Deep Earth Exploration Seismology

1
Inverse scattering of broad-band (seismic)
data New opportunities for detecting and imaging
interfaces in Earths deep interior
Rob van der Hilst, Ping Wang, Cao Qin, Dan Shim
(all at MIT), Maarten de Hoop (Purdue)
Deep Earth Exploration Seismology
2
Discontinuity spotting
Deuss Woodhouse (2002)
NB plot on left is not complete, especially in
last 15 years
Slide after Peter Shearer
3
Discontinuity spotting
Deuss Woodhouse (2002)
4
SS precursors probe layering near the SS bounce
point
Nice for global TZ studies since SS bounce points
are widely distributed
Slide courtesy Peter Shearer
5
S410S
S520S
S660S
Slide courtesy Peter Shearer
6
Deuss et al. (Science, 2006)
7
Long wavelength topography from SS precursors
8
Example Snake River Plane
Dueker Sheehan (1997)
9
First-order 410 and 660 observations are
consistent with mineral physics predictions for
olivine phase changes

• Absolute depths agree with expected pressures
• Variations in transition zone thickness
  consistent with Clapeyron slopes

Lebedev et al. (Science, 2002)
10
First-order 410 and 660 observations are
consistent with mineral physics predictions for
olivine phase changes

• Absolute depths agree with expected pressures
• Variations in transition zone thickness
  consistent with Clapeyron slopes

410
520
660
11
Long wavelength topography from SS precursors

• Excellent global data coverage
• but
• Long period data ? limited radial resolution
• Stacking over large areas ? limited lateral
  resolution
• Scale lengths of topography of and correlation
  between
• 410, 520, and 660 still debated
• (on global scale, poor correlation!)

12
Example Snake River Plane

• Higher radial and lateral resolution
• but
• large short-wavelength topo difficult to
  explain (3-D effects?)
• regional only (mainly continental)

Dueker Sheehan (1997)
13
Furthermore, phase transitions complex, not
instantaneous
(a)
??
(ß)
(?)
Irifune et al. (1998)
Precise shape of transition depends on
temperature, composition, water content, and
other factors
14
(No Transcript)
15
410 and 660 observations may be consistent with
mineral physics predictions for olivine phase
changes But many issues must be resolved
before temperature and composition can be
constrained accurately from seismic data

• Not yet well known
• Scale lengths of topography and correlation
  between various interfaces
• Character (regularity) of the transitions (and
  how this varies from place to place!)
• Requires new methods for systematic,
  multi-resolution imaging of both continental and
  oceanic mantle

16
Example 2 core-mantle boundary and D
Figure by Ed Garnero (ASU)
17
(Garnero, 2000)
(Forward) Modeling of selected waveforms has
been very successful.But- Simple geometries-
A priori assumptions about structure- Strict
data selection- Small data sets- Limited
geographical coverage
(Helmberger et al., 2005)
18
One can now use hundreds of thousands (soon,
millions) of high-quality records!!
Deep Earth (Industry-type) Exploration
Seismology Generalized Radon Transform (GRT)
of broad band ScS wavefield
Wang et al. (2006)
80,000 broad-band ScS waves
19
Recording of reflected waves
Earthquakes (sources of seismic waves)
660 km
Earths mantle
Earths core
2,900 km depth
CMB
20
GRT reconstructs structure at image point y from
scattered wave field ? Contributions from
specular and non-specular reflections
Specular
Non specular

u1
u2
r
r
s
s
y2
y1
s source r receiver y image point u
data
GRT Concept in a nut shell
21
Construct radial reflectivity profiles
u1
u2
r
r
s
s
Image points are predefined and need not be at
Core Mantle Boundary
22
Construct radial reflectivity profiles
u1
u2
r
r
s
s
? contrasts in elasticity
23
Reflection coefficient (R) and pulse-width depend
on opening (scatter) angle improve radial
resolution ? narrow wide angle (all
pre-critical)
Forward modeling uses only a small fraction of
available data
Wang et al (JGR, 2006) Van der Hilst et al
(Science, 2007)
75
80
43
epicentral distance (deg)
24
Figure Ping Wang
Van der Hilst et al. (Science, 30 March 2007)
25
Routine 3-D exploration of lowermost mantle now
possible!! along with calculation of formal
uncertainties Ma et al. (JGR, 2006)
Routine, large-scale, 3-D exploration of D now
feasible!
Van der Hilst et al. (Science, submitted)
26
Scatter images from GRT (80,000 broad band ScS)
27
Top Scatter images from GRT (80,000 broad
band ScS) Bottom S-wavespeed from tomography
(Grand, 2002)
28
Wavespeed increase (gt 75 confidence)
29

• Sidorin, Helmberger, Gurnis (Science,1999)
• Location of phase transformation estimated from
  assumed Clapeyron slope
• (6 MPa/K) and Grand et al. (1997) S model
• Agreement with scatter interface ? is remarkable!

30
Routine 3-D exploration of lowermost mantle now
possible!! along with calculation of formal
uncertainties Ma et al. (JGR, 2006)
Discontinuity height above CMB
(After Sidorin et al., Science, 1999)
Van der Hilst et al. (Science, 2007)
31
top of D
Consistent with previous studies!
(Thomas et al., JGR, 2004)
32
Wavespeed decrease (gt 75 confidence)
33
Hernlund et al (Nature,2005)
?
Are the wavespeed in- and decrease related to the
post-perovskite (pv?ppv) and the
back-transformation (ppv?pv)?
34
Hernlund et al (Nature,2005)
Do ppv-lenses in D only exist in cartoons?
Hutko et al (Nature, 2006)
35
Van der Hilst et al (Science, 2007)
ppv?
Hernlund et al (Nature,2005)
Do ppv-lenses in D only exist in cartoons? Or
are they real?
Hutko et al (Nature, 2006)
36
Lateral temperature variations
g10 MPaK-1
900o
Uncertainty Clapeyron slope
37
Thermal gradient
g10 MPaK-1
TBL
DT
Uncertainties Clapeyron slope, actual geotherm
in (non-steady state) boundary layer,
38
1) It works!  Results locally consistent with
waveform modeling.2) Can study regions that are
out of reach of techniques relying on near or
post critical reflection ? explore uncharted
territory! 3) Interface with positive
depth-temperature correlation detected over
entire study region (150-300 km above CMB) ?
consistent with ppv transition.4) Temperature
variations of order of 600-800 K over 1000-2000
km.  (NB this is fairly large ? Clapeyron slope
perhaps larger than 10 MPa/K?) 5) Multiple
interfaces ? multiple transitions? 6) Tcmb
estimated at 4,000K. 7) With many assumptions
and much uncertainty thermal gradient and heat
flux can be estimated dT/dz 6-16 K/km q
60-160 mWm-2 (for ? 10 Wm-1K-1). Similar
approach also possible with SKKS (CMB, D) and SS
(Transition Zone)
39
To download papers related to this
research See http//quake.mit.edu/hilstgroup/ro
bspage/ ? publications