The birth of spectral decomposition

Abstract

It was definitely a eureka moment, in fact several eureka moments along the way. The first came in 1991–1992 when I was working in the technology group at Amoco Canada. A key processing step before handing seismic data to the interpreters was shaping the source wavelet in each 2D line to a consistent spectral shape. Some of the data sets contained strong coal reflections overlying the geologic zone of interest. It was very important to stay away from such coal reflections when doing the spectral matching/shaping because the interference from these strong coal reflections would contaminate the wavelet information. Such contaminated spectra seemed to show more about the coals than about the wavelet we were trying to characterize. So the thought was, if a short window around the coals tells us something about the coals, then why don't we see if a short window around a target zone of interest tells us something about that zone of interest. A 2D testline across well control seemed like a good place to start investigating this idea. The resulting trace-by-trace, short window amplitude spectra revealed a great deal of interesting variability. The eureka moment occurred as soon as I saw that those variations in spectral content seemed to be associated with geologic heterogeneity. Repeating the experiment on a 2D stratigraphic seismic model showed interference patterns that were similar to those observed in the real data. At that point, I was hooked and knew there was so much more potential under that rock.