Dispersion Estimation By Nonlinear Optimization of Beam-Formed Fields

摘要

A method for estimating velocity dispersion in seismic surface waves in massive 3-D data sets (401) that improves upon auto-picking of a curve along the peak or ridge of the magnitude of the beam-formed field (402). The seismic data are transformed to the frequency-slowness domain, where nonlinear constrained optimization is performed on the transformed data. The optimization matches a nonlinear mathematical parametric model (403) of a beam-formed field to that in the transformed data, adjusting the parameters each iteration to reduce mismatch (404). Dispersion curves are determined by the center of the beam in the optimized models (405). A preferred nonlinear parametric mathematical model is a Gaussian-shaped beam or a cosine-tapered boxcar beam.