基于Poynting矢量行波分离的最小二乘逆时偏移

Least-squares reverse time migration with wavefield decomposition based on the Poynting vector

因为在逆时偏移中基于双程波动方程构建震源波场和检波器波场,所以在波场延拓过程中地震波遇到波阻抗界面时,背向发育的反射波会与正常传播的波场互相关产生较强振幅的低频噪声.这一特点使得以逆时偏移为基础的最小二乘偏移方法在梯度计算时同样存在着低频噪声的干扰,从而导致反演收敛的速度减慢.考虑到计算量和存储成本的因素,本文借助Poynting矢量良好的方向指示性实现波场的上下行波分离,并在早期迭代的梯度计算中只保留震源波场和检波器波场沿不同垂直方向传播的组分之间的互相关,有效避免了成像噪声的干扰,提高了算法收敛的速率.数值算例验证了方案的有效性.

Because the source wavefield and receiver wavefield are constructed based on the two-way wave equation in the reverse time migration,the reflection waves developed in the back direction can cross-correlate with the normally propagating waves and generate low-frequency noises with high amplitude when the seismic waves meet the impedance interface during the wavefield extrapolation.This feature makes the least-squares migration based on reverse time migration is also disturbed by low-frequency noise during gradient calculation,which leads to slower inversion convergence.Considering the calculation amount and storage cost,we utilize the Poynting vector that represents the wave propagation direction to achieve wavefield decomposition.Then only retaining the cross-correlation between the parts of the source wavefield and the receiver wavefield in different vertical directions in the gradient calculation of the early iteration can suppress the imaging noise and improve the rate of algorithm convergence.Numerical examples verify the effectiveness of this scheme.