Summary
At epicentral distances less than 30°, the steep‐angle reflections PcP and ScP from the core–mantle boundary (CMB) are normally difficult to identify and rarely reported. The reason is that the reflection coefficient at the CMB for these phases is relatively small and PcP and ScP onsets are often disturbed by interference with other more prominent phases. However, the absolute traveltimes of PcP and ScP in the steep‐angle range are the most sensitive seismic phases for depth undulations of the CMB and, in addition, of all CMB reflections, these phases are influenced least by lateral heterogeneities along their ray paths.
During this study, a set of more than 200 such reflections from the Earth's core, observed at European stations with reflection points below Europe, was collected and analysed. Starting from observations at the GERES array in Germany, PcP and ScP could also be identified at other European seismic stations. The stations used for this study were the Fennoscandian arrays NORSAR, NORES, and FINES, the Gräfenberg array in Germany, and the stations of the German Regional Seismic Network (GRSN). With the combined investigation of PcP and ScP, a map of reflection points of the CMB below Europe could be constructed. To reduce the influence of location and other errors, traveltime differences were measured between the onsets of P (or Pn), PcP and ScP. These traveltime differences show positive and negative deviations from theoretical values of up to several seconds. CMB undulations of up to ±6 km as discussed in the literature change the absolute traveltimes of PcP (or ScP) only by about 0.9 (or 1.2) s for an epicentral distance of 20°.
The observed residuals can be explained by several factors still unresolved location errors, lateral heterogeneities along the whole ray paths of all investigated phases, or undulations of the depth of the CMB. All observed traveltime effects must be known at least one order of magnitude better than the expected effects to resolve statistically reliable CMB undulations. Inverting the residuals for possible CMB undulations shows that if there is any CMB signal in the data, it is very small. Most of the observed residuals can easily be explained by smaller changes in event depth, lateral heterogeneities in the source region and along the whole ray paths, and/or station corrections but the observations do not require any CMB topography. If such undulations of up to ±6 km exist, all other traveltime effects will mask them. Coming to this disillusioning result by using the most sensitive data to resolve a CMB topography can only lead to this conclusion; all models of CMB undulations derived from traveltime data must be taken as very unsure and any far reaching interpretations of such models should be avoided.