奇异值分解技术在绕射波分离成像中的应用研究

Separation and imaging of seismic diffractions using singular value decomposition

实现绕射波分离成像,要利用反射波和绕射波的特征差异,不同方法所利用的特征差异不尽相同。在地下反射层位水平连续的假设条件下,基于反射波和绕射波在相干性方面的差异,提出一种基于奇异值分解的绕射波分离成像方法。首先通过炮域动校正拉平反射波,使其相干性增强,而绕射波呈现非线性形态,相干性较弱,两者在相干性方面差异明显。基于此,再利用奇异值分解重构出反射波,并将其从地震数据中减去,从而分离出绕射波。为了在有效压制反射波的同时,尽可能地保护绕射波能量,在重构反射波时采用了一种迭代的方式。最后对数据进行反动校,得到绕射波单炮记录,继而实现绕射波成像。数值模型和实际资料应用结果表明,该方法能够有效压制强反射,提高小尺度非均质体的识别精度,有良好的应用价值。

The difference between the characteristics of reflected waves and those of diffracted waves is utilized in current methods for separating and imaging of seismic diffractions.Under the assumption that the reflective horizons in the subsurface are horizontal and continuous,an approach based on singular value decomposition and using the coherence difference is proposed in this study.First,the NMO corrections in shot gathers were used to flatten the reflections and enhance lateral coherence;the diffractions showed weaker coherence due to their nonlinearity.Subsequently,singular value decomposition was applied to construct the reflections,which were then subtracted from the data to obtain the diffractions.In order to effectively suppress the reflections and better preserve the diffractions,an iterative method was utilized.Finally,after reversing the NMO corrections,the shot gathers that approximately contained the diffractions were obtained;subsequently,the information was used to produce diffraction images.The results obtained by processing both synthetic and actual data showed that the proposed method can effectively suppress strong reflections and improve the identification of small-scale non-homogeneous objects.