在全波形反演中,深层区域的反演效果明显不如浅层区域的反演效果好。针对这一问题,在不增加计算量的前提下提出了一种新的能量加权梯度方法,引入了反偏移方法,利用反偏移方法计算得到的加权能量对梯度进行预处理。这种改进方法克服了常规能量加权梯度方法计算加权能量不准确、迭代收敛慢的缺陷,可以在提高深、浅层整体反演效果的同时加快迭代收敛速度。模型试算结果表明,基于改进后能量加权梯度方法的全波形反演效果及迭代收敛速度明显优于常规的能量加权梯度方法。
In full waveform inversion,the inversion effect in deep layers is not as good as shallow layers.In order to solve this problem,we proposed a new energy weighted gradient method without increasing the computation amount,which introduced inverse migration to compute the weighted energy and then pre-process the gradient by using the energy.The method can slove the inaccurate weighted energy and slow iteration speed in conventional method,and improve the whole inversion effect in shallow and deep layers while speeding up iteration velocity.The model testing results indicate that our method can achieve better full waveform inversion effect and higher iteration speed compared to conventional energy weighted gradient method.