USArray - a tool for probing the continent

Göran Ekström, Gene Humphreys and Alan Levander

Submitted to the IRIS newsletter (10/25/98 version).

USArray is the working name for an envisioned facility for the seismological probing of the North American continent. The facility is currently conceived to consist of two main parts: (1) a densified network of permanent broad band stations providing uniform coverage across the contiguous US, and (2) a collection of more than one hundred seismometers configured in a transportable telemetered array. This tool is needed for a new style of systematic mapping of the continental lithosphere and upper mantle, with the goal of revealing structures which tell us about the evolution of the continent from the Precambrian to the present.

The scientific and technical design of USArray evolves from the successes of the IRIS PASSCAL program and a growing understanding of the value of combining local and regional, and short- and long-term observations in seismological imaging. A project of systematic seismological mapping of the US is likely to stimulate a broad Earth science investigation of the continent, and encourage the collection and (re)examination of geological, geochemical and geophysical observations and data sets. The new facility will also allow a significant improvement in the uniform detection and quantification of earthquakes in the US, providing additional constraints on active stresses and tectonics. The structure of the project would be designed to support broader objectives expressed by the seismological and geological communities, such as multidisciplinary cooperation and educational outreach.


The central goal of the project is to understand the structure and evolution of the North American continent. North America exhibits nearly every type of geologic setting. As shown in Figure 1, it includes one of the Earth's great orogenic plateaus and one of the great continental cratons, active plate margins bounded by major strike-slip, subduction, and rift fault systems, an active hot spot, modern passive margins, and the remnants of a Paleozoic mountain belt of Himalayan proportions. Its seismic structure has been explored on many scales using data from North American stations of the Global Seismographic Network and affiliated stations, a number of regional arrays, a growing number of PASSCAL experiments, and numerous reflection and refraction profiles. However, many important questions remain, in particular concerning the relationships between the smaller geological-province and crustal scale structures and the larger continental and lithospheric scale structures. Viewed slightly differently, a principal objective of the USArray project is to tie together the seemingly disparate tectonic provinces into a coherent model of the origin and evolution of continental lithosphere.


The experiment envisioned for the mobile portion of USArray is a 10-year-long roving deployment across the contiguous US with potential land extensions into Canada and Mexico, as well as seafloor extension onto the continental shelf. The target area of ten million square kilometers could be uniformly covered by, for example, 20 deployments of 100 seismometers, each of six months' duration and with the geometry depicted in Figure 1. The result of the experiment would be a uniform, internally consistent data set with well understood spatial sampling and aliasing properties, which would be used to image the entire continent in the same detail and resolution. Many natural opportunities would exist to encourage and coordinate add-on PASSCAL experiments with the deployment of USArray.

Results from recent regional PASSCAL experiments provide examples of the type of seismological mapping of the continent that could be achieved with the new array. Figure 2 shows results from P and S wave tomography of the western US using teleseismic arrival time data. A clear separation is seen between the fast cratonic core and the heterogeneous but largely slow orogenic belt. In the orogenic belt, low velocities correlate with areas of young volcanism, and near the plate margin, high velocities correspond with tectonic domains, indicating a surprisingly complex upper mantle structural geology. Isostatic calculations show that in the high western US interior thermal effects alone cannot explain the high velocities and low densities of the mantle, and that some compositional modification of the mantle, such as basalt depletion, may have occurred as well. Figure 3 shows results from receiver function studies across the Snake River Plain, revealing both significant velocity heterogeneity and topography on the 410 and 660 km discontinuities. These variations indicate lateral variations in temperature or composition that occur in the upper mantle beneath an active portion of North America.

In addition to a mapping of the isotropic velocity structure of the deep portion of the continent, constraining its temperature and composition, mapping of the anisotropic properties would provide us with a history of the deformation of the continental mantle. Figure 4 shows a compilation of S-wave splitting results across North America. Anisotropy beneath the tectonically inactive portion of North America, and beneath the Yellowstone swell, is aligned in a direction that is generally consistent with shearing of the continental lithosphere (or, for Yellowstone, the asthenosphere) in the direction of absolute plate motion. Results from much of the elevated western US, however, are complex and must represent small-scale, poorly understood deformation processes beneath this currently active area. The transportable array would allow us to map anisotropy uniformly across the continent.


The densification of the permanent network of broadband stations would involve cooperation among groups which operate high-quality observatories in the US, such as the USGS, the regional networks, DOD, and IRIS. The USGS National Seismic Network (NSN) already constitutes such a network, cooperating with all the parties mentioned above, and the permanent component of USArray would build on the existing NSN. The emphasis would be on high-quality data and efficient data distribution.

A denser network of high-quality stations will be valuable as a set of fixed reference points for the portable USArray deployments: current efforts to combine tomographic images from a variety of portable experiments have been hampered by the lack of a common baseline between studies. In addition, some tomography experiments benefit from the accumulation of many observations at a single site, since this allows for a better separation of local and distant wave propagation effects. These considerations suggest that an initial goal for the permanent component of USArray should be uniform coverage. To achieve a density of one high-quality station every 350 km within the conterminous US, approximately 30 stations would have to be added to the existing inventory of sites (see Figure 1); several existing stations would need to be upgraded.

An expanded network of stations, contributing data in near real time to the USGS NSN, would improve the detection, location, and source characterization capabilities of the NEIS for earthquakes and other seismic events in the US and surrounding areas. With this expanded network, moment tensor estimation of earthquake parameters from regional waveforms could include earthquakes to smaller magnitudes (approximately magnitude 3.5) anywhere in the continental US. Moment tensors of smaller earthquakes would provide information on current stresses and modes of deformation within the continent. For example, Figure 5 shows the normal faulting mechanism of the 1997 M=4.9 Alabama earthquake, which was large enough to be studied using data from IRIS GSN stations at far-regional distances. A dense national network would allow this type of characterization for much smaller earthquakes.

The permanent component of USArray could also be used to probe deep Earth structure on a global scale. Figure 6, for example, shows a record section across the MOMA array; the emergence of the diffracted phase SPdKS is indicative of an ultra-low velocity region above the core-mantle boundary in the central Pacific.


By itself, USArray is a facility and an experiment in seismology and geophysics. Its success would require the active participation of the broader IRIS community. It would provide a natural venue for pursuing IRIS education and outreach goals in seismology. In concept, USArray is also envisioned to be a key element of a coordinated program of broad, interdisciplinary Earth science study of the North American continent. Each of the USArray temporary deployments, moving across the rich variety of geologic provinces of North America, could be the observational core of an integrated field laboratory involving the full spectrum of geoscience investigations, and could also provide a prominent center for an exciting public education program.


This article summarizes presentations by the authors at the 1997 IRIS instrumentation workshop in Santa Fe and the 1998 IRIS annual workshop in Santa Cruz. Our attempt at formulating the USArray concept is based on earlier ideas and proposals of various groups and individuals. In particular, we acknowledge the contributions and assistance of Ray Buland, Ken Dueker, Karen Fischer, Bob Hutt, Art Lerner-Lam, Anne Meltzer, Tom Owens, Jeff Park, Peter Shearer, Anne Sheehan, Rob van der Hilst, Bob Woodward, Michael Wysession, and George Zandt.

Figure 1

Idealized tectonic map of North America. Many structure boundaries are gradational and poorly understood. The permanent station locations are the existing sites of the Canadian, Mexican and US national broadband networks, the California broadband networks and the IRIS/USGS GSN.
(Back to text.)

PostScript version of this figure.

Figure 2

Composite image of upper mantle seismic structure at 100 km depth beneath North America. Blue is high velocity mantle and red is low velocity mantle. The continental scale image is from the multi-bounce S-wave modeling of Grand (1997). Overprinting Grand's image are five regional array inversions by different authors. The scale and baseline of the different inversions have been adjusted and the total range in P-wave velocity is about 8%. Using standard scaling relations, red regions are partially molten and blue regions are subsolidus. (Figure courtesy of Ken Dueker.)
(Back to text.)

PostScript version of this figure.

Figure 3

Common-sample-point stacks of receiver functions across the Snake River Plain (Dueker and Sheehan, 1997). Discontinuities are seen at depths of about 250, 410, and 670 km. The feature at 150 km depth is the first reverberation from the Moho. Shown in the background is the S-wave velocity structure, with light areas being about 8% slower than dark areas. (Figure courtesy of Ken Dueker and Anne Sheehan.)
(Back to text.)

PostScript version of this figure.

Figure 4

S-wave splits from the U.S. and southern Canada compiled from several different studies and publications. Green lines indicate the fast direction and split time of split SKS arrivals. The greatest reported split time (in southern Canada) is about 2.5 s. Red lines indicate the backazimuths of null arrivals in areas where null arrivals are common.
(Back to text.)

PostScript version of this figure.

Figure 5

The October 24, 1997 Alabama earthquake. The map on the left shows the location and focal mechanism of the earthquakes, with triangles indicating the locations of stations used in the CMT analysis. The record section on the right shows comparisons of observed (top) and model (bottom) seismograms for the four closest stations. The seismograms are dominated by fundamental mode Love and Rayleigh waves.
(Back to text.)

PostScript version of this figure.

Figure 6

Left: Paths to the stations of the MOMA array from earthquakes in the western Pacific. The regions sampled by possible SPdKS phases on the source side of the path are shown by the four ellipses. Background shows the lowermost layer of the Grand et al. (1997) S-wave model. Paths from Tonga and New Britain require ultra-low velocity zones at the CMB, but paths from the Marianas do not. Right: Record section of SKS/SPdKS phases for the New Britain earthquake. SPdKS is clearly observed moving out from SKS, but its peak arrives up to 5 seconds late with respect to the times predicted from PREM (dashed line). These observations are consistent with the existence of a thin, ultra-low velocity zone, possibly partially molten mantle. (Figure courtesy of Karen Fischer.)
(Back to text.)

PostScript version of this figure.

Göran Ekström, Department of Earth and Planetary Science, Harvard University, copyright ©1998, all rights reserved