AGU Fall Meeting 2012 | Talk T13H-03 |
3 December | 14:10 - 14:25 | Moscone South
Mantle hydration at the Middle America Trench: Constraints from
measurements of seismic anisotropy
Harm Van Avendonk,
Water carried to depth by subducting oceanic lithosphere is the primary source of mantle hydration, an
essential component of many arc- and global-scale geochemical and geodynamic processes. The upper mantle
may account for a majority of water fluxing into arcs, yet there are few constraints on the degree of
hydration of the subducting oceanic upper mantle (Figure 1). The upper mantle is often assumed to be
efficiently dehydrated by melting at ridges. However, recent seismic-reflection images of bending-induced
normal faults that extend into the upper mantle [e.g., Ranero, 2003], as well as significantly
reduced upper-mantle seismic velocities under the outer rise and trenches of arcs [e.g., Ivandic et al.,
2008; Contreas-Reyes et al., 2008; Van Avendonk et al., 2011], have been interpreted as evidence
that the subducting mantle is pervasively hydrated via serpentinization at the outer rise by seawater
penetrating through the crust along plate-bending-induced faults. This seawater may fill cracks in the upper
mantle with free water; react strongly with olivine in upper mantle peridotite, filling cracks and fault
zones with serpentinite; and/or diffuse between fault zones, pervasively serpentinizing the upper mantle.
The hypothesis that seawater flowing along outer-rise faults commonly hydrates the upper mantle is largely
based on isotropic seismic velocity analyses
[Ivandic et al., 2008; Contreas-Reyes et al., 2008; Van Avendonk et al., 2011] that assume observed slow
velocity anomalies can be entirely attributed to serpentinization.
Seismic velocities in serpentinized rocks are much slower than in unaltered rocks [e.g., Christensen, 1966],
enabling seismic-travel-time-based estimates of mantle serpentinization and thus the flux of water carried
into subduction zones by serpentinite. However, the outer-rise normal faults themselves, as well as an
inherited, strain-induced alignment of mineral grains (i.e., lattice-preferred orientation or LPO) in the
upper mantle and can produce azimuthally dependent seismic wave speeds that are up to ~0.5 km/s
slower in one direction than in another [e.g., Shearer and Orcutt, 1986; Hudson, 1981], an effect comparable
to the change in velocity due to ~20% pervasive serpentinization [Christensen, 1966] (Figure 2). To accurately
estimate the degree of serpentinization at the outer rise and trench using seismic travel times, the azimuthal
variation of seismic wave speed must be determined and the competing effects of LPO, cracks,
and hydration, each with their own azimuthal dependence, must be distinguished.
Schematic diagram of the subduction zone water cycle. The subducting
lithosphere is thought to deliver 3.0 to 4.4 x 105 kg of water per m2 of seafloor to the
mantle wedge and 0.2-2.4 x 105 kg/m2 to the deep mantle. Of this, sediments
and oceanic crust account for ~1.2x105 kg/m2 [Staudigel et al., 1996; Plank
and Langmuir, 1998; Kerrick and Connolly, 2001]. Flow of seawater along bending-induced faults at the
outer-rise may serpentinize the oceanic upper mantle, however estimates of this potentially significant
component of the input water flux are poorly constrained and vary from 0.6 to 4.6 kg/m2
[Schmidt and Poli, 1998; Kerrik, 2002; Ranero et al, 2003]. After Rupke et al.  with instability
models of Gerya et al. .
Comparison of the wave speed and delay time (x=50 km) effects of
anisotropic fabric and pervasive serpentinization on quassi-compressional waves (qP)
travels through the upper mantle (dunite): (a) Best-fit azimuthal qP-wave anisotropy models
for the Pacific Ocean uppermost mantle [Kawasaki and Konno, 1984], the south Pacific upper
mantle [Shearer and Orcutt, 1986], and calculated velocities predicted for an olivine matrix
with 22% LPO [Shearer and Orcutt, 1986]. (b) Effective media theory [Hudson, 1981] calculations
for qP in dunite with aligned, water- and serpentinite-filled cracks. (c) Geometrical effect
of large, serpentinite-filled, planar joints. (d) Effect of pervasive serpentinization
[Christensen, 1966]. Olivine crystal drawing is from Lev and Hager .
We are using active-source OBS data from the Middle America Trench (Figures 3 and 4) to measure wave speed
variation with azimuth in the upper mantle there. Our goal is to seperate this signal from the velocity
reduction due to serpentinization alone (Figure 5), providing an estimate for the extent and distribution
of serpentinite and thus the amount of water suducting into the mantle.
Map of OBS and airgun shots from cruise MGL0807 offshore Nicaragua.
Example OBS data from instrument site ANE03. See map in Figure 3
for instrument location.
Theoretical delay-time curves calculated for an anisotropic
upper mantle (red line) with: pervasive serpentinization (solid green); cracks filled
with serpentinite (dashed light green) or water (dashed blue); and large, serpentinized
joints (dashed dark green). These velocity effects can be separated by first measuring the
anisotropy of the unaltered mantle seaward of the outer rise (1). Then, the degree of
pervasive serpentinization is found by measuring the change in the mean delay time between
the unaltered and altered mantle (2). Finally, the extent and character of cracking and/or
jointing can be estimated by measuring the change in the amplitude and period of the
delay-time curve (3).