不依赖源子波的跨孔雷达时间域波形反演

波形反演是近年来较热门的反演方法,其分辨率可以达到亚波长级别.在波形反演的实际应用中,源子波的估计十分重要.传统方法使用反褶积来估计源子波并随着反演过程更新,该方法在合成数据波形反演中效果较好,但在实际数据反演过程中存在一系列的问题.由于实际数据信噪比较低,在源子波估计过程中需要大量的人为干涉,且结果并不一定可靠.本文使用一种基于褶积波场的新型目标函数,令反演过程不再依赖源子波.详细推导了针对跨孔雷达波形反演的梯度及步长公式,实现介电常数和电导率的同步反演.针对一个合成数据模型同时反演介电常数和电导率,结果表明该方法能够反演出亚波长尺寸异常体的形状和位置.接着,将该方法应用到两组实际数据中,并与基于估计源子波的时间域波形反演结果进行比较.结果表明不依赖源子波的时间域波形反演结果分辨率更高,也更准确.     

频率域全波形反演方法研究进展

全波形反演方法利用叠前地震波场的运动学和动力学信息重建地下速度结构,具有揭示复杂地质背景下构造与岩性细节信息的潜力.根据研究需要,全波形反演既可在时间域也可在频率域实现.频率域相对于时间域反演具有计算高效、数据选择灵活等优势.近十几年来频率域全波形反演理论在波场模拟方法、反演频率选择策略、目标函数设置方式、震源子波处理方式、梯度预处理方法等方面取得了进展.目标函数存在大量局部极值的特性是影响反射地震全波形反演效果的重要内在因素之一.如果将Laplace域波形反演、频率域阻尼波场反演、频率域波形反演三种方法有机结合,可以降低反演的非线性程度.     

基于多尺度迭代求解三维频率-空间域声波方程的全波形速度反演研究-理论模型

使用广义最小残量方法迭代求解三维频率-空间域声波方程,反演时使用多尺度、多重网格的策略,探讨了如何快速实现高分辨率的三维频率-空间域迭代法声波全波形速度反演.通过对理论模型进行三维频率-空间域迭代法声波全波形反演数值试验,证实该方法的计算速度快、计算效率高,反演所得速度的分辨率高.从而为基于多尺度迭代求解三维频率-空间域声波方程的全波形速度反演成像打下方法基础.     

基于反射地震数据的时频域包络反演

包络信号含有丰富的低频分量,即使在地震数据缺失低频条件下,包络目标函数也能有效缓解全波形反演的周期跳跃现象.但是,当初始速度模型较为平滑时,观测数据中的反射地震事件在模拟数据中没有与之相对应的波形,导致包络反演初期无法很好地利用反射波信号进行速度建模.本文提出基于反射地震数据的时频域包络反演方法,通过结合反射波全波形反演理论,构建反射波时频域包络目标函数,来提高包络反演的速度建模精度.本文首先利用Gabor变换获取时频域地震数据,并提取振幅信息,即为时频域包络信号.然后,推导反射波时频域包络反演的伴随震源和梯度算子.Marmousi模型数据测试结果表明,基于反射地震数据的时频域包络反演方法可以为全波形反演提供一个较好的初始速度模型.     

基于反射地震数据的时频域包络反演

包络信号含有丰富的低频分量,即使在地震数据缺失低频条件下,包络目标函数也能有效缓解全波形反演的周期跳跃现象.但是,当初始速度模型较为平滑时,观测数据中的反射地震事件在模拟数据中没有与之相对应的波形,导致包络反演初期无法很好地利用反射波信号进行速度建模.本文提出基于反射地震数据的时频域包络反演方法,通过结合反射波全波形反演理论,构建反射波时频域包络目标函数,来提高包络反演的速度建模精度.本文首先利用Gabor变换获取时频域地震数据,并提取振幅信息,即为时频域包络信号.然后,推导反射波时频域包络反演的伴随震源和梯度算子.Marmousi模型数据测试结果表明,基于反射地震数据的时频域包络反演方法可以为全波形反演提供一个较好的初始速度模型.     

基于自适应非稳态相位校正的时频域多尺度全波形反演

本文提出非稳态相位校正时频域目标函数,通过缩小观测数据与模拟数据在波形相位上的差异来缓解全波形反演过程中对应波形匹配错位的问题（周波跳跃）.同时引入自适应相位校正因子,可以根据观测数据与模拟数据的差异来调整相位校正量的大小.在构建非稳态相位校正时频域全波形反演目标函数的基础上,利用链式法则详细推导了对应的伴随震源,并从理论上证明了该方法的可行性与优越性.数值测试过程中结合了低通滤波多尺度反演策略,进一步缓解全波形反演过程中的强非线性问题.缺失低频分量测试结果表明,利用自适应非稳态相位校正时频域多尺度全波形反演方法结合常规全波形反演方法在缺失7 Hz以下低频分量的地震数据中仍然能够得到高精度的反演结果.震源不准确测试结果表明,即使震源子波相位差异较大,利用非稳态相位校正方法仍然能够一定程度上缓解周波跳跃现象.测试结果综合证明了本文提出的方法在构建初始速度建模,缓解周波跳跃等方面具有一定的优势.     

基于广义模型约束的时间域航空电磁反演研究

由于航空电磁具有海量数据,因此快速有效的成像和反演手段至关重要.本文针对层状介质模型推导与实现了广义模型约束条件下时间域航空电磁一维反演.从正则化反演的目标函数出发,通过改变模型约束项构造Lp范数反演和聚焦反演,进而通过改变模型求解域构造出基于小波变换的稀疏约束反演.针对不同反演方法目标函数的构建方式,本文进一步从数学原理上分析不同反演方法的预期效果,并通过理论模型和实测数据进行验证.结果表明L0.8范数反演、聚焦反演和基于小波变换的稀疏约束反演可以得到更符合地下层状介质陡变界面的反演结果.     

相位空间内的解缠绕相位反演

全波形反演方法是一种数据域高精度反演方法,该方法通过匹配观测数据与模拟数据的地震波形,利用梯度法准确反演地下介质参数的分布情况.由于观测数据普遍缺少低频信息,该方法易受周期跳跃现象影响.特别是当地下存在大尺度强反射界面的构造时,地下介质的反演转化为强非线性问题求解.该情形下,即使观测数据包含充足的低频信息,全波形反演也难以给出准确的反演结果.一般可以通过减弱反演对初始模型参数的依赖性来克服上述问题,具体表现为使用新变量(例如瞬时相位、包络等)代替目标函数中的采样后波场,以增强新目标函数的凸性.但是,对该新目标函数进行反演时,伴随状态方程中存在关于新变量和波场的一个链式微分项,该项保留了反演问题的非线性,导致新的反演方法难以处理包含大尺度构造的强非线性反演问题.此外,基于新变量的反演问题依然在波场空间中计算模型梯度,难以充分利用新变量与模型参数之间的弱非线性关系.因此,本文提出用频率域波动方程的相位形式代替传统的波动方程来消除伴随状态方程中的链式微分项,用解缠绕的相位代替目标函数中采样前波场并在相位空间进行反演.该方法可以最大程度地利用地下介质参数和解缠绕相位之间的弱非线性关系,从而削弱反演的非线性性.由于基于频率域波场计算得到相位有严重的缠绕问题,本文采用基于振幅排序的多聚类算法来对相位进行解缠绕.虽然将介质参数到波场的映射替换为介质参数与解缠绕相位的映射,会导致反演结果的分辨率有所下降,但该方法可以在相位空间恢复介质参数的大尺度低波数分量.Marmousi模型测试证明了该方法的有效性和准确性,针对部分BP模型的测试也证明了该方法处理强非线性问题的能力.     

基于K-SVD字典学习的稀疏约束编码多震源方向:全波形反演（英文）

全波形反演(FWI)是一种较为重要的速度建模方法,但计算量巨大是阻碍其实用化。业已证明通过多震源策略减少模拟单炮次数,可以大大提高全波形反演计算效率,但引入了交叉串扰噪音。为解决上述问题,本文提出一种基于K-SVD字典学习的稀疏约束编码多震源全波形反演方法。首先,增加不同单炮的差异性引入极性编码策略减少串扰噪音;其次基于FWI不同迭代次数反演结果特征引入K-SVD字典学习方法计算变换基函数,推导了基于稀疏约束的目标泛函;进一步我们引入基于维纳滤波的时间域多尺度反演方法,提高反演方法的稳定性。最后,通过洼陷模型和Marmousi模型测试验证表明:1)本文的基于K-SVD字典学习的多震源编码反演方法,在减少全波形反演计算量的同时,能有效克服反演串扰噪音,提高反演精度;2)新方法能灵活的与时间域多尺度反演方法结合,降低反演过程陷入局部极小值,增强反演稳定性,对复杂模型也具有较好的适应性。     

基于归一化能量谱目标函数的全波形反演方法

基于L2范数的常规全波形反演目标函数是一个强非线性泛函,在反演过程中容易陷入局部极小值.本文提出归一化能量谱目标函数来缓解全波形反演过程中的强非线性问题,同时能够有效地缓解噪声和震源子波不准等因素的影响.能量谱目标函数是通过匹配观测数据与模拟数据随频率分布的能量信息来实现最小二乘反演的,其忽略了地震数据波形与相位变化的细节特征,这在反演的过程中能够有效缓解波形匹配错位等问题.数值测试结果表明,基于归一化能量谱目标函数在构建初始速度模型、抗噪性和缓解震源子波依赖等方面都优于归一化全波形反演目标函数.金属矿模型测试结果表明,即使地震数据缺失低频分量,基于归一化能量谱目标函数的全波形反演方法在像金属矿这样的强散射介质反演问题上同样具有一定的优势.     

2D acoustic‐elastic coupled waveform inversion in the Laplace domain

Although waveform inversion has been intensively studied in an effort to properly delineate the Earth's structures since the early 1980s, most of the time‐ and frequency‐domain waveform inversion algorithms still have critical limitations in their applications to field data. This may be attributed to the highly non‐linear objective function and the unreliable low‐frequency components. To overcome the weaknesses of conventional waveform inversion algorithms, the acoustic Laplace‐domain waveform inversion has been proposed. The Laplace‐domain waveform inversion has been known to provide a long‐wavelength velocity model even for field data, which may be because it employs the zero‐frequency component of the damped wavefield and a well‐behaved logarithmic objective function. However, its applications have been confined to 2D acoustic media. We extend the Laplace‐domain waveform inversion algorithm to a 2D acoustic‐elastic coupled medium, which is encountered in marine exploration environments. In 2D acoustic‐elastic coupled media, the Laplace‐domain pressures behave differently from those of 2D acoustic media, although the overall features are similar to each other. The main differences are that the pressure wavefields for acoustic‐elastic coupled media show negative values even for simple geological structures unlike in acoustic media, when the Laplace damping constant is small and the water depth is shallow. The negative values may result from more complicated wave propagation in elastic media and at fluid‐solid interfaces. Our Laplace‐domain waveform inversion algorithm is also based on the finite‐element method and logarithmic wavefields. To compute gradient direction, we apply the back‐propagation technique. Under the assumption that density is fixed, P‐ and S‐wave velocity models are inverted from the pressure data. We applied our inversion algorithm to the SEG/EAGE salt model and the numerical results showed that the Laplace‐domain waveform inversion successfully recovers the long‐wavelength structures of the P‐ and S‐wave velocity models from the noise‐free data. The models inverted by the Laplace‐domain waveform inversion were able to be successfully used as initial models in the subsequent frequency‐domain waveform inversion, which is performed to describe the short‐wavelength structures of the true models.     

基于GPU并行的时间域全波形优化共轭梯度法快速GPR双参数反演

探地雷达（GPR）时间域全波形反演计算量巨大,内存要求高,在微机上计算难度大.本文中作者基于GPU并行加速的维度提升反演策略,采用优化的共轭梯度法,避免了Hessian矩阵的计算,在普通微机上实现了时间域全波形二维GPR双参数（介电常数和电导率）快速反演.论文首先推导了二维TM波的时域有限差分法（FDTD）的交错网格离散差分格式及波场更新策略.然后,基于Lagrange乘数法,将约束问题转化为无约束最小问题,构建了共轭梯度法反演目标函数,采用Fletcher-Reeves公式与非精确线搜索Wolfe准则,确保了梯度方向修正因子及迭代步长选取的合理性.而GPU并行计算及维度提升反演策略的应用,数倍地提升了反演速度.最后,开展了3个模型的合成数据的反演实验,分别从观测方式、梯度优化及天线频率等方面,分析了这些因素对雷达全波形反演的影响,说明双参数的反演较单一的介电常数反演,能提供更丰富的信息约束,有效提高模型重建的精度.     

Improvements in crosshole GPR full-waveform inversion and application on data measured at the Boise Hydrogeophysics Research Site

Crosshole ground penetrating radar (GPR) tomography has been widely used and has the potential to improve the obtained subsurface models due to its high spatial resolution compared to other methods. Recent advances in full-waveform inversion of crosshole GPR data show that higher resolution images can be obtained compared to conventional ray-based GPR inversion because it can exploit all information present in the observed data. Since the first application of full-waveform inversion on synthetic and experimental GPR data, the algorithm has been significantly improved by extending the scalar to a vectorial approach, and changing the stepped permittivity and conductivity update into a simultaneous update. Here, we introduce new normalized gradients that do not depend on the number of sources and receivers which enable a comparison of the gradients and step lengths for different crosshole survey layouts. An experimental data set acquired at the Boise Hydrogeophysics Research Site is inverted using different source–receiver setups and the obtained permittivity and conductivity images, remaining gradients and final misfits are compared for the different versions of the full-waveform inversion. Moreover, different versions of the full-waveform inversion are applied to obtain an overview of all improvements. Most improvements result in a reducing final misfit between the measured and synthetic data and a reducing remaining gradient at the final iteration. Regions with relatively high remaining gradient amplitudes indicate less reliable inversion results. Comparison of the final full-waveform inversion results with Neutron–Neutron porosity log data and capacitive resistivity log data show considerably higher spatial frequencies for the logging data compared to the full-waveform inversion results. To enable a better comparison, we estimated a simple wavenumber filter and the full-waveform inversion results show an improved fit with the logging data. This work shows the potential of full-waveform inversion as an advanced method that can provide high resolution images to improve hydrological models.     

基于改进全变差正则化的GPR多尺度全波形双参数同步反演

针对探地雷达（GPR）双参数全波形反演中电导率反演精度差、双参数存在串扰现象、反演计算量大、易陷入局部极值等问题.作者将具有多参数调节功能的L-BFGS算法引入到GPR时间域全波形反演中,它避免了对Hessian矩阵的直接存储与精确求解,减小了存储量和计算量.结合参数调节因子的选取,有效减小了同步反演时介电常数与电导率的串扰影响,在不降低介电常数反演精度的前提下,提高电导率参数的反演精度.通过在反演目标函数中加载改进全变差正则化方法,提高了反演的稳定性,使目标体边缘轮廓更加清晰.首先以简单模型为例,对比了单尺度反演与多尺度串行反演策略的优劣,说明多尺度串行反演有利于逐步搜索全局最优解;而开展参数调节因子的选取实验,说明合适的参数调节因子可以有效改善介质电导率的反演精度;测试了不同正则化的反演效果,表明改进全变差正则化能提高反演稳定性,显著降低模型重构误差.最后,分别对含噪合成数据和实测数据进行了反演测试,说明本文提出的多尺度、双参数反演具有较强的鲁棒性,能提供更丰富的信息约束,重构图像界面清晰、反演效果好.     

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.     

2D Laplace-Domain Waveform Inversion of Field Data Using a Power Objective Function

The wavefield in the Laplace domain has a very small amplitude except only near the source point. In order to deal with this characteristic, the logarithmic objective function has been used in many Laplace domain inversion studies. The Laplace-domain waveform inversion using the logarithmic objective function has fewer local minima than the time- or frequency domain inversion. Recently, the power objective function was suggested as an alternative to the logarithmic objective function in the Laplace domain. Since amplitudes of wavefields are very small generally, a power <1 amplifies the wavefields especially at large offset. Therefore, the power objective function can enhance the Laplace-domain inversion results. In previous studies about synthetic datasets, it is confirmed that the inversion using a power objective function shows a similar result when compared with the inversion using a logarithmic objective function. In this paper, we apply an inversion algorithm using a power objective function to field datasets. We perform the waveform inversion using the power objective function and compare the result obtained by the logarithmic objective function. The Gulf of Mexico dataset is used for the comparison. When we use a power objective function in the inversion algorithm, it is important to choose the appropriate exponent. By testing the various exponents, we can select the range of the exponent from 5 × 10^?3 to 5 × 10^?8 in the Gulf of Mexico dataset. The results obtained from the power objective function with appropriate exponent are very similar to the results of the logarithmic objective function. Even though we do not get better results than the conventional method, we can confirm the possibility of applying the power objective function for field data. In addition, the power objective function shows good results in spite of little difference in the amplitude of the wavefield. Based on these results, we can expect that the power objective function will produce good results from the data with a small amplitude difference. Also, it can partially be utilized at the sections where the amplitude difference is very small.     

Comparison of scaling methods for waveform inversion

Waveform inversion can lead to faint images for later times due to geometrical spreading. The proper scaling of the steepest-descent direction can enhance faint images in waveform inversion results. We compare the effects of different scaling techniques in waveform inversion algorithms using the steepest-descent method. For the scaling method we use the diagonal of the pseudo-Hessian matrix, which can be applied in two different ways. One is to scale the steepest-descent direction at each frequency independently. The other is to scale the steepest-descent direction summed over the entire frequency band. The first method equalizes the steepest-descent directions at different frequencies and minimizes the effects of the band-limited source spectrum in waveform inversion. In the second method, since the steepest-descent direction summed over the entire frequency band is divided by the diagonal of the pseudo-Hessian matrix summed over the entire frequency band, the band-limited property of the source wavelet spectrum still remains in the scaled steepest-descent directions. The two scaling methods were applied to both standard and logarithmic waveform inversion. For standard waveform inversion, the method that scales the steepest-descent direction at every frequency step gives better results than the second method. On the other hand, logarithmic waveform inversion is not sensitive to the scaling method, because taking the logarithm of wavefields automatically means that results for the steepest-descent direction at each frequency are commensurate with each other. If once the steepest-descent directions are equalized by taking the logarithm of wavefields in logarithmic waveform inversion, the additional equalizing effects by the scaling method are not as great as in conventional waveform inversion.     

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.     

跨孔雷达全波形反演成像方法的研究

跨孔雷达全波形反演是一种使用全波形信息反演两钻孔之间地下信息的层析成像技术.常规的层析成像反演大部分采用射线追踪方法,其中基于初至时的射线追踪方法可以反演出速度剖面（介电常数剖面）,基于最大振幅的层析成像可以反演出衰减剖面（电导率剖面）.常规射线追踪方法有许多不足,究其原因是该方法仅使用了小部分的信号信息.为了进一步提高成像分辨率,本文全面推导了全波形跨孔雷达层析成像反演方法,该方法利用雷达波全幅度相位信息能够反演出地下高分辨率的介电常数和电导率图像.本文通过基于局域网的分布式并行算法,有效地解决了巨量数据正演计算问题.文中首先建立了基于单轴各向异性介质完全匹配层的时间域有限差分二维正演算法,进而通过应用包括时间维度在内的全波场信息与残场逆向传播的全波场信息乘积来计算梯度方向,通过求取以步长为自变量的目标函数的极值确定步长公式,并提出以第一次介电常数反演作为同步反演的初始模型,能够有效提高收敛速度.本文对多组模型进行成像实验,取得了较好的反演效果.