| Migration-based velocity analysis is the most efficient, and accurate velocity inversion technique. It generally involves time-consuming prestack depth migration, and picking of the depth residuals in common-image gathers (CIGs) in each iteration. Two modifications are proposed to minimize the time of prestack depth migration and the picking work in velocity analysis: one approach is to invert the velocity model in layer-stripping style; the other is based on a grid parametrization of the velocity model. Both approaches are based on the idea of parsimonious depth migration, which is the fastest depth migration currently available.; Both approaches have four basic steps: (1) Picking the primary, most consistent reflection events from one reference seismic section or volume. (2) Depending on whether the reference data is 2-D poststack, 2-D common-offset, 3-D poststack, or 3-D common-offset, the corresponding parsimonious depth migration is used to migrate all the picked time samples to their spatial locations and to give their orientations. (3) Ray-tracing defines the CRP gathers for each reflection point. (4) Velocity updating.; For the layer-stripping approach, a small (2-3) number of iterations converge to a 2-D model of layer shape and interval velocity. The computation time of this layer-stripping approach is of the same order as that of the standard (1-D) rms velocity scan method, and is much faster than current iterative prestack depth migration velocity analysis methods for typical field data.; For the grid-based approach, it is not necessary to define continuous reflectors and that the time at any offset (not only zero offset) can be used as the reference time for a reflection. Truncations, and multi-valued layers, which need much effort in the layer-stripping approach, are handled naturally and implicitly in the grid-based approach. Two important features of the proposed algorithms are: the traveltime picking is limited to only a stacked or common-offset section or volume, and it needs to be done only once; there is no need to do intensive prestack depth migration, hence the computation time is orders of magnitude faster than other migration-based velocity analyses. |
