High-Resolution Shallow Seismic Structure Imaging Using Grid-Based Nonlinear Refraction Traveltime Tomography

We develop a nonlinear refraction traveltime tomography technique that can rapidly reconstruct seismic<br>velocity distribution in a large 2-D gridded model. For the traveltime and raypath calculations, we improve<br>a graph-theoretical method by placing nodes on each grid boundary with an optimized distribution pattern.<br>This method greatly reduces errors in traveltime and raypath calculations for complex velocity models.<br>We pose a nonlinear objective function for inversion, which includes not only the traveltime misfit and<br>model curvature roughness norm, but also traveltime gradient misfit norm. Our numerical experiments show<br>that fitting the gradients of the traveltime curves in addition to the traveltime itself can better resolve the<br>velocity contrasts across interfaces. We apply Newton’s method to minimize the nonlinear objective function<br>iteratively. The use of Tikhonov regularization allows to perform a global inversion and reconstruct the whole<br>model with minimum model curvature roughness.<br>We apply this refraction traveltime tomography technique to image shallow bedrock topography at a<br>coastal site near Boston, Massachusetts. We will demonstrate that performing nonlinear tomography can<br>effectively reconstruct the subsurface image for complex velocity structure.