应用修正正交有限内存拟牛顿算法的全波形反演

全波形反演(FWI)通过最小化观测数据与计算数据之间的误差，得到高分辨率的地下模型参数。正则化技术常用来克服FWI的不适定性。复杂地质模型可能同时具有平滑特征及锐利边界，单一的正则化方法往往不能得到令人满意的反演结果。为此，针对带有混合正则化的目标泛函不可微性，提出一种修正正交有限内存拟牛顿方法，求解相应的正则化目标泛函。在具有复杂构造的修正Marmousi模型以及BG Compass模型上进行数值模拟试验，且与不带有正则化的全波形反演及邻近有限内存拟牛顿方法进行比较，结果表明所提修正正交有限内存拟牛顿算法在计算效率及定量分析上具有明显的优越性。

Modified Orthant-Wise Limited-memory Quasi-Newton algorithm for full-waveform inversion

Full-waveform inversion (FWI) obtains high-resolution parameters of subsurface models by minimizing the errors between observed data and calculated data. Regularization techniques are o-ften used to overcome the ill-posedness of FWI. However, complex geological models are likely to have sharp boundaries and smooth characteristics at the same time, in which case a single regularization method usually fails to deliver satisfactory inversion results. Given the non-differentiability of the objective function involving hybrid regularization, this paper proposes a modified Orthant-Wise Limited-memory Quasi-Newton method for solving the corresponding regularization objective function. Numerical simulation experiments are conducted on the complex-constructed modified Marmousi model and BG Compass model, and the proposed method is compared with FWI without regularization and the proximal Limited-memory Quasi-Newton method. The results show that the proposed algorithm has obvious advantages in computational efficiency and quantitative analysis.