Another development to enhance the BP approach is the 3D back-projection. It is particularly useful for deep-focus earthquakes, whose fault planes are not well defined. Standard BP project the source locations onto a 2D sub-horizontal plane such as the shallow dipping plate interface of the mega-thrust earthquakes, because the incident seismic rays are nearly vertical and a trade-off between source location and timing exists along the ay path of the seismic wave. This means the BP method usually provides better horizontal resolution than depth resolution, similar to the uncertainties of earthquake location. The fault planes of deep earthquakes are typically ambiguous, with a choice between a sub-horizontal and a sub-vertical fault plane, both consistent with the focal mechanism. Accurate depth information can also determine whether a rupture was confined to a single fault plane or crossed multiple fault planes at different depths. 3D back-projection (3DBP) improves the depth resolution by combining the BPs of the P and pP phases (Chen et al., Gcubed, 2018). The ray paths of the two phases intersect in space, which overcome the trade-offs seen when back-projecting rays of, for example, only the P phase.
Figure 1. 3D spatio-temporal rupture propagation of the 2013 Okhotsk earthquake, imaged by applying our 3DBP technique to USArray data of P and pP waveforms.(a) Map view of BP images. The inset plot displays the BP source power as function of time (b) Cross-section of BP images along AA’. (c) Cross-section of BP images along BB’. The colors and shape of the ellipses demonstrate the timing and the 95% confidence interval on the peak locations.
The benefit of the 3DBP is shown in the case study of the 2014 Mw 8.3 Okhotsk earthquake, the largest deep-focus earthquake ever recorded by modern seismology. In our previous 2D BP we find that the bilateral rupture possibly indicate different mechanisms activating at the same time, since the northern branch follows the interior of slab isotherms and the southern branch go beyond the possible width of metastable olivine wedge and therefore required the aid of thermal shear instability (Meng et al., GRL, 2014). The left figure shows the 3DBP images of the 2013 M8.3 Okhotsk earthquake. The overall north-and-south-ward bilateral rupture is consistent 2D BP onto a horizontal fault plane. However, the cross-sections along the two branches suggest the focal depth increases as the rupture propagates southward (panel c). This increased depth of the southern branch is consistent with multiple-source inversions (Chen et al., 2014). Since the focal mechanism indicates sub-horizontal fault plane, the increased depth in the southern branch suggested a cascading failure of rupture along sub-parallel horizontal planes in a en echelon fashion.