Discriminating dynamic rupture arrest in fluid-induced microearthquakes using spectral inversion

Determining the maximum possible magnitude of fluid-induced earthquakes requires to understand rupture arrest within or outside a fluid-pressurized patch. Recent studies have highlighted the importance of incorporating rupture physics into the study of injection-induced earthquakes. We perform 3D dynamic simulations of spontaneous ruptures propagating across a pressurized fault, stimulated by fluid injection within the nucleation zone. Our simulations unveil two end-member models describing a fluid-induced micro-earthquake: a self-arresting rupture that decelerates spontaneously and a run-away rupture that terminates abruptly at the fault edge. We compute synthetic waveforms radiated from both models and invert them using a probabilistic spectral inversion approach to identify characteristics that distinguish between the two rupture types. We find that self-arresting ruptures radiate less high-frequency waves (with γ > 3) and lack the typical P/S corner frequency shift. In contrast, run-away ruptures conform to the ω² model (γ ∼ 2, ). We interpret these differences as primarily arising from the rupture arrest mechanism, smooth arrest results in gradual variations in the moment-rate function, whereas abrupt arrest at a barrier causes a sharp changes in the moment-rate function. This abrupt arrest generates high-frequency radiation and a back-propagating stopping phase, playing a critical role in controlling the rupture duration and the radiated seismic waves. Our results demonstrate that spectral features such as high-frequency decay and P/S corner frequency shift may provide observational diagnostics to distinguish rupture arrest mechanisms, even in the absence of direct evidence from rupture kinematics.