Towards characterization of intrinsic and stratigraphic Q in VSP data with information measures

Siming Lv, Kristopher A. Innanen

The problem of distinguishing between intrinsic Q and stratigraphic filtering is a classical example of non-uniqueness in seismic data analysis. Two very different mechanisms affecting propagating waves - reverberations between thin layers and transformation of mechanical energy to heat - produce almost identical effects. A version of the Shannon entropy, defined on snapshots of a VSP wave field, has been proposed to discuss these two influences, but it so far has been used to argue that the distinction is not as meaningful as we might think. In this project the entropy calculation is extended in time domain and also to frequency domain, aiming to locate what can be separated between these two effects. Conditional probabilities in which correlation of wave field values with neighbouring values is incorporated, rather than a statistical PDF histograms of single instances of particular wave field values. 1D VSP modelling codes and a range of well logs are used to investigate the separability of intrinsic/extrinsic sources of attenuation and dispersion. Progress of this kind will have significant impact on reservoir characterization where viscosity changes are expected: such changes can be tied to intrinsic Q but not extrinsic Q. The results, in which the various processes produce noticeable differences in entropy, indicate that this is a promising line of inquiry. Parameters like bin size have a large effect on the entropy, especially at late times, so that the footprint of bin size is studied in detail.