Observational Constraints on Relief on the 410- and 660-km Discontinuities in the South American Subduction Zone

Subducting plates affect the characteristics of the deep seismic discontinuities in the mantle. The effects can be observed seismically when P-waves are converted to S-waves at the discontinuities. We studied the subduction zone beneath South America, analyzing receiver functions from the permanent station LPAZ and from 19 temporary stations in the BANJO network. Data were obtained from the IRIS Data Management Center. To achieve more robust signals, we stacked events from stations located in similar positions relative to the subducting plate. Multi-station stacks show clear conversions from the 410- and 660-km discontinuities in the transition zone oceanward of the subducting slab. Further east, stacks no longer show significant P-to-S conversion signals around 410 and 660 km. Discontinuity conversion points for station LPAZ are tightly clustered within the slab, while those for the multi-station stacks are more scattered. Sampling the slab interior at LPAZ shows a transition zone thickening to nearly 300 km. The transition zone progressively thins to the west, with the westernmost stations showing a normal transition zone thickness of about 250 km. The thickness variability agrees with predicted shallowing of the 410-km discontinuity and deepening of the 660-km discontinuity within a cool subducting slab. Further, both deep disontinuities exhibit a high likelihood that signal amplitude increases with decreasing frequency in most groups of stations, with lower probability for frequency dependence in the west. This dependence would imply either that the phase transitions take place across a greater depth interval or that the signal is scattered by relief on the discontinuities. Because we do not see the discontinuities even at long periods in the east, the phase transitions would have to be implausibly thick to explain these observations. Seeking an alternative explanation for frequency dependence, we employed a synthetic model with a Gaussian-shaped discontinuity (with a width of 60 km and amplitude of 50 km) for the slab`s interior to monitor effects of discontinuity relief on the P-to-S signal. The relief can explain frequency dependence on the order we observe. The explanation based on topography is consistent with our findings of a laterally varying transition zone thickness. The topography model also would explain less frequency dependence on the oceanward side and the lack of P-to-S signal in the east where the presence of the slab causes the greatest changes in relief.