【摘要】 采用新近提出的多次波射线追踪正演算法,结合共轭梯度法求解带约束的阻尼最小二乘最优化反演问题,分析讨论了利用多震相走时资料进行联合反演成像的方法及技术.考虑到不同震相走时的拾取误差不同,反演算法中引入了不同震相种类数据的权系数;由于同时反演速度模型和反射界面起伏中不同模型参数变化对走时影响程度的不同,Jacobi偏导矩阵元素中引入了不同参数的归一化因子;另外,为了克服射线密度过大(或过小)区域速度模型的过度(或欠)更新问题,反演算法中引入了等权射线密度的概念.几种数值模拟实例表明(含噪声敏感性试验):多波走时的联合或同时反演成像技术是一种提高走时成像空间分辨率,进而降低重建模型失真度行之有效的方法.更多还原

【Abstract】 The newly developed multiple (reflected, transmitted and converted) arrival tracking algorithm and the conjugated gradient method are used to solve the constrained, damped least squares problem. We discuss and analyze the simultaneous inversion algorithm with different travel time datasets. In the inversion process we introduce different weighting factors according to the different picking errors of different arrivals, and normalize different parameters in Jacobian matrix elements to balance the influence on travel time due to the different contributions of velocity variation and depth of reflector. In order to overcome the velocity overestimation (or underestimation) in some regions where the ray density is very high (or low), the concept of equal weighted ray density is introduced into inversion process. The numerical results (including the noise-sensitivity tests both in travel times and velocity model perturbation) show that it is a practical and efficient way to improve the spatial resolution and reduce the artifact in reconstructing velocity distribution and reflective interface.更多还原