Diffracted seismic waves may be used to help identify and track geologically heterogeneous bodies or zones. However, the energy of diffracted waves is weaker than that of reflections. Therefore, the extraction of diffracted waves is the basis for the effective utilization of diffracted waves. Based on the difference in travel times between diffracted and reflected waves, we developed a method for separating the diffracted waves via singular value decomposition filters and presented an effective processing flowchart for diffracted wave separation and imaging. The research results show that the horizontally coherent difference between the reflected and diffracted waves can be further improved using normal move-out (NMO) correction. Then, a band-rank or high-rank approximation is used to suppress the reflected waves with better transverse coherence. Following, separation of reflected and diffracted waves is achieved after the filtered data are transformed into the original data domain by inverse NMO. Synthetic and field examples show that our proposed method has the advantages of fewer constraints, fast processing speed and complete extraction of diffracted waves. And the diffracted wave imaging results can effectively improve the identification accuracy of geological heterogeneous bodies or zones.

中文翻译：

---

通过SVD滤波器分离衍射波

衍射地震波可用于帮助识别和跟踪地质异质的物体或区域。但是，衍射波的能量比反射的能量弱。因此，提取衍射波是有效利用衍射波的基础。基于衍射波和反射波传播时间的差异，我们开发了一种通过奇异值分解滤波器分离衍射波的方法，并提出了一种有效的衍射波分离和成像处理流程图。研究结果表明，使用法向偏移（NMO）校正可以进一步改善反射波和衍射波之间的水平相干差异。然后，使用带秩或高阶近似来抑制具有更好横向相干性的反射波。随后，在通过逆NMO将滤波后的数据转换为原始数据域之后，实现了反射波和衍射波的分离。综合和现场算例表明，该方法具有约束少，处理速度快，衍射波完全提取的优点。衍射波成像结果可以有效提高地质异质体或区域的识别精度。