基于曲线矢量空间的二维时频峰值滤波消减地震勘探随机噪声策略

It plays an important role in petroleum seismic to remove the seismic random noise as much as possible and recover the valid signal (reflection wave) with high fidelity, meeting the requirements of the high signal-to-noise ratio (SNR), the high resolution and the high accuracy under the complex surface and near surface situation for the increasingly complex seismic objects in cumulated reservoir. The two-dimensional time-frequency peak filtering in curve vector space method will be developed to attenuate the seismic noise and preserve the seismic signal accurately by exploring seismic lateral coherence in this proposal. This method constructs the curve vector space according to the local direction of the event. Therefore the linearity of the signal in vector space is improved and the curve information of the event is preserved. Consequently, this two-dimensional time-frequency peak filtering method based on the vector space enhances the unbiased filtering ability of time-frequency peak filtering to obtain the accurate filtered signal. Furthermore, we will study the relationship between the filtering parameters and the characteristics of the seismic noise, including nonstationarity, nonlinearity and quasi-similarity, etc. Then the temporal and spatial various two-dimensional time-frequency peak filtering method will be developed. It will reveal the key problems of the changes in two dimensional time-frequency characteristics of the seismic noise, the cross-term of time-frequency distribution and the semblance estimation, etc. This technique will provide the reliable processing method for suppressing the complex strong random noise in petroleum prospecting data.

面对能源地震勘探高信噪比、高分辨率、高准确度要求、复杂地表和近地表地质条件、以及赋存空间越来越复杂的勘探目标，尽可能多的消除随机噪声，高保真恢复有效信号（反射波），提高地震勘探记录信噪比,显得尤为重要。本申请项目拟研发一种利用多道地震记录横向空间相干特性的曲线空间二维时频峰值滤波技术，在消减复杂地震噪声的同时切实保真有效地震信号。通过分析地震同相轴局部走向，构建曲线矢量空间，从而增强矢量空间中有效信号的线性特征，同时保留复杂同相轴弯曲特征。这种基于曲线矢量空间的二维时频峰值滤波方法具有改善一维时频峰值滤波无偏估计的能力，从而达到滤波信号的保真。该技术研究随机噪声非平稳、非高斯、准相似性等复杂特性与二维时频峰值滤波参数之间的作用关系，构建时变加空变的二维时频峰值滤波技术，揭示随机噪声二维时频特征变化、时频分布互项干扰和一致性估计等关键问题，为油气资源探查中复杂强随机噪声消减提供可靠的处理手段

随着地震勘探向深部勘探及非常规油气勘探发展，地震勘探资料信噪比低、随机噪声性质复杂、有效信号能量弱，对地震信号处理技术提出新的要求和挑战。本项目将消减复杂随机噪声并保真恢复有效同相轴作为最终目标，主要研究基于曲线矢量空间的二维时空时频峰值滤波理论及消减复杂随机噪声的滤波技术。该项目利用多道地震记录横向空间相干特性，结合径向道变换思想研究近似同相轴形态的滤波曲线矢量模型及滤波曲线矢量变换方法，设法沿同相轴对地震数据时空时频峰值滤波，增大反射地震信号线性程度，优化同时消减随机噪声和保持有效信号不失真的时空时频峰值滤波算法。项目组系统地分析了匹配同相轴形态的曲线矢量滤波模型构建原理，形成了从局部拟合到整体逼近同相轴，从时空域构建到频域实现的曲线矢量滤波模型，并以此组构时间-曲线矢量空间，建立了基于曲线矢量的时空时频峰值滤波算法。通过定性定量地分析窗长参数、随机噪声特性等因素对滤波结果的影响，根据地震勘探信号邻域特征、结构特征、尺度特征及随机噪声统计特征量等特征建立时空时频峰值滤波算法的参数优化方案。对于实际陆地勘探地震资料中复杂特性随机噪声消减问题，提出具有高时频聚集性的时频分布，利用现代统计方法优化二维时空时频峰值滤波参数，进而提高滤波精度及消减指向性随机噪声、非平稳随机噪声等复杂随机噪声的能力。理论分析和地震记录试算结果表明，该项技术在复杂特性随机噪声情况下弱反射信号恢复方面具有明显优势，滤波结果在信号幅值、主频、连续性等方面均有明显的改善，信噪比提高10dB，项目取得的成果已在国内外期刊上发表。此项技术的研究与应用为油气地震勘技术提供消减复杂特性随机噪声的新方法。