Inversion of walkaway VSP data in the presence of lateral velocity heterogeneity

TL;DR

This paper develops a method to invert walkaway VSP data for anisotropic parameters while accounting for weak lateral heterogeneity, improving the accuracy of subsurface anisotropy characterization.

Contribution

It introduces a downward continuation technique that corrects for lateral heterogeneity, enabling stable inversion of anisotropic parameters in dipping fracture models.

Findings

Corrected VSP data yields accurate slowness surfaces at receiver depths.

Stable inversion of five anisotropic parameters is possible with sufficient source offset.

The method improves anisotropy estimation in heterogeneous subsurface conditions.

Abstract

Multi-azimuth walkaway vertical seismic profiling (VSP) is an established technique for the estimation of in situ slowness surfaces and inferring anisotropy parameters. Normally, this the technique requires the assumption of lateral homogeneity, which makes the horizontal slowness components at depths of downhole receivers equal to those measured at the surface. Any violations of this assumption, such as lateral heterogeneity or a nonzero dip of intermediate interfaces, lead to distortions in reconstructed slowness surfaces and, consequently, to errors in estimated anisotropic parameters. Here, we relax the assumption of lateral homogeneity and discuss how to correct VSP data for weak, lateral heterogeneity (LH). We describe a procedure of downward continuation of recorded travel times that accounts for the presence of both vertical inhomogeneity and weak lateral heterogeneity, which produces correct slowness surfaces at depths of downhole receivers. Once the slowness surfaces are found and the desired type of anisotropic model to be inverted is selected, the corresponding anisotropic parameters, providing the best fit to the estimated slowness can be obtained. We invert the slowness surfaces of $P$-waves for parameters of the simplest anisotropic model describing dipping fractures -- transversely isotropic medium with a tilted symmetry axis. Five parameters of this model -- the $P$-wave velocity $V_0$ in the direction of the symmetry axis, Thomsen's anisotropic coefficients $\epsilon$ and $\delta$, the tilt $\nu$, and the azimuth $\beta$ of the symmetry axis can be estimated in a stable manner when maximum source offset is greater than half of the receiver depth.