Thesis Defense: Ryley Hill, Ph.D.

Physics-based hydrogeologic models of fluid-fault interactions: implications of natural and anthropogenic poroelastic effects on induced seismicity

Abstract

Our understanding of the role of fluid in seismicity is extensive, ongoing, and growing. The accurate characterization of fluid-fault interactions is dependent on calculations of pore pressure and stress through space and time. The effects of fluid in the Earth’s crust is encompassed well by linear poroelasticity, yet the complexity of the fully coupled problem is limited to numerical solutions. Therefore, the necessity of physical-based models that can accurately resolve the spatio-temporal pore pressure and stress distribution is critical. In this work, I use the finite element method to solve the fully coupled poroelastic transient response within the Earth’s crust due to hydrological loads and wastewater injection. Our work has improved the understanding of fluid-fault interactions, seismic hazards associated with wastewater injection, and earthquake triggering processes.

We use a numerical model and paleoseismic data to study the poroelastic stress induced from ancient Lake Cahuilla on the Southern San Andreas Fault. We find that stress changes were likely sufficient for triggering past major earthquakes. Elsewhere, we investigate induced seismicity associated with wastewater injection at the Raton Basin, CO. We combine statistical methods and a physical model to produce a map of seismic hazard. Furthermore, we build an optimization framework that incorporates the seismogenic index in order to maximize wastewater injected while reducing the seismic hazard. Last, we combine machine learning random forests and a physical model to forecast seismicity at a ~20 year ongoing single well injection site at Paradox Valley Unit, CO. The forecasted seismicity is treated with a game theory analysis to procure feature contributions to help decipher earthquake triggering mechanisms of the induced earthquakes.