- Copyright © 2002 Society of Exploration Geophysicists
Azimuthal AVO (AVOA) has become a popular method of fracture detection in both carbonate and sandstone reservoirs. Previous methods often relied on stacking the multiazimuth data into offset and azimuth bins, often using only two azimuth sectors perpendicular to each other. This can mix the data over a significant azimuth range and does not allow for meaningful error estimates. Once the data have been stacked into azimuth bins, information on azimuthal distribution is lost. Each azimuth sector is treated as if it has equal weight, even though one sector may have significantly lower fold and thus a lower signal-to-noise ratio. It has also been suggested that the results of azimuth sectoring analyses are often influenced by acquisition geometry.
In addition to losing information on the acquisition geometry and potentially biasing the results, different estimates of the intercept will be obtained for different azimuths. This is nonphysical because it implies that the P-wave normal incidence reflectivity varies with source-receiver azimuth.
In this paper, I describe and demonstrate a method for computing the azimuthal variation in AVO response without azimuth sectoring the data. This results in each datapoint having equal weight in the inversion. Furthermore, because the equations governing the azimuthal variation of AVO are linear, error estimates can easily be computed and then used to infer the success (or lack thereof) of the AVOA inversion.
A disadvantage is that prestack migration (either time or depth) cannot be performed prior to the AVOA analysis. However, poststack time or depth migration can be used to correctly position events where the structure is not too complex and lateral velocity variations are not too severe. In cases where prestack migration is required, many potential problems must be addressed. For instance, it is likely that ray tracing will be required to compute incident angles. In addition, …