基于最小走时射线的LSQR成像反演方法研究

Study on LSQR Inversion of Tomography Method Based on Shortest Traveltime Ray Tracing

给出了基于最小走时射线追踪方法及最小二乘正交分解法(LSQR)反演算法的层析成像技术的具体实现过程;建立了有速度异常区域的假定物理模型并对反演结果进行误差分析。结果表明:异常区相对误差普遍高于平均相对误差,节点速度值往往偏近于模型背景值;模型平均相对误差为2.34%～2.59%,相关系数为0.84～0.85,成像结果能较好反映模型情况;对成像结果进行分析还表明由于射线分布不均及射线覆盖程度不足等原因将引起图像局部失真,可以通过完善射线布置情况等方法得以改善。

The concrete implement process of tomography,which based on shortest traveltime ray tracing technique and LSQR iterative method,was given. Assumed physical models with abnormal velocity areas were upbulit. Result from difference analysis of inversion indicated that： （i） relative difference of abnormal velocity areas is higher than average relative difference,velocity of nodes always leaned to background velocity; （ii） average relative difference is 2.34%～2.59% ,and correlation coefficient is 0.84～0.85. （iii） result of tomography can well reflect the models. （iv） asymmetry and shortage of radial may induce image distortion and should be meliorated by ameliorating radial disposal.