Generalized Differential Semblance Optimization
Generalized Differential Semblance Optimization: Wave-equation traveltime and waveform tomography first inverts the smoothly varying traveltimes of selected events to obtain the low-wavenumber estimate of the velocity model. Then, the rapidly varying waveforms are inverted during the later iterations to estimate the high-wavenumber portion of the velocity model. This physics-based multiscale procedure can mitigate the tendency for iterative waveform inversion to get stuck in a local minimum. Multiscale physics can be adapted to differential semblance optimization (DSO) in the image domain. That is, the space-lag offset can be identified in the subsurface-offset domain as an implicit function of velocity. It describes the smoothly varying moveout of the migration image in the subsurface-offset domain, which is analogous to the smoothly varying travel-time residual of a reflection event in a shot gather.