Source process of deep-focus earthquakes

Our study of the rupture process of large deep earthquakes (Antolik et al., 1998) has uncovered some very interesting charactericstics. Although much significance has been placed on the apparent rupture of the great 1994 Bolivia earthquake and the 1994 Fiji Island earthquake through the seismically active portion of the deep slab, we beleive that rupture of most deep events is confined to the slab core. Even the two above mentioned events contained the majority of their moment release within the region defined by the aftershocks and background seismicity. Thus temperature-controlled rupture mechanisms such as transformational faulting (or perhaps some sort of plastic instability) are likely to remain the leading candidate to explain seismic failure at depths below the brittle regime in the Earth.

However, both the Bolivia event and the 1996 large Flores Sea earthquake (M_w 7.9) are somewhat different from the smaller events which we have analayzed in that the principal portion of the moment release appears to consist of a continuous patch rather than disjointed asperities separated by regions of little or no slip. The gaps, which we have observed for nearly all of the events in the study (Figure 26.1), are likely to be real since we have not used a rupture parameterization involving multiple fault planes (the rupture is confined to a nodal plane of the centroid focal mechanism, and is represented by a continuous distribution of point sources rather than discrete subevents). Thus the source regions of deep-focus earthquakes may be more compact than previously recognized, and the stress drops could be much higher. Ogawa (1987) cited a shear heating instability mechanism as a possible candidate for deep earthquakes, and rupture propagation aided by melting has been discussed by Kanamori et al. (1998). We believe that shear heating may have played a role in rupture of the Bolivia and Flores Sea events, given their distribution of moment release. However this is not likely for events which contain large gaps in their slip distribution since it would be improbable for a shear instability to re-initiate quickly over distances of tens of km.

**Figure 26.1:** View of the slip distribution for the August 5, 1996 Tonga deep earthquake (M_w 7.3). Fault plane is sub-horizontal (dip 18^o). The moment release concentrates into three distinct subevents having very large slip. The maximum slip is nearly 15 m which is very large for am earthquake of this size. The estimated static stress drop is about 25 MPa. Background seismicity is shown projected onto the fault plane as the circles. The moment release aligns very closely with the strike of the slab as inferred from the background events.
$\begin{figure}% \begin{center} \epsfig{file=figs/bsl98_dsch_fig1.eps, width=9cm} % \end{center}\end{figure}$

We have also observed a tendency for rupture to propagate in a largely horizontal direction. Many recent events also occurred on sub-horizontal or low angle planes. It is not known what mechanism would favor such a rupture geometry, although some sort of isobaric process may play a role. If there is a large component of differential horizontal flow at the base of the upper mantle, perhaps because of layered convection, then rupture on sub-horizontal planes may be favored.

References

Antolik, M., D.S. Dreger, and B. Romanowicz, Rupture processes of large deep-focus earthquakes from inversion of moment rate functions, In press Journ. Geophys. Res., 1998.

Kanamori, H., D.L. Anderson, and T.H. Heaton, Frictional melting during the rupture of the 1994 Bolivia earthquake, Science, 279, 839-842, 1998.

Ogawa, M., Shear instability in a visco-elastic material as the cause of deep focus earthquakes, J. Geophys. Res., 92, 13,801-13,810, 1987.