Self-adaptive smoothing surface wave tomography based on model sensitivity: methodology and application

摘要

Surface wave tomography usually adds a smooth term into the objective function to make the edge smoothing. This paper will make an improvement on the inversion method under Gaussian probability distribution condition during the iterative process of near-surface inversion, which is based on ray tracing, using the sensitivity parameters in G matrix as spatial smoothing constraints. We try to select self-adapted parameters, following the laws of physics. These parameters include damping factor, model deviation, smoothing factor, and sensitivity parameters. Improved inverse algorithm will be used in numerical simulation and real data, from ambient noise of Taipei Basin of Taiwan, to achieve the near-surface shear wave velocity structure. Compared with the earlier research results in this region, we find the same points and differences. By the process above, we verify the reasonability of the selection of the self-adaption parameters, validity of the improved inverse algorithm and the reliability of the code. The improved algorithm provides a fast, self-adapted, less empirical and universal tool and method to study the near-surface shear wave velocity structure.