相似系数阈值滤波的数据驱动控制束偏移

高斯束偏移不仅具有接近波动方程偏移的精度，而且具有Kirchhoff偏移灵活、高效的特点。然而当实际地震采集数据中含有较强噪声时，易产生偏移假象而影响成像质量。为此，在传统高斯束偏移的基础上，根据有效信号和干扰信号在τ-p域中的相干性差异，发展了一种基于相似系数阈值滤波的数据驱动控制束偏移方法。采用数据驱动策略，在高斯束偏移成像过程中，先计算τ-p道集的相似系数，再通过设定相似系数阈值控制干扰信号，从而降低偏移剖面中的随机噪声；控制束偏移可以直接提取角度域共成像点道集，无需复杂的角度映射变换且具有更高信噪比。模型测试及实际资料处理结果表明：对于低信噪比数据，控制束偏移剖面的信噪比明显高于常规高斯束偏移，但会损失相对振幅的可靠性；尽管控制束偏移在τ-p道集的滤波过程增加了一定的计算量，但总体与常规高斯束偏移方法的计算效率相当；相似系数阈值参数选取十分关键，阈值较小时偏移噪声较强，但过大阈值也可能压制部分有效信息或产生偏移假象，选取合适的阈值参数才能得到较理想的偏移剖面。

A data-driven controlled beam migration based on the semblance threshold filtering

The Gaussian beam migration (GBM) has been recognized as a robust and versatile depth imaging tool with accuracy comparable to the wave-equation migration and with flexibility and efficiency comparable to Kirchhoff migration.However,migration artifacts may occur when strong ambient and random noise is usually mixed in seismic data and difficult to be distinguished in GBM,which will seriously affect the imaging.Aiming at the problem,we propose a data-driven controlled beam migration (CBM) method based on the semblance analysis in this paper.Based on the semblance difference between signal and disturbance in the τ-p domain,we first derive a formula of the semblance in the τ-p domain according to the expression in signal analysis,and then we develop a filtering method using the semblance threshold.Meantime,we adopt a data-driven strategy to eliminate the disturbance during the Gaussian beam imaging procedure,thereby reducing migration noise.Moreover,CBM can directly extract high signal-to-noise ratio (SNR) angle domain common image gathers (ADCIGs) without complex angle mapping transformations.Model and field data tests suggest that the proposed method can improve the migration SNR of poor SNR data to a certain extent with efficiency comparable to GBM,but CBM may lose the reliability of some amplitude.Furthermore,selecting an appropriate semblance threshold is very critical.If the threshold is too small,the noise may become strong.However,the excessive threshold may also suppress part of the effective information or generate artifacts.