盐下构造速度建模与逆时偏移成像研究及应用

盐丘速度建模及成像是盐下油气藏勘探有关技术瓶颈问题.盐下构造由于盐丘速度与围岩地层差异大,且厚度横向变化大,造成地震波场复杂及时间域构造畸变.针对H区复杂盐丘的地质特征,通过技术创新重新认识盐下油气藏.针对盐丘速度建模的难点,提出了"多信息约束层控实体建模技术",采用序贯高斯模拟及克里金趋势约束速度反演方法,较好解决了盐下速度异常问题,大大提高了速度建模的精度;针对盐下复杂构造成像, 基于有限差分方法研究了精确且高效的差分格式逆时波场外推算法.基于GPU/CPU协同平台,将波场延拓通过GPU实现.采用逆时偏移深度域成像技术,使高角度反射界面、甚至超过90°盐丘侧翼界面的反射波精确成像.通过盐丘理论模型试算验证算法及方法的正确性.上述方法解决了盐丘速度建模精度问题、盐丘侧翼的回转构造成像问题,实现了对盐丘边界及盐丘侧翼的准确归位.消除了速度异常造成的时间域构造畸变,使盐下地层在深度域能够准确成像.     

Wave-equation migration velocity analysis. II. Subsalt imaging examples

Subsalt imaging is strongly dependent on the quality of the velocity model. However, rugose salt bodies complicate wavefield propagation and lead to subsalt multipathing, illumination gaps and shadow zones, which cannot be handled correctly by conventional traveltime‐based migration velocity analysis (MVA). We overcome these limitations by the wave‐equation MVA technique, introduced in a companion paper, and demonstrate the methodology on a realistic synthetic data set simulating a salt‐dome environment and a Gulf of Mexico data set. We model subsalt propagation using wave paths created by one‐way wavefield extrapolation. Those wave paths are much more accurate and robust than broadband rays, since they inherit the frequency dependence and multipathing of the underlying wavefield. We formulate an objective function for optimization in the image space by relating an image perturbation to a perturbation of the velocity model. The image perturbations are defined using linearized prestack residual migration, thus ensuring stability, relative to the first‐order Born approximation assumptions. Synthetic and real data examples demonstrate that wave‐equation MVA is an effective tool for subsalt velocity analysis, even when shadows and illumination gaps are present.     

库车坳陷复杂高陡构造地震成像研究

复杂构造地震成像主要取决于叠前地震数据品质、偏移速度可靠性和偏移算子成像精度. 库车坳陷异常复杂的近地表条件导致极低信噪比的地震采集数据. 该区逆冲推覆高陡构造刺穿盐体大面积分布, 盐层厚度变化大、顶底面形态复杂, 盐下断裂带破碎、小断块发育, 形成异常复杂的地震成像问题. 本文重点研究三个关键环节:(1)精细的叠前地震预处理研究: 根据该区地震地质复杂性和地震资料特征, 采用一些新的方法技术和技术组合从振幅与时移的大、中、小尺度变化三个层次来解决资料信噪比问题, 重建深部反射信号; (2)三级偏移速度分析研究:利用库车坳陷盐刺穿逆冲推覆构造建模理论及变速成图配套技术解决叠前时间偏移速度场时深转换问题,利用井约束低频速度地震迭代反演技术解决连井层速度场与偏移速度场的融合问题,实现从DMO速度分析、叠前时间偏移速度分析到叠前深度偏移速度分析的有机衔接,建立拓扑结构相对保持的叠前深度偏移速度模型;(3)基于退化Fourier偏移算子的半解析波动方程叠前时间和深度偏移研究, 极大地改善了地震偏移过程中高波数波的成像问题. 通过对库车坳陷大北、博孜、却勒、西秋4和西秋10等复杂高陡构造的叠前时间和深度偏移地震成像处理,取得了较好的应用效果.     

地下复杂介质地震处理中的CFP技术

要简要介绍CFP(Common Focus Point)方法技术的基本原理和主要应用。CFP是复杂介质地震处理中的一项新技术，它把叠前偏移分成两个独立的步骤：首先对检波点(炮点)进行聚焦处理，产生共聚焦点道集(CFP道集)，然后再对炮点(检波点)进行聚焦，产生叠前偏移的输出。两个步骤中间的CFP道集则可以进行其它处理。如果是为了寻找构造信息，那么可以应用共焦点CFP偏移；如果是为了寻找岩石、孔隙或流体的信息，则要应用双焦点CFP偏移。目前，该技术主要应用于：(1)CFP两步聚焦法偏移；(2)叠前深度偏移速度模型的建立；(3)试图通过算于而不是速度来解决复杂地表的静校正；(4)消除全程或层间多次波；(5)CFP方法基准面的延拓和盐下成像；(6)多分量地震资料的偏移成像。     

A robust approach to time‐to‐depth conversion and interval velocity estimation from time migration in the presence of lateral velocity variations

The problem of conversion from time‐migration velocity to an interval velocity in depth in the presence of lateral velocity variations can be reduced to solving a system of partial differential equations. In this paper, we formulate the problem as a non‐linear least‐squares optimization for seismic interval velocity and seek its solution iteratively. The input for the inversion is the Dix velocity, which also serves as an initial guess. The inversion gradually updates the interval velocity in order to account for lateral velocity variations that are neglected in the Dix inversion. The algorithm has a moderate cost thanks to regularization that speeds up convergence while ensuring a smooth output. The proposed method should be numerically robust compared to the previous approaches, which amount to extrapolation in depth monotonically. For a successful time‐to‐depth conversion, image‐ray caustics should be either nonexistent or excluded from the computational domain. The resulting velocity can be used in subsequent depth‐imaging model building. Both synthetic and field data examples demonstrate the applicability of the proposed approach.     

叠前逆时偏移影响因素分析

反射地震勘探中的偏移成像技术是获取地下介质构造形态最有效的手段之一.在叠前深度域偏移方法中,目前工业界采用的方法包括基于射线理论的波动方程积分解法和基于波动理论的微分波动方程单程波解法,这两类方法难以处理地震波横向速度变化剧烈的高陡倾角构造成像问题.近年来勘探地震学研究领域发展起来的叠前逆时偏移采用了双程波求解微分波动方程的算法,这种方法具有相位准确、不受介质横向速度变化和高陡倾角构造的影响、成像精度高、可以利用回转波正确成像等优点,从理论上弥补了当前工业界常规地震偏移所面临的成像缺陷.然而,叠前逆时偏移成像方法从理论走向实用尚需解决如下问题:计算速度和数据存储空间的节省、初始速度模型的建立、震源子波的选择、数值模型边界条件的定义和假像的消除等等.对于计算速度和存储量大的问题,随着计算机硬件的快速发展,将会不断得到改善,同时可以采取一些计算技术和存储策略来加以缓解.本文主要针对初始速度模型的建立、震源子波的选择、数值模型边界条件的定义和假像的消除这些因素,利用简单模型进行了分析.对于反射波造成的传播路径上的假像,给出了一种振幅补偿滤波方法.对勘探地球物理学界给出的SEG/EAGE二维盐丘模型、Marmousi模型和本研究设计的崎岖海底模型进行了叠前逆时偏移成像,均取得了较好的成像效果.     

共炮检距高斯波束偏移的一种快速实现算法

与共炮高斯波束偏移相比,共炮检距高斯波束偏移具有直接抽取炮检距域共成像点道集的优势.过去,共炮检距高斯波束偏移以损失成像精度的代价采用最速下降法来降低积分的维数,从而提高计算效率.但经过最速下降近似简化的偏移公式仍是频率域的,需要在每个频点进行计算.为此,本文提出一种快速实现算法来避免采用最速下降法.本文通过分析一个水平层状速度模型的偏移过程和Marmousi速度模型的成像结果来检验不同插值方法对快速实现算法的成像精度和计算效率的影响,并建议采用二维三次卷积插值方法.同时本文在Marmousi速度模型下验证了快速实现算法相对于最速下降法在成像精度和计算效率上的优势.此外,本文将采用二维三次卷积插值的快速实现算法应用于Sigsbee2A模型并获得了清晰的盐下图像.     

层析成像反演速度建模方法研究

本文应用层析成像反演方法优化叠前深度偏移速度模型。阐述叠前深度偏移速度建模流程,详细介绍利用两种不同层析反演方法来优化深度域的层速度模型,对同一实际地震资料速度模型建立的应用效果表明,在资料信噪比较高的地区能提供准确解释层位信息的前提下,模型层析反演优化后的速度模型精度要比网格层析反演优化后的速度模型更高,且稳定性强。      

Stereotomography assisted by migration of attributes

Depth velocity model building remains a difficult step within the seismic depth imaging sequence. Stereotomography provides an efficient solution to this problem but was limited until now to a picking of seismic data in the prestack time un-migrated domain. We propose here a method for stereotomographic data picking in the depth migrated domain. Picking in the depth migrated domain exhibits the advantage of a better signal-to-noise ratio and of a more regular distribution of picked events in the model, leading to a better constrained tomographic inverse problem. Moreover, any improvement on the velocity model will improve the migrated results, again leading to improved picking. Our strategy for obtaining a stereotomographic dataset from a prestack depth migration is based on migration of attributes (and not on a kinematic demigration approach!). For any locally coherent event in the migrated image, migration of attributes allows one to compute ray parameter attributes corresponding to the specular reflection angle and dip. For application to stereotomography, the necessary attributes are the source/receiver locations, the traveltime and the data slopes. For the data slope, when the migration velocity model is erroneous, some additional corrections have to be applied to the result of migration of the attributes. Applying these corrections, our picking method is theoretically valid whatever the quality of the migration velocity model. We first present the theoretical aspects of the method and then validate it on 2D synthetic and real seismic reflection data sets.     

Migration methods for imaging different-order multiples

Multiples contain valuable information about the subsurface, and if properly migrated can provide a wider illumination of the subsurface compared to imaging with VSP primary reflections. In this paper we review three different methods for migrating multiples. The first method is model-based, and it is more sensitive to velocity errors than primary migration; the second method uses a semi-natural Green's function for migrating multiples, where part of the traveltimes are computed from the velocity model, and part of the traveltimes (i.e., natural traveltimes) are picked from the data to construct the imaging condition for multiples; the third method uses cross-correlation of traces. The last two methods are preferred in the sense that they are significantly less sensitive to velocity errors and statics because they use “natural data” to construct part of the migration imaging conditions. Compared with the interferometric (i.e., crosscorrelation) imaging method the semi-natural Green's function method is more computationally efficient and is sometimes less prone to migration artifacts. Numerical tests with 2-D and 3-D VSP data show that a wider subsurface coverage, higher-fold and more balanced illumination of the subsurface can be achieved with multiple migration compared with migration of primary reflections only. However, there can be strong interference from multiples with different orders or primaries when multiples of high order are migrated. One possible solution is to filter primaries and different orders of multiples before migration, and another possible solution is least squares migration of all events. A limitation of multiple migration is encountered for subsalt imaging. Here, the multiples must pass through the salt body more than twice, which amplifies the distortion of the image.     

Velocity analysis after migration

The double‐square‐root (DSR) equation used in pre‐stack migration is formulated in terms of velocity‐dependent and velocity‐independent terms. The velocity‐dependent term is shown to be the hyperbolic normal moveout (NMO) correction, whereas the velocity‐independent term is related to the recording geometry only. This separation of the velocity‐dependent term offers a means of applying vertical corrections to an initial migration velocity field. Using this concept, procedures are described both for velocity determination and for achieving improved structural imaging.
This decoupling is accurate both for constant‐velocity media and for media whose velocity varies as a function of depth. In media whose velocity varies as a function of both space and depth, a procedure is described for building velocity models through common‐image gather (CIG) stacking following prestack depth migration (PSDM) and time conversion (TC). This so‐called PSDM‐TC stack procedure provides a means of (a) incorporating both vertical and lateral velocity updates into an initial velocity model, (b) obtaining improved structural imaging by using a non‐optimal velocity model for the prestack depth migration, and (c) updating velocity by flattening CIGs and maximizing stack energy. The procedure can be applied to both P‐P wave and P‐SV wave migration.     

三维叠前深度偏移速度分析及蒙特卡洛自动层速度拾取

是否能够正确地建立深度域三维速度模型是三维叠前深度偏移成败的关键 .本文根据Deregowski循环 ,利用叠前深度域地震成像对速度模型变化的敏感性 ,采用偏移迭代逐次逼近最佳成像速度 ,研究开发了一套快捷有效的三维叠前深度偏移深度域速度模型建立技术 .借鉴时间域CDP(共深度点 )道集上常规叠加速度分析的策略 ,在深度域CRP(共反射点 )道集上 ,提出剩余慢度平方谱的概念并建立相应的实现技术 .导出深度域中均方根速度与层速度之间的关系 ;按照串级偏移原理确定偏移循环过程中初始速度、剩余速度及修改后速度之间的关系 ;采用蒙特卡洛非线性优化算法实现从剩余慢度平方谱中自动拾取层速度 ,讨论了其地质速度约束条件和蒙特卡洛非线性优化的收敛准则 ,使得所拾取的层速度模型具有合理的地质意义并获得最佳偏移成像效果 .SEG EAGE理论模型数值试算验证了方法的有效性 ,在海拉尔盆地霍多莫尔工区 ,5 8km2 三维资料的速度模型建立并获得满意的三维叠前深度偏移成像 .     

TI介质局部角度域射线追踪与叠前深度偏移成像

研究与实践表明,对于长偏移距、宽方位地震数据,忽略各向异性会明显降低成像质量,影响储层预测与描述的精度.针对典型的横向各向同性(TI)介质,本文面向深度域构造成像与偏移速度分析的需要,研究基于射线理论的局部角度域叠前深度偏移成像方法.它除了像传统Kirchhoff叠前深度偏移那样输出成像剖面和炮检距域的共成像点道集,还遵循地震波在成像点处的局部方向特征、基于扩展的脉冲响应叠加原理获得入射角度域和照明角度域的成像结果.为了方便快捷地实现TI介质射线走时与局部角度信息的计算,文中讨论和对比了两种改进的射线追踪方法:一种采用从经典各向异性介质射线方程演变而来的由相速度表征的简便形式;另一种采用由对称轴垂直的TI(即VTI)介质声学近似qP波波动方程推导出来的射线方程.文中通过坐标旋转将其扩展到了对称轴倾斜的TI(即TTI)介质.国际上通用的理论模型合成数据偏移试验表明,本文方法既适用于复杂构造成像,又可为TI介质深度域偏移速度分析与模型建立提供高效的偏移引擎.     

角度域剩余深度对剩余速度的敏感性分析

本文针对地震勘探深度域偏移速度建模研究,利用角度域共成像点道集(ADCIGS)建立了以剩余速度为自变量,剩余深度为目标函数的关系式,及目标函数的梯度公式.利用导出的两个公式分别对剩余深度与剩余速度的关系进行了定量分析.通过理论分析和模型试算证明初始速度模型的误差具有方向敏感性,即正误差较负误差对速度建模迭代收敛更敏感.利用此结论进行深度域速度建模既可以提高计算效率也可以提高建模精度.     

扩展成像条件下的最小二乘逆时偏移

逆时偏移(RTM)是复杂介质条件下地震成像的重要手段.因受观测系统限制、上覆地层影响以及波场带宽有限等因素的影响,现行的常规RTM所采用的互相关成像条件通常对地下构造进行模糊成像.最小二乘逆时偏移(LSRTM)通过最小化线性Born近似正演数据和采集数据之间的波形差异,采用梯度类反演算法优化反射系数模型,获得的成像结果具有更高的分辨率和更可靠的振幅保真度.然而,基于波形拟合的LSRTM对背景速度模型的依赖性很强.误差太大的速度模型容易产生周波跳跃现象,导致LSRTM难以获得全局最优解.为了克服这一问题,本文基于扩展模型的思想,在线性Born近似下,推导得到RTM扩展成像条件.并基于最小二乘反演理论,提出扩展成像条件下的LSRTM方法.理论模型试算表明,本文方法不仅可以提供分辨率更高、振幅属性更为可靠的成像结果,而且能够在一定程度上消除速度误差对反演成像的影响.     

地震多次波成像

地震资料含有各种类型多次波,而传统成像方法仅利用地震一次反射波成像,在地震成像前需将多次波去除.然而,多次波携带了丰富的地下结构信息,多次波偏移能够提供除反射波外的额外地下照明.修改传统逆时偏移方法,用包含一次反射波和多次波的原始记录代替震源子波,将SRME方法预测的表面多次波代替一次反射波作为输入数据,可将表面多次波成像.多次波成像的挑战和困难在于大量串扰噪声的产生,针对表面多次波成像中的成像噪声问题,将最小二乘逆时偏移方法与多次波分阶思想结合起来,发展可控阶数的表面多次波反演成像方法,有望初步实现高精度的表面多次波成像.在消除原始记录中的表面多次波后,通过逆散射级数方法预测得到层间多次波,将层间多次波作为逆时偏移方法的输入数据可将其准确归位到地下反射位置.数值实验表明,多次波成像能够有效地为地下提供额外照明,而可控阶表面多次波最小二乘逆时偏移成像方法几乎完全避免成像噪声.     

窄方位地震数据双平方根方程偏移方法探讨

简单回顾双平方根（DSR）方程偏移一般理论之后,介绍了适应偏移速度分析的角度域成像快速算法.针对地震勘探中窄方位三维地震数据成像问题,探讨了限定输入数据体或限定波场传播方向的DSR方程偏移方法.数值试验表明,共方位角偏移等DSR方程叠前深度偏移方法在盐体或礁体、古潜山等强横向变速地区具有很好的应用前景.     

3D post‐stack interval velocity analysis with effective use of datuming

Interval velocity analysis using post‐stack data has always been a desire, mainly for 3D data sets. In this study we present a method that uses the unique characteristics of migrated diffractions to enable interval velocity analysis from three‐dimensional zero‐offset time data. The idea is to perform a standard three‐dimensional prestack depth migration on stack cubes and generate three‐dimensional common image gathers that show great sensitivity to velocity errors. An efficient ‘top‐down’ scheme for updating the velocity is used to build the model. The effectiveness of the method is related to the incorporation of wave equation based post‐stack datuming in the model building process. The proposed method relies on the ability to identify diffractions along redatumed zero‐offset data and to analyse their flatness in the migrated local angle domain. The method can be considered as an additional tool for a complete, prestack depth migration based interval velocity analysis.     

Gabor‐frame‐based Gaussian packet migration

We present a Gaussian packet migration method based on Gabor frame decomposition and asymptotic propagation of Gaussian packets. A Gaussian packet has both Gaussian‐shaped time–frequency localization and space–direction localization. Its evolution can be obtained by ray tracing and dynamic ray tracing. In this paper, we first briefly review the concept of Gaussian packets. After discussing how initial parameters affect the shape of a Gaussian packet, we then propose two Gabor‐frame‐based Gaussian packet decomposition methods that can sparsely and accurately represent seismic data. One method is the dreamlet–Gaussian packet method. Dreamlets are physical wavelets defined on an observation plane and can represent seismic data efficiently in the local time–frequency space–wavenumber domain. After decomposition, dreamlet coefficients can be easily converted to the corresponding Gaussian packet coefficients. The other method is the Gabor‐frame Gaussian beam method. In this method, a local slant stack, which is widely used in Gaussian beam migration, is combined with the Gabor frame decomposition to obtain uniform sampled horizontal slowness for each local frequency. Based on these decomposition methods, we derive a poststack depth migration method through the summation of the backpropagated Gaussian packets and the application of the imaging condition. To demonstrate the Gaussian packet evolution and migration/imaging in complex models, we show several numerical examples. We first use the evolution of a single Gaussian packet in media with different complexities to show the accuracy of Gaussian packet propagation. Then we test the point source responses in smoothed varying velocity models to show the accuracy of Gaussian packet summation. Finally, using poststack synthetic data sets of a four‐layer model and the two‐dimensional SEG/EAGE model, we demonstrate the validity and accuracy of the migration method. Compared with the more accurate but more time‐consuming one‐way wave‐equation‐based migration, such as beamlet migration, the Gaussian packet method proposed in this paper can correctly image the major structures of the complex model, especially in subsalt areas, with much higher efficiency. This shows the application potential of Gaussian packet migration in complicated areas.     

Multipathing in three‐parameter common‐image gathers from reverse‐time migration

Reverse‐time migration gives high‐quality, complete images by using full‐wave extrapolations. It is thus not subject to important limitations of other migrations that are based on high‐frequency or one‐way approximations. The cross‐correlation imaging condition in two‐dimensional pre‐stack reverse‐time migration of common‐source data explicitly sums the product of the (forward‐propagating) source and (backward‐propagating) receiver wavefields over all image times. The primary contribution at any image point travels a minimum‐time path that has only one (specular) reflection, and it usually corresponds to a local maximum amplitude. All other contributions at the same image point are various types of multipaths, including prismatic multi‐arrivals, free‐surface and internal multiples, converted waves, and all crosstalk noise, which are imaged at later times, and potentially create migration artefacts. A solution that facilitates inclusion of correctly imaged, non‐primary arrivals and removal of the related artefacts, is to save the depth versus incident angle slice at each image time (rather than automatically summing them). This results in a three‐parameter (incident angle, depth, and image time) common‐image volume that integrates, into a single unified representation, attributes that were previously computed by separate processes. The volume can be post‐processed by selecting any desired combination of primary and/or multipath data before stacking over image time. Separate images (with or without artifacts) and various projections can then be produced without having to remigrate the data, providing an efficient tool for optimization of migration images. A numerical example for a simple model shows how primary and prismatic multipath contributions merge into a single incident angle versus image time trajectory. A second example, using synthetic data from the Sigsbee2 model, shows that the contributions to subsalt images of primary and multipath (in this case, turning wave) reflections are different. The primary reflections contain most of the information in regions away from the salt, but both primary and multipath data contribute in the subsalt region.