Research ArticleEARTH SCIENCES

Compositional mantle layering revealed by slab stagnation at ~1000-km depth

See allHide authors and affiliations

Science Advances  10 Dec 2015:
Vol. 1, no. 11, e1500815
DOI: 10.1126/sciadv.1500815
  • Fig. 1 Seismic evidence for slab stagnation in the uppermost lower mantle.

    Cross sections through the P-wave model GAP-P4 (16) (A and D) and the S-wave model S40RTS (15) (B and E) centered on South Asia (top row) and South America (bottom row). (C and F) Resolution kernels quantify the depth range, over which the S40RTS S-wave velocity structure is mapped for a hypothetical point anomaly at 900 km. These kernels demonstrate that flat slabs imaged in the uppermost lower mantle beneath Peru and Indonesia are well resolved (that is, independently resolved from MTZ heterogeneity).

  • Fig. 2 Density of subducted slabs and ambient mantle as a function of XLM.

    (A) Semianalytical solution for the density of subducted slabs relative to the ambient mantle. The black-to-gray curves show the thermal density anomaly of subducted slabs of ages at the trench as annotated (in millions of years) relative to pyrolitic mantle (blue reference curve). Thermal density anomalies are averages over the cool slab core of thickness 250 km, calculated using a parameterization for depth-dependent thermal expansivity and accounting for thermal diffusion during slab sinking (see “Methods for slab-sinking models”). Slab-sinking speeds in the upper and lower mantle are assumed to be 60 and 6 km/My, respectively (52). Any compositional density difference between the slab and pyrolite is ignored because of a density trade-off between the slab’s basaltic and harzburgitic domains. For comparison, colored curves show densities (relative to pyrolite) of lower mantle compositions enriched in basalt (see table S3). Dashed lines are extrapolations. (B) Colored curves are calculated from the absolute density profiles of mantle materials [taken from Xu et al. (22)]. Black arrows mark depth ranges, in which basalt is strongly negatively or positively buoyant.

  • Fig. 3 Numerical-model predictions for three regimes of slab-sinking behavior.

    (A to C) Arrows denote velocity vectors and colors indicate potential temperature in snapshots at 150 My (for time series, see fig. S3). The 5% iso-contours of harzburgite (black) and basalt fraction (gray) are labeled. Domains with XLM > 0 are shown with dark red hatching. Slab-sinking behavior varies as a function of XLM (as annotated) between the cases shown. (D) Map of slab-sinking regimes as a function of all parameters. Mantle parameters Γ and XLM are varied between big squares; slab parameters τ and β are varied between small squares (that is, within each big square). Cases (A to C) are labeled. For Γ = −3 MPa/K, slabs are always predicted to stagnate at the 660 (not shown). Bottom scale: Mg/Si of the lower mantle, calculated from XLM [according to Workman and Hart (2)].

  • Fig. 4 Compositional mantle layering predicted by a global-scale thermochemical mantle-convection model (case A1) after ~4.57 Gy model time.

    (A) Relative to the asthenosphere, the shallow lower mantle is enhanced by basalt (see translucent bars). (B) Model snapshot of composition (see fig. S6 for time series).

  • Fig. 5 Detections of lower-mantle seismic discontinuities in the depth range of 700 to 2000 km.

    (A) Global distribution of detections shows that most detections (>75%) occur near (<1000 km) subduction zones (green lines); only ~25% occur near hot spots (green dots). (B) Histogram of detection depths shows a peak at depths of about 1000 to 1100 km. See table S2 for data sources.

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/1/11/e1500815/DC1

    Fig. S1. Numerical-model predictions of slab descent through a mantle with a gradual increase in viscosity between 660- and 1500-km depths.

    Fig. S2. Histogram of predicted stagnation depths for slabs that stagnate in the uppermost lower mantle.

    Fig. S3. Numerical-model predictions of slab descent as a time series.

    Fig. S4. Initial condition of the center of the numerical-model box for a case with τ = 50 My, β = 45°, and XLM = 10%.

    Fig. S5. Compositional mantle evolution predicted by global-scale geodynamic models for different lower-mantle density profiles of basalt.

    Fig. S6. Compositional mantle evolution predicted by global-scale geodynamic models.

    Table S1. Notations.

    Table S2. Sources of data for Fig. 5.

    Table S3. Hypothetical (molar) abundances of major oxides in the lower mantle.

    References (8196)

  • Supplementary Materials

    This PDF file includes:

    • Fig. S1. Numerical-model predictions of slab descent through a mantle with a gradual increase in viscosity between 660- and 1500-km depths.
    • Fig. S2. Histogram of predicted stagnation depths for slabs that stagnate in the uppermost lower mantle.
    • Fig. S3. Numerical-model predictions of slab descent as a time series.
    • Fig. S4. Initial condition of the center of the numerical-model box for a case with τ = 50 My, β = 45°, and XLM = 10%.
    • Fig. S5. Compositional mantle evolution predicted by global-scale geodynamic models for different lower-mantle density profiles of basalt.
    • Fig. S6. Compositional mantle evolution predicted by global-scale geodynamic models.
    • Table S1. Notations.
    • Table S2. Sources of data for Fig. 5.
    • Table S3. Hypothetical (molar) abundances of major oxides in the lower mantle.
    • References (81–96)

    Download PDF

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article