SCEC2022 Plenary Talk, Earthquake Forecasting and Predictability (EFP)

Simulating seismic activity at the Coso geothermal field

Kayla Kroll

Oral Presentation

2022 SCEC Annual Meeting, SCEC Contribution #12462
It is understood that manipulation of the subsurface stress state by fluid injection or extraction can induce or possibly inhibit earthquake activity. In fact, it has been suggested that actively managing reservoir pressures via co-extraction of fluid during an injection operation may reduce the occurrence of seismicity by reducing the overpressure along subsurface faults and fractures. Modeling studies show that in highly idealized settings, co-extraction does reduce overpressures and generally reduces seismicity, but it does not guarantee complete seismic quiescence. There are few robust observations of this type of causal mechanism influencing operational control of induced seismicity, primarily due to lack of systematic experimentation at the field-scale and lack of robust seismicity catalogs. However, in 2019, two large earthquakes occurred near Ridgecrest, California, ~35 km southeast of the Coso Geothermal Plant. The M6.4 and M7.1 Ridgecrest earthquakes were separated by only 34 hours and generated vigorous aftershock sequences along the mainshock rupture planes. Curiously, a paucity of aftershock activity has been observed in the area surrounding the Coso geothermal field (CGF). The unexpected paucity of seismic triggering observed near the CGF provides an excellent field laboratory to study whether natural tectonic, viscoelastic relaxation or industrial activities relieved stress and/or overpressure on faults in the CGF, such that they are less likely to be triggered by coseismic stress changes. In this work, we couple 3D non-isothermal multiphase poromechanical simulations with 3D physics-based earthquake simulations to understand the mechanisms controlling aftershock activity in the GCF. Poromechanical simulations include site specific injection and production information and characteristics and are constrained by InSAR time-series of surface deformation. Earthquake simulations are constrained based on 30+ years of well-located seismicity and local fault slip-rate data.

Prepared by LLNL under Contract DE-AC52-07NA27344.