Constraining Geological Structures and Temporal Variations in Fluid Accumulation near Hydraulic Fracturing Injection Sites in the Western Canada Sedimentary Basin

Hydraulic fracturing (HF) in low-permeability tight shale in the Western Canada Sedimentary Basin (WCSB) has been associated with an increasing number of M3+ earthquakes in the last decade. One way to help quantify injection induced stress alterations near HF wells is to constrain areas of fluid infiltration and geological structures that facilitate fluid transport and possibly, fault slip. We use new seismic waveform data from a dense deployment of nine broadband seismometers starting in June 2017 in close proximity to numerous HF wells near Dawson Creek, British Columbia. We generate an earthquake catalog that will be used for joint velocity and hypocenter inversion to relocate earthquakes and highlight 3D velocity structures in different injection periods. The results will help highlight how faults and fluids interact to nucleate earthquakes. We use the open-source SeisComP3 software to detect and locate earthquakes in the study area and to test parameter settings during a time period from 12^th July 2017 to 31^st August 2017. We detect 491 earthquakes, of which 128 are false detections, compared to 11 detected earthquakes by the CNSN catalog for the same period. Parameter settings providing the sensitivity and false detection rate referenced above include a 3-20 Hz, 4^th order Butterworth band pass filter on continuous waveforms, an STA/LTA trigger with STA duration of 0.2 s, LTA duration of 10 s, a triggering threshold of 2.5, and de-triggering threshold of 1.5 for P-wave detections. The S-wave detections employ a 3-12 Hz, 3^rd order Butterworth band pass filter and an AIC-algorithm. The joint inversion of hypocenter location and velocity structure uses absolute arrival times and cross-correlation values with the tomoDD software. TomoDD also constrains V_P/V_S-ratios, where high values indicate a concentration of fluids in the crustal volume. Starting values for the inversion are based on the Crust1.0 1D velocity model. Preliminary earthquake hypocenters cluster into four groups, with two spatially separated western clusters striking approximately 135°, and two parallel eastern clusters, striking roughly 20°. Combined with a detailed microseismicity catalog, our 4D imaging results will illustrate spatial and temporal variations in fluid content and potential migration patterns near HF wells.