Full waveform inversion using oriented time‐domain imaging method for vertical transverse isotropic media

Full waveform inversion for reflection events is limited by its linearised update requirements given by a process equivalent to migration. Unless the background velocity model is reasonably accurate, the resulting gradient can have an inaccurate update direction leading the inversion to converge what we refer to as local minima of the objective function. In our approach, we consider mild lateral variation in the model and, thus, use a gradient given by the oriented time‐domain imaging method. Specifically, we apply the oriented time‐domain imaging on the data residual to obtain the geometrical features of the velocity perturbation. After updating the model in the time domain, we convert the perturbation from the time domain to depth using the average velocity. Considering density is constant, we can expand the conventional 1D impedance inversion method to two‐dimensional or three‐dimensional velocity inversion within the process of full waveform inversion. This method is not only capable of inverting for velocity, but it is also capable of retrieving anisotropic parameters relying on linearised representations of the reflection response. To eliminate the crosstalk artifacts between different parameters, we utilise what we consider being an optimal parametrisation for this step. To do so, we extend the prestack time‐domain migration image in incident angle dimension to incorporate angular dependence needed by the multiparameter inversion. For simple models, this approach provides an efficient and stable way to do full waveform inversion or modified seismic inversion and makes the anisotropic inversion more practicable. The proposed method still needs kinematically accurate initial models since it only recovers the high‐wavenumber part as conventional full waveform inversion method does. Results on synthetic data of isotropic and anisotropic cases illustrate the benefits and limitations of this method.     

苏北大陆科学钻探靶区深反射地震的叠前深度偏移

由于深反射地震数据具有信噪比低和记录长度长等特点,叠前深度偏移方法的应用有许多困难．为此,我们研究了一种适合于深反射地震的叠前深度偏移方法;包括：用逆风有限差分方法计算程函方程;在常规速度扫描的基础上,用协方差控制提高速度分析精度;用联合反演算法计算层速度,再插值后得到初始速度模型;用Ｋｉｒｃｈｈｏｆｆ法作为偏移速度分析工具,求得最终的速度模型;最终的速度模型作为有限差分深度偏移的输入,求得最终的偏移结果．用该方法对“中国大陆科学深钻工程”东海二维深反射地震数据ＤＨ－４线进行了叠前深度偏移,取得了良好的效果。     

The effectiveness of a pseudo-inverse extended Born operator to handle lateral heterogeneity for imaging and velocity analysis applications

Wave equation–based migration velocity analysis techniques aim to construct a kinematically accurate velocity model for imaging or as an initial model for full waveform inversion applications. The most popular wave equation–based migration velocity analysis method is differential semblance optimization, where the velocity model is iteratively updated by minimizing the unfocused energy in an extended image volume. However, differential semblance optimization suffers from artefacts, courtesy of the adjoint operator used in imaging, leading to poor convergence. Recent findings show that true amplitude imaging plays a significant role in enhancing the differential semblance optimization's gradient and reducing the artefacts. Here, we focus on a pseudo-inverse operator to the horizontally extended Born as a true amplitude imaging operator. For laterally inhomogeneous models, the operator required a derivative with respect to a vertical shift. Extending the image vertically to evaluate such a derivative is costly and impractical. The inverse operator can be simplified in laterally homogeneous models. We derive an extension of the approach to apply the full inverse formula and evaluate the derivative efficiently. We simplified the implementation by applying the derivative to the imaging condition and utilize the relationship between the source and receiver wavefields and the vertical shift. Specifically, we verify the effectiveness of the approach using the Marmousi model and show that the term required for the lateral inhomogeneity treatment has a relatively small impact on the results for many cases. We then apply the operator in differential semblance optimization and invert for an accurate macro-velocity model, which can serve as an initial velocity model for full waveform inversion.     

Migration velocity analysis using pre‐stack wave fields

Using both image and data domains to perform velocity inversion can help us resolve the long and short wavelength components of the velocity model, usually in that order. This translates to integrating migration velocity analysis into full waveform inversion. The migration velocity analysis part of the inversion often requires computing extended images, which is expensive when using conventional methods. As a result, we use pre‐stack wavefield (the double‐square‐root formulation) extrapolation, which includes the extended information (subsurface offsets) naturally, to make the process far more efficient and stable. The combination of the forward and adjoint pre‐stack wavefields provides us with update options that can be easily conditioned to improve convergence. We specifically use a modified differential semblance operator to split the extended image into a residual part for classic differential semblance operator updates and the image (Born) modelling part, which provides reflections for higher resolution information. In our implementation, we invert for the velocity and the image simultaneously through a dual objective function. Applications to synthetic examples demonstrate the features of the approach.     

Wave-equation migration velocity analysis. I. Theory

We present a migration velocity analysis (MVA) method based on wavefield extrapolation. Similarly to conventional MVA, our method aims at iteratively improving the quality of the migrated image, as measured by the flatness of angle‐domain common‐image gathers (ADCIGs) over the aperture‐angle axis. However, instead of inverting the depth errors measured in ADCIGs using ray‐based tomography, we invert ‘image perturbations’ using a linearized wave‐equation operator. This operator relates perturbations of the migrated image to perturbations of the migration velocity. We use prestack Stolt residual migration to define the image perturbations that maximize the focusing and flatness of ADCIGs. Our linearized operator relates slowness perturbations to image perturbations, based on a truncation of the Born scattering series to the first‐order term. To avoid divergence of the inversion procedure when the velocity perturbations are too large for Born linearization of the wave equation, we do not invert directly the image perturbations obtained by residual migration, but a linearized version of the image perturbations. The linearized image perturbations are computed by a linearized prestack residual migration operator applied to the background image. We use numerical examples to illustrate how the backprojection of the linearized image perturbations, i.e. the gradient of our objective function, is well behaved, even in cases when backprojection of the original image perturbations would mislead the inversion and take it in the wrong direction. We demonstrate with simple synthetic examples that our method converges even when the initial velocity model is far from correct. In a companion paper, we illustrate the full potential of our method for estimating velocity anomalies under complex salt bodies.     

Correlation‐based reflection waveform inversion by one‐way wave equations

Reflection full waveform inversion can update subsurface velocity structure of the deeper part, but tends to get stuck in the local minima associated with the waveform misfit function. These local minima cause cycle skipping if the initial background velocity model is far from the true model. Since conventional reflection full waveform inversion using two‐way wave equation in time domain is computationally expensive and consumes a large amount of memory, we implement a correlation‐based reflection waveform inversion using one‐way wave equations to retrieve the background velocity. In this method, one‐way wave equations are used for the seismic wave forward modelling, migration/de‐migration and the gradient computation of objective function in frequency domain. Compared with the method using two‐way wave equation, the proposed method benefits from the lower computational cost of one‐way wave equations without significant accuracy reduction in the cases without steep dips. It also largely reduces the memory requirement by an order of magnitude than implementation using two‐way wave equation both for two‐ and three‐dimensional situations. Through numerical analysis, we also find that one‐way wave equations can better construct the low wavenumber reflection wavepath without producing high‐amplitude short‐wavelength components near the image points in the reflection full waveform inversion gradient. Synthetic test and real data application show that the proposed method efficiently updates the background velocity model.     

Adaptation of prestack migration to multi-offset ground-penetrating radar (GPR) data

In adapting the prestack migration technique used in seismic imaging to the inversion of ground‐penetrating radar (GPR) from time‐ to depth‐sections, we show that the theoretical integral formulation of the inversion can be applied to electromagnetic problems, albeit with three assumptions. The first two assumptions concern the electromagnetic characteristics of the medium, primarily that the medium must be perfectly resistive and non‐dispersive, and the third concerns the antennae radiation pattern, which is taken to be 2D. The application of this adaptation of the inversion method is confirmed by migrating actual GPR measurements acquired on the test site of the Laboratoire Central des Ponts et Chaussées. The results show good agreement with the geometry of the structures in the medium and confirm that the possible departure from the assumption of a purely resistive medium has no visible effect on the information concerning the geometry of scattering and reflecting structures. The field experiments also show that prestack migration processing is sufficiently robust with regard to the assumption of a non‐dispersive medium. The assumption of a 2D antennae radiation pattern, however, produces artefacts that could be significant for laterally heterogeneous media. Nevertheless, where the medium is not highly laterally heterogeneous, the migration gives a clear image of the scattering potential due to the geometry of structural contrasts in the medium; the scatterers are well focused from diffraction hyperbolae and well localized. Spatial geometry has limited dimensional accuracy and positions are located with a maximum error equal to the minimum wavelength of the signal bandpass. Objects smaller than one wavelength can nevertheless be detected and well focused if their dielectric contrasts are sufficiently high, as in the case of iron or water in gneiss gravels. Furthermore, the suitability of multi‐offset protocols to estimate the electromagnetic propagating velocity and to decrease the non‐coherent noise level of measurements is confirmed. Our velocity estimation is based on the semblance calculation of multi‐offset migrated images, and we confirmed the relevance of this quantification method using numerical data. The signal‐to‐noise ratio is improved by summing multi‐offset results after the addition of random noise on measurements. Thus the adaptation of prestack migration to multi‐offset radar measurements significantly improves the resolution of the scattering potential of the medium. Limitations associated with the methods used here suggest that 3D algorithms should be applied to strongly laterally heterogeneous media and further studies concerning the waveform inversion are necessary to obtain information about the electric nature of the medium.     

Improving the gradient of the image‐domain objective function using quantitative migration for a more robust migration velocity analysis

Migration velocity analysis aims at determining the background velocity model. Classical artefacts, such as migration smiles, are observed on subsurface offset common image gathers, due to spatial and frequency data limitations. We analyse their impact on the differential semblance functional and on its gradient with respect to the model. In particular, the differential semblance functional is not necessarily minimum at the expected value. Tapers are classically applied on common image gathers to partly reduce these artefacts. Here, we first observe that the migrated image can be defined as the first gradient of an objective function formulated in the data‐domain. For an automatic and more robust formulation, we introduce a weight in the original data‐domain objective function. The weight is determined such that the Hessian resembles a Dirac function. In that way, we extend quantitative migration to the subsurface‐offset domain. This is an automatic way to compensate for illumination. We analyse the modified scheme on a very simple 2D case and on a more complex velocity model to show how migration velocity analysis becomes more robust.     

Image‐ray Tomography

Seismic tomography is a well‐established approach to invert smooth macro‐velocity models from kinematic parameters, such as traveltimes and their derivatives, which can be directly estimated from data. Tomographic methods differ more with respect to data domains than in the specifications of inverse‐problem solving schemes. Typical examples are stereotomography, which is applied to prestack data and Normal‐Incidence‐Point‐wave tomography, which is applied to common midpoint stacked data. One of the main challenges within the tomographic approach is the reliable estimation of the kinematic attributes from the data that are used in the inversion process. Estimations in the prestack domain (weak and noisy signals), as well as in the post‐stack domain (occurrence of triplications and diffractions leading to numerous conflicting dip situations) may lead to parameter inaccuracies that will adversely impact the resulting velocity models. To overcome the above limitations, a new tomographic procedure applied in the time‐migrated domain is proposed. We call this method Image‐Incident‐Point‐wave tomography. The new scheme can be seen as an alternative to Normal‐Incidence‐Point‐wave tomography. The latter method is based on traveltime attributes associated with normal rays, whereas the Image‐Incidence‐Point‐wave technique is based on the corresponding quantities for the image rays. Compared to Normal‐Incidence‐Point‐wave tomography the proposed method eases the selection of the tomography attributes, which is shown by synthetic and field data examples. Moreover, the method provides a direct way to convert time‐migration velocities into depth‐migration velocities without the need of any Dix‐style inversion.     

Quantitative imaging of complex structures from dense wide-aperture seismic data by multiscale traveltime and waveform inversions: a case study

An integrated multiscale seismic imaging flow is applied to dense onshore wide‐aperture seismic data recorded in a complex geological setting (thrust belt). An initial P‐wave velocity macromodel is first developed by first‐arrival traveltime tomography. This model is used as an initial guess for subsequent full‐waveform tomography, which leads to greatly improved spatial resolution of the P‐wave velocity model. However, the application of full‐waveform tomography to the high‐frequency part of the source bandwidth is difficult, due to the non‐linearity of this kind of method. Moreover, it is computationally expensive at high frequencies since a finite‐difference method is used to model the wave propagation. Hence, full‐waveform tomography was complemented by asymptotic prestack depth migration to process the full‐source bandwidth and develop a sharp image of the short wavelengths. The final traveltime tomography model and two smoothed versions of the final full‐waveform tomography model were used as a macromodel for the prestack depth migration. In this study, wide‐aperture multifold seismic data are used. After specific preprocessing of the data, 16 frequency components ranging from 5.4 Hz to 20 Hz were inverted in cascade by the full‐waveform tomography algorithm. The full‐waveform tomography successfully imaged SW‐dipping structures previously identified as high‐resistivity bodies. The relevance of the full‐waveform tomography models is demonstrated locally by comparison with a coincident vertical seismic profiling (VSP) log available on the profile. The prestack depth‐migrated images, inferred from the traveltime, and the smoothed full‐waveform tomography macromodels are shown to be, on the whole, consistent with the final full‐waveform tomography model. A more detailed analysis, based on common‐image gather computations, and local comparison with the VSP log revealed that the most accurate migrated sections are those obtained from the full‐waveform tomography macromodels. A resolution analysis suggests that the asymptotic prestack depth migration successfully migrated the wide‐aperture components of the data, allowing medium wavelengths in addition to the short wavelengths of the structure to be imaged. The processing flow that we applied to dense wide‐aperture seismic data is shown to provide a promising approach, complementary to more classical seismic reflection data processing, to quantitative imaging of complex geological structures.     

Seismic velocity analysis in the scattering-angle/azimuth domain

Migration velocity analysis is carried out by analysing the residual moveout and amplitude variations in common image point gathers (CIGs) parametrized by scattering angle and azimuth. The misfit criterion in the analysis is of the differential-semblance type. By using angles to parametrize the imaging we are able to handle and exploit data with multiple arrivals, although artefacts may occur in the CIGs and need to be suppressed. The CIGs are generated by angle migration, an approach based on the generalized Radon transform (GRT) inversion, and they provide multiple images of reflectors in the subsurface for a range of scattering angles and azimuths. Within the differential semblance applied to these CIGs, we compensate for amplitude versus angle (AVA) effects. Thus, using a correct background velocity model, the CIGs should have no residual moveout nor amplitude variation with angles, and the differential semblance should vanish. If the velocity model is incorrect, however, the events in the CIGs will appear at different depths for different angles and the amplitude along the events will be non-uniform. A standard, gradient-based optimization scheme is employed to develop a velocity updating procedure. The model update is formed by backprojecting the differential semblance misfits through ray perturbation kernels, within a GRT inverse. The GRT inverse acts on the data, subject to a shift in accordance with ray perturbation theory. The performance of our algorithm is demonstrated with two synthetic data examples using isotropic elastic models. The first one allows velocity variation with depth only. In the second one, we reconstruct a low-velocity lens in the model that gives rise to multipathing. The velocity model parametrization is based upon the eigentensor decomposition of the stiffness tensor and makes use of B-splines.     

Depth and morphology of reflectors from the non-linear inversion of arrival times and waveform semblance data. Part II: modelling and interpretation of real data acquired in Southern Apennines, Italy

In order to retrieve a 2D background velocity model and to retrieve the geometry and depth of shallow crustal reflectors in the Southern Apennines thrust belt a separate inversion of first arrival traveltimes and reflected waveforms was performed. Data were collected during an active seismic experiment in 1999 by Enterprise Oil Italiana and Eni-Agip using a global offset acquisition geometry. A total of 284 on-land shots were recorded by 201 receivers deployed on an 18 km line oriented SW–NE in the Val D'Agri region (Southern Apennines, Italy).
The two-step procedure allows for the retrieval of a reliable velocity model by using a non-linear tomographic inversion and reflected waveform semblance data inversion. The tomographic model shows that the P wave velocity field varies vertically from approximately 3 km/s to 6 km/s within 4 km from the Earth's surface. Moreover, at a distance of approximately 11 km along the profile, there is an abrupt increase in the velocity field. In this zone indeed, an ascent from 2 km depth to 0 km above sea level of the 5.2 km/s iso-velocity contour can be noted. The retrieved velocity can be associated with Plio-Pleistocene clastic deposits outcropping in the basin zone and with Mesozoic limestone deposits. The inversion of waveform semblance data shows that a P-to-P reflector is retrieved at a depth of approximately 2 km. This interface is deeper in the north-eastern part of the profile, where it reaches 3 km depth and can be associated with a limestone horizon.     

TI介质局部角度域高斯束叠前深度偏移成像

各向异性射线理论基础上的局部角度域叠前深度偏移方法能够为深度域构造成像与基于角道集的层析反演提供有力支撑,但是对于复杂地质构造而言,高斯度叠前深度偏移在不失高效、灵活等特点的情况下,具有明显的精度优势.为此,本文研究局部角度域理论框架下的高斯束叠前深度偏移方法.为提高算法效率与实用性,文中讨论了一种从经典弹性参数表征的各向异性介质运动学和动力学射线方程演变而来的由相速度表征的简便形式,并提出了一种比较经济的各向异性高斯束近似合成方案.结合地震波局部角度域成像原理,讨论一种适合高斯束偏移的角度参数计算方法.国际上通用的理论模型合成数据试验表明：相比局部角度域Kirchhoff叠前深度偏移成像方法,本文方法具有更高的成像精度与抗噪能力,既适用于复杂构造成像,也可为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.     

双平方根单程波动方程叠前τ偏移方法

本文将常规双平方根（DSR）单程波动方程从深度域变换到双程垂直走时（τ）域，由此推导出可从数学上实现“沉降观测”的单程波DSR传播算子. 其递归波场延拓算法包含波数域针对常速背景的相移处理和空间域针对横向速度扰动的相位校正，可以应对上覆地层速度横向变化对构造成像的影响. 结合零炮检距、零时间成像条件，提出了在τ域进行波场延拓与成像的DSR方程叠前偏移新方法. 为了克服其全三维偏移算法在实际应用中可能面临的困难，本文采用稳相近似，在crossline常炮检距偏移理论基础上推导了实用的共方位角叠前τ偏移方法. 数值试验表明，DSR方程叠前τ偏移在强横向非均匀介质中的成像精度与分辨率优于传统的时间域成像技术.     

Preserved‐amplitude angle domain migration by shot‐receiver wavefield continuation

We present preserved‐amplitude downward continuation migration formulas in the aperture angle domain. Our approach is based on shot‐receiver wavefield continuation. Since source and receiver points are close to the image point, a local homogeneous reference velocity can be approximated after redatuming. We analyse this approach in the framework of linearized inversion of Kirchhoff and Born approximations. From our analysis, preserved‐amplitude Kirchhoff and Born inverse formulas can be derived for the 2D case. They involve slant stacks of filtered subsurface offset domain common image gathers followed by the application of the appropriate weighting factors. For the numerical implementation of these formulas, we develop an algorithm based on the true amplitude version of the one‐way paraxial approximation. Finally, we demonstrate the relevance of our approach with a set of applications on synthetic datasets and compare our results with those obtained on the Marmousi model by multi‐arrival ray‐based preserved‐amplitude migration. While results are similar, we observe that our results are less affected by artefacts.     

Tomographic velocity inversion for ADCIGs in areas with a rugged surface

Pre-stack depth migration velocity analysis is one of the key techniques influencing image quality. As for areas with a rugged surface and complex subsurface, conventional prestack depth migration velocity analysis corrects the rugged surface to a known datum or designed surface velocity model on which to perform migration and update the velocity. We propose a rugged surface tomographic velocity inversion method based on angle-domain common image gathers by which the velocity field can be updated directly from the rugged surface without static correction for pre-stack data and improve inversion precision and efficiency. First, we introduce a method to acquire angle-domain common image gathers (ADCIGs) in rugged surface areas and then perform rugged surface tomographic velocity inversion. Tests with model and field data prove the method to be correct and effective.     

Time‐lapse seismic imaging using regularized full‐waveform inversion with a prior model: which strategy?

Full‐waveform inversion is an appealing technique for time‐lapse imaging, especially when prior model information is included into the inversion workflow. Once the baseline reconstruction is achieved, several strategies can be used to assess the physical parameter changes, such as parallel difference (two separate inversions of baseline and monitor data sets), sequential difference (inversion of the monitor data set starting from the recovered baseline model) and double‐difference (inversion of the difference data starting from the recovered baseline model) strategies. Using synthetic Marmousi data sets, we investigate which strategy should be adopted to obtain more robust and more accurate time‐lapse velocity changes in noise‐free and noisy environments. This synthetic application demonstrates that the double‐difference strategy provides the more robust time‐lapse result. In addition, we propose a target‐oriented time‐lapse imaging using regularized full‐waveform inversion including a prior model and model weighting, if the prior information exists on the location of expected variations. This scheme applies strong prior model constraints outside of the expected areas of time‐lapse changes and relatively less prior constraints in the time‐lapse target zones. In application of this process to the Marmousi model data set, the local resolution analysis performed with spike tests shows that the target‐oriented inversion prevents the occurrence of artefacts outside the target areas, which could contaminate and compromise the reconstruction of the effective time‐lapse changes, especially when using the sequential difference strategy. In a strongly noisy case, the target‐oriented prior model weighting ensures the same behaviour for both time‐lapse strategies, the double‐difference and the sequential difference strategies and leads to a more robust reconstruction of the weak time‐lapse changes. The double‐difference strategy can deliver more accurate time‐lapse variation since it can focus to invert the difference data. However, the double‐difference strategy requires a preprocessing step on data sets such as time‐lapse binning to have a similar source/receiver location between two surveys, while the sequential difference needs less this requirement. If we have prior information about the area of changes, the target‐oriented sequential difference strategy can be an alternative and can provide the same robust result as the double‐difference strategy.     

3D acoustic waveform inversion in the Laplace domain using an iterative solver

In this paper we propose a 3D acoustic full waveform inversion algorithm in the Laplace domain. The partial differential equation for the 3D acoustic wave equation in the Laplace domain is reformulated as a linear system of algebraic equations using the finite element method and the resulting linear system is solved by a preconditioned conjugate gradient method. The numerical solutions obtained by our modelling algorithm are verified through a comparison with the corresponding analytical solutions and the appropriate dispersion analysis. In the Laplace‐domain waveform inversion, the logarithm of the Laplace transformed wavefields mainly contains long‐wavelength information about the underlying velocity model. As a result, the algorithm smoothes a small‐scale structure but roughly identifies large‐scale features within a certain depth determined by the range of offsets and Laplace damping constants employed. Our algorithm thus provides a useful complementary process to time‐ or frequency‐domain waveform inversion, which cannot recover a large‐scale structure when low‐frequency signals are weak or absent. The algorithm is demonstrated on a synthetic example: the SEG/EAGE 3D salt‐dome model. The numerical test is limited to a Laplace‐domain synthetic data set for the inversion. In order to verify the usefulness of the inverted velocity model, we perform the 3D reverse time migration. The migration results show that our inversion results can be used as an initial model for the subsequent high‐resolution waveform inversion. Further studies are needed to perform the inversion using time‐domain synthetic data with noise or real data, thereby investigating robustness to noise.