基于Born波路径的高斯束初至波波形反演

Gaussian beam first-arrival waveform inversion based on Born wavepath

为了提高表层速度反演精度,本文提出了一种新的波形反演方法.该方法只利用初至波波形信息以减少波形反演对初始模型的依赖性,降低反演多解性与稳定性.由于只利用初至波波形信息,所以该方法利用高斯束计算格林函数和正演波场,以减少正演计算量.为了避免庞大核函数的存储,该方法基于Born波路径,利用矩阵分解算法实现方向与步长的累加计算.将此基于Born波路径的初至波波形反演方法应用于理论模型实验,并与声波方程全波形反演和初至波射线走时层析方法相对比,发现该方法的反演效果略低于全波形反演方法,但明显优于传统初至波射线走时层析方法,而计算效率却与射线走时层析相当.同时,相对于全波形反演,本文方法对初始模型的依赖性也有所降低.

We propose a new waveform inversion method to improve the accuracy of the near surface velocity structure.Only the first-arrival waveform is used in this method to decrease the nonlinearity between wavefield and model parameters and the dependency of waveform inversion on the starting model.To increase the computational efficiency of forward calculation,we use Gaussian Beam（GB）method to calculate the Green function and the first-arrival wavefield.To avoid the storage of the huge Fr6 chet matrix,based on Born wavepath,we propose the calculation of the direction and step length in waveform inversion using matrix decomposition,accomplished by accumulation.We then apply this new first-arrival waveform inversion method on a numerical experiment.The comparisons between this method and traditional acoustic full waveform inversion（FWI）and ray-based first-arrival traveltime tomography（RFTT）show that it can obtain a result a little worse than FWI while much better than traditional RFTT,while the computational efficiency is almost the same as RFTT.Meanwhile,the dependence of this method on the starting model is much weaker than FWI.