A New Phase Change in Earth's Deep Mantle: What Does This Mean for the Rest of Us? II
Presiding: T Rushmer, University of Vermont; A Hofmeister, Washington University
U32A-01 10:30h
Slab Detachment in the Deep Mantle
U32A-02 10:45h
Microscale Deformation of (Post) Perovskite-Dominated Sediment in the Uppermost Outer Core
Seismic and nutational data hint strongly at a layer comprised of silicate sediments several km thick confined to the top of the liquid outer core, directly beneath the core-mantle boundary and equating with observed ultra low velocity zones (ULVZs)1. Its origin is thought to be due to high pressure chemical reactions that take place between Fe-rich silicate in the lower mantle and liquid iron in the underlying outer core. We speculate that the Fe silicate is dominated by the newly discovered post-perovskite phase (ppv). Initial numerical investigations show that viscous compaction in the sediment layer will act to expel interstitial core metal liquid and reduce an initial 50% porosity to a residual value of < 0.1 on timescales of the order 80-100 Ma1. Using a modified form of Biot's equations2, we show that deformation of a poro-viscoelastic sediment pile will respond by drawing up Fe liquid metal into the layer from below at a rate proportional to the shear stress rate. Instead of a static, isolated residual porosity distribution in the most compacted upper regions of sediment, we envisage a more dynamic environment where fresh core liquid is emplaced periodically into the slowly accumulating pile. Estimates of the potential magnitude of the instability, including pressure changes, local fluid flow rates and timescales compare favourably with fluid motions in the convecting outer core. A key difference between both models, which are in fact complementary, relates to the rheology and microscale deformation behaviour of the assumed ppv-dominated sediment. The small grain size of the suspension is close to the limit of dilatant behaviour (c. 10-6 m) in granular materials. More information is required on the chemical and physical behaviour of the post-perovskite phase at lengthscales characteristic of geophysically interesting colloidal suspensions under high P-T conditions. 1.Buffet, BA., Garnero, EJ & Jeanloz, R. 2000. Science, 290, 1338-1342. 2.Koenders, MA & Petford, N. 2000. Geophys. Res. Lett. 27, 1231-1234.
U32A-03 INVITED 11:05h
Imaging Reflectors in the Deep Mantle; Post-Perovskite or Not?
The discovery of the post-perovskite phase transition presents a new context for interpreting seismological observations of deep mantle structure. As is the case for seismic velocity discontinuities detected in the upper mantle transition zone, interpretation as an expression of a phase change provides an opportunity to infer thermal and dynamical effects that are otherwise not directly sensed by seismology. Extensive observations of P and S wave velocity increases several hundred kilometers above the core-mantle boundary in different regions of the deep mantle appear to be generally compatible with the presence of a post-perovskite transition. The tendency for S wave reflections to be stronger and for the ratio of Vs/Vp to be higher in the D" layer beneath the velocity increases can be accounted for by the phase transition. If the post-perovskite transition is indeed the root cause of the seismic discontinuity, small scale variations in the discontinuity may provide a probe of thermal and chemical variations. One of the most intensively studied regions of the deep mantle is beneath the Cocos plate, where P and S waves from deep earthquakes under South America bottom in the deep mantle before being observed at the dense BDSN and TRINet networks of broadband stations in California. We apply simplified seismic migration procedures to image the P and S velocity structure under the Cocos plate. Use of 3D tomographic models to account for volumetric heterogeneity allows us to constrain topography on the seismic reflectors and their relative strengths. S reflections are stronger and more readily migrated than P reflections in this region, and an extensive lateral reflector is found in the S images. While this appears to be rather uniform in depth over a 700 km extent, an abrupt change in depth occurs near 5N, with perhaps 50 km change in depth of the reflector. Explaining this abrupt change in the framework of post-perovskite requires some chemical contribution beyond a thermal effect. We also find that there is no evidence for a radial decrease in velocity below the D" discontinuity. There is however, evidence for scattering from a localized low velocity structure offset from the reflector. Implications for post-perovskite interpretations will be considered.
U32A-04 11:25h
Origin of SdS and PdP and Implications for a D" Phase Change
Precursors to ScS and PcP seismic phases, often labeled as SdS or PdP, have been included in evidence for a sharp discontinuity in seismic velocities at depths between 100 and 300 km above the core-mantle boundary (the D"region). The poor lateral coherence of observed SdS and PdP may be consistent with either the effects of constructive interference of distributed heterogeneity in D"or topography on a discontinuity or both. P and S wavefields interacting with 2-D models of D" heterogeneity are modeled by a pseudospectral method. Structures examined include a discontinuity with topography, spatially continuous heterogeneity, and bimodal models of heterogeneity containing either high velocity inclusions (solid-solid phase changes) or low velocity inclusions (partial melt regions). Deterministic models of topography on a D" discontinuity are generated from tomograms, an assumed temperature derivative of velocity, and a Clapeyron slope of a perovskite phase change. Synthetics from these models are compared with 2-D waveform profiles to determine whether existing tomograms of the D" region can predict the observation of PdP and SdS phases, their amplitudes relative to PcP and ScP, and their lateral coherence.
U32A-05 INVITED 11:40h
Imaging Structure at and Near the Core Mantle Boundary With a Generalized Radon Transform of Broad Band ScS and SKKS Coda Waves
Seismic constraints on the core-mantle boundary (CMB) and on the character of any nearby structures and interfaces are of great importance for understanding the mineralogy, phase chemistry, and dynamics in Earth's lowermost mantle. The recently discovered post-perovskite transition is an obvious target of our research, but our technique could also reveal hitherto unknown structures. The exponentially growing data sets available through, e.g., IRIS call for powerful inversion methods. In order to produce accurate images of scatterers or interfaces near the CMB from seismic body waves such as ScS and SKKS - and their codas - we have developed a generalized Radon transform (GRT) from 'exact' asymptotic analysis and the theory of Fourier integral operators. We enhance the images, estimate uncertainty, and infer scaling properties of the interfaces using mixed-model statistics (i.e., properties of the GRT imaging operator are used for the characterization and reduction of noise). Our inverse scattering approach is set up to extract information from tens of thousands of broad-band waveforms. We present preliminary results of GRT imaging of the CMB beneath Central America (from -115 to --65W and -10S to 40N), using some 65,000 ScS phases generated by earthquakes with an Mb magnitude larger than 5.2. Specifically, we will present a ~2000 km great circle transect from (-105W, 0) to (-75W, 30N), showing a sharp CMB and a more gradual interface approx. 280 km above it, with hints of structure in between. From this we can begin to retrieve the lateral variation in depth (related to the Clapeyron slope in case of a phase transition) and the regularity of interfaces in the bottom 300 km or so of the mantle. These lateral variations will be discussed in the context of wavespeed variations inferred from global tomography.