Seismic Array Back-Projection

A focus of my research is to understand the physical processes of earthquakes. The goal is to address, the fundamental questions about the physics of earthquake ruptures through high-resolution and robust observations, (i.e., rupture initiation, its complex propagation, and the final arrest). Currently the greatest challenge in this field is that the observations are behind the modeling efforts, making testing and validations of the ever increasing rupture models impossible. State-of-art earthquake simulations predict the emergence of complicated patterns such as variable rupture speeds and isolated rupture fronts, rupture re-initiation and frequency-dependent behavior. However, efforts to extract reliable seismological constraints are limited by the non-uniqueness and low resolution of the standard earthquake source inversions. By developing the next-generation high-resolution earthquake source imaging techniques, the aim of my research is to narrow the gap in spatiotemporal resolution between computational earthquake simulations and source imaging based on seismological observations. Over the past decade, the development of large-scale dense seismic networks has enabled rapid progresses in a broad spectrum of seismological sciences. Back-tracing of seismic waves recorded by dense arrays allows us to track the source areas of strongest high-frequency radiation, which enables “Back-Projection” , an emerging technique for earthquake source imaging.

Geometrical complexities of ruptures plays a major role in controlling the characteristics of dynamic rupture propagation, especially for large continental and oceanic strike-slip earthquakes. As an example, we were able to use back-projection source imaging to show that the rupture of the 2012 M8.6 Off-Sumatra earthquake was remarkably geometrically complex, involving four orthogonally oriented fault planes (Fig. 1). The high resolution of BP provided an unprecedented view of the rupture, which branched twice into faults where dynamic stresses were compressional. This observation challenged the conventional understanding of the dynamic clamping effect (Meng et al, Science, 2012). Another interesting aspect is that the rupture steps from two parallel fault segments C and D near the westernmost area of the earthquake. The rupture transition from faults C and D across an offset of 20 km is extremely large, considering that fault stepping is typically not larger than 5 km in California (Wesnousky, 2006). The Understanding of these extreme fault stepping or branching is essential in seismic hazard assessment because it is the key mechanism controlling connection of different fault segments in a single earthquake.