Characterization of Upper Mantle structure beneath the US Array with Sp receiver functions
Katie Foster, Ken Dueker, Steve Hansen, Brandon Schmandt*: University of Wyoming, *California Institute of Technology
Synopsis: New images from 42,000 Sp and 72,000 Ps Receiver Functions
image lithospheric and upper mantle structure that has not previously been
well constrained, contributing greatly to our understanding of upper mantle
processes and continental evolution. We present a taxonomy of the negative
velocity gradient (NVG) arrivals beneath the Moho across the span of the
Transportable Array. The negative velocity gradient is here classified into three
categories:
a) Western Cordillera: Lithosphere-Asthenosphere Boundary (T > 1200 C)
b) Craton: Mid-Lithospheric Discontinuity (T< 900 C)
c) Colorado Plateau: doublet structure of metasomatized material within
lithosphere and melt staging area at base of lithosphere
Also imaged is the Lithosphere-Asthenosphere Boundary beneath the
Cratonic portion of study area, which is a new finding.
Discussion: The Lithosphere-Asthenosphere boundary has long been loosely defined and
unreliably imaged. The LAB beneath cratonic North America is constrained by shear wave velocity
and anisotropy models as well as xenolith thermobarometry, all finding 180 – 250 km depth. Thus
far, however, no published receiver function studies have imaged a velocity contrast manifesting the
cratonic LAB beneath NA. A shallower negative velocity gradient arrival at ~ 90 km depth has been
found (Rychert et al., [2010]; Abt et al., [2010]) with some resulting confusion in the community,
given that this signal is clearly too shallow to be the LAB and must be internal to the lithosphere. We
find this Mid-Lithospheric Discontinuity to be at large beneath the North American craton and believe
it to be a layer of metasomatized, volatile-rich material that has accumulated over Ga time span and
pooled at the solidus front. Our conclusion is supported by N. Sleep [2009] studies demonstrating
the high mobility of CO2 and a freezing front of carbonatite at 90 km depth. Another hypothesis is
that the MLD is a fossil LAB from Archean ages.
LA
A
GP
Co. Plateau
C
B
D
O
A
A
B
C
D
KE
FJ
LG
M
NH
B
C
E
CCP slices of Sp receiver functions across full volume to 300 km depth.
Solid black line is depth to first sharp negative velocity gradient as found by
automated picking algorithm. Blue and red contour lines represent positive
and negative 2%, 3%, and 4% P wave velocity perturbation (Schmandt and
Humphreys, [2010]). Axis labels represent latitude or longitude as
appropriate.
LAB
LAB
LAB
B
Cratonic MLD
Cascadia
E
RM B’
Co. Plateau
JDF
MLD
LAB
CP
Cratonic MLD
F
P
G
Q
H
R
A
C
I
D
A’
MLD
R S P T UQV W X Y Z
C’
Co. Plateau
MLD
LAB
Above: Topographic map with cross-section
locations- circles demark station locations, black
lines are geologic province boundaries as found
by Whitmeyer and Karlstrom (2007)
Left: CCP slices with corresponding locations
mapped in (F). All slices share the same color
pallet.
LAB: Lithosphere- Asthenosphere Boundary
MLD: Mid- Lithospheric Discontinuity
JDF: Juan de Fuca plate
RM : Rift Margin
D’
MLD
Temperature field at 100 km depth
Calculation of temperature field: We currently use the
olivine anelastic ity model of Jackson and Faul [2010] that
maps isotropic shear velocity as a function of wave period
directly to temperature. The leading free parameter is grain
size. A grain size of 1 mm is currently used consistent with
the olivine grain size from mantle xenoliths in New Mexico
(Satsukawa et al., 2011). Our velocity model is provided by of
Shen et al. [2012].
LAB
LAB
E
Great Plains
N Great Plains
MLD
LAB
Date and Methods: The Sp RF data volume is composed of all Transportable Array stations from project initiation
2004 – end 2011, such that each station has approximately two years of running time. In addition, we have included
the PASSCAL experiments SNEP, FAME, CREST, SEIDCAR, HLP, and CAFE. Teleseismic events of Mb ≥ 5.7 in 30120˚ delta range for are viewed and culled on all ~1200 components simultaneously. An initial estimate of the source
function amplitude-spectrum for each event is filtered to minimum phase. The three-component receiver function
amplitude-spectra for each station-event data bin are calculated via least-squares inversion using the estimated
source spectra as constraint equations, following the multi-channel deconvolution algorithm initially laid out by Baig,
A. M., M. Bostock, and J. P. Mercier (2005). The resulting receiver functions are migrated through a threedimensional velocity model from Shen et al., [2012], and back-projected along the incident raypath and imaged by
common conversion point (CCP). Bootstrapping methods are used for error estimation. Our current volume totals to
42,000 Sp/SKSp RFs and 72,000 Ps RFs.
Image quality remains an issue, with best arbitrator being convergence towards one image as we increase the data
fold. This is a luxury only available in select regions where we have PASSCAL experiment data. As shown below,
artifacts at depth in TA only image disappear in 4x fold image, which is overall significantly lower noise.
P wave velocity perturbation at Sp NVG depth
pick; model from Schmandt and Humphreys
[2010]
LA
CP
Vsv Tomography as surface at depth of NVG
pick. Model from Shen et al. [2012].
IA
GP
LAB
MLD
LAB
MLD
LAB
MLD
LAB
Map: Red
circles are
SNEP/FAME
stations, black
circles are TA
stations
T
Shear and Surface Wave Anisotropy
Image quality and data fold
ImageImage
S
I
J
LAB
(Top) Combined TA and
SNEP/FAME PASSCAL data.
(Middle) TA-only. A 110 km bin
size and 3 s low-pass filter
corner is used.
Shear wave velocity perturbation at Sp NVG
depth pick; model from Schmandt and
Humphreys [2010]
E’
LAB
MLD
Vs (m/s)
MLD
dlnG(%)
Above: Map view of depth to strongest negative velocity gradient
arrival in 60-140 km range for Sp and Ps respectively. Black
contour line denotes 900 ˚C at depth of transition; blue contour is
1150 ˚C.
Left: Probability density function of temperature and depth for all
Sp NVG arrivals. Note two clusterings- one at 750 ˚C, other at
1300 ˚C, corresponding to the two regions in above maps.
Right: Temperature and depth for Ps arrivals.
Fast Axis Direction
LAB
Temperature (˚C)
MLD
A(Sp/Sv)
D
Comparison along Los Angles to Chicago section (Fig. 1). A)
Isotropic shear velocity. B) dln(G) azimuthal anisotropy strength.
C) Azimuthal anisotropy fast polarization axis at 250 km depth. D)
Sp image. Anisotropic images are courtesy of H. Yuan from the
Yuan et al. (2011) model.
K
U
CP
V
CP
L
W
M
X
N
O
Y
CP
Z
CP
OA