High-order generalized screen propagator migration based on particle swarm optimization

Various migration methods have been proposed to image high-angle geological structures and media with strong lateral velocity variations; however, the problems of low precision and high computational cost remain unresolved. To describe the seismic wave propagation in media with lateral velocity variations and to image high-angle structures, we propose the generalized screen propagator based on particle swarm optimization (PSO-GSP), for the precise fitting of the single-square-root operator. We use the 2D SEG/EAGE salt model to test the proposed PSO-GSP migration method to image the faults beneath the salt dome and compare the results to those of the conventional high-order generalized screen propagator (GSP) migration and split-step Fourier (SSF) migration. Moreover, we use 2D marine data from the South China Sea to show that the PSO-GSP migration can better image strong reflectors than conventional imaging methods.     

波动方程的高阶广义屏叠前深度偏移

不同于常规广义屏传播算子的推导中使用散射理论,本文利用单平方根算子的渐近展开,推导出了单程波方程广义屏传播算子的高阶表达式.高阶广义屏传播算子不仅可提高常规广义屏传播算子的计算精度,而且还能改善广义屏传播算子对速度强横向变化介质的适应性.把高阶广义屏传播算子应用于波动方程叠前深度偏移,可得到比常规广义屏传播算子更好的效果.高阶广义屏传播算子的阶数越高,计算精度越高,但计算量也越多.以SEG EAGE二维盐丘模型数据的波动方程叠前深度偏移为例,二阶广义屏传播算子相对于常规（一阶）广义屏传播算子增加了30%的计算量.高阶广义屏传播算子是常规广义屏传播算子理论的发展和完善.     

A comparison of continuous mass‐lumped finite elements with finite differences for 3‐D wave propagation

The finite‐difference method on rectangular meshes is widely used for time‐domain modelling of the wave equation. It is relatively easy to implement high‐order spatial discretization schemes and parallelization. Also, the method is computationally efficient. However, the use of finite elements on tetrahedral unstructured meshes is more accurate in complex geometries near sharp interfaces. We compared the standard eighth‐order finite‐difference method to fourth‐order continuous mass‐lumped finite elements in terms of accuracy and computational cost. The results show that, for simple models like a cube with constant density and velocity, the finite‐difference method outperforms the finite‐element method by at least an order of magnitude. Outside the application area of rectangular meshes, i.e., for a model with interior complexity and topography well described by tetrahedra, however, finite‐element methods are about two orders of magnitude faster than finite‐difference methods, for a given accuracy.     

Wavefield interpolation in 3D large‐step Fourier wavefield extrapolation

Extrapolating wavefields and imaging at each depth during three‐dimensional recursive wave‐equation migration is a time‐consuming endeavor. For efficiency, most commercial techniques extrapolate wavefields through thick slabs followed by wavefield interpolation within each thick slab. In this article, we develop this strategy by associating more efficient interpolators with a Fourier‐transform‐related wavefield extrapolation method. First, we formulate a three‐dimensional first‐order separation‐of‐variables screen propagator for large‐step wavefield extrapolation, which allows for wide‐angle propagations in highly contrasting media. This propagator significantly improves the performance of the split‐step Fourier method in dealing with significant lateral heterogeneities at the cost of only one more fast Fourier transform in each thick slab. We then extend the two‐dimensional Kirchhoff and Born–Kirchhoff local wavefield interpolators to three‐dimensional cases for each slab. The three‐dimensional Kirchhoff interpolator is based on the traditional Kirchhoff formula and applies to moderate lateral velocity variations, whereas the three‐dimensional Born–Kirchhoff interpolator is derived from the Lippmann–Schwinger integral equation under the Born approximation and is adapted to highly laterally varying media. Numerical examples on the three‐dimensional salt model of the Society of Exploration Geophysicists/European Association of Geoscientists demonstrate that three‐dimensional first‐order separation‐of‐variables screen propagator Born–Kirchhoff depth migration using thick‐slab wavefield extrapolation plus thin‐slab interpolation tolerates a considerable depth‐step size of up to 72 ms, eventually resulting in an efficiency improvement of nearly 80% without obvious loss of imaging accuracy. Although the proposed three‐dimensional interpolators are presented with one‐way Fourier extrapolation methods, they can be extended for applications to general migration methods.     

3D Fourier finite-difference migration by alternating-direction-implicit plus interpolation

Conventional two‐way splitting Fourier finite‐difference migration for 3D complex media yields azimuthal anisotropy where an additional phase correction is needed with much increase of computational cost. We incorporate the alternating‐direction‐implicit plus interpolation scheme into the conventional Fourier finite‐difference method to reduce azimuthal anisotropy. This scheme retains the high‐order remnants ignored by the two‐way splitting in the form of a wavefield interpolation in the wavenumber domain. The wavefield interpolation for each step of downward extrapolation is implemented between the wavefields before and after the conventional Fourier finite‐difference extrapolation. As the Fourier finite‐difference migration is implemented in the space and wavenumber dual space, the Fourier transforms between space and wavenumber domain that were needed for the alternating‐direction‐implicit plus interpolation in frequency domain (FD) migration are saved in Fourier finite‐difference migration. Since the azimuth anisotropy in Fourier finite‐difference is much less than that in FD, the application of the alternating‐direction‐implicit plus interpolation scheme in Fourier finite‐difference migration is superior to that in FD migration in handling complex media with large velocity contrasts and steep dips. Impulse responses show that the presented method reduces the azimuthal anisotropy at almost no extra cost.     

Broadband constant-coefficient propagators

The phase error between the real phase shift and the Gazdag background phase shift, due to lateral velocity variations about a reference velocity, can be decomposed into axial and paraxial phase errors. The axial phase error depends only on velocity perturbations and hence can be completely removed by the split‐step Fourier method. The paraxial phase error is a cross function of velocity perturbations and propagation angles. The cross function can be approximated with various differential operators by allowing the coefficients to vary with velocity perturbations and propagation angles. These variable‐coefficient operators require finite‐difference numerical implementation. Broadband constant‐coefficient operators may provide an efficient alternative that approximates the cross function within the split‐step framework and allows implementation using Fourier transforms alone. The resulting migration accuracy depends on the localization of the constant‐coefficient operators. A simple broadband constant‐coefficient operator has been designed and is tested with the SEG/EAEG salt model. Compared with the split‐step Fourier method that applies to either weak‐contrast media or at small propagation angles, this operator improves wavefield extrapolation for large to strong lateral heterogeneities, except within the weak‐contrast region. Incorporating the split‐step Fourier operator into a hybrid implementation can eliminate the poor performance of the broadband constant‐coefficient operator in the weak‐contrast region. This study may indicate a direction of improving the split‐step Fourier method, with little loss of efficiency, while allowing it to remain faster than more precise methods such as the Fourier finite‐difference method.     

A central difference method with low numerical dispersion for solving the scalar wave equation

In this paper, we propose a nearly‐analytic central difference method, which is an improved version of the central difference method. The new method is fourth‐order accurate with respect to both space and time but uses only three grid points in spatial directions. The stability criteria and numerical dispersion for the new scheme are analysed in detail. We also apply the nearly‐analytic central difference method to 1D and 2D cases to compute synthetic seismograms. For comparison, the fourth‐order Lax‐Wendroff correction scheme and the fourth‐order staggered‐grid finite‐difference method are used to model acoustic wavefields. Numerical results indicate that the nearly‐analytic central difference method can be used to solve large‐scale problems because it effectively suppresses numerical dispersion caused by discretizing the scalar wave equation when too coarse grids are used. Meanwhile, numerical results show that the minimum sampling rate of the nearly‐analytic central difference method is about 2.5 points per minimal wavelength for eliminating numerical dispersion, resulting that the nearly‐analytic central difference method can save greatly both computational costs and storage space as contrasted to other high‐order finite‐difference methods such as the fourth‐order Lax‐Wendroff correction scheme and the fourth‐order staggered‐grid finite‐difference method.     

一阶弹性波方程交错网格高阶差分解法

提高计算精度和运算效率是所有波场正演方法所追求的目标,本文通过将速度 （应力）对时间的奇数阶高阶寻数转化为应力（速度）对空间的导数,运用时间和空间差分精度 均可达任意阶的高阶差分法,通过交错网格技术,对一阶速度－应力弹性波方程进行了数值求 解．波场快照以及实际模型的正演结果表明,这种求解一阶弹性波方程的高阶差分解法,和 常规的差分法相比网格频散显著减小,精度明显提高,而且可以取较大的空间步长,提高计算 效率。     

基于多参量的模拟退火全局优化傅里叶有限差分算子

傅里叶有限差分法(FFD)能够处理复杂地质构造中的波传播问题,但对陡倾角成像仍有明显的误差.优化参数的方法能够在保持计算效率的前提下进一步提高陡倾角的成像精度.本文在有理近似的基础上,将FFD算子展开式中的常系数由两个拓展为四个,然后采用模拟退火算法对这四个参数进行全局优化.本方法除了考虑速度对比度以外,还考虑了频率和延拓步长等参量的影响.理论误差分析和脉冲响应测试均表明该方法能极大地提高FFD算子的精确传播角度.二维SEG/EAGE盐丘模型实验表明本文方法对陡倾角以及盐下构造的成像精度明显高于未优化的FFD法.将本文的方法与交替方向加插值的方法结合应用于三维脉冲响应测试更进一步证实了本文方法的有效性.     

Pure acoustic wave propagation in transversely isotropic media by the pseudospectral method

Seismic wave propagation in transversely isotropic (TI) media is commonly described by a set of coupled partial differential equations, derived from the acoustic approximation. These equations produce pure P‐wave responses in elliptically anisotropic media but generate undesired shear‐wave components for more general TI anisotropy. Furthermore, these equations suffer from instabilities when the anisotropy parameter ε is less than δ. One solution to both problems is to use pure acoustic anisotropic wave equations, which can produce pure P‐waves without any shear‐wave contaminations in both elliptical and anelliptical TI media. In this paper, we propose a new pure acoustic transversely isotropic wave equation, which can be conveniently solved using the pseudospectral method. Like most other pure acoustic anisotropic wave equations, our equation involves complicated pseudo‐differential operators in space which are difficult to handle using the finite difference method. The advantage of our equation is that all of its model parameters are separable from the spatial differential and pseudo‐differential operators; therefore, the pseudospectral method can be directly applied. We use phase velocity analysis to show that our equation, expressed in a summation form, can be properly truncated to achieve the desired accuracy according to anisotropy strength. This flexibility allows us to save computational time by choosing the right number of summation terms for a given model. We use numerical examples to demonstrate that this new pure acoustic wave equation can produce highly accurate results, completely free from shear‐wave artefacts. This equation can be straightforwardly generalized to tilted TI media.     

自适应空间分区裂步傅立叶偏移方法

本文提出了一种新的偏移方法——自适应空间分区裂步傅立叶（ASDSSF）偏移方法。该方法将剥层相位移方法的思想推广到裂步傅立叶偏移方法，使之当速度场出现强间断时也能精确而有效地成像。原理上ASDSSF偏移属于多参考慢度（MRS）偏移方法，本文的重点是，在不损失精度的同时选取比同类MRS偏移方法更少的参考慢度。我们根据全局速度函数的变化和误差控制参数来选择参考慢度，同一个参考慢度所对应的速度构成一个分区，每一个空间分区可以由几个空间上不连续的子分区组成，从而有效地减小了参考慢度的个数。每一延拓步的参考慢度以及参考慢度的个数和如何构建分区都是根据速度函数自动生成，因此更为合理。为了消除速度场强间断产生的人为噪音，设计了简单有效的f-k域的光滑滤波。我们对一个生成的二维叠前模型和SEG／EAEG盐丘模型进行了试算。     

Accuracy‐constrained optimisation methods for staggered‐grid elastic wave modelling

The classical finite‐difference methods for seismic wave modelling are very accurate at low wavenumbers but suffer from inaccuracies at high wavenumbers, particularly at Nyquist wavenumber. In contrast, the optimisation finite‐difference methods reduce inaccuracies at high wavenumbers but suffer from inaccuracies at low wavenumbers, particularly at zero wavenumber when the operator length is not long and the whole range of wavenumbers is considered. Inaccuracy at zero wavenumber means that the optimisation methods only have a zeroth‐order accuracy of truncation and thus are not rigorously convergent. To guarantee the rigorous convergence of the optimisation methods, we have developed accuracy‐constrained optimisation methods. Different‐order accuracy‐constrained optimisation methods are presented. These methods not only guarantee the rigorous convergence but also reduce inaccuracies at low wavenumbers. Accuracy‐constrained optimisation methods are applied to staggered‐grid elastic wave modelling.     

Optimal implicit staggered‐grid finite‐difference schemes based on the sampling approximation method for seismic modelling

We propose new implicit staggered‐grid finite‐difference schemes with optimal coefficients based on the sampling approximation method to improve the numerical solution accuracy for seismic modelling. We first derive the optimized implicit staggered‐grid finite‐difference coefficients of arbitrary even‐order accuracy for the first‐order spatial derivatives using the plane‐wave theory and the direct sampling approximation method. Then, the implicit staggered‐grid finite‐difference coefficients based on sampling approximation, which can widen the range of wavenumber with great accuracy, are used to solve the first‐order spatial derivatives. By comparing the numerical dispersion of the implicit staggered‐grid finite‐difference schemes based on sampling approximation, Taylor series expansion, and least squares, we find that the optimal implicit staggered‐grid finite‐difference scheme based on sampling approximation achieves greater precision than that based on Taylor series expansion over a wider range of wavenumbers, although it has similar accuracy to that based on least squares. Finally, we apply the implicit staggered‐grid finite difference based on sampling approximation to numerical modelling. The modelling results demonstrate that the new optimal method can efficiently suppress numerical dispersion and lead to greater accuracy compared with the implicit staggered‐grid finite difference based on Taylor series expansion. In addition, the results also indicate the computational cost of the implicit staggered‐grid finite difference based on sampling approximation is almost the same as the implicit staggered‐grid finite difference based on Taylor series expansion.     

Lateral velocity variation related correction in asymptotic true‐amplitude one‐way propagators

It was mathematically proved that the asymptotic true‐amplitude one‐way wave equation could provide the same amplitude as the full‐wave equation in heterogeneous lossless media in the sense of high‐frequency asymptotics. Much work has been done on the vertical velocity variation related amplitude correction term but the lateral velocity variation related term has not received much attention, even being excluded in some asymptotic true‐amplitude one‐way propagator formulations. Here we analyse the effects of different amplitude correction terms in the asymptotic true‐amplitude one‐way propagator, especially the effect related to the lateral velocity variation, by comparing the wavefield amplitude from the one‐way propagator with that from full‐wave modelling. We derive a dual‐domain wide‐angle screen type asymptotic true‐amplitude one‐way propagator and evaluate two implementations of the amplitude correction. Numerical examples show that the lateral velocity variation related correction term can play a significant role in the asymptotic true‐amplitude one‐way propagator. Optimization of the expansion coefficients in the asymptotic true‐amplitude one‐way propagator can improve both the amplitude and phase accuracy for wide‐angle waves.     

基于波动方程的广义屏叠前深度偏移

地震波传播算子的计算效率和精度是制约三维叠前深度偏移的关键因素. 广义屏传播算子(GSP, Generalized Screen Propagator)是一种在双域中实现的广角单程波传播算子. 这一方法略去了在非均匀体之间发生的交混回响,但它可以正确处理包括聚焦、衍射、折射和干涉在内的各种多次前向散射现象. 通过背景速度下的相移和扰动速度下的陡倾角校正,广义屏算子能够适应地层速度的强烈横向变化. 这种算子可以直接应用于炮集叠前偏移,通过将广义屏算子作用于双平方根方程,还可以获得一种高效率、高精度的炮检距域叠前深度偏移方法,用于二维共炮检距道集和三维共方位角道集的深度域成像. 本文首先简述了炮检距域广义屏传播算子的理论,进而讨论了共照射角成像(CAI, Common Angle Imaging)条件,由此给出各个不同照射角(炮检距射线参数)下的成像结果,进而得到共照射角像集. 由于照射角和炮检距的对应关系,共照射角像集又为偏移速度分析和AVO(振幅随炮检距变化)分析等提供了有力工具.     

两种类型共炮点域相关型叠前深度偏移方法分析

研究不同偏移方法的成像机理是实现复杂构造高精度成像的前提,研究了两类共炮点域的相关型叠前深度偏移成像方法：基于单程波动方程的叠前深度成像方法和基于双程波动方程的叠前深度成像方法,同时对比了它们在计算效率、数据存储量、成像精度、成像机理、速度敏感性等方面的差异及其共性。以复杂构造模型为例,采用了傅里叶有限差分法（FFD）和逆时成像法（RTM）,这两种方法实现了121个共炮点道集的叠前深度偏移成像处理。计算结果表明,当速度准确时,两种深度域波动方程成像方法均可以恢复出各个地质反射界面,其中逆时偏移对陡倾角成像效果显著,当速度存在百分比误差和随机扰动情况时,逆时成像结果要差于单程波方法,因此,逆时偏移方法对速度的敏感性较大,且低频噪声较为严重。     

High‐accuracy practical spline‐based 3D and 2D integral transformations in potential‐field geophysics

Potential, potential field and potential‐field gradient data are supplemental to each other for resolving sources of interest in both exploration and solid Earth studies. We propose flexible high‐accuracy practical techniques to perform 3D and 2D integral transformations from potential field components to potential and from potential‐field gradient components to potential field components in the space domain using cubic B‐splines. The spline techniques are applicable to either uniform or non‐uniform rectangular grids for the 3D case, and applicable to either regular or irregular grids for the 2D case. The spline‐based indefinite integrations can be computed at any point in the computational domain. In our synthetic 3D gravity and magnetic transformation examples, we show that the spline techniques are substantially more accurate than the Fourier transform techniques, and demonstrate that harmonicity is confirmed substantially better for the spline method than the Fourier transform method and that spline‐based integration and differentiation are invertible. The cost of the increase in accuracy is an increase in computing time. Our real data examples of 3D transformations show that the spline‐based results agree substantially better or better with the observed data than do the Fourier‐based results. The spline techniques would therefore be very useful for data quality control through comparisons of the computed and observed components. If certain desired components of the potential field or gradient data are not measured, they can be obtained using the spline‐based transformations as alternatives to the Fourier transform techniques.     

面向目标的小束源照明和成像

结合小束源和Fourier传播子,应用Fourier传播子进行面向目标的小束源照明和成像.小束源的合成通过小波束变换中的小束函数获得,而Fourier传播子进行波场外推,完成小束源的波传播.小束源具有空间位置和方向的双局部特性,对于照明和成像,它更加灵活,并有更多的控制方法.通过照明分析,选择面向目标的有效照明小束源,进行部分小束源偏移,可以提供更好的成像质量和计算效率.作为数值试验,我们选用Fourier有限差分传播子,对 Marmousi模型和二维SEG-EAGE盐丘模型的数据,试验了小束源对目标结构的方向照明和成像.获得的结果表明,结合小束源和Fourier传播子进行面向目标的照明和成像是切实有效的.     

Full waveform inversion and the truncated Newton method: quantitative imaging of complex subsurface structures

Full waveform inversion is a powerful tool for quantitative seismic imaging from wide‐azimuth seismic data. The method is based on the minimization of the misfit between observed and simulated data. This amounts to the solution of a large‐scale nonlinear minimization problem. The inverse Hessian operator plays a crucial role in this reconstruction process. Accounting accurately for the effect of this operator within the minimization scheme should correct for illumination deficits, restore the amplitude of the subsurface parameters, and help to remove artefacts generated by energetic multiple reflections. Conventional minimization methods (nonlinear conjugate gradient, quasi‐Newton methods) only roughly approximate the effect of this operator. In this study, we are interested in the truncated Newton minimization method. These methods are based on the computation of the model update through a matrix‐free conjugate gradient solution of the Newton linear system. We present a feasible implementation of this method for the full waveform inversion problem, based on a second‐order adjoint state formulation for the computation of Hessian‐vector products. We compare this method with conventional methods within the context of 2D acoustic frequency full waveform inversion for the reconstruction of P‐wave velocity models. Two test cases are investigated. The first is the synthetic BP 2004 model, representative of the Gulf of Mexico geology with high velocity contrasts associated with the presence of salt structures. The second is a 2D real data‐set from the Valhall oil field in North sea. Although, from a computational cost point of view, the truncated Newton method appears to be more expensive than conventional optimization algorithms, the results emphasize its increased robustness. A better reconstruction of the P‐wave velocity model is provided when energetic multiple reflections make it difficult to interpret the seismic data. A better trade‐off between regularization and resolution is obtained when noise contamination of the data requires one to regularize the solution of the inverse problem.     

Improvements and Corrections to AT123D Code

Although based on exact analytical solutions, semi‐analytical solute transport models can have significant numerical error in applications with high frequency oscillatory source terms and when parameter value combinations cause series solution approximations to converge slowly. Methods for correcting these numerical errors are presented and implemented in the AT123D code, which employs Green's functions to represent point, linear, and rectangular prismatic source zones. In order to increase its computational accuracy, a Romberg numerical integration scheme was added to AT123D with prespecified error criteria, variable time stepping, and partitioning of the integral to handle rapidly changing source terms. More rapidly converging series solution approximations for the Green's functions were also incorporated to improve both accuracy and computational efficiency for finite‐depth aquifers. AT123D also has been modified to eliminate redundant calculations at points where approximate steady‐state conditions have been reached to improve computational efficiency during numerical integration. These modifications help to decrease computer run times that can be excessive for three‐dimensional problems with large numbers of computational points, small time steps, and/or long simulation time periods. Errors in the original AT123D code also were corrected in this modified version, AT123D‐AT, in order to accurately simulate finite‐duration (pulse) source releases.