SAFT Project Summary

We propose to study the detailed structure of the Tonga subduction zone using a one year deployment of two dense seismic arrays; one deployed in the active arc (Tonga) and the other in the backarc region (Fiji). The Tonga subduction zone has the highest rate of deep earthquake occurrence and the fastest subduction velocity, and thus represents the optimum location to study processes such as phase transformations in the downgoing oceanic crust, the effect of a cold subducting slab on the 410 and 670 km discontinuities, and the source processes of deep earthquakes. Both arrays will take advantage of a cooperative agreement with the Japanese SPANET (South Pacific broadband seismic NETwork) deployment (1998-2002), which has installed six sensors in this region.

The detailed structure of the subducting oceanic crust and mantle wedge will be studied using a tight array of 12 sensors in Tonga. The array will enable characterization and modeling of dispersed guided phases within the subducted crust, multipathed arrivals caused by the complex 3-D structure, and secondary converted and dispersed body waves that have been tentatively identified in an earlier (1993-1995) sparse deployment. The arrivals will be studied using array processing techniques, ray tracing, finite difference travel time and waveform modeling algorithms.

A second array of 13 sensors in Fiji will be used to study reflected and converted local arrivals that interact with mantle discontinuities near the Tonga slab. Determination of the ray parameter and azimuth of arrivals using the array will enable positive identification and enhancement of these arrivals, which have been tentatively identified during the earlier deployment. These results will be used to study the depth, velocity contrast, and sharpness of the discontinuities as a function of distance from the slab core, and will be compared with mineral physics results that predict changes in discontinuity elevation and sharpness. The results will help constrain the thermal structure, mineralogy and presence of water within the slab. The local array results will provide better resolution of these parameters than teleseismic results due to the smaller Fresnel zones and the higher frequency content of the arrivals.

The detailed geometry and source parameters of deep earthquakes will be studied using waveform cross-correlation location techniques and empirical Green's Function analysis. The occurrence of repeating deep earthquakes along deep fault-like structures, will also be investigated. The evolution of parameters like stress drop and inter-event times within repeating earthquake sequences on the same deep fault structure has the potential to provide unique constraints on the mechanical processes associated with deep earthquakes.

 back to SAFT home