SEARCH

SEARCH BY CITATION

Keywords:

  • Alaska;
  • dynamics;
  • mantle flow;
  • numerical modeling;
  • slab edge;
  • subduction

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Alaskan Subduction-Transform Boundary
  5. 3. Methodology
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[1] Away from subduction zones, the surface motion of oceanic plates is well correlated with mantle flow direction, as inferred from seismic anisotropy. However, this correlation breaks down near subduction zones, where shear wave splitting studies suggest the mantle flow direction is spatially variable and commonly non-parallel to plate motions. This implies local decoupling of mantle flow from surface plate motions, yet the magnitude of this decoupling is poorly constrained. We use 3D numerical models of the eastern Alaska subduction-transform plate boundary system to further explore this decoupling, in terms of both direction and magnitude. Specifically, we investigate the role of the slab geometry and rheology on the mantle flow velocity at a slab edge. The subducting plate geometry is based on Wadati-Benioff zone seismicity and tomography, and the 3D thermal structure for both the subducting and overriding plates, is constrained by geologic and geophysical observations. In models using the composite viscosity, a laterally variable mantle viscosity emerges as a consequence of the lateral variations in the mantle flow and strain rate. Spatially variable mantle velocity magnitudes are predicted, with localized fast velocities (greater than 80 cm/yr) close to the slab where the negative buoyancy of the slab drives the flow. The same models produce surface plate motions of less than 10 cm/yr, comparable to observed plate motions. These results show a power law rheology, i.e., one that includes the effects of the dislocation creep deformation mechanism, can explain both observations of seismic anisotropy and decoupling of mantle flow from surface motion.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Alaskan Subduction-Transform Boundary
  5. 3. Methodology
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[2] Plate boundaries are inherently three-dimensional (3D) tectonic features with variations in geometry and physical properties along their length [Jarrard, 1986; Gudmundsson and Sambridge, 1998; Bird, 2003; Lallemand et al., 2005; Sdrolias and Muller, 2006; Syracuse and Abers, 2006; Rychert et al., 2008]. For example, subduction zones have significant changes in slab dip along strike, including flat slab segments, as in the eastern Alaska subduction zone, the Peru-Chile Trench, the Cascadia subduction zone, and the southwestern Japan Trench [Gudmundsson and Sambridge, 1998; Gutscher et al., 2000; Ratchkovski and Hansen, 2002; Lallemand et al., 2005; Syracuse and Abers, 2006; Tassara et al., 2006]. In addition, many plate boundaries are not simply end-member subduction zones, transform faults, or spreading centers in isolation, but rather are hybrid systems. For example, plate boundary corners with slab edges formed by the intersection of a subduction zone with a transform fault occur in eastern Alaska, the northern Kuril-Kamchatka subduction zone, the southernmost Mariana trench near the Yap trench, the northern Tonga trench, and the north and south boundaries of the Antilles and the Scotia subduction zones [Gudmundsson and Sambridge, 1998; Bird, 2003]. Subduction-transform intersections also occur in non-corner geometries, as in the termination of the Kermadec Trench against the Alpine fault in New Zealand and the north and south terminations of the Cascadian subduction zone [Gudmundsson and Sambridge, 1998; Bird, 2003]. The three-dimensionality in both surface and subsurface structure can influence the relative motion of the mantle with respect to the surface plates. Thus, in many cases a two-dimensional (2D) approach to studying plate boundary processes can miss critical aspects of the dynamics resulting purely from the geometry.

[3] Away from subduction zones, the surface motion of oceanic plates is well correlated with mantle flow direction, as inferred from seismic anisotropy [Kreemer, 2009; Conrad et al., 2007; Becker et al., 2003]. However, observations of seismic anisotropy from numerous subduction zones display seismic fast directions that are non-parallel to plate motions, e.g., the Peru-Chile, Kamchatka, Tonga, Marianas, South Sandwich, Cascadia, Costa Rica-Nicaragua, and eastern Alaska subduction zones [Russo and Silver, 1994; Peyton et al., 2001; Smith et al., 2001; Pozgay et al., 2007; Muller et al., 2008; Zandt and Humphreys, 2008; Long and Silver, 2008; Abt et al., 2009; Christensen and Abers, 2010]. If the seismic fast directions are tracking mantle flow [Savage, 1999; Kaminiski and Ribe, 2002; Kneller and van Keken, 2007; Long and Silver, 2009; Jadamec and Billen, 2010], this implies that local decoupling between the plates and underlying mantle may be a common feature in subduction zones. This also indicates differential motion, at least in terms of direction, of the mantle close to subduction zones versus that farther away from them.

[4] Several 3D semi-analytic, numerical, and laboratory experiments have investigated mantle flow dynamics in subduction zones [Zhong and Gurnis, 1996; Zhong et al., 1998; Buttles and Olson, 1998; Hall et al., 2000; Kincaid and Griffiths, 2003; Schellart, 2004; Funiciello et al., 2006; Piromallo et al., 2006; Royden and Husson, 2006; Schellart et al., 2007; Kneller and van Keken, 2008; Giuseppe et al., 2008; Jadamec and Billen, 2010]. For example, 3D subduction studies, commonly in the context of slab rollback, indicate a toroidal component of mantle flow dominates near the slab edge, where material is transported from beneath the slab around the slab edge and into the mantle wedge [Kincaid and Griffiths, 2003; Schellart, 2004; Funiciello et al., 2006; Piromallo et al., 2006; Stegman et al., 2006]. This component of flow cannot be captured in 2D models of subduction [Tovish et al., 1978; Garfunkel et al., 1986]. In addition, numerical models investigating the effect of variable slab dip and curvature on flow in the mantle wedge show that the variable geometry can induce pressure gradients that drive trench parallel flow [Kneller and van Keken, 2007, 2008]. Thus, both observations of seismic anisotropy and 3D fluid dynamics simulations indicate a difference between the direction of mantle flow and the surface plate motion, indicating local decoupling of the surface plates from the mantle in subduction zones [Jadamec and Billen, 2010].

[5] In this paper, we use 3D instantaneous regional models of the eastern Alaska subduction-transform plate boundary system to further explore this decoupling of mantle flow from surface plate motion, in terms of both direction and magnitude. The 3D models of the subduction-transform plate boundary system in southern Alaska use a subducting plate geometry based on Wadati-Benioff zone seismicity and a 3D thermal structure for both the subducting and overriding plates, constrained by geologic and geophysical observations. No kinematic boundary condition is prescribed to the subducting plate, rather the negative buoyancy of the slab drives the flow. We employ a composite viscosity, which includes both the diffusion and dislocation creep mechanisms (i.e., Newtonian and non-Newtonian viscosity).

[6] This study builds on the 3D numerical models presented by Jadamec and Billen [2010] that investigated the effect of the slab shape and Newtonian versus non-Newtonian rheology on rapid mantle flow and compared the results to observations of seismic anisotropy from Christensen and Abers [2010]. In this paper, we explore a larger range in model parameters and specifically address how these affect the velocity magnitude in the mantle surrounding the slab, the overall pattern of mantle flow around the slab edge, as well as implications of the non-Newtonian rheology as a mechanism for plate-mantle decoupling in subduction zones. In addition, we discuss implications specific to Alaskan tectonics. The approach taken for models presented in this paper, as well as those of Jadamec and Billen [2010], differs from that in previous 3D numerical models of subduction in that previous models [Moresi and Gurnis, 1996; Zhong and Gurnis, 1996; Billen et al., 2003; Billen and Gurnis, 2003; Piromallo et al., 2006; Stegman et al., 2006; Schellart et al., 2007; Kneller and van Keken, 2007; Giuseppe et al., 2008; Kneller and van Keken, 2008] either use a simplified plate geometry, prescribe a velocity boundary condition to the subducting plate, do not include an overriding plate, or use a Newtonian rheology in the mantle.

2. Alaskan Subduction-Transform Boundary

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Alaskan Subduction-Transform Boundary
  5. 3. Methodology
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[7] The Aleutian-Alaska subduction zone accommodates convergence between the northernmost Pacific plate and the Alaskan part of the North American plate (Figure 1). The subduction zone length is greater than 3000 km, and along this length the sense of curvature changes from convex to the south in the central Aleutians to convex to the north in south central Alaska [Page et al., 1989; Gudmundsson and Sambridge, 1998; Ratchkovski and Hansen, 2002; Bird, 2003]. Offshore south central Alaska, the observed Pacific plate motion with respect to North America is N16°W at 5.2 cm/yr [DeMets and Dixon, 1999]. The easternmost part of the subduction zone forms a plate boundary corner where the Aleutian trench terminates against the northwest trending Fairweather-Queen Charlotte transform fault (Figure 1) [Bird, 2003]. In this plate boundary corner, the dip of the subducting plate shallows to form a flat slab [Ratchkovski and Hansen, 2002], and the Yakutat block, a buoyant oceanic plateau, is subducting beneath the North American plate [Bruns, 1985; Pavlis et al., 2004; Ferris et al., 2003] (Figure 1).

image

Figure 1. Surface of subducted portion of the Pacific plate for the two slab shapes (a) slabE325 and (b) slabE115 tested in the models. Seismic data used to generate slab surfaces are superimposed. Location map included in lower right. See Table 4 for additional data used to constrain slab morphology.

Download figure to PowerPoint

[8] East of approximately 212° longitude, where the subduction zone approaches the transform boundary, there is an abrupt decrease in seismicity at depths greater than 25 km (Figure 1). Despite the decrease in seismicity, several seismic studies suggest there is a steeply dipping slab, the Wrangell slab, in this region of the plate boundary corner beneath the active adakitic Wrangell Volcanoes (Figures 1 and 2) [Stephens et al., 1984; Skulski et al.,1991; Preece and Hart, 2004]. There is debate over the geometry of the Wrangell slab as well as its continuity with the main Aleutian slab to the west [Perez and Jacob, 1980; Stephens et al., 1984; Page et al., 1989; Zhao et al., 1995; Ratchkovski and Hansen, 2002; Eberhart-Phillips et al., 2006; Fuis et al., 2008]. Therefore, we construct two slab geometries, one with and one without the Wrangell slab, to test the influence of the slab geometry on the mantle flow in this plate boundary corner (Figure 1).

image

Figure 2. Temperature structure for 3D geodynamic model. (a) Seafloor ages for Pacific plate and thermal domains in upper plate used in half-space cooling model. Seafloor ages from Muller et al. [1997]. See text for constraints on upper plate age assignment. (b) Temperature profiles for domains in upper plate: I - forearc, II - magmatic arc, III - Cordillera, IV - Ancestral North America. Note temperature profiles for domains I and IV overlap because the two regions have the same effective age. (c) Radial slice through temperature at 26 km depth, which is the same for models using slabE325 and slabE115. Radial slice at 152 km depth for models with (d) slabE325 and (e) slabE115. (f) Cross section through flat slab region (AA′), which is the same for models using slabE325 and slabE115. Cross section through the Wrangell Volcanics (BB′) for models using (g) slabE325 and (h) slabE115. Note, Figures 2b–2h are shown for subset of model domain.

Download figure to PowerPoint

3. Methodology

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Alaskan Subduction-Transform Boundary
  5. 3. Methodology
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[9] We present eighteen 3D numerical fluid dynamics experiments of the subduction-transform plate boundary system in southern Alaska, in which we varied the viscosity structure (Newtonian versus composite viscosity), the subducting plate geometry (slabE325 versus slabE115), the yield stress (σy), the plate boundary coupling (ηwk), and the thermal structure (Table 1). Because the approach taken here is to model a regional plate boundary, it was necessary to map regional observations of slab geometry and the plate boundary configuration, as well as the thermal structure of the overriding plate and slab, onto a model grid. A C and C++ code, referred to as SlabGenerator, was written to generate and map this initial configuration onto a 3D model mesh that was locally refined in regions of geometric complexity and where large viscosity contrasts were expected to occur [Jadamec, 2009; Jadamec and Billen, 2010]. The 3D plate boundary configuration and thermal structure were then used as input to the finite element mantle convection code, CitcomCU [Moresi and Solomatov, 1995; Moresi and Gurnis, 1996; Zhong, 2006], as will be described in the sections that follow.

Table 1. Summary of Model Parameters and Resultsa
ModelSlabTopRheologyσy Max. (MPa)PBSZ (Pa s)Vpacsurf (cm/yr, Azimuth)Vop (cm/yr, Azimuth)Vwedge (cm/yr, Azimuth)
  • a

    Top is temperature structure of overriding plate, either (var) laterally variable effective age or (unif) laterally uniform effective age with both using an age-dependent temperature formulation. Vpacsurf is the Pacific plate surface velocity offshore south central Alaska at 217.3° longitude, 57.0° latitude. Vop is the surface velocity in the overriding plate at 210.5° longitude, 64.5° latitude. Vwedge is the velocity in mantle wedge at 210.5° longitude, 64.5° latitude, and 100 km depth. At this location in the mantle wedge, horizontal velocities dominate with vertical velocities less than 3% of the horizontal velocity magnitude. Vertical velocities increase with proximity to the pivoting slab tip.

A1n325varNewtonian5001 × 10202.11, 338.91°0.23, 263.09°2.56, 185.80°
A2n325varNewtonian5001 × 10210.76, 328.64°0.18, 285.15°2.43, 183.03°
A3n325varNewtonian10001 × 10202.10, 338.39°0.23, 261.17°2.25, 187.02°
A4n325varNewtonian10001 × 10210.74, 327.29°0.18, 282.83°2.07, 183.67°
A1c325varComposite5001 × 10205.28, 345.92°0.06, 280.92°31.68, 189.84°
A2c325varComposite5001 × 10212.58, 345.89°0.06, 290.08°28.52, 189.51°
A3c325varComposite10001 × 10205.17, 342.93°0.06, 275.49°15.40, 180.79°
A4c325varComposite10001 × 10212.47, 342.73°0.06, 283.39°12.12, 179.42°
B1n115varNewtonian5001 × 10201.92, 332.63°0.24, 264.99°3.15, 204.87°
B2n115varNewtonian5001 × 10210.69, 320.38°0.19, 279.15°2.86, 203.13°
B3n115varNewtonian10001 × 10201.92, 332.11°0.24, 263.61°2.88, 207.91°
B4n115varNewtonian10001 × 10210.68, 318.90°0.19, 277.28°2.52, 206.48°
B5n115unifNewtonian5001 × 10210.74, 318.45°0.26, 264.21°3.16, 203.42°
B1c115varComposite5001 × 10204.40, 339.59°0.07, 276.41°65.78, 225.59°
B2c115varComposite5001 × 10212.04, 339.45°0.07, 284.45°79.79, 226.71°
B3c115varComposite10001 × 10204.40, 336.57°0.07, 273.33°40.82, 225.39°
B4c115varComposite10001 × 10212.02, 336.18°0.06, 279.12°32.32, 225.24°
B5c115unifComposite5001 × 10212.10, 340.39°0.08, 273.39°94.76, 225.81°

3.1. 3D Model Domain and Driving Forces

[10] The 3D regional models of the subduction-transform plate boundary system in southern Alaska include an overriding plate (the North American plate), a subducting plate (the Pacific plate), and the underlying mantle (Figure 3). A viscous shear zone separates the Pacific and North American plates. The driving forces in the system are the negative thermal buoyancy of the subducting slab (slab pull) and the positive thermal buoyancy of the Juan de Fuca ridge (ridge push), which lies along the southeast boundary of the Pacific plate within the model domain (Figure 3). The resisting forces are the viscous stresses in the mantle, the plate boundary shear zone, and within the interior of the slab. There are no driving velocities applied anywhere in the model, rather the models predict the flow velocities for the plates and mantle. We note that while the crust and harzburgitic residue of melting making up the lithosphere have a net positive buoyancy, once the crust is converted to eclogite the net compositional buoyancy is zero because the positive buoyancy of the thicker (2.5 times) harzburgitic layer cancels out the effect of the eclogitized crust [Oxburgh and Parmentier, 1977; Hacker et al., 2003]. Therefore, we choose not to include these compositional effects as the imposed shape of the slab essentially accounts for the positive buoyancy of the crust at shallow depths.

image

Figure 3. 3D model set-up. NAM- North American plate; PAC- Pacific plate; Slab- subducted part of Pacific plate with portion of slab geometry that is varied outlined with short-dashed black line; PBSZ- plate boundary shear zone; SMSZ - southern mesh boundary shear zone; and JdFR - Juan de Fuca Ridge. Cross sections A and B correspond to those in Figure 2.

Download figure to PowerPoint

[11] The 3D model domain spans from 185° to 240° longitude, 45°N to 72°N latitude, and 0 to 1500 km in depth (Figure 3). The finite element mesh varies in resolution from 0.04° to 0.255° in the longitudinal direction, 0.0211° to 0.18° in the latitudinal direction, and 2.35 km to 25 km in the radial direction, with the highest resolution centered on the plate boundary corner in south central Alaska. The mesh contains 960 x 648 x 160 elements in the longitudinal, latitudinal, and radial directions, respectively. Models were run using 360 processors on Lonestar, a Dell Linux cluster at the Texas Advanced Computing Center, for approximately 48 hours per job in models with the composite viscosity and for less time in models with the Newtonian only viscosity.

[12] Reflecting (free-slip) boundary conditions are used on all boundaries. Because the North American plate has a high viscosity, it is essentially fixed horizontally at the model domain sides by the free-slip boundary conditions providing a fixed reference frame for the relative motion of the Pacific plate. In contrast, we use a low viscosity zone to decouple the Pacific plate from the southern edge of the model domain (see below). This allows the Pacific plate to move freely in response to the local driving forces. 2D tests were also used to determine the necessary box depth and width in order to minimize boundary condition effects on the flow in the subduction zone. These are all instantaneous flow simulations designed to explore the present-day balance of forces, lithosphere and mantle structure.

3.2. Subducting Plate Geometry

[13] We construct two 3D slab shapes, slabE325 and slabE115, to test competing hypotheses for the geometry and continuity of the subducted oceanic lithosphere beneath south central Alaska (Figure 1). We assume that the Wadati-Benioff zone represents the shape of the subducting lithosphere, and use seismic data to constrain the shape and depth extent of the slab (Figure 1; Table 4; and references therein). In models using slabE325, the depth of the slab is everywhere 325 km, and thus the Aleutian and Wrangell slabs are continuous. In contrast, in models using slabE115, east of 212° longitude the slab surface extends to only to 115 km depth. Thus in models using slabE115, the slab is shorter beneath the Wrangell Mountains and barely protrudes beneath the lithosphere of the overriding plate in this region of the model (Figure 1; Table 4).

[14] The 3D slab surfaces (Figure 1) were generated using a tensioned cubic spline algorithm in GMT [Wessel and Smith, 1991]. The slab surface was then mapped onto the finite element grid using the SlabGenerator code by calculating the perpendicular distances of the finite element mesh points to the slab surface. In this way, the varying strike, dip and depth of the slab are smoothly represented on the model grid. From these perpendicular distances, the initial 3D thermal structure and 3D plate boundary shear zone structure are constructed.

3.3. Subducting Plate Thermal Structure

[15] A 3D thermal structure constrains the temperature-dependent viscosity and the density anomaly that drives the flow in the geodynamic models. The thermal structure is based on geologic and geophysical observables, for both the subducting and overriding plates, thereby capturing the regional variability in the plate boundary system (Figure 2).

[16] The 3D thermal structure for the subducting plate, generated by the SlabGenerator code, is constructed using a semi-infinite half-space cooling model in which the thickness of the thermal lithosphere is a function of the plate age [Turcotte and Schubert, 2002]:

  • display math

where Tm = 1400°C, Ts = 0°C, κ = 1 × 10−6 m2/s, d is depth perpendicular from the slab surface, and t is the plate age. The half-space cooling model is an appropriate approximation of the thermal structure of oceanic lithosphere where the plate age is less than approximately 80 Myr [Turcotte and Schubert, 2002; Hillier and Watts, 2005], which is the case for the majority of the Pacific plate included in our models (Figure 2a) [Muller et al., 1997].

[17] Along the length of the Aleutian trench, the seafloor age decreases eastward, from approximately 70 Myr at 185° E longitude to approximately 30 Myr in the plate boundary corner (Figure 2a). The seafloor age at the trench is extrapolated onto the subducted part of the Pacific plate so that there is also an eastward decrease in the age of the slab. The eastward decrease of plate age creates an the eastward decrease in thermal and mechanical thickness of the Pacific plate. Note, that we do not include a compositional density anomaly corresponding to the Yakutat block in our models, and therefore, in the plate boundary corner, we assume the age of the seafloor beneath the Yakutat block is the same as that just southwest of the Transition fault (Figure 2).

[18] To simulate conductive warming of the slab in the mantle, a depth-dependent correction factor based on a length-scale diffusion analysis is applied to d, the distance perpendicular from the slab surface used in equation (1). We adjust the minimum value of d as a function of increasing vertical depth in the model, such that the minimum value for d on the slab surface increases from 0 km for the unsubducted part of the Pacific plate to 15 km on the slab surface at 300 km depth. Before this correction is added, T = 0°C everywhere on the slab surface, even on the parts of the slab surface that are at 300 km depth. After this correction, the minimum temperature on the slab surface smoothly increases with increasing slab depth, simulating the warming that would occur as the slab is immersed within the mantle, with the deeper parts having been immersed longer and hence warmed longer. The slab thermal field is then blended into the ambient mantle thermal field using a sigma-shaped smoothing function.

3.4. Overriding Plate Thermal Structure

[19] The 3D thermal structure for the overriding North American plate, constructed with SlabGenerator, also assumes a semi-infinite half-space cooling model (equation (1)), but uses effective thermal ages assigned to the continental regions as a proxy to capture the temperature structure within the continental lithosphere (Figures 2a and 2b). We elect this approach rather than using surface heat flow observations to directly invert for the temperature as a function of depth, because there are few coupled surface heat flow and radiogenic heat production measurements in Alaska. Without coupled measurements, the relative contribution of heat from radioactive decay within the crust cannot be quantitatively separated from the contribution of heat from the underlying mantle [Turcotte and Schubert, 2002].

[20] Thus, using equation (1), we construct a spatially variable depth-dependent thermal structure for the overriding plate, with the overriding plate subdivided into four thermal domains: the Cordilleran region, the magmatic arc, the forearc, and ancestral North America (Figures 2a and 2b). We base these subdivisions on a synthesis of regional geophysical and geologic observations, including the surface heat flow [Blackwell and Richards, 2004], location of Neogene volcanism (Alaska Volcano Observatory) [Plafker et al., 1994], seismic profiles [Fuis et al., 2008], moho temperature estimates, major terrane boundaries [Greninger et al., 1999], and the integrative work characterizing the thermal structure of the lithosphere in western Canada by Currie and Hyndman [2006] and Lewis et al. [2003]. For the Cordilleran domain, we assume the relatively warm Cordilleran back arc lithosphere in western Canada [Currie and Hyndman, 2006; Lewis et al., 2003] extends into mainland Alaska (Figures 2a and 2b), and assign this region an effective age of 30 Myr, corresponding to the lithospheric thickness of a continental mobile belt [Blackwell, 1969; Hyndman et al., 2005]. The magmatic arc is assigned an effective age of 10 Myr, giving it the thinnest (warmest) lithosphere with the 1200°C isotherm at approximately 40 km. The forearc and ancestral North America (effective ages of 80 Myr) have the thickest lithosphere reaching 1200°C at approximately 110 km depth (Figure 2b). The resultant 3D thermal structure is shown in Figures 2c–2e.

[21] To test the effect of the lateral variations in the overriding plate thermal structure, we also construct a depth-dependent thermal structure that uses a single effective age and thus has a laterally uniform thermal field throughout the entire overriding plate. We choose an effective age of 30 Myr, which corresponds to a lithospheric thickness of approximately 60 km at the 1200°C isotherm and 90 km at the 1350°C isotherm (Figure 2b), and is comparable to that of warm continental back arcs [Blackwell, 1969; Hyndman et al., 2005].

3.5. Plate Boundary Shear Zone

[22] In numerical models of subduction the plate interface is typically represented either as a discrete fault surface [Kincaid and Sacks, 1997; Zhong et al., 1998; Billen et al., 2003] or a narrow low viscosity layer [Gurnis and Hager, 1988; Kukacka and Matyska, 2004; Billen and Hirth, 2007]. We model the plate interface as a narrow 3D low viscosity layer in which the imposed reduction in viscosity represents a number of processes including the shear heating, variation in grain size, and the presence of water and sediments. We use the SlabGenerator code to define the plate interface, referred to as the plate boundary shear zone (PBSZ), that is at least 40 km wide and extends to 100 km in depth [Jadamec, 2009; Jadamec and Billen, 2010]. The rheological implementation of the narrow 3D low viscosity layer (PBSZ) is described in section 3.7. The orientation of the PBSZ is spatially variable following the strike and dip of the subducting plate and is continuous with the vertically-oriented Fairweather-Queen Charlotte transform boundary (Figures 1 and 3). The PBSZ location is fixed in space, which is appropriate for these models of the present-day deformation [Sdrolias and Muller, 2006]. We also include a narrow vertical low viscosity layer located along the southern edge of the Pacific plate in our model to decouple the Pacific plate from the southern model boundary. This low viscosity layer is implemented by the same method as for the PBSZ. No low viscosity layer is imposed on the Juan de Fuca ridge, because the localized warm temperatures at the spreading center decouple the plates in that region. In addition, no weak zone is imposed on the western-most boundary as this was found to have little affect on the model results.

3.6. Governing Equations and Numerical Method

[23] The open source finite element code, CitcomCU [Zhong, 2006], based on CITCOM [Moresi and Solomatov, 1995; Moresi and Gurnis, 1996], is used to solve for the viscous flow in the 3D models of the Alaska plate boundary corner. CitcomCU solves the Navier-Stokes equation for the velocity and pressure, assuming an incompressible fluid with a high Prandlt number, given by the conservation of mass:

  • display math

and conservation of momentum:

  • display math

where u, σ, ρo, α, T, To, g, and δrr are the velocity, stress tensor, density, coefficient of thermal expansion, temperature, reference temperature, acceleration due to gravity, and Kronecker delta, respectively [Moresi and Solomatov, 1995; Zhong, 2006]. The models are instantaneous, therefore the energy equation is not solved.

[24] The constitutive relation is defined by

  • display math

where P is the dynamic pressure, ηeff is the effective viscosity defined in equation (9), and inline image is the strain rate tensor. The models are defined and solved in spherical coordinates. The equations of motion are non-dimensionalized by the Rayleigh number, Ra, is defined by

  • display math

where ΔT = ToTsurf and ρo, α, g, R, ηref, and κ are as defined in Table 2.

Table 2. Dimensionalization Parameters
ParameterDescriptionValue
RaRayleigh number2.34 × 109
gacceleration due to gravity, m/s29.8
Toreference temperature, K1673
Tsurftemperature on top surface, K273
Rradius of Earth, m6371 × 103
ηrefreference viscosity, Pa⋅s1 × 1020
ρoreference density, kg/m33300
κthermal diffusivity, m/s21 × 10−6
αthermal expansion coefficient, K−12.0 × 10−5

3.7. 3D Model Rheology

[25] Experimental studies and the observation of seismic anisotropy throughout the upper mantle, suggest that dislocation creep is the dominant deformation mechanism of olivine in the upper mantle [Hirth and Kohlstedt, 2003]. The non-linear relationship between stress and strain rate for deformation by dislocation creep leads to a faster rate of deformation (higher strain rates) for a given stress, which can be expressed as a reduction in the effective viscosity (i.e., inline image). Previous 3D numerical modeling studies using only Newtonian viscosity (diffusion creep) have shown that imposing a region of low viscosity in the mantle wedge provides a better fit to the geoid and topography [Billen and Gurnis, 2001; Billen et al., 2003; Billen and Gurnis, 2003]. Because mantle flow strain rates are high in the corner of the mantle wedge, the imposed low viscosity region in these previous models may be due to the weakening effect of the dislocation creep mechanism, providing one example of the possible importance of including the non-Newtonian (dislocation creep) viscosity. Therefore following the implementation of Billen and Hirth [2007], in our 3D models of the Alaska subduction zone, we modified CitcomCU to use a composite viscosity in the upper mantle, that includes the effects of both dislocation (ds) and the diffusion (df) creep deformation mechanisms [Jadamec and Billen, 2010].

[26] For deformation under a fixed stress (driving force) the total strain rate is the sum of the contributions from deformation by the diffusion and dislocation creep mechanisms:

  • display math

where inline image without the subscripts i-j refers to the second invariant of the strain rate tensor, inline image. Substituting, inline image, the composite viscosity, ηcom, can be defined by

  • display math

[27] The viscosity components, ηdf and ηds, are defined assuming the experimentally-determined viscous flow law governing deformation for olivine aggregates [Hirth and Kohlstedt, 2003] such that

  • display math

where Pl is the lithostatic pressure, R is the universal gas constant, T is non-adiabatic temperature, Tad is the adiabatic temperature (with an imposed gradient of 0.3 K/km), and A, n, d, p, COH, r, E, and V are as defined in Table 3, assuming no melt is present [Hirth and Kohlstedt, 2003].

Table 3. Flow Law Parameters, Assuming Diffusion and Dislocation Creep of Wet Olivinea
VariableDescriptionCreepdfCreepds
Apre-exponential factor1.09 × 10−20
nstress exponent13.5
dgrain size, μm (if A in μm)10 × 103
pgrain size exponent3
COHOH concentration, H/106 Si10001000
rexponent for COH term11.2
Eactivation energy, kJ/mol335480
Vactivation volume, m3/mol4 × 10−611 × 10−6

[28] For diffusion creep, the strain rate depends linearly on the stress (n = 1) but non-linearly on the grain size (p = 3) [Hirth and Kohlstedt, 2003]. A grain size of 10 mm is used for the upper mantle giving a background viscosity of 1020 Pa s at 250 km. In contrast, for dislocation creep of olivine, the strain rate dependence is non-linear for dislocation creep of olivine (n = 3.5), but there is no grain-size dependence, giving the same background viscosity at 250 km for a strain rate of 10−15 s−1 [Hirth and Kohlstedt, 2003]. For higher strain rates, the composite viscosity will lead to lower effective viscosity due to most of the deformation being accommodated by the dislocation creep mechanism.

[29] The lower mantle viscosity is included using the Newtonian flow law for olivine, with a larger effective grain size (70 mm) in order to create a viscosity jump by a factor of 30 from the upper to lower mantle. This simplified viscosity structure for the lower mantle is consistent with the magnitude of the viscosity jump constrained by post-glacial rebound [Mitrovica, 1996] and models of the long wavelength geoid [Hager, 1984], as well as the lack of observations of seismic anisotropy in the lower mantle and the few experimental constraints on the viscous behavior of perovskite.

[30] Close to the earth's surface and within the cold core of the subducted slab, the viscosity values determined by equation (7) become unrealistically large in that they imply a rock strength much greater than that predicted by laboratory experiments [Kohlstedt et al., 1995]. To allow for plastic yielding where the stresses calculated in the model exceed those predicted from laboratory experiments, the stresses calculated in the model are limited by a depth-dependent yield stress. The yield stress increases linearly with depth, assuming a gradient of 15 MPa per km, from 0.1 MPa at the surface to a maximum value of either 500 MPa or 1000 MPa (Table 1), with the maximum value based on constraints from experimental observations, low-temperature plasticity, and the dynamics in previous models [Kohlstedt et al., 1995; Weidner et al., 2001; Hirth, 2002; Billen and Hirth, 2005, 2007].

[31] We thus define the effective viscosity, ηeff:

  • display math
  • display math

where ηcom, σII, inline image, and σy are the composite viscosity, second invariant of the stress tensor, second invariant of the strain rate tensor, and yield stress, respectively. The effective viscosity is solved for as an additional solution loop in CitcomCU that iterates until the global difference between the velocity field of consecutive solutions is less than some specified value, typically 1% [Billen and Hirth, 2007].The models allow for viscosity variations of up to seven orders of magnitude, with a minimum viscosity cutoff value of 5 × 1017 Pa s and a maximum value of 1 × 1024 Pa s.

[32] To incorporate the 3D PBSZ defined by SlabGenerator into CitcomCU, we modified CitcomCU to read in a weak zone field, such that the weak regions indicative of the plate boundaries are smoothly blended into the background viscosity, ηeff, using

  • display math

where ηeff is the effective viscosity as defined in equation (9), and ηo is the reference viscosity equal to 1 × 1020 Pa s [Jadamec, 2009; Jadamec and Billen, 2010]. Awk is the scalar weak zone field, defined using SlabGenerator, assuming a sigma-function with values ranging from 0 to 1 [Jadamec, 2009; Jadamec and Billen, 2010]. Values of 1 correspond to fully weakened regions in the center of the shear zone and values of 0 correspond to unweakened regions. The ηwk value serves as an upper bound on the viscosity in the shear zone and will be overwritten if the viscosity calculated by equation (9) is lower. Thus, the final form of the viscosity, ηf, becomes

  • display math

[33] Although we cannot compare the complex 3D numerical models presented in our paper to an analytic solution, previous 2D modeling by Moresi and Solomatov [1995] and Moresi et al. [1996] using CITCOM indicates that for models with large viscosity variations, error can be reduced by limiting the viscosity jump across each element, preferably to a factor of 3 or less. CitcomCU implements the full multigrid method to accelerate convergence [Zhong, 2006], which was found to save on valuable compute time in the high resolution regional models of the Alaska plate boundary containing large viscosity variations [Jadamec, 2009]. In our models, to limit the viscosity jump in the PBSZ, the PBSZ width varies with the resolution of the finite element mesh so that the PBSZ always spans a minimum number of elements. We found that numerical stability could be obtained using 8 elements to span four orders of magnitude of viscosity, i.e., we allow for viscosity jumps of up to a factor of 5 across the elements in the PBSZ. This implementation was tested on a series of 3D models with a simplified subduction zone geometry and we found that viscosity contrasts on this order led to good convergence behavior [Jadamec, 2009].

3.8. 3D Visualization

[34] The increasing incorporation of high performance computing and massive data sets into scientific research has led to the need for high fidelity tools to analyze and interpret the information [Erlebacher et al., 2001; Kreylos et al., 2006; Chen et al., 2008; Kellogg et al., 2008]. Immersive 3D visualization facilities provide one approach to fill this gap in the workflow, especially when working with complex spatially varying data and nonlinear model behavior [e.g., Kreylos et al., 2006; Billen et al., 2008].

[35] The open source software 3DVisualizer [Kreylos et al., 2006; Billen et al., 2008; Jadamec et al., 2008] was used in the Keck Center for Active Visualization in the Earth Sciences (KeckCAVES) for rapid inspection and interactive exploration of the 3D plate boundary geometry and thermal structure output from the SlabGenerator code. In this way, the quality and smoothness of the features mapped onto the model grid, containing over 100 million finite element nodes, could be assessed efficiently. This was especially useful because of the large model size (which amounted to inspecting almost two thousand 2D model slices for each 3D model) and because of the geometrically complex slab and plate boundary configuration.

[36] 3DVisualizer was also used in the KeckCAVES for inspection and exploration of the composite viscosity field, velocity field, and strain rate output from CitcomCU for each viscous flow simulation. Interactive exploration of the model output in a 3D immersive environment allowed for on-the-fly query of streamlines, velocity vectors, and of isosurfaces of scalar variables such as velocity magnitude, viscosity, and pressure, allowing for real-time assessment of hypotheses. Using the interactive 3D virtual reality facility enabled the massive amounts of data (several Gigabytes per model run) to be conceptualized and allowed for identification of key areas to target for more quantitive analysis with MATLAB or other tools like GMT. To preserve the choices and features that were explored and identified, numerous sessions were recorded with a playback feature making the sessions portable to a movie format. 3D model results were also visualized with the desktop version of 3DVisualizer.

4. Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Alaskan Subduction-Transform Boundary
  5. 3. Methodology
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[37] We present results based on eighteen 3D models in which we varied the subducting plate geometry, upper mantle viscosity structure (Newtonian versus composite viscosity), yield stress, plate boundary coupling, and the overriding plate lithospheric thickness (Table 1). When describing the models, we refer to the viscosity structure as either Newtonian or composite. Let the reader note that, as indicated in section 3, both of these formulations include a depth-dependent yield stress.

4.1. 3D Mantle Flow and Subducting Plate Geometry

[38] The incorporation of realistic slab geometries into 3D models of subduction leads to a complex pattern of poloidal and toroidal flow in the mantle, with the spatial positions of the mantle flow components sensitive to the slab geometry. Figures 4a–4f illustrate this complex flow field for three representative models: Model A1c with slabE325 and the composite rheology (Figures 4a and 4d), Model B1c with slabE115 and the composite rheology (Figures 4b and 4e), and Model B1n with slabE115 and the Newtonian rheology (Figures 4c and 4f). These three models use a depth dependent yield stress of 500 MPa and a weak zone viscosity of 1020 Pa s (Table 1).

image

Figure 4. Flow field, viscosity and strain rate for models using (left) slab325 with composite viscosity, (middle) slab115 with composite viscosity, and (right) slabE115 with Newtonian viscosity. Viscosity isosurface and velocity slices for (a, d) model A1c, (b, e) model B1c, and (c, f) model B1n. Cross section of viscosity and second invariant of the strain rate tensor for (g, j) model A1c, (h, k) model B1c, and (i, l) model B1n. Profiles at 64.5°N (location of vertical line in Figures 4g–4l) for (m) models A1c-A4c, (n) models B1c-B5c, and (o) models B1n-B5n. Figures show subset of model domain.

Download figure to PowerPoint

[39] Figures 4a–4c and Figures 5a–5c show counterclockwise toroidal flow in the mantle, where the slab edge provides an opening for the return flow from beneath the slab, around the slab edge, and into the mantle wedge. Here, the low viscosity mantle flows around the higher viscosity slab. In models with slabE325, the toroidal flow is centered around 222° longitude (Figures 4a and 5a), whereas, in models with slabE115, the toroidal flow is centered at approximately 212° longitude (Figures 4b, 4c, 5b, and 5c). This occurs because in models with the slabE115, the depth of the slab east of 212° is comparable to the depth of the base of the overriding lithosphere, and thus is too short to induce toroidal flow in the mantle east of 212° (Figures 1b, 2h, and 4b). Thus, one of the effects of slabE115 versus slabE325 is to shift the locus of the toroidal cell approximately 10° westward so that it is focused around the most northern part of the slab. This leads to a stronger component of westward directed trench parallel flow beneath the magmatic gap in south central Alaska for models with slabE115.

image

Figure 5. Horizontal and vertical velocity at 100 km depth. Horizontal velocity arrows show position of toroidal flow for (a) model A1c using slab325 with composite viscosity, (b) model B1c using slab115 with composite viscosity, and (c) model B1n using slabE115 with Newtonian viscosity. Vertical velocity along cross section BB′ at 100 km for (d) models A1c-A4c, (e) models B1c-B5c, and (f) models B1n-B5n. Shown here is the extraction at 100 km depth, but note the magnitude of the vertical upwelling as well as horizontal velocity varies with depth in the 3D model. Figures show subset of model domain.

Download figure to PowerPoint

[40] Figures 4d–4f show the poloidal component of flow, where the sinking of the slab draws material into the mantle wedge toward the slab. Beneath the Pacific plate, a broader region of poloidal flow is induced in the mantle where the surface part of the Pacific plate is pulled into the subduction zone. Cross sections AA′ through the poloidal flow in the mantle wedge show velocity vectors that are steeper than the slab dip (Figures 4d–4f), indicating steepening of the slab dip with time, or retrograde slab motion [Garfunkel et al., 1986] if the models were to run forward in time. However, in cross section BB′ for models with slabE115, there is essentially no poloidal flow component in the mantle wedge above the Wrangell slab (Figures 4e and 4f). This is because slabE115 is too short in this region of the model to induce the poloidal motion of mantle material.

[41] The toroidal and poloidal flow in the mantle around the subduction zone are not independent. Figures 4a4c and Figures 5a5c show there is motion along strike associated with the poloidal flow and motion in the vertical direction associated with the toroidal flow. For example, in all models, the poloidal flow in cross section AA′ contains a westward trench parallel velocity component in the mantle wedge, suggesting the flow of material away from the slab nose toward the arcuate central Aleutian subduction zone (Figures 4d4f). The trench parallel flow component around the northernmost part of the slab is stronger in models with slabE115, because the center of the toroidal flow is more westward, i.e., at 212° rather than 222° longitude.

[42] In addition, in all models, the toroidal flow around the slab edge, whether it is centered at 212° or 222° longitude, is associated with an upward component of flow (Figures 4a–4c and Figure 5a–5c). This upward flow component occurs where the toroidal flow emerges from beneath the slab, i.e., on the eastern side of the counterclockwise flow pattern in Alaska. The magnitude of the upward flow varies depending on slab shape, slab strength, and mantle rheology, and can have values greater than 10 cm/yr (Figures 5a–5c). Cross section BB′ shows the upward component of flow at 100 km depth for models that use slabE325 with the composite viscosity (Figure 5d), slabE115 with the composite viscosity (Figure 5e), and slabE115 with the Newtonian viscosity (Figure 5f). Note the upward component of flow associated with the slab edge forms a 3D lenticular feature, and thus its magnitude varies laterally as well as with depth, with the maximum vertical component occurring at approximately 200 km depth.

[43] The upward component of mantle flow associated with the toroidal flow at the slab edge combines with the upward component of the poloidal flow in the mantle wedge above the slab. The vertical motion of mantle material near slab edges may play a role in forming melts from both decompression melting and melting of the slab edge (i.e., adakites), as this part of the subduction zone is continuously exposed to upward advected warm mantle material. The implications for the adakitic Wrangell Mountains in eastern Alaska, which are situated at approximately 216° longitude, are discussed in section 5.4.

4.2. 3D Variable Viscosity and Mantle Flow

[44] Cross sections through the viscosity structure and strain rate field are shown for the three representative models: Model A1c (Figures 4g and 4j), Model B1c (Figures 4h and 4k), and Model B1n (Figures 4i and 4l). Figure 4 also shows vertical profiles through the viscosity and strain rate for all models using slabE325 and the composite viscosity (Figure 4m), slabE115 and the composite viscosity (Figure 4n), and slabE115 and the Newtonian viscosity (Figure 4o).

[45] Models using the composite viscosity formulation have higher strain rates and lower mantle viscosities than models using the Newtonian viscosity. In these models using the composite viscosity, a laterally variable viscosity emerges in the mantle as a consequence of the lateral variations in the flow (Figures 4g, 4h, 4j, 4k, 4m, and 4n). Viscosities in the mantle wedge can be lower than 1018 Pa s, where strain rates are high, on the order of 10−12 s−1. The weakest viscosities in the mantle wedge occur as elliptical regions immediately beneath the overriding plate and immediately above the slab, again correlating to where strain rates are high ( Figures 4g, 4h, 4j, 4k, 4m, and 4n). Note that although shown in cross section only, these low viscosity (high strain rate) regions form three-dimensional ellipsoids in the mantle wedge that follow the strike of the slab.

[46] For models using slabE115 and the composite viscosity, the high strain rate and low viscosity region in the mantle wedge along cross section AA′ is broader than for models using slabE325 (Figure 4h versus Figure 4g). This is because the velocity gradients and hence strain rates are higher where both toroidal and polodial flow occur leading to the lower viscosity, as shown in cross section AA′ for model B1c. Models that use the composite viscosity and either slab shape also have high strain rates and low viscosity beneath and subparallel to the downgoing Pacific plate and slab (Figures 4g, 4h, 4j, and 4k).

[47] In the Newtonian models, viscosities in the mantle wedge are higher (1019 to 1020 Pa s) and strain rates are lower (10−13 to 10−14 s−1). In these models, the mantle viscosity is laterally uniform, except for temperature-dependent variations, because there is no power law relation between the stress and strain rate. Therefore, the viscosity is not reduced in regions of high strain rate as was the case for models using the composite viscosity (Figure 4).

[48] The weakening effect of the strain rate dependent viscosity in models using the composite viscosity formulation leads to viscosity contrasts that can be up to seven orders of magnitude between the mantle and lithosphere. However, away from the driving force of the slab, strain rates decrease, and mantle viscosities increase to on the order of 1020 Pa s. The localized regions of low viscosity and high strain rate around the slab and beneath the plates significantly reduces the coupling between the lithosphere and the mantle near the subduction zone.

[49] It is the magnitude of the driving forces that ultimately limits the amount of weakening that occurs in the models with the composite viscosity formulation, with the greater amount of weakening in the mantle close to where the sinking slab is driving the flow. In this way, using this viscosity formulation redistributes how the stresses are accommodated in the model by weakening in some regions (the mantle) while strengthening in others (the lithosphere). In other words, the rheologic flow law is an additional constraint on the equations of motion that determines the viscosity distribution, but that viscosity distribution must still satisfy the driving forces and boundary conditions.

4.3. Localized Fast Velocities, Mantle Viscosity, and Slab Strength

[50] The 3D models of the Alaska subduction-transform boundary system predict spatially variable velocity magnitudes in the mantle, with localized fast velocities close to the slab where the negative buoyancy of the slab drives the flow (Figures 4, 5, and 6). The velocities decrease with increasing distance from the driving force of the slab (Figures 6a–6f). The rate of mantle flow, which can be over 80 cm/yr, depends on several factors, such as the slab buoyancy, slab strength, and the viscous support of the mantle surrounding the slab.

image

Figure 6. Horizontal velocity and strain rate at 100 km depth. (a–c) Percent of maximum horizontal velocity for models A1c, B1c, and B1n respectively. The inset in these figures shows the maximum absolute horizontal velocity magnitude, which can be greater than 50 cm/yr. Thin black lines are contours of the percent of maximum horizontal velocity in 20% intervals. Gray line is 1000°C isotherm of the slab. Absolute horizontal velocity and second invariant of the strain rate tensor along cross section AA′ for (d, g) model A1c, (e, h) model B1c, and (f, i) model B1n. The fastest velocities in the mantle wedge occur within 500 km from slab, decreasing with increasing distance from the slab. Figures show subset of model domain.

Download figure to PowerPoint

[51] Figures 6a–6c show the percent of the maximum horizontal velocity magnitude at 100 km depth and show the differential motion of the mantle produced by subduction of the Pacific plate beneath Alaska, with the fast flow velocities occurring as an arcuate region, encompassing both the mantle wedge and the mantle around the slab edge. The velocities decrease to approximately 10% of the maximum value by approximately 500 km from the slab surface (Figures 6d–6f). Note these plots of the horizontal velocity magnitude also illustrate the conversion vertical sinking motion of the slab into localized regions of horizontal flow in the surrounding mantle.

[52] In models with slabE325 and the composite viscosity, the high velocity region extends eastward to approximately 225° longitude, just east of the slab edge. Whereas, in models using slabE115, the region of fast velocities only extends to approximately 215° longitude. This is because for models using slabE115, east of 212° longitude the slab extends only to approximately 115 km depth which is too short with respect to the thickness of the overriding plate to generate any significant mantle flow.

[53] Spatially variable motion in the mantle occurs for models using the Newtonian viscosity and for models using the composite viscosity. However, models using the composite viscosity have larger ranges in the velocity magnitudes as well as sharper velocity gradients (Figures 5 and 6). The localized weakening of the mantle viscosity around the slab in models using the composite viscosity reduces the viscous resistance in the mantle to the sinking of the slab, leading to faster mantle flow velocities (Figure 6), as seen by Billen and Hirth [2005], Jadamec and Billen [2010], and Stadler et al. [2010]. For example, mantle wedge velocities in models using the composite viscosity can be greater than ten times that in models using the Newtonian viscosity (66.8 cm/yr versus 6.5 cm/yr in maximum horizontal velocity magnitude for models B1c versus B1n at 100 km depth) (Figure 6). This implies greater decoupling of mantle flow from surface plate motions for models with the composite viscosity. Note that for both types of upper mantle rheology, surface plate motions are less than 6 cm/yr (see section 4.4).

[54] Figures 46 illustrate how the velocities are sensitive to the rheology, including both to the slab strength and the surrounding mantle viscosity. For models that use the composite rheology, high strain rates located where the slab bends into the mantle, lead to viscous yielding within the slab and deformation by dislocation creep on the underside of the slab, resulting in a narrower and weaker slab core (Figures 4g, 4h, 4j, and 4k). For example, whereas the slab core remains on the order of 1024 Pa s in models with the Newtonian viscosity (Figure 4i), in models with the composite rheology, the viscosity of the slab is reduced to on the order of 1022 Pa s in the slab hinge (Figures 4g and 4h). As a result, less weight of the slab can be supported by the slab strength, allowing for the slab to pivot faster and generate faster velocities in the mantle (Figures 46).

[55] When varying the yield strength, models with a weaker slab (yield strength of 500 MPa) generate mantle velocities that are almost twice as fast as the equivalent models with a stronger slab (yield strength of 1000 MPa) (black lines versus gray lines in Figures 6d–6f). Again, models with the lower yield strength allow less of the weight of the slab to be supported by the slab strength, thus leading to a slab that is sinking faster and thereby generates more vigorous mantle flow.

[56] Mantle wedge velocities also vary depending on whether toroidal and poloidal flow are both occurring. For example, where both poloidal and toroidal flow are acting together flow velocities in the mantle are stronger (Figures 5 and 6). This can be seen for example, in comparison of Model B1c (slabE115) and A1c (slabE325) along cross sections AA′ in Figure 6. The velocities are higher for Model B1c in which the flow from the toroidal and poloidal cells combine in the mantle wedge.

[57] In addition, models that use the uniform overriding plate thermal structure and the composite viscosity have faster mantle velocities, because the overriding plate is relatively thinner and thus the shallow mantle is warmer, allowing for lower viscosities and less resistance to flow (Figure 6e, models B2 versus B5).

4.4. Predicted Surface Motions

[58] The predicted Pacific plate velocity vectors offshore south central Alaska are shown in Figure 7. For models that use slabE325 and the Newtonian viscosity, the predicted velocity of the Pacific plate ranges from 327.3° at 0.74 cm/yr to 338.9° at 2.11 cm/yr (Figure 7a). For models that use slabE115 and the Newtonian viscosity, the predicted plate motion ranges from 318.9° at 0.68 cm/yr to 332.6° at 1.92 cm/yr (Figure 7b). Thus, models that use slabE325 have a more northerly and faster Pacific plate velocity than the same models using slabE115. This is because models using slabE325 have more slab pull in the eastern part of the slab where it extends to325 km beneath the Wrangell Mountains (Figure 1). This pattern is also seen in models using the composite viscosity, where the predicted Pacific plate velocity ranges from 343° at 2.47 cm/yr to 346° at 5.28 cm/yr for models using slabE325 (Figure 7c) and from 336° at 2.02 cm/yr to 340° at 4.4 cm/yr for models using slabE115 (Figure 7d).

image

Figure 7. Predicted Pacific plate motion vectors offshore south central Alaska for Newtonian models using (a) slabE325 and (b) slabE115 and for composite viscosity models using (c) slabE325 and (d) slabE115. Observed NUVEL–1A Pacific plate motion vector, assuming North America fixed, shown with thick gray solid line [DeMets and Dixon, 1999].

Download figure to PowerPoint

[59] Comparison of models using the composite viscosity with those using the Newtonian viscosity show that, on average, models using the composite viscosity move in a more northerly direction and about 2.5 times faster (Figures 7a and 7b versus Figures 7c and 7d). This is because the strain rate dependence in models using the composite viscosity leads to a viscosity reduction in the mantle surrounding the slab, reducing the viscous support of the slab by the mantle. In addition, high strain rates along the plate boundary in models using the composite viscosity lead to a lower viscosity and broader plate boundary shear zone, allowing less resistance along the plate boundary interface and consequently faster surface plate motions. Imposing a higher viscosity of 1021 Pa s versus 1020 Pa s along the plate interface decreases the surface plate motion by approximately a factor of two, regardless of whether the composite or Newtonian viscosity is used (Figure 7, dashed versus solid lines). There is little difference in predicted plate motion for models that use the uniform overriding plate thermal structure versus those that use the variable overriding plate thermal structure (Figure 7, models B2 versus B5).

[60] Comparison of the predicted Pacific plate velocity to the observed NUVEL–1A Pacific plate velocity offshore Alaska [DeMets and Dixon, 1999] is shown in Figure 7. There is good agreement between the predicted and observed Pacific plate motion offshore south central Alaska for models using the composite viscosity and plate boundary viscosity of 1020 Pa s, regardless of slab shape (Figure 7). However, before considering how the model parameters affect the match between the observed and predicted surface plate motion, it is important to realize that the entire Pacific plate is not included in the models because doing so would exceed the computational resources available. Instead, the model domain was designed such that the bounds of the Pacific plate included in the models are approximately representative of those of the actual Pacific plate, that is, a subduction zone to the north (Aleutian-Alaska trench), transform and spreading center to the east (Queen Charlotte-Fairweather transform and Juan de Fuca ridge), and a spreading center to the south (vertical weak zone at the southern boundary approximates the orientation of the east-west trending Antarctic spreading center). The main exception is the western boundary, where the Pacific plate has westward dipping subduction zones that are not included in the models.

[61] In order for a plate to remain rigid and move in response to slab-pull and ridge-push forces on multiple boundaries, the orientations of the individual plate boundaries must be roughly oriented such that the local driving force is mostly consistent with the overall plate motion. In the case of southern Alaska, plate reconstructions indicate this plate boundary configuration has been stationary for the last 40 Myr [Sdrolias and Muller, 2006] implying that the local driving forces and orientation of the plate boundary are dynamically consistent with the overall Pacific plate motion. Therefore, if the other model parameters are appropriate (e.g., mantle rheology), then the section of the Pacific plate and Aleutian slab included in our models should result in plate motions that are close to Pacific plate motion.

[62] Thus, although we do not expect the models to achieve an exact match with observed Pacific plate velocities, that there is a good fit in the same models that produce rapid flow in the mantle indicates that rapid mantle flow is not incompatible with surface plate motions. In other words, the same 3D geodynamic models of the Alaska subduction-transform system that predict velocity magnitudes of greater than 80 cm/yr in the mantle close to the subduction zone, also predict surface plate velocities comparable to observed plate motions.

5. Discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Alaskan Subduction-Transform Boundary
  5. 3. Methodology
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[63] The results from the 3D models of the Alaska subduction-transform system demonstrate the ranges in mantle velocities that can be produced assuming an experimentally determined flow law for olivine and a realistic slab geometry and density structure. The results show larger velocity ranges as well as sharper velocity gradients in models using the composite viscosity formulation. The maximum mantle velocity predicted by the 3D models of the Alaska subduction-transform system ranges from less than 10 cm/yr in the Newtonian models to over 80 cm/yr in the composite viscosity models. The maximum velocity that occurs in the mantle wedge, around the slab edge, and underneath the slab depends on several factors: the (1) slab density anomaly, i.e., the driving force; (2) slab strength, which depends in part on the yield strength; (3) mantle viscosity; (4) plate boundary coupling, i.e., viscosity along the plate interface; and (5) the proximity of the mantle to the slab edge.

5.1. Spatially Variable Slab and Mantle Viscosity

[64] The consequence of using an experimentally derived Newtonian and composite viscosity formulation in the 3D numerical models of the Alaska subduction-transform system is that the slab strength depends on the thermal structure, and on the strain rate for the composite viscosity formulation [Billen and Hirth, 2007; Jadamec and Billen, 2010]. Therefore, a strong slab core emerges from the rheological flow law [Jadamec and Billen, 2010]. The depth dependent yield stress places a limit on the maximum strength of the slab determined from the experimentally derived flow law.

[65] In models with the composite rheology, the slab hinge area becomes locally weakened, with viscosities reduced from 1024 Pa s to on the order of 1022 Pa s (Figures 4g and 4h). This gives a viscosity difference of three to four orders of magnitude between the slab hinge and the mantle surrounding the slab, and of up to six orders of magnitude between the stronger parts of the slab and the weakest part of the mantle wedge (Figures 4g and 4h). In models using the Newtonian viscosity, the slab is not weakened in the slab hinge and maintains a viscosity on the order of 1024 Pa s in the core, with a viscosity difference between the slab and mantle on the order of four orders of magnitude (Figure 4i). The viscosity ratios in our models are in the upper range of those suggested from previous 2D and 3D models of subduction, where the viscosity contrasts between the slab and the mantle range from two to six orders of magnitude [Piromallo et al., 2006; Stegman et al., 2006; Royden and Husson, 2006; Billen and Hirth, 2007; Schellart et al., 2007; Kneller and van Keken, 2008; Schellart, 2008; Giuseppe et al., 2008; Jadamec and Billen, 2010; Ribe, 2010].

[66] The models of the Alaska subduction-transform subduction zone that use the composite viscosity formulation exhibit a localized lower viscosity in the mantle surrounding the slab, and therefore locally reduced viscous support of the slab. This local reduction of the mantle viscosity does not occur in Newtonian models, such as those of Piromallo et al. [2006], Stegman et al. [2006], and Schellart et al. [2007]. Thus, although 3D Newtonian models of short slabs undergoing slab rollback or slab steepening do show faster velocities close to the slab [Kincaid and Griffiths, 2003; Schellart, 2004; Funiciello et al., 2006; Piromallo et al., 2006; Funiciello et al., 2006; Stegman et al., 2006; Schellart et al., 2007], the maximum velocities and the velocity gradients are larger for models using the composite rheology because of the local reduction in mantle viscosity [Jadamec and Billen, 2010; Stadler et al., 2010]. This is consistent with a non-Newtonian viscosity behavior where a power law dependence between stress and strain rate is characterized by larger velocity gradients than in a Newtonian viscosity [Turcotte and Schubert, 2002].

[67] In order to match surface observables, several previous models have called upon mantle velocities in the mantle wedge and back-arc that are faster than plate motions, although the mechanism for the lower viscosities is typically attributed to the presence of water or to higher temperatures, rather than from the weakening effects of a strain rate dependent viscosity [Hyndman et al., 2005; Currie and Hyndman, 2006]. Although we use a thermal model with a warm back-arc lithosphere, as do Blackwell [1969] and Hyndman et al. [2005], the fast velocities in the mantle predicted by our models are not a result of the warm, thin lithosphere imposed for our overriding plate, because the thermal structure is the same for the instantaneous Newtonian and composite viscosity models. The low viscosity region in the mantle wedge that emerges from the strain rate dependent viscosity suggests that in some subduction zones, additional water or heat [Billen and Gurnis, 2001; Hyndman et al., 2005; Currie and Hyndman, 2006] may not be required to reduce the viscosity and that the strain rate dependent rheology may be as important of a contributor to reducing the viscosity in the mantle wedge.

5.2. Short Slabs and Fast Velocities

[68] The deepest part of the slab in the 3D models of the Alaska subduction-transform system extends to approximately 325 km based on seismic constraints (section 3.2). Therefore, although the model depth extends to 1500 km, and the models include an increase in mantle viscosity for the lower mantle, the slab is not supported by the lower mantle. Thus, the slab in our model is considered a short slab, and how the model parameters affect the slab dynamics should be considered in the context of that for short slabs in this early phase of subduction, as identified by Schellart [2004], Funiciello et al. [2006], Piromallo et al. [2006], Billen and Hirth [2007], Giuseppe et al. [2008], and Jadamec and Billen [2010]. Within this transient state, as a slab descends into the mantle it does not maintain its initial dip indefinitely, rather the slab will steepen due to the slab pull force [Spence, 1987; Conrad and Hager, 1999; Lallemand et al., 2008; Ribe, 2010]. This pivoting can be accentuated by the presence of a slab edge, where return flow can increase the rate of steepening, as indicated by the occurrence of steeper slab dips near slab edges [Lallemand et al., 2005; Schellart et al., 2007].

[69] Viscous yielding and deformation by dislocation creep which weaken the slab hinge allowing the slab to bend more easily also reduce the amount of slab pull that can be directly transmitted to the subducting plate. This effect is included in all the models and contributes to why the surface part of the subducting plate moves slower than the slab is sinking. Some amount of yielding is required to allow the slab to pivot and sink into the mantle, but the reduction in slab strength in the hinge limits the amount of slab pull force that can be directly transmitted to the surface plate.

[70] We expect that it is slabs in the transient state, not supported by the higher viscosity mantle below the transition zone, that are more likely to induce localized rapid mantle flow. Thus, other subduction zones with short slabs where rapid mantle flow may be expected include central America, Cascadia, the lesser Antilles, Scotia, and Vanuatu. Seismic and geochemical studies suggest this may be the case beneath Costa Rica-Nicaragua. In this region, local S wave and teleseismic SK(K)S wave anisotropy measurements contain trench parallel fast axes in the shallow and deeper parts of the mantle wedge, indicative of trench parallel flow [Abt et al., 2009, 2010]. Along the volcanic front there is a systematic northwestward increase in 143/144Nd and decrease in 208/204Pb suggesting northwestward flow of mantle [Hoernle et al., 2008]. Using the isotopic ages and with assumptions about the transport distance, velocity bounds of 6.3–19.0 cm/yr have be placed on the trench parallel component of mantle flow in the mantle wedge [Hoernle et al., 2008], which is either slightly slower than or more than twice as fast as the surface motion vector for the Cocos plate. We also point out that localized rapid flow may not be limited to subduction zones with slab edges, as previous studies that include a non-Newtonian rheology predict along strike flow of up to 50 cm/yr in a low viscosity channel within the mantle wedge [Conder and Wiens, 2007] as well as mantle plume ascent velocities on the order of 100 cm/yr [Larsen et al., 1999].

5.3. Seismic Anisotropy and Plate-Mantle Decoupling

[71] Beneath south-central Alaska teleseismic shear wave splitting observations show east-west and northeast-southwest trending (trench-parallel) fast directions in the mantle wedge north of the slab [Christensen and Abers, 2010]. The seismic fast directions rotate to northwest-southeast (trench-perpendicular) to the south of the 100 km Aleutian slab contour beneath south central Alaska and are northeast-southwest (trench perpendicular) to the southeast of the slab nose [Christensen and Abers, 2010]. In order to use the seismic anisotropy observations of Christensen and Abers [2010] as a constraint on the mantle flow field predicted by the 3D Alaska subduction-transform models, Jadamec and Billen [2010] calculated infinite strain axes (ISAs) from the predicted mantle flow field, according to the method of Conrad et al. [2007], and compared the ISAs to the observations of seismic anisotropy.

[72] Jadamec and Billen [2010] found that models using slabE115 and the composite rheology provided the best fit between the predicted ISA orientations and the seismic fast directions from the observed anisotropy. None of the models with slabE325, where the toroidal flow is located further to the east, were able to match the observed pattern of anisotropy. Models with slabE115, but that used the Newtonian viscosity, displayed less vigorous flow including a smaller toroidal flow component and lead to a poor fit to the anisotropy. Thus, the pattern of observed seismic anisotropy in this region of south central Alaska is consistent with rapid toroidal flow around the slabE115 edge. This implies decoupling of the mantle from surface plate motion in terms of both direction and speed within the Alaska subduction-transform system.

[73] Shear wave splitting studies that probe the character of the mantle in many other subduction zones find seismic fast directions that are nonparallel to the direction of subducting plate motion [Russo and Silver, 1994; Fischer et al., 1998; Hall et al., 2000; Smith et al., 2001; Kneller and van Keken, 2007; Pozgay et al., 2007; Hoernle et al., 2008; Long and Silver, 2008; Abt et al., 2010]. In addition, shear wave splitting measurements from the subduction-transform junctures in northern Kamchatka and southern Cascadia also display a curved pattern of the seismic fast directions, suggestive of toroidal flow in the uppermost mantle around a slab edge [Peyton et al., 2001; Zandt and Humphreys, 2008]. This suggests that partial decoupling of the surface plate motion from the underlying mantle flow field, at least in terms of direction, may be common in subduction zones. Moreover, in some subduction zones the mantle may flow at rates significantly faster than the surface plate motion [Conder and Wiens, 2007; Hoernle et al., 2008; Jadamec and Billen, 2010; Stadler et al., 2010], implying decoupling in speed as well.

[74] A power law rheology, i.e., one that includes the effects of the dislocation creep deformation mechanism, can explain both observations of seismic anisotropy and the decoupling of mantle flow from surface motion [Jadamec and Billen, 2010]. Dislocation creep of olivine in the upper mantle leads to lattice preferred orientation (LPO) in olivine which results in anisotropy in seismic waves [Savage, 1999; Kaminiski and Ribe, 2002; Karato et al., 2008; Long and Silver, 2009]. In addition, dislocation creep of olivine leads to a lowering of the viscosity in regions of high strain rate in the mantle, which in turn facilitates the decoupling of the mantle from the surface plates [Billen and Hirth, 2005; Jadamec and Billen, 2010]. This is because the power law relationship between stress and strain rate for deformation by dislocation creep leads to a faster rate of deformation (higher strain rates) for a given stress, which can be expressed as a reduction in the effective viscosity [Hirth and Kohlstedt, 2003; Karato et al., 2008]. This reduction can occur in regions of variable velocity, or high strain rate, such as in the mantle wedge and near a slab edge [Billen and Hirth, 2005, 2007; Jadamec and Billen, 2010].

5.4. Implications for Volcanics Above a Slab Edge

[75] In the 3D models of the Alaska subduction-transform system, where the slab edge is deep enough to induce toroidal flow, the models indicate there is associated mantle upwelling within approximately 500 km outward of the slab edge. This is consistent with mantle flow patterns predicted by other 3D subduction models where there is a slab edge and slab steepening [Funiciello et al., 2006; Piromallo et al., 2006; Stegman et al., 2006; Jadamec and Billen, 2010; Schellart, 2010b]. Thus, in these tectonic environments, warm mantle is expected to be transported from underneath the slab into the mantle wedge in an upward and arcuate pattern around the edge of the slab, which could lead to decompression melting within several hundred kilometers outward of the slab edge as well as contribute to melting of the slab edge [Yogodzinski et al., 2001]. In addition, because toroidal flow is associated with rapid mantle velocities [Kincaid and Griffiths, 2003], which are significantly faster for a composite viscosity [Jadamec and Billen, 2010; Stadler et al., 2010], this may contribute to the preservation of primitive magmas that can be brought to the surface [Durance-Sie, 2009; McLean, 2010; P. M. J. Durance and M. A. Jadamec, Magmagenesis within the Hunter Ridge Rift Zone resolved from olivine-hosted melt inclusions and geochemical modelling with insights from geodynamic models, submitted to Australian Journal of Earth Sciences, 2012].

[76] Adakitic volcanics have been identified at several subduction-transform plate boundaries, including the eastern Alaska subduction-transform boundary, the Kamchatka-Aleutian plate boundary corner, the Cascadia-San Andreas fault juncture, in southern New Zealand where there is a subduction-transform transition, and in the New Hebrides trench-Hunter fracture zone region [Skulski et al., 1991; Peyton et al., 2001; Yogodzinski et al., 2001; Durance-Sie, 2009]. However, the position of the volcanics with respect to the slab edge and upwelling in the mantle flow field has only recently been tested in 3D geodynamic models [Jadamec and Billen, 2010; Schellart, 2010b; McLean, 2010; Durance and Jadamec, submited manuscript, 2012].

[77] The Wrangell volcanics in Alaska, characterized by adakitic geochemical signatures indicative of melting of the slab edge, are located east of and physically separate from the Aleutian-Alaska magmatic arc (Figure 2) [Skulski et al., 1991; Preece and Hart, 2004]. In our 3D models that use slab115, the Wrangells are located above the upward stream in the counterclockwise quasi-toroidal flow associated with the deeper slab edge. In models using slabE325, the Wrangells are located above the center line of the return flow from the slab edge. We suggest the Wrangell volcanics in Alaska may be due in part to the upwelling associated with the deeper edge of slabE115 at approximately 212° to 215° longitude. These models do not investigate the link between melt migration and solid state flow of the mantle, which is an important and complex process especially in 3D, but beyond the scope of this paper.

5.5. Slab and Trench Geometry in Alaska

[78] The toroidal flow of the mantle around the eastern Alaska slab edge and steepening of the slab dip predicted by the 3D models of the Alaska subduction-transform system are suggestive of slab rollback [Garfunkel et al., 1986; Kincaid and Griffiths, 2003; Schellart, 2004; Funiciello et al., 2006; Stegman et al., 2006] and thereby retreat of the trench in eastern Alaska. 3D laboratory experiments of free subduction indicate that during trench retreat, the trench geometry evolves into an arcuate shape with the concave side directed away from subducting plate [Schellart, 2004; Funiciello et al., 2006; Schellart, 2010a]. Along the greater than 3000 km length of the Aleutian-Alaska subduction zone, the sense of curvature of the plate boundary changes concavity [Page et al., 1989; Gudmundsson and Sambridge, 1998; Ratchkovski and Hansen, 2002; Bird, 2003]. Although in the central Aleutians the trench is concave to the north, i.e., away from the subducting plate consistent with that expected for trench retreat, the trench in eastern Alaska is concave to the south, i.e., toward the subducting plate inconsistent with that expected for trench retreat.

[79] A comparison of global plate motion models [Schellart et al., 2008] indicates that although the trench in the central Aleutians may be retreating, the trench in eastern Alaska shows little to no trench retreat, despite that it terminates into the Queen Charlotte-Fairweather transform fault forming a slab edge. In addition, plate reconstructions indicate this portion of the plate boundary has been stationary for at least 10 Myr [Sdrolias and Muller, 2006]. These observations are unexpected, as along many plate boundaries the rate of trench retreat tends to be greatest near a slab edge, and 3D numerical models indicate that the magnitude of trench retreat along a subduction zone is greatest closest to a slab edge [Stegman et al., 2006; Schellart et al., 2007, 2008]. Therefore, other features of the subduction zone may control the shape of the trench in eastern Alaska or the subduction zone may be in a transient state.

[80] One such feature is the Yakutat terrane, the oceanic plateau that is actively colliding with North America in the plate boundary corner of southern Alaska [Lahr and Plafker, 1980; Bruns, 1983; Fletcher and Freymueller, 1999; Pavlis et al., 2004; Meigs et al., 2008] and may be subducted to a depth of 150 km [Ferris et al., 2003]. 3D laboratory and numerical models of the subduction of an oceanic plateau oriented at a high angle to a trench indicate that the shape of the trench becomes arcuate with the concave side toward the subducting plate [Martinod et al., 2005; Mason et al., 2010]. This is opposite to that predicted for slab rollback and consistent with the trench shape in eastern Alaska. At depth, these 3D laboratory and numerical models predict that after a plateau is subducted, the uppermost part of the slab shallows in dip becoming a flat slab in the vicinity of the oceanic plateau with the flanks of the slab dipping more steeply and away from the central shallow slab [Martinod et al., 2005; Mason et al., 2010]. This geometry is consistent with that observed in the subducted plate beneath south central Alaska,where the flat slab is flanked by the more steeply dipping Aleutian portion of the slab to the west and the steeply dipping, but shorter, Wrangell slab to the east [Stephens et al., 1984; Page et al., 1989; Gudmundsson and Sambridge, 1998; Ratchkovski and Hansen, 2002; Fuis et al., 2008]. Thus, the toroidal flow around the slab edge in south-central Alaska may not lead to the formation of a retreating trench, because the positive buoyancy of the Yakutat terrane may prevent sinking of this upper portion of the slab and thus inhibit trench retreat. We point out that, although the geometry of the subducting plate in our models is based on Wadati-Benioff zone seismicity, which we expect is due in part to the subducted Yakutat terrane, the models do not include an additional compositional buoyancy force representative of this plateau.

[81] Alaska has long-lived subduction history [Wallace and Engebretson, 1984; Lonsdale, 1988; Madsen et al., 2006; Sdrolias and Muller, 2006; Qi et al., 2007] suggesting that a slab could be present to the 660 km discontinuity. However, the seismicity and seismic tomography indicate that, although the slab beneath the Aleutians may reach the transition zone, the slab beneath south central Alaska likely does not extend deeper than 300 km, and that the slab beneath easternmost Alaska is likely even shorter (see Table 4 and references therein). The lack of a slab continuous to the transition zone beneath south central Alaska could result from previous detachment of the deeper part of the subducted plate due to the subduction of the Yakutat terrane or due to a weakness in the subducting plate at depth because of subduction of the failed Kula-Farallon spreading center [Madsen et al., 2006; Qi et al., 2007]. Investigating the potential detachment of a slab with depth over the course of the subduction history of southern Alaska is beyond the scope of this paper, but would be interesting to investigate using future time-dependent modeling studies and would likely be more tractable with future codes using adaptive mesh refinement [e.g., Stadler et al., 2010].

Table 4. Seismic Constraints on Slab Deptha
StudyData Typedaleut (km)dakp (km)dscak (km)dwr (km)
  • a

    Abbreviations, daleut, dakp, dscak, and dwr, refer to the depth of the subducting plate beneath the Aleutians, the Alaska Peninsula, south central Alaska, and the Wrangell Mountains. Data suggest a decrease in slab depth from west to east. Note, there is a range in depth resolution for studies listed in the table.

Qi et al. [2007]3D teleseismic tomography>400300–40090
Eberhart-Phillips et al. [2006]3D tomography>200160–18050–60
Zhao et al. [1995]3D tomography>190165>90
Kissling and Lahr [1991]3D tomography120–150
Fuis et al. [2008]Seismic reflection, refraction earthquake hypocentral locations240175100
Ferris et al. [2003]Teleseismic receiver function analysis150
Ratchkovski and Hansen [2002]Earthquake hypocentral locations210165–185
Page et al. [1989]Seismic reflection, refraction earthquake hypocentral locations>100
Engdahl and Gubbins [1987]Simultaneous time travel inversion400
Stephens et al. [1984]Earthquake hypocentral locations85
Boyd and Creager [1991]Local seismicity and teleseismic residual sphere analysis600
Gudmundsson and Sambridge [1998]RUM seismic model300250150

[82] The seismicity, seismic tomography, and the agreement between the models using slabE115 and the seismic anisotropy from the northern mantle wedge, suggest slabE115 is representative of the actual slab shape beneath mainland Alaska, and that slabE325 is not (Table 4 and Jadamec and Billen [2010]). The numerical models show that the relative depth of the slab tip and base of the overriding lithosphere influences whether there will be flow induced in the mantle. Thus, for models using slabE115, although this slab shape has two slab edges, one at approximately 212° longitude (beneath south central Alaska) and the other at 223° longitude (near the northern end of the Queen Charlotte-Fairweather fault), only the deeper slab edge at 212° induces toroidal flow. The lack of induced flow near the northern end of the Queen Charlotte-Fairweather fault implies that there is little strain induced in the mantle wedge here and therefore we would not expect a well developed LPO in the mantle wedge in the vicinity of the easternmost Wrangell slab. If, however, the actual slab shape were two-tiered as in slabE115 but with the slab beneath the Wrangells extending to 175 km, rather than to 115 km, we would expect the tiered shape to produce two regions of toroidal flow in the mantle, one centered at 212° longitude and the second centered farther to the east at 223° longitude. In this case, one would expect a well developed LPO, and thus organized pattern in the seismic anisotropy, in the mantle wedge near both 212° and 223° longitude. Additional seismic anisotropy measurements near the northern end of the Queen Charlotte-Fairweather fault would be useful to better constrain the depth of the shorter slab segment beneath the Wrangell mountains.

6. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Alaskan Subduction-Transform Boundary
  5. 3. Methodology
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[83] We have constructed 3D instantaneous regional models of the eastern Alaska subduction-transform plate boundary system to investigate the role of slab geometry and rheology on the decoupling of mantle flow from surface plate motion, in terms of both direction and magnitude, at subduction zones. The 3D models use a subducting plate geometry based on Wadati-Benioff zone seismicity and seismic tomography and a 3D thermal structure for both the subducting and overriding plates, constrained by geologic and geophysical observations. The models test the effect of a Netwonian viscosity and a composite viscosity, which includes both the diffusion and dislocation creep mechanisms (i.e., Newtonian and non-Newtonian viscosity). The models in this paper differ from previous 3D numerical models of subduction in that previous models either use a simplified plate geometry, prescribe a velocity boundary condition to the subducting plate, do not include an overriding plate, or use a Newtonian rheology in the mantle. However, our models are limited in that they are instantaneous, examining the spatial distribution of the flow rather than the time-dependent evolution.

[84] Models using the composite viscosity formulation have higher strain rates and lower mantle viscosities than models using the Newtonian viscosity. In models using the composite viscosity, a laterally variable mantle viscosity emerges as a consequence of the lateral variations in the mantle flow and strain rate. Spatially variable mantle velocity magnitudes are predicted, with localized fast velocities (greater than 80 cm/yr) close to the slab where the negative buoyancy of the slab drives the flow. The same models produce surface plate motions of less than 10 cm/yr, comparable to observed plate motions. We expect that it is short slabs undergoing slab steepening, not supported by the higher viscosity mantle below the transition zone, that are more likely to induce the localized rapid mantle flow. These results show that a power law rheology, i.e., one that includes the effects of the dislocation creep deformation mechanism, can explain both observations of seismic anisotropy and the decoupling of mantle flow from surface motion.

[85] The incorporation of realistic slab geometries into 3D models of subduction leads to a complex pattern of poloidal and toroidal flow in the mantle, with the spatial positions of the mantle flow components sensitive to the slab geometry. We conclude that slabE115 is representative of the actual slab shape beneath south central Alaska. The vertical motion of mantle material near slab edges may play a role in forming melts from both decompression melting and melting of the slab edge (i.e., adakites), as this part of the subduction zone is continuously exposed to upward advected warm mantle material. The Wrangell volcanics may be due in part to the upwelling associated with the deeper edge of slabE115.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Alaskan Subduction-Transform Boundary
  5. 3. Methodology
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[86] We thank C. Matyska, W. P. Schellart, and an anonymous reviewer for thoughtful reviews of the manuscript. We thank D. Turcotte, L. Kellogg, S. Roeske, D. Eberhart-Phillips, T. Taylor, and L. Moresi for thoughtful discussions. We thank Oliver Kreylos and the UC Davis Keck Center for Active Visualization in the Earth Sciences (KeckCAVES) for enabling the incorporation 3D immersive data visualization into this research. We thank the Computational Infrastructure for Geodynamics (CIG) for the CitcomCU source code. Figures were made with GMT and 3D Visualizer. High resolution models were run on the TeraGrid site, Lonestar, at the Texas Advanced Computing Center (TACC) through TG-EAR080015N. This work was supported by National Science Foundation grants EAR-0537995 and EAR-1049545.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Alaskan Subduction-Transform Boundary
  5. 3. Methodology
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Alaskan Subduction-Transform Boundary
  5. 3. Methodology
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information
FilenameFormatSizeDescription
jgrb16995-sup-0001-t01.txtplain text document3KTab-delimited Table 1.
jgrb16995-sup-0002-t02.txtplain text document0KTab-delimited Table 2.
jgrb16995-sup-0003-t03.txtplain text document0KTab-delimited Table 3.
jgrb16995-sup-0004-t04.txtplain text document1KTab-delimited Table 4.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.