Stress Drops of Hydraulic Fracturing Induced Microseismicity in the Horn River Basin: Challenges at High Frequencies Recorded by Borehole Geophones

Abstract:

Source parameters of seismicity induced by fluid injections are important for identifying potential differences in the rupture mechanics of tectonic and atectonic earthquakes, for probing self-similar rupture mechanics to very small magnitudes, and for improving operational strategies of seismic risk mitigation and safer geo-energy development. Corner frequencies and stress drops can also influence ground motions and affect seismic hazard estimates. Here, we assess corner frequencies and stress drops of seismicity induced by hydraulic fracturing in the Horn River Basin, British Columbia, over a particularly low magnitude range (Mw -2 to +0.5) to complement existing studies at higher magnitudes.

Our data were recorded by two deep borehole arrays of 15 hz geophones that detected over 90,000 events. To improve the signal to noise ratio, however, we restrict our analysis to 129, M_w > -0.8 events that show clear phase arrivals. Along both borehole arrays, we observe amplifications in the phase arrival spectra at high frequencies (300 to 1000 Hz) that increase in deeper stations. The high frequency amplifications are also present in the pre-event noise and in the continuous data but vary in frequency bands and amplitude across stations and during the phase arrivals. These high frequency features are more likely due to receiver-side effects than source or path effects. Their origin(s) could include instrument self-noise, poor coupling to the borehole casing, or spurious frequencies excited by initial phase arrivals. We restrict our analysis to the SH components of M_w > -0.8 events at the shallower stations, which are less affected by high frequency amplifications.

To estimate stress drops, we apply a Madariaga circular crack model to the SH spectra. Visually and quantitatively, most spectra still do not fit the model well. Systematic differences include a more abrupt transition from low-frequency plateau to high-frequency decay and a faster decay than predicted by Brune or Boatwright models, even including a kappa term to account for high-frequency attenuation effects. To obtain robust first-order estimates, we fix Q and the low-frequency amplitude and only invert for corner frequency, and we impose additional goodness-of-fit criteria. Based on our best estimates, we calculate a mean corner frequency of 208 ± 49 Hz and a mean static stress drop of 6.8 ± 3.7 MPa. The corner frequency appears approximately constant with magnitude and the stress drop appears to scale with M_w. These are most likely a result of our inability to resolve higher (and lower) corner frequencies due to the high frequency amplifications and lack of model fit. However, in combination with datasets of induced and tectonic earthquakes at higher magnitudes, our stress drop and corner frequency estimates lie in the expected range if self-similarity extends into the negative magnitude range. Our results highlight the challenges in retrieving high frequency source parameters from borehole geophones.