论文详情
基于超完备子波库的单道匹配追踪方法在自适应分解地震道时未能分离有效信号和噪声,故难以直接、准确地去除噪声。为此提出了基于稀疏反演的多道匹配追踪去噪方法,以多个连续地震道为研究对象,首先采用多方向矢量中值滤波法计算同相轴倾角,以中间道为标准将同相轴校正后再水平叠加;然后采用基于基追踪降噪的稀疏反演方法分解叠加道,选择与叠加道中有效信号匹配的子波作为中间道匹配追踪分解的子波库;最后根据其它道相对中间道的校正量,对中间道匹配追踪分解的子波库进行反校正,分别构建各道匹配追踪分解的子波库。将基于有效信号匹配子波构建的子波库代替超完备子波库,在搜寻最佳匹配子波时选择性地提取地震道中有效信号,可直接分离有效信号和噪声。单道信号测试结果验证了该方法去除噪声的可行性;合成和实际地震数据应用结果表明,该方法相较于单道匹配追踪方法具有更好的去噪效果,且去噪后剖面的横向连续性更好。
Single-channel matching pursuit algorithm based on the over-complete wavelet library does not distinguish effective signals and noise in the process of adaptively decomposing seismic traces.Hence,it is difficult to remove noise directly and accurately.This paper,therefore,proposes a multi-channel matching pursuit denoising method based on sparse inversion.The method takes multiple continuous seismic traces as the research objects.Firstly,multi-directional vector median filtering method is used to calculate the inclination angle of seismic events.These events are corrected to be horizontal using the central trace as the standard,and then stacked.The sparse inversion method based on the basis pursuit denoising is then used to decompose the stack trace.Next,the wavelets matching the effective signal in the stack trace are selected as the wavelet library of the central trace matching pursuit decomposition.Finally,according to the correction amount of other traces relative to the central trace,the wavelet library of the central trace is inversely corrected to construct the wavelet library of each trace matching pursuit decomposition.The wavelet library constructed by matching wavelets of the effective signal replaces the over-complete wavelet library.In the process of searching for the best matching wavelet,the effective signal components in the seismic trace are selectively extracted,so that the effective signal and noise can be directly distinguished.The single-channel signal test result verified the effectiveness of the proposed method to remove noise.The results of the synthetic and real seismic data showed that the proposed method is better at denoising than the single-channel matching pursuit algorithm and can improve the lateral continuity of the seismic sections after denoising.
中石化科技部“顺北油田碳酸盐岩断裂体系及缝洞体刻画研究”项目(P17057-9)资助。