Trade-off of Elastic Structure and Q in Interpretations of Seismic Attenuation

Abstract

The quality factor Q is an important phenomenological parameter measured from seismic or laboratory seismic data and representing wave-energy dissipation rate. However, depending on the types of measurements and models or assumptions about the elastic structure, several types of Qs exist, such as intrinsic and scattering Qs, coda Q, and apparent Qs observed from wavefield fluctuations. We consider three general types of elastic structures that are commonly encountered in seismology: (1) shapes and dimensions of rock specimens in laboratory studies, (2) geometric spreading or scattering in body-, surface- and coda-wave studies, and (3) reflectivity on fine layering in reflection seismic studies. For each of these types, the measured Q strongly trades off with the (inherently limited) knowledge about the respective elastic structure. For the third of the above types, the trade-off is examined quantitatively in this paper. For a layered sequence of reflectors (e.g., an oil or gas reservoir or a hydrothermal zone), reflection amplitudes and phases vary with frequency, which is analogous to a reflection from a contrast in attenuation. We demonstrate a quantitative equivalence between phase-shifted reflections from anelastic zones and reflections from elastic layering. Reflections from the top of an elastic layer followed by weaker reflections from its bottom can appear as resulting from a low Q within or above this layer. This apparent Q can be frequency-independent or -dependent, according to the pattern of thin layering. Due to the layering, the interpreted Q can be positive or negative, and it can depend on source–receiver offsets. Therefore, estimating Q values from frequency-dependent or phase-shifted reflection amplitudes always requires additional geologic or rock-physics constraints, such as sparseness and/or randomness of reflectors, the absence of attenuation in certain layers, or specific physical mechanisms of attenuation. Similar conclusions about the necessity of extremely detailed models of the elastic structure apply to other types of Q measurements.