Abstract

The neotectonics of southern Alaska (USA) are characterized by a several hundred kilometers–wide zone of dextral transpressional that spans the Alaska Range. The Denali fault system is the largest active strike-slip fault system in interior Alaska, and it produced a Mw 7.9 earthquake in 2002. To evaluate the late Quaternary slip rate on the Denali fault system, we collected samples for cosmogenic surface exposure dating from surfaces offset by the fault system. This study includes data from 107 samples at 19 sites, including 7 sites we previously reported, as well as an estimated slip rate at another site. We utilize the interpreted surface ages to provide estimated slip rates. These new slip rate data confirm that the highest late Quaternary slip rate is ∼13 mm/yr on the central Denali fault near its intersection with the eastern Denali and the Totschunda faults, with decreasing slip rate both to the east and west. The slip rate decreases westward along the central and western parts of the Denali fault system to 5 mm/yr over a length of ∼575 km. An additional site on the eastern Denali fault near Kluane Lake, Yukon, implies a slip rate of ∼2 mm/yr, based on geological considerations. The Totschunda fault has a maximum slip rate of ∼9 mm/yr. The Denali fault system is transpressional and there are active thrust faults on both the north and south sides of it. We explore four geometric models for southern Alaska tectonics to explain the slip rates along the Denali fault system and the active fault geometries: rotation, indentation, extrusion, and a combination of the three. We conclude that all three end-member models have strengths and shortcomings, and a combination of rotation, indentation, and extrusion best explains the slip rate observations.

INTRODUCTION

Continental interiors are commonly deformed during collision (e.g., Tapponnier et al., 1982). The style of deformation is influenced by numerous physical factors, such as crustal thickness, thermal structure, preexisting weaknesses, rheology of the upper and lower crust, the size and geometry of the colliding plates, and the obliquity of the collision. The Denali fault system in Alaska includes the Denali, the eastern Denali, and the Totschunda faults and deforms the interior of Alaska as it accommodates far-field strain from the ongoing Yakutat microplate–North America collision (Fig. 1). This is dominantly a right-lateral strike-slip system, with significant contractional deformation (Haeussler, 2008; Bemis et al., 2012, 2015; Fitzgerald et al., 2014), which has also led to exhumation of the Alaska Range in the past 30 m.y. (e.g., Fitzgerald et al., 1995; Benowitz et al., 2011; Finzel et al., 2011; Lease et al., 2016). The fault is seismically active, as expressed by the Mw 7.9 Denali fault earthquake in 2002 (e.g., Eberhart-Phillips et al., 2003). The Denali fault has long been one of the most significant faults in central Alaska, because it is inferred that the fault developed in a Late Jurassic to Early Cretaceous suture zone (Ridgway et al., 2002). Both the fault and the suture zone approximately follow the trace of the Alaska Range (Hickman et al., 1977; Nokleberg et al., 1985). There has been more than 370 km of right slip along the Denali fault system since Cretaceous time (Lowey, 1998), and a 38 Ma pluton offset 38 km along the fault near Denali (Reed and Lanphere, 1974) indicates a time-averaged slip rate of ∼1 mm/yr since middle Cenozoic time. However, Nockleberg et al. (1985) proposed 400 km of offset on the Denali fault since ca. 56 Ma, which would imply a slip rate of 7 mm/yr for most of Cenozoic time. Benowitz et al. (2012a) gave preliminary evidence of other offsets that would imply a slip rate of 12 mm/yr for the past 25 m.y. Thus, there is considerable variability in the range of slip rate values over the time span from ca. 100 to 25 Ma.

There is a significant branch in the Denali fault system in the eastern Alaska Range (Fig. 1), where the central Denali fault splits into the eastern Denali fault and the Totschunda fault (Richter and Matson, 1971; Schwartz et al., 2012). The eastern Denali fault extends in bedrock into the Yukon (Fig. 1). A metamorphic complex with ca. 17 Ma biotite K-Ar cooling ages formed along, and is offset by, this eastern part of the fault (Richter et al., 1975). Thus, the eastern Denali fault was active in early Miocene time. The other branch, the Totschunda fault, has ∼17–27 km of offset of Paleozoic rock units (MacKevett, 1978), but it is not clear when the fault initiated. Milde et al. (2013) and Milde (2014) inferred that the fault has been active since ca. 25 Ma based on analysis of new geochronology and thermochronology along the fault. The fault is notably colinear with the plate-bounding Fairweather fault (Richter and Matson, 1971), but thus far no unequivocal connector fault between the Totschunda and the Fairweather faults has been identified, because the region is heavily glaciated. Nevertheless, given the geometry of the fault system, it seems almost certain to exist (see Elliott et al., 2013). In Matmon et al. (2006) it was shown that the Totschunda and eastern Denali fault have subequal Quaternary slip rates close to their intersection. Understanding deformation of interior Alaska due to plate motion requires detailed temporal and spatial analyses of slip amount and rate along the entire Denali fault and its main branch.

In Matmon et al. (2006) it was shown that the Quaternary slip rate decreases westward along the central part of the Denali fault system, from ∼13 to 9 mm/yr. These rates were obtained using cosmogenic exposure dating of moraines and other features offset along an ∼325 km length of the central Denali fault system within the past 17,000 yr. Mériaux et al. (2009) added two cosmogenic slip rate estimates along the Denali fault; their eastern site was the same moraine dated in Matmon et al. (2006; site DFCR), and they obtained a similar result. The western site (Bull Creek) of Mériaux et al. (2009) was to the west of previously published results, where they determined a slip rate of 7 mm/yr over the past 12,000 yr, which further established the westward-decreasing slip rate on the Denali fault.

In this paper we provide additional slip rate data at 12 sites, particularly from the eastern and western ends of the fault. The new sites are as much as 197 km west of Denali (formerly Mount McKinley), including the western locality from Mériaux et al. (2009) (DFBU; Fig. 2; Tables 1 and 2). From our perspective, we sampled the best cosmogenic sampling sites from long 154.7°W eastward to the U.S.-Canada border, a fault length of 765 km. At four of the sites we estimate slip rates based on measured offset, but infer the age from other arguments. Three of these sites (DFMB, DFES, and DFPB) have cosmogenic ages that we interpret as not reflecting the primary age of the feature, and another is from a site where we did not collect samples for cosmogenic exposure dating (DFKL). In total, the cosmogenic work presents 107 exposure ages collected from 23 offset features located at 19 different sites.

Our new data, combined with data at 7 sites obtained by Matmon et al. (2006) and by Mériaux et al. (2009) at 2 of these same sites, better reveal the overall pattern of decreasing slip rate toward the west. However, the data also show variability along the length of the fault. Many strike-slip faults in continental interiors show variations in total slip or slip rate along their length (Dead Sea, San Andreas, Alpine, and Tintina faults). The causes of these changes in total slip or slip rate are commonly related to structures that bleed slip off the master fault. In the case of the Denali fault, our large chronologic data set enables us to discuss mechanisms that may decrease slip along the fault, structural constraints on these mechanisms, and the implications for tectonic models of southern Alaska.

METHODS

Slip rates along the Denali and Totschunda faults were determined by exposure age dating of offset geomorphic features and measurement of their offset. The amount of offset at each site was determined by one or more of three methods: measuring by tape in the field, calculating the distance using a handheld global positioning system (GPS), or by measuring distances on high-resolution imagery. Geomorphic features were projected to the fault trace to obtain the offset measurement. Offset errors are typically assumed to be symmetrical, although in some cases we estimated asymmetrical errors, particularly where the offset features cross the fault at a low angle. Samples were collected from 23 offset features at 19 locations. These include new locations as well as those previously studied by Matmon et al. (2006) (Fig. 2; Tables 1 and 2; Supplemental File S11). The sampled geomorphic features include lateral moraines, alluvial and colluvial fans, a surface incised by a now-abandoned river meander, an esker, and inactive rock glaciers. Dating was done by measuring 10Be concentrations in boulders and sediment collected from the surface of each feature. We collected and analyzed 107 samples. Sediment samples included hundreds of 1–3 cm clasts collected at the surface. Boulder samples consist of chips from their top surfaces. Some sites along the central and western Denali fault have numerous exposed boulders. Boulder rock types include quartz pegmatite, granite, granodiorite, schist, gneiss, and indurated conglomerate. Other sites along the western Denali and the eastern Denali and Totschunda faults are composed of fragments of fine-grained dolomitic shale and vein quartz clasts.

Postdepositional processes acting on geomorphic surfaces affect the accumulation of 10Be in surface samples and can result in underestimated exposure ages. Periglacial processes (cryoturbation) mix dosed surface and less-dosed subsurface material. Erosion of moraine crests removes well-dosed material from moraine surfaces. Boulder erosion removes the most heavily dosed outer surface of the boulder. These processes ultimately result in the exposure of material that has not been continuously exposed over the entire life of the geomorphic feature. In an attempt to avoid such young material, we did not sample boulders that showed any indication of recent rotation and spalling; rather, we sampled boulders with smooth, presumably glacially molded, surfaces. The boulders we sampled did not show signs of weathering, thus reducing the possibility of recent mass loss from the boulder faces. Nevertheless, we are aware that weathering and erosion could have occurred. Boulder erosion rates have been estimated in previous studies in the Arctic to range between 1 and 3 mm k.y.–1 (e.g., Briner et al., 2005). Our calculations show that the impact of such rates of boulder erosion on age calculations of Holocene and late Pleistocene samples is 1%–8%. Due to the limited influence of boulder erosion on the exposure ages and the lack of an unambiguous correction methodology, we use boulder, cobble, and clast ages calculated without any erosion correction.

Shielding of exposed rock surfaces by dust, snow, or topographic obstacles reduces the production rate of cosmogenic isotopes and will result in an underestimated exposure age. Snow cover measurements obtained from stations located along roads near the study area indicate a maximum annual snow cover of 100 cm. This snow thickness (at a density of 0.2 g cm–3) is equivalent to a thickness of 7–8 cm of rock, and the maximum effect of such shielding (which occurs only a few months a year) on our exposure ages is 5%–6%, although it is smaller for most samples. This effect was demonstrated by Benson et al. (2004) in the Rocky Mountains of Colorado. Thus, in our age calculations we make no corrections for snow shielding.

Samples for 10Be dating were processed at the cosmogenic laboratory at the Hebrew University of Jerusalem (Israel), following the method modified from Bierman and Caffee (2001). The 10Be/9Be ratios were measured at the Center for Accelerator Mass Spectrometry (AMS) at Lawrence Livermore National Laboratory (Livermore, California, LLNL). The 10Be/9Be ratios were normalized to the LLNL KNSTD3110 standard with a value of 2.85 × 10–12. The 10Be ages were calculated considering a sea-level, high-latitude production rate of 4.5 10Be atoms g–1 yr–1 (Balco et al., 2008). Latitude/altitude scaling of the nucleonic production rate was done using methods in Stone (2000). We use Equation 1 to calculate simple model exposure ages (Lal, 1988): 
graphic
where N(t) = measured concentration (atoms g–1 quartz), P = production rate (atoms g–1 quartz yr–1), and l = 10Be decay constant (4.99 × 10–7 yr–1), which is calculated from the currently accepted 10Be half-life of 1.39 ± 0.01 Ma (Chmeleff et al., 2010; Korschinek et al., 2010). Because the moraines across the Denali and Totschunda faults were derived from relatively small drainage basins (1–5 km2), we assume that cosmogenic nuclide inheritance is low. Therefore, ages were not corrected for possible predepositional cosmogenic nuclide inheritance. The cosmogenic data are presented in Supplementary File S22.

RESULTS

Exposure Ages and Offsets

We divide the studied length of the fault in six segments, from west to east, for ease of discussion. The results are summarized in Tables 1 and 2 and shown in Figure 2.

Segment 1 includes the westernmost two sites, DFMB and DFRC (Fig. 2). Site DFMB is a large flat-crested moraine offset by the Denali fault. One sediment sample and three boulder samples were collected from this site. The three boulder samples yielded concentrations that correlate with simple exposure ages that range between 22.0 ± 2.4 ka and 11.0 ± 1.2 ka. The sediment sample yielded a simple exposure age of 7.0 ± 0.8 ka. The estimated offset of this moraine is ∼425 m. Site DFRC is a steep alluvial fan with large boulders, currently incised by several small channels that are offset by the Denali fault. We collected eight boulder samples, four north of the fault and four south of it. Simple calculated exposure ages range between 11.3 ± 1.2 ka and 7.7 ± 0.8 ka; the average is 9.0 ± 0.3 ka. The average offset measured at this site is 48 ± 5 m.

Segment 2 includes the two sites west of Broad Pass, DFBU and DFCA (Fig. 2). DFBU is an offset lateral moraine. This site was also studied by Mériaux et al. (2009), who obtained an age of 12.3 ± 2.0 ka. Two small boulders, two single quartz clasts, and two sediment samples were collected from this site. All samples yielded similar ages with an average of 11.5 ± 0.5 ka, which is within 1σ of the result of Mériaux et al. (2009). Based on the retrodeformation of features on aerial photographs and imagery, we find the offset to be 87 +10/–28 m. Our best estimate of the offset is nearly identical to the value of 85 ± 5 m determined by Mériaux et al. (2009). Site DFCA is a planar, gently tilted alluvial surface. Two sediment samples from this site yielded similar exposure ages of 8.7 ± 0.9 ka and 9.2 ± 1.0 ka. The offset at this site is not well constrained because the piercing points were not easily determined; nevertheless, the best estimate derived from matching piercing points on Google Earth imagery yielded an offset of 109 +14/–26 m.

Segment 3 includes five sites east of Broad Pass, DFSC, DFPP, DFWC, DFNR, and DFTH (Fig. 2). Site DFSC includes two offset features, an abandoned meander and a moraine, which were described in Matmon et al. (2006). Three additional samples from this site were analyzed in our study and their corresponding exposure ages agree with those in Matmon et al. (2006). The average age of all samples from this site is 14.1 ± 0.8 ka. The average offset was measured to 157 ± 10 m. At site DFPP one quartz clast, one boulder, and one sediment sample were collected. The average age of all three samples is 14.2 ± 0.9 ka and the measured offset is 57 ± 14 m. The proximity of the site to the bank of a large river raises the possibility that the observed offset is only a fraction of the total offset because bank erosion may have removed some of the geomorphic expression of the offset. Site DFWC includes an offset alluvial fan, and was described in Matmon et al. (2006). Two additional samples from this site were analyzed for this study and their exposure ages agree with those in Matmon et al. (2006). The average age of all four boulder samples from this site is 2.6 ± 0.2 ka. The offset is 25.5 ± 4 m. Site DFNR is a steep alluvial surface incised by a primary channel that is offset by the Denali fault. Six boulder samples were collected from this site. They yield an average age of 13.0 ± 0.6 ka. The average offset at this site is 100 ± 10 m. Site DFTH is an offset lateral moraine with a particularly sharp crest. Six boulders, three from each side of the fault, were sampled for exposure dating and yield a corresponding average age of 3.2 ± 0.3 ka. The average offset at this site is 14 ± 1 m.

Segment 4 includes three sites west of the Denali-Totschunda fault intersection, DFCR, DFMF, and DFES (Fig. 2). The first two sites were thoroughly described in Matmon et al. (2006). Site DFCR yielded an average age of 11.0 ± 0.4 ka and an average offset of 144 ± 14 m. This site was also studied by Mériaux et al. (2009), who obtained a well-constrained age of 12.0 ± 2.1 ka, within 1σ of the age obtained by Matmon et al. (2006). Site DFMF yielded an average age of 12.2 ± 0.5 ka and an average offset of 154 ± 12 m. We infer that site DFES was reset by recent rapid erosion, as discussed in the following.

Segment 5 includes the eastern Denali fault sites, DFTR, DFPB, and DFKL (Fig. 2). DFTR was described in Matmon et al. (2006). It includes two offset moraines, a higher one and a lower one. All samples from this site yielded an average age of 14.5 ± 0.6 ka and an average offset of 115 ± 25 m was measured. Site DFPB is a flat and gently sloping alluvial fan surface, which was cut by a stream and then offset by the Denali fault. It is devoid of boulders and quartz-bearing clasts are rare. Only one sample of three quartz clasts (1–3 cm) was collected. This amalgamated sample yielded an age of 27.8 ± 3.1 ka. The offset at this site is 75 ± 8 m. Site DFKL is located ∼250 km to the southeast, where the fault is characterized by en echelon mounds that were offset by the fault. The offset is distinct, but we found no way to exposure-date this feature. The origin and age of this feature are discussed further in the following.

Segment 6 includes the Totschunda fault sites, DFDP, DFNM, DFTC, DFCM, and DFNC (Fig. 2). Sites DFDP, DFNM, and DFTC were described in Matmon et al. (2006). Additional samples collected for our study yielded exposure ages that agree well with those previously reported. Site DFDP is an offset rock glacier. Samples from this site yielded an average exposure age of 11.2 ± 0.6 ka. The offset at this site is 64 ± 7 m. Two offset features were dated at site DFNM, an offset lateral moraine and an offset rock glacier. Both features yielded similar ages, with an average age of 10.8 ± 0.6 ka. However, the moraine is offset 80 ± 14 m, while the rock glacier is offset ∼65 m. Site DFCM is a flat and gently tilted alluvial surface shallowly incised by a small channel offset by the Totschunda fault. Four granitic boulders yielded an average exposure age of 1.3 ± 0.1 ka. The offset at this site is 11 ± 1 m. Site DFNC includes two offset moraines, a high one and a low one. Boulder and sediment samples from the high moraine yielded an average exposure age of 13.8 ± 1.3 ka and boulders from the low moraine yielded an average age of 7.4 ± 0.8 ka. The offset of the low moraine ranges between 54 ± 1 m and 74 ± 3 m. The offset on the high moraine could not be confidently determined.

DISCUSSION

Age Distribution

The distribution of cosmogenic ages (Fig. 3) shows a pronounced peak ca. 11 ka. Other smaller peaks are ca. 3ka, 16 ka, and 20 ka. Moraines did not yield ages younger than 8 ka and all ages younger than 8 ka are derived from either alluvial features, rock glaciers, or a neoglacial moraine crest. In addition, no ages older than 28 ka (site DFPB) were calculated. The peak of ages is at the very beginning of marine isotope stage 1 (MIS 1; Fig. 4). The influence of climatic perturbations (such as glacial-interglacial cycles) on exposure age dating of moraines on the northern slope of the Alaska Range was discussed in Matmon et al. (2010), who showed that exposure ages of boulders and sediment samples on MIS 3–MIS 4 age Delta moraines exhibited only a few ages that represent the actual time of moraine stabilization (ca. 60 ka). Instead, many of the ages were from the last glacial maximum (LGM, ca. 20 ka) and Younger Dryas (ca. 12 ka). Similarly, boulders and sediment samples collected from the MIS 2 Donnelly moraine exhibited some ages that represent the actual time of moraine stabilization (ca. 20–17 ka) while the rest of the ages were Younger Dryas. Matmon et al. (2010) inferred that intensified moraine exhumation, boulder exposure, and boulder erosion occur during glacial episodes later than the depositional time of the dated samples. These major phases of erosion are likely caused by ice-driven mechanical weathering of exposed bedrock and sediment (Hales and Roering, 2007) perched above the ice and the consequent exposure of shielded rock fragments. In terms of cosmogenic exposure age dating, these processes may partially or completely reset the exposure age of the investigated glacial feature, and result in the underestimation of the moraine age and a multipeak age distribution that correlates with glacial periods that followed the initial deposition. Along the southern slope of the Alaska Range, this resetting is expected to be more intense and complete because the amount of precipitation is much greater than along the northern slope. Thus surface processes may act more rapidly and effectively to obliterate the original exposure age of the moraines.

Moraines and glacial features were deposited during glacial periods older than 23 ka along the southern slope of the Alaska Range (e.g., Karlstrom, 1964) and along the southern Alaska margin (Eyles et al., 1991). In this study, one offset glacial feature at site DFMB is considered to be older than the LGM, based on its morphology and amount of offset. All other investigated sites were initially considered to be LGM age or younger. However, even the apparently old glacial feature at site DFMB yielded ages younger than 23 ka. Site DFMB is on a very flat and broad moraine crest with several huge erratic boulders, spaced several hundreds of meters from each other and dotting the otherwise flat surface of the moraine. These boulders had strong hematitic alteration and were the most weathered of any boulders in our study. The best estimate of the offset of this moraine is 425 m based on aerial photographs and ground measurements. Given the strong weathering and the very large fault offset, we expected old ages, perhaps at the end of MIS 6 at 130 ka (Lisiecki and Raymo, 2005; Fig. 4). However, three boulders yielded exposure ages that range between 22.0 ± 2.4 ka and 11.0 ± 1.2 ka and a sediment sample yielded an exposure age of 7.0 ± 0.8 ka. All four ages are significantly younger than expected, based on the large offset, and indicate rapid erosion during late Pleistocene and early Holocene time. We infer that this erosion reset the exposure age of this moraine. If we infer a constant slip rate of ∼5 mm/yr for this site, as indicated by the rate at the nearest site (DFRC, 136 km to the west), we calculate an age of ca. 90 ka for this moraine. This age does not correlate with documented growth of glaciers in Alaska. Alternatively, if the moraine has an age of 130 ka, the age of the MIS 5e-MIS 6 boundary (Lisiecki and Raymo, 2005; Fig. 4), then a slip rate of 3 mm/yr is inferred. We prefer this interpretation for calculation of an inferred slip rate, as the large size of this moraine is consistent with the penultimate regional glaciation.

Another site, DFES, yielded unrealistically young ages. The esker at site DFES is offset 180 m. It is located just 14 km east of sites DFCR and DFMF, where the slip rate is well constrained at ∼13 mm/yr (Matmon et al., 2006; Table 1; Fig. 2). Its proximity to these sites suggests the slip rate at site DFES should also be ∼13 mm/yr. Considering the offset of 180 m, its age should be ca. 14 ka. However, of the four samples that were collected from this site, three yielded 10Be/9Be ratios indistinguishable from blank and a fourth sample yielded a 10Be concentration of 0.13 ± 0.003 × 105 atoms g–1 quartz, correlating to an age of 0.9 ± 0.1 ka. Together these four results suggest very rapid and recent erosion of the esker crest and the nearly total resetting of the cosmogenic exposure age. We therefore do not consider this site in the discussion related to offsets along the Denali fault (Table 2).

We infer that one site, DFPB, yielded an unrealistically old age (27.8 ± 3.1 ka) for the offset measured. The samples were collected from an alluvial fan in a valley bottom, which was subsequently incised by a stream and produced a riser, and it was then offset by the eastern Denali fault. The four nearby sites (DFTR, DFDP, DFNM, and DFTC) with moraine or other postglacial features have cosmogenic ages between 18.4 and 10.8 ka, with an average of 13.7 ± 0.4 ka (Table 2). Thus, with this perspective on deglaciation, it seems unlikely that site DFPB in a valley bottom would have an exposure age nearly 10 k.y. older than anywhere else. One explanation is that the age is correct, but the surface is a relict fan that predates the LGM glaciation. Support for this idea comes from comparing the morphology of this fan, which has a smooth undulatory surface, to the modern alluvial fans, which are more incised and have distinct channels (site DFPB in Supplemental File S1; see footnote 1). Thus, the age of the surface may be correct, but the surface does not mark the initiation of the incision of the fan after LGM time. Alternatively, the single analysis for the age is the smallest sample size in our collection. It came from amalgamating three difficult-to-find quartz clasts. In this case, if even one clast had exceptional cosmogenic nuclide inheritance, our assumption that there is no inheritance of would be incorrect, and an apparently old age would be produced. In either case, it seems reasonable to infer that the valley was filled with ice at the LGM because of the extent of glacial deposits in nearby valleys, and we thus estimate the age from an average age of the nearby postglacial sites. This is the only change we make to the results presented in Matmon et al. (2006).

The southeasternmost site on the Denali fault (DFKL) provides a minimum slip rate that is based on field-measured offset and estimates of its age from geomorphic and Quaternary history, but not from surface-exposure ages. This site, near Kluane Lake, Yukon, is important because it is 250 km southeast of the nearest site on the eastern Denali fault (DFTR). This section of the Denali fault has been referred to as the Shakwak fault or Shakwak Valley fault (Bostock, 1952; Clague, 1979).

The geomorphology of the eastern Denali fault in the Yukon is unusual and somewhat unique. Along its ∼100 km length, this southeasternmost part of the fault is primarily expressed as a linear and continuous series of spruce-covered mounds that are typically several tens of meters across and may reach lengths >100 m (Bostock, 1952; Clague, 1979; Figs. 5A, 5B). The tops of the mounds are ∼3–8 m above their surrounding surfaces. The surface relief results in the mounds being better drained, which results in less acidic soils that allows trees to grow larger. These mounds give the fault trace the appearance of a string of pearls, or a large and continuous mole track (Fig. 5A). We are aware of similar mounds along the Magallanes-Fagnano fault in the Tierra del Fuego region of Argentina; although their origin has not been specifically addressed (Costa et al., 2006), it is a similar high-latitude region that was recently glaciated.

Google Earth imagery (from 2003, 2004, 2007, and especially higher resolution data from 2010) shows geomorphic and structural details of the fault zone. The mounds vary somewhat in shape, but are predominantly ellipsoidal with long axes commonly oriented ∼20° ± 8° counterclockwise of the trend of the main fault. They are consistently arranged en echelon. Spacing can be relatively even with common gaps of 30–50 m between adjacent mounds (Fig. 5B), and they also occur in tighter clusters or can be more widely separated. The gaps serve as preferential locations for surface runoff originating upslope from the mounds. As the fault traverses from glacially scoured terrain into younger alluvium in inset drainages and across postglacial fan surface, the mounds decrease in amplitude (61°28.175′N, 139°21.490′W), although the same en echelon arrangement is retained.

The fault trace is largely subparallel to glacial features, but also crosscuts these at angles ranging from 8° to 37° (Fig. 5C). Locally, there are suggestions that these features are offset (61°34.810′N, 139°39.655′W). Steps and bends of various scales are also present along the fault trace. The largest is a left step of as much as 1.6 km associated with a 6.5-km-long section (extending from 61°33.920′N, 139°37.621′W on the north to 61°32.172′N, 139°31.221′W on the south) between two linear strands of the fault trace connected by arcuate scarps (around 61°33.051′N, 139°34.009′W) that indicate reverse faulting across the transpressional stepover. Short splays also extend from minor steps and bends, indicating local structural complexity. In addition to these structural elements, which are indicative of lateral slip, there are arcuate to semiarcuate scarps suggestive of small thrusts along the trace.

Based on the geometric and geomorphic features that characterize this section of the Denali fault, we interpret the mounds as contractional structures, most likely en echelon folds, that formed between Reidel shears in a dextral shear zone. The oblique orientation of the mounds relative to the overall strike of the fault trace reflects classical wrench fault geometry in which folds form above lateral shears that do not extend completely to the surface. The mounds could be bounded on one or both sides by thrust faults that at some locations could link adjacent structures. The mound orientation with respect to the main fault trace is opposite of that expected for Riedel shears, but is in the expected orientation of shortening between R and R′ shears.

We interpret a sequence of mounds at site DFKL (61°18.426′N, 139°0.930′W) as having been laterally offset along the Denali fault (Seitz et al., 2008; Fig. 5D). Based on field mapping, the best measurement of offset is 22.5 +7.5/–2.5 m. Estimating slip rate requires an understanding of when the mounds formed and when right-lateral offset began to accumulate. One of the uncertainties is the observation that the mounds are not clearly offset at other locations along the fault. An alternative interpretation of the geomorphology is that what appears to be an upslope, and now offset, extension of the mounds may be material that was originally ponded behind them and was later incised and subsequently offset.

The age of initiation of the mounds is not known, but the observation that they define the Denali fault where it cuts across glacial features (Fig. 5C) indicates they are postglacial; if not, they would have been erosionally removed by the moving ice. The Denali fault trace commonly is near the base of a U-shaped glacial valley, and Quaternary mapping indicates that the region was covered with ice at the LGM. On the mounds that we examined, we found till exposed beneath overturned tree root wads, and we infer that the near-surface sediment consists of a well-consolidated lodgement till and/or permafrost. We concur with Bostock (1952) and Clague (1979) that these mounds occur on glacial drift.

Regional Quaternary geology considerations indicate that deglaciation began ca. 14 ka and retreated into the mountains ca. 13 ka (Bond and Lipovsky, 2009). After deglaciation, surface-rupturing earthquakes along the eastern Denali fault would be required to build the topography of the mounds between Riedel shears. We suggest that the strength of the lodgement till and/or permafrost prevented the immediate development of a single throughgoing fault trace after deposition, as is observed along the Alaskan sections of the Denali fault. At some time after the mounds formed, faulting broke through to the surface, at least locally, to produce the offset. At other locations terrace risers (61°21.986′N, 139°8.153′W) and channels (61°45.992′N, 140°3.680′W) are right-laterally offset ∼20–40 m.

A minimum slip rate can be estimated using the mound offset measurements and general fault zone observations. Using our measured mound offset of 22.5 +7.5/–2.5 m, and the age of deglaciation (13 ka) as a starting date, the offset at DFKL yields a rate of 1.5–2.3 mm/yr with a preferred valued of 1.7 mm/yr. An unknown amount of time passed between the initiation of mound formation and offset of the mounds and associated landforms at DFKL. Thus the actual rate remains unconstrained, but it must be higher. We use 2 mm/yr in our tables, but if the offset of the mounds is at the young end of the time frame we consider, then the inferred slip rate must be higher. We note that evidence for the trace of the fault can be seen as much as 6 km southeast of Kluane Lake. Farther to the southeast, there is no surface expression of the eastern Denali fault, although we have spent quite a bit of time looking for one on imagery and in the field. On this basis, it seems likely the fault has a low slip rate in the region of the offset mounds.

Offset Uncertainty

The determination of reliable piercing points, as well as the ability to measure accurate offsets, is sometimes hindered by surface processes such as riverbank erosion and solifluction. These and other processes obliterate the original morphology offset by faulting. We attribute different levels of reliability to the measured offset at each site depending on (1) our ability to locate a real and precise piercing point, and (2) the accuracy of measuring the offset between the piercing points, which depends on the size and geometry of the geomorphic feature that forms the piercing points. In most cases, the detection of offset geomorphic features is relatively simple and the level of accuracy depends only on the size of the measured features. At most sites, the offset features of choice are relatively small (5–10 m wide) channels on top of, or adjacent to, moraines or alluvial fans. In such cases the measurement uncertainty relates to the width of the stream and ranges between 10 and 20 m. The level of uncertainty increases when other features, such as large perennial rivers or the boundaries of rock glaciers, are considered. In most cases, we assumed symmetrical uncertainty in the offsets, but at some sites we were able to observe and measure asymmetrical offset uncertainties. The average offset uncertainty among all sites is 11% ± 5%. The locations that include uncertainties ≥11% ± 5% are DFNM (18%), DFTR (22%), DFCA (13%–24%), and DFPP (25%). Site DFNM is a rock glacier. Site DFPP is flanked by a perennial river that eroded some of the offset bank. Site DFCA was, and may still be, significantly affected by solifluction. We discuss the possible reasons for such large uncertainties and their influence on the calculated slip rate.

Sites DFCR and DFBU were also investigated by Mériaux et al. (2009). In their study, the measured offset in site DFCR ranged between 122 ± 5 m and 183 ± 30 m. This range is similar to the offset measured in our study (144 ± 14 m). Similarly, the offset for site DFBU measured by Mériaux et al. (2009) was 82 ± 5 m, nearly identical to our measurement (87 +10/–28 m). The good agreement in offset estimates among the two studies, which were done independently, demonstrates the reproducibility of the measured offsets of these two sites in particular and of all sites in general.

Slip Rates

Considering the sites that yielded both reliable ages or reasonable inferred ages and well-constrained offsets, we obtain a slip rate distribution along the entire Denali and Totschunda faults (Fig. 2B). Site DFRC, which is well dated and has a well-measured offset, yields a slip rate along the westernmost segment of only ∼5 mm/yr. Although our westernmost site, DFMB, did not yield a reliable cosmogenic age, we infer a slip rate estimate of ∼3 mm/yr. Thus, both westernmost sites yield relatively low, but nonzero, slip rates. Eastward, slip increases to ∼8–12 mm/yr in segments 2 and 3 and increases to a maximum rate of ∼13 mm/yr in segment 4 (red line, Fig. 2B). East of segment 4, slip is generally partitioned between the eastern Denali fault and the Totschunda fault. Along the eastern Denali fault (segment 5) the highest slip rate is ∼8 mm/yr, near the intersection with the Totschunda fault, while along the Totschunda fault (segment 6) the rates range between 6 and 9 mm/yr. We infer that it is reasonable to combine the slip rates along the eastern Denali fault (segment 5) and the Totschunda fault (segment 6) near their intersection, which sums to ∼14–16 mm/yr. This value is similar to the slip rate of segment 4, which is ∼13 mm/yr just west of the intersection. Overall, there is a systematic slip rate decrease from east to west. This slip rate decrease was suggested by Matmon et al. (2006) and by Mériaux et al. (2009).

Superimposed on the eastward slip rate increase is some scatter in the data (Fig. 2B). Such variability may be apparent and caused by errors in age calculations and/or offset measurements. We assume that small sections of faults cannot move faster than larger lengths of the fault. Therefore, assuming that offset features cross the entire fault zone, anomalously high slip rates may only be explained by such errors in age or offset measurement. Conversely, anomalously low slip rates may be actual natural fluctuations, perhaps related to aspects of strain partitioning (e.g., Spotila et al., 2007) or distributed shear. If the observed offset was not measured on the main trace of the fault, additional slip may be accommodated on other active strands that we missed. For example, the low inferred slip rate for site DFTH indicates that we may have missed an active strand of the Denali fault. The sharp-crested neoglacial moraine at this site has a clear offset of 14 ± 1 m measured both on the ground and using digital photogrammetric methods. In the 2002 earthquake the Denali fault offset this moraine 1.2–1.4 m (Haeussler et al., 2004), so we know this strand of the fault is active. However, the calculated slip rate (4.4 ± 0.4 mm/yr) is far lower than adjacent slip rates that are in the range of 10–13 mm/yr (Fig. 2B). It seems unlikely that a 6–9 mm/yr decrease represents a real drop in long-term slip rate for this short section of the fault. We looked for additional active strands both from the air and on imagery but found none. Another possibility to consider is that there was significant lateral slip on the nearby Susitna Glacier thrust, which ruptured in the 2002 earthquake. However, the focal mechanism for the subevent that included slip on the thrust indicates pure thrust motion, with no lateral component of slip (Eberhart-Phillips et al., 2003). Moreover, in a detailed postearthquake survey of the surface rupture, Crone et al. (2004, p. S15) specifically noted that there was “…no compelling evidence of significant tectonic lateral slip anywhere along the Susitna Glacier fault.”

We suggest that there is either distributed strike-slip deformation in the region of this site, or there is an unidentified strand, perhaps farther north in the higher mountains. In summary, we infer that the overall slip rate is near the upper end of the values we determined (red line, Fig. 2B).

Slip Rates through Time

Comparing average slip rates calculated over various time spans may shed light on the temporal behavior of the Denali fault. With respect to large and long-term offsets along the Denali fault system, significant right-lateral strike slip is inferred to have occurred in middle to Late Cretaceous time (e.g., Hickman et al., 1977). Lowey (1998) summarized prior long-term (tens of millions of years) offsets and found the most precise piercing point, which indicates that mid-Cretaceous rock units were offset ∼370 km across the fault (asterisks in Fig. 1). Nokleberg et al. (1978) found evidence of 400 km of right-lateral strike slip of 56 Ma metamorphic and intrusive rocks. These observations suggest long-term average slip rates that range between 4 and 7 mm/yr. Reed and Lanphere (1974) reported a relatively precise 38 km offset of a 38 Ma pluton located near Denali (plus symbols in Fig. 1). This implies an average slip rate of 1 mm/yr for the time period from pluton emplacement to the present. Plafker et al. (1977) noted that some of the major glaciers crossed by the Denali fault exhibit dogleg morphology (i.e., a bend in the valley accommodating the glacier) that coincides with the fault trace. c, p. B69) suggested that the ∼4 km bend in these glaciers represents the total offset that accumulated since glaciation initiated: “Displacement rates of 1–2 cm/yr suggest that major dextral displacement on these segments could have begun as recently as 200,000 to 400,000 years ago.” However, Plafker et al. (1977) did not have any hard data to back up the displacement rates or the age of initiation of glaciation. If we assume for a moment that the offset was initiated at the beginning of Pleistocene time at 1.8 Ma, then the inferred slip rate would be ∼2 mm/yr.

In contrast to the million-year slip rates presented here, the 103–104 yr average slip rates are faster. Slip rates in this study range between 2 and 13 mm/yr. This slip rate range suggests that although mountain building began along the Alaska Range in Oligocene time in response to the Yakutat microplate collision, the slip rate has been increasing toward the present day. It is reasonable to infer that the far-field consequences of terrane collision would change from diffuse deformation to more clearly organized as the Yakutat microplate slab is thrust beneath and into the Alaskan margin and as the orogen builds.

Contemporary GPS slip rates show the Denali fault as a profound boundary between geodetic sites north and south of the fault. Fletcher (2002) presented the most complete analysis of GPS geodetic slip rates across the central Denali fault, also summarized in Freymueller et al. (2008). Freymueller et al. (2008) showed that GPS vectors south of the fault parallel the fault, and they found a slip rate of ∼6 mm/yr assuming a 6 km locking depth (Fig. 6). These observations are limited to the time period before the 2002 earthquake, as after slip and viscoelastic relaxation dominates the postearthquake signal. The earliest campaign geodetic measurements bearing on this estimate of the slip rate are from 1995 to 2002 and most are from 1997 to 2002. This GPS slip rate of 6 mm/yr is about half of our 103 to 104 yr slip rates (Fig. 2B). We suggest that the short time span of observations prior to the 2002 earthquake yielded a lower geodetic slip rate. Despite the value of the GPS slip rate not matching our 103–104 yr slip rate well, the geodetic model is very consistent with right-lateral strike slip. Biggs et al. (2007) developed and used a multi-interferogram method utilizing InSAR (interferometric synthetic aperture radar) data to deduce the slip rate of the part of the central Denali fault for a 10 yr period prior to the earthquake. They find a slip rate of 10.5 ± 5.0 mm/yr (see Fig. 2). Leonard et al. (2007) and Marechal et al. (2015) used GPS geodesy to examine modern deformation in southwestern Yukon (Fig. 2B). Modeling the GPS data is not as straightforward as in central Alaska, due to significant and somewhat uncertain glacial isostatic adjustment corrections. Leonard et al. (2007) first documented significant (5–10 mm/yr) contractional deformation across the eastern Denali fault system; their study was superseded by that of Marechal et al. (2015), who had significantly better distribution of survey monuments. Marechal et al. (2015) found ∼5 mm/yr contraction across the Denali fault system, and for their northern transect, they found ∼7 ± 2 mm/yr of strike-slip deformation across the eastern Denali fault, and ∼4 ± 2 across a southern extension of the Totschunda fault. Neither value closely matches values that we infer; however, the total slip across the two fault systems is similar. Marechal et al. (2015) also discussed the possibility that distributed off-fault deformation may be considerable.

The slip rates at the ends of the Denali fault system are either low or zero. At the southeast end of the Denali fault system in the Yukon, Canada, there is no evidence of surface faulting along the bedrock trace of the fault more than 6 km southeast of Kluane Lake. In western Alaska, west of the Alaska Range, the Denali fault forms a striking lineament across the tundra, but there is no unequivocal evidence for recent movement. The Alaska active fault compilation by Koehler (2013) categorizes this section of the Denali fault as suspicious, meaning that linear features along the fault make it look like it may be active. Regardless, this section of the fault clearly has either a very low slip rate or a slip rate of zero.

Regional Active Fault Patterns

In considering the regional setting of the Denali fault system, it is important to consider the behavior of other active faults in the region (Fig. 7). North of the Alaska Range is the northern Alaska Range thrust system (Bemis and Wallace, 2007; Bemis et al., 2012, 2015; Burkett et al., 2016). These thrust faults extend along a 500 km length of the Alaska Range from Denali eastward to the Denali-Totschunda fault junction, and are oriented subparallel to the Denali fault system. They have been active throughout Quaternary time, and deformation likely began during Oligocene–Miocene time (Ridgway et al., 2007; Benowitz et al., 2011). South of the Denali fault, fewer active thrust faults have been identified, but the known or suspected active faults are southeast vergent and connect with, or horsetail into, the Denali fault system (Haeussler, 2008; Bemis et al., 2015). The best known of these is the Susitna Glacier thrust, which ruptured in, and was only discovered after, the 2002 Denali fault earthquake (Crone et al., 2004; Riccio et al., 2014). Other likely active or potentially active thrusts include the Broad Pass fault, the Broxson Gulch thrust, the Kahiltna thrust, and the McCallum Creek thrust (Plafker et al., 1994; Haeussler et al., 2017; Koehler et al., 2012; Waldien, 2015). The Colorado Creek thrust also has the same geometry, was active in Oligocene time (Trop et al., 2004), and could remain active. These thrusts south of the Alaska Range have the proper orientation and geometry to lead to a westward decrease in slip rate along the Denali fault system (Matmon et al., 2006; Haeussler, 2008). They also appear to have initiated during early to middle Miocene time (Haeussler et al., 2008; Riccio et al., 2014), indicating that they are related to the Yakutat collision, and may remain active today as the collision is ongoing.

The principal geometric relationship is that thrust faults north of the Denali fault system are subparallel to it, where as those to the south diverge from it in a horsetail fashion (Haeussler, 2008; Bemis et al., 2015). Thus, the thrust faults north of the Alaska Range are indicative of shortening roughly perpendicular to the Denali fault, whereas the thrusts on the south side of the Alaska Range are indicative of right-transpressive shortening oblique to the Denali fault. The reason for these differences are likely related to the larger scale setting of southern Alaska, discussed in the following section.

MODELS FOR SOUTHERN ALASKA NEOTECTONICS

The Denali fault system is the most significant intracontinental fault in Alaska, so the measured or inferred slip rates along the fault place significant constraints on geometric models of present-day deformation. We assess four models that include rotation, indentation, extrusion, and a combination of these (Fig. 8). We compare the predicted slip distributions for these models to our data (Fig. 7B). Each model infers that the Yakutat microplate collision is driving the deformation in interior Alaska along the Denali fault system (e.g., Haeussler, 2008) due to its northwestward movement along the Fairweather fault system and into interior Alaska along the Totschunda fault, and a hypothesized connector fault (Richter and Matson, 1971; Freymueller et al., 2008; Elliott et al., 2013). Although we focus here on geometrical models, there is widespread agreement that ongoing deformation in southern and central interior Alaska is driving flat slab subduction of the Yakutat microplate (e.g., Haeussler et al., 2000, 2008; Matmon et al., 2006; Abers, 2008; Haeussler, 2008; Freymueller et al., 2008; Mériaux et al., 2009; Benowitz et al., 2011; Finzel et al., 2011; Jadamec et al., 2013; Lease et al., 2016).

Our discussion of these models has one important assumption, which is that the highest slip rate on the Denali fault (∼13 mm/yr) is the value of the total shear strain across the Denali fault system. The geometry of the fault system supports this assumption. The present parallelism between the Totschunda fault and the North America–Pacific plate relative motion demands that there is nearly pure strike slip along the Totschunda fault and a greater degree of shortening across the Denali fault system west of the Totschunda fault intersection (Haeussler, 2008; Bemis et al., 2012, 2015). Moreover, recent examination of the Quaternary development of the thrust belt on the north side of the Alaska Range (Bemis et al., 2012, 2015) only shows contraction westward of the Totschunda-Denali fault junction. Aftershocks from the 2002 earthquake only show significant compression immediately west of the Totschunda-Denali fault junction (Vallage et al., 2014). Therefore, we infer that the combined slip rate on the Totschunda and eastern Denali fault must equal the maximum slip rate of ∼13 mm/yr for the Denali fault west of the junction (orange line, Fig. 7B). Combining the highest slip rate site on the eastern Denali fault and the adjacent site along the Totschunda fault (sites DFTR and DFNM) gives a slip rate of 15 mm/yr, close to the 13 mm/yr value to the west of the fault intersection. However, as most Totschunda slip rates are 7–9 mm/yr, this implies that the eastern Denali fault slip rate must quickly decrease eastward to 4–6 mm/yr (13 mm/yr – 9 mm/yr = 4 mm/yr; or 13 mm/yr – 7 mm/yr = 6 mm/yr), as shown by the brown line in Figure 2B. Thus, it appears that only a short length of the eastern Denali fault close to the intersection with the Totschunda has a relatively high slip rate.

The rotational model for the Denali fault system (Fig. 8A) was first proposed by St. Amand (1957) and was next examined by Stout and Chase (1980). This model explains many of the primary characteristics of the Denali fault system. The Denali fault fits a small-circle geometry and slip rate predictions from the Totschunda fault to the Denali region (Fig. 7B). Modern geodesy also demonstrates counterclockwise rotation of the region south of the Denali fault (Fig. 6; Fletcher, 2002; Freymueller et al., 2008).

The rotational model has some significant shortcomings. It assumes that the boundary of the southern Alaska block is a perfect small circle, but it is not (Stout and Chase, 1980). The Denali fault system cannot fit a small circle geometry in the region west of Denali. The rotational model also does not explain the geometry of the Denali fault system in western Alaska, or the low slip rate west of Denali, or GPS geodetic observations (Fig. 6). Moreover, the model predicts that the slip rate along the entire active Denali fault system should be constant, but we find that it is not (Fig. 7B).

The indentation model of the Denali fault system was posited by Mériaux et al. (2009), and it provides an explanation for the westward decrease in slip rate along the fault system (Fig. 8B). This model predicts a fundamental structural transition in the Denali fault system, in that it would change from dominantly right-lateral slip along the Totschunda fault, to right transpression in the middle section, to thrust faulting to the west. This model predicts that the slip rate on the Denali fault system should go to zero near Denali (Fig. 7B), that the western Denali fault should be inactive, and that there should be numerous thrust faults parallel to the Denali fault on both sides of the Alaska Range in the Denali region where the highest rates of contractional deformation are predicted. This model’s strength is that it provides an explanation for a westward decrease in the Denali fault slip rate, and it also provides an explanation for the nearly pure strike slip along the Totschunda fault and a greater degree of shortening or compression across the Denali fault system west of the Totschunda fault intersection (Haeussler, 2008; Bemis et al., 2012, 2015; Vallage et al., 2014).

The indentation model does not explain all the observations when examined in detail. For example, the fold and thrust belt north of the Alaska Range should be best developed and should extend westward of the Denali region. However, that is where it dies out. If the thrust belt is absent north of the Denali fault in the Denali area, an alternative possibility is that it is better developed on the south side of the Alaska Range. Although some active thrusts, such as the Broad Pass thrust and the Foraker thrust, are likely present (Haeussler et al., 2017), there is no evidence that these two structures are large enough or have a high enough slip rate to accommodate the entire decrease in Denali fault slip rate. In addition, there is no evidence for deformation causing widespread young exhumation of the western Alaska Range (Lease et al., 2016; Haeussler, 2008; Benowitz et al., 2011, 2012b, 2015). Moreover, in a study to assess both the Denali fault slip rate and the slip rates on the northern foothills fold and thrust belt, Bemis et al. (2015) found that the vector sum of both rates stays roughly constant at ∼9 mm/yr and that the vector rotates westward at a constant ∼20° angle to the Denali fault system. If a pure indentation model is correct, we would expect that the angle of shortening relative to the Denali fault should increase westward and the rate of shortening should increase westward, neither of which is observed (Bemis et al., 2015; green line, NAR thrusts in Fig. 7B). The indentation model does not explain the evidence for activity on the western Denali fault, because it predicts the slip rate should go to zero.

An extrusion model for Alaska Range tectonics (Fig. 8C) was proposed by Redfield et al. (2007), in which all of the accreted terranes of western Canada and Alaska were, and still are, moving northward and westward. The motion of these terranes is proposed as being driven by relative plate motions along the major right-lateral faults, including the Denali fault; and the terranes are ultimately driven toward the free face of the system, the Bering Sea margin. Redfield et al. (2007) envisioned that this model explains aspects of the framework geology from Cretaceous time through the present and that material is moving rapidly westward along the major dextral strike-slip faults of Alaska. The extrusion model partly encompasses the rotational model, in that it acknowledges the movement of material through the large-scale arc of the Denali fault system.

The extrusion model does the best in explaining the motion on the Denali fault system west of the Denali region, although it does not consider the pre-Miocene validity of the model. However, the extrusion model, as envisioned by Redfield et al. (2007), predicts a high slip rate on the western Denali fault, with a suite of active right-lateral faults extending to the Bering Sea margin; their model predicts that the sum of the slip rates on all right-lateral faults is consistent from eastern Alaska to western Alaska. The model also envisages westward-escaping blocks of continental Alaska thrusting over the Bering Sea margin. There is no evidence for a high slip rate on the western Denali fault or the Kaltag fault (Plafker et al., 1994), and there is no evidence for large-scale translation of blocks west of mainland Alaska over the Bering Sea margin (e.g., Worrall, 1991).

Given that simple end-member geometrical models for the neotectonics of southern Alaska are inadequate, we infer that a combination of rotation, indentation, and extrusion is the best explanation for all the slip rate observations (Haeussler, 2008; Fig. 8D). No geometric aspect of these models precludes combining them. We consider rotation of the southern Alaska block as important in explaining the right-lateral character of the Denali fault system (Fig. 6). We infer that indentation is occurring at some level to explain the westward decrease in the Denali fault system slip rate. We also infer that the entire southern Alaska block must be moving to the north-northwest to produce the northern Alaska Range thrust system. We infer that there is a small amount of extrusion in order to explain the recent activity on the western Denali fault, but it is not occurring on the scale and magnitude proposed by Redfield et al. (2007).

We explain some of the slip rates and their uncertainties in the context of the combination of the rotation, indentation, and extrusion models (Figs. 7B and 8D). The highest slip rates along the Denali fault system are ∼13 mm/yr. Given the geodetic observations of the Yakutat microplate moving northwestward at 50 mm/yr (Elliott et al., 2013), a total of ∼37 mm/yr of shortening (50 mm/yr – 13 mm/yr) is occurring across the entire St. Elias orogen. We infer that the highest slip rates of ∼13 mm/yr along the central Denali fault system are from counterclockwise rotation of the southern Alaska block. We infer that the westernmost slip rate of ∼3 mm/yr is the rate of westward extrusion of the southern Alaska block. However, we suggest that this rate is likely a maximum, as the fault trace west of our westernmost site is not well expressed, which suggests that the slip rate may be lower. Bemis et al. (2015) estimated ∼3 mm/yr as the maximum rate of shortening across the entire northern Alaska Range thrust system north of the Denali fault (Fig. 7B). We concur with this interpretation and infer that this is a maximum value for northward indentation of the southern Alaska block. The arc of the fold and thrust belt is indicative of north (not northwest) directed shortening, and we, along with Bemis et al. (2015), do not infer that this fold and thrust belt developed in the style suggested by Mériaux et al. (2009), but rather it is related to large-scale simple shear between the southern Alaska block and stable North America (Page et al., 1995; Haeussler, 2008; Bemis et al., 2015). The difference between the maximum slip rate value of ∼13 mm/yr and the maximum westward extrusion rate of ∼3 mm/yr is 10 mm/yr, which seems to demand that this is the value for the total shortening rate on all thrust faults that splay off the south side of the Denali fault system. These thrusts would be between the Denali-Totschunda junction and our westernmost site on the Denali fault (site DFMB), a distance of ∼575 km. The thrusts would accommodate shortening within the southern Alaska block related to indentation. A number of thrusts with the correct geometry have been identified, such as the Broad Pass thrust, the Susitna Glacier thrust, the McCallum Creek thrust, and the Broxson Gulch thrust. However, the only one with unequivocal activity is the Susitna Glacier thrust that ruptured in the 2002 Denali fault earthquake. There are compelling reasons to consider the Broad Pass thrust as active (see Haeussler et al., 2017). The McCallum Creek thrust is considered “suspicious[ly]” active (Koehler et al., 2012), although there is no evidence that demands that it is active, and also none that the Broxson Gulch thrust, exposed in bedrock, is active. Thus, there are not enough known active thrust faults with a high enough slip rate to explain satisfactorily the westward decrease in the Denali fault slip rate. Future work is needed to identify and characterize active thrusts on the south side of the Denali fault, or to refute aspects of this analysis (e.g., Waldien, 2015). Given that the Susitna Glacier thrust fault was only discovered after surface rupture in the 2002 earthquake, it seems likely that additional thrusts remain to be discovered. Regardless, our assessment of geometric models of deformation is important for anticipating the types of active faults, inferring rates of deformation, and determining earthquake hazards found in different parts of Alaska. Moreover, it highlights the intrinsic complexity of natural systems, as in other collisional orogens around the world (Tapponnier et al., 1982; Le Pichon et al., 1992; Wang et al., 2003).

ACKNOWLEDGMENTS

We thank Keith Labay for assistance in drafting some figures and Jeff Freymueller for sharing Figure 6. We appreciate reviews by Sean Bemis, an anonymous reviewer, and guest editor Jeff Benowitz that greatly improved the manuscript.

1Supplemental File S1. Site descriptions for features offset along the Denali fault system and error calculations. Please visit http://doi.org/10.1130/GES01447.S1 or the full-text article on www.gsapubs.org to view Supplemental File S1.
2Supplemental File S2. Cosmogenic 10Be and 9Be data from samples near the Denali fault system and calculations of age. Please visit http://doi.org/10.1130/GES01447.S2 or the full-text article on www.gsapubs.org to view Supplemental File S2.

Supplementary data