This section was contributed by Michael Leyton [email protected]

Geoneutrino flux predictions depend on assumptions about how heat-producing elements are distributed throughout the crust and mantle. Seismic tomography is one of the few direct constraints we have on large-scale mantle structure. This page hosts an interactive 3D viewer built from two independent tomographic datasets, alongside an explanation of what each one shows and how they compare.

Interactive 3D Viewer

Rotate, zoom, and filter two global mantle models — plus a model-agreement comparison — rendered as a 3D point cloud from the core-mantle boundary up through the transition zone.

Launch the interactive viewer »

What am I looking at?

The viewer plots seismic velocity anomalies at thousands of points inside the mantle, colored by value, with present-day coastlines drawn as a thin overlay for geographic reference. Two prominent features to look for are the Large Low-Velocity Provinces (LLVPs) - two continent-sized regions of anomalously slow shear velocity sitting on the core-mantle boundary beneath Africa and the Pacific - and fast, sheet-like anomalies elsewhere that are generally interpreted as subducted oceanic lithosphere (old slabs). The viewer includes a “reveal LLVPs” preset that jumps straight to filter settings where these structures stand out.

Model A - S40RTS (δV/V)

A whole-mantle shear-velocity model, expressed as percent deviation from a reference Earth model (δVs/Vs). Built from Rayleigh wave dispersion, teleseismic body-wave traveltimes, and normal-mode splitting functions.

45 depth slices from 24-2891 km (crust to core-mantle boundary), downsampled to a 2°×2° grid.

Model B - SubMachine Vote Map

A tomographic vote count (0-18): the number of independently published tomography models that agree an anomaly exists at a given point, after each is thresholded to a binary fast/slow mask and stacked. A high vote count means broad cross-model consensus, not a larger anomaly.

45 depth slices from 1111-2871 km, downsampled to the same 2°×2° grid as Model A.

Comparing the two models

δV/V and vote count are different physical quantities, so subtracting them directly would not mean anything. The viewer instead offers a normalized difference: S40RTS is interpolated onto the vote map’s depth levels, both fields are independently z-scored (converted to standard deviations from each model’s own mean), and then subtracted (diff = z(S40RTS) − z(vote map)). The result highlights where the two models disagree about how anomalous a region is, in unit-free terms, rather than claiming a literal difference in physical units.

Using the viewer

  • Model - switch between S40RTS, the vote map, and the normalized difference. Each remembers its own threshold, depth range, and polarity settings, so you can flip back and forth without losing your place.
  • Threshold - hides points below a minimum anomaly strength (or vote count) so only the strongest signal remains visible.
  • Polarity - isolate slow anomalies (LLVPs) or fast anomalies (slabs) for the two signed models.
  • Depth range - restrict the point cloud to a shell of the mantle, e.g. just the lowermost few hundred kilometers above the core.
  • Drag to rotate, scroll to zoom, and toggle the coastline overlay, translucent surface shell, and core-mantle boundary reference ring independently.

References

Ritsema, J., Deuss, A., van Heijst, H.J., and Woodhouse, J.H. (2011), S40RTS: a degree-40 shear-velocity model for the mantle from new Rayleigh wave dispersion, teleseismic traveltime and normal-mode splitting function measurements, Geophysical Journal International, 184(3), 1223-1236.

Hosseini, K., Matthews, K.J., Sigloch, K., Shephard, G.E., Domeier, M., and Tsekhmistrenko, M. (2018), SubMachine: Web-Based Tools for Exploring Seismic Tomography and Other Models of Earth’s Deep Interior, Geochemistry, Geophysics, Geosystems, 19, 1464-1483.

Natural Earth, 1:110m Physical Vectors - Coastline, used for the present-day continent overlay.