Microseismic monitoring of hydraulic fracturing stimulations can be performed using either a downhole array in a nearby observation well or a large-aperture surface array of sensors. Borehole microseismic offers high sensitivity but possesses poor lateral accuracy and source characterization capabilities due to limited acquisition geometry. While surface monitoring's large aperture yields good epicentral accuracy and source mechanism characterization via adequate focal sphere sampling, it is limited in sensitivity by ambient noise and higher attenuation. These two techniques are complementary; by jointly processing the arrays, sensitivity and accuracy of the monitoring can be increased while still providing accurate source mechanisms. However, combining the two approaches presents some challenges. The signal content differs significantly between surface and borehole receivers due to attenuation of the higher frequencies, and the presence of different noise sources at the surface. Moreover, it is non-trivial to construct a velocity model which explains both sets of arrivals.
In this paper, we introduce a technique aimed at jointly processing microseismic data recorded from both surface and borehole arrays. We adopt a "relative migration" scheme in which the migration operation is undertaken via cross-correlation with one (or several) template event(s) recorded on both networks. Residual moveouts induced by the difference in location between the template and the target events are then minimized through an exhaustive parameter-space search. This procedure yields a 4D-migrated cube in which values exceeding a given threshold are potential events. The benefits of using cross-correlation in a joint-processing scheme are threefold:
using events locally occurring as templates reduces the a priori velocity model dependence, as ray paths are almost similar;
the sensitivity of the resulting detection is strongly increased and
constraints on the stack of the migration can be reduced since the source mechanism-induced phase variations can be accounted for directly.
We will first review the methodology, along with its benefits and limitations. Then, a supporting case study will be presented of a multi-stage stimulation operation recorded with a large aperture, surface array and a multi-level borehole array.