预测反褶积去除多次波的几个理论问题的探讨  

Discuss of several theoretical questions using predictive deconvolution to remove multipule

作　　者：张军华[1] 缪彦舒[1] 郑旭刚[1] 周振晓[1] 单联瑜[2] 梁鸿贤[2] 

机构地区：[1]中国石油大学(华东)地球资源与信息学院 [2]胜利油田分公司物探研究院

出　　处：《油气地球物理》2008年第3期15-19,共5页Petroleum Geophysics

摘　　要：以模型研究为基础,讨论了预测反褶积去除多次波应注意的几个理论问题,得出了一些有意义的结论:①道集中多次波时差是变化的,要去除整道集的多次波,预测步长应以远道为参照;②预测反褶积去除多次波还不完全等同于一般的提高分辨率处理,步长太小的多次波无法准确预测和去除;③算子长度要大于多次波周期,长度太小会使预测结果混入假能量;④级联反褶积效果不理想,建议慎用此技术;⑤层间多次波衰减很快,小于50m的层间多次波无法分辨、难以去除。This article is based on the research of models, and discusses several theoretical issues about using predictive deconvolution to remove multipule. 1） In trace gather, the moveout of multiple is changed. To remove multiple in entire trace gather, the selection of predictive length must take remote traces into account, 2） Predictive deconvolution to remove multiple is not completely 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 researching cascade of predictive deconvolution is far from our perfection, so we must use it carefully. If it is necessary we suggest that using long predictive length first, 5） Peg leg multiple in layers are attenuated very quickly, and it is impossible to remove those multiple that the interval between layers is less than 50m. To remove multiple in high precision, we must use some other methods.

关 键 词：多次波 预测反褶积 级联反褶积 预测步长 算子长度 

分 类 号：P631.443[天文地球—地质矿产勘探] P618.13[天文地球—地质学]