Calibrating well logs to VSP attributes: interval velocity and amplitude

Abstract

The vertical seismic profile (VSP) occupies a unique niche in the discipline of reservoir characterization. It provides a level of resolution and a volume of investigation in between those offered by surface seismic data and well logging tools. The relationship between VSP and surface seismic experiments is obvious: The most common use of VSP is in high-resolution reflection imaging and tying surface seismic sections to depth. The relationship between VSP and acoustic well logging is less well explored. Interval velocities from check shot VSP are often used as coarse equivalents of sonic logs, but significant discrepancies are often seen. There are several reasons why we should not expect VSP interval velocities and sonic logs to match perfectly. Many of these reasons are discussed in Stewart et al. (1984) and include the volume of rock sampled by each method, instrumental biases, and wave propagation effects such as anisotropy, short-path multiples, and velocity dispersion. Because there is some interesting physics behind some of these potential discrepancies, several people have proposed using the differences between sonic log and VSP velocities to extract information about the formation. Because of the larger volume of investigation, most of which may be considered to be undisturbed by the presence of the borehole, VSP is often considered to be more reflective of the true formation properties. Thus, the calibration of sonic logs to VSP is also of importance. Another use of VSP which is gaining in popularity is the direct extraction of attenuation information, or Q values. This typically involves comparing downgoing wave amplitudes at successive receiver locations (often in the frequency domain), and computing a Q value for the interval. Unfortunately, scattering and other elastic effects often dominate the amplitude fluctuations, frequently resulting in nonphysical negative Q values. Recently, there has been progress in using elastic models …