Tens of thousands of global digital seismograms are available on nine optical disks (CD-ROMs) distributed by the National Earthquake Information Center (NEIC). The data currently span over a seven year period (January 1980 to March 1987). We focus on two subsets of the NEIC data base in this study: a set of 37,710 short-period vertical-component seismograms (sampled at 20 Hz or 40 Hz) from 3,046 earthquakes and a set of 63,853 three-component long-period records (sampled at 1 Hz) from 2,995 earthquakes. The typical periods of the short- and long-period body waves are one and twenty seconds respectively.
These seismograms do not have uniform start and stop times. The short-period
data are triggered and have high record densities surrounding the P, PKP, and
PP branches (Figure 2a). The temporal coverage is superior for the long-period
records (Figure 2b). Both the long- and short-period data suffer from sparse
coverage at distances near the epicenter (0
to 5 ) and near the
antipode (170 to 180 ).
To determine which phases are resolved by our algorithm from the NEIC short- and long-period seismograms, we plot the pick times versus range in order to image global travel-time curves (Figures 3-6). A weighting scheme is used to overcome the uneven time-distance coverage seen in Figures 2a and 2b. The images consist of a time-distance grid in which the darkness of each pixel is proportional to the number of picks divided by the number of seismograms in that time-distance range; thus we plot the pick density rather than pick frequency. Low confidence picks are included in order to image weak phases, at the cost of introducing noise triggered picks. To minimize complications due to depth phases, results are plotted only for earthquakes occurring at depths less than 90 km. To help identify the phases imaged in Figures 3-6, we also include theoretical travel-time curves, calculated from the iasp91 earth model [Kennett and Engdahl, 1991]. Figure 7 shows phases imaged in the vertical- and radial-component autopicks (Figures 3-6) and Figure 8 shows the phases most prominent in the transverse-component autopicks (Figure 6).
