数据驱动型层间多次波预测方法研究  

Study of data-driven interbed multiple prediction.

作　　者：叶月明 赵昌垒[1,2] 姚根顺 胡冰 庄锡进 章学刚[3] 

机构地区：[1]中国石油杭州地质研究院 [2]中国石油勘探开发研究院 [3]长江大学工程技术学院

出　　处：《石油地球物理勘探》2014年第2期244-251,217-218,共8页Oil Geophysical Prospecting

摘　　要：采用表面相关的多次波预测算法(SRME)时,对于层间多次波,是将表面接收到的波场利用基准面重建或CFP算子延拓到产生层间多次波的界面,然后利用SRME方法预测,该方法的缺陷是需要知道产生多次波层界面以上的速度模型。而数据驱动型层间多次波预测(IMP)算法是将层间多次波分解为三个波场分量,即无层间多次波的一次反射波,产生层间多次波界面的一次反射波,及该层界面以下的总反射,此三个波场通过互相关和褶积即可预测出层间多次波。第二个分量和第三个分量均可通过简单的切除得到,第一个分量可通过类似于SRME的基于最小平方能量准则的自适应迭代算法得到,其初始值与第三个分量相同,因此该方法是完全数据驱动的。模型数据试验表明,该方法可有效地预测层间多次波。For interbed multiple with SRME(surface-related multiple elimination),one can think of extending SRME method to interbed multiple predicting(IMP)by applying a full wave-filed redatuming or CFP operator continuation to a interbed multiple generating boundary,however,the velocity-depth model between surface and the layer generating multiple is needed.With the data-driven interbed multiple prediction algorithm,interbed multiple was decomposed into three wave-field components, which are the primaries,primaries of the layer which generating interbed multiple and the reflections below the interbed multiple generating layer. The cross-correlation and convolution of these three wave-fields can predict interbed multiple. The second and third component can be easily obtained by muting,and the first component can be obtained based on least-square energy criterion adaptive algorithm like SRME.Initial value of the first component is the same as the third components,so this method is completely data-driven. Model test shows this method can predict interbed multiple effectively.

关 键 词：数据驱动 IMP 层间多次波 SRME 多次波预测 

分 类 号：P631[天文地球—地质矿产勘探]