应用预测反滤波迭代法消除多次波

ELIMINATION OF MULTIPLES BY PREDICATIVE INVERSE FILTER ITERATION

地质矿产部石油物探研究所

本文叙述了消除复杂多次波的预测反滤波迭代法。如果某个地区多次波的周期性较差,则用常规预测反滤波方法消除多次波的效果就很不理想。本文通过分析地震记录的自相关函数,利用“线性分解——拟合”的简易方法,对可能存在的周期性较差的不易识别的短程多次波,进行了不同预测距离的预测反滤波迭代处理,使之减弱到最低程度。通过对EM和J地区资料处理表明,本方法对压制短程多次波具有一定效果。迭代法效果的好坏,与预测反滤波参数的选择有很大关系。

This paper mainly describes the method of eliminating the multiples by predictive inverse filter iteration. Generally speaking, the conventional predictive inverse filter won’t be able to give an ideal result to the elimination of multiples with wicked periodicity. But by the inverse filter iteration of different predictive distance, the short-path multiples with wicked periodicity could be reduced to the minimum.Of course, it should be done on the basis of analyzing the autocorrelation function of seismo-grams and making the "linear decomposition-fitting" .The data processing for EM and J areas have shown that the short-path multiples could be removed by the method recommended and the results achieved are fairly good. The selection of parameters will produce a great impact to the results of iteration.