Lithospheric S Wave Velocity Variations Beneath the Mackenzie Mountains and Northern Canadian Cordillera

Abstract

The Mackenzie Mountains (MMs) in the Yukon and Northwest Territories, Canada, are an enigmatic mountain range. They are currently uplifting (Leonard et al., 2008, https//doi.org/10.1029/2007JB005456), yet are about 700 km from the nearest plate boundary. Their arcuate shape is distinct and extends over 100 km eastward from the general trend of the Northern Canadian Cordillera. To better assess the cause and conditions of the current uplift, we processed ambient seismic noise data from a linear array of broadband seismographs crossing the mountains, along with other regional seismic stations, to estimate Rayleigh wave phase velocities between 6 and 40 s periods. From this, we estimated phase velocity dispersion and performed a tomographic inversion to estimate V_S. Tomography reveals a low-velocity structure that extends upward from the base of the ∼50–66 km thick lithosphere to the upper crust, and we hypothesize that inferred low density and low rigidity associated with the V_S anomaly localizes the ongoing uplift and thrust-dominated seismicity of the MMs. Additionally, we find relatively low crustal velocities that extend to the west of the MMs, suggesting that strain transfer from the Gulf of Alaska plate boundary plays a driving role as the crust translates to the northeast and buckles up against the craton consistent with the orogenic float hypothesis of Mazzotti and Hyndman (2002, https//doi.org/10.1130/0091-7613(2002)030〈0495:YCASTA〉2.0.CO;2). Finally, we observe lithospheric azimuthal anisotropy with an NW-SE fast direction. This is nearly orthogonal to teleseismic shear wave splitting measurements in the central MMs, and suggests that asthenosphere flow and lithospheric strain are not aligned in this region.

Key Points

• We image an 5% reduction in V_S beneath the highest extent of the Mackenzie Mountains that extends from the asthenosphere to the upper crust

• Lithospheric V_S anisotropy shows a dominant NW-SE fast orientation and contrasts with bulk upper mantle SKS splitting results

• V_S shows marked heterogeneity including a low-velocity lower-crustal channel west of the Cordilleran Deformation Front

Plain Language Summary

The Mackenzie Mountains (MMs) in the Yukon and Northwest Territories of NW Canada are actively uplifting, but lie approximately 700 km east of the nearest plate boundary where primary forces that cause mountain uplift are expected to originate. To investigate why these mountains are currently rising, we processed background seismic noise to extract fundamental mode seismic surface (Raleigh) wave phase velocities for raypaths between pairs of seismographs. We inverted these data to shear wave velocity to obtain a 3-D image of seismic shear wave velocity, which is sensitive to the rigidity, and thus the strength, of the crust and uppermost mantle. We find a notable volume of low shear wave velocities beneath the highest extent of the mountains that extends from the upper crust (several km depth) to near the bottom of the tectonic plate (around 50–66 km depth). We conjecture that this compositional, thermal, and/or fluid-based anomaly is intimately related to current uplift and its unusual location east of the general Cordillera in that its weak nature focuses deformation and uplift in the MMs.

1 Introduction

The Mackenzie Mountains (MMs) fold and thrust belt forms a distinctly arcuate eastward salient of the Northern Canadian Cordillera (NCC). The tectonic history of this continental and plate boundary region spans approximately 750 Myr (Monger & Price, 2002). Important past events include Neoproterozoic rifting of Rodinia, which may have left significant fabric signatures in today's lithosphere (Audet et al., 2016) and Middle Devonian (∼390 Ma) onset of subduction and arc magmatism, which led to the establishment of a coherent continental margin by the Late Cretaceous (∼90 Ma). Starting around 150 Ma, the region began a period of extensive right-lateral transpression and the accretion of exotic terranes, resulting in over 800 km of cumulative dextral slip, first on the Teslin, then the Tintina, and still later the on the Denali faults (Nelson et al., 2013; Figure 1). The MM lies in a region of pericratonic affinity and may have a sloping wedge of Precambrian crust beneath their west-thinning sedimentary cover (Cook et al., 2004). Prior studies indicate thin, hot, and weak lithosphere extending east from the western edge of the MM deformation front (Audet et al., 2019; Hyndman, 2017; Schaeffer & Lebedev, 2014). Petrological, seismological, and heat flow studies suggest that the lithosphere to the west of the Cordilleran deformation front (CDF; Figure 1) is only around 50–66 km thick (Audet et al., 2019; Francis et al., 2010; Harder & Russell, 2006; Lewis et al., 2003) consistent with low velocities observed in global seismology models (Schaeffer & Lebedev, 2014), and thus has low effective elastic thickness (Audet et al., 2007; Fluck et al., 2003). This state is consistent with mantle upwelling due to the opening of a slab window around 20–30 Ma (Thorkelson et al., 2011).

Figure 1

Left: terranes and major faults of the Mackenzie Mountains region; from east to west: North American craton, ancestral North America, Intermontane, Insular and Outboard terranes (Nelson et al., 2013). These are crosscut by the dextral Tintina, Teslin, and Denali faults (labeled in white). The easternmost black line indicates the Cordilleran deformation front (CDF), the easternmost extent of Cenozoic or newer deformation. Yellow triangles show the sites deployed during the 2017–2019 Mackenzie Mountain EarthScope Project (Baker et al., 2019). Additional triangles show the location of other seismographs used in the study. Top right: location of the study region at left (indicated by red box).

Unlike much of the Cordillera, the MM is actively uplifting through crustal shortening, as indicated by NE-SW thrust earthquake focal mechanisms (Mazzotti et al., 2013) and Global Navigation Satellite System (GNSS) deformation data (Hyndman et al., 2005; Leonard et al., 2007). Limited thermochronological evidence indicates rapid cooling intervals between 75–67 and 60–55 Ma (Enkelmann et al., 2019; Powell et al., 2016), which suggests that Cordilleran deformation in the westernmost MMs started before the Paleocene. A third phase of uplift is inferred around 33–20 Ma (Enkelmann et al., 2019), which would make the current NE-SW shortening the fourth period of deformation since the mountains formed. These uplift phases have been linked to North America plate movement (Enkelmann et al., 2019; McKay et al., 2021) but there is yet no thermochronological evidence of recent uplift and erosion (Enkelmann et al., 2019). This may indicate that the current episode has not created enough uplift to produce a thermal signature, or that sample availability excludes areas of recent active uplift. Two hypotheses have been advanced for the cause of the ongoing crustal shortening: (a) convergent tractions at the base of the lithosphere (Finzel et al., 2014, 2015); and (b) long-distance strain transfer from the Yakutat Indentor subducting off the Gulf of Alaska to the eastern margin of the mobile belt (Hayward, 2019; Hyndman, 2017; Mazzotti & Hyndman, 2002). Testing these two models requires high-resolution information of crustal and lithosphere-scale structure beneath and around the MM. Far-field crustal shortening in convergent or transpressional systems is an important process across extended plate boundaries with broad implications, and has been widely noted in past and presently active orogenesis (Erslev et al., 2022; Li et al., 2021).

Elucidating the cause of the shortening and associated uplift and its relationship to the geologic history and ongoing processes of the region is a goal of the Mackenzie Mountain EarthScope Project (MMEP), which included a roughly 2 yr deployment of 40 broadband seismographs and ongoing GNSS measurements (Baker et al., 2019). Prior to MMEP, this large and mostly roadless region was sparsely instrumented, and lithosphere-scale resolution was thus very limited. Pre-MMEP studies include McLellan et al. (2018) who used ambient noise and teleseismic earthquake data from a sparse network of seismic stations to calculate fundamental mode Rayleigh wave phase velocity maps at periods of 8–80 s. Recent studies incorporating data from MMEP include an inversion of ballistic Rayleigh wave group velocities from periods of 10–60 s (Estève et al., 2021), which noted low velocities at crustal and uppermost mantle depths below the NCC. We build on these studies here by using all available data from MMEP, EarthScope, Canadian, and other seismograph deployments, and extending Estève et al. (2021) with analysis of ambient seismic noise phase velocities between 6 and 40 s period, which can resolve velocity structure at depths of approximately 5–60 km. Interpreted in combination with topography, Bouguer gravity anomaly, and effective elastic thickness, we interpret structural controls on current dynamics and propose a refined conceptual model for the uplift of the MMs.

2 3-D Shear Wave Velocity Inversion

2.1 Ambient Noise Processing

Ambient noise data are obtained from 184 continuously recording seismographs operational in the region of 57°–70°N and 141°–114°W between January 2000 and May 2021 (Figure 1). The concurrence of deployment of the MMEP quasi-linear array (Baker et al., 2019), the Yukon-Northwest Seismograph Network (University of Ottawa, 2013), and the EarthScope Transportable Array (TA), for the latter part of this time period, produces substantial sampling of crossing inter-station surface-wave raypaths for imaging lithosphere-scale structure to at least 60 km depth.

Data processing follows the method of Bensen et al. (2008). We first download day-long vertical-component seismograms from all available stations. Conditioning of the data includes demeaning, detrending, tapering, removing the instrument response, and spectral whitening. We remove transient signals from ballistic sources by applying a running absolute-mean filter, followed by cross-correlating and stacking, to produce empirical Green's functions (EGFs) between all station pairs. We then perform time-frequency analysis on the EGFs to measure the fundamental mode Rayleigh wave phase velocities at periods of 6–40 s (Dziewonski et al., 1969; Levshin et al., 1989; Lynner & Porritt, 2017). Dispersion curves with a signal-to-noise ratio of less than 4 dB are culled, as are shorter station-station paths of less than 1.5 wavelengths. We also remove data that do not have continuously good signal-to-noise between 10 and 30 s. Lastly, we remove data corresponding to all paths connecting two MMEP stations to improve consistency in raypath coverage and minimize azimuthal artifacts arising from the quasi-linear MMEP deployment geometry. Resulting station-station path numbers range from 1,771 for 40 s wave period to 3,737 for 10 s wave periods (Figure S1 in Supporting Information S1).

Raypaths color-coded by interstation average wave speed are plotted for select wave periods in Figure S2 in Supporting Information S1. We note desirable consistency for nearby raypaths. For instance, at 15 s, the velocities for the MM12-C36M and MM13-C36M raypaths are 3.147 and 3.144 km s⁻¹, respectively (Figure S3 in Supporting Information S1). A 5–15 dB increase in winter microseismic noise levels occurs due to characteristic seasonal increased winter storm activity in the Gulf of Alaska (Aster et al., 2008, 2010; Baker et al., 2019; Figure S4 in Supporting Information S1). We attempt to minimize any corresponding bias by averaging cross-correlations over the maximum number of months for each station-station pair.

2.2 Phase Velocity Maps

Assuming hexagonal symmetry with a horizontal axis of symmetry, phase velocity can be represented by:

(1)

where C₀(T) is the isotropic phase velocity, the 2ψ and 4ψ terms account for variations in phase velocity at a given period (T) and wave-propagation azimuth (ψ) with π and π/2 periodicity, respectively (Smith & Dahlen, 1973). Rayleigh wave dispersion typically has low sensitivity to anisotropy with π/2 periodicity at longer periods, such as those considered here (Estève et al., 2021). Thus, in this study, we simplify our computations by limiting our analysis to only the isotropic and π periodic components of azimuthal anisotropy.

We invert surface-wave phase velocity dispersion measurements at seven distinct periods between 10 and 40 s to estimate azimuthally anisotropic phase velocities and associated 1 − σ uncertainties (Figure 2 and Figures S5–S6 in Supporting Information S1). We use a Bayesian (probabilistic) trans-dimensional tomography method described by Gosselin et al. (2021) that nodally parameterizes the 2-D maps. Each node is characterized by five parameters: two describing geographical location, one for isotropic velocity, and two for describing azimuthal anisotropy (as discussed above). The number of the nodes is also considered unknown. This trans-dimensional inversion procedure considers the data as travel times, with associated data errors represented by a linear function with respect to path length (Figure S10 in Supporting Information S1). See Bodin et al. (2012) and Gosselin et al. (2021) for further details.

The trans-dimensional Bayesian inversion uses a reversible-jump Markov chain Monte Carlo (rjMcMC) algorithm to construct an ensemble of models that approximates the Bayesian posterior probability density of the model parameters (Green, 1995). We set wide and uniform prior distributions for the model parameters (Table S1 in Supporting Information S1). A random initial velocity model was drawn from the prior. We ran 24 parallel chains, each producing five million samples of which the initial half were discarded as “burn-in” samples. The mean and standard deviation of the remaining samples were used as a representative model with associated uncertainty. See Estève et al. (2021) and Gosselin et al. (2021) for further details.

2.3 Pseudo 3-D Inversion

Phase velocities, with estimated 1 − σ uncertainties, at each point in the 2-D tomography maps were extracted to derive shear wave velocity (V_S) variation with depth using a 1-D inversion scheme following the approach of Estève et al. (2021). For these 1-D inversions, the crust and mantle are approximated by two Bernstein polynomials (Estève et al., 2021; Gosselin et al., 2017), where the Moho is defined as the depth of the discontinuous transition from one polynomial to the other. Several inversion tests were performed with different combinations of polynomial orders for the crust and mantle velocity structure. In this study, 1-D inversions are parameterized with a second-order polynomial in the crust and a third-order polynomial in the mantle, as we qualitatively determined that this parameterization fits the dispersion data with the fewest model parameters. Furthermore, it has been shown that the inversion of surface-wave data suffers from a trade-off between Moho depth and crust-mantle velocity structure, without prior constraint on Moho depths (e.g., Lebedev et al., 2013). Here, we use previously published Moho depth estimates for our study area as prior constraints (Audet et al., 2020; Miller et al., 2018; Postlethwaite et al., 2014). V_S profiles are parameterized with a minimum velocity at the surface, have a positive velocity contrast across the Moho, and are corrected for topography using ETOPO1 (Amante & Eakins, 2009).

The 1-D inversions at each location were performed using a Markov chain Monte Carlo (McMC) algorithm. We use broad uniform priors for the V_S in the crust and mantle (Table S2 in Supporting Information S1), with more informative priors on Moho depth (as discussed above). For each inversion, we produced an ensemble of 100,000 model samples that approximates the Bayesian posterior distribution of the model parameters (i.e., the polynomial coefficients, as well as Moho depth). The first 10,000 samples were discarded as “burn-in” samples. The mean and standard deviation of the remaining ensemble of models were used as the representative 1-D V_S profile with associated uncertainty. Finally, we combine all 1-D profiles to produce the pseudo 3-D V_S model of the region using the surface command from Generic Mapping Tools (Wessel et al., 2019). We produce a model of V_S extending to 100 km depth, but note that our data set has reduced sensitivity beyond approximately 60 km depth.

2.4 Results

Figure 2 shows the recovered phase velocity maps. Low phase velocities at all periods are found near the center of the MMEP transect, which is book-ended by higher velocities to the NW and SE. Low phase velocities are also found in areas to the west of the CDF, especially for periods of 25–40 s. Anisotropic fast axes are consistently oriented NW-SE, and align with the large-scale transpressional tectonic fabric (Figure 3). Resolution test results from a checkerboard synthetic input structure are shown in Figures S8–S9 in Supporting Information S1. We note that areas of low anisotropy cannot be resolved where the anisotropy resolution is particularly poor, so we only interpret the (orange or yellow; σ_θ ≤ 10°) regions where anisotropy is well resolved at 20 s.

Figure 2

(a–f) Fundamental-mode Rayleigh wave phase velocities at 10, 15, 20, 25, 30, and 40 s periods estimated via Bayesian trans-dimensional tomography. The Denali, Teslin-Thibert, and Tintina faults are shown (from west to east). The thrust contour with triangles indicates the eastern limit of Cordilleran deformation.

Figure 3

(a–f) Azimuthal anisotropy of phase velocities at 10, 15, 20, 25, 30, and 40 s periods shown as a vector field. The colored background shows the estimated uncertainty in fast axis orientation determined from the inversion, and lines give the direction and magnitude of the fast axes. Other characteristics are as in Figure 2.

Our 3-D shear wave velocity model is shown in six depth slices in Figure 4 and three cross-sections in Figure 5. Line A-A′ is aligned with the MMEP transect, and lines B-B′ and C-C′ encompass the northern and southern NCC, respectively (see Figure 4a). In cross-sections, the input and recovered Moho depth is almost the same as the priors because the surface waves have low resolution with regards to this (Figure 5). We find a distinctly low shear wave velocity (∼5%) anomaly within the center of the MM (Figures 4 and 5a). This anomaly extends to the base of the model at 80 km depth with a southwestward trend at greater depths (Figure 5a). The lowest crustal V_S also corresponds to areas of highest elevation within the MM. At 20–30 km depth, the low V_S anomaly also extends southwestward to the coast. Additionally, in cross-section A-A′ (Figure 5a), we observe a high V_S feature in the upper mantle (40–80 km) that is bounded by the Teslin and Tintina faults. Much of the lower crust (at 30 km depth) to the west of the CDF is characterized by low shear wave velocity (Figure 4). This extends into the mantle, although the imaged velocity variations are not uniform.

Figure 4

(a–f) Shear wave velocity estimated at indicated depths (note that the color bar range changes between subfigures). Gray areas indicate zones where the uncertainty is high (standard deviation > 0.3 km/s). Locations of cross-sections A-A′, B-B′, and C-C′ are indicated on plot (a). Other characteristics are as in Figure 2.

Figure 5

(a–c) Cross-sections A-A′, B-B′, and C-C′ (see Figure 4). V_S is plotted as deviations from the mean at each depth in the cross-section. The black dashed nearly horizontal lines show the Moho derived from the receiver function analyses (Audet et al., 2020) which were used as the input for the 1-D V_S inversions. Nearly coincident white dashed lines show the Moho depth recovered from the inversion. The elevation as shown by the gray filled profile, is given in meters above sea level.

3 Discussion

3.1 Comparison With Past Work

McLellan et al. (2018) used ambient noise and teleseismic earthquake data (without the MMEP data) set to obtain azimuthally anisotropic phase velocity maps of northwestern Canada. This study highlights the thin, weak, and hot Cordilleran lithosphere and the thick, rigid, and cool cratonic lithosphere of the Canadian Shield. Their model revealed low-velocity anomalies underlying the NCC from the crust to the uppermost mantle, which is in contrast to the broad high-velocity anomaly within the adjacent Canadian Shield to the east. At upper crustal depths, their model shows three low-velocity anomalies across the NCC that are interpreted to reflect thicker sedimentary cover. This is consistent with our model, which shows low phase velocities (<3.2 km/s) at 10 s (Figure 2a). At uppermost mantle depths, the authors suggest that the low-velocity anomalies within the Cordilleran mantle reflect elevated widespread temperatures and therefore would support the thermal isostasy model for the region (Hyndman & Currie, 2011).

More recently, Estève et al. (2021) calculated a pseudo 3-D V_S model of northwestern Canada obtained from azimuthally anisotropic Rayleigh wave group velocity maps. Their pseudo 3-D V_S model revealed several detailed features across the NCC due to a better seismic station coverage than McLellan et al. (2018). In general, their V_S model shows good agreement with ours with regards to large-scale structures in the NCC. In particular, at mid-to-lower crustal depths (15–30 km depth), a low V_S structure (<3.8 km/s) is observed from the NA-PA plate boundary to the CDF (Figure S13 in Supporting Information S1; transects A-A′ and D-D′′). This is interpreted to likely reflect elevated crustal temperatures that buoyantly support regional high elevations, in agreement with our results and that of McLellan et al. (2018). At uppermost mantle depths (40–60 km depth), a high-velocity zone (V_S > 4.5 km/s) is observed north of the Tintina fault in northern Yukon, which is interpreted to be the seismic signature of the Mackenzie craton. Our V_S model shows a high-velocity anomaly located in the uppermost mantle (Figure 5, transect BB′) and, therefore confirms that northern Yukon is underlain by a thick cratonic root.

The most notable discrepancies between our V_S model and that of Estève et al. (2021) are observed at uppermost mantle depths. First, their V_S model revealed a low-velocity region (V_S < 4.2 km/s) underlying Yukon. This low-velocity region is also characterized by sharp V_S contrasts at uppermost mantle depths beneath the surface expression of the Tintina and Denali faults. The authors interpret this feature as upwelling of hot asthenosphere in the area. Our V_S model also reveals a low-velocity anomaly in this area, however, it is offset to the southeast and we do not observe equally sharp V_S contrasts beneath the surface expression of the Tintina and Denali faults within the uppermost mantle. When converted to V_S perturbation (in percent) to the average background V_S, Estève et al. (2021) image the same low-velocity anomaly within the crust beneath the MM along transect A-A′ as in our velocity model (Figure S13 in Supporting Information S1 transect A-A′). However, at uppermost mantle depths, the amplitude of the low-velocity anomaly is less than in our V_S model (δV_S/V_S = ∼−2%).

Our estimates of seismic anisotropy at short periods are in agreement with that of Estève et al. (2021) and McLellan et al. (2018), and reflect crustal-scale features. Interestingly, at a period of 40 s, fast-axis orientations are oriented NW-SE across the MM and they progressively rotate counterclockwise further north, across western Yukon, to be NE-SW. This pattern of seismic anisotropy is consistent with that of Estève et al. (2021). This could be related to basal traction in the mantle or could reflect the influence of the Aleutian-Yakutat subduction zones in the region. At 40 s, the sensitivity kernels are maximum around 40–80 km depth (Figure S7 in Supporting Information S1), which is likely to straddle the lithosphere-asthenosphere boundary (Audet et al., 2019).

3.2 Co-Interpretation With Other Geophysical Data

To understand what controls the variations in the observed seismic velocity structure, we further extract grids of topography, Bouguer gravity anomaly (Hayward, 2019), and effective elastic thickness (T_e, from Audet & Burgmann, 2011) data over the region (Figures 6a–6c). At the broadest scale, we observe that high elevations are associated with negative Bouguer anomalies, especially at long wavelengths, and with low T_e. In the following discussion, we compare these data with the phase velocity maps instead of extracting V_S at various depth slices, since they sample smoothly varying and broad depth ranges. These data sets are first compared quantitatively by calculating the correlation coefficients (CC) between all pairs of data (Figure 7a), except between the various phase velocity maps.

Figure 6

(a) Topography, (b) Bouguer gravity anomaly, and (c) estimated elastic thickness (Audet & Burgmann, 2011) for the study region. (d) Bouguer anomaly and averaged phase velocity anomaly at indicated periods for profile A-A′.

Figure 7

(a) Correlation coefficient of topography, Bouguer anomaly, and elastic thickness with topography, Bouguer anomaly, elastic thickness (T_e) and phase velocities over the study area. (b) Correlation coefficient of same quantities extracted along profile A-A′.

We observe that the topographic and Bouguer anomaly data are well correlated (CC <−0.8), which suggests local isostatic compensation below the region. However, this relationship is known to vary as a function of wavelength (i.e., the admittance and coherence functions), which can be modeled as the loading and flexural compensation of an effectively elastic plate (Watts, 2001). The effective elastic thickness (T_e) can be calculated by mapping the spectral relationships over the region, for example, using a wavelet transform (e.g., Audet & Mareschal, 2007). The T_e data shown in Figure 6c is extracted from the global continental T_e model of Audet and Burgmann (2011). This map shows that T_e is low (<40 km) in the southern part of the NCC, and remains high (>50 km) north of approximately 63°. T_e is mostly sensitive to the geothermal gradient through temperature-dependent creep, as well as the state of mechanical coupling (i.e., yield strength contrast) across the rheological layers (Burov & Diament, 1995; Lowry & Smith, 1995). The T_e map is moderately correlated with the Bouguer gravity anomaly (CC ∼ 0.6), pointing to a potential thermal effect on crustal density structure, as predicted by the thermal isostasy model (Hyndman & Currie, 2011).

Comparing these data sets with the phase velocity maps (Figure 7a), we find the highest CC (∼0.8) between T_e and phase velocity anomaly at 40 s. This relationship suggests that seismic velocities of the deep crust and uppermost mantle dominantly reflect temperature variations that lead to commensurate changes in T_e. The moderately high CC values (0.67) between the Bouguer anomaly and phase velocities at periods of 30–40 s are consistent with this interpretation.

3.3 Variations Along the MMEP Line

We now focus our discussion to the variations in these parameters along the MMEP line, and plot the same geophysical data (topography, Bouguer anomaly, T_e) projected along this line (Figure 6d). We again calculate the correlation coefficient between all pairs of data (Figure 7b). We note that most of the previously discussed relationships hold, except that CC values for all combinations (except topography) that include T_e are slightly lower.

Figure 5 shows that the lower crust consistently displays low-velocity anomalies below high elevations, and that the seismic velocity structure of the uppermost mantle is more heterogeneous. In particular, we find negative V_S anomalies beneath the MM, and a localized fast seismic velocity anomaly between the Tintina and Teslin faults, consistent with high V_P and V_S anomalies that have been reported west of the Tintina Fault using teleseismic body wave data (Estève et al., 2020a, 2020b).

3.4 Upwelling of Fluids Beneath the MM?

Taken together, the previously established relationships suggest that high elevations are associated with a low-density crust and uppermost mantle (Brocher, 2005) that is likely reflecting elevated temperatures below the MM (Hyndman & Currie, 2011; Figure 5). However, while it is challenging to associate changing crustal V_S with a particular compositional trend (Hacker et al., 2015; Shinevar et al., 2018), we note that our observations could also be consistent with an increase in bulk silica content in the crust underlying the MM (Lowry & Pérez-Gussinyé, 2011). This, in turn, would reduce lithospheric strength and generate low T_e (Burgmann & Dresen, 2008; Lowry & Pérez-Gussinyé, 2011). However, the regional tectonic history does not indicate a terrane boundary approximately 100–200 km east of the Tintina Fault (Figure 5a), so perhaps the crustal composition is fairly homogeneous. Additionally, the continuity between the mantle low V_S and the crustal low V_S suggests a common origin, and mantle velocities are primarily modulated by temperature or partial melt (Schutt & Lesher, 2006). Based on our previous discussion, we prefer to interpret our results in terms of elevated lower crustal and upper mantle temperature below the MM.

3.5 Constraining the Two Hypotheses for the MM Uplift

Imaged low velocities in the lower crust to the west of the deformation front are consistent with a rheologically weak layer, as required for orogenic float induced by the Yakutat indentor (Mazzotti & Hyndman, 2002; Oldow et al., 1990). However, the variation in velocity, particularly the higher velocities to the south at 30 km depth (Figure 4) but still to the west of the deformation front, suggests that the process is geographically confined. Moreover, the strong variations in velocity from 10 to 60 km indicate that 3-D variations are important for uplift and broader deformational localization.

The anisotropy at 40 s (Figure 3) is perpendicular to the direction one would predict from the Finzel et al. (2014) hypotheses. If the 40 s anisotropy were localized at the base of the lithosphere, this would tend to invalidate lithospheric tractions as a cause for uplift of the Mackenzies. However, it is also possible that the anisotropy due to lithospheric tractions occurs below the sensitivity of 40 s waves, and is therefore unresolved in our imaging. We note that 40 s Rayleigh waves are most sensitive to structure around 30–70 km in depth (Figure S7 in Supporting Information S1).

Both of these hypotheses would explain the NE-SW shortening indicated by GNSS and earthquake focal mechanisms (Hyndman et al., 2005; Leonard et al., 2007, 2008; Mazzotti et al., 2013); however, it is likely that the low seismic velocity and inferred rheologically weak lithosphere beneath the MM is key to localized uplift.

4 Conclusion

To the west of the CDF, we find that V_S structures vary widely, in contrast to earlier lower-resolution studies that showed uniformly low velocities to the west and high velocities to the east. Most notably, we imagine a plume-like low-velocity (∼−5%) anomaly that extends from the asthenosphere to the upper crust. This probably contributes to the current uplift of the Mackenzies, and could indicate return flow related to lithospheric delamination produced by the opening of the slab window at 20–30 Ma (Thorkelson et al., 2011). However, the nature of the “plume”, be it nonmagmatic fluids, partial melt, or high temperatures, is at present undetermined, but the complete lack of Neogene-Quaternary volcanism and magmatism in the Mackenzies (Edwards & Russell, 2000) would seem to preclude an extensive magmatic component. We also image low seismic velocities in the lower crust that extend from the MMs to the Gulf of Alaska, and which are consistent with NE translation of the crust along a lower-crustal décollement system driven by Yakutat block convergence, and with the orogenic float hypothesis (Mazzotti & Hyndman, 2002). Imaged low velocities are also consistent with low viscosity and relatively deformable lithosphere beneath the MMs, as inferred by Finzel et al. (2014). Finally, we also confirm the finding by Estève et al. (2020a) of a lithosphere-scale high-velocity feature translated by the Tintina fault.

Acknowledgments

This project was supported by NSF EarthScope awards 1460536 and 1460533. Rob Anthony, Michael Baker, Sharon Busby, Julien Chaput, Hugh Harper, David Heath, Richard Karstens, Max Kaufman, Stefan Krogseng, Brandon Rassmussen, Nealy Sims, John West, Derek Witt, and Nesha Wright assisted with fieldwork, and the authors particularly appreciate the work of Michael Baker and Derek Witt in assisting with all aspects of the experiment. Brandon Rassmussen was supported by the IRIS Intern REU program (NSF EAR-1156739). The authors thank the Yukon and Northwest Territories Geological Surveys for their logistical support and advice, the Yukon Geological Survey for making their data publicly available, and Y. Jeff Gu and colleagues at the University of Alberta for invaluable staging assistance. Our instruments were deployed on the traditional territories of many First Nations peoples; the authors wish to acknowledge and thank these communities for their essential assistance. Seismic instruments, training, and field assistance were provided by the IRIS PASSCAL Instrument Center at New Mexico Tech. The facilities of the IRIS Consortium are supported by the National Science Foundation under Cooperative Agreement EAR-1851048 and the DOE National Nuclear Security Administration. Yukon University provided critical staging and lodging in Whitehorse. North Wright Air and Alpine Aviation provided essential logistical services in Yukon and Northwest Territories.

Open Research

Data Availability Statement

The facilities of IRIS Data Services, and specifically the IRIS Data Management Center, were used to access waveforms, metadata, and data products used in this study (https://doi.org/10.7914/SN/7C_2015). IRIS Data Services are funded through the Seismological Facilities for the Advancement of Geoscience and EarthScope (SAGE) Proposal of the National Science Foundation under Cooperative Agreement EAR-1851048. GPS data from this project are archived at UNAVCO (https://doi.org/10.7283/T58G8J34).