Along age-progressive hotspot volcano chains, the emplacement rate of igneous material varies through time. Time-series analysis of changing emplacement rates at a range of hotspots finds that these rates vary regularly at periods of a few to several tens of millions of years, indicative of changing melt production within underlying mantle plumes. Many hotspots exhibit at least one period between ∼2 and 10 Myr, consistent with several proposed mechanisms for changing near-surface plume flux, and thus melting rate, such as small-scale convection, solitary waves and instability formation in tilted plume conduits. Here, we focus on quantifying instability growth within plumes tilted by overlying plate motion. Previous studies using fluids with constant or temperature-dependent viscosity suggest that such instabilities should not form under mantle conditions. To test this assertion, we use a modified version of the finite element code ASPECT to simulate 400 Myr of evolution of a whole-depth mantle plume rising through the transition zone and spreading beneath a moving plate. In a 2-D spherical shell geometry, ASPECT solves the conservation equations for a compressible mantle with a thermodynamically consistent treatment of phase changes in the mantle transition zone and subject to either a temperature- and depth-dependent linear rheology or a temperature-, depth- and strain-rate dependent non-linear rheology. Additionally, we examine plume evolution in a mantle subject to a range of Clapeyron slopes for the 410 km (1–4 MPa K–1) phase transitions. Results suggest that plume conduits tilted by >67° become unstable and develop instabilities that lead to initial pulses in the transition zone followed by repeated plume pulsing in the uppermost mantle. In these cases, pulse size and frequency depend strongly on the viscosity ratio between the plume and ambient upper mantle. Based upon our results and comparison with other studies, we find that the range of statistically significant periods of plume pulsing in our models (∼2–7 Myr), the predicted increase in melt flux due to each pulse (3.8–26 × 10−5 km3 km−1 yr−1), and the time estimated for a plume to tilt beyond 67° in the upper mantle (10–50 Myr) are consistent with observations at numerous hotspot tracks across the globe. We suggest that pulsing due to destabilization of tilted plume conduits may be one of several mechanisms responsible for modulating the melting rate of mantle plumes as they spread beneath the moving lithosphere.
- Award ID(s):
- 1849949
- NSF-PAR ID:
- 10387859
- Date Published:
- Journal Name:
- Geophysical Journal International
- Volume:
- 225
- Issue:
- 2
- ISSN:
- 0956-540X
- Page Range / eLocation ID:
- 906 to 925
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
SUMMARY -
SUMMARY Within the Western Branch of the East African Rift (EAR), volcanism is highly localized, which is distinct from the voluminous magmatism seen throughout the Eastern Branch of the EAR. A possible mechanism for the source of melt beneath the EAR is decompression melting in response to lithospheric stretching. However, the presence of pre-rift magmatism in both branches of the EAR suggest an important role of plume-lithosphere interactions, which validates the presence of voluminous magmatism in the Eastern Branch, but not the localized magmatism in the Western Branch. We hypothesize that the interaction of a thermally heterogeneous asthenosphere (plume material) with the base of the lithosphere enables localization of deep melt sources beneath the Western Branch where there are sharp variations in lithospheric thickness. To test our hypothesis, we investigate sublithospheric mantle flow beneath the Rungwe Volcanic Province (RVP), which is the southernmost volcanic center in the Western Branch. We use seismically constrained lithospheric thickness and sublithospheric mantle structure to develop an instantaneous 3D thermomechanical model of tomography-based convection (TBC) with melt generation beneath the RVP using ASPECT. Shear wave velocity anomalies suggest excess temperatures reach ∼250 K beneath the RVP. We use the excess temperatures to constrain parameters for melt generation beneath the RVP and find that melt generation occurs at a maximum depth of ∼140 km. The TBC models reveal mantle flow patterns not evident in lithospheric modulated convection (LMC) that do not incorporate upper mantle constraints. The LMC model indicates lateral mantle flow at the base of the lithosphere over a longer interval than the TBC model, which suggests that mantle tractions from LMC might be overestimated. The TBC model provides higher melt fractions with a slightly displaced melting region when compared to LMC models. Our results suggest that upwellings from a thermally heterogeneous asthenosphere distribute and localize deep melt sources beneath the Western Branch in locations where there are sharp variations in lithospheric thickness. Even in the presence of a uniform lithospheric thickness in our TBC models, we still find a characteristic upwelling and melt localization beneath the RVP, which suggest that sublithospheric heterogeneities exert a dominant control on upper mantle flow and melt localization than lithospheric thickness variations. Our TBC models demonstrate the need to incorporate upper mantle constraints in mantle convection models and have global implications in that small-scale convection models without upper mantle constraints should be interpreted with caution.
-
Abstract The Earth's long‐ and intermediate‐wavelength geoid anomalies are surface expressions of mantle convection and are sensitive to mantle viscosity. While previous studies of the geoid provide important constraints on the mantle radial viscosity variations, the mantle buoyancy in these studies, as derived from either seismic tomography or slab density models, may suffer significant uncertainties. In this study, we formulate 3‐D spherical mantle convection models with plate motion history since the Cretaceous that generate dynamically self‐consistent mantle thermal and buoyancy structures, and for the first time, use the dynamically generated slab structures and the observed geoid to place important constraints on the mantle viscosity. We found that non‐uniform weak plate margins and strong plate interiors are critical in reproducing the observed geoid and surface plate motion, especially the net lithosphere rotation (i.e., degree‐1 toroidal plate motion). In the best‐fit model, which leads to correlation of 0.61 between the modeled and observed geoid at degrees 4–12, the lower mantle viscosity is ∼1.3–2.5 × 1022 Pa⋅s and is ∼30 and ∼600–1,000 times higher than that in the transition zone and asthenosphere, respectively. Slab structures and the geoid are also strongly affected by slab strength, and the observations prefer moderately strong slabs that are ∼10–100 times stronger than the ambient mantle. Finally, a thin weak layer below the 670‐km phase change on a regional scale only in subduction zones produces stagnant slabs in the mantle transition zone as effectively as a weak layer on a global scale.
-
Abstract The timing and rate of decompression melting of a compositionally heterogeneous mantle during continental rifting are assessed with a new one‐dimensional geodynamic code, MELT1D. MELT1D computes pressure and temperature in the extending lithosphere and rising asthenosphere and calculates the resulting melt fraction and eruption rates for different lithologies. A series of models simulate syn‐rift melt production from (a) dry and wet depleted lherzolite similar to the mantle that underlies most mid‐ocean ridges, (b) dry and wet relatively fertile ultramafic compositions representing plume or primitive mantle material, (c) pyroxenite representing recycled ultramafic oceanic crust or magmatic metasomes, and (d) basalt representing recycled mafic crust or metasomes. The models predict sequential melting of the different compositions that is broadly consistent with basalt eruption histories in many Phanerozoic rifts. Results show a progressive transition in magma sources as the lithosphere thins, beginning with melting of wet mantle and compositionally fertile mafic components near the lithosphere‐asthenosphere boundary during the earliest stages of extension. This transitions to magmatism dominated by melting of relatively fertile ultramafic components (pyrolitic and pyroxenitic compositions) as extension progresses, and finally to melting of ambient lherzolite asthenosphere as lithosphere thinning approaches breakup. Mantle composition, pre‐rift lithosphere thickness, and mantle temperature exert the greatest controls on the timing and volumes of magmas produced from each lithology. In general, a cool or thick lithosphere has a greater capacity to sequester fertile lithologies than thin or warm lithosphere, and thus has a greater capacity to produce early syn‐rift magmas without requiring a hot mantle plume.
-
The structure of the lithosphere-asthenosphere boundary (LAB) beneath oceanic plates is key to understanding how plates interact with the underlying mantle. Prior contradictory geophysical observations have been used to argue for a thin, melt-rich boundary that decouples the plate from the rest of the mantle, or for a much broader anisotropic and thermally controlled boundary that indicates significant coupling with the rest of the mantle. The predictions of models based on these interpretations can be tested most easily in a subduction zone setting where the steady increase in pressure at the base of the subducting plate’s LAB will have differing effects on melt and anisotropy. Melt remains stable within the mantle to ~150-250 km (for carbonate melt) or to ~330 km (for silicate melt), while anisotropy induced by different processes should have no significant change until ~250 km to ~440 km depth. We calculate P-to-S receiver functions (PRFs) using varying frequency bands at broadband seismic stations with >4 years of data from the Servicio Geológico Colombiano’s Red Sismológica Nacional de Colombia to investigate the characteristics of the LAB of the subducting Nazca oceanic plate from the coast to the Andean foreland (corresponding to slab LAB depths of ~50 km to >400 km). The use of PRFs permits identification and analysis of anisotropy across the boundary while calculation at a range of frequency bands permits tuning of the PRFs to differing spatial scales to determine the size and abruptness of the boundary. We find that the P-to-S converted phase of the subducted Nazca plate’s LAB is detectable 4-5 seconds after the converted phase of the plate’s Moho to at least ~150 km depth. Assuming the slab has an average Vp/Vs of 1.75 to 1.78 and Vp of 8.2 km/s (+2.5% dVp), this corresponds to a plate thickness of ~50 km, matching the expected thickness given the Nazca plate’s age in the region (~10-20 Myrs). We find that the Nazca plate’s LAB is most consistently detectable in the <0.24 Hz band and largely undetectable in the <2.4 Hz band, indicating the LAB is gradational and between 10 and 30 km in thickness. Amplitude variations and complexities in the LAB converted phases further indicate that the boundary marks a change in anisotropy most consistent with the LAB representing a sheared zone between the plate and underlying mantle.more » « less