|
|
Missouri to Massachusetts Array of Broadband Seismometers
Slicing Into the Earth: Seismic Mapping with the Missouri-to-Massachusetts Broadband Deployment
Michael E. Wysession (1), Karen M. Fischer (2), Timothy J. Clarke (3),
Ghassan I. Al-eqabi (1), Matthew J. Fouch (2), Patrick J. Shore (1),
Raul W. Valenzuela (1), Aibing Li (2), and Julia M. Zaslow (2)
- Department of Earth and Planetary Sciences,
Washington University, St. Louis, Missouri
- Department of Geological Sciences,
Brown University, Providence, Rhode Island
- New Mexico Institute of Mining and Technology,
Socorro, New Mexico
Eos, 77, 477, 480-482, 1996.
Introduction
Regional arrays of seismometers provide a powerful means of mapping the
details of deep-Earth structure. Our understanding of the geological
processes at work within our planet depends upon our ability to examine them;
seismology remains the best tool available. However, spatial aliasing due to
the less-than-optimal distribution of global seismometers has long made it
difficult to determine deep-Earth structure from teleseismic waves. The
temporary deployment of portable broadband seismometers can help by providing
high-resolution windows into our planet. Patterns of global mantle convection
create seismically observable features such as anisotropy at the top and
bottom of the mantle, topography of upper mantle discontinuities, and
heterogeneous structure at the core-mantle boundary. We hope to provide
additional constraints on the style of mantle convection by quantifying these
observations.
April, 1996, marked the end of a 1-year deployment of 18 seismometers as part
of the Missouri-to-Massachusetts (MOMA) IRIS PASSCAL Broadband Deployment.
The seismometers were evenly spaced between permanent stations CCM (Cathedral
Caves, Missouri) and HRV (Harvard, Massachusetts), providing a 20-station
array spanning 1740 km with a 91.5 km average interstation distance
(Figure 1).
The geometry of these stations is providing important information not
only about the crust and lithosphere of the Eastern United States, but also of
a 2-D slice across the entire mantle and into the core.
Deployment
The deployment of the MOMA stations was logistically challenging. Sixteen of
the 18 temporary stations were borrowed from the IRIS PASSCAL pool, and
because of shortages in instrument availability, a variety of types of
broadband sensors, clocks and diSKS were used. The deployment began in the
heavy snows of mid-January (1995) with an incomplete set of sensors, and was
completed by late March using sensors borrowed from the Joint Seismic Program.
Eighteen evenly-spaced locations between stations CCM and HRV were initially
targeted, and strong attempts were made to site the stations within 5 km of
these initial targets. Vaults were usually placed on bedrock, though this was
not possible for many of the western stations where layers of glacial loess
are many tens of meters thick. Cloudy winter skies sometimes prevented the
solar panels from recharging the deep-cycle batteries. Stations had to be
visited often. The spring and early summer of 1995 were very rainy, so large
drains had to be installed at several sites to prevent recurrent flooding.
Despite problems with instrument failure, negative power budgets and flooding,
the average MOMA station uptime was very high, and overall data quality is
excellent.
Two data streams were recorded: 20 sps and 1 sps. While the 1 sps stream is
redundant, since all information is contained in the 20 sps stream, having the
1 sps stream allows for convenient examination of the data and facilitates
analyses that only require long-period data (such as with surface and
core-diffracted waves). The data are in the process of being converted into
SEED volumes through the use of newly-available database software written by
PASSCAL engineers, and will be available from the IRIS Data Management Center
in April, 1997.
Core-Mantle Boundary
The lowermost mantle has been the site of several exciting seismological
discoveries over the past decade. The rock in the D'' region, the few hundred
kilometers of mantle overlying the core, contains a greater degree of lateral
variation than any part of the Earth other than its surface. There is a
pattern to the long-wavelength variation in D'' velocities that greatly
resembles the locations of Mesozoic paleosubduction, suggesting some degree of
mass transport between the upper and lower mantles. Many regions display a
discontinuous velocity increase atop D''. The depth of this increase varies
by location, and there is some suggestion that it may be different for P and S
waves. Because D'' may be both a conductive thermal boundary layer between
the core and mantle as well as a chemical boundary layer containing
differentiated mantle phases, more than one discontinuity is possible. The
core-mantle boundary (CMB) causes significant seismic scattering. This could
be a result of small-scale heterogeneity within D'' or in CMB topography;
there is even evidence of an extremely slow velocity layer at the bottom of
D''.
The MOMA array was designed to examine several of the current questions about
the CMB. Its stations recorded roughly 53% of the world's large earthquakes in
the a distance range of 100° - 140°, nearly double that of an average global
station location. Most of these earthquakes came from the southwest Pacific.
This range is ideal for core phases that diffract along and refract across the
CMB. Waves from earthquakes in the Tonga/Kermadec region arrive nearly
along-strike to the MOMA array.
Figure 2a
shows core-refracted Sdiff waves
from transverse MOMA seismograms for a Kermadec earthquake on July 3, 1995.
The ray parameter (slowness) of the Sdiff arrivals is the slope of the arrival
times with respect to distance, and is a direct result of the velocity
structure at the very base of the mantle. The change in the shape of the
diffracted waves can also be used to model the vertical structure of D''
through an analysis of the rate of decay of amplitudes during diffraction as a
function of frequency [Valenzuela et al., 1996]. The velocity structure N710
shown in
Figure 2b
obtained by forward-modeling the SHdiff amplitude decay,
shown in
Figure 2c.
This preliminary model a discontinuous increase at a
depth of 2710 km, 180 km above the CMB, underlain by a parabolic decrease. It
is similar to the discontinuous increase found by Garnero et al. [1993],
though with very different data and techniques. A preliminary analysis of the
variation of this slowness of the core-diffracted Pdiff waves for the Kermadec
earthquake as a function of frequency shows the presence of a minimum value at
a mid-frequency range is indicative of fast velocities at the top of D'',
underlain and overlain by slower velocities, supporting the type of model
shown in
Figure 2b.
We are also examining the polarization of core-diffracted shear waves as an
indication of anisotropy at the base of the mantle, in the manner of Maupin
[1994]. Several other CMB studies are making use of the SKS phase, that
travels through the mantle as an S wave but through the core as a P wave: (1)
differential times of core-diffracted and core-refracted waves provide
constraints on velocities above and below the CMB [Garnero and Helmberger,
1995; Wysession, 1996]; (2) an analysis of SKS waves that partially diffract
around the core has revealed an ultra-slow velocity D'' layer north of its
identification by Garnero and Helmberger [1996]; and (3) relative amplitudes
of multiple SKS phases examine the S velocities at the very base of the mantle
[Silver and Bina, 1992]. Work is also underway to invert the entire body wave
train across the array for a long-wavelength image of a 2D slice through the
mantle.
Subducting Slab Velocity Structure
The seismic velocity structure of subducting lithospheric slabs is fundamental
to our understanding of mantle dynamics. The overall geometry and extent of
subducting lithosphere is a powerful constraint on patterns of convective flow
in the mantle. For instance, recent travel-time-based images of subducting
slab velocity anomalies indicate a wide range of slab morphology within the
mantle transition zone, suggesting that interactions between subduction flow
and mantle structure are highly variable. Theoretical studies demonstrate
that broadband waveforms should provide a strong complement to travel-time
data in the resolution of slab structure, but the sparse distribution of
permanent broadband stations has in general hindered use of waveforms to
constrain slabs in the transition zone and lower mantle. Broadband waveforms
recorded by the MOMA array will improve resolution of the geometry and extent
of subducting slabs beneath the Caribbean, South America, and the northwest
Pacific Ocean.
For example, two intermediate-focus earthquakes in Colombia generated
body-wave phases that have excellent signal-to-noise ratios at stations across
the array. These phases largely miss the lithosphere currently subducting
beneath South America, but travel through a slab-like high-velocity anomaly
that has been imaged in the lower mantle beneath the Caribbean and interpreted
to be the subducted Farallon plate [Grand, 1994; and others]. The data
contain strong core-reflected shear wave travel-time anomalies that are
consistent with the existence of the Caribbean anomaly to great depth in the
lower mantle, and they also manifest large variations in amplitude and
waveform complexity. We plan to model body waveforms from these and other
South and Central American events with synthetic seismograms to constrain both
radial and lateral variations in the upper mantle and D''. This analysis
should considerably enhance resolution of the width and dip of the lower
mantle Caribbean slab anomaly.
Upper Mantle Structure and Crust
The geometry of MOMA is ideally suited to provide a transect of the crust and
upper mantle across the Eastern United States. As seen in
Figure 1, the
stations extend from the stable midcontinent craton across the Appalachians to
the Eastern terranes. This represents a large variation in crustal and
lithospheric structure. MOMA can detail the transition of the lithosphere
among these very different tectonic regions, including a variation in the
depth to the MOHO. In addition, we can resolve variations in the depths of
upper mantle discontinuities, that are the result of thermal variations
associated with mantle convection.
The CORE method [Clarke, 1993] is one technique we are using to address these
questions. This is a tool for the analysis of body-wave seismograms that not
only allows interpretation of arrival times, but also provides a framework for
waveform inversion. Whole body-wave trains from multiple seismograms can be
inverted to reveal a three-dimensional image of mantle velocities, including
discontinuity locations. In addition, receiver functions are being generated
for the MOMA stations, providing information about the thickness and velocity
of the crust and additional constraints on the depths of mantle
discontinuities.
Surface waves travelling across the MOMA stations can be inverted for upper
mantle structure because of variations in group and phase velocity dispersion.
Large earthquakes both in the Southwest Pacific and in Western North America
(Northern Mexico, Southern California, Texas) have occurred nearly
along-strike, providing an opportunity to make a 2-D image along a slice
through the crust and mantle. An example is shown in Figure 3 for the West
Texas Earthquake of April 14, 1995 [Al-eqabi et al., 1996].
Figure 3a shows
the propagation of the body and surface wavetrains across MOMA as seen on the
vertical component.
Figures 3b,c
show an example of a waveform inversion for
the vertical component record at station MM04, in Le Raysville, PA, using an
algorithm of Nolet [1990]. The solid line represents the data and the dashed
line depicts the model, both before (3b) and after (3c) inversion. The best
fitting model for the West Texas - MM04 path is shown as a solid line in
Figure 3d,
with the initial model represented as a dashed line [Iyer and
Hitchcock, 1989]. The final model has a deeper MOHO depth (38 km) and faster
sub-crustal velocities. Work is underway to produce a 2-D model of the crust
and mantle beneath MOMA using the West Texas and other earthquakes in the
manner of van der Lee [1995].
Upper Mantle Anisotropy
We are using an analysis of shear wave splitting to constrain the extent and
orientation of seismic anisotropy beneath the eastern United States [Fouch et
al., 1996], which can be an indication of lithospheric genesis and convective
flow directions. The MOMA array provides detailed sampling of variations in
anisotropic mantle structure across several different lithospheric provinces:
the interior North American craton, the Grenville Province, the Appalachian
Plateau and Range, and the complex orogenic zones of New England. A number of
large, deep earthquakes at a variety of back azimuths yield very clear
teleseismic core phases such as SKS that provide well-constrained splitting
parameters (fast polarization direction, ¢, and splitting time, *t) at 15 of
the 18 MOMA stations
(Figure 1).
Stations at which we have not yet documented
splitting are marked by crosses, indicating an absence of anisotropy beneath
the station, very weak anisotropy that cannot be detected, or anisotropy with
a fast direction that is parallel or perpendicular to the back azimuths of
events analyzed.
The observed splitting parameters represent anisotropy on nearly vertical
paths from the core-mantle boundary to the station. Our data are the first
measurements of shear wave splitting in Ohio, Indiana, and Illinois, and
greatly increase lateral resolution of shear wave splitting in northern
Pennsylvania and southern New York. Images of upper mantle shear wave
velocity structure contain high velocities to depths of more than 300 km
beneath the cratonic interior of the North American continent [Grand, 1994;
van der Lee, 1995]. The eastern margin of this fast continental root
intersects the MOMA array in the vicinity of MM07 - MM05 and coincides with a
shift in the pattern of observed fast directions. From MM07 to the west , the
fast directions are within 20° of the direction of absolute plate motion
(∼WSW) (Figure 1).
In contrast, the fast directions are more variable at
stations MM06 - MM01, located within the Paleozoic Appalachian orogenic belt.
From MM06 east to MM03, ¢ rotates from W to ∼NW, and from MM02 east to HRV,¢
repeats this pattern, rotating from WSW to ∼NW. These splitting measurements
in general are consistent with the findings of previous studies over length
scales of ∼100 km [Levin et al., 1995; Barroul et al., 1996; and others].
However, the more densely spaced MOMA stations in northern Pennsylvania and
southern New York document a much stronger variation in fast direction
orientation than is apparent in previous studies.
The magnitude of the splitting times (0.5 - 1.4 s) indicates that a
substantial portion of the observed splitting originates from anisotropy at
subcrustal depths. Assuming that the anisotropy is controlled by
strain-induced preferred orientation of upper mantle minerals like olivine,
the variations in fast direction observed across the MOMA array are strong
evidence that shearing in the asthenosphere parallel to the absolute motion of
the North American plate is not the sole source of the anisotropy. We are
investigating a variety of models to explain the splitting observed. In one,
the translation of a mechanically strong North American continental root
drives asthenospheric flow around the continental root and produces an
asthenospheric fabric and fast direction orientation parallel to the root
margin. In this model, some of the complexity in the observed fast directions
from MM05-MM01 could be related to asthenospheric flow around irregularities
in root morphology. Another model we plan to evaluate is one where the
anisotropy is largely located in the lithosphere and reflects ancient
deformation beneath the western segment of the array and more recent and
complicated strain events (the Appalachian orogeny, and rift events in the
Jurassic and Triassic related to the opening of the Atlantic Ocean) beneath
the eastern segment of the array. A third possible explanation is that the
variable fast directions, particularly over the eastern segment of the array,
reflect the superposition of significant asthenospheric and lithospheric
anisotropy.
Acknowledgments
We would like to thank the many people at IRIS who made this experiment
possible, and in particular, Paul Friberg, Carl Ebeling, Sid Hellman, Gennady
Pratusevich, Tom Jackson, John Weber, Bob, Busby, Tim Ahern, and Rick Benson.
We thank Tom Owens for helpful advice about station deployments and Francis Wu
for his help in locating the MM04 site. We thank the many land owners, private
and public, who allowed us to dig up their back yards.
References
Al-eqabi, G. I., M. E. Wysession, P. J. Shore, K. M. Fischer, and T. J.
Clarke, Upper mantle structure under the northeastern United States from the
inversion of regional and teleseismic Rayleigh waves recorded by the MOMA
array, Seismol. Res. Lett., 67, 29, 1996.
Barroul, G., P.G. Silver, and A. Vauchez, Seismic anisotropy in the eastern
U.S.: Deep structure of a complex continental plate, submitted to J. Geophys.
Res., 1996.
Clarke, T. J., The complete ordered ray expansion 2: Multi-phase body wave
tomography, Geophys. J. Int., 115, 435-444, 1993.
Fouch, M. J., K. M. Fischer, M. E. Wysession, and T. J. Clarke, Shear wave
splitting beneath the Eastern United States, Abstracts from the 8th Ann. IRIS Workshop, Blaine, WA, 56, 1996.
Garnero, E. J., and D. V. Helmberger, Seismic detection of a thin laterally
varying boundary layer at the base of the mantle beneath the central Pacific,
Geophys. Res. Lett., 23, 977-980, 1996.
Garnero, E. J., D. V. Helmberger, and S. Grand, Preliminary evidence for a
lower mantle shear wave velocity discontinuity beneath the central Pacific,
Phys. Earth Planet. Int., 79, 335-347, 1993.
Grand, S. P., Mantle shear structure beneath the Americas and surrounding
oceans, J. Geophys. Res., 99, 11,591-11,621, 1994.
Iyer, H. M. and T. Hitchcock, Upper-mantle velocity structure in the
continental U.S. and Canada, in Pakiser,L. C., and W. D. Mooney, Geophysical
framework of the continental United States: Boulder, Colorado, Mem. Geol. Soc.
Am., 172, 681-710, 1989.
Levin, V., W. Menke, and A. Lerner-Lam, Seismic anisotropy in northeastern
U.S. as a source of significant teleseismic P traveltime anomalies, submitted
to Geophys. J. Int., 1996.
Maupin, V., On the possibility of anisotropy in the D'' layer as inferred from
the polarisation of diffracted S waves, Phys. Earth Planet. Int., 87, 1-32,
1994.
Nolet, G., Partitioned waveform inversion and two dimensional structure under
the Network of Autonomously Recording Seismographs. J. Geophys. Res., 95,
8499-8512, 1990.
Silver, P. G., and C. R. Bina, An anomaly in the amplitude ratio of SKKS/SKS
in the range 100-108° from portable teleseismic data, Geophys. Res. Lett., 20, 1135-1138, 1993.
Valenzuela, R. W., M. E. Wysession, G. I. Al-eqabi, P. S. Shore, T. J. Clarke,
and K. M. Fischer, Determination of lateral and radial velocity structure of
the base of the mantle from diffracted waves recorded using the
Missouri-to-Massachusetts IRIS PASSCAL Array, Eos Trans. AGU, 77(17), Spring
Meeting suppl., S178, 1996.
Wysession, M. E., Large-scale structure of the core-mantle boundary from
core-diffracted waves, Nature, 382, 244-248, 1996.
back to MOMA home
|