Figure 7 . (a) Maps of the shear-velocity variation at (top) 150 km and (bottom) 500 km depth according to S40RTS. The east-west contrast across the US is similar for SEMUCB-WM1 and TX2015. (b) Harmonic undulations of the 410-km and 660-km discontinuities for (from top to bottom) models T8, T5, and T2 with spatial wavelengths of 8°, 5°, and 2°. (c) Distribution of hypothetical earthquakes (stars) and stations (circles). For models PREM, S40RTS, T8 and T5 we compute waveforms for the twelve earthquakes indicated by red stars. For T2, we compute waveforms for these earthquakes and the additional 36 earthquakes indicated by black stars.
For each of the eight structures, we compute waveforms at periods longer than 10 s for 462 stations in a rectangular 2° 2° grid between longitudes 130°–65°W and latitudes 25°–50°N (Figure 7c). We calculate waveforms for 12 earthquakes uniformly distributed at a distance of 75° from [-100°E, 40°N]. We use 48 earthquakes distributed in a spiral for structure T2. All earthquakes have the same dip-slip source mechanism (source parameters for event 080596G in the Global CMT catalog) to ensure strong radiation of SH waves to teleseismic stations. The uniform data coverage is sufficient to investigate the effects of velocity heterogeneity on Ssds-S traveltimes and the resolution of undulations of the 410-km and 660-km discontinuities using long-period Ssds reflections. Because of the high computational cost, we cannot afford to reproduce the source-station combinations in the data and, therefore, we cannot estimate CRP mapping artifacts due to inhomogeneous slowness and azimuthal sampling.