Common Shot, Prestack Inversion and Mode Conversion Analysis of Physical Model Seismic Data.

摘要

Imaging physical model data provides a good test for an inversion algorithm. The physical model data are real wave fields and do not include the simplifications of synthetic data. Also, the parameters of the model are known beforehand so that it is easy to determine how well the inversion works. Here, inversion is a true amplitude Kirchhoff depth migration in the sense that the amplitude of the imaged reflections is proportional to the reflection coefficient. Each shot record in a physical model data set is inverted separately with a common shot, prestack inversion routine with a laterally and depth variable velocity function. Each shot record inversion forms a partial image of the subsurface. The results are then stacked to form a full image of the subsurface. The physical model data set is inverted twice. For the second inversion, the output trace spacing is half the spacing for the first inversion and the output aperture is three times wider than in the first inversion. In both cases, the background velocity field is nearly identical to the actual model. This tests the inversion procedure independent of velocity analysis. Both inversions accurately position reflectors in the model but each performs better on different portions of the data. With a larger inversion output zone, steeper events are imaged better but the increased migration smile noise obliterates some deeper events. Theses. (jhd)