 |
| Titel |
Predicting lower mantle heterogeneity from 4-D Earth models |
| VerfasserIn |
Nicolas Flament, Simon Williams, Dietmar Müller, Michael Gurnis, Dan J. Bower |
| Konferenz |
EGU General Assembly 2016
|
| Medientyp |
Artikel
|
| Sprache |
en
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 18 (2016) |
| Datensatznummer |
250132999
|
| Publikation (Nr.) |
 EGU/EGU2016-13558.pdf |
|
|
|
|
|
| Zusammenfassung |
| The Earth’s lower mantle is characterized by two large-low-shear velocity provinces
(LLSVPs), approximately ∼15000 km in diameter and 500-1000 km high, located under
Africa and the Pacific Ocean. The spatial stability and chemical nature of these LLSVPs are
debated. Here, we compare the lower mantle structure predicted by forward global mantle
flow models constrained by tectonic reconstructions (Bower et al., 2015) to an analysis of five
global tomography models. In the dynamic models, spanning 230 million years, slabs
subducting deep into the mantle deform an initially uniform basal layer containing 2% of the
volume of the mantle. Basal density, convective vigour (Rayleigh number Ra), mantle
viscosity, absolute plate motions, and relative plate motions are varied in a series of model
cases. We use cluster analysis to classify a set of equally-spaced points (average
separation ∼0.45º) on the Earth’s surface into two groups of points with similar
variations in present-day temperature between 1000-2800 km depth, for each model
case. Below ∼2400 km depth, this procedure reveals a high-temperature cluster in
which mantle temperature is significantly larger than ambient and a low-temperature
cluster in which mantle temperature is lower than ambient. The spatial extent of
the high-temperature cluster is in first-order agreement with the outlines of the
African and Pacific LLSVPs revealed by a similar cluster analysis of five tomography
models (Lekic et al., 2012). Model success is quantified by computing the accuracy
and sensitivity of the predicted temperature clusters in predicting the low-velocity
cluster obtained from tomography (Lekic et al., 2012). In these cases, the accuracy
varies between 0.61-0.80, where a value of 0.5 represents the random case, and the
sensitivity ranges between 0.18-0.83. The largest accuracies and sensitivities are
obtained for models with Ra ≈ 5 x 107, no asthenosphere (or an asthenosphere
restricted to the oceanic domain), and a basal layer ∼ 4% denser than ambient mantle.
Increasing convective vigour (Ra ≈ 5 x 108) or decreasing the density of the basal
layer decreases both the accuracy and sensitivity of the predicted lower mantle
structure.
References:
D. J. Bower, M. Gurnis, N. Flament, Assimilating lithosphere and slab history in 4-D
Earth models. Phys. Earth Planet. Inter. 238, 8-22 (2015).
V. Lekic, S. Cottaar, A. Dziewonski, B. Romanowicz, Cluster analysis of global lower
mantle tomography: A new class of structure and implications for chemical heterogeneity.
Earth Planet. Sci. Lett. 357, 68-77 (2012). |
|
|
|
|
|
|