Source Processes of Global Major Deep-focus Earthquakes

Systematically studying source processes of global major deep-focus earthquakes provides a good opportunity to reveal the physics of deep-focus earthquakes. We apply a multiple point source inversion method based on waveform modeling of both direct P and SH waves and near-surface reflected pP and sSH waves. The method is used to study source processes of 29 deep-focus earthquakes from 1994 to 2013, with Mw > 7.0 and depth > 400 km. 26 of 29 deep-focus earthquakes can be modeled by our method reasonably well, using one to six point sources. The results are summarized as follow: (1) The 2013 Mw 8.3 Okhotsk and 1994 Mw 8.2 Bolivia earthquakes, the two largest ever-recorded earthquakes, have off-plane source distributions. The two earthquakes consist of several subevents distributed off a particular plane. (2) 16 of these earthquakes have subevents which could be distributed in a plane but with different fault plane orientations that are inconsistent with a rupture plane. Most of the variations of fault plane orientation are about 20˚, but some are up to 45˚. (3) A relatively uniform propagating velocity cannot be found for each earthquake. The subevents are randomly distributed in space and time. A triggering mechanism may be required to interpret the feature. The fast and slow triggering is uncorrelated with the cold and hot subduction slab. (4) Most of these deep-focus consists of normal faults, however, thrust faults exist in two regions, north Tonga-Kermadec Slab and the junction of Japan slab and Mariana slab. We suggest that these deep-focus earthquakes have their subevents occurring in a pre-existing weak zone and, triggering each other.