Abstract

Crustal structure was derived from EarthScope Idaho-Oregon (IDOR) controlled-source seismic data across the Precambrian continental margin in the Idaho and Oregon region of the U.S. Cordillera. Refraction and wide-angle reflection traveltimes were inverted to derive a seismic velocity model that constrains the contact between oceanic accreted terranes and craton. The seismic data reveal that the boundary is a near-vertical, through-going feature of the crust, represented by the transpressional western Idaho shear zone (WISZ). The WISZ separates crust with different seismic velocities at all depths, implying a contrast in lithology, and extends to an ∼7 km offset of the Moho. The thinner, ∼32-km-thick accreted terrane crust to the west is characterized by faster seismic velocities that correspond to an intermediate composition. We interpret a high-velocity layer below a high-amplitude seismic reflection as mafic magmatic underplating associated with the feeder system of the Columbia River Basalts. The cratonic crust east of the WISZ is 37–40 km thick, with a felsic composition to ∼29 km subsurface depth, underlain by an intermediate-composition layer above the Moho. The strong contrasts in lithology and crustal thickness across the WISZ have influenced subsequent magmatism and extension in the region. The northwestern extent of the Archean Grouse Creek cratonic block beneath the Atlanta lobe of the Idaho batholith is interpreted based on continuity of crustal architecture in the seismic model. The velocity structure and crustal thickness east of the WISZ are consistent with the Atlanta lobe melting within a thickened crust.

INTRODUCTION

In western Idaho, a steep tectonic boundary marks the remnants of a suture zone that has been fundamentally reshaped since oceanic-island-arc terranes were accreted during the Jurassic–Cretaceous. This part of the North American Cordillera has undergone a complex sequence of tectonic events that have formed, deformed, overprinted, and altered the crustal structure of the former passive margin. Following subduction and terrane accretion, a period of arc magmatism and intense transpressional deformation consumed much of the suture zone and juxtaposed Precambrian North American craton against relatively juvenile accreted oceanic terranes less than 10 km away (e.g., McClelland et al., 2000; Tikoff et al., 2001; Giorgis et al., 2008). The current demarcation of the intervening boundary is the steeply dipping western Idaho shear zone (WISZ; Fig. 1). Subsequent tectonic events in the region, including the emplacement of the adjacent Idaho batholith, the outpouring of voluminous Columbia River flood basalts, and Basin and Range extension, were influenced by the preexisting crustal architecture.

The transition from oceanic-affinity strontium isotope signatures (87Sr/86Sr ≤ 0.704) to continental-affinity signatures (87Sr/86Sr ≥ 0.708) typically occurs gradually over >100 km elsewhere in the western Laurentia boundary characterized by suturing (e.g., Giorgis et al., 2005). This isotopic transition occurs over only 5–10 km in Late Cretaceous plutons across the WISZ in the study area (Armstrong et al., 1977; Fleck and Criss, 1985; Manduca et al., 1993). The close proximity of these isopleths indicates a near-vertical contact between the accreted terranes and the North American craton at least to the magma-source depth in the midcrust. However, geochemical data record a second sharp 87Sr/86Sr transition in mantle-sourced Miocene volcanic rocks that lie ∼120–150 km west of the WISZ in southern Oregon (e.g., Hart, 1985; Leeman et al., 1992; Evans et al., 2002). Models that have been proposed to explain the presence of the two subparallel Sr = 0.706 isopleths involve a detachment crosscutting the WISZ and creating a horizontal offset in the lower crust or upper mantle (Fig. 2; Leeman et al., 1992; Tikoff et al., 2008). Understanding the geometry and implications of this boundary at depth was a primary goal of the multidisciplinary EarthScope IDaho-ORegon (IDOR) project (Tikoff et al., 2017).

The IDOR controlled-source seismic project collected a 430-km-long seismic refraction and wide-angle reflection data set across the WISZ and ∼200 km on either side of the boundary (Fig. 1) to constrain seismic velocities in the accreted terranes, Idaho batholith, and craton; determine crustal thickness on either side of the boundary; and look for reflections that could be associated with a horizontal detachment structure in the lower crust or uppermost mantle. Results from refraction and reflection traveltime inversion provide new constraints on the deep structure of the boundary between oceanic-arc terranes and the North American craton, and on the structure and seismic velocity (wave speed) of the adjacent crust.

Tectonic and Geologic Background

The modern boundary between the Blue Mountains accreted terranes and Precambrian North American craton in west-central Idaho lies along the steeply dipping, 5–10-km-wide WISZ. Prior to terrane accretion, the western edge of the North American craton had been a dominantly passive margin since the Cambrian (Bond et al., 1984). Since the Jurassic, the margin has undergone multiple phases of tectonic modification associated with the formation of the North American Cordillera and subsequent events, including subduction and terrane accretion, transpressional deformation accommodated by a crustal-scale shear zone, multiple phases of voluminous magmatism, and extension. The complex sequences of events that have modified the Precambrian continental margin in the area of western Idaho and eastern Oregon have culminated in a boundary that is unusually steep and narrow (e.g., Manduca et al., 1993; McClelland et al., 2000; Tikoff et al., 2001; Giorgis et al., 2008).

The sequence of tectonic events that eventually led to the formation of the WISZ began with the amalgamation and accretion of oceanic-arc terranes during the Jurassic and Early Cretaceous (Selverstone et al., 1992; Getty et al., 1993). The accreted Blue Mountains Province terranes (Fig. 1) include oceanic-island arcs and arc-related sedimentary basins amalgamated outboard of North America during the Late Triassic (Vallier, 1977; Brooks, 1979; Dickinson, 1979; Vallier and Brooks, 1995; Dorsey and LaMaskin, 2007), and docked to North America at ca. 160 Ma south of their current location in Oregon and Idaho (e.g., Schwartz et al., 2011; LaMaskin et al., 2015; Gaschnig et al., 2016). The Salmon River suture zone formed by ca. 130 Ma as a result of these accretionary processes along the western edge of the craton (Lund and Snee, 1988; Gray and Oldow, 2005). Estimates for the total northward translation of the Blue Mountains terranes vary from ∼100 km to >1000 km (e.g., Wyld and Wright, 2001; Gehrels, 2001; Dickinson, 2004; Housen and Dorsey, 2005; LaMaskin et al., 2011).

After accretion, a subduction-zone continental arc was active in the area of the suture from ca. 120 to 87 Ma (e.g., Manduca et al., 1993; Giorgis et al., 2008; Gaschnig et al., 2010). Strong transpressional deformation produced the Late Cretaceous WISZ within the active continental arc (Fig. 1; e.g., Tikoff et al., 2001; Giorgis et al., 2016). Arc magmatism and the near-vertical shear zone truncated and largely consumed the Salmon River suture zone, producing the sharp Sr isotopic break and resulting in the narrow zone of deformation that represents the modern surface boundary between continental and terrane crust (Lund and Snee, 1988; Manduca et al., 1993; Giorgis et al., 2005). The suture zone and the shear zone are spatially coincident, but they likely represent distinct phases of deformation (McClelland et al., 2000; Tikoff et al., 2001; Giorgis et al., 2008). Remnants of the Salmon River suture zone at the latitude of the IDOR controlled-source seismic line are buried beneath Cenozoic volcanic flows and sediment, but there are exposures of suture-related belt rocks nearby to the north (Fig. 1; e.g., Hamilton, 1963; Gray and Oldow, 2005; Blake et al., 2009). The timing of motion on the WISZ is well constrained by age dates, ductile deformation patterns, and structural relationships in plutons near the shear zone. North of the seismic line, a single pluton exhibits variation from magmatic to solid-state foliations, which constrains initiation of motion on the WISZ to after ca. 104 Ma (Braudy et al., 2016). Cessation of motion is constrained to ca. 90 Ma by U-Pb zircon dating, pluton fabric analysis, and truncation of the WISZ by the Orofino shear zone (e.g., Manduca et al., 1993; McClelland and Oldow, 2007; Giorgis et al., 2008; Benford et al., 2010).

Within and immediately adjacent to the WISZ, the Idaho batholith includes a narrow zone of 110–87 Ma plutons that were coeval with motion on the WISZ (Manduca et al., 1993; Giorgis et al., 2008; Gaschnig et al., 2010). Major- and trace-element geochemistry of this phase revealed components of crust and mantle, indicative of arc magmatism (Gaschnig et al., 2011). A few plutons of similar age also exist further east. The most voluminous parts of the Idaho batholith, however, are the large 83–67 Ma Atlanta lobe, which is crossed by our seismic line, and the smaller 66–53 Ma Bitterroot lobe to the north (Fig. 1; Armstrong et al., 1977; Hyndman, 1983; Vallier and Brooks, 1986; Foster et al., 2001; Gaschnig et al., 2010). The composition of these lobes is more felsic, dominated by peraluminous mica granites, as opposed to the metaluminous amphibole-bearing granodiorite that dominates pre–85 Ma rocks. U-Pb zircon core data indicate that the peraluminous plutons cannibalized earlier plutons and melted the craton (Gaschnig et al., 2010, 2013, 2016). Petrology, metamorphic relationships in the Bitterroot lobe (Foster et al., 2001), and isotope geochemistry (Gaschnig et al., 2011) are all consistent with post–85 Ma melting occurring within thickened continental crust, with little or no contribution from the mantle.

Proterozoic–Paleozoic cratonic platform sediments are exposed to the east of the Idaho batholith near the seismic line (Fig. 1). Cratonic basement rocks are not exposed along the seismic profile, but Gaschnig et al. (2013) observed Archean zircon cores in the southern part of the Atlanta lobe that are consistent with the Grouse Creek unit as defined by Foster et al. (2006) in southern Idaho. U-Pb zircon ages farther north in the Atlanta lobe are consistent with a younger Proterozoic-age cratonic basement similar to that exposed north of the Atlanta lobe. The boundary between these cratonic blocks is poorly defined in the center of the batholith by the geochemical results, but it likely crosses the seismic line obliquely within the Atlanta lobe (Fig. 1).

The tectonic regime in the area became extensional in the Eocene. Eastern phases of the Idaho batholith and cratonic platform sediments are covered by Challis volcanic flows and are intruded by related shallow plutons in central-eastern Idaho (Fig. 1; Hyndman, 1983; Moye et al., 1988; Lewis and Kiilsgaard, 1991; Gaschnig et al., 2010). Challis magmatism spanned 53–43 Ma, with diverse compositions from mixed mantle and crustal sources (Gaschnig et al., 2011). The magmatism has a clear relationship with active extension (e.g., Bennett, 1986; Armstrong and Ward, 1991; Gaschnig et al., 2011).

Large portions of Oregon, Washington, and westernmost Idaho were covered in the Miocene by widespread lava flows of the Columbia River Basalt Group (Fig. 1; Waters, 1961; Reidel and Hooper, 1989; Reidel et al., 2013a, 2013b). Source dikes for the 17–15 Ma Columbia River Basalt flows are restricted to a few areas. The seismic line crosses the major Chief Joseph dike swarm near the Idaho-Oregon border (Fig. 1). The Columbia River Basalts are associated with regional extension and are often attributed to a mantle plume origin (e.g., Camp and Ross, 2004; Hooper et al., 2007; Camp, 2013).

Basin and Range extension, which began ca. 16 Ma (Oldow et al., 1989), dominates the Miocene to ongoing tectonic deformation. The seismic line crosses extensional ranges and valleys in eastern Idaho that strike northwest (Fig. 1), oblique to the northern strike of the Nevada Basin and Range to the south. Extension also occurs in Long Valley immediately east of the WISZ, which has tilted the WISZ to ∼70°–80° dip to the east (Tikoff et al., 2001).

IDOR REFRACTION AND WIDE-ANGLE REFLECTION SEISMIC SURVEY

A 430-km-long controlled-source seismic survey was acquired in 2012 across eastern Oregon and Idaho as part of the multidisciplinary EarthScope IDOR project. The east-west refraction and wide-angle reflection profile crosses perpendicular to the boundary separating the Blue Mountains Province accreted terranes and the Precambrian North American craton (Fig. 1). Distance along the profile was sufficient to record seismic waves that sampled the lower crust and uppermost mantle on either side of the boundary. The controlled-source profile was collocated with a broadband teleseismic array deployed as part of the larger IDOR project (Stanciu et al., 2016).

The controlled-source survey included eight explosive shots recorded over two nights. Shot spacing was nominally 40 km, with an ∼120 km shot gap across National Forest lands in Idaho due to permitting restrictions. This gap corresponds to the Idaho batholith, but stations in the batholith and ray paths undershooting the batholith provide reasonable coverage. The shots were each composed of ∼900 kg of explosives contained in a 30 cm borehole. The explosive sources were designed to maximize seismic wave generation while minimizing permanent rock deformation, so the explosive medium was buried to an ∼21 m center-of-mass depth, and the boreholes were plugged with cuttings and bentonite clay. The wide borehole diameter reduced the vertical length of the medium, increasing the peak pressure of the explosion and thereby increasing seismic energy (Harder et al., 2011).

The shots were recorded at 2555 stations utilizing 4.5 Hz vertical-component geophones and REF TEK 125A “Texan” seismographs recording at a 4 ms sampling rate. The dominant frequency observed from the shots was ∼12 Hz. The recording stations were spaced at 100–200 m along roads, 500 m on hiking trails, and ∼1 km where access was limited by road closures due to forest fires. There were two gaps of ∼5 km each where stations were not deployed due to lack of access by road or hiking trail, and an ∼20 km gap on shots 5 and 7 where an encroaching forest fire required stations to be removed prior to the second night of shooting. Instrumentation and field support were supplied by IRIS PASSCAL (Incorporated Research Institutions for Seismology, Portable Array Seismic Studies of the Continental Lithosphere), and deployment was conducted by a field crew of 53, mostly volunteers. An emphasis was placed on education and inclusion during the volunteer selection process, so the field crew included 22 undergraduate students recruited from schools with limited opportunities for research experience, including 4 yr colleges and minority-serving institutions.

SEISMIC DATA

Signal quality varied with the local surface conditions and geology for each source site. The western shots 1, 2, and 3 in Oregon were located in surface outcrops of accreted Blue Mountains Province, overlying Columbia River Basalt volcanics, or overlying Cenozoic sediment (Fig. 1). Source-to-ground coupling for these shots was good due in part to a high water table, and energy from these shots was clearly visible out to the easternmost recording stations, up to 420 km source-receiver offset (Fig. 3). Shot 5 was located in thick basalt flows west of the WISZ in Idaho (Fig. 1). Shot 7 was located on the eastern edge of the WISZ (Figs. 1 and 4). Energy from both of these shots was strongly attenuated in the western ∼90 km of the line, likely due to near-surface basins in that region. East of the Idaho National Forest shot gap, shots 8, 9, and 10 provided sources on the cratonic side of the continental margin (Fig. 1). Signal quality for these shots was not as high as the western shots, with usable signal to ∼200 km offset (Fig. 5). Shot 8 was located in a water-saturated stream valley near Challis volcanic flows and the eastern extent of the Idaho batholith. Shot 9 was above the water table in unconsolidated alluvial sediments on the western flank of the Lost River Range, which resulted in poor coupling. Shot 10 was located on Proterozoic crystalline rocks at the edge of the Pahsimeroi rift basin. The signal of both shots 9 and 10 was rapidly attenuated by propagation across ridge and valley structures associated with the Basin and Range extension in this area.

Multiple P-wave seismic phases are visible in the shot gathers for each of these shots (Figs. 35), as well as S-wave phases that are not considered in this analysis. P-wave phase information is summarized in Table 1. Upper-crust turning waves (Pg) have a high signal-to-noise ratio to 80–100 km offsets on every shot, except shots 9 and 10, which deteriorated beyond 25 and 45 km offset, respectively. The inverse of the time/offset slope of the Pg arrival provides an estimate of the apparent seismic velocity of the layers through which the rays traveled. Very short-offset arrivals for several of the shots show apparent velocities between 3 and 4 km/s, indicating the presence of sedimentary basins. The longer-offset Pg phases on the western side of the line rapidly increase in apparent velocity from ∼5.0 km/s for the shallower rays to 6.3–6.4 km/s beyond ∼50 km offset (Figs. 3 and 4). On the eastern side of the line, the maximum Pg apparent velocities are 6.0–6.2 km/s to >170 km offset (Figs. 4 and 5). This contrast is most obvious on the shot located at the WISZ, where the slope of the Pg arrivals to the west is visibly different from the slope of the Pg arrivals to the east (Fig. 4). The Pg phase becomes a secondary arrival behind seismic energy refracted from the uppermost mantle (Pn) at offsets greater than 150–160 km for stations west of the shots and ∼165–175 km for stations east of the shots. This Pn phase is observable at an apparent velocity of 7.8–8.0 km/s on all shots (Figs. 35) except shot 9.

Wide-angle reflections from the midcrust and Moho beneath both the eastern and western portions of the line are observed as secondary arrivals on all of the shots (Figs. 35). Reflected phases were identified on shot gathers based on the characteristic quasi-hyperbolic curvature of arrival time as a function of offset (“moveout”), and their largest amplitudes occur near the reflection critical angle. Refractions from below a reflector are tangential to the reflection hyperbola and have an apparent velocity associated with the layer beneath the reflector. Reflections from the Moho (PmP) were identified based on this tangential relationship to the ∼8 km/s apparent velocity Pn refracted phase. The Moho reflection observed on shot 7 for stations to the east is unusual in that it is observable at near-vertical angles but has lower amplitude approaching the critical angle. Near-surface features further obscure the association of PmP to Pn on this shot, so the identification of PmP east of the station gap is based on reciprocity with shot 8. PmP traveltimes were only picked in areas where the phase was clearly identifiable.

Reflections from discontinuities within the crust typically have a lower amplitude compared to PmP because the impedance contrast is not as strong as the contrast between crust and mantle. However, on most of the IDOR shots, the PmP arrivals did not have significantly higher amplitude than the intracrustal reflections. This lower amplitude could indicate energy attenuation through the crust, layering at the base of the crust, or a Moho that is gradational at the wavelengths of the data. The largest-amplitude reflection observed in the data is a midcrustal reflection labeled P1P on shot 2 that is much stronger than the Moho reflection (Fig. 3). This reflection is also observed on shots 3, 5, and 7. Midpoints for this arrival are beneath the accreted terranes in Oregon. Modeling indicates that any refracted phase associated with this reflector is not observed because it is not a first arrival and is not detectable in the coda of the Pg and Pn phases. There are two other reflectors observed over a relatively long range of offsets; both are lower amplitude than the P1P reflection. P2P and P3P are both observed before the PmP phase at stations on the eastern side of the line (Figs. 4 and 5). Additional moderate-amplitude reflections are recorded over shorter-offset ranges, including upper-crust reflections observed on shots 2, 7, and 9, midcrust reflections on shots 9 and 10 at the eastern end of the line, and a high-amplitude, short-offset reflection from the lower crust on shot 7 (P4P; Figs. 3 and 4). Arrivals that were reflected from the same intracrustal surface were identified based on reciprocity on reversed shot data.

P-wave traveltimes were picked manually for Pg, Pn, and reflected Moho (PmP) and midcrustal (PxP) phases. Data were filtered using an Ormsby band-pass filter with corner frequencies of 2, 8, 16, and 32 Hz. Crooked road geometry, ∼2 km of elevation variation, and narrow near-surface basins produced irregularities in the Pg arrival times of up to 250 ms delay between stations <200 m apart. To correct for these near-station delays, a static time correction based on the first arrival times was temporarily applied to aid picking of secondary arrivals. These time shifts were then removed to obtain absolute traveltimes.

The traveltime pick uncertainty on first-arrival Pg picks is estimated to be no more than half of the dominant period (<40 ms) for offsets <50 km on shots 1–8, and <100 ms for shots 9 and 10 due to the near-surface complexities and low-amplitude first arrivals on those shots (Table 1). Longer-offset and secondary arrivals have more variability in the picking uncertainty due to vastly different signal quality, attenuation, and noise on each arrival (Table 1). Secondary arrivals are also subject to a greater chance of cycle skipping (∼80 ms/cycle). These picking errors are usually systematic across a range of offsets, and there is a bias toward picking later cycles, and therefore modeling slightly deeper reflector depths, due to the difficulty of identifying emergent energy. Secondary arrivals observed between shots 7 and 8 are most prone to systematic picking errors due to the lack of reciprocity resulting from the lack of shots across this gap. Picks in this region were accepted with higher uncertainty than in other parts of the seismic line to ensure that the large-scale structure modeled beneath the batholith is consistent with the data. Velocities below Pg depths were based primarily on the curvature of midcrustal and Moho reflected arrivals, not their absolute times, so systematic cycle-skipping errors of a few hundred milliseconds on traveltimes of tens of seconds have a minor effect upon velocity.

TRAVELTIME INVERSION

Methodology

Traveltimes of the observed P-wave phases were inverted to produce a seismic velocity model of the crust and uppermost mantle using a three-dimensional (3-D) inversion procedure that incorporates direct, turning, reflected, and refracted arrivals. The method forward models first-arrival traveltimes to every point throughout a gridded 3-D velocity volume using a finite-difference solution to the eikonal equation (Vidale, 1990; Hole and Zelt, 1995). This algorithm is computationally more efficient than ray tracing to a large number of stations. Ray paths through the gridded time volume are found by following the steepest traveltime gradient from the receiver back to the source. Traveltime misfits are inverted by simple back-projection (Hole, 1992), distributing them equally along the ray path. The slowness (inverse of velocity) perturbation of each grid cell is then calculated from the mean misfit of all the rays that intersect that cell. This perturbation volume is smoothed and added to the velocity model. The entire procedure, including ray tracing in the updated model, is repeated until the model reaches an acceptable traveltime misfit.

The algorithm was expanded by Hole and Zelt (1995) and Zelt et al. (1996) to incorporate reflected arrivals by calculating traveltimes through the velocity volume in two steps, where the reflection surface acts as an intermediate “source” for calculating the second part of the ray path. Reflection traveltime misfits are inverted for the depth of the reflector, and then the overlying layer velocity, and iterated until a satisfactory misfit is obtained.

The modeling philosophy emphasizes minimum structure and a layer-stripping approach to building the model. The minimum structure approach (Zelt, 1999) avoids imposed features that are not strictly required by the data, even if a model that includes the imposed structure could be designed to match the data. Long-wavelength velocity structure is determined in the earliest iterations of the modeling process by applying smoothing to the slowness perturbation at a scale that is only slightly smaller than the length of the model. Details in the velocity structure are allowed to emerge through subsequent iterations as the horizontal and vertical smoothing is reduced gradually to maintain model stability. This approach pushes traveltime misfits into large, smooth structure, such that the final model is a smooth version of the true velocity structure. A range of reasonable smoothing schemes will produce velocity models with the same major features, while the optimal scheme minimizes ray path streaking and artifacts in areas of poor ray coverage.

The layer-stripping philosophy (Parsons et al., 1996; Zelt et al., 1996; Zelt, 1999) attempts to build a robust model by using the highest-quality data first, and only incorporating lesser-quality data once the better data have been modeled. This process is accomplished by beginning with the shortest shot-receiver offset arrivals, which usually have the highest signal-to-noise ratio and the least uncertainty, and using them to model the shallowest layers of the model with the highest resolution. The short ray paths mean that high-quality traveltime misfits are only being applied to a small number of grid cells at shallow depth, which reduces uncertainty in any given cell. After this layer of the model has reached a satisfactory misfit, the shallow portion of the model is kept fixed during modeling of deeper layers. The next set of traveltimes is incorporated to model the next deeper layer, and the procedure is continued until all phases have been incorporated into the model. Larger smoothing is maintained in the deeper layers of the model as compared to the shallower layers, corresponding to the larger pick uncertainty and longer ray paths. This process forces the misfits of each phase into the deeper layers of the model that have not already been constrained by more reliable traveltimes.

Implementation

The orientation of the IDOR controlled-source seismic profile was approximately east-west, with road and trail access resulting in a crooked line geometry with ∼40 km of north-south variation. The 3-D model box was 430 km × 45 km × 50 km, with a 500 m grid size, to allow ray tracing through the velocity volume with true shot and receiver locations. The x-y coordinate system used for the traveltime tomography was based on a Transverse Mercator coordinate system centered on −115° longitude. The z-axis was depth below sea level, and true shot and receiver elevations were included in the model.

Upper Crust

The preferred starting model for the traveltime inversion was a very smooth one-dimensional (1-D) model with no sharp changes in velocity that was based on an average of the first-arrival traveltime picks for all shots (Fig. 6). Alternative models with a faster/slower average velocity and a higher/lower velocity gradient with depth were also tested. None of the starting models included lateral variation in velocity. All of the 1-D starting models produced final 3-D models after inversion that displayed similar structure, including a lateral velocity contrast in the upper-crust velocity.

The initial smoothing for the tomography (Hole, 1992) was 240 km in x and y and 40 km in depth. The smoothing was allowed to decrease in successive iterations while maintaining a fixed horizontal-to-vertical ratio. Several x-z ratios were tested, including 2:1, 3:1, 4:1, 5:1, and 8:1. The 4:1 horizontal-to-vertical ratio was chosen because it provided the best traveltime misfits without creating obvious artifacts in the model. This ratio also mimics the dominant shape of the ray paths, since they travel much farther horizontally than they do in depth. At most scales, the crooked line geometry did not provide enough ray coverage to constrain north-south velocity variation, so the model was forced to be two-dimensional (2-D) by smoothing across the entire model volume in the y direction.

Reliable ray paths from Pg traveltimes constrained the first layer of the IDOR seismic velocity model to a subsurface depth of ∼10 km (Fig. 7A). Due to the shot spacing, the highest signal-to-noise and short-offset (<40 km) portion of these arrivals did not have sufficient intersecting rays to model independently as the first layer. To incorporate the higher-quality information from stations at <40 km offset, the initial Pg upper-crustal model derived from all offsets was updated by tracing the traveltimes from these stations through the model with less smoothing. For this near-surface layer alone, variation in the y direction was allowed due to the north-south geologic variability and crooked line at the scale of the final 2 × 2 × 1 km smoothing (Table 1). These high-resolution updates to the model reveal structures in the upper ∼2 km, such as sedimentary basins, that can be correlated with known surface geology. The final root-mean-square (RMS) traveltime misfit of 43 ms for all Pg picks (Fig. 8A; Table 1), including longer offsets, was improved from >60 ms without the high-resolution near-surface layer.

A nongeologic 2-D boundary was extracted from near the bottom of the reliable Pg ray coverage, near 10 km subsurface depth. The model above this boundary was held fixed when modeling the deeper crust. The velocities at this boundary were extended to greater depth and smoothed to produce the starting model for the middle crust.

Middle Crust and Reflectors

Once the upper-crust velocity was well resolved, wide-angle reflection traveltimes from the midcrust were used to invert for reflector depth and velocity structure between the fixed upper crust and the reflector. Velocity was primarily constrained by the curvature of the reflected arrival times, so phases observed over a long-offset range had little trade-off between depth and the velocity of the overlying layer. This resolution was improved when the reflected phase was observed on more than one shot.

Reflection traveltimes were initially inverted for the best-fitting constant reflector depth in the extended Pg velocity model. Next, the reflection was forward modeled using a range of constant velocities, starting with the velocity at the base of the overlying Pg model and increasing in 0.1 km/s increments until the misfits stopped improving and began to deteriorate. This velocity model was then used to invert for reflector depth, and the process was repeated until a best-fitting combination of constant velocity and constant depth was found. For a shallowly dipping reflector, an incorrect average velocity produces misfits that are strongly dependent upon distance. This pattern allows the average depth and overlying velocity of reflectors observed over a long range of offsets and/or on multiple shots to be well constrained. The best velocity and depth combination for each reflector was used as the starting model for the 2-D inversion.

Reflection traveltimes were inverted for laterally varying reflector depth in this best-fitting starting velocity model, and then for 2-D velocity between the reflector and the overlying fixed layer, and then again for reflector depth (Zelt et al., 1996). Reflector depth was smoothed horizontally during the inversion to maintain geologically reasonable structures and eliminate topography that is not required by the data. For most of the reflected phases from the midcrust, the reflectors were smoothed 10–20 km horizontally to incorporate ray paths from multiple shots (Fig. 7B; Table 1). However, across the Idaho batholith, they were smoothed 40–60 km horizontally due to the gap in shot coverage. Velocity in the layers above the reflectors and below the shallower fixed layer was smoothed 20–60 km in x, with larger smoothing used for deeper layers and larger shot offsets, 3 km in z, and forced to be constant in y. Where multiple midcrustal reflectors were observed, reflector depth and velocity were determined for the shallower reflector first, and then this structure was fixed to model the deeper reflection traveltimes. While the ray coverage on the reflectors is good (Fig. 7B), traveltime misfits of 200–300 ms (Fig. 8B; Table 1) represent probable cycle skips.

Most of the crustal reflections observed on the IDOR shots do not have associated refractions from beneath the reflector, so there are not well-constrained velocity contrasts across any of the reflectors. The high-amplitude reflection recorded on the western end of the line probably has an associated refracted phase, but modeling indicates this phase would never emerge as a first arrival, so it is probably hidden in the coda of the earlier arrivals. The reflections without associated refractions were modeled as “floating” reflectors (Zelt, 1999). Although seismic energy can be reflected back to the surface due to the impedance contrast between two layers with different bulk properties (seismic velocity and/or density), reflections may also result from the layering of thin beds, sills, or shear fabrics. These reflections do not require a change in bulk properties across the reflection surface, and therefore midcrustal reflections are modeled as “floating” within a smooth velocity gradient unless a refraction is observed from beneath the reflector, or deeper data require much higher velocity beneath the reflector. For most of these reflectors, the deeper data allow at most a small velocity contrast across the reflector, such that the next layer’s average velocity is the same. Exceptions to these average velocities occurred in the lower crust on either side of the line.

The strong, deep P1P reflector beneath the western end of the line implies a velocity contrast based on the reflector strength on shots 2, 3, and 5 (Fig. 3; Data Repository Fig. S1A1). The underlying Moho PmP reflection is not observed over a very long offset range, so velocity beneath the P1P reflector is not well constrained by this shot. However, the difference in curvature between the P1P and PmP arrivals observed on multiple shots implies a different average velocity above each reflector, indicating a higher-velocity layer must exist between the two reflectors. The velocity assigned to this layer is the minimum average velocity required to match the data.

On the eastern end of the line, the P2P and P3P reflections have very similar curvature, indicating there is not a strong velocity contrast between these reflectors. The curvature of the PmP reflection is different from that of P2P and P3P, and it requires a higher velocity in the lower crust between P3P and the Moho. East of shot 7, the difference in curvature between the P2P and PmP reflections requires a higher velocity in the lower crust, which is consistent with the westward extrapolation of the P3P velocity discontinuity. However, the shot gap, missing stations, and complexities across the WISZ disrupt the continuity of arrivals in this area, preventing this reflector from being identified farther west. The same average velocity can be achieved with a smooth gradient from the P2P reflector to the Moho, which is also consistent with the data, so the velocity discontinuity was restricted to the region beneath the P3P reflector, and the remaining lower crust was modeled with a gradient.

Lower Crust, Moho, and Uppermost Mantle

Once all of the midcrustal reflectors were fixed in place, the velocity structure above those reflectors was fixed, and the PmP phase reflected from the Moho was incorporated into the model. The initial ray path calculations indicated significant variability in the Moho depth, and smoothing the Moho beneath the entire seismic line resulted in severe velocity artifacts. To isolate the source of the artifacts, PmP arrivals from each shot were inverted independently for reflector depth. Based on these reflector depths, the Moho reflections were divided into two separate phases, PmP-w and PmP-e. Traveltimes from shots 1, 2, 3, 5, and 7-west were designated PmP-w, and traveltimes from shots 7-east, 8, 9, and 10 were designated PmP-e. These reflectors were modeled independently.

The ray coverage at the Moho on the western side (PmP-w) is dense for most of the model (Fig. 7B). The PmP-e ray coverage on the eastern side is sparser, with modest coverage and higher estimated picking errors beneath the Idaho batholith. Due to the shot distribution on the east and the shadow-zone effect for shots from the west, the western end of the PmP-e reflector is only constrained by the short-offset PmP arrivals from shot 7. This relatively low-amplitude arrival is clearly visible behind the high-amplitude midcrust P4P reflection (Fig. 4).

Inversion for reflector depth and lower-crustal velocity followed the same procedures as for the shallower reflectors. The Moho on the western side of the line is better constrained due to smaller shot spacing and higher-amplitude PmP-w reflections. Traveltime misfits for the lower crust above the PmP-w reflector are ∼107 ms. The PmP-e reflector is less well constrained due to the shot gap across the batholith and near-surface complexities from the shots in the Basin and Range. In the area of the batholith, average misfits are ∼180 ms, indicating probable cycle skips due to the lack of reciprocity in this area. On the eastern end, misfits are ∼220 ms (Fig. 8B; Table 1).

Once the crustal velocity model and Moho depth were fixed, Pn arrivals were inverted for velocity in the uppermost mantle lid (Data Repository Fig. S1B). The limited range of geologically feasible velocities in the uppermost mantle means the Pn phase also serves to verify the crustal velocity model and Moho depth. The Pn arrivals on shots 1, 2, and 3 have good signal quality. Pn phases are observed on the other shots (except shot 9), but the signal-to-noise ratio is not as high, so the picking uncertainty is greater for these arrivals (Table 1). Pn traveltimes calculated through the final velocity model using a constant 8.0 km/s uppermost mantle velocity provide a good fit for the observed Pn phases at less than ∼220 km offset on all shots. Longer-offset Pn arrivals are difficult to identify on most shots. A very large smoothing was utilized to invert for velocity in the uppermost mantle, which did not vary significantly from 8.0 km/s.

Seismic Velocity Model

The whole-crust seismic velocity model produced from the IDOR controlled-source data reveals a strong contrast in seismic properties beneath the WISZ (Figs. 9 and 10). The western half of the line is characterized by faster velocities throughout the entire crust, while the eastern side has slower velocities (Figs. 9 and 10). These velocities are consistent with those measured by Christensen and Mooney (1995) and Christensen (1996) under laboratory conditions for intermediate-to-mafic and felsic compositions, respectively. In addition, there is a large, steep step on the Moho beneath and slightly east of the WISZ. These structures indicate the presence of a boundary that coincides with the surface expression of the WISZ.

On the western and central portions of the line, there are several small, surface low-velocity bodies that are modeled using the short-offset Pg traveltimes recorded on densely deployed stations (Fig. 10). These correlate with Cenozoic sediment-filled basins identified at the surface (Fig. 1). The eastern end of the profile also reveals sediment-filled basins, but these Basin and Range–related fault-bounded basins are less well resolved due to significant local north-south geologic variation and a lower signal-to-noise ratio.

Below ∼5 km depth on the western half of the profile in the accreted terranes, velocities of 6.3–6.5 km/s in the middle crust (Figs. 9 and 10) are consistent with rocks of felsic to intermediate composition (Christensen and Mooney, 1995). The very high-amplitude P1P reflection observed on the western shots (Figs. 3 and 4) indicates a boundary at 21–22 km subsurface depth (∼20 km below sea level), 8–10 km above the Moho (Fig. 10). The lower crust beneath the reflector did not produce a first-arrival refraction, but the relative curvatures of the P1P reflection and the PmP phase reflected from the Moho constrain the layer to 6.8 ± 0.1 km/s. The fastest portion of this layer corresponds to the highest-amplitude section of the reflector (Fig. 3) and the thinnest crust (Fig. 10), strongly suggesting a discontinuity across the boundary. This velocity is interpreted as evidence of a mafic lower crust (Christensen and Mooney, 1995). Another smaller-amplitude reflection, P5P, is west of shot 2 at ∼10 km subsurface depth. It has limited lateral extent and cannot be shown to correspond to a velocity contrast and thus has been modeled as a “floating” reflector.

On the eastern half of the profile, the seismic velocity in the upper 24–30 km of the crust is ≤6.2 km/s, which is consistent with a felsic lithology (Christensen and Mooney, 1995). The upper 10 km is ≤6.0 km/s, except in a region of slightly faster velocity at 5–8 km depth from model km 270 to 325 (Fig. 10). In this area, a slightly faster velocity of 6.2–6.3 km/s is required by the data, based on velocity tests, but this is in the region of poorest ray coverage due to the shot gap and fire-related station gap (Fig. 7), so it is heavily smoothed. With the exception of this feature, there is little velocity difference between the Idaho batholith and the Precambrian craton to the east.

A continuous reflector, P2P, is observed on shots 7 and 8 (Figs. 4 and 5), which are reversed across the batholith. This phase corresponds to a reflector at 20–24 km subsurface depth, dipping gently to the east (Fig. 10). A deeper reflector, P3P, is observed on shots 8, 9, and 10 and overlaps laterally with the P2P reflector. This reflector is at ∼29 km subsurface depth, 8–10 km above the Moho (Fig. 10). The seismic velocity below the P3P reflector is 6.6 ± 0.1 km/s, constrained by curvature of the P3P and PmP reflectors. This velocity is consistent with that of intermediate-composition rocks at lower-crustal pressures and temperatures (Christensen and Mooney, 1995). The western end of P3P is not constrained by shot 7 due to shot and receiver gaps and data quality, so it is possible this reflector could extend further west. There is no evidence in the data for an extension, but there is also not clear evidence of the reflector’s termination. The curvature of the P2P and underlying PmP reflectors is consistent with either a westward extension of the velocities above and below the P3P reflector, or a smooth gradient with an equivalent average velocity (Fig. 10). A strong, short P4P reflector exists immediately east of the WISZ at ∼27 km subsurface depth (Fig. 10). Velocity below this reflector, in the lower-crustal corner near the WISZ, is not well constrained, but it is inconsistent with mantle velocity. Several shallow reflectors underlie the eastern half of the line, but none requires an associated large increase in velocity.

The Moho was modeled independently on the east and west sides of the profile to avoid artificially imposing structure when it became apparent there was significant lateral variation. On the west under the accreted terranes, the Moho is 31–32 km below the land surface and deepens modestly toward the western end of the line (Fig. 10). The Moho beneath the batholith and craton is at 37–40 km depth below the higher-elevation land surface (Fig. 10). An ∼7 km offset on the Moho exists within <15 km beneath the WISZ (Fig. 10).

The upper-mantle seismic velocity is constrained by the Pn arrivals. After ray tracing through the crust’s strong lateral variations, the Pn traveltimes are consistent with ∼8.0 km/s constant velocity along the entire line.

GEOLOGIC AND TECTONIC INTERPRETATION

Western Idaho Shear Zone

Analysis of the IDOR controlled-source seismic profile reveals the WISZ to be a through-going feature of the crust, indicating that the juxtaposition of oceanic- and continental-affinity rocks extends in a near-vertical orientation through the entire crust, at least to the depth of the Moho and possibly deeper into the lithosphere (Figs. 9 and 10). At the surface, the steeply dipping WISZ represents the boundary that separates relatively young oceanic-island-arc terranes of the Blue Mountains Province from the Precambrian North American craton and Idaho batholith (e.g., McClelland et al., 2000; Tikoff et al., 2001; Giorgis et al., 2008). We determine that this strong contrast in terrane age and composition extends throughout the entire crust, manifested as changes in the seismic character and contrasting seismic wave speeds observed on either side of the shear zone (Figs. 9 and 10). Below the surface expression of the WISZ, the PmP reflections are not well constrained for a distance of <15 km between the shallower western and deeper eastern portions of the Moho. This gap begins directly beneath the WISZ and extends eastward under Long Valley (Fig. 10). This location is consistent with surface mapping of the WISZ, which indicates a current dip of 70°–80° to the east, while estimates of Miocene-to-modern extension restore the boundary to near-vertical prior to the Miocene (Tikoff et al., 2001). The contrasting crustal seismic velocities, the sharp Moho offset, and the coincidence of these features with the surface WISZ indicate that the WISZ is a through-going crustal boundary.

Transpressional deformation on the WISZ resulted in significant tectonic shortening of the broad Salmon River suture zone between the accreted terranes and craton (e.g., Giorgis et al., 2005), resulting in a very narrow boundary between oceanic and continental geochemical signatures across the WISZ (Armstrong et al., 1977; Fleck and Criss, 1985; Manduca et al., 1993). Syn–WISZ continental-arc plutons record an unusually narrow 87Sr/86Sr isotopic break coincident with the WISZ, indicating a steep boundary between oceanic and continental source rocks to at least the lower crust. A similarly sharp Sr isotope transition was observed in Miocene volcanic rocks ∼120–150 km to the west of the WISZ in eastern Oregon, indicating a similar steep boundary in the lithospheric upper mantle (Hart, 1985; Leeman et al., 1992; Evans et al., 2002). Based on the differing depths of the source rocks, the dual steep ocean/continent boundaries were interpreted as evidence for a “shelf” of continental lithosphere extending west beneath the oceanic terranes visible at the surface (Fig. 2; Leeman et al., 1992). The top of this shelf was hypothesized to be a younger detachment crosscutting the WISZ and related to Sevier orogeny deformation far inboard. Our seismic image indicates that the WISZ extends steeply through the entire crust, and no detachment crosscuts it within the crust or at the Moho. Subhorizontal ductile detachments often produce strong seismic reflections, but none was observed in the IDOR data beneath the Moho. If compression was transmitted inboard via a horizontal detachment, it must be deeper in the mantle.

The strong contrast in age, composition, and rheology between the oceanic accreted terranes and the cratonic lithosphere had a significant influence on tectonic processes since 90 Ma. Post–WISZ granitic plutons of the Idaho batholith were emplaced close to the shear zone on the cratonic side (Fig. 1) and appear to be the result of melting in a thickened crust (Gaschnig et al., 2011; Byerly et al., 2016; Fayon et al., 2017). The lack of widespread intracrustal melting west of the shear zone implies a preferential susceptibility in the cratonic crust, probably due to greater thickness and the lower temperature required to melt more felsic rocks.

Miocene extension associated with the Columbia River Basalts is pervasive west of the WISZ (e.g., Camp and Ross, 2004; Hooper et al., 2007; Camp, 2013). The feeder dike system for the Columbia River Basalt volcanics is widespread in the accreted terranes and only crosses the WISZ in small isolated locations (Fig. 1; Reidel et al., 2013b), although some surface flows extend more broadly east of the WISZ (Reidel et al., 2013a). Miocene-to-modern extension exists in easternmost Idaho, but only affects the Idaho batholith in isolated locations such as Long Valley, immediately to the east of the WISZ in the study area, where extension has tilted the WISZ from near vertical to its current steep 70°–80° eastward dip (Tikoff et al., 2001; Giorgis et al., 2006). This distribution of extensional features indicates that the thinner accreted terrane side of the boundary is more susceptible to extension than the more felsic crust east of the WISZ. The compartmentalization of magmatism and tectonic deformation to either side of the WISZ likely results from the juxtaposition of crustal architectures that respond independently to tectonic stresses and heating events due to significantly different composition, thickness, and deformation history. This effect demonstrates the importance of structural and rheological inheritance on the response of the crust to subsequent tectonic and magmatic events.

Blue Mountains Province Accreted Terranes

The accreted terranes observed in outcrops west of the WISZ consist of oceanic island arcs and intervening sedimentary basins (Vallier and Brooks, 1995; Dorsey and LaMaskin, 2007). The seismic velocities in this region (Figs. 9 and 10) are inferred to represent intermediate compositions from 5 to 7 km depth through the middle crust, consistent with other accreted oceanic-island-arc terranes in the North American Cordillera (e.g., Spence et al., 1985; Spence and Asudeh, 1993; Hammer and Clowes, 2004; Stephenson et al., 2011). Variations in the near-surface velocity are caused by local sedimentary basins associated with Cenozoic extensional valleys and by sedimentary units of the amalgamated terranes.

The crust in this region is ∼32 km thick, and the Moho deepens toward the west. Crustal thickness of ∼35 km at the west end of the line is consistent with that observed on controlled-source seismic data ∼35 km to the west in the High Lava Plains (Cox et al., 2013). Broadband seismic data reveal a similar trend in crustal thickness across eastern Oregon and western Idaho (Eagar et al., 2011).

The thinnest crust in the IDOR controlled-source profile, ∼31 km, occurs west of the WISZ near the Snake River at the boundary between Oregon and Idaho (Fig. 10). This area is where the line crosses the southern end of the Chief Joseph dike swarm, which fed the outpouring of Columbia River Basalt flows (Fig. 1; Reidel et al., 2013b). Wolff and Ramos (2013) identified this area as the primary source location for mantle upwelling that supplied a significant volume of the Columbia River Basalts via dikes that emerged in the massive Grande Ronde lava flows up to 400 km to the north. The seismic data include an unusually high-amplitude P1P reflection at this location (Fig. 3) from the top of an 8–10-km-thick, lower-crustal layer with ∼6.8 km/s velocity (Figs. 9 and 10). We interpret the high-velocity lower-crustal layer beneath this strong seismic reflector (Fig. 10) as basaltic intrusion and/or underplating in the source region of the Columbia River Basalt basalts.

North American Craton

East of the WISZ, the Precambrian craton is 37–40 km thick, averaging ∼5 km thicker than the accreted terrane crust west of the shear zone (Figs. 9 and 10). On the craton side, the velocity is slower compared to the accreted terranes and is interpreted to indicate felsic composition in the upper and middle crust and intermediate composition in the lower crust. Crust at all depths east of the WISZ has lower seismic velocity and is interpreted to be more felsic than average continental crust, while crust to the west is faster and therefore likely more mafic than average (Fig. 9; Christensen and Mooney, 1995). There is no significant difference in velocity between the Idaho batholith and the host cratonic blocks (Figs. 9 and 10).

The Idaho batholith extends approximately from the WISZ to the location of shot 8, with the dominant Atlanta lobe between approximately model km 225 and 320 on the seismic line (Figs. 1 and 10). Beneath the Atlanta lobe, the velocities are interpreted to indicate a felsic composition to >24 km depth below surface and are consistent with granite-granodiorite plutons and/or granitic metamorphic lithologies (Christensen and Mooney, 1995). Beneath the eastern half of the Atlanta lobe, the velocity gradient is slightly higher from 5 to 15 km depth compared to the adjacent batholith or craton. The cratonic platform east of the Idaho batholith is partially intruded by Challis magmatism at a shallow depth (Moye et al., 1988). We did not observe a velocity signature associated with the Challis plutons or volcanics. The eastern end of the seismic line crosses extensional valleys and ridges associated with the Basin and Range. Velocity in this region is similar to the western part of the craton and the Idaho batholith, consistent with felsic lithologies to ∼28 km depth. Reflectors in this region could be detachments or other features related to Basin and Range extension. None of the reflectors imaged east of the WISZ exhibits a high amplitude or lateral extent consistent with a large-scale Sevier thrust detachment that could have offset the WISZ and transmitted Late Cretaceous compression into Montana and Wyoming.

A modest-amplitude reflector at 22–25 km subsurface depth correlates with the lateral extent of the Atlanta lobe, while a reflector at 28–29 km depth extends east from the center of the batholith and also underlies the unintruded craton (Fig. 10). The relative curvatures of the reflections from these features constrain the composition to be felsic below the western part of the Atlanta lobe at least to the depth of the shallower reflector at ∼22 km, and to the depth of the deeper reflector at ∼29 km beneath the eastern half of the Atlanta lobe. Where the two reflectors overlap laterally, the velocity between them must be as slow as above the shallower reflector. The relative curvatures of these reflections and the reflection from the Moho constrain the lower crust to have a faster average velocity; therefore, the composition is likely intermediate.

Beneath the deeper P3P reflector, the velocity average of the 8–10-km-thick layer above the Moho is 6.6 km/s, which is consistent with the P-wave velocity of diorite and in the range of velocities between felsic and mafic granulite (Christensen and Mooney, 1995). This velocity is not consistent with a mafic-to-ultramafic composition, such as mafic garnet granulite (Vp ∼7.0 km/s) or eclogite (Vp ∼7.9 km/s), which would be left behind by the melting of a mafic protolith in a continental arc (Beard and Lofgren, 1991; Rapp and Watson, 1995). The velocity and inferred intermediate composition are consistent with a residual left by intracrustal melting of a felsic protolith. A velocity discontinuity is the preferred model for the P3P reflector based on the strength of the velocity contrast across the reflector.

Beneath the western Atlanta lobe, near the offset in the WISZ, the curvatures of the ∼22 km P2P reflector and Moho reflections require an increase in the average velocity between 22 and 40 km depth. A velocity discontinuity is not required, since the data can be matched by either a layered zone or a gradient with the same average velocity. The deeper ∼28–29 km P3P reflector and associated intermediate-composition layer could extend westward and connect to the P4P reflector at the WISZ step, but there is no evidence in the data to support this connection due to the shot and station gaps. If the P3P arrival is present on shot 7 at longer offsets, it is obscured by the P2P and PmP reflections; alternatively, it may be truly absent. Additionally, P4P has high-amplitude, near-zero offset and is not identifiable at >30 km offset. It would be very unusual for a reflection from the top of a strong velocity contrast to decrease in amplitude as it approaches the critical angle. For these reasons, the velocity between P2P and the Moho has been modeled without a velocity discontinuity, and P4P has been considered an independent reflector adjacent to the WISZ (Fig. 10).

Idaho Batholith

The mostly felsic to intermediate composition of the Idaho batholith is similar to other exhumed Cordilleran Cretaceous batholiths in the California Sierra Nevada and Peninsular Ranges, and the British Columbia Coast Mountains (Spence and McLean, 1998; Fliedner et al., 2000; Hammer et al., 2000; Morozov et al., 2003). These other batholiths, like the western border suite of the Idaho batholith, are predominantly metaluminous granodiorites, whereas the Atlanta and Bitterroot lobes are predominantly peraluminous mica granites. Subduction-related continental arc batholiths are typically formed by dehydration partial melting of a mafic protolith, leaving behind a pyroxene-enriched residual (Beard and Lofgren, 1991; Rapp and Watson, 1995). Lack of evidence for this residual beneath exhumed Cretaceous batholiths in North America has been interpreted as evidence that it was delaminated into the mantle (Kay and Mahlburg Kay, 1993). The granites of the Atlanta lobe are primarily crustal-source granites (Gaschnig et al., 2011), however, and do not necessitate the existence of a pyroxene-rich residual. We did not observe evidence of a pyroxene-rich residual in the lower crust beneath the Idaho batholith, which we interpret as further evidence that the Idaho batholith melted in a thickened crust.

The lateral extent of the reflector at ∼22 km subsurface depth coincides with the surface extent of the Atlanta lobe of the Idaho batholith. Major -and trace-element geochemistry indicates that the Atlanta lobe melted in the garnet stability field of the lower crust, ≥35 km depth (Gaschnig et al., 2011). This reflector is roughly consistent with the minimum Late Cretaceous melting depth when ≥10 km of exhumation of the Idaho batholith (Fayon et al., 2017) is considered. Adding the estimated minimum exhumation to the current crustal thickness of 37–40 km yields a total crustal thickness of >50 km at the time of magmatism. This thickness is consistent with a thickened-crust melt source for the Idaho batholith, as proposed by Foster et al. (2001) for the northern Bitterroot lobe and by Gaschnig et al. (2011) for the Atlanta lobe, rather than a subduction-related arc source. Thickened crust is also consistent with observed petrographic fabrics (Byerley et al., 2016). The spatial extent, lack of velocity contrast, and low amplitude of the reflection from the ∼22 km reflector are consistent with a structural or textural feature related to the formation of the batholith, rather than a velocity discontinuity. We interpret this reflector to be the top of the source zone of melting for the Atlanta lobe in a Late Cretaceous thickened crust. Melting of a felsic protolith produce a smaller volume of less-felsic residual than the melting of a mafic protolith. This residual could be distributed in the lower crust, represented by the underlying rocks interpreted as intermediate composition.

The source zone for melting of the Atlanta lobe was ≥35 km, but the relationship of the Atlanta lobe to plutons of the early metaluminous suite indicates the depth of emplacement was much shallower, in the mid-to-upper crust (Gaschnig et al., 2010), consistent with estimates of exhumation (Fayon et al., 2017). There is evidence that the Bitterroot lobe was emplaced in the crust as a relatively thin sill, because the floor of the batholith is exposed in a metamorphic dome (Vallier and Brooks, 1986). Beneath the Idaho batholith, the velocity and average crustal thickness are very similar to the unintruded craton to the east, leading to the possibility that the Atlanta lobe is also relatively thin and extends to <10 km current depth.

The lateral extent of the ∼29-km-deep P3P reflector corresponds to the region of higher velocity gradient in the upper crust, and to a change in the depth of the Moho at the base of the crust (Figs. 9 and 10). These lateral changes in the velocity model are all coincident with the Archean-Proterozoic basement boundary at model km 275 ± 20 (Fig. 1) interpreted by Gaschnig et al. (2011) based on inherited zircon core data in Atlanta lobe plutons. We interpret the lateral change in crustal architecture beneath the center of the Atlanta lobe to be the northeastern margin of the Archean Grouse Creek craton. The continuity of crustal structure below the eastern surface boundary of the Idaho batholith combined with lateral contrasts at multiple depths in the crust at the interpreted Archean craton boundary indicate that the batholith has not strongly overprinted the crustal architecture of the preexisting craton.

CONCLUSIONS

Transpressional deformation in the WISZ formed a steep boundary between oceanic accreted terranes and the Precambrian North American craton. Traveltime inversion of the IDOR controlled-source seismic data indicates significant contrasts in velocity and crustal thickness that reveal the boundary to be a through-going feature of the crust that offsets the Moho ∼7 km. This juxtaposition of crustal units with significantly different compositions, ages, strengths, and thicknesses provided a mechanism for preferential distribution of post–WISZ deformation and magmatism, leading to structural inheritance influencing subsequent tectonic events.

West of the steep boundary, the accreted terrane crust is ∼32 km thick, with seismic velocities inferred to represent intermediate composition in the upper and middle crust, consistent with the oceanic-arc terranes. We interpret a high-velocity, 8–10-km-thick, lower-crustal layer to be a mafic magmatic intrusion and/or underplating associated with the Chief Joseph dike swarm feeder system of the Columbia River Basalt Group. On the eastern, cratonic side of the WISZ, the crust is 37–40 km thick. Slower seismic velocities that are interpreted as indicative of felsic lithologies persist to 25–28 km depth. There is not a significant difference in upper- and middle-crustal velocity below the granitic Idaho batholith and the unintruded craton to the east. A reflector at 22–25 km depth beneath the Atlanta lobe of the Idaho batholith is interpreted as the top of the zone of melting that fed the intrusion. Continuity of crustal architecture across the eastern margin of the Idaho batholith and a change of crustal architecture beneath the center of the batholith are interpreted to map the Archean Grouse Creek cratonic block extending beneath the batholith. Thick felsic-to-intermediate crust, combined with exhumation, indicates a Late Cretaceous crustal thickness of ≥50 km, which is consistent with the voluminous Atlanta lobe of the Idaho batholith forming by internal melting within a thickened crust.

Data acquisition and analysis for this project were funded by National Science Foundation (NSF) EarthScope Program grants EAR-0844264 and EAR-1251724 to Virginia Tech, EAR- 0844260 and EAR-1251877 to University of Wisconsin–Madison, and EAR-0844187 and EAR-1251814 to University of Florida. The seismic instruments were provided by the Incorporated Research Institutions for Seismology (IRIS) through the PASSCAL Instrument Center at New Mexico Tech. Data collected will be available through the IRIS Data Management Center. The facilities of the IRIS Consortium are supported by the NSF under Cooperative Agreement EAR-1261681 and the U.S. Department of Energy National Nuclear Security Administration. Special thanks go to George Slad for field management. We gratefully acknowledge the survey crew, shooter crew, and field crew, including ∼45 student volunteers, who made the data acquisition possible, as well as the landowners and public land managers who supported the project. This research used software provided by Landmark Software and Services, a Halliburton Company. We wish to thank Richard Gaschnig, Annia Fayon, Christian Stanciu, Jeffrey Vervoort, Paul Kelso, and Paul Bremner for many helpful conversations that aided the interpretation.

1GSA Data Repository Item 2017082, Figure S1, example of the model region constrained by a reflected phase, is available at http://www.geosociety.org/datarepository/2017, or on request from editing@geosociety.org.