Rupture Directivity of Fluid-Induced Seismic Events

The hazards associated with induced seismicity have been hard to predict, especially when new technologies have been deployed, or when extant technologies have been deployed or up-scaled in new settings. The failure to adequately manage induced seismicity hazards has had major economic and environmental implications in many countries, where regulators have been forced to shut down industry operations. Recent examples include the Groningen gas field in the Netherlands, hydraulic fracturing of the Bowland shale in the UK, the CASTOR offshore UGS site in Spain, and the Pohang and Basel deep geothermal projects in South Korea and Switzerland respectively, with the latter one being seismically active several years after the injection well was shut-in.

To better understand the complex rupture processes of induced earthquakes, we first compile the seismic waveforms of multiple seismic events induced by different fluid-injection operations, as hydraulic fracturing (HF) in shale gas wells and enhanced geothermal systems, wastewater disposal (WWD), and gas storage. We then implement the Empirical Green’s Function (EGF) method to measure for each analysed induced earthquake their rupture time and directivity, which complement other key source mechanism parameters of the same events already calculated, as moment magnitudes and focal mechanisms.

The EGF method allows the measurement of the Source Time Function (STF) of a seismic event at each seismic station that recorded it, by deconvolving its seismic signal with the one from another seismic event of smaller magnitude but with very similar hypocentral location. This deconvolution between seismic waveforms from Target and EGF events allows a numerical removal of all factors that affect the recorded seismic waveform (like the earth structure, the near surface, and the instrument response) except for the event’s source itself, obtaining a STF for each target event at every seismic station.

We then measure for each target event the rupture time at each station from the obtained STF, and invert the rupture times with respect to each station’s azimuth. We use the coefficient of determination (R2) of the inverted rupture time variation with azimuth to determine whether an induced earthquake had a predominant unilateral or bilateral rupture. In most cases, we observed a consistent self-similar scaling of the earthquake’s magnitude and mean rupture time, and a predominant unilateral rupture for higher magnitude events. In each case, we also examine whether the rupture directivity is driven by the local stress conditions, or by the fluid injection that induced them. Finally, we discuss the possibility of implementing the same methods to monitor the recently awarded CCS licenses in the UK Continental Shelf.