SHIPS - Upper Mantle Structure and Subducting Slab Geometry
Blakely, R.J., T.M. Brocher, and R.E. Wells, 2005, Subduction zone magnetic anomalies and implications for hydrated forearc mantle, Geology, 33, 445-448. PDF
Continental mantle in subduction zones is hydrated by release of water from the underlying oceanic plate. Magnetite is a significant byproduct of mantle hydration, and forearc mantle, cooled by subduction, should contribute to long-wavelength magnetic anomalies above subduction zones. We test this hypothesis with a quantitative model of the Cascadia convergent margin, based on gravity and aeromagnetic anomalies and constrained by seismic velocities, and find that hydrated mantle explains an important disparity in potential-field anomalies of Cascadia. A comparison with aeromagnetic data, thermal models, and earthquakes of Cascadia, Japan, and southern Alaska suggests that magnetic mantle may be common in forearc settings and thus magnetic anomalies may be useful in mapping hydrated mantle in convergent margins worldwide.
Brocher, T.M., T. Parsons, A.M. Tréhu, C.M. Snelson, and M.A. Fisher, 2003, Seismic evidence for widespread serpentinized forearc upper mantle along the Cascadia margin, Geology, 31, no. 3, 267-270. PDF
Petrologic models suggest that dehydration and metamorphism of subducting slabs release water that serpentinizes the overlying forearc mantle. To test these models, we use the results of controlled-source seismic surveys and earthquake tomography to map the upper mantle along the Cascadia margin forearc. We find anomalously low upper-mantle velocities and/or weak wide-angle reflections from the top of the upper mantle in a narrow region along the margin, compatible with recent teleseismic studies and indicative of a serpentinized upper mantle. The existence of a hydrated forearc upper-mantle wedge in Cascadia has important geological and geophysical implications. For instance, shearing within the upper-mantle, inferred from seismic reflectivity and consistent with its serpentinite rheology, may occur during aseismic slow slip events on the megathrust. In addition, progressive dehydration of the hydrated mantle wedge south of the Mendocino triple junction may enhance the effects of a slab gap during the evolution of the California margin.
Calvert, A.J., M.A. Fisher, K. Ramachandran, and A.M. Tréhu, 2003, Possible emplacement of crustal rocks into the forearc mantle of the Cascadia subduction zone, Geophys. Res. Lett., 30(23), 2196, SDE 3-1 to 3-4, doi:10.1029/2003GL018541. PDF
Seismic reflection profiles shot across the Cascadia forearc show that a 5–15 km thick band of reflections, previously interpreted as a lower crustal shear zone above the subducting Juan de Fuca plate, extends into the upper mantle of the North American plate, reaching depths of at least 50 km. In the extreme western corner of the mantle wedge, these reflectors occur in rocks with P wave velocities of 6750–7000 m/s. Elsewhere, the forearc mantle, which is probably partially serpentinized, exhibits velocities of approximately 7500 m/s. The rocks with velocities of 6750–7000 m/s are anomalous with respect to the surrounding mantle, and may represent either: (1) locally high mantle serpentinization, (2) oceanic crust trapped by backstepping of the subduction zone, or (3) rocks from the lower continental crust that have been transported into the uppermost mantle by subduction erosion. The association of subparallel seismic reflectors with these anomalously low velocities favours the tectonic emplacement of crustal rocks.
Calvert, A.J., 2004, Seismic reflection evidence for two megathrusts in the northern Cascadia subduction zone, Nature, 428, 163-167. PDF
At convergent continental margins, the relative motion between the subducting oceanic plate and the overriding continent is usually accommodated by movement along a single, thin interface known as a megathrust. Great thrust earthquakes occur on the shallow part of this interface where the two plates are locked together. Earthquakes of lower magnitude occur within the oceanic plate, and have been linked to geochemical dehydration reactions caused by the plate’s descent. Here I present deep seismic reflection data from the northern Cascadia subduction zone which show that the inter-plate boundary is up to 16 km thick, and comprises two megathrusts that bound a >5 km-thick, ~110 km-wide region of imbricated crustal rocks. Earthquakes within the subducting plate occur predominantly in two geographic bands where the dip of the plate is inferred to increase as it is forced around the edges of the imbricated inter-plate boundary zone. This interpretation implies that seismicity in the subducting slab is controlled primarily by deformation in the upper part of the plate. Slip on the shallower megathrust, which may occur by aseismic slow slip, will transport crustal rocks into the upper mantle above the subducting oceanic plate, and may, in part, provide an explanation for P wave velocities as low as 6800 m/s that are observed there.
Calvert, A. J.,, K. Ramachandran, Honn Kao, and M. A. Fisher (2006). Local Thickening of the Cascadia Forearc Crust and the Origin of Seismic Reflectors in the Uppermost Mantle, Tectonophysics, in press. PDF
Seismic reflection profiles from three different surveys of the Cascadia forearc are interpreted using P wave velocities and relocated hypocentres, which were both derived from the first arrival traveltime inversion of wide-angle seismic data and local earthquakes. The subduction decollement, which is characterized beneath the continental shelf by a reflection of 0.5 s duration, can be traced landward into a large duplex structure in the lower forearc crust near southern Vancouver Island. Beneath Vancouver Island, the roof thrust of the duplex is revealed by a 5-12 km thick zone, identified previously as the E reflectors, and the floor thrust is defined by a short duration reflection from a <2 km thick interface at the top of the subducting plate. We show that another zone of reflectors exists east of Vancouver Island that is approximately 8 km thick, and identified as the D reflectors. These overlie the E reflectors; together the two zones define the landward part of the duplex. The combined zones reach depths as great as 50 km. The duplex structure extends for more than 120 km perpendicular to the margin, has an along-strike extent of 80 km, and at depths between 30 km and 50 km the duplex structure correlates with a region of anomalously deep seismicity, where velocities are less than 7000 m s -1. We suggest that these relatively low velocities indicate the presence of either crustal rocks from the oceanic plate that have been underplated to the continent or crustal rocks from the forearc that have been transported downward by subduction erosion. The absence of seismicity from within the E reflectors implies that they are significantly weaker than the overlying crust, and the reflectors may be a zone of active ductile shear. In contrast, seismicity in parts of the D reflectors can be interpreted to mean that ductile shearing no longer occurs in the landward part of the duplex. Merging of the D and E reflectors at 42-46 km depth creates reflectivity in the uppermost mantle with a vertical thickness of at least 15 km. We suggest that pervasive reflectivity in the upper mantle elsewhere beneath Puget Sound and the Strait of Georgia arises from similar shear zones.
Creager, K.C., L.A. Preston, R.L. Crosson, T. van Wagoner, A. Tréhu, and the SHIPS Working Group, 2002, Three-dimensional reflection image of the subducting Juan de Fuca plate, U.S. Geological Survey Open-File Report 02-328 and Geological Survey of Canada Open File 4350, p. 37-42. http://geopubs.wr.usgs.gov/open-file/of02-328/
The dynamics and environment under which intraslab earthquakes occur have been debated during the past several decades. Earthquake nucleation dynamics proposed for the shallow crust, involving brittle fracture when shear stresses overcome normal stresses, are difficult to extend to the high lithostatic pressures encountered in subduction zones. A process known as transformational faulting has been invoked to explain catastrophic faulting of ice at high confining pressure as ice undergoes a phase transition to ice II [Kirby, 1987]. The olivine to spinel phase transition exhibits similar behavior in the laboratory, offering a reasonable explanation for deep earthquakes [e.g., Green and Burnley, 1989; Houston and Green, 1995; Kirby, 1995; Kirby et al., 1991]. An important phase transformation in the shallowest 100 km of subduction zones is the transition of the basalt in the subducting oceanic crust to eclogite. Dehydration embrittlement, faulting associated with the release of fluids as a result of this phase transition, provides an especially attractive explanation for these earthquakes [Kirby, 1995]. The depth range over which this reaction proceeds is strongly temperature dependent because the equilibrium mineral assemblages depend strongly on the temperature [Peacock and Wang, 1999] and because the transition may be thermally inhibited in a cold, fast moving slab [Kirby, 1995]. For example, the oceanic crust of the Pacific plate subducting under northeast Japan is very old and fast moving. Earthquakes apparently in the subducted crust occur to depths as great as 100 km, as well as in a more active zone clearly within the mantle part of the subducted plate that extends deeper. In contrast, the Philippine Sea plate subducting below southwest Japan is young and slow, and contains shallow Wadati–Benioff seismicity down to only 60 km. Geotherms calculated by thermal modeling of these two subduction zones can explain the first order differences in observed seismicity in terms of basalt to eclogite phase transitions in equilibrium [Peacock and Wang, 1999].
An obvious test of the hypothesis that earthquakes are associated with the basalt to eclogite phase transition is to determine whether the earthquakes occur within the subducted crust or the subducted mantle. For example, do the earthquakes occur above or below seismic reflectors associated with the subducted oceanic crustmantle boundary, or with the interplate interface? P–to– S and S–to–P conversions from local earthquakes to seismometers in northern Honshu, Japan show convincing evidence for a strong discontinuity at the upper edge of the seismicity, probably associated with the interplate interface [e.g., Hasegawa et al., 1978; Matsuzawa et al., 1990; Zhao et al., 1997]. They infer that earthquakes occur within both the crustal and mantle portions of the old subducted Pacific plate.
In contrast, the width of the Cascadia seismogenic zone is much narrower, perhaps as narrow as five kilometers. Our goal is to determine the relative position of intraslab earthquakes to reflector surfaces estimated using data from the 1998 active-source Wet SHIPS experiment [Fisher et al., 1999]. We observe strong secondary arrivals (Figure 1) with mid-points under a broad area of the northern Olympic Peninsula (Figure 2). The amplitudes and polarities of reflected arrivals vary systematically over spatial scales as short as one to two kilometers (Figure 1). Beyond offsets of 80 km, the reflected arrivals are typically larger than direct arrivals. We picked approximately 500 travel times from stacks of reflected arrivals after applying a low-pass filter at 14 Hz. Using three-dimensional ray tracing through a three-dimensional tomography model obtained from SHIPS and earthquake first arrival data [Crosson et al., 1999], we determine the locus of all point scatterers which would correspond to an observed travel time for a given source and receiver. This defines an approximately ellipsoidal surface from which a ray could reflect and produce the observed travel time. For coherent arrivals, a reflector must be locally tangent to this ellipsoid. We invert for a smooth reflective surface that optimally is tangent to each of the data ellipsoids.
Kao, H., S.-Ju. Shan, G. Rogers, H. Dragert, J. F. Cassidy, and K. Ramachandran (2005). Depth distribution of seismic tremors along the northern Cascadia margin, Nature, 436, 841-844, doi:10.1038/nature03903. PDF
The Cascadia subduction zone is thought to be capable of generating major earthquakes with moment magnitude as large as Mw 5 9 at an interval of several hundred years. The seismogenic portion of the plate interface is mostly offshore and is currently locked, as inferred from geodetic data. However, episodic surface displacements—in the direction opposite to the long-term deformation motions caused by relative plate convergence across a locked interface—are observed about every 14 months with an unusual tremor-like seismic signature. Here we show that these tremors are distributed over a depth range exceeding 40kmwithin a limited horizontal band. Many occurred within or close to the strong seismic reflectors above the plate interface where local earthquakes are absent, suggesting that the seismogenic process for tremors is fluid-related. The observed depth range implies that tremors could be associated with the variation of stress field induced by a transient slip along the deeper portion of the Cascadia interface or, alternatively, that episodic slip is more diffuse than originally suggested.
Nedimovic, M.R., R.D. Hyndman, K. Ramachandran, and G.D. Spence, 2003, Reflection signature of seismic and aseismic slip on the northern Cascadia subduction thrust, Nature, 424, 416-420. PDF
At the northern Cascadia margin, the Juan de Fuca plate is underthrusting North America at about 45 mm/yr, resulting in the potential for destructive great earthquakes. The downdip extent of coupling between the two plates is difficult to determine because the most recent such earthquake (thought to have been in 1700) occurred before instrumental recording. Thermal and deformation studies indicate that, off southern Vancouver Island, the interplate interface is presently fully locked for a distance of 60km downdip from the deformation front. Great thrust earthquakes on this section of the interface (with magnitudes of up to 9) have been estimated to occur at an average interval of about 590 yr. Further downdip there is a transition from fully locked behaviour to aseismic sliding (where high temperatures allow ductile deformation), with the deep aseismic zone exhibiting slow-slip thrust events. Here we show that there is a change in the reflection character on seismic images from a thin sharp reflection where the subduction thrust is inferred to be locked, to a broad reflection band at greater depth where aseismic slip is thought to be occurring. This change in reflection character may provide a new technique to map the landward extent of rupture in great earthquakes and improve the characterization of seismic hazards in subduction zones.
Preston, L. A., 2003, Simultaneous inversion of 3D velocity structure, hypocenter Locations, and reflector geometry in Cascadia, Ph. D. thesis, University of Washington, Seattle, 135 pp. PDF
We present results from a non-linear inversion of direct and wide-angle reflection travel times for 3-D P-wave velocity structure, earthquake hypocenters, and reflector geometry under NW Washington focusing on the structure of the subducting Juan de Fuca plate. The first-arrival travel times are derived from both active-source experiments and from local earthquakes. The reflection arrivals were picked from data collected during the 1998 Wet SHIPS active-source experiment, which consisted of air-gun sources detonated within the inland waterways of NW Washington and SW British Columbia to land-based stations. As part of this research, we have developed a method of incorporating the reflection and first-arrival travel times into a simultaneous non-linear iterative inversion scheme for reflector geometry, 3-D velocities and earthquake relocations. This procedure reduces the well-known trade-off between reflector position and the velocities above it by including independent first-arrival information. Results indicate the wide-angle reflector to be the Moho of the subducting Juan de Fuca slab. The relocated intraslab earthquakes separate into two groups: those located up-dip of the 45km reflector depth contour generally lie below the reflector in the subducting mantle, while those down-dip of this contour primarily occur within the subducted oceanic crust. These results are consistent with the subducted mantle events being associated with serpentine dehydration embrittlement and the subducted crustal events being associated with the basalt to eclogite transformation. Error and resolution analyses demonstrate we have the necessary resolvability and can distinguish the relative locations among the velocities, reflector, and intraslab hypocenters within the subducting slab with sufficient precision to make our interpretations. Our results have important implications for our general understanding of the causes of intraslab earthquakes, earthquake hazards, and fluid processes within the shallowest portion of a warm subduction zone.
Preston, L.A., K.C. Creager, R.S. Crosson, T.M. Brocher, and A.M. Tréhu, 2003, Intraslab earthquakes: Dehydration of the Cascadia slab, Science, 302, no. 5648, p. 1197-1200. PDF
We simultaneously invert travel times of refracted and wide-angle reflected waves for three-dimensional compressional-wave velocity structure, earthquake locations, and reflector geometry in northwest Washington state. The reflector, interpreted to be the crust-mantle boundary (Moho) of the subducting Juan de Fuca plate, separates intraslab earthquakes into two groups, permitting a new understanding of the origins of intraslab earthquakes in Cascadia. Earthquakes up-dip of the Moho’s 45-kilometer depth contour occur below the reflector, in the subducted oceanic mantle, consistent with serpentinite dehydration; earthquakes located down-dip occur primarily within the subducted crust, consistent with the basalt-to-eclogite transformation.
Ramachandran,K., and R.D. Hyndman, 2005, P- and S-wave velocity structure beneath SW British Columbia: constraints on serpentinized forearc mantle wedge, Geophy. Res. Lett., in press. PDF
This article describes 3-D high-resolution P-and S-wave velocity models (Vp, Vs) of the forearc crust and upper mantle beneath SW British Columbia,constructed from earthquake travel-time tomography. The forearc mantle wedge exhibits low velocities inferred to be due to hydration and serpentinization of the mantle peridotite by the fluids rising from the dehydrating Juan de Fuca slab. Low Poisson’s ratio is inferred in the forearc crust just above the mantle wedge, possibly due to quartz deposited from rising silica-rich fluids. The Moho transition is inferred from a small increase in Vp and Vs at a depth of 36 km. The east dipping interface between the subducting oceanic crust and overlying forearc mantle is identified by higher values of Vp/Vs and Poisson’s ratio in the forearc mantle, close to the interface. The serpentinization in the forearc mantle wedge decreases landward from 45 to 20% volume, using peridotite-antigorite Vp and Vs versus serpentinization relations. The total water content in a unit column of the forearc mantle increases landward from 500 m3/m2 near the mantle wedge corner to 2500 m3/m2 approaching the arc. About 25 m.y. of subduction is estimated to produce this amount of water from dehydration of the downgoing slab.
Ramachandran, K., R. D. Hyndman, and T. M. Brocher, 2006, Regional P-wave velocity structure of the Northern Cascadia Subduction Zone, J. Geophys. Res., in review. PDF
This paper presents the first regional three dimensional P-wave velocity model for the Northern Cascadia Subduction Zone (S.W. British Columbia and N.W. Washington State) constructed through tomographic inversion of first-arrival times from active source experiments together with earthquake travel-time data recorded at permanent stations. The velocity model images the structure of the subducting Juan de Fuca plate, megathrust, and the forearc crust and mantle. Beneath southern Vancouver Island, the megathrust above the Juan de Fuca plate is characterized by a broad zone (25–35 km depth) having relatively low velocities of 6.4–6.6 km/s. These low velocities are inferred to represent subducting sedimentary rocks of the accretionary wedge. This relative low velocity zone coincides with the location of most of the episodic tremors recently mapped beneath Vancouver Island and its low velocity may also partially reflect the presence of trapped fluids and sheared lower crustal rocks. Serpentinization of the upper forearc mantle provides evidence for slab dewatering and densification. Tertiary sedimentary basins in the Strait of Georgia and Puget Lowland imaged by the velocity model lie above the inferred region of slab dewatering and densification and may therefore partly result from a higher rate of slab sinking. In contrast, sedimentary basins in the Strait of Juan de Fuca lie in a synclinal depressionin the Crescent Terrane. The correlation of inslab earthquake hypocenters M > 4 with P-wave velocities greater than 7.8 km/s suggest that they originate near and possibly below the oceanic Moho of the subducting Juan de Fuca plate.
Tréhu, A.M., T.M. Brocher, K. Creager, M. Fisher, L. Preston, G. Spence, and the SHIPS98 Working Group, 2002, Geometry of the subducting Juan de Fuca plate: New constraints from SHIPS98, U.S. Geological Survey Open-File Report 02-328 and Geological Survey of Canada Open File 4350, p. 25-32. http://geopubs.wr.usgs.gov/open-file/of02-328/
We have processed seismic reflection profiles acquired in the Strait of Juan de Fuca, inverted travel times of first arrivals from onshore recordings of the offshore shots to determine the velocity of the upper crust in this region and inverted the travel times of first and secondary arrivals that undershoot the central core of the Olympic Mountains. The seismic reflection profiles show a pattern of crustal reflectivity similar to that recorded beneath Vancouver Island. Inversion of first arrivals, interpreted to be diving waves through the upper and mid-crust indicate a 5–7 kilometer-deep linear, northwest-trending basin beneath the southwestern shore of the Strait and uplift of the basement Crescent terrane rocks beneath the northwestern shore. Velocities beneath the central Olympic Peninsula at 15–25 kilometers depth show strong lateral variation, with higher velocities underlying the surface exposures of the Crescent terrane and lower velocities at these depths beneath the Olympic core rocks. Inversion of secondary seismic arrivals interpreted to be reflections from the base of the crust of the subducted Juan de Fuca plate indicate that the Moho is at a depth of about 34 kilometers beneath the western Strait of Juan de Fuca and dips about 7o to the east, reaching a depth of about 46 kilometers beneath the eastern boundary of Olympic National Park. Correlating these results with results of previous experiments on Vancouver Island and in southwestern Washington confirm earlier interpretations of an arch in the subducted slab beneath the Olympic peninsula but indicate that the arch is asymmetric and less pronounced than previously thought. We attribute both the northwest-trending folding of the Crescent terrane and the asymmetry in the shape of the subducting slab to resistance to the northward motion of the Paleocene-age Cascadia forearc terrane by the thick lithosphere of the pre-Tertiary terranes of British Columbia.