论文部分内容阅读
全波形反演中构建常规梯度算子过程中需要三步骤:震源子波的正向传波场传播,波场残差的反传波场和波场互相关构建梯度算子,其过程存在数据量大、效率低等缺点,为提高反演的效率,本文针对常规时间域梯度算子进行优化,提出了基于特征能量的梯度算法.在正传过程中计算每一个网格点上的正传波场的最大激发能量及其对应的时间步,保存一个子波时间长度的利用特征能量以构建梯度算子.在构建梯度算法中利用保存的子波长度的特征能量进行构建梯度算子.该算法无需保存震源的正传波场,可以减少运算过程中的磁盘读写,提高全波形反演的计算效率.在Mamousi模型梯度测试和实际资料的反演测试中表明:该算法在可以保证梯度算子的精度,具有数据读写量小的优点,效率高的优点.
Three steps are required in the process of constructing a conventional gradient operator in a full waveform inversion: the forward propagating wavefield of the source wavelet, the inverse wavefield of the wavefield residual and the cross-correlation of the wavefield to construct a gradient operator, and the data of the process exists In order to improve the efficiency of inversion, aiming at the optimization of conventional time domain gradient operator, a gradient algorithm based on characteristic energy is proposed, and the normal distribution of each grid point The maximum excitation energy of the wave field and its corresponding time step, the use of feature energy of a wavelet time length is saved to construct a gradient operator, and the constructed gradient operator is constructed by using the preserved characteristic energy of the wavelet length in the construction gradient algorithm. The algorithm does not need to preserve the orthostatic wavefield of the source, which can reduce disk read and write during the operation and improve the computational efficiency of the full waveform inversion.In the Mamousi model gradient test and actual data inversion test, it is shown that this algorithm can guarantee the gradient The accuracy of the operator, with the advantages of small data read and write, high efficiency.