Dynamic origins of seismic wavespeed variation in
Highlights
► 2-D convection models of with a phase transition. ► Synthetic waveform modeling of the lower mantle triplication. ► Significant thermal and phase heterogeneity over small distances. ► Advected isotherms enable multiple crossings of the phase boundary. ► Seismic data may reveal the evolutionary stage of slabs and plumes.
Introduction
The discontinuity is characterized by a seismic velocity increase of 1–3% approximately 250 km above the core–mantle boundary (CMB) (see review by Wysession et al. (1998)). Seismic waveform modeling detects a velocity jump beneath locations of inferred paleosubduction, including Alaska, the Caribbean, Central America, India, and Siberia. Furthermore, in seismic tomography the high velocity and deep anomalies in these regions are interpreted as slabs (e.g., Grand, 2002).
The discontinuity can explain a triplicated arrival SdS (PdP) between S (P) and ScS (PcP) observed in waveform data (Lay and Helmberger, 1983). Between epicentral distances 65–83 degrees, SdS is a composite arrival from a discontinuity reflection Sbc (Pbc) and a ray turning below the interface Scd (Pcd). The horizontally-polarized S-wave (SH) triplicated arrival is often analyzed at 10 s period. SH is usually not contaminated by other phases or mode conversions and modern array processing techniques using broadband data can recognise energy on the tangential component that originates from other phenomena such as SKS splitting (e.g., Garnero and Lay, 2003, Wookey and Kendall, 2007). Shorter periods (1 s) can detect the P-wave triplication (PdP) but data stacking is required to suppress noise.
There are generally fewer or weaker SdS (PdP) detections outside high-velocity areas which suggests subduction history influences the presence and strength of the arrival. In contrast to seismic observations, a pre-existing basal chemical layer generates strong (weak) SdS below upwellings (downwellings) (Sidorin and Gurnis, 1998, Sidorin et al., 1998) and does not generate short wavelength heterogeneity on the discontinuity (Tackley, 1998). However, detections of SdS beneath the seismically slow central Pacific suggest the discontinuity height above the CMB or velocity increase may be modulated by composition in some regions (Garnero et al., 1993, Lay et al., 2006).
Early dynamic calculations and waveform modeling propose a thermal slab interacting with a phase change with a positive Clapeyron slope can explain the origin of the discontinuity (Sidorin et al., 1999a, Sidorin et al., 1998). Incident rays are refracted by a high velocity thermal slab above and turn beneath the discontinuity in the higher velocity phase to produce the SdS arrival. Furthermore, a high-temperature thermal boundary layer generates a negative seismic velocity gradient at the CMB. This counteracts the velocity step to ensure that differential travel times (e.g., ScS–S) and diffracted waveforms match data (Young and Lay, 1987, Young and Lay, 1990). Sidorin et al. (1999b) unify regional seismic models from waveform studies by proposing a global discontinuity with an ambient phase change elevation of 200 km above the CMB and a Clapeyron slope of 6 MPa K−1. The actual boundary locally is thermally perturbed upward and downward.
The subsequent discovery of a phase transition in MgSiO3 from silicate perovskite (Pv) to “post-perovskite” (pPv) at CMB conditions supports the phase change hypothesis (Murakami et al., 2004, Oganov and Ono, 2004). Ab initio calculations at 0 K suggest that the transformation of isotropic aggregates of Pv to pPv increases the S-wave velocity by 1–1.5% and changes the P-wave velocity by −0.1% to 0.3% (Oganov and Ono, 2004, Tsuchiya et al., 2004, Iitaka et al., 2004). First principle calculations at high temperature produce similar wavespeed variations (Stackhouse et al., 2005, Wentzcovitch et al., 2006), and high pressure experiments resolve a 0–0.5% increase in S-wave velocity (Murakami et al., 2007).
Lattice-preferred orientation in the pPv phase may reconcile the small variation in wavespeeds from mineral physics with the 1–3% increase across the discontinuity deduced from seismic data (e.g., Wookey et al., 2005). Fe content in Pv may decrease (Mao et al., 2004) or increase (Tateno et al., 2007) the phase transition pressure, and Al may broaden the mixed phase region (Catalli et al., 2010, Akber-Knutson et al., 2005). However, other researchers find that compositional variations have little effect (Hirose et al., 2006, Murakami et al., 2005). Independent of the width of the mixed phase region, strain-partitioning into weak pPv can produce a sharp seismic discontinuity due to the rapid change in seismic anisotropy (Ammann et al., 2010). Thomas et al. (2011) reconcile a negative P-wave contrast from beneath the Caribbean with a positive P-wave contrast beneath Eurasia by proposing a Pv–pPv phase transition with a fraction of 12% alignment in pPv.
The phase transformation destabilizes the lower thermal boundary layer, which produces more frequent upwellings (Sidorin et al., 1999a, Nakagawa and Tackley, 2004). A dense basal layer interacting with a chemically heterogeneous slab and the Pv–pPv transition produces strong lateral gradients in composition and temperature that may also explain finer structure within (Tackley, 2011). A “double crossing” of the phase boundary occurs when pPv transforms back to Pv just above the CMB and may explain neighboring seismic discontinuities (Hernlund et al., 2005). However, the precritical reflection from the second putative phase transition is a relatively low amplitude arrival and therefore difficult to detect (Flores and Lay, 2005, Sun et al., 2006). The Clapeyron slope and transition temperature (or pressure) control the topography of the pPv layer (e.g., Monnereau and Yuen, 2007),
The CMB beneath Central America is illuminated through the propagation of seismic waves along a narrow corridor from South American subduction zone events to broad-band networks in North America. Furthermore, seismic tomography and plate reconstructions indicate deeply penetrating slab material (e.g., Grand, 2002, Ren et al., 2007). Under the Cocos Plate, the discontinuity can be modeled at constant height above the CMB with S-wave variations of 0.9–3.0% (Lay et al., 2004, Ding and Helmberger, 1997) or as an undulating north–south dipping structure from 300 to 150 km above the CMB with constant velocity (Thomas et al., 2004). The Pv–pPv transition may account for the positive jump seen in S-wave models and smaller negative P-wave contrast (Hutko et al., 2008, Kito et al., 2007, Wookey et al., 2005). Furthermore, its interaction with a buckled slab may explain an abrupt 100 km step in the discontinuity (Sun and Helmberger, 2008, Hutko et al., 2006, Sun et al., 2006). A second deeper negative reflector appears in some locations about 200 km below the main discontinuity. This may originate from the base of a slab, a plume forming beneath a slab (Tan et al., 2002), back-transformation of pPv–Pv, chemical layering (Kito et al., 2007, Thomas et al., 2004), or out-of-plane scatterers (Hutko et al., 2006).
It is now timely to reanalyze the role of slabs in the deep mantle and their seismic signature following discovery of the Pv–pPv transformation and a proposed “double crossing” of the phase boundary. In this study we follow a similar approach to Sidorin et al. (1998) using a compressible formulation and viscoplastic rheology. Compressibility promotes an irregular, more sluggish, flow field and encourages greater interaction between upper and lower boundary instabilities (Steinbach et al., 1989). Viscous heating redistributes buoyancy sources (Jarvis and McKenzie, 1980) and latent heat can reverse phase boundary distortion caused by the advection of ambient temperature (Schubert et al., 1975). These non-Boussinesq effects could play an important role in determining the lateral variations in temperature and phase that may give rise to rapidly varying waveforms. Using constraints from experimental and theoretical mineral physics we predict seismic wavespeed structure from the temperatures and phases determined by the convection calculations. Finally, we compute synthetic seismograms to analyze S, SdS, and ScS arrivals.
Section snippets
Equations
We employ the truncated anelastic liquid approximation (TALA) for infinite-Prandtl-number flow (Jarvis and McKenzie, 1980, Ita and King, 1994) with a divariant phase change using CitcomS (Zhong et al., 2000, Tan et al., 2007). The conservation equations of mass, momentum, and energy (non-dimensional) are:where is density, u velocity, P dynamic pressure, deviatoric stress
Seismic waveform modeling
We express the relative perturbation to a parameter X using the notation where is the change in X. The perturbation to the P-wave velocity, , and S-wave velocity, , can be expressed as (Tan and Gurnis, 2007):where K is the adiabatic bulk modulus, G shear modulus, density, and is equal to 0.45, which is similar to PREM at 2700 km depth. is computed directly from the geodynamic calculations (A). The perturbations to the elastic
Overview of slab dynamics
We describe typical slab dynamics using model S2 (Fig. 2) as all models exhibit similar behavior. The pPv phase boundary is 300 km above the CMB and the Clapeyron slope is 7.6 MPa K−1. Initially, the upper boundary layer thickens and the CMB region warms (Fig. 2A). The pPv region has constant thickness except where a few thin stationary plumes emanate from the lower thermal boundary layer and the phase boundary is perturbed to higher pressure. A slab forms from the upper thermal boundary layer
Discussion
Slabs rapidly advect and blanket the CMB from episodic flushing of the upper thermal boundary layer. The maximum thermal anomaly of slabs at the CMB is 0.33 (1200 K), which is at the upper bound of estimates (Steinbach and Yuen, 1994). Additionally, the simple two-phase model (Pv and pPv) and uncertainties in material properties preclude an accurate mapping of temperature and phase to seismic wavespeed. However, we focus on changes in seismic velocity gradient because the absolute magnitude of
Conclusions
The models reveal complex interaction of slabs, plumes, and the Pv–pPv transition which produces significant thermal and phase heterogeneity in over small distances. Slabs deflect the phase boundary and disturb the lower thermal boundary layer, pushing aside pre-existing upwellings. Plumes regularly develop beneath slabs and distort the phase boundary and slab morphology as they erupt from the CMB. Plume formation occurs less frequently at slab edges because a few stationary upwellings drain
Acknowledgements
We thank the Computational Infrastructure for Geodynamics (CIG) for distributing CitcomS, Don Helmberger for helpful discussions, and Jennifer M. Jackson for comments. We also thank Allen K. McNamara and an anonymous reviewer for constructive comments. This work was supported by NSF Grant EAR-0855815, and D.S. was supported by a CIW/DTM postdoctoral fellowship. Figures were produced using GMT (Wessel and Smith, 1998).
References (73)
The region
Phys. Earth Planet. Inter.
(1974)- et al.
Modelling structure beneath Central America with broadband seismic data
Phys. Earth Planet. Inter.
(1997) - et al.
Preliminary reference Earth model
Phys. Earth Planet. Inter.
(1981) - et al.
Preliminary evidence for a lower mantle shear wave velocity discontinuity beneath the central Pacific
Phys. Earth Planet. Inter.
(1993) - et al.
shear velocity heterogeneity, anisotropy and discontinuity structure beneath the Caribbean and Central America
Phys. Earth Planet. Inter.
(2003) - et al.
Sound velocity of MgSi post-perovskite phase: a constraint on the discontinuity
Earth Planet. Sci. Lett.
(2007) - et al.
Interpreting seismic structure using synthetic waveforms computed from dynamic models
Earth Planet. Sci. Lett.
(1998) - et al.
The effect of temperature on the seismic anisotropy of the perovskite and post-perovskite polymorphs of MgSi
Earth Planet. Sci. Lett.
(2005) - et al.
Effects of depth-dependent properties on the thermal anomalies produced in flush instabilities from phase transitions
Phys. Earth Planet. Inter.
(1994) Living dead slabs in 3-D: the dynamics of compositionally-stratified slabs entering a slab-graveyard above the core–mantle boundary
Phys. Earth Planet. Inter.
(2011)