For unconventional resource plays, the delineation of fractures can be critical in positioning drilling locations. Fluid-filled fractures and cracks directly modify the effective impedance of rocks, attenuate amplitude and distort seismic spectrum. Thus, attenuation estimation is a promising way to characterize fracture system.
Existing methods for seismic attenuation are usually based on "Constant Q" model, which ignores the interference from reflectivity anomalies. For unconventional reservoirs, the spectrum of the reflected wave may be affected by the presence of thin beds (shales) in the formation, which makes Q estimates less reliable.
We employ a non-stationary Q model to characterize attenuation, and correct the reflected spectrum by using inverted reflectivity sequence based on well logs to remove local thin-bed effects from seismic reflection data. In synthetic examples, significantly less variance in the estimated values and fewer unphysical negative Q values are obtained. Following the workflow, we also applied attenuation estimation on Barnett shale data. The recovered Q estimates have a good correspondence with the production data. Though, the attribute is the average over a target horizon, this may be sufficient to find evidence of fluid-filled fractures, or change in lithology.