Full waveform inversion (FWI) can be viewed as an iterative cycle involving forward modelling, pre-stack migration, impedance inversion and velocity model updating. At each stage of the process there are many factors affecting the outcome. Among the most important are the type of modelling (acoustic versus elastic), derivation of the initial velocity model, the inherent differences between field data and numerically modelled data, and the conditioning of the field seismic data to be inverted.
Our attempts to derive an initial velocity model suggest that the integration of a refraction tomography velocity model with well log data provides a better initial velocity model than well log data alone as the sub-weathering velocities need to be included to help make a match between the first breaks of the field and modelled data. Initial comparisons of field and modelled shot gathers confirms that conditioning of the field seismic data plays a large role in the successful matching of field and modelled data.
View full article as PDF (0.52 Mb)