基于奇异值分解的角度域去噪方法

受地层调谐效应和地震数据品质等因素的影响,角度域共成像点道集(angle domain common image gathers,ADCIGs)存在不同程度的波形拉伸和随机噪声干扰。为了提高ADCIGs及其叠加剖面的成像效果,提出了基于奇异值分解的角度域去噪方法。首先对叠前偏移输出的ADCIGs进行奇异值分解,然后对奇异值进行归一化修正,采用累计贡献率的方法确定降噪阶次,从而实现角度域内的信噪分离和噪声压制。在确定降噪阶次时,采用累计贡献率的方法可以直观地判断各奇异值分量对数据的贡献,便于快速选择降噪阶次。理论模型和实际数据的测试处理结果表明,基于奇异值分解的角度域去噪方法适用于具有水平同相轴的ADCIGs,它能有效分离角度域内的随机干扰,并且能压制高角度处的频率畸变,改善大角度数据的品质。对ADCIGs进行基于奇异值分解的角度域去噪,可进一步提高该叠前道集的精度,从而有效改善角度域叠加剖面的信噪比和分辨率,也为基于叠前道集的速度分析和叠前反演提供了更为准确的数据基础。

Due to stratigraphic tuning and the quality of seismic data,angle-domain common image gathers (ADCIGs) are affected by the interference of wave tensions and random noise.A denoising method in the angle domain based on singular value decomposition (SVD) is discussed in this paper,with the aim of improving the quality of ADCIGs and its stack section.First,the SVD is applied to ADCIGs.Subsequently,the order of noise reduction is quantified by the cumulative contribution rate method.All the singular values are normalized and weighted in this method.The contribution of each singular value component to the data can be evaluated intuitively,and the level of noise reduction can be rapidly selected.Finally,signal/noise separation and denoising were performed in the angle domain.Tests on synthetic and field data showed that the proposed denoising method is applicable to ADCIGs with horizontal events,as it can effectively separate the random interference in the angle domain,suppress the frequency distortion,and improve the data quality at high angles.Filtering can further improve the accuracy of ADCIGs,and thus effectively improve the signal-to-noise ratio and the resolution of the imaging profile in the angle domain.This,in turn,can provide more accurate data for velocity analysis and pre-stack inversion based on pre-stack common image gathers.

国家自然科学基金项目(41874123)和中央高校基本科研业务费专项资金(300102268401)共同资助。