Figure 5 . The recorded (a) and predicted (b, c, d) traveltime delays of S waves by tomographic mantle models S40RTS (in b), SEMUCB-WM1 (in c), and TX2015 (in d) with crustal model CRUST1.0. Each circle indicates the location of a seismic station. Its color indicates the mean of the S-wave traveltime delays with respect to the PREM model for at least five S waves. (e) Histograms of the S wave traveltime delay in the data (gray fill) and predictions by S40RTS (green line), SEMUCB-WM1 (blue line), and TX2015 (red line) for the stations in panels a–d.
Figure 5a shows how shear-wave velocity variations in the crust using CRUST1.0 (Laske et al. 2013) and in the upper mantle affect the traveltimes of S waves. Plotted are the average S-wave delay with respect to the PREM velocity model of at least five S-waves recorded at seismic stations from the USArray. The delay times have been corrected for ‘source terms’, representing the effects of a potential mislocation of the earthquake location and origin time on the absolute S wave traveltime. S waves recorded by USArray stations in the western US (the tectonically active region) arrive on average 5–6 seconds later than at stations in the central and eastern US (the stable platform). The global-scale mantle models S40RTS (Ritsema et al. 2011), SEMUCB-WM1 (French & Romanowicz, 2014), and TX2015 (Lu & Grand, 2016) predict a similar traveltime pattern (Figures 5b–d) but the range is slightly smaller than in the data (Figures 5e). Our calculations indicate that the crustal structure from CRUST1.0 enhances the east-west contrast only slightly, so wave speed variations in the mantle are primarily responsible for the S-wave traveltime differences.
The imperfect match between the recorded and the predicted S wave traveltime is expected because tomographic models do not perfectly explain the recorded traveltime variation of any shear wave (e.g., Ritsema et al. 2004). Nevertheless, it is obvious that shear-velocity heterogeneity affects teleseismic S wave traveltimes across the USArray. Since Ssds has two additional propagation legs through the upper mantle, the Ss410s-S and Ss660s-S difference times are likely to be double the variation shown in Figure 5a due to shear velocity heterogeneity only. If shear-wave speed variations in the upper mantle beneath North America are ignored in the modeling, a variation of the Ss410s-S and Ss660s-S difference times of more than 10 s would imply that the depths of the 410-km and 660-km discontinuities vary by about 18 and 20 km or more. This is of the same magnitude as resolved in Figure 4.