非线性K-L滤波及其在反射地震资料处理中的应用

Nonlinear K-L filtering and its application to seismic data processing

廊坊地矿部物化探研究所, 廊坊102849

Institute of Geophysical and Geochemical Exploration, Langfang 102849

K－L变换可以用来提高地震资料的信噪比。本文分析了基于K－L变换的分类特征提取方法及其与奇异值分解算法的关系后指出,在资料信噪比小于1时,应用K－L滤波方法不能取得良好效果；引入二次波场等非线性运算方法可以进一步改进K－L滤波方法,使其在资料信噪比很低时,仍然能压制噪音成分,保持反射地震信号不受明显畸变。利用噪音极强的深层海洋地震资料进行非线性K－L滤波,结果揭露了海洋地壳消减带上的断裂和破碎区带。

The K-L transform has been introduced to seismic data processing for enhancement of s/nratio of seismic records. I analysis the method of characteristics classification based on the K-Ltransform and its relation to the singular value decomposition,pointing out that it fails when therecords contain very strong noises. Introducing some nonlinear transform,such as quadratic wavetransform,before performing the K-L filtering can improve the filtering procedure, which cansuppress very strong noise while keeps signal clear without serious distortion. The procedure istested by deep reflection marine data of very low s/n ratio,resulting in clear interpretation of acomplicated structure pattern at a subduction zone of the marine crust.