最小平方滤波法时间域全波形反演  

Time-domain full waveform inversion based on least square filter

作　　者：白璐[1] 韩立国[1] 张盼[1] 胡勇[1] 

机构地区：[1]吉林大学地球探测科学与技术学院,吉林长春130011

出　　处：《石油地球物理勘探》2016年第4期721-729,4,共9页Oil Geophysical Prospecting

基　　金：国家"863"计划重大项目(2014AA06A605);国家自然科学基金项目(41374115)联合资助

摘　　要：全波形反演是一种利用叠前地震记录的全波场信息重建地下介质物性参数的高精度速度建模法,但其反演效果常会受到"跳周"现象的影响。本文提出一种用最小平方滤波做数据预处理的反演方法,有效减弱"跳周"现象对反演结果的影响。该方法利用最小平方滤波减小模拟波场与观测波场之间的相位差,由滤波后波场构建一种新目标函数,使反演过程稳定地朝着全局极小点处收敛。数值模拟试验结果表明:基于最小平方滤波的全波形反演能有效改善反演效果,尤其在模型深部或大炮检距范围的效果更明显;新目标函数相比常规L2范数能更稳定地下降,不易陷入局部极小,避免了常规全波形反演中易出现的严重"跳周"现象。Conventional full waveform inversion often suffers from cycle skipping. In order to solve this problem, we propose a time-domain full waveform inversion based on least square filter. First we apply the least square filter in the preprocessing stage for full waveform inversion and construct a new objective function to make the inversion process toward global minimum point. This approach effectively avoids the problem so that the inversion process gets into local minima. Experimental results show that this full waveform inversion based on least square filter can better reconstruct velocity model, especially for deep layers and large offsets. The new objective function can stably descent and greatly reduces the possibility of suffering from cycle skipping. © 2016, Editorial Department OIL GEOPHYSICAL PROSPECTING. All right reserved.

关 键 词：全波形反演 最小平方滤波 目标函数 矩阵快速正演 最速下降法 共轭梯度法 

分 类 号：P631[天文地球—地质矿产勘探]