With the improvement of automatic CIG picking tool, the accuracy of tomography is mainly limited by the model representation (e.g. model grid size) and the regularization of the inversion solver. In practice, a dense model grid with the grid size less than 20 meters and a 100 meters’smoothing are applied to achieve high resolution velocity updates in some well-conditioned areas (e.g. shallow sediment areas). The small grid size and small model smoothing may cause inversion artefacts in poorly conditioned areas (deep part) and increase the computational cost dramatically. In this abstract, we introduce a non uniform grid, which has a dense sampling in shallow and a sparse sampling in deep to reduce the computing cost.Also we propose a scheme for smoothing along geological structures to stabilize the inversion. We demonstrate with a production example in the Gulf of Mexico that our high resolution tomography is able to improve the subsurface images.
展开▼
