a.  Physical mechanisms of fluid injection induced earthquakes

Hydraulic fracturing (HF) through fluid injection alters stresses near injection sites as the volume of fluid increases, thus produces earthquakes in some cases. While HF injection induced earthquakes (IIE) are typically minor (M<4), some recent events have exceeded M4. To understand the properties of induced earthquakes associated with hydraulic fracturing (HF), and the interaction between the injected fluids and the shale rock formation, we designed an experiment to isolate the effect of HF by deploying eight broadband seismometers at close proximity to the well pad for a period of pre-, co- and post-HF stimulation, in the Montney Play, BC, Canada. In collaboration with Prof. Rebecca Harrington at Ruhr University Bochum, Prof. Yajing Liu at McGill University, and postdoc fellow Bei Wang at Victoria University, we use multi-station matched filter detection and double-difference earthquake relocation methods to develop a catalog of 350 events associated with HF stimulation (M ~ -2.8 to 1.8) [Yu et al., 2019]. Our work finds that pore pressure change and poroelastic stress transfer could both be dominant mechanisms responsible for producing HF induced earthquakes, depending on proximity to the well and the elapsed time.
Moreover, earthquake static stress drop, as a measure of the stress released by fault slip, helps us quantitatively understand the causal relation between hydraulic fracturing (HF) and earthquake source properties. With the additional contribution from Prof. Honn Kao (Natural Resources Canada and University of Victoria) and Prof. Rachel E. Abercrombie (Boston University), we apply both spectral ratio and clustered single‐spectrum fitting methods to aforementioned induced earthquakes [Yu et al., 2020]. We find that stress drop increases with distance to the injection well but is generally invariant within clusters either proximal (~0.1–1 MPa) or distal (~1–10 MPa) to the well. Clustered single spectrum fitting also suggests that seismic energy loss during wave propagation (seismic attenuation) is higher near the well. The observations lead us to interpret that either the higher fracture density and/or elevated pore pressures near the well prevent the crustal rocks from storing and releasing a larger amount of stress at distances approximately <1 km from the well.
However, with the classical concept of pore pressure and/or poroelastic stress change, it is challenging to explain the occurrence of M3+ earthquakes induced over the short operational period of hydraulic fracturing stimulations but located several kilometers away from the wellbore. Aseismic slip loading has recently been proposed as a complementary mechanism, which is validated by laboratory work and numerical modeling. However, aseismic/slow slip signals linked to fluid injection-induced earthquakes remain largely undocumented to date. We find a new type of induced seismic signal in Montney Play that possibly manifests the source process that bridges the slow (aseismic) slip inferred by recent modeling and observations near the wellbore to seismic slip at greater distances. This signal consists of an impulsive broadband onset followed by protracted low-frequency ringing. They have broader P and S-pulses (implying longer source durations) and lower corner frequencies (implying either slower rupture speeds, lower stress drop values, or a combination of both), which are identical to low-frequency earthquakes found in plate boundary zones.


b.  Relation between geological settings and intraplate earthquakes

Intraplate earthquakes often occur in reactivated, pre-existing weak zones and can pose significant risk to nearby communities. It is helpful to relate seismicity to tectonic structures in assessing seismic hazard. I look at source properties of earthquakes in interseismic zones by relating the seismicity with geological settings. I study the seismicity in the Charlevoix Seismic Zone (CSZ) as an example. It is an active seismic zone in eastern Canada overprinted by an extensive fracture network due to a meteorite impact of Devonian age. In Yu et al. [2016], we obtain an increased number of accurate relative locations of earthquakes in the CSZ to allow the examination of fine-scale crustal structures. With Dr. Maurice Lamontagne at Natural Resources Canada sharing his understanding on the geological settings at CSZ, we conclude that: a) the diffusely distributed seismicity inside the meteorite impact structure suggests it consists of highly fractured, weak crustal material relative to the surroundings, b) the alignment of earthquakes outside the impact structure highlighting the rift system structure, c) the paucity of seismicity in between two SE-trending linear bands along the St. Lawrence River may result from the presence of volumes of high-seismic-velocity material. Prof. Yajing Liu and Prof. Rebecca Harrington at McGill University (now at Ruhr University, Bochum) contributed to interpretations on tectonic structures and seismic hazards in the CSZ.


c.  Influence of fault geometry heterogeneity on mega-thrust earthquake rupture segmentations

Are slab geometry anomalies (e.g., subducting seamount chains) one of the key factors in controlling rupture segmentation along subduction faults? Previous studies have drawn empirical relationships between megathrust rupture extent and fault dipping angles (or its gradient in two dimensions), but the geometrical influence on megathrust earthquakes along-strike rupture propagation has not yet been thoroughly quantified. In collaboration with Dr. Bunichiro Shibazaki at the Building Research Institute, Japan and Dr. Takanori Matsuzawa at National Research Institute for Earth Science and Disaster Resilience, Japan, I developed a numerical code in the framework of the rate- and state-dependent friction law to simulate multiple earthquake sequences, including interseismic strain accumulation, earthquake nucleation, rupture propagation, and afterslip behavior along a non-planar fault. I take a 3D numerical modeling approach to investigate the influence of subducting slab geometry on earthquake rupture patterns [Yu et al., 2018]. The modeling result demonstrates that, to first-order, fault geometry heterogeneity controls earthquake rupture behavior. In addition, fault shear stress evolution is clearly modulated with the width of seismogenic zone (W). At a constant plate convergence rate and a larger W indicate an average lower interseismic stress loading rate and longer rupture recurrence period, and could slow down or sometimes stop ruptures that initiated from a narrower portion of the fault. I have also collaborated with Prof. Hongfeng Yang at the Chinese University of Hong Kong, China and Prof. Yajing Liu at McGill University on this work, specifically on model design and interpretation.


d.  Stress adjustment following Mega-thrust earthquake revealed by aftershocks

The up to tens of meters coseismic slip distance and aftershock relaxation should dominate seismicity near the source area of a large slip event. However, the spatial extent of the stress perturbation is unclear. Comparing detailed seismic patterns before and after rupture could help reveal the corresponding stress change. In order to understand the influence of the largest earthquakes in the world on stress adjustment at local and regional scales, we study the effects of the 2011 MW 9.0 Tohoku-Oki earthquake by looking into the seismicity rate and focal mechanism solutions (FMSs) of fore/aftershock sequences [Yu et al., 2013], as well as seismicity in northeast China (NEC) [Yu et al., 2016]. We find that the main shock lowered the E-W compressional stress in Japan Sea and Northeastern China by having Pacific Plate and North America Plate temporarily decoupled in the main rupture zone. As a result, the seismic hazard of roughly E-W compressive events decreases, while the activity of NNE-SSW strike-slip events increases. Moreover, mantle-depth events show P axes along the dip direction of the subducting Pacific Plate in E-W vertical cross-sectional view and in the WNW–ESE directional map view, suggesting a possible down-dip transfer of compressional stress along the subducting plate.
I worked with Prof. Li Zhao at Academia Sinica in Taiwan (now at Peking University), using the GCAP method to determine the focal mechanism solutions of earthquakes before and after the megathrust earthquake, with Prof. Jian Lin at Woods Hole Oceanographic Institution to calculate the coulomb failure stress change due to the coseismic slip, and with Prof. Jieyuan Ning and Prof. Yanbin Wang at Peking University, Prof. Yajing Liu at McGill University to interpret how stress transfer at both crustal and mantle depths in the regional scale.