S43D-01
Imaging Mantle Convection Processes Beneath the Western USA Using the EarthScope Transportable Array
High resolution velocity models beneath western USA can provide important clues to mantle convection processes in this tectonically active region, e.g., the subduction of the Juan de Fuca plate, the upwelling of the Yellowstone plume, and their possible interactions. In this study, we apply the tomography technique using the Transportable Array data complemented by regional networks data resulting in a total of 732 stations. In our preliminary models we use 57 earthquakes sources. We derived two preliminary Vs models and one preliminary Vp model using tangential, radial, and vertical components respectively. Our preliminary tomographic images show some common features which have been imaged before such as the high velocity anomaly beneath the Cascades and the low velocity anomaly beneath the Yellowstone National Park. However, the unprecedented dense station distribution allows us to see deeper and reveals some new features: (1) the imaged Juan de Fuca subduction system goes deeper than previously been imaged. It reaches more than 500 km depth in Washington and northern California while in Oregon it seems break off and is segmented, implying a possible interaction with the proposed Yellowstone plume; (2) immediately south of the Juan de Fuca subduction system, we image low velocity anomalies down to ~{400} km depth, coincident with the proposed location of the slab gap; (3) we image the low velocity anomaly beneath the northeast Oregon down to ~{300} km depth, deeper than has previously been imaged, which has been hypothesized as the depleted mantle after the eruption of the Columbia River flood basalts, a result of delamination of the Wallowa plutonic roots [Hales, et. al., 2005]; (4) we see the high velocity Pacific plate abutting against the low velocity North American plate along the trace of the San Andreas Fault System. These observations suggest we are only just beginning to image the complex interactions between geologic objects beneath the western USA.
S43D-02 INVITED
Community Models of Mantle P-wave Heterogeneity Beneath North America From Multiscale Travel-time Tomography Using Transportable Array Data
As USArray peregrinates across the western United States, the volumes of data it produces offer unique opportunities for seismic imaging. Constraining structures on a range of length scales and their physical and chemical causes is a prerequisite for understanding the relationship between near surface and deeper mantle processes. We produce 3-D models of P-wavespeed variations in the mantle from linearized inversion of travel time residuals from USArray combined with the global EHB catalog. We perform global inversions, but an adaptable parameterization enhances resolution in regions where more data are available. This adaptability allows us to create an internally consistent model while at the same time reaching a resolution in the western United States nearly equivalent to that of regional tomography studies. The addition of USArray data results in a significant improvement of spatial resolution of upper mantle structure beneath the western United States. This will help us understand, for example, the structural transitions from the central craton to the tectonic domains further west, the deep structure of the Cascadia subduction zone, and the relationship between the Yellowstone hotspot and processes deeper in the mantle. Model updates will be made available to the community and will also be useful as starting models for higher resolution wave equation tomography studies.
S43D-03
Array Processing of Teleseismic Body Wave Phases Recorded by the Transportable Array
We have begun application of a new array processing procedure to teleseismic body waves recorded by the USArray. Unlike standard slant stack processing used in high-frequency arrays this procedure begins by assuming the source is already known. This allows an abstraction of the data as an arrival time gather in which time zero for each seismogram is the predicted arrival time of the phase of interest. We use a novel, robust stacking algorithm to compute a beam and align the data. The analyst is required to pick an initial trace to use as a starting estimate of the array beam. This trace is used to provide initial alignment of data by cross-correlation with this initial beam estimate. A median stack is then used to provide a robust starting estimate of the array beam. The final beam is computed as a weighted stack with a penalty function derived from coherence of each trace with the array beam used to improve the robustness of the algorithm. An iterative loop continues to refine the array beam estimate until lags computed by cross-correlation with the array beam do not change. This procedure is found to normally converge in 2 to 5 iterations and is remarkably effective in automatically discarding problem data. The primary outputs of the procedure are an array beam, arrival time residuals, and relative amplitudes at each station relative to the array beam. We have successfully applied this procedure to P, S, pP, PP, and Pdiff body wave phases, but other phases could be handled with the same program. We find remarkable coherence of the initial P and S phase in the 30 to 90 degree range routinely used for body wave tomography. Large events commonly show coherence of 0.8 or higher over the entire array. Smaller events often require subarray processing that requires a least squares procedure to produce a consistent set of residuals for the entire array. This different behaviour for small events is largely due to a well-known frequency dependence in spatial coherence. Small events are often only well recorded at short periods which are correlated over shorter distance scales than longer periods that dominate larger events.
S43D-04 INVITED
Progress in Ambient Noise Surface Wave Measurements in the Western US
Ambient noise has been shown to produce accurate surface wave dispersion maps on multiple scales over a broad period band, with resolution limited primarily by the location and spacing of instruments. The USArray Transportable Array (TA) component of Earthscope provides a nearly ideal network for carrying out ambient noise processing. Prior to USArray, much of the western United States (US) had poor station coverage. Current station spacings across the region are about 70 km, and over 400 stations return continuous data. We have carried out ambient noise surface wave measurements on emerging data from the USArray TA since October 2004 and present current short- to intermediate-period dispersion maps. Ambient noise processing produces higher resolution dispersion maps within the footprint of the USArray TA than has been possible using earthquake measurements. Resolution within the footprint of the USArray TA, and across much of the western US, now matches the average inter-station distance. Strong velocity anomalies in the dispersion maps are correlated with the Sierra Nevada, Great Valley, Peninsular and Cascade Ranges, the Columbia Basin and flood basalts, and the Snake River Plain. Recent work has demonstrated the existence of Love wave energy in the ambient noise wavefield and allows for the generation of both Rayleigh and Love wave dispersion maps. The dispersion maps provide the basis to invert for crustal and upper mantle shear-wave velocities across California and Nevada. We present a preliminary crustal and upper mantle velocity model for this region. The short- to intermediate-period dispersion measurements from ambient noise are complemented by longer-period two-plane wave measurements and result in better constraints on mantle velocities. The velocity model allows for examination of crustal structures in the Basin-and-Range, Sierra Nevada, and the hypothesized southern edge of the Juan de Fuca plate. As the USArray TA begins to roll eastward, emerging data promises to provide high resolution dispersion maps for the entire country and will allow for the construction of a single, high resolution 3D velocity model of the crust and upper mantle across the continental US.
S43D-05
Inferences on Upper Mantle Seismic Velocity and Anisotropy in Western North America From Surface Wave Analyses
In this study, we examine surface wave dispersion to determine seismic velocity variations in the crust and upper mantle beneath the Great Basin. This will improve our understanding of the relationship between lithospheric and asthenospheric processes in the western United States, and what caused the relatively recent extension episode in the Basin and Range. We employ a two-station method using the excellent lateral and good azimuthal coverage enabled by USArray Transportable Array (TA) broadband seismic stations. We use data from TA stations to generate Rayleigh wave dispersion curves between periods of 16s and 170s, which enable modeling of seismic shear wave velocities to ~300km depth. In order to reduce trade-offs between structure in the crust and upper mantle, we supplement our measurements with published results of surface wave phase velocity data between 8s and 40s determined from ambient seismic noise analysis. Constraints on Moho depth are incorporated using the routinely determined receiver function results (http://www.seis.sc.edu/EARS). To model upper mantle shear wave velocities from the measured dispersion curves, we apply a forward modeling technique which allows us to determine quantitative model uncertainties and parameter trade-offs. As of summer 2007, we have found 22 events with high quality Rayleigh wave dispersion curves for ~900 station pairs with inter-station distances between ~70 and 400km. We combine phase velocity dispersion measurements for several sub-regions within the Great Basin and calculate an average and standard deviation for each group. We find a clear increase per group in phase velocities from north to south for periods above ~35s. In the northern part of the Great Basin (N41°- N42.5° latitude), dispersion curves for measurements along E-W paths show a well-defined reduction in phase velocity with respect to a slightly modified Tectonic North America (mTNA) shear wave velocity model between periods of ~35s and 100s. Dispersion curves for southern regions (between N39° and N41° latitude) in the E-W direction do not differ from the predictions of mTNA, and phase velocities for periods longer than 35s are generally larger than those for the northern paths. This increase in phase velocity from N to S is much less clearly visible for paths that are not in the E-W direction, most of which exhibit phase velocities lower than predictions from model mTNA around 35s-100s. Given that the sensitivity to shear wave velocities of 50s Rayleigh waves peaks at about 70km depth, these results suggest a reduction in shear wave velocities compared to mTNA, located below a thin (<100 km) lithospheric lid across the region. Our finding of a likely isotropic phase velocity structure in the northern Great Basin is similar to what is observed in most of the High Lava Plains to the north. Our results also suggest azimuthal anisotropy in the southern Great Basin with a fast direction aligned approximately E-W, which is generally consistent with regional shear-wave splitting results, though new splitting results within the Great Basin show shear wave splitting complexity that is not clearly manifested in the surface wave data. These variations are likely the result of extension across the Great Basin combined with (or the result of) well-organized asthenospheric flow across the region.
S43D-06
Seismic Velocity Structure of the Pacific Northwestern United States From Tomographic Inversion of Teleseismic P-wave Travel Time Residuals
The goal of this study is to examine the structure and dynamics of the Cascadia subduction system as they relate to the evolution of tectonomagmatism across the backarc, including the Columbia River basalts, the High Lava Plains, and the time-progressive Newberry and Yellowstone hotspot tracks. We evaluate relative delay times from several broadband seismic networks, including the USArray Transportable Array (TA), the linear Cascadia array operated by Oregon State University in 1993-1994, and the currently operating High Lava Plains (HLP) broadband seismic array operated by Arizona State University and the Carnegie Institution of Washington. To date, we have made 8852 high-quality relative delay time measurements from 186 teleseismic events recorded at 359 stations spanning across Oregon and Washington, as well as parts of western Idaho, northern California and northwest Nevada. We applied a linear inversion using the VanDecar [1990] method to obtain a relative P- wave velocity perturbation model from the delay time residual data. The primary features of our model highlight the complex nature of this region. First, we find clear evidence for large lateral variations in the subducting Juan de Fuca slab as illuminated by higher than average seismic wavespeeds. The slab extends to a depth of perhaps greater than 400km beneath Washington, and appears to extend laterally to the east. Conversely, the slab is noticeably absent below a depth of approximately 250km beneath much of central and southern Oregon. Adjacent to this "slab window" is a region of lower than average seismic velocities located east of the slab in the back-arc that does not correlate directly with either the Columbia River basalt trend or the High Lava Plains. We also find preliminary evidence for the western extent of the Precambrian continent, generally coincident with the 87Sr/86Sr=0.706 line, a demarcation of the geochemical edge of the cratonic North American lithosphere. The strongly reduced velocities are likely the result of high asthenospheric temperatures in the mantle wedge, and perhaps provide evidence for anomalously elevated temperatures produced by a localized mantle upwelling in the region. In addition, the absence of a strong slab signature in this area suggests that elevated temperatures have significantly modified the slab in this region and is perhaps the cause for the dearth of subduction-related seismicity in the area.
S43D-07
Imaging Subduction, Episodic Tremor and Slip in the Pacific Northwest: Cascadia Arrays For Earthscope (CAFE)
Subduction delivers fluids into the Earth's mantle by transport of hydrated crust downward in subducting plates. These fluids are released at depth and may be responsible for a wide variety of phenomena including weakened thrust faults, episodic tremor and slip (ETS), intraslab earthquakes, forearc serpentinization, and arc magmatism. Cascadia is the volcanic arc associated with the youngest subducting plate, and hence a primary EarthScope target. In 2006 we launched Cascadia Arrays For Earthscope (CAFE), an EarthScope effort utilizing Flexible Array, Transportable Array, and PBO facilities, and integrating these data with complementary constraints from geodynamics and geochemistry. Seismic imaging, the emphasis of this presentation, is employed to illuminate (i) the descending oceanic plate, from where fluids are expelled by metamorphism, and (ii) the mantle wedge, where fluids migrate to produce hydrous phases such as serpentine or, beneath the volcanic arc, primary magmas, and (iii) the interface between them where ETS may be produced. The experiment traverses a section of the Cascadia system where earthquakes extend to nearly 100 km depth, thus permitting an investigation of the relationship between the release of fluids and the generation of Wadati-Benioff-zone earthquakes, and crosses regions of ETS excitation. The basic experiment has four components: (1) a 47-element broadband imaging array of Flexible Array instruments integrated with Bigfoot; (2) three small-aperture seismic arrays with 15 additional short-period instruments near known sources of ETS; (3) analysis of the PBO and PANGA GPS data sets to define the details of episodic slip events; and (4) integrative modeling. Sixty-two seismographs were deployed in July 2006; here we present a first look at the experiment and the data collected. Initial data recovery has been excellent, with approximately 12 months of continuous data recovered as of this writing, most delivered to the IRIS DMC. This time window includes an ETS episode in Jan. 2007. Given the success of this deployment, we expect to make good progress toward understanding the relationship between subduction, ETS, and fluid cycling.