Figure S1. Stacks of synthetic seismograms from the 3-D model in which the velocity boundary is shifted (see Fig. 10e), shown for four azimuth and distance bins. The cross marks the predicted arrival time of S660S precursor. The number at the top right corner of each panel indicates the number of traces in that stack. When the bin is far away from the velocity boundary, only one precursor with smaller amplitude is seen, whereas when the bin is at or near the boundary, ‘double precursors’ with larger amplitudes are seen. This shows the artificial ‘precursor’ cannot be suppressed by stacking.

Figure S2. ‘Rose diagrams’ (angle histograms) showing the global azimuthal coverage that could possibly be achieved by stacking all SS precursor data before the days of USArray. To compute the ray paths, the earthquakes are all shallow (depth ≤ 75 km) events with 5.8 ≤ Mb ≤ 7.0 from the Global CMT catalogue from 1976 to 2004, the stations are all seismometers available from IRIS before the deployment of USArray and the distance range is limited in 100°–160°. The globe is subdivided into 412 nearly equal areas, 10°-radius spherical caps for common SS midpoint stack. Within each cap, the 360° azimuth range is divided into 15° bins for plotting the angle histogram. (a) At each spherical cap (i.e. for each ‘rose’), the radius of each ‘rose petal’ (azimuth bin) is proportional to the number of ray paths sampling that direction, and the maximum radius of each ‘rose’ is normalized to 1. (b) At each spherical cap, all ‘petals’ have a radius of 1, with the darkness of the petal's colour indicating the number of traces sampling that direction.

GJI_5683_sm_Supplement.zip745KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.