Spectral recomposition is an improved methodology for generating forward seismic models of outcrop analogs that make use of the full range of frequencies found within real-world seismic volumes. It overcomes the issue common to conventional forward seismic models in which single-peak frequencies are modeled, with the resultant models not capturing the range of frequency content that is found in the real world image. This variable frequency content of real-world seismic volumes has long been known to contain more detailed information within different spectral bands, which forms the basis of the spectral decomposition process, in which different spectral frequencies can correlate to different stratigraphic thicknesses, for example.

In this article, we introduce and illustrate the use of a new method termed spectral recomposition. Spectral recomposition is a poststack seismic method operating in the frequency domain that recomposes forward seismic models generated at separate peak frequencies to derive a final image, which has a frequency spectrum similar to that of a targeted real-world image. Using this method, forward seismic models can be tailored to specific real-world images and be used as a more appropriate comparison between the known outcrop geometries and those that are imaged in the subsurface.