Upper-mantle structures beneath USArray derived from waveform complexity

Tomographic imaging of the crust and upper mantle beneath the western United States has greatly improved with the addition of USArray data. These models display many detailed images of both fast and slow blobs penetrating into the transition zone. To study such features, we apply a newly developed technique, called MultiPath Detector analysis, to the SH waveform data. The method simulates each observed body waveform by performing a decomposition; by [S(t) + C x S(t - Delta(LR))]/2, where S(t) is the synthetics for a reference model. Time separation Delta(LR) and amplitude ratio C are needed to obtain a high cross-correlation between a simulated waveform and data. The travel time of the composite waveform relative to the reference model synthetics is defined as Delta(T). A simulated annealing algorithm is used to determine the parameters Delta(LR) and C. We also record the amplitude ratio (Amp) between the synthetics for the reference model relative to the data. Generally, large Delta(LR) values are associated with low Amp's. Whereas the conventional tomography yields a travel time correction (Delta(T)), our analysis yields an extra parameter (Delta(LR)), which describes the waveform complexity. With the array, we can construct a mapping of the gradient of Delta(LR) with complexity patterns. A horizontal structure introduces waveform complexity along the distance profile (in-plane multipathing). An azimuthally orientation Delta(LR) pattern indicates a vertical structure with out-of-plane multipathing. Using such maps generated from artificial data, we can easily recognize features produced by dipping fast structures and slow structures (DSS). Many of these features display organized waveform complexity that are distinctly directional indicative of dipping sharp-edges. Here, we process the array data for events arriving from various azimuths and construct maps of multipathing patterns. The similarity between tomographic features and complexity maps is striking. When features are dipping such as the slab structures beneath the Cascade Range and Nevada, strong complexity is observed from Southeastern events arriving along these ray paths with split pulses separated up to 6 s for both. This requires extended slab segments to at least 600/300 km with a 4/8 per cent velocity jump along the edges. One of the most dramatic set of DSS observations is associated with a slow northwest dipping conduit beneath Yellowstone that extends into the transition zone. A number of forward modelling experiments are included for the strongest patterns formed by sharpening present tomographic images.