Seismic Sequence Analysis and Attribute Extraction

The variation of frequency content with time is an important seismo-stratigraphic indicator. This paper discusses the analysis of local frequency content of seismic reeection data with quadratic joint time-frequency representations. Signal adaptive quadratic time-frequency representations have considerably better time-frequency localization properties (resolution) than more standard approaches such as the sliding window Fourier transformation or wavelet transform. Two applications of the quadratic type of time-frequency representations are demonstrated: seismic sequence analysis and seismic attribute extraction. The joint time-frequency representation of a seismic reeection pattern is often much more easily interpreted in terms of subsurface stratiication than a time or frequency domain description alone. We show how the time-frequency representation can be used to delineate seismic sequences on the basis of the time-frequency characteristics of the signal. There exists a close relation between complex-trace attribute analysis and quadratic time-frequency representations. In the time-frequency approach the seismic attributes are characteristics of the local spectrum. Extraction of the attributes via the time-frequency representation of the seismic trace leads to considerable improvement of signal to noise ratio the attributes. Furthermore, the classic set of seismic attributes of instantaneous amplitude, phase and frequency, can be easily extended with other parameters describing the local spectrum, such as instantaneous bandwidth and higher order measures.