Seismic Anisotropy in the Transition Zone of the Mantle: Implications for Mantle Dynamics and Deep Earthquakes

Michael R. Brudzinski* and Wang-Ping Chen
University of Illinois at Urbana-Champaign
* Now at University of Wisconsin-Madison

We demonstrate a rare, unequivocal case of seismic anisotropy in the transition zone beneath the Tonga-Fiji Region. In many triplicate, broadband waveforms, horizontally polarized shear-waves (SH) arrive up to 3 s earlier than vertically polarized shear-waves (SV). Using waveforms from earthquakes of different depths, we isolate the anisotropic region to lie mainly in the transition zone where the speed of SV is approximately 1% slower than that of SH. 

The anisotropic zone coincides with the occurrence of outboard earthquakes - a subhorizontal swath of deep (>300 km) seismicity extending nearly 800 km farther to the west of the Wadati-Benioff zone. Cold temperature, as evident from deep earthquakes, is expected to increase seismic wave speeds.  However, P and SH wave speeds are not fast in the source region of outboard earthquakes [Brudzinski and Chen, JGR, 21661, 2000; Chen and Brudzinski, Science, 2475, 2001], and SV wave speeds are even slower than SH wave speeds.  Thus the effect of cold temperature must have been counteracted by a petrologic anomaly.  Furthermore, seismic anisotropy at depths where weakly anisotropic gamma-phase should dominate suggests the presence of a highly anisotropic phase instead.

Two leading candidates for the petrologic anomaly are the presence of metastable olivine (alpha-phase) or volatiles brought down by subduction. Both would lower seismic wave speeds in the transition zone and could trigger deep earthquakes. For metastable olivine, it must also develop lattice-preferred orientations to account for the observed anisotropy. For hydrous minerals or free fluids, their shear-wave speeds may be slow enough to provide a large contrast (~10%) against coexisting minerals, which is needed to develop a fine-scale, laminated structure. However, the lack of a consistent pattern in fault plane solutions of outboard earthquakes does not support preferred orientations of fluid pockets as a mechanism to trigger these earthquakes. Moreover, the amount of hydrous minerals needed to reduce seismic wave speeds may also make the entire slab buoyant, prohibiting subduction.