预测反褶积去多次波几个理论问题探讨

DISCUSSION OF SEVERAL THEORETICAL QUESTIONS TO REMOVE SEISMIC MULTIPLES USING PREDICTIVE DECONVOLUTION

这里以模型研究为基础,探讨了预测反褶积去多次波应注意的几个理论问题,得出了有一定价值的几条结论:①道集中多次波时差是变化的,要去除整个道集的多次波,预测步长应以远道为参照;②预测反褶积去多次波还不完全等同于一般的提高分辨率处理,太小步长的多次波是无法准确预测并去除的;③算子长度要大于多次波周期,太小算子长度会使预测结果混入假的能量;④模型研究级联反褶积效果并不理想,因此要慎用此项技术,如果要做级联反褶积,建议先大步长后小步长;⑤地层中的层间多次波衰减很快,要分辨小于50 m的层间多次波,几乎是不可能的,要进行高精度的多次波去噪,还须借助其它方法。

This article deals with predictive deconvolution to remove seismic multiples based on numerical modeling,and discusses several theoretical issues in this research with the conclusions as:(1) In trace gather,the moveout of multiple is changeable and to remove multiples in entire trace gather,the selection of predictive length must take far traces into account;(2) Predictive deconvolution to remove multiple is not completely the same as other processing of enhancing resolution,and it is difficult to attenuate multiple exactly using a very short predictive length;(3) The length of predictive operator must larger than the period of multiple,because it maybe increases fake energy after processing;(4) The result about the cascade of predictive deconvolution is far from the perfection,so that it must be used carefully and it is also suggested to use long predictive length first if it is necessary;(5) Peg leg multiple in layers are attenuated very quickly,and it is impossible to remove those multiple that the interval thickness between layers is less than 50m,and some other methods are in need to remove the multiples in high precision.