Back-projection technique takes advantage of the dense array of seismometers that are available around the world such as the Transportable Array in the United States and Hi-net array in Japan. The wavefront observed by the array is collapsed back in space and time (back-projected) to the target region to determine the timing and location of the energy source that generated the seismic waves. If an earthquake has large enough spatial and temporal extent, the rupture propagation can be imaged with great detail using this technique.
The simplicity of the back-projection technique allows rupture property analysis to be performed very rapidly as soon as the data from the array become available. The analysis also requires only limited a priori knowledge, i.e., hypocentral location, and can be performed at any frequency. On the other hand, the method is limited in determining the mechanism or depth of rupture. The simple approach of the back-projection method can be improved in various ways. For example, combination of multiple arrays and seismic phases poses certain challenges, but in some cases, can improve the resolution of the source significantly (e.g., 2009 Samoa earthquake).

Even though the rupture extent and duration are not large enough to image details, smaller earthquakes can be detected using back-projection analysis and seismic arrays. Application of the approach to times immediately after the 2011 Japan earthquake shows that earthquakes as small as magnitude 4's can be detected using seismic networks in North America (with main contribution from the USArray Transportable Array). Some of these earthquakes are not listed in the local (i.e., Japan Meteorological Agency) or global (e.g., National Earthquake Information Center) catalogues. More information about how to hide these earthquakes from dense local monitoring network (e.g., Hi-net array in Japan) can be found here.


(left) Comparison of earthquake locations in the first 25 hours following the 2011 Japan earthquakes. The yellow dots are epicenters from the JMA catalogue, and red dots are events not in the JMA catalogue but have been detected by the back-projection analysis of North American data.

For deep (depth > 100 km) earthquakes, the depth phases can be combined with the direct phase to achieve good resolution in depth of the rupture. Our analysis of 22 intermediate-depth (100-400 km depth) earthquakes with magnitudes above 6.5 and at teleseismic P distances from the Japanese Hi-net array show that many of these earthquakes rupture on a nearly horizontal plane (very little depth extent). Furthermore, a good fraction of these earthquakes (8 out of 22 earthquakes) consist of multiple sub-horizontal rupture planes that are separated in depth. The timing and depth separation of these events suggest that the secondary plane is dynamically triggered by P waves. We propose a new mechanism to explain these observations that includes activation of pre-existing zones of weakness, hydrous minerals, and focusing of water due to grain-size differences.