非稳态地震稀疏约束反褶积研究（英文）

传统Robinson褶积模型主要受缚于三种不合理的假设,即白噪反射系数、最小相位地震子波与稳态假设,而现代反射系数反演方法（如稀疏约束反褶积等）均在前两个假设上寻求突破的同时却忽视了一个重要事实：实际地震信号具有典型的非稳态特征,这直接冲击着反射系数反演中地震子波不随时间变化的这一基础性假设。本文首先通过实际反射系数测试证实,非稳态效应造成重要信息无法得到有效展现,且对深层影响尤为严重。为校正非稳态影响,本文从描述非稳态方面具有普适性的非稳态褶积模型出发,借助对数域的衰减曲线指导检测非稳态影响并以此实现对非稳态均衡与校正。与常规不同,本文利用对数域Gabor反褶积仅移除非稳态影响,而将分离震源子波和反射系数的任务交给具有更符合实际条件的稀疏约束反褶积处理,因此结合两种反褶积技术即可有效解决非稳态特征影响,又能避免反射系数和地震子波理想化假设的不利影响。海上地震资料的应用实际表明,校正非稳态影响有助于恢复更丰富的反射系数信息,使得与地质沉积和构造相关的细节特征得到更加清晰的展现。     

Nonstationary deconvolution using maximum kurtosis optimization

Deconvolution is an essential step for high-resolution imaging in seismic data processing. The frequency and phase of the seismic wavelet change through time during wave propagation as a consequence of seismic absorption. Therefore, wavelet estimation is the most vital step of deconvolution, which plays the main role in seismic processing and inversion. Gabor deconvolution is an effective method to eliminate attenuation effects. Since Gabor transform does not prepare the information about the phase, minimum-phase assumption is usually supposed to estimate the phase of the wavelet. This manner does not return the optimum response where the source wavelet would be dominantly a mixed phase. We used the kurtosis maximization algorithm to estimate the phase of the wavelet. First, we removed the attenuation effect in the Gabor domain and computed the amplitude spectrum of the source wavelet; then, we rotated the seismic trace with a constant phase to reach the maximum kurtosis. This procedure was repeated in moving windows to obtain the time-varying phase changes. After that, the propagating wavelet was generated to solve the inversion problem of the convolutional model. We showed that the assumption of minimum phase does not reflect a suitable response in the case of mixed-phase wavelets. Application of this algorithm on synthetic and real data shows that subtle reflectivity information could be recovered and vertical seismic resolution is significantly improved.     

基于广义高斯分布的地震盲反褶积方法研究

Bussgang算法是针对褶积盲源分离问题提出的,本文将其用于地震盲反褶积处理.由于广义高斯概率密度函数具有逼近任意概率密度函数的能力,从反射系数序列的统计特征出发,引入广义高斯分布来体现反射系数序列超高斯分布特征.依据反射系数序列的统计特征和Bussgang算法原理,建立以Kullback-Leibler距离为非高斯性度量的目标函数,并导出算法中涉及到的无记忆非线性函数,最终实现了地震盲反褶积.模型试算和实际资料处理结果表明,该方法能较好地适应非最小相位系统,能够同时实现地震子波和反射系数估计,有效地提高地震资料分辨率.     

基于PSO的地震盲反褶积方法

本文基于地层反射系数非高斯的统计特性,在反褶积输出单位方差约束下,将反褶积输出的负熵表示为非多项式函数,作为盲反褶积的目标函数,然后采用粒子群算法优化目标函数寻找最佳反褶积算子,实现地震信号的盲反褶积.数值模拟和实际资料处理结果表明,与传统反褶积方法相比,本文方法同时适应于最小相位子波及混合相位子波的反褶积,能够更好地从地震数据中估计反射系数,有效拓宽地震资料的频谱,得到高分辨率的地震资料.     

基于带状混合矩阵ICA实现地震盲反褶积

基于对地震反褶积本质上是一个盲过程的认识，引入高阶统计学盲源分离技术——独立分量分析（ICA）实现地震盲反褶积.在无噪声假设条件下，利用地震记录时间延迟矩阵和地震子波带状褶积矩阵，将地震褶积模型转化为一般线性混合ICA模型，采用FastICA算法，将带状性质作为先验信息，实现所谓带状ICA算法（B\|ICA），得到个数与子波算子长度相等的多个估计反射系数序列和估计子波序列，最后利用褶积模型提供的附加信息从中优选出最佳的反射系数序列及相应的地震子波.模型数据和实际二维地震道数值算例表明：对于统计性反褶积，在不对反射系数作高斯白噪假设，不对子波作最小相位假设的所谓“全盲”条件下，基于ICA方法（反射系数非高斯分布，地震子波非最小相位）可以较好解决地震盲反褶积问题，是基于二阶统计特性的地震信号统计性反褶积方法的提升，具有可行性和应用前景.     

用遗传算法实现地震信号反褶积

遗传算法作为寻优手段具有全局优化和很好的稳定性．本文将遗传算法用于地震信号反褶积处理，与已往方法相比它具有更好的分辨率和稳定性我们采用Bernoulli－Gaussian模型和ARMA模型分别描述地震反射系数序列和地震子波，用最大似然和最小预测误差准则分别构造用于估计反射系数序列和地震子波的目标函数，用遗传算法优化目标函数，以实现地震信号反褶积．     

多分辨率地震信号反褶积

基于二进小波变换提出了一种新的反褶积方法─-多分辨率地震信号反褶积．在地震信号二进小波变换域中的各尺度上分别进行其分辨率随小波尺度变化的反褶积,利用不同分辨率反褶积结果之间的相关性,以及测量噪声随尺度的衰减特性,从低分辨率反褶积结果逼近高分辨率反褶积结果．理论分析和实验表明,该方法有较高的精度,并且在较低信噪比情况下有好的效果．     

CORRECTING FOR COLOURED PRIMARY REFLECTIVITY IN DECONVOLUTION1

Statistical deconvolution, as it is usually applied on a routine basis, designs an operator from the trace autocorrelation to compress the wavelet which is convolved with the reflectivity sequence. Under the assumption of a white reflectivity sequence (and a minimum-delay wavelet) this simple approach is valid. However, if the reflectivity is distinctly non-white, then the deconvolution will confuse the contributions to the trace spectral shape of the wavelet and reflectivity. Given logs from a nearby well, a simple two-parameter model may be used to describe the power spectral shape of the reflection coefficients derived from the broadband synthetic. This modelling is attractive in that structure in the smoothed spectrum which is consistent with random effects is not built into the model. The two parameters are used to compute simple inverse- and forward-correcting filters, which can be applied before and after the design and implementation of the standard predictive deconvolution operators. For whitening deconvolution, application of the inverse filter prior to deconvolution is unnecessary, provided the minimum-delay version of the forward filter is used. Application of the technique to seismic data shows the correction procedure to be fast and cheap and case histories display subtle, but important, differences between the conventionally deconvolved sections and those produced by incorporating the correction procedure into the processing sequence. It is concluded that, even with a moderate amount of non-whiteness, the corrected section can show appreciably better resolution than the conventionally processed section.     

气枪震源资料反褶积方法及处理流程研究

气枪震源具有极高的可重复性,可用于地下介质变化的监测。但不同工作条件下气枪震源产生的信号会存在细微差异,反褶积方法能在一定程度上消除由震源变化引起的记录信号变化。为了去除气枪震源子波信号,获取气枪源到台站之间的格林函数,通常需要选取一种恰当的方法对地震波形数据进行反褶积处理。频率域水准反褶积和时间域迭代反褶积是在接收函数等领域已被广泛使用的2种反褶积方法。本文以云南宾川主动源资料为例,对比了利用这2种方法处理气枪震源资料的效果,结果表明：在计算效率方面,频率域水准反褶积方法更具优势;在处理结果的信噪比方面,时间域迭代反褶积方法表现更好,P波初至也更清晰。此外,进一步讨论了在多炮资料的处理过程中反褶积和叠加等操作的顺序问题,最后提出了从气枪震源资料中提取气枪源到台站之间的格林函数的一般流程。     

地震子波处理的二步法反褶积方法研究

针对玛湖斜坡区三块三维地震资料和赛汉塔拉凹陷二块三维地震资料连片处理中的特点,结合地质任务和处理目标要求,提出了地震数据连片处理中的地震子波处理的方法.该方法主要体现了两次反褶积,一次是采用地表一致性反褶积,将不同震源的频带拓宽到一个标准上;再一次采用相位校正反褶积,将不同震源的数据校正到相同相位上.为了保证提取的相位校正反褶积算子稳定,采用叠后地震道提取（主要考虑到叠后地震道信噪比高,算子稳定性强）,然后将该算子应用到叠前地震道,进行相位校正.     

ESTIMATION OF MARINE SOURCE SIGNATURES FROM DIRECT ARRIVALS TO HYDROPHONE GROUPS*

Deterministic deconvolution of seismic data recorded with a non-minimum phase source, such as a sparker, requires an estimate of the source signature. Present methods of deterministic signature estimation and deconvolution require additional field equipment (near- or far-field hydrophones), or else make a deterministic estimate of the effective source signature (the free-field source signature convolved with source and receiver ghosts). By analyzing the direct arrival signal to normal hydrophone groups it is shown that extraction of the free-field source signature from this signal is possible for a spherically symmetric source, and it is demonstrated that the use of this signature for a preliminary deterministic deconvolution gives better results on sparker data than minimum-phase whitening deconvolution applied on its own. The applicability of the method to non-spherically symmetric sources, such as arrays, is also discussed. This paper is published with the permission of the Director, Bureau of Mineral Resources, Geology and Geophysics, Canberra, Australia.     

Sparsity‐enhanced wavelet deconvolution

We propose a three‐step bandwidth enhancing wavelet deconvolution process, combining linear inverse filtering and non‐linear reflectivity construction based on a sparseness assumption. The first step is conventional Wiener deconvolution. The second step consists of further spectral whitening outside the spectral bandwidth of the residual wavelet after Wiener deconvolution, i.e., the wavelet resulting from application of the Wiener deconvolution filter to the original wavelet, which usually is not a perfect spike due to band limitations of the original wavelet. We specifically propose a zero‐phase filtered sparse‐spike deconvolution as the second step to recover the reflectivity dominantly outside of the bandwidth of the residual wavelet after Wiener deconvolution. The filter applied to the sparse‐spike deconvolution result is proportional to the deviation of the amplitude spectrum of the residual wavelet from unity, i.e., it is of higher amplitude; the closer the amplitude spectrum of the residual wavelet is to zero, but of very low amplitude, the closer it is to unity. The third step consists of summation of the data from the two first steps, basically adding gradually the contribution from the sparse‐spike deconvolution result at those frequencies at which the residual wavelet after Wiener deconvolution has small amplitudes. We propose to call this technique “sparsity‐enhanced wavelet deconvolution”. We demonstrate the technique on real data with the deconvolution of the (normal‐incidence) source side sea‐surface ghost of marine towed streamer data. We also present the extension of the proposed technique to time‐varying wavelet deconvolution.     

ZERO MEMORY NON-LINEAR DECONVOLUTION*

A type of iterative deconvolution that extracts the source waveform and reflectivity from a seismogram through the use of zero memory, non-linear estimators of reflection coefficient amplitnde is developed. Here, we present a theory for iterative deconvolution that is based upon the specification of a stochastic model describing reflectivity. The resulting parametric algorithm deconvolves the seismogram by forcing a filtered version of the seismogram to resemble an estimated reflection coefficient sequence. This latter time series is itself obtained from the filtered seismogram, and so a degree of iteration is required. Algorithms utilizing zero memory non-linearities (ZNLs) converge to a family of processes, which we call Bussgang, of which any colored Gaussian process and any independent process are members. The direction of convergence is controlled by the choice of ZNL used in the algorithm. Synthetic and real data show that, generally, five to ten iterations are required for acceptable deconvolutions.     

基于反射地震记录变子波模型提高地震记录分辨率

本文给出了地震记录变子波模型的一种近似数学表达式．基于该表达式研究了反射系数序列不满足白噪假设和子波在地下传播时发生变化这两种情况下地震道谱的组成及结构,讨论了谱白化及反褶积方法在这两种情况下效果不佳的原因．然后基于变子波模型,提出了一种新的提高地震记录分辨率的方法：第一步,用自适应于地震记录的Gabor分子窗把地震记录恰当地划分成若干片断,每段内信号近似平稳,然后将地震记录变换到时间-频率域;第二步,在变换域对每个分子窗内信号的振幅谱进行处理以拓宽频带;最后把处理后的时间-频率域函数反变换回时间域得到提高分辨率后的结果．本文提出的方法具有能较好地适用于反射系数不满足白噪假设的情况及提高分辨率后的地震记录能较好地保持原地震记录的相对能量关系等优点,模型和实际资料算例结果均表明,本文方法在拓宽地震资料频带及保持地震记录局部能量相对关系方面均明显优于谱白化方法．     

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.     

薄互层地震成像中高分辨率处理方法

地震子波在叠前时间偏移前、偏移过程中以及偏移后具有不同特性和非稳态性,使得基于子波压缩原理的反褶积和基于绕射叠加原理的叠前时间偏移不能有效地提高薄互层地震分辨率.本文基于叠前时间偏移前后以及偏移过程中地震子波频率特性的分析,联合三种高分辨率处理方法对薄互层进行成像,即偏移前CMP道集应用反Q滤波补偿高频衰减、偏移过程中应用最优加权Kirchhoff叠前时间偏移降低高频损失、偏移后在CRP道集上应用子波调谐反褶积拓展频带.数值分析和实例证明,本文采用的三种高分辨率处理方法是必要的,对刻画薄互层厚度与边界能够取得好的效果,利于后续的岩性储层预测与AVO/AVP/AVA反演.     

Estimation of the seismic wavelet through homomorphic deconvolution and well log data: application on well-to-seismic tie procedure

Wavelet estimation and well-tie procedures are important tasks in seismic processing and interpretation. Deconvolutional statistical methods to estimate the proper wavelet, in general, are based on the assumptions of the classical convolutional model, which implies a random process reflectivity and a minimum-phase wavelet. The homomorphic deconvolution, however, does not take these premises into account. In this work, we propose an approach to estimate the seismic wavelet using the advantages of the homomorphic deconvolution and the deterministic estimation of the wavelet, which uses both seismic and well log data. The feasibility of this approach is verified on well-to-seismic tie from a real data set from Viking Graben Field, North Sea, Norway. The results show that the wavelet estimated through this methodology produced a higher quality well tie when compared to methods of estimation of the wavelet that consider the classical assumptions of the convolutional model.     

FURTHER THOUGHTS ON POPPERIAN GEOPHYSICS—THE EXAMPLE OF DECONVOLUTION*

Popper's demarcation criterion should be applied to all our theories in geophysics to ensure that our science progresses. We must expose our theories to tests in which they stand some risk of being refuted. But if we have a theory which has no rivals it may be difficult in practice to devise a test in which the theory risks being refuted conclusively. The example of the deconvolution problem for seismic data is considered for the case where the source wavelet is unknown. It is shown that all our existing theories of deconvolutions are not scientific in Popper's sense; they are statistical models. We cannot compare these models in a way that is independent of the geology, for each model requires the geology to have a different set of statistical properties. Even in our chosen geology it may be extremely difficult to determine the most applicable model and hence determine the “correct” deconvolution theory. It is more scientific to attempt to solve the deconvolution problem (a) by finding the source wavelet first, deterministically, or (b) by trying to force the wavelet to be a spike—that is, by devising a “perfect” seismic source. A new method of seismic surveying, which has been proposed to tackle the deconvolution problem by the first of these approaches, is based on a theory which is open to refutation by a simple Popperian test. Since the theory makes no assumptions about the geology, the test has equal validity in any geology. It pays to frame our theories in such a way that they may easily be put at risk. Only in this way will we establish whether we are on firm ground. The alternative is simply to take things on trust.     

Klauder wavelet removal before vibroseis deconvolution

The spiking deconvolution of a field seismic trace requires that the seismic wavelet on the trace be minimum phase. On a dynamite trace, the component wavelets due to the effects of recording instruments, coupling, attenuation, ghosts, reverberations and other types of multiple reflection are minimum phase. The seismic wavelet is the convolution of the component wavelets. As a result, the seismic wavelet on a dynamite trace is minimum phase and thus can be removed by spiking deconvolution. However, on a correlated vibroseis trace, the seismic wavelet is the convolution of the zero-phase Klauder wavelet with the component minimum-phase wavelets. Thus the seismic wavelet occurring on a correlated vibroseis trace does not meet the minimum-phase requirement necessary for spiking deconvolution, and the final result of deconvolution is less than optimal. Over the years, this problem has been investigated and various methods of correction have been introduced. In essence, the existing methods of vibroseis deconvolution make use of a correction that converts (on the correlated trace) the Klauder wavelet into its minimum-phase counterpart. The seismic wavelet, which is the convolution of the minimum-phase counterpart with the component minimum-phase wavelets, is then removed by spiking deconvolution. This means that spiking deconvolution removes both the constructed minimum-phase Klauder counterpart and the component minimum-phase wavelets. Here, a new method is proposed: instead of being converted to minimum phase, the Klauder wavelet is removed directly. The spiking deconvolution can then proceed unimpeded as in the case of a dynamite record. These results also hold for gap predictive deconvolution because gap deconvolution is a special case of spiking deconvolution in which the deconvolved trace is smoothed by the front part of the minimum-phase wavelet that was removed.