3D free‐boundary conditions for coordinate‐transform finite‐difference seismic modelling

New alternative formulations of exact boundary conditions for arbitrary three-dimensional (3D) free-surface topographies on seismic media have been derived. They are shown to be equivalent to previously published formulations, thereby verifying the validity of each set of formulations. The top of a curved grid represents the free-surface topography while the interior of the grid represents the physical medium. We assume the velocity–stress version of the viscoelastic wave equations to be valid in this grid before transforming the equations to a rectangular grid. In order to perform the numerical discretization we apply the latter version of the equations for seismic wave propagation simulation in the medium. The numerical discretization of the free-surface topography boundary conditions by second-order finite differences (FDs) is shown, as well as the spatially unconditional stability of the resulting system of equations. The FD order is increased by two for each point away from the free surface up to eight, which is the order used in the interior. We use staggered grids in both space and time and the second-order leap-frog and Crank– Nicholson methods for wavefield time propagation. An application using parameters typical of teleseismic earthquakes and explosions is presented using a 200 × 100 km² area of real topography from southwestern Norway over a homogeneous medium. A dipping plane wave simulates a teleseismic P-wave incident on the surface topography. Results show conversion from P- to Rg- (short period fundamental mode Rayleigh) waves in the steepest and/or roughest topography, as well as attenuated waves in valleys and fjords. The codes are parallelized for simulation on fast supercomputers and PC-clusters to model high frequencies and/or large areas.     

Three-parameter Radon transform in layered transversely isotropic media

Radon transform is a powerful tool with many applications in different stages of seismic data processing, because of its capability to focus seismic events in the transform domain. Three-parameter Radon transform can optimally focus and separate different seismic events, if its basis functions accurately match the events. In anisotropic media, the conventional hyperbolic or shifted hyperbolic basis functions lose their accuracy and cannot preserve data fidelity, especially at large offsets. To address this issue, we propose an accurate traveltime approximation for transversely isotropic media with vertical symmetry axis, and derive two versions of Radon basis functions, time-variant and time-invariant. A time-variant basis function can be used in time domain Radon transform algorithms while a time-invariant version can be used in, generally more efficient, frequency domain algorithms. Comparing the time-variant and time-invariant Radon transform by the proposed basis functions, the time-invariant version can better focus different seismic events; it is also more accurate, especially in presence of vertical heterogeneity. However, the proposed time-invariant basis functions are suitable for a specific type of layered anisotropic media, known as factorized media. We test the proposed methods and illustrate successful applications of them for trace interpolation and coherent noise attenuation.     

Out‐of‐plane geometrical spreading in anisotropic media

Two-dimensional seismic processing is successful in media with little structural and velocity variation in the direction perpendicular to the plane defined by the acquisition direction and the vertical axis. If the subsurface is anisotropic, an additional limitation is that this plane is a plane of symmetry. Kinematic ray propagation can be considered as a two-dimensional process in this type of medium. However, two-dimensional processing in a true-amplitude sense requires out-of-plane amplitude corrections in addition to compensation for in-plane amplitude variation. We provide formulae for the out-of-plane geometrical spreading for P- and S-waves in transversely isotropic and orthorhombic media. These are extensions of well-known isotropic formulae.
For isotropic and transversely isotropic media, the ray propagation is independent of the azimuthal angle. The azimuthal direction is defined with respect to a possibly tilted axis of symmetry. The out-of-plane spreading correction can then be calculated by integrating quantities which describe in-plane kinematics along in-plane rays. If, in addition, the medium varies only along the vertical direction and has a vertical axis of symmetry, no ray tracing need be carried out. All quantities affecting the out-of-plane geometrical spreading can be derived from traveltime information available at the observation surface.
Orthorhombic media possess no rotational symmetry and the out-of-plane geometrical spreading includes parameters which, even in principle, are not invertible from in-plane experiments. The exact and approximate formulae derived for P- and S-waves are nevertheless useful for modelling purposes.     

Multiple attenuation: coping with the spatial truncation effect in the Radon transform domain

The limited size of the spatial aperture of a seismic gather causes multiple- and primary-reflection energy to spread out and cross-hatch the image of the parabolic Radon transform. This spatial truncation effect devastatingly impairs the separation of primary and multiple reflections in a Radon demultiple process. It is difficult to suppress the spatial truncation effect completely, but it is at least possible to model or to identify implicitly such an effect and then to design automatically a mute function, i.e. a 2D mask filter in the Radon transform domain. This is referred to as the adaptive surgical mute scheme. By using this scheme, it is possible to extract the multiple-reflection energy cleanly from the Radon transform image, so that the multiple reflections can be more effectively attenuated. It is also possible to preserve the diffused primary-reflection energy after the surgical mute, so that the final multiple-attenuated seismic profile is amplitude-preserved.     

基于改进线性Radon变换的井孔地震上下行波场分离方法

反射波场分离是井孔地震资料处理中极其重要的一个环节,波场分离的质量直接影响成像结果的精度.不管是VSP还是井间地震资料,其反射波时距曲线都近似直线型,根据这一特征,本文提出一种改进的线性Radon变换方法来进行井孔资料的反射波上下行波场分离.该方法基于频率域线性Radon变换,通过引入一个新的变量λ来消除变换算子对频率的依赖性,避免了求取每一频率分量对应的不同变换算子,显著降低了计算成本;文中在求解该方法对应的最小二乘问题时,引入了发展较为成熟的高分辨率Radon变换技术来进一步提高波场分离的精度.采用本文方法进行井孔地震资料的上下行波场分离可以在保证分离精度的前提下有效地提高计算效率.根据上下行波在λ-f域内分布的特殊性,设计简单的滤波算子就可实现上下行波场的分离.最后通过合成数据试算以及实际资料处理（VSP数据和井间地震数据）验证了该方法的可行性和有效性.     

基于多路径Radon变换的地震数据噪声压制和波型分离方法研究

Radon变换是一种稀疏变换,被广泛应用于地震数据处理,其中线性Radon和抛物Radon最为常用.在实际地震数据中,直达波和面波的同相轴形态为线性,反射波为双曲型,单独使用线性Radon或抛物Radon变换时,不能确保所有同相轴在变换域的系数都是稀疏的,影响地震数据处理效果.本文提出的多路径Radon变换联合了线性Radon变换和抛物Radon变换,每个同相轴都有两种不同的变换参数来与积分路径相适应,能够兼顾不同形态的同相轴;然后利用最小二乘稀疏反演方法对不同形态同相轴匹配最佳积分路径,使其自动分离到两个不同的Radon域剖面且保持系数同时稀疏.从多路径Radon域剖面上,能够很容易地识别不同系数所对应的同相轴形态、时间截距以及速度,这些特征有利于提高利用Radon变换方法进行随机噪声压制、面波压制以及波型分离等技术的处理效果,该变换在模型数据和实际数据中的应用结果证明了本文方法有效性.

     

Avoiding pitfalls of least‐squares inversion by using the energy‐flux error criterion

Seismic inversion by modelling and data fitting depends on the criterion chosen to measure the misfit between observed and modelled data. The popular least‐squares error criterion has an important drawback: it is sensitive both to the shape of the recording surface and to velocity variations along this surface. Tests on synthetic seismic reflection data show that least‐squares inversion may work surprisingly poorly in situations where (i) the range of angles between reflected rays and the acquisition surface is large, (ii) the velocity varies significantly along this surface, or (iii) a compensation for the effects of dissipation is applied to the gradients. In these situations, the gradients may contain important artefacts and have incorrect amplitudes. The outgoing flux of energy of the residual wavefield across the acquisition surface provides an alternative measure of the data misfit which is independent of the recording surface, provided this surface is closed, and which is only sensitive to the aperture in the practical situation of an open surface or line of receivers. Energy‐flux inversion presents a strong resemblance to reverse‐time migration, but with the additional possibility of iteratively improving the images. In all the tests, energy‐flux inversion provided better images than least‐squares inversion.     

Logarithm of short‐time Fourier transform for extending the seismic bandwidth

Improving seismic resolution is essential for obtaining more detailed structural and stratigraphic information. We present a new algorithm to increase seismic resolution with a minimum of user‐defined parameters. The algorithm inherits useful properties of both the short‐time Fourier transform and the cepstrum to smooth and broaden the frequency spectrum at each translation of the spectral decomposing window. The key idea is to replace the amplitude spectrum with its logarithm in each window of the short‐time Fourier transform. We describe the mathematical formulation of the algorithm and its testing on synthetic and real seismic data to obtain broader frequency spectra and thus enhance the seismic resolution.     

基于拉东变换的瑞雷面波频散分析与应用

为克服面波谱分析法(SASW)提取频散曲线抗干扰能力差、不能得到多模式频散曲线等缺点,对拉东变换法进行了改进,不对原始记录进行数字处理,避免了数字处理效应的影响.通过对信号的频谱分析、结合场地的地质条件,选择频散分析的频率、速度范围,来达到规避高视速度的直达波、反射波.理论模型合成记录和实际资料的处理表明:研究的频散分析方法是有效的且适应性强,取得了较好的效果.     

基于一维卷积神经网络的高分辨率Radon变换反演方法研究

高分辨率Radon变换是地震资料处理常用的方法之一,其反演通常涉及矩阵求逆、多次迭代等环节,这些因素导致Radon变换反演计算量大,收敛速度慢等问题.本文在分析Radon变换分辨率降低原因基础上,提出基于一维卷积神经网络（Convolutional Neural Network,CNN）的高分辨率Radon变换反演方法.该方法通过卷积神经网络的非线性表征能力实现低分辨率Radon参数到高分辨率Radon参数的映射,分析了基于反褶积原理的串联映射模型和基于残差学习的并联映射模型提高分辨率的原理.将上述CNN网络得到的特定频率Radon参数约束其他频率参数的反演,避免了分频训练的弊端.模拟数据和实际数据的多次波压制实验表明,本文提出的基于一维卷积神经网络的高分辨率Radon变换可以较好地压制多次波,且计算效率高.

     

基于波动方程预测和双曲Radon变换联合压制表面多次波

基于波动方程预测的表面多次波压制方法可处理复杂地下介质的地震资料,但计算成本较高.基于滤波的多次波压制方法计算效率较高,但其成功应用仅局限于一次波和多次波有明显时差差别的地震数据,对来自速度逆转等复杂介质数据则较难获得满意的压制效果.本文将波动方程预测的反馈迭代法和滤波法有效结合,采用GPU(图形处理器)和CPU协同并行加速计算粗略预测表面多次波,随后在双曲Radon域比较分析原始数据和预测的多次波,设计合理有效的Butterworth型自适应滤波器,滤出原始数据Radon域中的多次波能量,进行Radon反变换后,在时空域将多次波从原始数据中减去,得多次波压制结果.文中对理论模拟的单炮数据、复杂的SMAART模型以及实际地震数据进行了计算,结果表明,结合基于波动方程预测和双曲Radon变换的方法有效突破了两种方法各自的局限性,可高效高精度地压制复杂地下介质的表面多次波.     

Improved methods to transform frequency‐dependent complex stiffness to time domain

A method to transform the frequency‐dependent complex stiffness to the impulse response in the time domain was proposed in the previous paper. However, there is a problem in that the accuracy and the convergence of the transformed impulse response are not good in some cases. Moreover, the hysteretic damping was not considered in the previous study although it is essential for practical purposes. In this paper, transform method improvements are proposed. First, the accuracy and the convergence are improved by taking the concept of virtual mass into account. Then, a more improved method for transforming the complex stiffness with large hysteretic damping to the time domain is proposed using the least square method. It is well known that the rigorous transform of the hysteretic damping is impossible because it is non‐causal. So this method is thought to be an approximate causalization process. Copyright © 2006 John Wiley & Sons, Ltd.     

Iterative resolution estimation in least‐squares Kirchhoff migration

We apply iterative resolution estimation to least‐squares Kirchhoff migration. Reviewing the theory of iterative optimization uncovers the common origin of different optimization methods. This allows us to reformulate the pseudo‐inverse, model resolution and data resolution operators in terms of effective iterative estimates. When applied to Kirchhoff migration, plots of the diagonal of the model resolution matrix reveal low illumination areas on seismic images and provide information about image uncertainties. Synthetic and real data examples illustrate the proposed technique and confirm the theoretical expectations.     

Seismic inversion with generalized Radon transform based on local second-order approximation of scattered field in acoustic media

Sound velocity inversion problem based on scattering theory is formulated in terms of a nonlinear integral equation associated with scattered field. Because of its nonlinearity, in practice, linearization algorisms (Born/single scattering approximation) are widely used to obtain an approximate inversion solution. However, the linearized strategy is not congruent with seismic wave propagation mechanics in strong perturbation (heterogeneous) medium. In order to partially dispense with the weak perturbation assumption of the Born approximation, we present a new approach from the following two steps: firstly, to handle the forward scattering by taking into account the second-order Born approximation, which is related to generalized Radon transform (GRT) about quadratic scattering potential; then to derive a nonlinear quadratic inversion formula by resorting to inverse GRT. In our formulation, there is a significant quadratic term regarding scattering potential, and it can provide an amplitude correction for inversion results beyond standard linear inversion. The numerical experiments demonstrate that the linear single scattering inversion is only good in amplitude for relative velocity perturbation ( \( \delta_{c}/c_{0} \) ) of background media up to 10 %, and its inversion errors are unacceptable for the perturbation beyond 10 %. In contrast, the quadratic inversion can give more accurate amplitude-preserved recovery for the perturbation up to 40 %. Our inversion scheme is able to manage double scattering effects by estimating a transmission factor from an integral over a small area, and therefore, only a small portion of computational time is added to the original linear migration/inversion process.     

True amplitude imaging by inverse generalized Radon transform based on Gaussian beam decomposition of the acoustic Green's function

True amplitude migration is one of the most important procedures of seismic data processing. As a rule it is based on the decomposition of the velocity model of the medium into a known macrovelocity component and its sharp local perturbations to be determined. Under this decomposition the wavefield can be considered as the superposition of an incident and reflected/scattered waves. The single scattering approximation introduces the linear integral operator that connects the sharp local perturbations of the macrovelocity model with the multishot/multioffset data formed from reflected/scattered waves. We develop the pseudoinverse of this operator using the Gaussian beam based decomposition of acoustic Green's functions. The computation of this pseudoinverse operator is done pointwise by shooting Gaussian beams from the target area towards the acquisition system. The numerical implementation of the pseudoinverse operator was applied to the synthetic data Sigsbee2A. The results obtained demonstrate the high quality of the true amplitude images computed both in the smooth part of the model and under the salt body.     

广域电磁法接收机数字信号FFT实时实现技术

针对广域电磁法（WFEM）接收机各频组数字信号快速傅里叶变换（FFT）点数不统一、低频组点数大不适合现场可编程门阵列（FPGA）实时实现的技术难题,本文采用插值和抽取相结合的抽样率转换方法,提出了满足广域电磁法接收机要求的等长度FFT变换技术,将各频组FFT变换点数统一变换到1024点,并以广域电磁法接收机实际采集到的电场信号进行了测试,最终在EP2C35F484C8 FPGA芯片上实现了1024点FFT变换,最高响应频率达到了93.84 MHz,完成1024点广域电磁信号FFT运算最快只需34 μs,并且能够有效提取广域电磁信号的振幅和相位信息.结果表明：在对广域电磁法接收机信号进行FIR低通滤波后,能够通过插值和抽取相结合的方法,将各个频组FFT变换的长度统一变换成1024点,实现广域电磁信号的FFT实时变换.本文提出的等长度FFT变换技术能够满足新一代高性能广域电磁法接收机FFT实时变换的需求.     

Hiibert-Huang transform and wavelet analysis of time history signal

The brief theories of wavelet analysis and Hilbert-Huang transform (HHT) are introduced firstly in the present paper. Then several signal data were analyzed by using wavelet and HHT methods, respectively. The comparison shows that HHT is not only an effective method for analyzing non-stationary data, but also is a useful tool for examining detailed characters of time history signal.     

BG2015闪烁室测氡仪器应用于地震气氡测试运行实验

对BG2015闪烁室测氡仪器在江西九江地震台2号井、甘肃平凉地震台安国井和云南省开远地震局滇18井不同井水条件进行实验,针对不同井条件下气氡测值的稳定性、连续性和可靠性进行比较分析,得到BG2015闪烁室测氡仪器应用于地震行业的可行性。