Conventional wave equation migrations correct the reflector shape using extrapolation and imaging with initial boundary condition on the surface. The wave equation migration using paraxial approximation can not migrate multiples, S wave, and common sh...
Conventional wave equation migrations correct the reflector shape using extrapolation and imaging with initial boundary condition on the surface. The wave equation migration using paraxial approximation can not migrate multiples, S wave, and common shot gather data. One way to complete this defect is to use a full waveform inversion technique which needs much computing time in calculating the wave fields and partial derivative seismograms for each model guessed iteratively.
In this thesis two new methods are studied which image the subsurface by the partial derivatives, with respect to layer parameters, of the cross-correlation of observed and model data. The one method is to image the subsurface by inner product, cross-correlation, of the observed data and partial derivative seismograms which are calculated from the virtual sources determined from synthetic seismogram for a guessed model and partial derivatives of the guessed model with respect to velocity. The other method is to image the subsurface by crossprrelation of the virtual sources and the back propagated wave fields calculated from the observed data. The cross-correlation in each of the two method has large values at layer interfaces even if the initially guessed model is quite different from true model. Therefore, no iterative modification of the guessed model is necessary by this methods. This fact gains an advantage over the migration using iterative ful! waveform inversion.