This project evaluates the robustness and resolution of earthquake properties obtained from back-projection of seismic array data, and examines methods for improving the imaging ability. A rupture model, especially its speed and duration, is one of the main ingredients for predicting ground motion. It is also a key factor in studies of the physics of earthquake initiation, propagation, and termination. Despite such importance, rupture models are currently available for a limited number of well-studied events such as the 1992 Landers, California earthquake. A large number of high-quality data from a dense seismic array introduces an opportunity for unraveling detailed earthquake rupture processes, and the back-projection analysis can potentially lead to a catalog of energy radiation and propagation models. The method could be applied to any array, but this project focuses on data from the High Sensitivity Seismograph Network in Japan. Application of the method to this data set has provided constraints on properties such as extent and duration of the 2004 Sumatra-Andaman and 2005 Nias earthquakes. While these results demonstrate the power of the technique and high-quality array data, the estimates of uncertainty and applicability of the method to other earthquakes have not been fully explored. This project addresses these issues, and investigates strategies for improving the back-projection procedure. The aim of the project is to formalize our understanding of the capability and limitations of the method as well as determining its feasibility in hazard mitigation through detection and characterization of earthquakes in near real time. Among the broader impacts are the support and education of a graduate student and an undergraduate.
