A PKP precursor is a few seconds of faint energy arriving ahead of the stronger core-passing waves, and it is weak enough that finding one by eye in a single earthquake record is unreliable work; the usual remedies are enhancing signal-to-noise, suppressing weaker arrivals and trying to locate where the scattering happened [3][4]. Which makes the denominator the interesting number: 174,929 detections out of more than two million recordings is a pipeline yield of at most about 8.7 percent [5][8][1]. Most of the classifier's labour was discarding records and rejecting waveforms, not finding structure, with samples repeatedly rechecked and corrected by hand to improve it [6].
Ten times a combined global baseline implies that baseline sits somewhere near 17,500 detections [8][2]. Yet a team last year reported more than 30,000 precursor signals using a neural network of its own, leaning on dense seismic arrays in a way that limited how much of the boundary it could survey [10]. Both numbers hold only if the array-based count sits outside the global baseline, which is consistent with the coverage limit described; measured against that study alone, the new catalog is about 5.8 times larger [3]. Coverage is the advance being claimed, and the authors describe the result as the most spatially complete and densest global map of core-mantle boundary scattering so far [11].
What the catalog does not deliver is an edge. The output is two probability maps, one for how consistently precursors appear and one for where strong scatterers are likely to sit [7], and the authors are explicit that these mark probable scattering regions rather than outlines, since a precursor recorded at a single station cannot pin down the scatterer that produced it [14]. Earthquakes and seismometers are unevenly distributed, so precursor coverage is uneven too [15]. A belt drawn through that sampling has to survive the possibility that the space between two older patches was never listened to rather than genuinely continuous.
The mechanism the authors read out of the pattern, a lowermost mantle shaped by ancient subducted slabs, chemical segregation and localized melting [13], is inference from where the scattering is, not a measurement of what the material is. It matters because that boundary regulates heat flow, mantle circulation and the rise of volcanic plumes [17]. Converting six probable belts into six described structures needs receivers where the sampling is thinnest, and other seismic wave types to constrain depth and size, which is the follow-up work the authors name [16].
