algorithm based on visibility analysis utilizes the effective information from the target zone adequately.Introducing the illumination intensity of the elastic wavefield to EFWI by transmission coefficients between two layers can also balance the energy distribution of wavefields adaptively.Both of the two inversion strategies with illumination can improve the accuracy of imaging.The EFWI method can accurately reconstruct deep and multi-parameter models for metallic deposits,including density,P and S wave velocity,which makes it possible to obtain high-resolution migration imaging about deposit structures.