Geoscience Reference
In-Depth Information
These critical scales, below which porous flow
(hence channel flow that is much faster) domi-
nates, correspond to almost entire or significant
fraction of the mantle wedge, and RT-instability
(hence diapiric ascent that is much slower) could
be unrealistic. However, since viscosity and other
physical properties are not tightly constrained,
the results of scaling analysis allow variable in-
terpretations. Accordingly, for subduction zones,
many authors argue for different mechanisms,
e.g., hot partially molten diapirs (Tatsumi
et al
.,
1983), RT instability of serpentinite layer to gen-
erate a cold plume (Gerya & Yuen, 2003), a
focused fluid flow towards a volcanic front by
porous flow (Spiegelman & McKenzie, 1987) or
through a systematically oriented dyke system
(Furukawa, 1993).
Seismic velocity structures are useful to con-
strain the distribution of aqueous fluid and melt,
including the fluid fraction (
φ
) and the micro-
scopic geometrical configuration of fluid-solid
mixture represented by an equivalent aspect ra-
tio (
α
) of the fluid phase (Takei, 2002). Based
on the results of
V
P
-
V
S
tomography beneath NE
Japan, Nakajima
et al
. (2005) argued that, in the
mantle wedge, a few percent of connected melt
exists along the grain boundary (i.e., equilibrium
geometry), suggesting porous flow is dominant
in the region above 90 km depth. They use the
average
V
S
/V
P
ratio to infer the equilibrium
geometry, yet the data show a significant scat-
ter (Figure 13.4(a)). The wide scatter may be
explained by coexistence of porous and channel
flows in a fractured porous media (Sahimi, 1995),
since a relatively large
V
S
/V
P
∼
assumed to be partitioned into both the channels
and the grain boundaries, the calculated decrease
in
V
P
and
V
S
(Figure 13.4(b)) broadly reproduce the
observations (Figure 13.4(a)). In addition to the
seismic velocity decrease by the presence of fluid
as above, the presence of water may modify the
seismic properties of the NAMs. In Figure 13.4(b),
this effect has been neglected as the water content
in NAMs hence the velocity reduction is small in
this case, except for the anelasticity effect which
is not well constrained at present (Karato, 2011).
In this case, segregation of fluid phases occurs
in a relatively short time scale from hydrated
or molten regions, suppressing the local fluid
fraction to be consistent with geochemical mass-
balance based on the isotopic compositions of
the arc basalts, typically 0.1 to 1 wt % in the
mantle wedge (e.g., Ishikawa & Nakamura, 1994;
Taylor & Nesbitt, 1998; Nakamura
et al
., 2008;
Nakamura & Iwamori, 2009). Consequently, RT
instability or hot molten diapir may not work
effectively.
The above arguments suggest that solid and
fluid phases migrate separately: i.e., corner flow
of solid is induced by the subducting slab and
fluid phases ascend upwards through the solid
flow. Solid materials passing through the corner
region beneath the magmatic-hydrothermal arc
undergo significant hydration and are saturated
with H
2
O,andmayretain0.1to0.4wt%H
2
O
in nominally anhydrous phases even after break-
down of the major hydrous phases. As a result,
a hydrous boundary layer (hereafter referred to
as HBL) of 20-40 km thick almost enevitably
develops just above the subducting slab (Iwamori,
2007). The HBL continues to subduct and can
carry water to the transition zone as discussed in
the next section.
2 is associated
with cracks/dykes (
α
∼
0.001), while a relatively
small
V
S
/V
P
∼
1 is associated with porous
flow along the grain boundaries (
α
∼
1) (Takei,
2002). Figure 13.4(b) shows calculated
V
P
and
V
S
, based on the fluid distribution beneath the
NE Japan arc predicted by numerical modeling
(Iwamori & Zhao, 2000) and the theoretical rela-
tionship among
φ
,
α
,
V
P
and
V
S
(Takei, 2002). In
the numerical modeling, both aqueous fluid and
melt occur below the
P
13.3
Water Transport to Deep Mantle
13.3.1 Mantle transition zone and
stagnant slab
Presence of HBL just above the subducting slab
has been supported by seismological studies, in
particular those utilizing the receiver function
T
condition of second
critical endpoint (Mibe
et al
., 2008). When these
fluids, ranging mostly from 0.1 to 1 wt %, are
−
