Subtle lithoacoustic inversion of seismic data

Subsurface description can be greatly enhanced by creating detailed models of the spatial variation of lithology and porosity, with layer thicknesses less than 1/4 of the seismic wavelength, integrating seismic, geological and log data. This type of interpretation is possible only on the basis of a subtle acoustic model refined to match a set of seismic traces. Nevertheless, it is very easy to get a good match between reflection seismic and synthetic traces derived from quite different earth models. Furthermore, a slight change in the layer parameters of the initial model or in the application of the inversion technique might produce significant changes in the derived model. Accurate characterization of the subsurface media is highly dependent on a priori knowledge of the lithological and acoustic features of the area, on the quality of the seismic data (frequency range, noise level, etc.) and on the sophistication of the inversion methodology (skilful use of geological information for initial model creation, characteristics of the inversion techniques applied, and choice of inversion parameters). Up to now there have been many positive and many negative results obtained in attempting to incorporate inversion techniques into geological interpretation. It has become necessary to assess in advance the likelihood of a successful modelling result and to recognize situations where the application of inversion is useless or even of negative value, despite a fine match to trace data.