This project was focused on the use of dense arrays to unravel the role of nonlinear elastic effects associated with geophysical processes, such as earthquakes, landslides and volcanic eruptions. We have performed such work at the laboratory scale, under well-controlled experimental conditions such that (1) various dense array approaches could be tested and compared, and (2) the observed linear and nonlinear processes could be appropriately interpreted. The carried-out work can be broadly divided in two main categories. The first one is related to experiments where active sources were used either in transmission or reflection to image faults and fractured rock samples. The second one is related to experiments where acoustic emission (AE) events were detected by dense arrays during friction experiments, and analysed with regard to nonlinear elastic processes.
Imaging of fractured rocks conducted in transmission has shown that nonlinear effects are not necessarily located along the main, larger fractures. This suggests that each individual contact between the grains contribute to the overall nonlinearity. Transposed to field observations, this indicates that small, co-seismic reduction in wavespeed that are observed along many tectonic faults are not precisely located on the fault, but rather affect the overall region surrounding the epicentre. Using a double beamforming approach on our transmission data and a simple cubic-ray assumption, we were able to locate low-velocity regions without conducting any inversion. We found that these low-velocity zones corresponded to open fractures. Further, we found that the ray tracing was not significantly affected by large dynamic disturbances – equivalent to distant earthquakes in the field. This confirms that under these conditions, dynamic disturbances are large enough to transiently affect the elasticity (small reduction in wave-speed corresponding to the nonlinear effects), but small enough to remain in an elastic regime. Following this work in transmission, we explored the possibility of imaging fractured rocks in reflection, using an ultrasonic imaging device whose primary use is the imaging of soft tissue in the medical domain. This approach is particularly promising to monitor laboratory faults during friction experiments, and in the domain of rock mechanics in general. Assuming a homogeneous wave-speed within the rock matrix – which is typically the case in laboratory experiments, one can use built-in, reconstruction tools (e.g. beamforming, compounding) that greatly improve image quality, often in real-time. With this approach, we estimated the longitudinal stiffness and used it to produce a 2D image of a complex fracture. We also confirmed results obtained in transmission showing that the nonlinearity of closed fractures was not larger than the nonlinearity arising from inter-grain contacts within the rock matrix.
A second objective of this project was to investigate acoustic emission data (equivalent to seismic data in the laboratory) during friction experiments. Prior to using dense ultrasonic arrays along a laboratory fault, we conducted friction experiments with a couple of sensors only and determined the evolution of the frequency-magnitude b-value during stable and unstable frictional sliding experiments, which provides insights on the relative scaling of small versus large earthquakes. We found an inverse correlation between b and shear stress. The reduction of b occurred systematically as shear stress rises prior to stick–slip failure and indicates a greater proportion of large events when faults are more highly stressed. Following this work, we conducted friction experiments with a dense array of ultrasonic sensors along the fault and varied normal stress and sliding velocity to produce a variety of seismic events, from tremor-like signals and slow slip events to ordinary, fast earthquakes. We started to implement a matched field processing technique to localize micro-seismic events that precede main slip events. We use these spatio-temporal maps to determine whether similar patterns repeat themselves from one stick-slip event to the next. In addition, this will help us answering the question whether or not fast and slow slip events obey the same physics and share the same source scaling mechanisms.
