基于多次波和自适应预测误差滤波器的地震近炮检距数据重建方法  

Reconstruction of near offsets based on the multiple and adaptive prediction error filter

作　　者：刘财[1] 刘一[1] 刘洋[1] 王典[1] 陈常乐[1] 

机构地区：[1]吉林大学地球探测科学与技术学院,长春130026

出　　处：《地球物理学报》2017年第5期1988-1999,共12页Chinese Journal of Geophysics

基　　金：国家自然科学基金项目(41522404;41430322;41274119)资助

摘　　要：地震数据采集过程中,受野外施工条件的制约,往往很难获得完整的地震波场,尤其近炮检距数据的缺失尤为严重.当前,很多地震数据处理方法的应用都依赖于近炮检距数据,如何对近炮检距的缺失数据进行重建,是一个重要的研究课题.本文通过多次波和一次波的互相关构建虚拟一次波,利用数据本身的波场信息,对缺失的近炮检距数据进行插值重建.由于通过多次波构建的一次波与真实的一次波存在振幅和相位方面的差异,提出通过基于非平稳自回归过程的自适应预测误差滤波器来表征虚拟一次波的能量谱,利用最小二乘反演方法重建近炮检距缺失数据,自适应预测误差滤波器通过求解正则化约束下的数学欠定问题来实现局部自适应特征.通过对Sigsbee2B模型和实际数据的测试结果表明新方法可以合理地重建复杂的近炮检距缺失数据.Abstract：In seismic data acquisition, it is usually difficult to obtain the complete seismic wave field due to the restriction of field work conditions and the missing data at near offset are particularly valuable. Currently, many applications of seismic data processing methods depend on the reconstruction of near offset data, and how to reconstruct the missing data at near offset is an important issue of research. In this paper, we generate the pseudoprimaries through the cross-correlation of multiples and the primaries and reconstruct the missing near offset data depending on the data itself. Due to the differences between the pseudoprimaries and the ideal primaries in amplitude and phase, we propose using the adaptive prediction-error filter to characterize the energy spectrum of the pseudoprimaries, which is based on nonstationary autoregression, and then using the least-squares inversion to reconstruct the missing near offset data. The adaptive prediction-error filter achieves local adaptive features by solving underdetermined problems with regularization constraints. The test results of Sigsbee2B model data and the field data show that the new method can reasonably reconstruct the complex near offset missing data.

关 键 词：近炮检距数据重建 虚拟一次波 自适应预测误差滤波器 非平稳自回归 

分 类 号：P631[天文地球—地质矿产勘探]