Rupture Termination in Laboratory‐Generated Earthquakes

Abstract

Earthquakes are dynamic rupture events that initiate, propagate, and terminate on faults within the Earth's crust. Understanding rupture termination is essential for accurately estimating the maximum magnitude earthquake a region might experience. We study termination on sequences of M − 2.5 earthquakes that rupture a 3‐m granite laboratory sample. At this large scale, nucleation, propagation, and termination are either completely or partially confined within the sample–unique observations for experiments on rock. We compare measured termination locations to estimates from a fracture mechanics‐based model to quantify the fracture energy of the laboratory earthquakes, which compare well with estimates from small natural quakes. Our results provide a mathematical framework that links micrometer‐scale friction parameters to meter‐scale earthquake mechanics, shows that a 3‐m slab of granite can behave similar to a 200‐mm sheet of glassy polymer, and demonstrates how small events can prime a fault for larger, damaging ones.

Laboratory Earthquakes in Large‐Scale Rock Experiments Reveal Singular Crack‐Like Rupture Dynamics

“…The rock friction experiments considered here (Kammer & McLaskey, ; Ke et al, ; Xu et al, ), together with previous experiments in acrylic glasses (Svetlizky et al, ), establish an extensive applicability of the fracture mechanics framework to our understanding of frictional rupture fronts. Whether these results can be scaled up to earthquakes in nature, however, deserves further study as natural faults are known to be complex entities that include significant heterogeneity of fracture energy, friction laws, stresses, and fault geometry.…”

“…In the classical fracture mechanics framework, cracks arrest once the elastic energy being released per unit of newly created surface area G is insufficient to overcome the fracture energy, that is, G < Γ. By inferring G from measurements, Ke et al () were able to show that fracture mechanics predictions agree well with rupture arrest locations. Importantly, the values of Γ constrained from the arrest criterion are consistent with those inferred from the measured strains (section ) on similar rock samples (Kammer & McLaskey, ).…”

Contained Laboratory Earthquakes Ranging From Slow to Fast

“…Across all Poisson loading sequences, contained and partially contained events nucleated at x ≅ 2 m, where the peak τ / σ ratio occurred (see Figure d and Ke et al, ). This relatively slow nucleation can be identified in Figure by the dense set of contours that indicates slow slip in the nucleation region near x = 2 m. For event #29, the slip rate never exceeded 0.5 mm/s; rupture propagation velocity is poorly defined since the entire fault patch slipped nearly simultaneously.…”

“…Here, τ r ( x )= μ d σ N ( x ) where μ d is a dynamic friction level. (In Ke et al (), τ r ( x ) was derived from the final shear stress state τ f ( x ) after a complete rupture event.) Drawing upon this framework, we denote the fault sections where Δτ pot > 0 as a “favorable rupture patch” of length p .…”

“…As a result, local shear stress on the fault increased at the south end and decreased at the north end (Figure b, τ 2 ), creating heterogeneous shear stress profile τ = τ 1 + τ 2 . The bowl‐shaped σ ( x ), which is common to these type of experiments (Ke et al, ; Yamashita et al, ), results from (1) the torque exerted on the moving block by the application of shear stress, (2) compliance of the steel loading frame, and (3) edge effects. As a result, the peak τ / σ ratio consistently occurs ~2 m from the north end of the fault (Figure d).…”