Earthquake nucleation and recurrence in laboratory, reservoir and tectonic settings

Abstract

Numerical models are essential for bridging the spatial and temporal scales between natural observations and laboratory experiments, enabling the validation and generation of scientific hypotheses. Recent advancements in instrumentation have significantly improved the quality of observations, especially in the Groningen gas field since the 2012 Huizinge earthquake (ML 3.6). Experiments on borehole samples under in-situ conditions provide critical data on the physical properties of geological layers. With numerical models mastering homogeneous setups, these enriched datasets now allow the development and testing of models to understand how earthquake nucleation and sequences are influenced by heterogeneous material parameters and dimensions, from laboratory experiments to the Groningen gas reservoir. I developed and validated (quasi-)dynamic earthquake sequence models across various dimensions and time scales, considering factors such as tectonic motions, gas extraction, reservoir compaction, and fluid flow.

A key focus is on the impact of spatial dimension reduction. Simplifying models by reducing spatial dimensions can qualitatively and quantitatively affect outcomes like recurrence intervals, coseismic slip, and rupture speeds. I developed a theoretical framework to explain these changes and demonstrated that lower-dimensional models can replicate key higher-dimensional results with efficiency. Additionally, I investigated the influence of normal stress heterogeneity on earthquake nucleation. Five regimes of nucleation and slip behaviors were identified, governed by the ratio of heterogeneity wavelength to nucleation length. Lastly, I tackled the paradox of induced earthquakes on supposedly stable faults and apply the findings to build a more realistic heterogeneous model for Groningen. I emphasize the significance of fault healing over geological timescales, suggesting that even velocity-strengthening faults can host induced earthquakes. Overall, this thesis enhances our understanding of earthquake nucleation and sequences, providing insights for seismic hazard assessments across different scales and tectonic settings.