OPTIMAL ESTIMATION OF CRACK‐STRIKE 1

A bstract The accuracy of estimating crack‐strike from the algebraic equivalent of a popular technique, the dual source cumulative technique (DCT), for analysing shear‐wave splitting in seismic experiments is evaluated for earth models permeated by different alignments of micro‐cracks. A complementary analysis is performed using another analysis procedure, the dual‐independent source‐geophone technique (DIT), to investigate any benefits of the alternative formulation. The investigation considers synthetic vertical seismic profile (VSP) and reflection data for an earth model with two layers over a half‐space, and three different classes of crack‐strike variation with depth: uniform crack‐strike, an abrupt change of crack‐strike between the upper and lower layer, and a continuous increase over both layers. The synthetic data for zero‐offset and near‐offset VSPs and a reflection profile are computed using a full‐wave modelling package in which equivalent anisotropic media simulate distributions of aligned vertical, parallel, water‐filled microcracks. Estimates from the two techniques agree for the constant crack‐strike model, but differ for the VSP data with crack‐strike changes. The asymptotic behaviour of the two angular parameters θ G and θ S from DIT suggest that it may be used to determine crack‐strike under appropriate circumstances in these VSPs, when the time‐delay between the split shear‐waves for the layer of interest exceeds the peak period of the wavelet. In this limit, θ G tends to follow the crack‐strike change with θ S tending to a constant value, whereas DCT will give a misleading value between the upper and lower crack‐strike. Although the behaviour of DIT is not understood in all cases, θ G and θ S values from the VSP data always appear to diverge near the point where an abrupt crack‐strike change takes place. This could be used as a qualitative indicator for layer stripping. Both techniques agree for the reflection data as the recorded data matrix is necessarily symmetric, but still give misleading results for deeper layers in the presence of crack‐strike changes. This study suggests that more care should be taken when designing and analysing experimental configurations for detecting crack properties in reservoir rocks, to consider the response and resolution limits of the analysis techniques. A note of caution is offered to those who directly interpret polarization estimates as crack‐strike.