PKPPKP

The phase named PKPPKP is often abbreviated as P'P'. A ray diagram of this phase is displayed in figure 3.3 a).

Figure 3.3: a) Ray paths of the phases PKPPKP (P'P') and the underside reflection at the 660. The underside reflections are written as P$^d$P, where d denotes the depth of the discontinuity (here: P$^{660}$P). The crust, the 660, the core-mantle boundary (CMB) and inner-core boundary (ICB) are marked. P´P´ consists of two legs of the phase PKP and is reflected once at the free surface
b) Same as a) for phases PP and P$^d$P (P$^{660}$P). PP consists of two legs of P and is reflected once from the free surface between source and receiver. Instead of the reflection from the free surface, P$^d$P is reflected from the upper mantle discontinuities. The ray path for an upperside reflection from the 660 (Pp$_{660}$p) is added. The wave is reflected at the free surface near the receiver and from the upper side of the upper mantle discontinuity.

A part of the P'P' energy is reflected at the discontinuities beneath the surface reflection point and appears as precursor to P'P' in the seismogram.
The study of P'P' precursors (P'$^d$P', d denotes the depth of the discontinuity) has the advantage that the P'$^d$P' phases arrive in a quiet time window before the P'P' phase, where no other phases arrive.
The reflections of P'P' within the crust and mantle were first mentioned by Gutenberg and Richter (1934). Several detections of reflections from deeper reflectors followed (Gutenberg, 1960; Adams, 1968; Engdahl and Flinn, 1968).
In contrast, Haddon et al. (1977) showed that most precursors to P'P' can be explained by scattering of energy in the upper mantle beneath the receiver or in the lowest part of the mantle (D'') and are inconsistent with underside reflections at the discontinuities beneath the reflection point. Only the precursors assigned to the 660 might be consistent with the P'$^d$P' interpretation.
The spectrum of P'P' shows very high frequencies. Since the high frequencies can be reflected only by sharp reflectors, the existence of P'$^{660}$P' arrivals are interpreted as the existence of a sharp discontinuity at this depth (Richards, 1972).
A detailed study of digital broad-band (bb) array data (Davis et al., 1989) showed little evidence of P'$^{660}$P' phases. This points to a gradient zone with a thickness of 15 - 30 km, or to the existence of topography with an amplitude of $\pm$10 km on scalelengths of $\sim$300 km. The existence of P'$^{660}$P' phases is supported by array studies of Nakanishi (1988, 1989) and Xu et al. (1999) which show strong evidence for the existence of this phase. Despite the long time and the different methods used to study P'$^d$P' phases, the existence of these phases is still under discussion.