Moment tensors for microseismicity are frequently used to tie to the geomechanics of completions, as they inform on the fracture sets that are being stimulated and they are expected to be congruent with the dynamic (perturbed) stress state. These quantities, being a bridge from the microseismic data to the geomechanics, can be critical to extracting value out of microseismic acquisitions. However, the automation of moment tensor data is not without challenges, the one we focus on in this study is the ability to resolve the sense of first motion on the seismic traces and how they may be reversed from the maximum amplitudes that are much more easily determined. As such, maximum amplitude moment tensors may be reversed in sign from reality, and convey incorrect interpretations on the stress state.
A hydraulic fracture completion was monitored with a combination of near-surface borehole-deployed geophone arrays and an array of “superstations'' on the surface. All of the detected events (more than 17000 complete to −Mw1.2) were inverted for their moment tensors and a quality score was assigned to each inversion to determine the high-confidence (>95%) dataset. Mechanisms incongruent with the background stress are explicitly flipped due to a general uncertainty of the first motion: we wish to understand the impact of this assumption with comparison to the subset of mechanisms where first motions can be unambiguously assigned. The highest magnitude events are examined from the dataset to observe clear first motions and are compared to the automatically inverted and then potentially flipped dataset.
When the mechanisms are inverted from unambiguous first motions, we observe large-scale agreement with the mechanisms inverted automatically from strongest amplitudes and selectively flipped to be congruent with our notion of the background stress. Furthermore, by applying our quality scoring to the first-motion dataset, we observe that these mechanisms are much better fit than the automatic solutions. Disagreements do exist on subdominant mechanisms, generally not as well-aligned (or anti-aligned) with the background stress regime, complicating the decision to flip. Our work suggests that moment tensors should be largely congruent with the background stress regime, and that mechanisms that disagree need not imply a complete flip of this stress regime, but may show difficulty in picking first motion.
Assuming a stress state to invert for moment tensors which then informs on the stress state sounds like the very definition of circular reasoning. However, what we are suggesting is more subtle. We assume that the stress state is dominantly one regime, but we use that information to only resolve the sign of that mechanism: for example, is a given mechanism normal or thrust? Effectively, we assume that the microseismic event around the hydraulic fracture is responding to perturbations in the stress regime that do not represent a complete overturning of it. With that (we think modest) caveat, we resolve stress fields nuanced in space and time during the hydraulic fracture rapidly without the arduous task of first motion picking, which still can be used for validation.