EduQuakes Logo
January/February 2001

PKIKP ... and Those Mysterious Precursors

A simple question about one seismogram can open the door into the wonderful world of wave propagation. Seismographs in North America are far away from earthquakes that occur in the subduction zone along Indonesia, and thus the first arrivals from these events will be core phases--P waves that have traveled through the Earth's core. On 7 August 2000, there was a fairly large (Mw = 6.5) deep earthquake beneath the Banda Sea region of the Indonesia arc. This deep event is notable in producing sharp single-pulse P-wave seismograms (for example, see the source time function at the STF Web site When this event occurred, I was working with a student on a project to measure teleseismic P-wave residuals at the MichSeis and OhioSeis stations (see "OhioSeis", EduQuakes column in SRL 70:3). Core phases can sometimes look as simple as the teleseismic P waves. However, the MichSeis and OhioSeis seismograms appeared quite complex with multiple pulses and odd emergent first arrivals from the Banda Sea event. These stations are at epicentral distances between 137° and 140° and there is no "official" seismic phase in the travel-time tables that corresponds to the odd first arrivals. Welcome to the wonderful and sometimes confusing world of core phases! This column represents another installment in the "EduPhase" series. The supporting graphics of travel-time curves, ray tracing plots, and seismograms are at the "PKIKP" link of the EduPhase Web page

Seismologists use a compact and efficient notation to designate body-wave arrivals, but there is some unfortunate confusion over the details of core phase identification. A first-arriving core phase clearly must be some variation on "PKP"; that is, a "P" wave through the mantle, P wave through the outer core ("K"), and a "P" wave back through the mantle. Since the outer core is a low-velocity zone with respect to the overlying mantle, we expect to see a shadow zone followed by the PKP caustic point at a distance of about 143°. Two distinct arrivals with different apparent velocities emerge from the PKP caustic point, and hence the distance ranges from 143° to 180° two PKP arrivals. Raytracing plots of these core phases can be seen at the EduPhase Web site. To distinguish between these two PKP's, seismologists name the slower PKP as "PKPab" and the faster branch as "PKPbc." If our Earth had no inner core, then the two branches of the PKP caustic structure would be the extent of core phases. But there is an inner core, and the velocity increase at the outer-inner core boundary produces a "triplication" travel-time structure that lies on top of the PKP caustic structure. As we follow the PKPbc branch out to larger distances, the PKPbc rays are bottoming more deeply into the core. Eventually, at a distance of about 155°, the PKP rays hit the inner core and they are "reflected" back. As we shoot more rays at the inner core with a steeper angle, they continue to turn at the inner-outer core boundary and emerge at closer distances. This back-branch of the triplication extends all the way back to a distance of 120° or so. As we continue to shoot rays at the inner core with nearly vertical angles, these P ray paths are able to enter the inner core. Thus from about 120° all the way out to 180°, we have the PKIKP travel-time branch. This PKIKP branch streaks right across the PKPab and PKPbc branches, which makes it difficult to identify these phases in seismograms.

To return to the back branch of the triplication, this arrival can have three different names. If the velocity model has a rapid but continuous velocity increase at the inner-outer core boundary, then this back branch is viewed as a ray that turns within the inner core and hence it should be dubbed a PKIKP branch. On the other hand, if the velocity model has a jump in velocity at the inner core, then this same back branch is now viewed as a reflected wave and should be dubbed PKiKP. The phase identification scheme that I use in the EduPhase graphics names this back branch PKiKP. But many seismologists prefer to use the PKP notation for all four core phases, hence the PKiKP and PKIKP branches are referred to as PKPcd and PKPdf-yes, "df" rather than "de!" Use of the PKPab through PKPdf notation frees our observational identification from any statement about the details of the inner-outer core boundary.

Let us now look at those mysterious precursors that arrive several seconds before PKIKP in the distance range from about 130° to 143°, where the precursors merge into the PKP caustic. See the seismogram record sections for a nice display of the apparent association of these precursors with the PKP caustic. To see all the details of these multiple arrivals-including the precursors which are frequently confused with PKIKP-we need to have a dense array of seismographic stations in the distance range from 135° to 145° or so. The Mich/OhioSeis stations give very dense coverage in the range from 137° to 140° for the 7 August Banda Sea event. To extend this range, I have added both IRIS and USNSN seismograms from stations located in North America. It is fairly easy to put together these record sections: All the Mich/OhioSeis seismograms are available on the station Web servers as ASCII-formatted files; ASCII-formatted IRIS seismograms are available on the Web through the Wilbur interface; and USNSN seismograms in GSE format can be obtained from the USGS by an e-mail request system. Furthermore, all these seismograms are from "broadband" instruments, which means that we see a detailed and similar wave appearance from different networks. When you click at the EduPhase Web pages to look at these seismograms, note that the PKP amplitudes are quite large at the PKP caustic; this is expected from basic wave-propagation principles.

Early seismologists, such as Sir Harold Jeffreys, noticed that the PKIKP precursors seem to extend from the PKP caustic with the same apparent velocity as the PKP waves. Jeffreys did a classic analysis of waves diffracted from a caustic point and concluded that the amplitudes of the PKIKP precursors are too large to be explained by this diffraction effect. Several other seismologists then proposed various discontinuities above the inner-outer core boundary to explain these precursors. With the availability of dense seismic arrays in the 1960's and 1970's, several seismologists analyzed the details of the PKIKP precursors. It was proposed by Haddon and others that these precursors are generated by lateral variations in the Earth's mantle just above the core-mantle boundary, the D'' layer. Several papers argued that slight variations in the lowermost mantle velocity can deflect some PKP waves away from their expected ray paths so that they arrive at closer distances before PKIKP. Nearly forty years later, there is still active research on the topic of lateral variations in the D'' layer, though most of these recent studies do not use PKIKP precursors. Perhaps the advent of easy-to-use dense regional digital networks might provide new opportunities to track the details of PKIKP precursors. Just one curious student looking at seismograms might uncover a new twist in the mystery story of the PKIKP precursors!

SRL encourages guest columnists to contribute to "EduQuakes." Please contact Larry Ruff with your ideas. His e-mail address is

Posted: 21 January 2001