Geoscience Reference
In-Depth Information
show a broadened peak around 520 km depth in
the same region.
For a long time, it was thought that the 520 km
discontinuity was characterized by a wide tran-
sition interval, making it impossible for short
period data to see this discontinuity. Indeed, it
is not seen in short period PP and P
P
precur-
sors (Rost & Weber, 2002; Xu
et al
., 2003), and it
is very strong in long period PP precursors, per-
haps even stronger than in SS precursors (Deuss,
2009). Reflections from the 520 km discontinu-
ity are occasionally seen in short period receiver
function data (van der Meijde
et al
., 2005), and
Figure 10.5c shows that even splitting can some-
times be observed in Pds receiver functions. The
520 km discontinuity has also been seen in short
period P wave triplications by Ryberg
et al
. (1997).
In general, though, the 520 km discontinuity is
most easily found in long period data.
The phase transition from
β
(wadsleyite) to
γ
olivine (ringwoodite) appears over a depth inter-
val of 20 km (Frost, 2003), which would only be
visible to long period data such as SS and PP pre-
cursors, which indeed show the largest number of
observations of this discontinuity. However, the
observation of splitting cannot be explained by a
single phase transition in olivine from wadsleyite
to ringwoodite only, but requires the presence
of multiple phase transitions on a global scale
in the transition zone (Weidner & Wang, 2000;
Saikia
et al
., 2008) (Figure 10.6). A phase tran-
sition from garnet to Ca-perovskite appears in
the mid transition zone, at pressures very sim-
ilar to the wadsleyite to ringwoodite transition
but with a different Clapeyron slope. The com-
bined phase transitions of olivine and garnet in a
pyrolite mantle explain the seismic observations
of double peaks (or splitting) at the 520 km dis-
continuity. At lower temperatures, these phase
transitions appear at different depths, resulting
in two discontinuities if enough Ca is present.
At higher temperatures, there is a single reflec-
tion. To explain splitting in higher temperature
regions, compositional heterogeneity is required
in the middle of the transition zone (Saikia
et al
.,
2008); low Ca content will lead to single reflec-
tions in low temperature regions.
10.3.3 660 km discontinuity
The 660 km discontinuity is important as it
marks the boundary between the upper and
lower mantle. Its detailed structure may provide
information about the style of mantle convection
and the fate of subducted slabs, and it is not a
surprise that the most complex seismic structure
is found near the 660-km discontinuity at the
bottom of the transition zone. It has been
observed routinely since its discovery using P
wave array studies in the 1960s, using both
triplications and P
P
precursors (e.g. Niazi &
Anderson, 1965; Johnson, 1967; Engdahl & Flinn,
1969). It has also been seen using Pds receiver
functions (e.g. Paulssen, 1988; Chevrot
et al
.,
1999) and SS precursors (e.g. Shearer, 1991).
While the 660 km discontinuity is clearly vis-
ible in SS precursors (Figure 10.4a), it was found
to be absent in global stacks of PP precursors
(Figure 10.4b). These two data types sample very
similar regions of the Earth, thus it is surpris-
ing that the 660 is not seen in PP precusors on
a global scale. It was suggested that somehow
the jump in P wave velocity at this discon-
tinuity must be smaller than the jump in S
wave velocity (Estabrook & Kind, 1996). How-
ever, more recently it was discovered that PP
precursors do show the 660-km discontinuity,
but only in certain regions (Deuss
et al
., 2006;
Thomas & Billen, 2009). For example, it is not
seen in Alaska (Figure 10.5a), but a strong P660P
signal is seen in Asia (Figure 10.5c) and in the
Indian Ocean (Figure 10.5d). The structure of the
'660' also appears to be rather complicated, with
two peaks at 660 and 720 km depth being visible
in the PP precursors in some places (Figure 10.5b).
The story becomes even more confusing
when adding other data types, in particular
short period P
P
precursors which consistently
observe a 660 km discontinuity (Adams, 1971;
Benz & Vidale, 1993; Xu
et al
., 1998, 2003).
Unfortunately, it is difficult to find a location
with both PP and P
P
precursor data coverage so
the difference might just be a case of not sampling
in the same location. Even so, if a discontinuity
is seen at short period, then we would interpret
it as being sharp. We would then also expect to
