The Thermal Regime of NW Canada and Alaska, and Tectonic and Seismicity Consequences

NW Canada and Alaska are the continuation of the North American Cordillera through Mexico, western USA and western Canada. I show that they have similar thermal regimes and thermal control of tectonics and seismicity. I first summarize the multiple constraints to crust and upper mantle temperatures and then discuss some consequences. There are bimodal crust and upper mantle temperatures characteristic of most subduction zones: cool forearc, uniformly hot backarc (Yukon Composite Terrane to Southern Brooks Range, and Mackenzie Mountains), and stable cratonic backstops (Arctic Alaska Terrane and Canadian Shield). The main constraints are as follows: (a) Heat flow measurements, (b) Temperature‐dependent upper mantle velocities and seismic attenuation, (c) Temperature‐dependent topographic elevations; thermal isostasy, (d) Depth and temperature of the seismic lithosphere‐asthenosphere boundary (LAB), (e) Origin temperature and depth of craton kimberlite xenoliths, (f) Geochemically inferred source temperature and depth of recent volcanic rocks. (g) Depth to the magnetic Curie temperature, (h) Depth extent of seismicity. The backarc lithosphere is thin, LAB at 50–85 km and ∼1,350°C ± 25°C. Moho temperatures at 35 km are 850°C ± 100°C compared to cool cratonic areas of 400°C–500°C. The consequences include the following: (a) Thin and weak backarc lithosphere that accommodates pervasive tectonic deformation indicated by wide‐spread seismicity and GPS‐defined motions, in contrast to the stable cratonic regions; (b) Weak backarc lower crust that flattens the Moho and allows detachment and thrusting of the upper crust over the cold strong Arctic Alaska Terrane and Canadian Shield. This article provides a model for how to estimate deep temperatures from multiple constraints.

. Alaska and northwestern Canada topography and tectonic elements; the hot backarc and cold cratonic Arctic Alaska Terrane and Canadian Shield. The blue and red dotted lines demarcate the cold forearc, hot backarcs, and cold cratonic areas from this study.

Figure 2.
Schematic cross-section of hot backarcs illustrating the thin lithosphere and lower crust ductile detachment compared to the thick cold craton (modified from Hyndman, Currie, & Mazzotti, 2005). The upper backarc crust overthrusts the strong craton of the Arctic Alaska terrane at the Brooks Range and the Canadian Shield at the Mackenzie Mountains.
In southeastern Alaska and southwestern Yukon, there are very high near-coastal elevations in the St Elias-Chugach Range that sometimes have been associated with the Yakutat terrane collision. However, this terrane collision is located on the southern Alaska margin, some distance to the west and initiated later. An explanation for this immediate area is that the high elevations are a consequence of a small area of continental collision resulting from a change in Pacific-America motion of 15°-20° at ∼10 Ma (e.g., DeMets & Merkouriev, 2016). Oblique convergence, underthrusting, and uplift of the Queen Charlotte margin to the south started at this time (e.g., Hyndman, 2015b). This reorientation resulted in a small rotation and landward jump in the Pacific-America transform fault boundary, and the initiation of the Fairweather fault system just inland of the margin. A sliver of the continent was transferred from the American to the Pacific plate. This sliver that may have been part of the former cold forearc is now moving north and colliding with the southern Alaska margin, producing the thick crust and high elevations. This interpretation is consistent with the main St Elias-Chugach range uplift starting at 5-10 Ma (e.g., Enkelmann et al., 2009;von Huene et al., 1987).
To the east in the Yukon and western Northwest Territories, the seismicity and GPS relative motions show that deformation is complex, continuing far inland toward the edge of the craton at the Mackenzie Mountains thrust front and along the north-south mainly transcurrent Richardson Mountains extending to the margin of the Beaufort Sea (e.g., Cassidy et al., 2005;Estève et al., 2022;Hyndman, Flück, et al., 2005;Leonard et al., 2007Leonard et al., , 2008Marechal et al., 2015). As in the Alaska Brooks Range, the motion inland from the coastal collision has been interpreted to be transferred on a horizontal ductile detachment in the hot and weak backarc lower crust ( Figure 2). This strain transfer results in thrusting of the upper crust and foreland sediments to the east over the adjacent cold strong craton (e.g., Mazzotti & Hyndman, 2002). I show below that these tectonic deformations and seismicity are strongly controlled by the differences in lithosphere strength, which in turn results from the different thermal regimes of the crust and upper mantle (Figures 2 and 3).
The Pacific continental margin of the Yukon and northern British Columbia is a former subduction zone cut off by plate reorganization and the initiation of the Queen Charlotte-Fairweather transform fault system in the Eocene, ∼43 Ma (e.g., Engebretson et al., 1985;Hyndman & Hamilton, 1993;Schellart et al., 2010). The decay of the characteristic backarc high temperatures is sufficiently slow that the high temperatures in the crust and upper mantle in this area are indistinguishable from those of the current active subduction zones. The decay after termination of subduction is estimated to be 300-500 m.y. (e.g., Chapman & Pollack, 1975;Currie & Hyndman, 2006;R. C. Porter et al., 2019;Sleep, 2005), so there has been little cooling since subduction was cut off on this margin. Also, slab windows may delay the initial cooling (e.g., Thorkelson et al., 2011), and there may be initial extension (Batir & Blackwell, 2020).

Thermal Constraints and Transients
In this article, I first summarize the most important temperature constraints in the deep crust and upper mantle, and then describe some of the consequences of the laterally variable temperature distribution across the region. The eight main constraints include: (a) Geothermal heat flow measurements (and crustal radioactive heat generation) and inferred deep temperatures, (b) Temperature-dependent seismic velocities in the upper mantle, including the velocity of the Moho seismic Pn phase, and seismic attenuation. (c) Thermal isostasy, the thermally controlled elevation-crustal thickness relation. (d) The seismic depth to the base of the backarc lithosphere-asthenosphere boundary (LAB) which is at a constrained temperature (craton LAB depths are often poorly constrained). (e) The depth (pressure) and temperature at the origin of craton mantle kimberlite xenoliths and some crustal xenoliths (Cordillera mantle xenoliths usually give temperatures but limited constraints on origin depths). (f) Geochemical constraints to source equilibration temperatures and depths of recent volcanics, concluded to be at the LAB. (g) The temperature-controlled depth of the magnetic Curie temperature. (h) The thermally controlled depth extent of seismicity. For most of the constraints, I give estimates of the Moho temperature, which strongly constrains crustal temperatures, if there is a thermal steady state. As discussed below, the laterally variable thermal regimes often do not follow the principal geological terrane boundaries. In a few areas, thermal transients may be important. Most crustal thermal processes, including surface sedimentation and erosion, have time constants of a few 10's of m.y. (e.g., model analyses by Ehlers, 2005). The time constant for crustal temperatures from thinning of a thick stable lithosphere to thin backarc thicknesses is up to ∼50 m.y. (e.g., modeling by . However, the time constant for cooling from a thin lithosphere (hot backarc mobile belt) toward a stable craton is much longer, requiring 300-500 m.y. from heat flow and mantle velocity data and thermal modeling (Chapman & Pollack, 1975;Currie & Hyndman, 2006;R. C. Porter et al., 2019;Sleep, 2005). The thermal transients involved in rapid thinning of the lithosphere have been described for the Sierra Nevada in USA and Baja California in Mexico (e.g., Erkan & Blackwell, 2009), and suggested for an area of the northern Canadian Cordillera . Bao et al. (2014) proposed that the lithosphere in the Cordillera in southern British Columbia thinned in the Eocene, ∼50 Ma. However, Canil and Russell (2022) provided evidence that the lithosphere was thin much earlier. For most other areas of our study, the shallow (e.g., surface heat flow), intermediate (e.g., Curie temperature, effective elastic thickness, and depth extent of seismicity) and deep (e.g., upper mantle velocities and LAB depths) thermal constraints are in agreement within the uncertainties and steady state is likely a reasonable approximation, but thermal transients warrant further study.

Temperatures From Heat Flow
A key deep temperature constraint comes from geothermal heat flow measurements combined with estimates of crust and upper mantle radioactive heat generation and thermal conductivity. The methods of heat flow determination and the sources of error in their use have been described extensively (e.g., Beardsmore et al., 2001;Eppelbaum et al., 2014;Jessop, 1990). The downward extrapolation of heat flow data to estimate temperatures in the deep crust and upper mantle has been discussed in detail by Hasterok and Chapman (2011) and Goes et al. (2020), including the role of radioactive heat generation, especially in the upper crust, and thermal conductivity.
There are a limited number of Alaska-NW Canada heat flow measurements, but they are sufficient to define the regional patterns. In Alaska, the heat flow compilation and interpretation by Batir et al. (2016) delineate the different hot and cold Alaska thermal regimes (Figure 4a). More details are provided in an associated report (Batir et al., 2013). The data in NW Canada have been provided by Lewis et al. (2003), Majorowicz and Grasby (2010a, 2010b, and Grasby et al. (2012); see also the global data base by Fuchs and Norden (2021) (Figure 4b).  Batir et al., 2016). The hot backarc extends from the cold forearc and the hot volcanic arc to the Brooks Range and the cold Arctic Alaska terrane. (b) Contoured heat flow in NW Canada. The Wopmay orogen is cool and stable but has high upper crust radioactive heat generation, so high heat flow (after Blackwell et al., 1991;Lewis et al., 2003;Majorowicz & Grasby, 2020).
The data show the characteristic high backarc heat flow and mainly lower heat flow in the adjacent cratonic areas. In places, the detailed boundary between the hot and cold thermal regimes is somewhat complex and poorly defined. A primary difference in temperature control between the hot backarc Cordillera and the adjacent cool stable areas is the lithosphere thickness and resulting difference in mantle heat flux. However, generally of second order importance are variations in crustal thickness and heat generation (e.g., Hasterok & Chapman, 2011). The exception is the stable Wopmay orogen of the Canadian Shield, which has very high upper crust heat generation and resulting high heat flow but inferred cool deep temperatures, as discussed later. A number of other sources or sinks of heat have been proposed to play a limited role, including frictional heating and metamorphic reactions.

Heat Flow Pattern in Alaska
The Alaska forearc has low heat flow due to underthrusting of the cool upper oceanic plate, with an average of ∼40 mW/m 2 . The heat flow measurements in the region of the volcanic arc are irregular with some very high values, but most are about 80 mW/m 2 . The values for the backarc interior of Alaska between the Alaska Range and the Brooks Range vary between 61 and 106 mW/m 2 with most values between 70 and 80 mW/m 2 , indicating the continuation to the northwest of the margin-parallel backarc high heat flows in the Cordillera to the south (e.g., Blackwell et al., 1991). Most of the heat flow variability in the backarc likely arises from variations in upper crust radioactive heat generation that have only a limited effect on deep temperatures, since other indicators show quite laterally constant high temperatures in the upper mantle. In the Arctic Alaska terrane, the heat flows are low, as expected for this stable cratonic terrane. A few values in the coastal North Slope are a little higher, perhaps because of the residual heat from the original margin rifting.

Heat Flow Pattern in the Backarc of NW Canada
In the coastal zone of the Alaska Panhandle, there are a few low heat flows that may represent part of the former cool forearc before it was cut off from subduction by the Fairweather-Queen Charlotte Fault system at about 43 Ma. The low heat flows are comparable to the low values found to the south in the former forearcs of the Sierra Nevada of California and Baja California (e.g., Erkan & Blackwell, 2009). Most of the heat flow values are high in the Cordillera backarc of NW Canada (Fuchs & Norden, 2021;Lewis et al., 2003;Majorowicz & Grasby, 2010a, 2010b, showing the continuation of the high heat flow that extends throughout the North American Cordillera (Figure 4b). For central and southern British Columbia, data have been provided by Lewis et al. (1992) and Hyndman and Lewis (1995), for northwestern USA by Blackwell et al. (1982), Blackwell, Steele, Kelley, and Korosec (1990), Blackwell, Steele, Frohme, et al. (1990), and in the USA further to the south by Blackwell et al. (1991). Such regional high heat flow and inferred crust and upper mantle high temperatures are characteristic of the 200-1,000 km wide continental subduction backarcs globally (Currie & Hyndman, 2006;Hyndman, Currie, & Mazzotti, 2005). Much of the local heat flow variation is due to the critical variable of heat generation in the upper crust that may not greatly affect deep temperatures. The high heat flows include regions inboard of the former subduction zones cut off by the Queen Charlotte Fault Zone in the Eocene and to the south by the San Andreas Fault zone in the Miocene. As noted above, these backarcs have not significantly cooled since subduction was cut off.
In the northernmost Cordillera, there is one area of low heat flow associated with the Mackenzie craton that extends the Canadian Shield westward to near the Alaska-Yukon border. However, the data are inadequate to define its area well.
In Figure 5, I provide steady state Cordillera temperature-depth profiles for a characteristic backarc heat flow of 75 mW/m 2 with several average upper crust heat generations based on the detailed analyses of Hasterok and Chapman (2011) (e.g., Figure S2 in their Supplement). They deal with variations in thermal conductivity, which has a quite small effect for a reasonable range, and how surface heat flow affects mantle heat flow. For a heat 6 of 33 flow of 75 mW/m 2 , their preferred parameters of average heat generation and thermal conductivity with depth give a 65 km LAB, as shown in our figure. For a reasonable range of parameters as they discuss and a 1,350°C LAB, there is a range of less than 25°C at the Moho. For this regional average heat flow and a 35 km backarc crustal thickness, the Moho temperature is ∼850°C for common 10 km upper crust heat generation of 1.5 μW/m 3 , ∼750°C for 3 μW/m 3 (i.e., more in crust, less in mantle) and ∼950°C, for 0.75 μW/m 2 (i.e., less in crust, more in mantle). A steady state Moho at above 950°C is unlikely since lower crust melting is then expected (e.g., Collins et al., 2020). For the other limiting case of fixed upper crust radioactive heat generation, variations in surface heat flow indicate variable mantle heat flow and deep temperatures, and a depth to the LAB that is directly correlated with the surface heat flow. For the geotherms shown, the heat flows are approximately 80, 75, 68, and 60 mW/m 2 . The Moho temperatures are correspondingly lower for especially thinner crust, that is, 700°C for the inferred ∼25 km thick crust of parts of central Alaska.
The heat flows in much of the Cordillera backarc of NW Canada, north of 59°N are unusually high, ranging from 60 to 138 mW/m 2 with an average of 105 ± 22 mW/m 2 in crystalline rocks (Lewis et al., 2003), and are similarly high in associated sedimentary basins (Majorowicz, 1996). To the south between 57 and 59°N, the average is a more normal Cordillera value of 73 ± 11 mW/m 2 . The high heat flows north of 59°N are associated with measured high near-surface radioactive heat generation (U, Th, K); therefore, deep temperatures may not be much higher than the Cordillera backarc average (Lewis et al., 2003). The average measured surface heat generation in the area is very high, 4.4 ± 1.9 μW/m 3 , in contrast to a more normal Cordillera average of 1.6 ± 0.8 μW/m 3 just to the south. For a 10 km upper crust layer, this high heat generation gives an extra 28 mW/m 2 , compared to the south, for an expected heat flow of 101 mW/m 2 , similar to that observed. Very high near-surface heat generation has limited effect on deep temperatures compared to deeper heat generation, but does imply somewhat higher than normal backarc Moho temperatures, ∼900°C at the Moho .
One indication that deep temperatures may be unusually high is the very thin, ∼50 km lithosphere concluded in this region based on seismic receiver functions . Such thin lithospheres imply very hot temperatures in the deep crust and likely melting if there is thermal steady state, and the authors suggest that there may have been a short-term transient lithosphere thinning such that the lower crust is not yet hot. However, there are no recent volcanics in the region, as might be expected for such a recently thinned lithosphere. Also, deformation and exhumation in the region indicative of a hot and weak lithosphere occurred during the Late Cretaceous to early Eocene with some continuing to the Eocene-Miocene (Enkelmann, 2020;McKay et al., 2021). To the south Canil, Russell, and Fode (2021) found that volcanic xenolith data indicate slow cooling rates that are more consistent with the thin hot mantle lithosphere of the Canadian Cordillera being ancient, maintained for at least the past 30 Myr, and not cooled from asthenosphere temperatures in the past ∼50 Myr. The thin lithosphere in the region inferred from seismic receiver functions remains a puzzle.

Heat Flow and Temperatures in the Craton of Northwestern Canada; the Wopmay Orogen
For most of the stable areas of the northwestern Canadian Shield, regional average heat flows of 40-55 mW/m 2 give temperatures at a 35-40 km Moho of 400°C-500°C. Heat flow further to the east in the Yellowknife area of the Archean Slave Craton has a mean of 53 mW/m 2 , implying low lithosphere temperatures in agreement with other thermal constraints (e.g., Jaupart et al., 1998;Mareschal & Jaupart, 2004;H. K. C. Perry et al., 2006;C. Perry et al., 2010). However, immediately to the east of the Cordillera hot backarc, the region of the Paleoproterozoic Wopmay orogen ( Figure 4b) has a remarkably high heat flow for such an old and stable region (e.g., see global stable areas in database by Fuchs & Norden, 2021). The average value of Lewis et al. (2003) is 90 ± 15 mW/m 2 at sites in crystalline rocks, north of 59°N, most between 60°N and 61°N with lower more normal heat flows to the south. Similar high heat flows were found in hydrocarbon exploration and development wells in the sedimentary basins of the area by Majorowicz (1996) with estimated basement heat flows of greater than 80 mW/m 2 . Such basins are subject to thermal disturbance from the vertical component of subsurface water flow, but there is no significant correlation with the topography or hydraulic head patterns and the measurements are concluded to be reliable estimates of the deeper heat flow.
A number of other thermal constraints, as discussed below, show a cool thick lithosphere, including upper mantle seismic velocities, and elevation-crustal thickness (thermal isostasy), so the high heat flows must represent mainly upper crust processes. The likely explanation is that the measured high near-surface heat generation results in high heat flow but does not represent the cool deep crust and upper mantle temperatures (e.g., Hasterok 7 of 33 & Gard, 2016). The near-surface heat generation values of Lewis et al. (2003) are very high north of 59°N, 4.8 W/m 2 in contrast to the south of 59°N, where the average is more normal. They concluded that the observed heat flow variation is primarily due to variations in upper crust heat generation. However, deep temperatures may be somewhat higher than typical for the craton to the east, 550°C-700°C at the Moho (e.g., Flück et al., 2003).
There is an interesting correspondence noted by Hasterok and Gard (2016) between the northern Canada Wopmay orogen and the northeastern Australia Proterozoic region, which has similar high heat flows for a stable terrane, and where there are other indicators of a thick cool lithosphere (McLaren & Powell, 2014;McLaren et al., 2003). The Australian region also has an upper crust with very high radioactive heat generation. The two areas of northeastern Australia and Laurentia of western Canada have been concluded to have been adjacent prior to the breakup of supercontinent Nuna between about 1.3 and 1.2 Ga (e.g., Kirscher et al., 2021). The two areas may have a common origin for the upper crust high radioactive heat generation before the supercontinent breakup.

Temperature Constraints From Mantle Seismic Velocity
For regional mapping of deep temperature variations over Alaska and northwestern Canada, an important estimator is the temperature-dependent seismic velocity in the upper mantle. Within the continental crust, seismic velocities are controlled mainly by rock compositions, generally slow in the felsic high silica upper crust and faster in the mafic lower silica deep crust. The effect of varying temperatures on crustal velocities is relatively small, less than 2%, ∼0.1 km/s, for the range of temperatures in the lower crust of backarcs and stable areas (e.g., Christensen & Mooney, 1995). In contrast, in the upper mantle where composition variations are smaller, velocity is controlled mainly by temperature. Higher temperatures give lower velocities (e.g., Goes et al., 2000;R. C. Porter et al., 2019;Priestley & McKenzie, 2006). Composition variations have a much smaller effect. One composition effect that could be important is the degree of melt depletion in the mantle by removal of basaltic partial melt, the degree it is refractory. However, D. Schutt and Lesher (2006) and Priestley and McKenzie (2006) showed that the degree of mantle depletion has very little effect on mantle velocity. The degree of hydration may also be important (e.g., Lowry et al., 2021), although volcanic xenoliths indicate a generally quite dry lithosphere (e.g., Demouchy & Bolfan-Casanova, 2016;Kilgore et al., 2018). In contrast, in the asthenosphere, even a small concentration of partial melt can have a large velocity effect (e.g., Hammond & Humphreys, 2000;Yamauchi & Takei, 2016).
Velocity-temperature relations are somewhat nonlinear at higher temperatures, so absolute velocities must be used, not deviations from a reference velocity-depth model. The estimated temperatures depend on two factors, the velocity-depth models (allowing for anisotropy) and the velocity-temperature relations that allow for the composition, grain size, volatiles and, at high temperatures, for partial melt. An important calibration discussed below is the independently determined temperatures and depths at the base of the lithosphere (the LAB) from the geochemistry of recent volcanics and the depth of the LAB from seismic receiver functions (e.g., Hyndman & Canil, 2021b;Plank & Forsyth, 2016).
Most upper mantle seismic velocity data have come from large-scale shear wave tomography using distant earthquake and ambient noise sources (e.g., Babikoff & Dalton, 2019). An excellent first order mapping of upper mantle seismic velocities and spatially smoothed temperatures has been provided by global tomography analyses (e.g., Schaeffer & Lebedev, 2014;Tesauro et al., 2014) (Figure 3) (also temperatures by Priestley & McKenzie, 2006). The hot backarc (and arc) compared to cold cratonic Arctic Alaska terrane and Canadian Shield is well defined by these upper mantle velocities. There are now much more data allowing higher local resolution. For comparison with other areas, teleseismic tomography velocities using the USArray/EarthScope grid and other stations in western USA have been reported by numerous authors (e.g., Hansen et al., 2015 and comparisons therein) and some resultant crust and upper mantle temperatures. I provide only a first order analysis of the Vs maps, but the systematic difference in upper mantle velocities and inferred temperatures between the hot backarc and the adjacent stable cratonic area is clear. There is a need for more detailed and rigorous conversion of velocity to temperature and for mapping upper mantle temperatures, for example, by the velocity to temperature relations by Yamauchi and Takei (2016), Priestley and McKenzie (2006), and Cammarano et al. (2003).
In Alaska, seismic velocity data from several profiles provide thermal constraints, especially the Trans-Alaska Crustal Transect (TACT), which involved a detailed explosion-source wide-angle P-wave seismic survey that extended 1,350 km north-south from the Aleutian Trench to the Arctic coast, with a series of branch lines. The 8 of 33 results have been extensively analyzed and interpreted in terms of the parameters that can be used to estimate deep temperatures, including uppermost mantle Pn velocity, and crustal thickness for thermal isostasy estimation (e.g., Fuis et al., 1991Fuis et al., , 1997Fuis et al., , 2008Moore et al., 1997;Wissinger et al., 1997). The Broadband Experiment Across the Alaska Range (BEAAR) project operated seismic stations across the Alaska Range in southern Alaska that give crustal thicknesses and upper mantle S-wave tomography upper mantle velocities (Veenstra et al., 2006). Finally, the major Alaska EarthScope/USArray program provided 3-D passive broad band seismic data over most of Alaska and part of adjacent NW Canada, involving a ∼75 km spacing array of 200 long-period seismic stations operating for 3 or more years (e.g., Miller et al., 2020). This array allows constraints to temperatures over the entire region, especially through tomography S-wave velocity-depths and crustal thicknesses. There has also been an irregular network of ongoing seismic stations operated mainly for seismic hazards. These various station arrays allow three main seismic methods to map crust and upper mantle velocities and the crustal thickness, and some data for lithosphere thickness: (a) earthquake surface wave tomography defining shear and compressional wave velocity depth, especially for mantle temperatures; and Pn, (b) seismic receiver function mapping of the Moho and the LAB, (c) ambient noise tomography mapping of crustal thickness, and crust and uppermost mantle velocities.
Upper mantle S-wave velocities in Alaska from teleseismic earthquake tomography have been provided by Martin-Short et al. (2018), Feng & Ritzwoller, 2019;Berg et al. (2020), Nayak et al. (2020), and Gama et al. (2021). In NW Canada, S-wave tomography velocities have been provided by Estève et al. (2020). P-wave teleseismic body wave tomography has been provided by Mercier et al. (2009) and Estève et al. (2021). D. Schutt et al. (2021) and Kao et al. (2013) gave results from ambient noise tomography. I show the velocity maps of Berg et al. (2020), and D. Schutt et al. (2021; the results of Estève et al. (2021) are similar) at 60 km in Figures 6a and 6b. This depth is generally in the lithosphere where partial melt is not expected but close to the backarc LAB and gives a reasonable spatial definition. The inferred temperatures may be slightly lower than those at the LAB. In Alaska, there is a well-defined upper mantle low velocity backarc. In NW Canada, only the western part and two profiles have adequate data for good resolution. The thermal structure is complex and the detailed backarc-craton boundary in NW Canada is poorly constrained by this method (Figure 6b). Estève et al. (2020) showed a better continuity of low and high velocity areas across the Tintina Fault if the estimated ∼430 km displacement is restored. In the southern Yukon, adjacent Northwest Territories, and northern British Columbia, two multidisciplinary active-source seismic structure transects were carried out by the Lithoprobe SNORCLE program (e.g., Clowes et al., 2005), including Vibroseis multichannel deep seismic reflection and explosion-source seismic refraction/wide angle surveys that gave crustal thicknesses and temperature-dependent upper mantle Pn velocities. Broadband seismic recording was carried out along a line across the northern Canadian Cordillera, the Mackenzie Mountains EarthScope Project (MMEP) seismic array line (e.g., Baker et al., 2020), and the CANOE seismic array lines to the south (e.g., Audet et al., 2020;Schaeffer & Bostock, 2010), and at a number of other temporary sites Dalton et al., 2011;Tarayoun et al., 2017). The long-term earthquake network stations have provided additional data (e.g., Hyndman, Cassidy, et al., 2005). Upper mantle shear wave velocities have also been obtained from ambient noise tomography, although the method is especially sensitive to crustal velocities. From this method, crustal thickness and uppermost mantle velocities have been estimated in NW Canada and in the adjacent craton (Kao et al., 2013;.

Hot Backarcs and Low Upper Mantle Velocities
From the tomography data, I look especially at the depth range in the backarc of 60-100 km, which is mainly in the backarc lowermost lithosphere and in the upper asthenosphere where there is nearly constant adiabatic temperature. The effect of small concentrations of partial melt in the asthenosphere is needed to explain the lowest velocities. The temperatures are therefore constrained to the mantle solidus at depths of these low velocities. The shear wave velocity maps from these authors give a range for the backarcs, of 4.15-4.25 km/s. Using the velocity to temperature conversions, especially of Yamauchi and Takei (2016) (see also Cammarano et al., 2003;Faul & Jackson, 2005; Goes & van der Lee, 2002;Priestley & McKenzie, 2006 and discussions by Ball et al. (2019), Tesauro et al. (2014), and Klöcking et al. (2018), the inferred asthenosphere temperature at these depths almost everywhere in the backarc is 1,350°C ± 50°C, which gives a potential temperature T p extrapolated to surface pressure, of 1,325°C ± 50°C. These temperatures are very similar to the western Cordillera of the USA, excluding the Yellowstone hot spot trace, which may be hotter, and the Colorado Plateau, which has recently thinned lithosphere. The velocities and inferred temperatures are very similar to those given in the compilation by Klöcking et al. (2018) of 1,350°C ± 25°C for four tomography velocity studies in western USA, away from the Yellowstone hot spot. For this temperature at the LAB, the steady state backarc temperature at a 35 km Moho is ∼850°C, and ∼700°C for a very shallow 25 km Moho as in parts of central-western Alaska ( Figures 5 and 7). In Alaska, there is a clear indication of partial melt where velocities are as low as ∼4.05 km/s centered at 100-120 km depth (e.g., cross-sections of Berg et al., 2020). A velocity of ∼4.1 km/s indicates 2%-4% partial melt using the velocity versus melt fraction relation of Yamauchi and Takei (2016). Therefore, we need to exclude depths of 80-140 km for accurate velocity to temperature conversion. The presence of partial melt constrains the temperature to be near the solidus at ∼1,350°C.

Cool Cratonic Areas and High Upper Mantle Velocities
For the cratonic Arctic Alaska Terrane, the mantle velocities at depths of 60-100 km mostly range from 4.50 to 4.65 km/s ( Figure 6a) (Berg et al., 2020;Gama et al., 2021;Martin-Short et al., 2018;Nayak et al., 2020). In NW Canada, D. Schutt et al. (2021) and Estève et al. (2021) found very similar velocities of 4.45 to 4.65 km/s for the Canadian Shield adjacent to the northern Cordillera. Using the relation of Yamauchi and Takei (2016), these velocities in the two cratonic regions give low temperatures but over a considerable range of 500°C-900°C at these depths, and Moho temperatures of 400°C-600°C. These temperatures are in general agreement with the temperature estimates from geothermal studies at 80-100 km (e.g., C. Perry et al., 2010) and kimberlite xenoliths in the craton to the east of the northern Cordillera of 600°C-750°C (Figure 7) (e.g., Canil, 2008;Kopylova & Caro, 2004). The implied cratonic temperature at a 35-40 km Moho is 400°C-500°C.  2009)). Cordillera xenoliths from Greenfield et al. (2013); craton xenolith temperatures from Canil (2008) and Kopylova and Caro (2004).

Temperatures From Pn Velocities
The temperature at the Moho may be estimated from the refracted head wave Pn velocities just below the Moho. Most Pn velocities come from explosion-source wide-angle seismic surveys, but recently also with seismic array data. The velocities are generally similar over substantial areas, but there are local variations, likely because of seismic anisotropy, by Moho topography, and where the Moho velocity contrast is over a substantial thickness relative to the wavelengths used (e.g., Kao et al., 2013). D. L. Schutt et al. (2018) developed a method to estimate Pn from USArray grid seismograph data and mapped the Moho temperatures in western USA, but results from this method have not yet been reported for Alaska or NW Canada. A correlation between Pn velocity and heat flow has been recognized for many years, with low velocities in tectonically active hot areas, and high velocities in stable cratonic areas; for North America (Black & Braile, 1982), for India (Kubik et al., 1988), and for a global compilation (Sharma et al., 1991). These relations may be used to estimate crust and upper mantle temperatures. Pn velocities from the wide-angle refraction compilation of Christensen and Mooney (1995) and from the station array analysis of Buehler and Shearer (2017) are about 7.9 km/s in the North America Cordillera to the south and other tectonically active regions and 8.15 to 8.2 km/s in stable cratonic areas. These velocities may be converted to temperature (e.g., H. K. C. Perry et al., 2006;D. L. Schutt et al., 2018), giving inferred Moho temperatures of 700°C-900°C in backarcs and 450°C-700°C in cratonic areas, with significant uncertainties. The latter is a little higher than some other estimates for cratonic Mohos.
In Alaska, Fuis et al. (1991Fuis et al. ( , 1997 found that the cool forearc (Peninsular and Wrangellia terranes) have Pn velocities of 8.l-8.3 km/s. Most of the hot backarc has Pn velocities of ∼7.9 km/s. In contrast, the cool cratonic Arctic Alaska terrane has Pn velocities of ∼8.2 km/s. These velocities correspond to Moho temperatures of about 800°C and 500°C. Wissinger et al. (1997) found Pn in the Brooks Range at the northern edge of the backarc that are not well constrained but approximately 7.99 km/s, with a transition to the northern Brooks Range, of 8.12 km/s, for a Moho temperature of ∼550°C.
In northwestern Canada, from the detailed Lithoprobe SNORCLE explosion source transects, Welford et al. (2001), Fernández-Viejo et al. (2005), and Clowes et al. (2005) found Pn in the backarc mostly at about 7.9 km/s, which gives ∼850°C at the Moho using the above referenced relations. This corresponds to a surface heat flow of 75-85 mW/m 2 . In the adjacent stable areas, east of the Tintina Fault, the Pn velocities are mostly 8.2 km/s, that is, ∼450°C at the Moho. In both the northwestern Canada and Alaska, the Pn velocity data define the division of hot backarc and cold cratonic terranes reasonably well, although the absolute temperatures have substantial uncertainty. Further array processing should provide better Pn mapping, especially from USArray data in Alaska and western Yukon.

Temperatures From Mantle Seismic Attenuation
Another low resolution but reliable indicator of the areas with high temperatures in central Alaska is seismic attenuation (e.g., Stachnik et al., 2004). High temperatures are associated with high attenuation, low Q. Areas of very low Q likely indicate partial melt in the upper asthenosphere (e.g., Soto Castaneda et al., 2021), in agreement with some exceptionally low shear wave velocities. The attenuation boundary on the BEAAR profile from the high Q cold forearc to their low Q "hot wedge" backarc to the north is near the Denali Fault (and adjacent Hines Creek fault). Their estimate for the depth to the LAB is ∼80 km which is within the range of receiver function estimates. However, as discussed above, this depth may be the top of the partial melt zone, not the LAB, which may be shallower. Their asthenosphere temperature of ∼1,250°C in the Yakutat collision zone is ∼100°C lower than most other estimates that they report in Andes, NE Japan, Tonga, and lower than western USA and Canada from other LAB temperature estimates (Hyndman & Canil, 2021a;Plank & Forsyth, 2016). This unusually low temperature estimate could be related to why there is no arc volcanism in this area.

Thermal Isostasy: Regional Elevation and Crustal Thickness
Thermal isostasy relating elevation to deep temperatures, allowing for crustal thickness and density, should be a good method for detailed mapping the boundary in deep temperatures between the hot backarc and cool cratonic areas. I am not aware of it being used. The Cordillera-craton difference in average crust and upper mantle temperatures between the two regions results in a systematic topographic elevation difference (e.g., Artemieva, 2019;Hasterok & Chapman, 2011;Hasterok & Gard, 2016;Hyndman & Currie, 2011). There is a large difference for the same crustal thickness. I provide only a rough analysis; a detailed rigorous study will require substantial work but should be very productive. As I noted above, unusually good data are available from EarthScope-USArray and other seismic structure studies.
The three main contributions to topographic elevation are as follows: (a) Crustal thickness; thick crust floats higher than thin crust, Airy isostasy, (b) Average crustal density; light crust floats higher than dense crust, Pratt isostasy, and (c) The temperature effect on crust and upper mantle density; high temperatures result in lower density by thermal expansion and therefore in higher elevations, thermal isostasy (e.g., Hasterok & Chapman, 2007;Hasterok & Gard, 2016). Dynamic motions in the mantle may also be important, but this contribution to elevation has not been resolved in our studies. We find that the elevation difference between the Cordillera backarc and the adjacent craton can be accounted for by thermal isostasy, and variations in elevation within these two areas can be accounted for by Airy and Pratt isostasy, within the recognized uncertainties (e.g., Hyndman & Currie, 2011) without the effect of mantle dynamic motions, especially for the hot backarc. Using the data and analyses of Hasterok and Chapman (2011), Hyndman and Currie (2011) found that, allowing for crustal thickness and density variations, the hot backarc Cordillera is consistently 1,600 m higher than the adjacent stable craton (Figure 8). Alternatively, a crustal thickness difference of about 12 km is required for the same elevation. The elevation difference corresponds well to that expected from thermal expansion density reduction, using the average temperature-depth relations to a depth of ∼200-250 km where the two temperature profiles converge (Figure 7). The temperature difference is related to the lithosphere thickness difference of the western USA Cordillera ∼65 km compared to craton 200-250 km. A similar thermal isostasy effect using heat flow data was found by Hasterok and Chapman (2011) and Hasterok and Gard (2016).
It is important to average the topography and crustal thicknesses over appropriate horizontal spatial wavelengths since the crust/upper mantle to the depth of the brittle-ductile transition acts as an elastic plate (or equivalent plate if several strong layers) with a characteristic horizontal spatial scale. The scale can be estimated from the effective elastic thickness, "Te" (from gravity-topography coherence or admittance), giving horizontal lengths of 100-200 km for the hot backarc, and as much as 1,000 km for the cratonic stable areas (e.g., Flück et al., 2003). In this initial study, for lateral crustal density variations (Pratt isostasy), I have used the approximate relation between crustal thickness and average density from Hyndman and Currie (2011; their Supplement, Figure A4) rather than densities from the average crustal velocities. The remaining parameter required for isostasy is crustal thickness, which is becoming well defined for our study region.

Alaska Crustal Thickness and Thermal Isostasy
For Alaska, the crustal thickness has been mapped in detail and is becoming well constrained by wide-angle explosion source surveys, for example, TACT line N-S from Pacific to Arctic ocean and other seismic refraction surveys (Fuis et al., 1991(Fuis et al., , 1997(Fuis et al., , 2008Wissinger et al., 1997), from teleseismic earthquake and ambient noise tomography (Berg et al., 2020;Haney et al., 2020;Martin-Short et al., 2018;Miller & Moresi, 2018;O'Driscoll & Miller, 2015;Y. Zhang et al., 2019). The depth resolution is commonly 1-3 km, and horizontal spatial resolution ∼75 km for Alaska and easternmost Yukon, with lower resolution to the east. To the north, there is thick crust, 40-50 km, and 1,000-1,500 m elevations under the northern Brooks Range, which is mainly stable cratonic, and 35-40 km crust and low elevation, 0-200 m under the cratonic Arctic Alaska Terrane. This is only a rough analysis, but it shows the bimodal relationships of the hot high backarcs compared to low cratonic areas for the same crustal thickness. A more thorough analysis is needed. I note that caution is needed because the average crustal velocity and crustal thickness estimates need to be independent. For example, crustal thickness from Moho reflection times requires an independent estimate of average velocity.
The central Alaska lowlands are unusual compared to the Cordillera to the south, having thin crust, 25-30 km and resulting low elevations in spite of the high temperature thermal expansion effect. No significant area with crust that thin exists to the south, and no Cordillera hot backarc area to the south has such low elevation. If the Alaska hot backarc region of thin crust (central YCT) had a cold thick lithosphere, this area would be well below sea level. In contrast, most of the hot backarc in NW Canada has high elevations and average ∼35 km crustal thicknesses, similar to the Cordillera to the south.
A local example of thermal isostasy contrasts is that the Moho is vertically offset by 10 to 15 km in the area of the central 600 km of the Denali fault (Allam et al., 2017) or adjacent Hines Creek fault (Y. Wang & Tape, 2014). These faults correspond well to the forearc-backarc thermal boundary. The Moho depth on the southern side is ∼35 km and the northern side ∼25 km with similar elevations. The 10 km crustal thickness difference across the faults approximates the ∼12 km crustal difference expected due to hot backarc versus cold cratonic thermal isostasy (Figure 8). At this boundary, no  lower crust flow is expected to smooth the Moho step because to the south the forearc crust is cold and strong with no weak lower crust layer. The thermal contrast also explains the N-S seismic structure profile across Alaska of Fuis et al. (1997) that has been interpreted to show low compensation of the thick crust in the Brooks Range relative to central Alaska to the south. The thick crust to the north is balanced by the high densities resulting from low temperatures.
The BEAAR experiment across the high Alaska Range in southern Alaska also found typical 35-45 km crust (up to 50 km locally on the TACT line; Brocher et al., 2004) in contrast to beneath the northern-central lowlands where the crust is exceptionally thin, ∼26 km (Veenstra et al., 2006). In Figure 9a, I show the regional crustal thicknesses for Alaska by Haney et al. (2020) using Rayleigh-wave dispersion (very similar by Jiang et al., 2018, and others). The thermal isostasy boundary agrees well with the heat flow boundaries. The Denali and adjacent Hines Creek faults define the southern boundary of the hot backarc (Figure 9a). The Kobuk Fault at the southern Brooks Range is the approximate northern boundary of the hot backarc with the stable craton. There is no crustal thickness change resolved across the different elements of the backarc, the Ocean Domain Terrane (ODT) and the YCT. The position of the Alaska hot and cold regions is shown on the crustal thickness versus elevation plot in Figure 8. There is a clear division between elevations for the same crustal thickness between the hot and cold regions.

NW Canada Crustal Thickness and Isostasy
For the western part of NW Canada, there is USArray grid coverage over the region near the Alaska border with a spacing of ∼70 km. To the east, the resolution is poorer, several 100 km except where there are two detailed profiles. In contrast to Alaska, the Yukon-western NWT Cordillera backarc and adjacent craton both have mostly uniform thickness crust, 32-35 km, with less than 2 km Moho topography (Figure 9b) (e.g., Audet et al., 2020;Kao et al., 2013;Tarayoun et al., 2017;Q. Zhang et al., 2021). One area of substantial crustal thickening and high elevations is at the Yakutat and Saint Elias-Chugach collision zones near the coast, thickness ∼50 km. There are also slightly increased crustal thicknesses under the higher elevations of the Richardson Mountains and the Mackenzie thrust front. The very thin crust and low elevations of central Alaska do not extend to the backarc Cordillera of NW Canada. This average thickness of ∼32-35 km and the small variability are similar to the backarc Cordillera in southern Canada, but in the south, the adjacent craton tends to have a thicker crust than the Cordillera (e.g., Kao et al., 2013) and the elevations are a little higher than to the north, ∼500 m. To the south in northern Alberta, the craton crustal thickness is ∼40 km (Figure 9b). In NW Canada, the thermal isostasy hot backarc generally extends east near the Alaska border to the Mackenzie Mountain thrust front and there are quite uniform high elevations, ∼1,500 m (e.g., Audet et al., 2020;Q. Zhang et al., 2021) (see Figure 1), with slightly higher elevations near the thrust front. There are lower average elevations in northern Yukon associated with the cold Mackenzie craton, the westward extension of the Canadian Shield. Much lower elevations, 100-200 m, but similar crustal thicknesses of 30-35 km are associated with the cold high-density crust and upper mantle of the adjacent Canadian Shield. Across the boundary between the hot backarc and cold craton, there is very little change in crustal thickness (Figure 9b). The elevation change is explained by the difference in upper mantle temperature ( Figure 10). Locally, the Mackenzie and Richardson Mountains also have slightly thicker crust and higher elevations (e.g., Audet et al., 2020;Kao et al., 2013). In northeastern British Columbia, there is a pronounced crustal thickening eastward from the thrust front of ∼35 to ∼45 km over a distance of 100 km, where the elevation is much lower to the east, in the wide-angle seismic refraction study of Welford et al. (2001). Thrusting of the Cordillera upper crust has loaded and bowed down the craton lithosphere similar to the Western Canada sedimentary basin in southwestern Alberta (e.g., Price, 1990) but of smaller magnitude. It is unusual for the thermally defined hot backarc to extend very close to thrust front as in this example, in contrast to the Brooks Range in northern Alaska and the Rocky Mountain thrust front in southern Alberta, where there has been substantial overthrusting of the craton by more than 100 km. It is possible that the lithosphere in this region has been thinned only recently. However, there is geological evidence for thrusting and for uplift and erosion starting in the Cretaceous (e.g., Enkelmann, 2020).
The low elevation area with normal 30-35 km crustal thickness of northern Yukon extending to near the Alaska border ( Figure 1) has inferred cool cratonic temperatures from thermal isostasy and has high upper mantle velocities, as noted above. Geologically, it is defined as a crystalline cratonic basement beneath the sedimentary cover, 2.1 to 1.8 Ga, by Ross (1991) and Nelson and Colpron (2007). This westward extension of the Canadian Shield has been called the Mackenzie Craton by Estève et al. (2021). There is slightly thicker crust at the Richardson Mountains but also higher elevation so has similar inferred deep temperatures. The details of the boundary between the cold craton and hot backarc in this region are not yet well defined. Also, Estève et al. (2022) interpreted thick-skinned crustal deformation locally beneath the eastern part of the Romanzof Uplift of northern Yukon. This region appears to have weakened thick crust at the Beaufort Sea continental margin due to high temperatures in the crust and upper mantle.

Depth and Temperature at the Base of the Lithosphere (LAB)
The temperature and depth at the (LAB) provide a very strong constraint to temperature-depths in the crust and upper mantle, especially for the Moho and upper mantle temperatures, if steady-state thermal regimes can be assumed (see earlier discussion of thermal transients). Some estimates of crustal radioactive heat generation and crust/upper mantle thermal conductivity are required for greatest precision in interpolating to shallower depths and to the surface (e.g., Hasterok & Chapman, 2011). A few transient regimes have been concluded elsewhere, for example, Sierra Nevada in USA and Baja California in Mexico (e.g., Erkan & Blackwell, 2009), and suggested for an area of the northern Canadian Cordillera . The time constant for the hot backarc is a few 10's m.y., and for the cratonic areas, a few 100's m.y. LAB temperatures and depths can be obtained from both seismic structure data and the geochemistry of recent backarc volcanics. Craton LAB depth and temperature estimates are also possible from downward extrapolation of kimberlite xenolith temperature-depths to the asthenosphere adiabat (e.g., Figure 7).

Temperature at the Backarc LAB
If the LAB is defined by a partial melt layer, it is at or near the upper asthenosphere solidus. The base of the backarc lithosphere, the LAB, as seen in receiver functions has been concluded to represent by a partial melt layer with a sharp top and gradational base (e.g., Hopper & Fischer, 2018). Fischer et al. (2010) estimated a velocity decrease in the western U.S. of 10% ± 4.5% distributed over 30 ± 15 km in depth. The recent basalt geochemical data, as described below, also requires that the LAB be at or near the solidus. The underlying asthenosphere appears to be very close to adiabatic temperatures. At greater depths, Katsura et al. (2010) and Katsura (2022) estimated an adiabatic temperature gradient based on their temperature estimate at the 410-km discontinuity, giving a potential temperature of 1,337-1,370°C ± 35°C, using the expected adiabatic gradient of 0.4-0.5°C/km. This result provides support that the whole asthenosphere to the 410-km discontinuity is usually nearly adiabatic with only a very small temperature gradient. To maintain an adiabatic gradient, most of the Cordillera asthenosphere must be well mixed by small-scale solid-state convection. This T p agrees well with most other adiabat estimates away from hot spots like Yellowstone, Iceland, and Hawaii. The Cordillera asthenosphere also has the same temperature as the global adiabat within the uncertainties of about ±50°C, as estimated from volcanics geochemistry and seismic tomography velocities, except near hotspots. In the well-studied western USA, from seismic tomography velocity data, Hansen et al. (2015) estimated the average temperature at the LAB to be ∼1,345°C (T p ∼ 1,320°C). Klöcking et al. (2018) found that very small adiabatic temperature gradients continue down to ∼150 km with a T p ∼ 1,350°C based on four published seismic velocity-depth models in the western USA region. Although the absolute temperatures from seismic velocities have considerable uncertainty, ±50°C, the seismic studies consistently show a very low, adiabatic temperature gradient in the asthenosphere.
As described below, the geochemical equilibration temperature and pressure estimates for western USA are mostly 1,350°C ± 50°C. The asthenosphere T p under cratons and other stable areas appears to be very similar to that of backarcs and oceans away from hot spots. Xenolith data for the Slave Province in Canada (e.g., Canil, 2008;Kopylova & Caro, 2004;Kopylova et al., 1999) (Figure 7) indicate a conductive geotherm in the lithosphere that extrapolates to an average underlying asthenosphere adiabat potential temperature T p ∼ 1,325°C at depths of 200-250 km. Hamza and Vieira (2012) estimated an asthenosphere adiabat T p of about ∼1,300°C as generally representative of Precambrian cratons based on cratonic mantle xenolith temperatures. In conclusion, most estimates of the asthenosphere adiabat under the Cordillera and elsewhere globally away from hot spots are within the uncertainties of the potential temperature T p = 1,325°C ± 25°C, giving ∼1,350°C at the 65 km Cordillera LAB and a temperature of ∼1,425°C at the deep cratonic LAB.

Depth of the Backarc LAB From Seismic Data
The seismic depth of the LAB can be most accurately determined from seismic receiver functions where they are well defined, but the depth can also be estimated with lower resolution from seismic tomography velocities. The seismic receiver function method uses waves from distant earthquakes recorded at broad band seismic stations, recording phase conversion from S to P or P to S at the boundary and average P and S velocities from the conversion depth to the surface recording stations. Reliable and accurate resolution of the conversions requires good seismic data, especially from grid arrays such as EarthScope/USArray or detailed linear profiles (e.g., Kind et al., 2012). Approximate LAB depths are also possible from inferred temperatures from velocity-depth profiles taking the LAB to be at the break from conductive temperature gradients in the lithosphere to the very small adiabatic temperature gradients in the asthenosphere (e.g., Klöcking et al., 2018; R. C. Porter et al., 2019, for western USA).

Depth of the Seismic LAB in Alaska and NW Canada
The seismic LAB depth determined by receiver functions is a work in progress by several groups and we expect a more accurate and complete mapping in the future, especially using USArray grid data in  (2015) using receiver functions on an initial part of the USArray found the LAB at 80-90 km on several profiles. Gama et al. (2021) reported initial analyses for all Alaska and found most values in the backarc of 80-85 km. In this and other Alaska seismic studies, there is a question of whether the receiver function boundary marks the top of the LAB ponded partial melt layer or represents the deeper more gradual velocity contrast at the top of the main partial melt layer, commonly ∼85 km (e.g., Berg et al., 2020). This is work still to be done.
The step boundary to a much deeper LAB to the north at the Brooks Range is quite well defined. In the Arctic Alaska terrane, the referenced studies found 120-150 km, but the weak phase conversions in cratonic areas give depths that are poorly determined. Also, Selway et al. (2015) reported that in cratonic areas there often are midlithosphere discontinuities of composition or state origin, and that the base of the lithosphere is deeper, the LAB, 200-250 km, as determined by kimberlite xenoliths and seismic velocity-depth constraints (e.g., Figure 7). For the Alaska backarc, a 65 km LAB gives a temperature of 850°C at a 35 km Moho, and ∼750°C for an 85 km LAB. Figure 5 illustrates temperature-depths for LAB depths from 56 to 87 km, giving 35 km Moho temperatures from 950 to 730°C. There are lower temperature estimates for shallower Mohos. For the crust in the central Alaska backarc as thin as 25 km, the estimated Moho temperatures are significantly lower, 650°C-750°C.
One study in NW Canada estimated a very thin lithosphere, LAB ∼50 km, based on seismic receiver functions . Such thin lithospheres imply very hot temperatures in the deep crust and likely melting if there is thermal steady state, as the authors recognized, and they suggest that there may have been a short-term transient lithosphere thinning such that the lower crust is not yet hot. However, there are no recent volcanics in the region, as might be expected for such a recently thinned lithosphere. Also, deformation and exhumation in the region indicative of a hot and weak lithosphere occurred during the Late Cretaceous to early Eocene with some continuing to the Eocene-Miocene (Enkelmann, 2020;McKay et al., 2021). This shallow LAB estimate remains a puzzle.

Effective Elastic Thickness
The thickness of the strong upper layer of the lithosphere and the total lithosphere strength are strongly controlled by temperature with second order effects of strain rate, composition, and hydration (e.g., Audet & Bürgmann, 2011;Hyndman et al., 2009;Kaban et al., 2014;Ranalli, 1995). At small deformations, this layer may deform elastically, with frictional and faulting at large deformations. The effective elastic thickness, usually designated Te, therefore is a strong temperature constraint having good correlation with heat flow, upper mantle seismic velocities, and other indicators of deep temperature. Te is closely related to the geological brittle-ductile transition and to the thermally controlled base of seismicity, although the correspondence is not exact. For high temperature gradients and plate tectonic strain rates, there is a strong layer only in the upper crust. For low gradients, there is a significant strength well down into the mantle. For moderate gradients, there may be a strong layer in the upper crust, a weak lower crust, and a second strong layer in the upper mantle, the mantle rocks having a higher activation temperature. The two-layer case may be represented by a single effective elastic layer, Te. Te provides low resolution temperature estimates (e.g., Hyndman et al., 2009). For crustal compositions and plate tectonic strain rates, Te is at ∼400°C. For mantle compositions, Te corresponds to 600°C-800°C Locally Te can be estimated through the horizontal wavelengths of lithosphere downward flexure due to local loads, such as foreland thrust belts, but for regional mapping Te can best be estimated using topographic loads, from the correlation between gravity and topography as a function of horizontal wavelength (e.g., Burov & Diament, 1996;Lowry & Smith, 1994). I illustrate the results from the study of Flück et al. (2003) for the NW Canadian Cordillera in Figure 11. The estimates are quite bimodal with Te of 15-20 km in the Cordillera hot backarc and mostly over 60 km in the adjacent cratonic stable areas. In both areas, modeling suggests only a single strong layer that is in the upper crust for the Cordillera and extends well into the mantle for the craton (e.g., Afonso & Ranalli, 2004;Hyndman et al., 2009), recognizing that the detailed strength versus depth depends on the strain rates and rheology assumed. In NW Canada, the craton east of the Cordillera has an effective elastic thickness Te of 60 to over 150 km. In the Cordillera of NW Canada extending to the west in central Alaska (limit of the Flück et al. (2003) data at ∼143°W), the Te is thin, 10-20 km, characteristic of the Cordillera hot backarc. The area just to the west of the Mackenzie Mountains thrust front has thick Te, contrary to the generally high temperatures from some other constraints. The thick cratonic thickness Te extends several 100 km southwest of the mountain thrust front, further than indicated by the other estimates of the westward limit of cold cratonic temperatures. It has not yet been explained why the uppermost mantle in this region should be so strong from the Te data.
At the edge of their large-scale study of Te in the Amerasia Basin, Arctic Ocean, Ling et al. (2021) show that there is a southerly transition from a thick Te in northern Alaska to a much thinner Te in central Alaska to the south. In central-western Alaska, where the backarc crust is unusually thin, ∼25 km and therefore Moho temperatures are unusually low, there may be a thin strong layer in the upper mantle, and total lithosphere strength may be greater. In general, the strength estimates indicate that the backarc is weak enough to be deformed by plate boundary forces, whereas the cratonic areas are too strong.

Character of Magnetic Anomalies
Regional aeromagnetic anomalies provide, first, a mapping of the boundary between the Cordillera hot backarc and the adjacent cool craton. Second, crustal temperatures can be estimated from the depth to where crustal rocks reach their Curie temperature and become largely nonmagnetic, using spectral analyses of aeromagnetic data. The boundary between the North American Cordillera hot backarc and the adjacent cratonic regions is generally well defined by the spatial wavelength and character of the magnetic variations ( Figure 12) (e.g., Bankey et al., 2002;Geological Survey of Canada, 2020). In northwestern Canada, the Cordillera backarc has generally shorter wavelength variations, from some combination of small scale magnetic mafic oceanic and arc terranes, and from the magnetization thermal limit being shallower (e.g., Cook et al., 2004Cook et al., , 2005Lynn et al., 2005;Pilkington & Saltus, 2009;. The foreland basins are usually quite nonmagnetic. The boundary is commonly up to 100 km seaward of the Rocky and Mackenzie Mountains thrust front, illustrating the thrusting of the Cordillera upper crust over the strong craton lithosphere. Gaudreau et al. (2019) showed that the change in the spectral content of the Cordillera magnetic variations is visible from the magnetic anomaly map.  noted high amplitude magnetic anomalies in the two embayments in the Cordillera mountain front and argued that these embayments Figure 11. Effective elastic thickness, Te, from the coherence of topography and gravity as a function of horizontal wavelength (after Flück et al., 2003). Note the large Te to ∼300 km west of the mountain front. represent areas of strong mafic lithosphere that buttress the Mackenzie Mountains and Beaufort thrust belt, resulting in westward and northern thrust front extensions. The magnetic map of Saltus et al. (1999) and analysis of  show a similar contrast across the Brooks Range in northern Alaska, from short spatial wavelength variations of low amplitude in the backarc, and longer wavelength high amplitude variations in the Arctic Alaska terrane. The southern backarc boundary with the foreland belt from these characteristics is near the Denali Fault.

Bouguer Gravity
Although not a quantitative temperature estimator, the Bouguer gravity map clearly shows the contrast between the hot thin lithosphere backarcs compared to the cold cratonic areas. In NW Canada, the backarc-craton boundary is seen quite clearly in regional values, from less than −80 mgal in the backarc to over −50 mgal in the craton. The large gravity contrast primarily reflects the ∼1,600 m change in elevation with little change in crustal thickness. There is only a small change in Free Air gravity. The high elevations are not compensated by thick crust. Figure 13 shows the low Bouguer gravity across the backarc with transition to higher gravity on the craton a few 10's of km west of the thrust front. This gravity boundary likely represents the location of the underlying deep lithosphere thermal and lithosphere thickness boundaries. In Alaska, where the backarc crust is especially thin, the backarc-craton contrast is not as clear in the gravity data (e.g., , although the thick crust Brooks Range is well defined. In NW Canada, both the gravity and magnetic data suggest that the Cordillera-craton thermal boundary is just west of the deformation front (Figures 12 and 13). This is in contrast to the Te data that suggest the boundary is further west. The latter remains an unresolved question.

Depth to the Magnetic Curie Temperature
Crustal temperatures may be constrained by the depth to the magnetic Curie temperature, below which the crust becomes largely nonmagnetic. For the Curie temperature-depth, magnetite is concluded to be the primary mineral control, so the limiting temperature is ∼580°C (e.g., Dunlop & Özdemir, 2001), although other magnetic minerals could be important. The Curie depth can be estimated from aeromagnetic data using the radially averaged power spectral density function (e.g., Bouligand et al., 2009). Curie depths have been estimated for western USA by Bouligand et al. (2009), who found a range of 10-23 km for much of the main Cordillera backarc, in general agreement with heat flow and other backarc temperature constraints (e.g., Figure 7). For western North America, J. Wang and Li (2015) found quite scattered depths but again generally consistent with other temperature estimates. For the Yukon in northwestern Canada, J. Witter & Miller (2017) and J. B. Witter et al. (2018) estimated depths a little deeper of 25-35 km for an area that includes part of the cold Mackenzie craton. For western Canada, Curie depths have been estimated by Gaudreau et al. (2019) and Bao et al. (2014), which I discussed below. Audet and Gosselin (2019) described the difficulties and ambiguities. In each study, the Curie depth contrast between the hot Cordillera and cool craton is clear, but absolute temperatures require the choice of a poorly constrained source spectral statistics parameter that needs to be estimated independently.
For quantitative temperatures, the critical control is the spectral characteristics of the magnetic anomaly field. The parameter β describes the fractal distribution of crustal magnetization. Gaudreau et al. (2019) found that a β value of 2.5 gives the best agreement with heat flow data as a calibration in the Cordillera and Slave craton, and 2.0 for the Canadian Shield south of the Slave craton. With these preferred parameters, their average Curie depth is 15 ± 1 km in the Canadian Cordillera and 32 ± 3 km in the Slave craton, and 34 ± 3 km in the North American craton to the south. These estimates give over 1,000°C for steady state at a 35 km Moho in the Cordillera, and 600°C-650°C at a 35 km Moho for the craton (Figures 5 and 7). These temperatures are significantly higher than those from other constraints, especially for the Cordillera. Such high temperatures would be expected to result in substantial crustal melting (e.g., Collins et al., 2020). Our estimates for a Curie temperature of 580°C is at ∼20 km from the other constraints for the hot backarc Cordillera and ∼70 km for the craton areas. Since we now recognize that the very high heat flow in the northern Cordillera and adjacent craton region is likely a consequence of near-surface high heat generation and not indicative of especially hot deep temperatures, a lower temperature calibration may be appropriate. However, even with a lower temperature normal backarc calibration, the magnetic anomaly temperatures are higher than those indicated by other constraints. Further study of this temperature method is needed, with more detailed temperature calibration using all available temperature constraints.

Depth Extent of Seismicity
Another parameter that is strongly controlled by temperature, so is a temperature constraint, is the depth extent of seismicity. There is secondary influence by composition and strain rates (e.g., Watts & Burov, 2003;Wong & Chapman, 1990). Wong and Chapman (1990) estimated that for crustal seismicity in western USA, brittle failure is limited to 350°C ± 100°C. Uppermost mantle earthquakes have a higher temperature limit of 600°C to 800°C, but these are uncommon. Figure 14 shows the number of earthquakes versus depth for limited areas of NW Canada hot backarc areas, near the Beaufort Sea margin from Estève et al. (2022), and for the Denali fault zone area by Choi et al. (2021). For hot backarc comparison, I show the depth distribution for the hot crust of California (Hauksson & Meier, 2019) and for the Cascadia western backarc . For comparison of cold areas, I show the depth distributions of the well-studied Fennoscandian shield (Veikkolainen et al., 2017), and the cold Cascadia forearc . The decline in the number of events with depth is over ∼10 km. The depth extent is often taken to where there are very few events, ∼10%. However, I prefer to use the 50% depth, which is much better defined. The latter is about 5 km shallower for hot backarc areas and about 10 km shallower for cold cratonic areas. The NW Canada locations have very similar depth extents to the reference backarc examples; 50% is slightly shallower, at ∼10 km, rather than 12 km, but the difference is not significant. The hot backarc and cold cratonic area are markedly different, about a factor of two at 50% depth. In both regions, the 50% depth is where the temperatures are about 350°C based on our other temperature constraints. The depth of seismicity proves to be an excellent thermal discriminator, provided that there is sufficient seismicity with good depths. The effects of strain rates and crustal compositions have not yet been examined.

Lab Temperature and Pressure (Depth) From the Geochemistry of Backarc Volcanics
The equilibration geochemistry of recent volcanic rocks at their source in subduction backarcs provides an important estimation of the depth and temperature of the LAB in the Cordillera. Sporadic widespread backarc volcanism is common for most current and recent subduction zones, including Alaska and NW Canada. The temperatures in backarcs at a relatively shallow depth, less than ∼150 km, are hot enough for partial melting in the asthenosphere. It has recently been concluded that the geochemical equilibration recorded in many mafic volcanics with their host mantle matrix occurs primarily at the LAB, not at their deeper partial melting source depth (e.g., Hyndman & Canil, 2021b;Plank & Forsyth, 2016). Earlier, it was commonly thought that equilibration depth estimates represented an average of equilibration along an upward progressive partial melting path. It is now evident that rising grain boundary partial melt percolates upward in the permeable asthenosphere until it ponds under the impermeable lithosphere where it geochemically reequilibrates. The partial melt collects in a layer with a sharp top and a gradational base that is seismically detectable (e.g., C. B. Till et al., 2010). When there are sufficient low density melt concentrations to be gravitationally unstable, the melt breaks through and rises rapidly locally in conduits into the upper mantle and crust. Some transit times may be as short as hours to days based on the presence of upper mantle xenoliths (e.g., O'Reilly & Griffin, 2010). The detailed locations of rising melt concentrations through the lithosphere are likely controlled by local conditions such as preexisting structures, varying stress, slab windows, and concentration of water. As noted earlier, the nearly constant temperature asthenosphere is very close to an adiabat gradient. The temperatures are commonly specified by the potential temperature T p , which is the temperature extrapolated to the surface using the adiabatic gradient of about 0.04°C/km. The regional T p of 1,325°C ± 25°C (1,350°C at a 65 km LAB) as discussed earlier is sufficiently constant and well-determined to result in negligible uncertainty in Moho and other deep temperatures from the LAB temperature and depth.
There are widespread Neogene-Recent volcanic rocks in the Cordillera backarc from the northern Cordillera available for analysis (e.g., Canil, Hyndman, & Fode, 2021;Greenfield et al., 2013;Harder & Russell, 2006;Hyndman & Canil, 2021a, 2021bKlöcking et al., 2018;Lee et al., 2009;Plank & Forsyth, 2016). It is notable that much of the western USA is in tectonic extension, whereas western Canada and Alaska are in compression, indicating that the extension is not required for recent backarc volcanics. In the western USA Cordillera, both the major element geochemistry (e.g., Plank & Forsyth, 2016; and references therein) and trace element geochemistry (Klöcking et al., 2018) indicate a surprisingly uniform temperature and depth at the LAB of 65 ± 5 km, and a LAB temperature of ∼1,350°C ± 25°C (extrapolated surface potential temperature of ∼1,325°C) (with the exception of the Yellowstone hot spot area and recently thinned lithosphere Colorado Plateau) (also R. Porter & Reid, 2021). One hypothesis is that this depth is constrained by the spinel peridotite to garnet peridotite pressure-controlled phase boundary (e.g., Hyndman & Canil, 2021b). In the Canadian Cordillera, for the Yukon, Canil, Hyndman, and Fode (2021) found an average depth of 62 ± 7 km and for all of the Canadian Cordillera, 65 ± 5 km and ∼1,350°C ± 50°C (T p 1,325°C) using the major element method of Plank and Forsyth (2016) (Figure 15). These depths and temperatures give a 35 km Moho temperature of 800°C-850°C in the Cordillera compared to 400°C-500°C in the cratons.
For Alaska, although the geochemistry of recent volcanics has not yet been reported in detail, initial analyses have been conducted by D. Canil (Hyndman & Canil, 2021a) employing the method described by Plank and Forsyth (2016) and Canil, Hyndman, and Fode (2021). The Alaska data include those from the Bearing Sea islands (Akinin et al., 2013;Andronikov & Mukasa, 2010;Chang et al., 2009;Winer et al., 2004). The results are compared to analyses from the Canadian Cordillera in Figure 15. The means and standard deviations of temperature and depth are 67 ± 12 km, and 1,338°C ± 26°C. This is a larger variation but very similar average to the Cordillera to the south in western Canada and western USA. The larger uncertainty in the Alaska data is likely due to the lack of constraints for the source H 2 O and other geochemical indicators that allow sample selection for the least altered in upward transit. At a mean depth of 67 km and a LAB temperature of 1,350°C, the 35 km Moho temperature is estimated to be ∼840°C ( Figure 5).  (Hyndman & Canil, 2021b) compared to estimates for western Canada, with standard deviation ellipses (after Canil, Hyndman, & Fode, 2021). One discriminator for the least altered samples is illustrated for W. Canada.

Distribution of Seismicity
As discussed earlier, the strength of the lithosphere is primarily controlled by the thermal regime, the large-scale regions where there is strong seismicity are expected to be similarly thermally defined. The details of local seismicity are determined by the local weak faults, by the local strength of the crust and upper mantle, and the nature of the driving forces. Notably, established fault zones like the Denali and Tintina tend to be long-term weak zones. A comprehensive summary of the recent tectonics of Alaska and adjacent NW Canada has been provided by Haeussler et al. (2008), based on geologically defined tectonics, by seismicity and by geodesy data (see also P. Bird, 1996;Estabrook et al., 1988). Neogene and active faults have been mapped by Koehler et al. (2012) and Bemis and Wallace (2007). D'Alessandro and  outlined current seismicity. Ruppert et al. (2008) presented a stress map based on earthquake mechanisms. Ruppert and West (2020) outlined the current seismicity and the large improvement in event detection and location with the use of EarthScope USArray data, especially in northern Alaska where previously there had been few seismic stations. The new data confirm the almost complete lack of seismicity in the cold strong Arctic Alaska terrane ( Figure 16). Hyndman, Flück, et al. (2005) described seismicity and seismic hazard for NW Canada, again showing the low seismicity in the cratonic Canadian Shield compared to the Cordillera. The principal concentrations are the mainly convergent Mackenzie Mountains thrust belt, and the mainly strike-slip Richardson Mountains (Figure 16) (also Drooff & Freymueller, 2021). Leonard et al. (2008) estimated deformation rates based on catalog seismicity (seismic moment) rates of a few mm/yr across these two belts.
The most prominent large-scale feature of the seismicity in Alaska (Figure 16) is the intense seismicity in the immediate area of the Alaska subduction zone. The seismicity includes Wadati-Benioff deep earthquakes in the subducting plate and those in the overlying crust (e.g., Ruppert et al., 2007Ruppert et al., , 2012. Although the forearc is cold, it is a tapering wedge and the mantle part has been concluded to be serpentinized, so this crustal area is weak and seismically active. To the north, most seismicity is limited to the hot and weak backarc (e.g., Ruppert et al., 2008). The active Denali Fault approximately follows the boundary between the cold forearc and the hot arc and backarc (e.g., Figure 1). Another prominent seismicity feature is the moderate band of seismicity in the Brooks Range at the southern edge of the Arctic Alaska block, where the upper crust of the backarc is interpreted to be slowly thrusting over the strong cratonic lithosphere (Figure 16). This thrusting is discussed in a later section. Similarly, in NW Canada, the various components of the cold strong Canadian Shield to the east have very little seismicity. Most earthquakes occur within the hot and weak former subduction backarc, especially in the terrane collision zone near the coast, in the Mackenzie Mountain belt, and in the Richardson Mountains.
Some of the backarc seismicity of southwestern Alaska may be driven by the subduction thrust tractions. However, the largest deformation source is the crustal indentors, the Yakutat terrane and St. Elias-Chugach collisions over a large area in the corner of the Gulf of Alaska. An especially intense collision source of seismicity is the Saint Elias-Chugach area in the immediate corner of the Gulf of Alaska, as noted above. These collisions are expressed in the dextral Denali strike slip fault zone, which is highly seismically active and has large geologically defined displacement. In NW Canada, two strong concentrations of seismicity are the broad zone of the Mackenzie Mountains thrusting, the principal backarc boundary with the craton to the east, and the mainly dextral strike slip Richardson Mountains belt extending to the Beaufort Sea margin in the north. It is not clear how the Richardson seismic belt is controlled; the detailed thermal regime in this area is not yet sufficiently constrained. There is also minor seismicity on the Tintina fault, which currently is moving very slowly but has major geologically defined earlier displacement.
Extending north to the Beaufort Sea margin, there is an interesting region of crust that appears to be converging to the north on to the Beaufort Sea ( Figure 16). Its eastern boundary is the seismically active dextral Richardson shear zone (e.g., Hyndman, Cassidy, et al., 2005, Hyndman, Flück, et al., 2005 and its western boundary is the broad sinistral shear Canning displacement zone (e.g., Buurman, 2018; Ruppert  , 2008). There is a geologically well-defined Cretaceous to the recent thrust belt along the Beaufort Sea margin with substantial shortening (e.g., Lane, 2002). The current convergence is only a few mm/yr from the seismicity rate and the GPS data (Leonard et al., 2007. Offshore in the Beaufort Sea, there is a cluster of earthquakes ( Figure 16) that may be a consequence of the upper margin crust slowly thrusting over the oceanic plate and the load of the Mackenzie River Delta sediments (e.g., Audet & Ma, 2018).
A practical application of the lateral variation in thermal regimes is seismic hazard characterization. First, as noted earlier, deformation and seismicity are concentrated in the hot weak regions. Second, the seismic attenuation of ground motion with distance is much faster in hot areas than in cold areas.

Current Deformation Rates From Geodetic GPS
Geodetic GPS data are an important tool for defining the distribution of current deformation. Elliott and Freymueller (2020) presented a summary compilation of data and the results of a GPS relative velocity analysis for Alaska and northwestern Canada, relative to a stable North America reference. They developed a regional current tectonics model ( Figure 17) (see also Freymueller et al., 2008;Leonard et al., 2008). They concluded that along the southern Alaska margin there are large northerly relative motions resulting from the Pacific plate subduction, especially driven by the Yakutat collisional indenter. In the immediate subduction zone area, the GPS-defined motions are dominated by the buildup of elastic strain that will be released mainly in large thrust earthquakes. Inland, southcentral Alaska is rotating counterclockwise and in western Alaska there are westerly motions toward the Bering Sea region following the trend of the hot and weak backarc. There is no internal deformation evident within the cold and strong Arctic Alaska Terrane, but this terrane is moving slowly relative to their stable North America reference. The displacement of ∼2 mm/yr shortening north-south across the Brooks Range has been estimated using GPS data by Elliott and Freymueller (2020) and seismicity rates by Leonard et al. (2008).
In the Yukon and adjacent hot backarc areas, Leonard et al. (2007Leonard et al. ( , 2008, Mazzotti et al. (2008) and Marechal et al. (2015Marechal et al. ( , 2018 estimated displacement rates from GPS and seismicity of up to 10's of mm/yr near the coast in the southwest around the eastern Yakutat indentor from GPS and seismicity. There is rapid displacement on the Denali and related faults. A convergence rate of a few mm/yr using GPS and seismicity rates was estimated across the Mackenzie Mountains by Leonard et al. (2007Leonard et al. ( , 2008. However, the GPS rates are near the resolution limit, especially with the uncertainty in current postglacial rebound horizontal motion. The Mackenzie craton appears to be moving with the Canadian Shield to the east.

Lithosphere Strength Versus Depth and Total Lithosphere Strength
As discussed earlier, the strength of the lithosphere, with depth and total, is strongly controlled by the thermal regime (e.g., Ranalli, 1995;Tesauro et al., 2015). A simplified model of the strength versus depth of the lithosphere in hot backarcs and cold cratons is using temperatures as in Figure 7 is shown in Figure 18 (adapted from Hyndman et al., 2009). A model Te based on estimated temperatures is also shown. The model Te depth is in good agreement with the observed Te from gravity-topography coherence (Figure 11). Figure 18b shows the estimated total strength of the lithosphere in the two regions, compared to estimates of the magnitude of plate boundary forces. Cratons are much too strong to be deformed under normal conditions by a factor of about 20, consistent with the lack of seismicity and of GPS defined deformation (Figures 16 and 17). The backarcs are weak enough to be deformed, as indicated by the seismicity and the GPS defined deformation. Figure 17. GPS defined deformation velocities in Alaska and NW Canada, with stable North America reference (after Elliott & Freymueller, 2020). Subduction zone events are shaded. There is 1-3 mm/yr current convergence across both backarc-craton boundaries (also see Leonard et al., 2008).

Some Consequences of Temperatures at the Base of Thin and Thick Crusts
The Moho lower temperatures for a thin crust have significant implications for crust and uppermost mantle strength. The crust of the central Alaska backarc is unusually thin, 25-30 km, thinner than most of the North American Cordillera to the south. Assuming the same backarc temperature-depth and lithosphere thicknesses as adjacent areas, and to the Cordillera to the south (e.g., Figure 5), this depth has a cooler Moho temperature of about 700°C-750°C compared to 800°C-850°C at 35 km. There are several consequences: (a) There may be a less of a hot weak layer in the lower crust for there to be lower crust lateral flow, and that accommodates the detachment of the upper crust from the mantle. (b) For this situation of thin crust, the uppermost mantle is also cooler and may have significant strength compared to a weak upper mantle for a 35 km Moho. The total lithosphere strength is therefore likely greater. The backarc crust thins slightly and the elevation decreases to the west toward the Bering Sea continental shelf. Seaward under the continental shelf, the surface becomes progressively below sea level. The thinning could be partly the result of crustal extension in the mid-Cretaceous (e.g., Miller & Hudson, 1991), but from GPS data, the thinning appears to be ongoing. The region may be in the initial stages of a backarc basin. In addition to normal faulting extension, the westward thinning may be the result of lower crust flow toward the low elevation and thin crust of the Bering Sea. In the Cordillera to the south, the higher elevation and thicker crust is buttressed by the strong cratonic lithosphere to the east, by the strong cold forearc and subduction convergent forces to the west. In contrast, the Alaska backarc is tectonically open to lower elevations in the west, which have low gravitation potential. A related process is the high elevation Alaska Range located at the southern edge of the hot backarc. Because the crust is thick, ∼40 km, the lower crust is unusually hot and is likely to be tectonically weaker than to the north. The crust may readily shorten and thicken under subduction convergence forces and the forces of the Yakutat terrane collision.

The Flat Moho
An important consequence of the high temperatures in the backarc areas of central Alaska and the northern Canadian Cordillera is a very weak lower crust that allows lateral flow of the lower crust that flattens the Moho toward a gravitational equipotential. Hyndman (2017), McKenzie et al. (2000), P. Bird (1991), and Kruse et al. (1991) have argued that if the lower crust is very weak, the Moho flattens toward a gravitational equipotential. Lower crust flow explains the observed regionally constant thickness crust in backarc Alaska and NW Canada (Figures 9a  and 9b) where the high temperatures (Moho 800°C-850°C) result in very low viscosities in the lower crust (e.g., Shinevar et al., 2015). The flat Moho is in spite of periods of large Cenozoic tectonic shortening and extension (e.g., Colpron et al., 2007;Plafker & Berg, 1994). In places, lower crustal rocks have been brought to the surface. As noted above, there is one large-scale backarc crust thickness difference. The Alaska backarc has an unusually thin crust, mostly 25-30 km, with a small thinning trend to the east toward the Bering Sea (Figure 9a). In the NW Canada backarc, the crustal thickness is slightly thicker but very constant, mostly 30 ± 2 km (Figure 9b), in spite of periods of substantial shortening and extension. Formerly deeply buried metamorphic rocks have been exposed (e.g., Colpron et al., 2007;Miller & Hudson, 1991;Monger & Gibson, 2019). This constant crustal thickness continues to the south in the Cordillera of southern Canada and western USA, where the average may be slightly thicker, ∼33 km. In Alaska, the cold stable Arctic Alaska terrane has a thicker crust, up to 50 km in the Coville forearc basin just to the north of the Brooks Range collision. In contrast to Arctic Alaska, in NW Canada, the Canadian Shield has very similar thicknesses to the backarc, ∼30 km, but with lower elevation as expected from thermal isostasy.

Backarc Thrusting Over Cratonic Regions
The second important consequence of the high temperatures in the backarcs of Alaska and the northern Canadian Cordillera is lower crust detachment that allows the upper crust to be thrust a considerable distance over the adjacent strong cratonic terranes, mainly driven by plate boundary forces. This process is common globally (e.g., Hyndman, 2019a). The thrusting of the Rocky Mountains fold and thrust belt in southern Alberta (Figure 19) was earlier shown by Hyndman (2015b), van der Velden & Cook (1996), and Marquis and Hyndman (1992). The thrust structure in that region was defined geologically by Carr and Simony (2006). In the Alaska Brooks Range, the backarc side of the thrust shows the characteristic lower crust parallel reflectors interpreted to represent horizontal shearing of the weak lower crust (e.g., Hyndman, 2017).
In the Brooks Range of northern Alaska, GPS and seismicity data indicate a slow current convergence, and substantial past north-south shortening has been defined geologically. The thrust structure has been shown especially well by the wide-angle seismic program of the north-south TACT (Fuis et al., 1991(Fuis et al., , 1997(Fuis et al., , 2008Moore et al., 1997;Wissinger et al., 1997). This program defined an asymmetric crustal root beneath the Brooks Range. From the south, the Moho deepens from 35 km with increasing dip. From the north, the Moho deepens from 36 km beneath the North Slope to 49 km beneath the Brooks Range. The inferred geometry has the upper backarc crust to the south of the crustal root, thrust over the North Slope strong cratonic lithosphere. The latter is depressed by the load forming the ∼10 km thick foreland Colville Basin. The structure of the thrust zone is shown in Figure 19, adapted from Fuis et al. (1997), compared to the southern Rocky Mountains thrust of southern Alberta (Hyndman, 2015b).
In the NW Canada Cordillera, the Lithoprobe reflection profile 2b south of the Mackenzie Mountains illustrates thrusting of the upper crust over the craton Cook et al., 2004). Wide angle refraction shows >10 km foreland basin sediment deposition with inferred low velocity Cook et al., 2004;Welford et al., 2001). The foreland basin is less prominent in the seismic structure data for the Mackenzie Mountains but Moho depths indicate local depression of the Moho in places by overthrust loading (Kao et al., 2013). Geologically, there is a broad foreland sedimentary basin 200-500 km wide (e.g., R. I. Thompson, 1981).
In NW Canada to the south, Hayward (2019) showed the evidence for large scale overthrusting of the Canadian Shield by the backarc, based on the westward increasing detachment depth of granitic plutons estimated from the depth of truncated deep gravity sources. This projection from the foreland across much of the Cordillera implies that Precambrian or Paleozoic rocks may project at depth that far as well. This is a conclusion similar to that of Cook et al. (2004) in the area based on multichannel seismic reflection data. If this interpretation is correct, the lithosphere in the eastern Cordillera must have been subsequently thinned and heated to the current state, as proposed by Bao et al. (2014). However, Canil, Russell, and Fode (2021) have argued against the recent lithosphere foundering or replacement based on peridotite xenolith geothermometry. Also, McKay et al. (2021) provided evidence that most deformation along the eastern margin of the Northern Cordillera was Paleocene- early Eocene (>65-50 Ma) and late Eocene-early Oligocene (40-30 Ma), based on low-temperature thermochronology data. The timing of Cordillera lithosphere thinning is still uncertain.

Consequence 3: Regional Barrovian Metamorphism
Another expression of the hot and weak backarc that I do not discuss in detail is the explanation of high crustal temperatures inferred in exposures of formerly deeply buried high grade metamorphic rocks, the Barrovian metamorphism. The Barrovian sequence involves a structurally downward succession of metamorphic mineral zones that reflect high vertical temperature gradients approximately double those of stable continental crust. A long-debated question is the origin of the high temperatures. From Jamieson et al. (1998), "Barrovian regional metamorphism: where's the heat?" (see also Ryan & Dewey, 2019). Hyndman (2019b) argued that this metamorphism commonly represents deformation exposure of already hot backarc crust, not the result of deformation-related heating. A number of authors have concluded that the high temperature gradients predate major deformation events and therefore are not a consequence of the deformation (e.g., Brown, 2009;Collins, 2002;Johnson & Strachan, 2006;A. B. Thompson et al., 2001). The backarc current temperature gradients in Alaska and NW Canada are very similar to those indicated from the characteristic Barrovian regional metamorphism ( Figure 20). Therefore, most regional metamorphic rocks tectonically exposed in the backarc since the assembly of Alaska and NW Canada and the start of the current subduction configuration may come from the deep crust of the already hot backarc. One example of such Barrovian metamorphism in the Alaska backarc is given by O'Brien and Grove (2020). Further testing of this metamorphism model in Alaska and NW Canada should be productive.

Conclusions: The Bimodal Thermal Regimes
1. The first order thermal configuration of Alaska and NW Canada are generally bimodal, with uniformly hot backarcs and cold cratonic regions of the Arctic Alaska terrane and the Canadian Shield (In Figure 1, the large dotted lines denote red-hot side; blue-cold side of thermal boundaries). Estimated temperatures at a ∼35 km Moho are 800°C-850°C and ∼400°C-500°C, respectively (cooler where a shallower Moho). The temperatures are consistent within the uncertainties from Heat flow measurements, temperature dependent upper mantle velocities, temperature-dependent topographic elevations (thermal isostasy), the depth and temperature of the LAB, the origin temperature and depth of craton kimberlite xenoliths, the geochemically inferred source temperature and depth of recent backarc volcanic rocks, the depth to the magnetic Curie temperature, and the depth extent of seismicity. The detailed cold forearc, hot arc-backarc and craton (north from Brooks Range) are quite clear in Alaska.
The hot and cold areas are not as well resolved in NW Canada, with several complexities. The cold Mackenzie craton extends the Canadian Shield locally from the Canadian Shield westward to near the Alaska-Yukon border, where the backarc-craton boundary trends north to near the Beaufort Sea coast. However, there is an intervening weak zone at the Richardson Mountains seismicity. Also, the depth of seismicity suggests high temperatures in the Mackenzie Delta region. Just west of the Alaska-Yukon border, there is a tongue of probable weak backarc crust that extends northward to ongoing slow thrusting on the Beaufort Sea margin in NW Canada and eastern Alaska. The region to the west of the Mackenzie thrust front is indicated to be a hot backarc from mantle seismic velocities and from the elevations with normal crustal thicknesses. However, there is a quite thick effective elastic thickness, Te, that suggest cooler temperatures.
2. The high elevations in the NW Canada Cordillera without a crustal root are compensated by reduced densities resulting from high temperatures in the thin lithosphere. The low elevations in the Alaska backarc are similarly isostatically balanced by the very thin crust, despite the high temperatures. There is a contrast of ∼1,600 m compared to stable areas to the north and east, for the same crustal thickness. 3. The backarc lithosphere has estimated a significant strength only in the upper crust, and the total lithosphere is weak enough to be deformed by plate boundary forces. In contrast, the cratonic areas have strength well down into the upper mantle, and they are much too strong for significant deformation. The seismicity and Figure 20. P-T of regional Barrovian metamorphism compared to P (depth)-T of Alaska-NW Canada backarcs (modified from Hyndman, 2019b). Note correspondence to temperature-depth for Alaska & NW Canada backarcs, 800°C-850°C at Moho, in Figure 9.
GPS-defined deformation are largely limited to the hot and weak backarc in both Alaska and the NW Canadian Cordillera, with very limited seismicity and deformation in the cold cratonic areas. 4. In the hot backarcs, there is a very weak low viscosity lower crust that allows lateral flow that flattens the Moho to a gravitational equipotential over short geological times. A weak lower crust means that the mantle is commonly detached from the crust, such that thrust and extension deformation structures generally detach in the lower crust. Lower crust detachment allows the backarc upper crust to be thrust 10's of kilometers over the strong cratonic areas in the Brooks Range in Alaska and the Mackenzie Mountains front in NW Canada. The Brooks Range and the Mackenzie Mountains have ongoing current convergences of a few mm/yr from GPS and seismicity rates. 5. There is widespread Neogene to Recent backarc volcanism in both Alaska and NW Canada, where the lithosphere is very thin. The inferred source is at the LAB, at ∼1,350°C and ∼65 km. 6. Most regional metamorphic rocks exhumed in the backarcs since the start of the current stage of subduction likely represent rocks that were tectonically exposed from already hot deep crust, as is characteristic of most backarcs. The current backarcs have temperature gradients that correspond to metamorphism. The high temperatures are concluded not to be a consequence of the deformation.

Potential for Additional Work
Most of the temperature constraints that I have discussed could be improved substantially with more work. Suggestions include: 1. The heat flow data in Alaska and their hot backarc and cold cratonic interpretations are quite straight forward, but data on upper crust radioactive heat generation would allow more accurate and reliable deep temperatures. The heat flows in part of NW Canada north 59°N are exceptionally high in both the Cordillera backarc and in the adjacent craton, especially the Wopmay orogen. More constraints on crustal heat generation and lithosphere thickness are needed for confident crust and upper mantle temperature estimates. 2. For upper mantle temperatures derived from temperature-dependent velocities, there is an opportunity for 3D mapping of temperatures using the recently acquired seismic structure data, recognizing the role of partial melt in the asthenosphere mainly below ∼65 km. A reference no-melt velocity-depth is important. 3. Our simple regional thermal isostasy estimates confirm that there is a bimodal thermal regime in Alaska and NW Canada, with high hot backarcs and low cold cratonic areas for the same crustal thickness. However, this thermal isostasy method could be profitably mapped in detail over the whole of Alaska and NW Canada using the recently available distributions of crustal thickness and average crustal velocities. 4. The depth and temperature of the seismic LAB is a very strong constraint to deep crust and upper mantle temperatures, assuming a steady state. The inconsistencies between different seismic LAB depth estimates in the backarcs from receiver functions, tomography velocities, and from the recent volcanics geochemistry, 50-85 km, need to be resolved. 5. The depth estimates to the magnetic Curie temperature in NW Canada need better calibration using all of the available temperature constraints; most Curie temperature depth estimates so far are deeper than those from the other constraints, so the temperature estimates are hotter. Analyses for central-western Alaska are also needed. 6. The inferences that the craton extends under much of the Cordillera in NW Canada from multichannel seismic reflection data and from modeling of deep gravity sources is surprising, since most of the current Cordillera is not thermally "cratonic," having a thin lithosphere and high temperatures in the crust and upper mantle. A previous cratonic lithosphere in this now hot backarc region may have been thermally eroded and thinned in the past, similar to the recent thinning of the Colorado Plateau in western USA. 7. This article provides an example for the multiple deep temperature constraints that could be productively applied in many other places.

Conflict of Interest
The authors declare no conflicts of interest relevant to this study.

Data Availability Statement
No new data were created in this work, and no software has been developed. No original data of the referenced sources were accessed for interpretation in this work.