Abstract：In this paper, we present a new method for multi-component seismic data imaging, called elastic Gaussian beam migration. Its imaging accuracy is higher than that of Kirchhoff migration, and its computational efficiency higher than reverse time migration. This migration method first decompose the multi-component data into local plane-wave components with different wave modes and different initial directions by local slant stacks, and then achieves downward-continuation imaging of the vector wavefield using the traveltime, amplitude and polarity of elasto-dynamic Gaussian beams. Based on 2-D elastic Kirchhoff-Helmholtz integral, we derive an elastic inverse wavefield extrapolation formula in terms of Gaussian beams, and extract the imaging result of reflected and converted waves using cross-correlation imaging condition. For the polarization reversal on converted waves imaging sections, we offer a correction method on the propagation and polarization of converted waves near the reflectors. We finally make trials on numerical models to prove the correctness and availability of the presented elastic migration method.
