基于波长平滑算子的逆时偏移回转波假象去除方法

基于相关成像条件的常规逆时偏移（RTM）不仅会产生低频假象，而且在强速度梯度带会导致通过简单叠后拉普拉斯滤波手段不能去除的回转波假象。本文引入上、下行波分离成像条件，分析得知RTM中回转波假象的产生是缘于强速度梯度带处震源波场上行波与检波点波场下行波的互相关。为了消除该假象且兼顾计算效率，提出一种基于波长平滑算子的回转波假象去除方法。该方法考虑了地震波长因素，通过对速度模型进行波长内平滑，使波场在强速度梯度处的传播特征更接近于真实速度中的波场传播特性。数值试算结果表明，该波长平滑算法能很好地压制RTM中的回转波假象，可在经典的RTM成像框架下进行高效的成像。

Turning wave false image removal based on wavelength-dependent smoothing operator in reverse time migration

The conventional reverse time migration (RTM) based on cross-correlation imaging conditions not only produces low-frequency artifacts,but also generates the false images of turning wave in the strong velocity gradient regions,and the false images cannot be removed by a simple Laplacian filter.In this paper,we introduce the imaging condition by separating up-going and down-going wavefields to analyze the causes of the false images in RTM and it does indicate that the false images of turning wave are generated by the cross-correlation of the up-going source and down-going receiver wavefields in the strong velocity gradient areas.In order to eliminate the false images and take into account the computational efficiency,we propose a false image removal method based on the wavelength-dependent smoothing operator.This method includes the wavelength effect which makes the propagation characteristics of wavefields in the strong velocity gradient areas more close to the propagation characteristics in the true velocity.Numerical experiments demonstrate that the proposed wavelength-dependent smoothing algorithm can suppress the turning wave false images and perform RTM efficiently under the classical imaging framework.