| dc.description.abstract | Hydraulic fracturing has been found to be a major contributor to the increase in induced
seismicity worldwide, with pore pressure, poroelasticity, and coulomb stress transfer identified
as the three main triggering mechanisms. However, there is still much to be learned about how
these mechanisms operate in hydraulic fracturing-induced earthquakes. The accurate
discrimination of these mechanisms requires a complete and precise earthquake catalog,
particularly with regards to focal mechanisms, which provide insight into the changes in stress in
the area surrounding the hypocenter. Determining the polarities of first motions by hand is a
traditional method for identifying earthquake focal mechanisms, but it is not suitable for
microearthquakes due to their low signal-noise ratio and the large volume of data involved.
Machine learning, on the other hand, provides a reliable and efficient way to classify polarities.
Thus, in this study we apply a machine learning-based first motion classifier to automatically
invert focal mechanisms for induced earthquakes in the Tony Creek Dual Microseismic
Experiment (ToC2ME). We then discuss the accuracy and efficiency of the application of machinelearning-
based first motion classifier – DiTingMotion for hydraulic fracturing-induced
earthquakes and investigate the associated mechanisms for earthquake triggering during the
hydraulic fracturing. We have demonstrated that DiTingMotion is capable of classifying the
polarities of earthquake first motions and characterizing focal mechanisms for induced
earthquakes. By analyzing three major earthquake sequences during the ToC2ME experiment,
our results illustrate that pore pressure, poroelasticity, and coulomb stress transfer can coexist
during the hydraulic fracturing, although each may dominate during different stages. We suggest
that a comprehensive understanding of geological settings, hydraulic fracturing operations, and
the distribution of pre-existing faults/fractures is critical to comprehending the triggering of
induced earthquakes. These factors play important roles in seismic activity and comprehending
them is essential to mitigate the seismic hazard associated with hydraulic fracturing and optimize
shale gas production.
Keywords: Hydraulic Fracturing, Induced Seismicity, Machine Learning, Focal Mechanisms,
Earthquake Triggering Mechanisms, Microearthquakes | en_US |
