Estimation of Q from surface-seismic reflection data in data space and image space

Advantage

Using central freqency shift to analyze QVO/QVA has several advantages over the previous Q estimation methods (Rickett, 2007; Plessix, 2006; Rickett, 2006; Quan and Harris, 1997), which analyze Q from a stacked trace. First, the spectral amplitude of a stacked trace has a distorted attenuation signature, since the path lengths (hence accumulated attenuation), spectral distortions from NMO stretch, and reflectivity-transmissivity effects vary from one trace to the next for a given reflection event. Performing an analysis along the offset or angle may help reduce such distortion. Second, with the relevant information from different offsets or angles, I do not need the reference/source information that is a prerequisite for the previous methods. Third, a number of applications have already demonstrated the effectiveness of QVO for discrimination of lithology types (Hackert and Parra, 2004; Dasgupta and Clark, 1998), azimuthal discrimination of fractures (Moffat et al., 2009; Clark et al., 2009), and discrimination of time-lapse changes (Clark et al., 2001; Blanchard et al., 2009). In addition, there are also several potentially promising applications for this method. For example, such analysis may reduce the ambiguity between velocity and attenuation, because velocity does not have a frequency-dependent effect on the amplitude, and thus has no central-frequency shift. QVO may also help analyze amplitude variation with offset (AVO), as attenuation effects are always superimposed on AVO signature.

Migration-based Q analysis is used in this proposed method, instead of the ray-based Q estimation methods that were used in previous methods. One advantage of using a migration-based method is that in the presence of complex wave propagation (i.e. multipathing and geometrical dispersion), data-based methods often cannot produce high-quality images, while migration-based methods yield better images because they can handle the complex structure in the subsurface. In addition, the signal-to-noise ratio is higher in the image domain than in the data domain, leading to a higher-quality result from the central-frequency shift.

Estimation of Q from surface-seismic reflection data in data space and image space