Fault scarps of strongly varying height cut glacial and alluvial sequences mantling the faulted front of the Teton Range (western USA). Scarp heights vary from 11.2 to 37.6 m and are systematically higher on geomorphically older landforms. Fault scarps cutting a deglacial surface, known from cosmogenic radionuclide exposure dating to immediately postdate 14.7 ± 1.1 ka, average 12.0 m in height, and yield an average postglacial offset rate of 0.82 ± 0.13 m/k.y. using simple scarp height (average 11.2 m, offset rate 0.76 ± 0.11 m/k.y. using vertical separation). We apply the offset rate to higher fault scarps to develop preliminary age estimates for the geomorphically older landforms, with an initial assumption of constant offset rate through time. The landform age estimates of 16.2 ± 3.9 ka to 45.9 ± 11.0 ka imply that glaciation and alluviation influenced the range front during marine isotope stages 2 and 3. However, fault offset rate variability, suggested by previous work to be attributable to Yellowstone ice cap deglacial processes, suggests that the fault scarp height pattern might also be interpreted as a reflection of strongly variable offset rates in landforms of only slightly contrasting age. These results demonstrate the need for detailed geochronology of isochronous landforms and sediments of multiple ages, in order to understand both faulting and glaciation on faulted range fronts.
Glacial and alluvial sequences mantling mountain range fronts frequently host scarps of active faults and are used to evaluate spatial and temporal variation of fault motion (e.g., McCalpin, 1996, in New Zealand; Scott et al., 1983, on the Wasatch fault, Utah; Ansberque et al., 2016, in Tibet). Glacial and alluvial sequences provide geomorphic and stratigraphic markers that are typically amenable to radiocarbon, luminescence, and cosmogenic exposure dating (e.g., Chen et al., 2012; Kenworthy et al., 2014; Licciardi and Pierce, 2008). Because glacial and alluvial sequences in the ranges often respond simultaneously and repeatedly to climate shifts in multiple valleys, they create multiple isochronous markers for evaluation of spatial and temporal variation of fault motion (Gillespie and Molnar, 1995; McCalpin, 1996; Howle et al., 2012; Thackray et al., 2013).
In some cases, faults of known slip rate can also be used to evaluate ages of glacial and alluvial sequences. However, this process is hampered by spatial and temporal variability of offset along individual faults and fault segments (e.g., Z. Lifton et al., 2015; Pierce and Morgan, 1992) and can lead to erroneous interpretations (Gillespie and Molnar, 1995).
Fault activity in glaciated areas may also be complicated by influences of deglacial unloading and meltwater release and infiltration (e.g., Hampel et al., 2007). These processes can influence fault slip rates during the deglacial period, confounding studies that focus exclusively on deglacial and postglacial landforms.
Understanding fault offset rates is often limited by sparse glacial and alluvial chronologies. In many locations, the chronology includes only a single advance, typically an advance correlative with the last glacial maximum (LGM, 26–18 ka; Mix et al., 2001) or deglacial events, rendering difficult the evaluation of temporal fault slip variability. This question is particularly relevant in regions with complex glacial histories spanning the last glacial cycle. For example, glacial chronologies document prominent advances during marine isotope stages (MIS) 4 and/or 3 (71–57 ka and 57–29 ka; age ranges from Lisiecki and Raymo, 2005) in addition to MIS 2 (29–14 ka) in Central Asia (Rother et al., 2014), New Zealand (Shulmeister et al., 2010; Putnam et al., 2013), southern South America (Darvill et al., 2015), western North America (e.g., Thackray, 2001; see following discussion), and elsewhere. Longer glacial chronologies, with multiple events separated by 10 k.y. or more, allow analysis of longer term fault behavior and, potentially, the use of fault offset rates to hypothesize ages of earlier glacial events.
This project, focusing on the Teton fault, uses a base of published cosmogenic radionuclide (CRN) glacial chronologic data (Licciardi and Pierce, 2008), coupled with high-quality lidar (light detection and ranging) topographic data (Teton National Park, 2014; Teton Conservation District, 2008; EarthScope Intermountain Seismic Belt LiDAR Project, 2008), to reveal probable variability of glacial landform ages and fault offset rates in time and space. Our study reveals challenges with using glacial chronologies, which are typically incomplete, for understanding fault offsets. The project also reveals new understanding of this major fault of the Yellowstone–Snake River Plain region, in which normal faulting and volcanism, influenced by migration of the Yellowstone volcanic field, dominate the Neogene and Quaternary geologic history (e.g., Pierce and Morgan, 1992). The project develops the best possible postglacial fault offset rate, given current knowledge of fault scarp topography and glacial geochronology, points to gaps in our understanding of the fault’s mechanisms and hazards, and suggests the types of data that are needed to understand this and other faults more fully. In particular, the Teton Range front is an ideal setting in which to examine the effects of advance and retreat of glacial ice on fault slip rates. Our results also point to possible pre-LGM mountain glacier advances not yet documented in the CRN chronologies, as well as the potential impacts of fault motion on glacial landform preservation. These findings inform the study of faulting in a variety of landscapes influenced by episodic glacial and alluvial deposition.
The Teton Range (Fig. 1) is in the northeastern Basin and Range province. The range exposes Archean metasedimentary rocks and Archean and Proterozoic intrusive igneous rocks, overlain unconformably by west-dipping Paleozoic sedimentary and Quaternary volcanic rocks (Love et al., 1992). The Teton fault is a N10°E-striking, east-dipping normal fault that separates the Teton Range from Jackson Hole. The fault has undergone 2.5–3.5 km of slip over the past 2–3 m.y. (Byrd et al., 1994), raising preexisting topography to form the highest elevations in the region. The Teton fault represents an anomaly among faults in the Snake River Plain–Yellowstone region. Surface scarps cutting Late Pleistocene glacial landforms are higher (12–36 m) and inferred fault offset rates greater (to 2.4 m/k.y.; Byrd, 1995) than along other regional normal fault systems. The Lost River fault in central Idaho, for example, displays scarps of 2–15 m in Late Pleistocene glacial and glacial-fluvial landforms (Staley, 2015) and yields trench-based slip rates of 0.18–0.3 mm/yr (Haller and Wheeler, 2010). Scarps of the Sawtooth fault are 4–9 m high in Late Pleistocene landforms and yield scarp-derived slip rates of 0.5–0.9 m/k.y. (Thackray et al., 2013). The height and variability of Teton fault scarps reflect older glacial landforms, very rapid offset rates, and/or highly variable offset rates since the peak of the last glaciation.
The Teton Range rises to 4200 m, 2400 m above Jackson Hole, the adjacent structural basin (Fig. 1). The Teton Range presents a major orographic barrier to moist westerly winds funneled along the eastern Snake River Plain, contributing to the relatively moist, montane climate (1500–2000 mm/yr) and supporting several extant cirque glaciers.
Jackson Hole has been occupied by glacial ice from two sources (Figs. 1 and 2; Pierce and Good, 1992; Licciardi and Pierce, 2008). The southern outlet lobe of the Yellowstone ice cap entered Jackson Hole from the north during the last two glaciations and imposed strong geomorphic effects on the basin and adjacent range front. Mountain glaciers formed in the Teton Range and expanded down several major valleys to cross the range front (Figs. 2 and 3). The glaciers entered western Jackson Hole during the last (Pinedale) glaciation and, presumably, during previous glaciations. As is discussed herein, the contrasting geomorphic effect of these two ice sources is an important basis for our examination of the fault scarp and associated glacial landforms.
This study builds upon a wealth of prior and ongoing research concerning faulting and glaciation on the Teton Range. The fault and fault mechanics were studied in detail during the 1990s and have been revisited occasionally since then. Highly relevant to our analysis is the work of Byrd (1995) and Byrd et al. (1994), who documented the neotectonics and mechanics of the fault and studied the variability of scarp height in some detail, and Smith et al. (1993) and Ostenaa et al. (1993), who reported the long-term and short-term fault offset history. O’Connell et al. (2003) assessed a variety of evidence from scarp analysis and geophysical and geological data to determine fault history, mechanics, and potential seismic ground motion for the Jackson Lake Dam. Most of these studies characterized and/or discussed the surface fault scarp patterns and the variability in fault scarp height that are the focus of this study. However, the chronology of glaciation was not well understood at the time, and various assumptions were made regarding ages of glacial landforms.
Byrd (1995) reported findings from a shallow trench in the southern portion of the fault zone at Granite Creek (Fig. 4). Notably, for logistical reasons the trench was excavated through a low scarp (3.5 m) in alluvial landforms at the mouth of the canyon, rather than through the high postglacial scarps (12–14 m) nearby, and thus reveals only the mid-late Holocene history of the fault. The trench site is in the southern portion of the fault, where O’Connell et al. (2003) suggested the slip rate is lower than in the central portion of the fault. Byrd (1995) and Byrd et al. (1994) concluded that only two fault rupture events have occurred since 7.9 thousand calibrated radiocarbon years (cal ka), representing 4.1 m total slip (7.9 cal ka, 2.8 m slip; 7.0–4.8 cal ka, 1.3 m slip). Using the trench results and the 13.1 m scarp height cutting the deglacial surface nearby, Byrd (1995) determined that the mid-Holocene offset rate was lower (0.3–1.1 mm/yr) than the average offset rate between deglaciation (17–13 ka) and 7.9 ka (1.3–2.4 mm/yr). This inference was cited by Hampel et al. (2007) in assessing their model of deglacial influences on Teton fault slip rates. Byrd (1995) also measured a fault plane striking N35°E and dipping 85°E.
The Granite Canyon trench was discussed by Machette et al. (2001), who compiled published offset rates and recurrence interval estimates for several fault sections, including an inferred offset rate of 2 mm/yr. Hampel et al. (2007) modeled the impact of deglacial meltwater infiltration and crustal unloading on fault offset rates, suggesting that either or both processes contributed to apparently rapid early postglacial fault offset rates, as inferred from the postglacial scarp height and trench results by Byrd (1995).
More recently, geophysical studies have revealed shallow subsurface structure of the fault and surrounding sediments. Shallow seismic surveys determined a 70° fault dip at Taggart Lake (Figs. 2 and 3) and 13 m offset of subsurface horizons at Taggart Lake and Granite Creek (Fig. 4; Thackray et al., 2015; Zellman et al., 2016).
The history of Teton mountain glaciers and especially the Yellowstone ice cap in Jackson Hole have been well studied. Pierce and Good (1992) described details of landforms and sediments of the southern lobe of the Yellowstone ice cap. Licciardi and Pierce (2008) documented ages of the last two major advances of the Yellowstone ice cap into Jackson Hole, during MIS 6 [136 ± 13 ka (10Be)] and MIS 2 (ca. 14.6 ± 0.7 ka (10Be) and earlier], as well as ages of two Teton mountain glacier moraines discussed in detail in the following. Additional moraine sequences on the eastern range front and on the western flank of the range have been dated, and the chronology of the ice cap moraines has been improved (Licciardi et al., 2014; J. Licciardi, 2015, personal commun.).
The timing of range-front deglaciation is very important to this study. The youngest moraine at Jenny Lake dates to 14.7 ± 1.1 ka (10Be) [Licciardi and Pierce, 2008; ages recalculated by J. Licciardi (2015, personal commun.), as described in Methods discussion]. Licciardi and Pierce (2008) also dated moraine boulders at Lake Solitude, in the cirque at the head of that drainage, and found that that moraine was constructed within several hundred years of the Jenny Lake moraines, the ages overlapping within analytical uncertainties. Therefore, the range front at Jenny Lake was deglaciated shortly after 14.7 ± 1.1 ka (10Be). Larsen et al. (2016), in a study of range-front and high-elevation lakes, documented onset of range-front lake sedimentation centuries before 13.8 ka, and continuing retreat until at least 13.5 ka. Those results are broadly consistent with the 10Be chronology.
Current knowledge of Teton glaciation is within a broad context of glacial geology and geochronology spanning the western United States. In the western continental interior, most CRN glacial chronologies primarily document dominant MIS 2 glacial advances and not MIS 4 or 3 glacial advances (Gosse et al., 1995; Licciardi et al., 2004; Licciardi and Pierce, 2008; Laabs et al., 2009, 2011, 2013; Sherard, 2006). However, 14C, 36Cl, optically stimulated luminescence (OSL), U-series, and relative weathering data indicate that MIS 4 and/or MIS 3 ice advances were also widespread. A 14C chronology and coastal stratigraphy were used (Thackray, 2001) to document MIS 4 and 3 glaciation in the coastal Olympic Mountains that was far more extensive than MIS 2 glaciation, a result confirmed and expanded by OSL dating (Wyshnytzky et al., 2015; Marshall, 2013; Staley, 2015). In the eastern Cascade Range, the Porter and Swanson (2008),36Cl CRN chronology documented extensive pre–MIS 2 glaciation, and Colman and Pierce (1992) suggested, using weathering rinds, an extensive MIS 3 ice advance. The Colman and Pierce (1992) weathering rind study also suggested extensive MIS 4 glaciation at McCall, Idaho, represented by a moraine more recently bracketed by MIS 6 and MIS 2 10Be CRN dates by Phillips et al. (2007; W. Phillips, 2015, personal commun.). In Thackray et al. (2004), moraine morphometry, soil data, and lacustrine cores were used to suggest extensive MIS 3 glaciation in the Sawtooth Mountains. Alluvial sediments also have yielded ages indicative of MIS 3–correlative glaciation, in the Lost River Range, Idaho (OSL dating; Kenworthy et al., 2014), the Wind River valley, Wyoming (U-series dating of pedogenic carbonate; Sharp et al., 2003), and West Yellowstone, Montana (obsidian hydration dating; Pierce et al., 1976).
We exploit a unique glacial geomorphic scenario to evaluate the Teton fault scarp and landforms constructed by Teton mountain glaciers. As noted above, ice from two main sources has dominated the study area. The southern outlet glacier of the Yellowstone Plateau ice cap has entered Jackson Hole at least twice (Fig. 1), while mountain glaciers originating in the Teton Range fluctuated beyond the range front multiple times during the Pinedale glaciation, as evidenced by extensive terminal moraine belts and numerous lateral moraines (Figs. 2–4). During the MIS 6 Bull Lake glaciation, the Yellowstone outlet glacier filled Jackson Hole to 2700–3050 m elevation on the range front and eroded Teton mountain glacier landforms of MIS 6 age and older (Fig. 1; Pierce and Good, 1992; Licciardi and Pierce, 2008). During the MIS 2 Pinedale glaciation, the Yellowstone outlet glacier reached only as far as a moraine belt immediately south of Jackson Lake (Figs. 1 and 2). Thus, only post–MIS 6 mountain glacier landforms are preserved at the Teton Range canyon mouths south of Jackson Lake, as MIS 2 Yellowstone ice did not scour the range front.
To measure offset on the Teton fault and to map range-front glacial landforms, we utilized bare-Earth lidar topographic data sets from Teton National Park (2014), as well as from the Teton Conservation District (2008) and the EarthScope Intermountain Seismic Belt LiDAR Project (2008). We analyzed the topography of the 1 m bare-Earth data sets in ArcMap 10.2 (ESRI, www.esri.com) and measured fault offset by deriving topographic profiles perpendicular to the fault scarp using the ArcMap profiler tool.
We focus on simple scarp heights to evaluate variability of along-strike fault offset. We also calculated offset via slope regression-derived vertical separation methods (e.g., Thompson et al., 2002; Amos et al., 2010; Supplemental Items 1–31). In valley-floor settings where geomorphic surfaces are continuous and slope gently, the vertical separation method produces consistent results. Vertical separation measurements are only slightly lower than are simple scarp height measurements, and the resulting slip rates are indistinguishable. However, vertical separation methods are problematic on the higher lateral moraine and fan landforms, as those surfaces are highly irregular and yield vertical separation measurements that are highly variable and, in some cases, in conflict with fundamental geomorphic relationships (see Results discussion).
We make several observations regarding uncertainty in our scarp-height measurements. First, the fault scarp cuts complex glacial, alluvial, and hillslope landforms and has diffuse top and bottom edges, leading to uncertainty in picking the top and bottom elevations of the scarp. Second, the fault-zone geomorphology is complex, and ensuring that we are measuring the scarp across isochronous surfaces is a particular challenge. Third, the ArcGIS profiler tool interpolates the profile from the lidar digital elevation model (DEM), at a resolution chosen by the user; however, that process should introduce only minor uncertainty. Fourth, where the fault scarp height is large, the uncertainty is greater.
The complex geomorphology local to the profiles requires that the profile locations be carefully chosen to avoid modifications by hillslope instability, alluvial deposition or erosion, or lacustrine deposition. Therefore, our rigorously chosen profiles are singular profiles at the most suitable location on each landform, and measuring repeat profiles to determine uncertainty is untenable. As described here, we rejected more than half of our original data set because we considered the local geomorphology, and thus the measurements, unsuitable.
We quantify fault-scarp height uncertainty using detailed analysis in a glacial-alluvial and structural setting very similar to that in Thackray et al. (2013), in which scarp height on the Sawtooth fault was measured (1) from bare-Earth lidar DEMs (resolution similar to that of the Teton data set), (2) from unfiltered non-bare Earth lidar DEMs that included vegetation returns, and (3) from field survey, the presumed most accurate method. It was determined (Thackray et al., 2013) that the scarp height measurements varied within 0.2 m for a 2.5-m-high scarp. That measured uncertainty is 8% of the fault scarp height. As the sources of uncertainty scale with fault scarp height, in this study we estimate measurement uncertainty of 8% of scarp height, and incorporate that uncertainty into our analyses.
We are able to calculate fault-plane slip at our Taggart Lake study site (Fig. 3) using a preliminary, geophysics-based fault dip estimate. The dip of the Teton fault is generally poorly known, but recent shallow seismic surveys across the scarp in the basin that contains Taggart Lake (Fig. 3) allow estimation of near-surface fault dip of 70°, to 220 m depth (Zellman et al., 2016). This fault-dip estimate permits calculation of fault-plane slip at that location. Byrd (1995) determined a fault dip of 85° from the shallow trench at Granite Creek, but only to ∼4 m depth. Because the fault dip is unconstrained elsewhere, we generally rely on scarp heights as the most accurate expression of offset. Given the inferred steep dip of the fault plane, as inferred from these two locations, the calculated fault-plane slip is only slightly larger than the vertical scarp height.
We chose relatively simple, single-strand scarp morphologies in valley floors, lateral moraines, and alluvial fans, and we field checked the fault scarp heights and their geomorphic contexts in several locations. In several locations, grabens are bounded on their basinward edges by antithetic scarps, and we subtracted the height of the antithetic scarps from the overall scarp height measurements. We rejected scarp heights from our data set in locations that we determined to be compromised by postglacial slope failure, alluviation, or fluvial erosion, as well as from scarps of otherwise ambiguous morphology.
To determine the fault offset rate, we measured the scarp height at 15 locations in valley-bottom deglacial moraines and alluvial surfaces (5 used, 10 ambiguous and discarded). All data and locations are in Supplemental Items 1 and 2 (see footnote 1). The 10 discarded scarp profiles were found to have been altered to varying degrees by postglacial alluvial or hillslope processes. It is notable that the average measurements of those scarps were very similar to the average of the retained scarp heights. We also measured offset using vertical separation methods on the five retained profiles. We utilized the locally determined timing of deglaciation, from range-front moraine CRN ages of 15.9 ± 0.8 and 14.7 ± 1.1 ka (10Be) at Jenny Lake [Licciardi and Pierce, 2008; recalculated by J. Licciardi (2015, personal commun.) using revised 10Be production rates of N. Lifton et al. (2015), and the Lm scaling scheme (Lal, 1991; Balco et al., 2008)]. The age of the younger moraine is within analytical uncertainty of unpublished ages of the youngest moraines at Taggart Lake and other moraine systems on the Teton Range front (J. Licciardi, 2015, personal commun.). The deglacial timing, confirmed by moraine boulder ages from the cirque upvalley of Jenny Lake (see discussion of Previous Work), correlates with widely documented regional glacier retreat, reviewed in Young et al. (2011). We calculated an average fault offset rate spanning the 14.7 ± 1.1 ka (10Be) period since deglaciation reset the valley-floor deglacial surfaces, which variably include ground moraine, ablation moraine, and glacial-fluvial sediment. We applied the geomorphically derived offset rate to higher scarps cutting geomorphically older landforms.
We assumed initially that the average rate of fault offset has been constant through time and with distance along the length of the fault. With that assumption, we applied the postglacial offset rate to scarp heights to calculate provisional age estimates for the older fault-offset landforms, as a starting point for discussion. One of us (Staley, 2015) performed a similar analysis for alluvial terraces cut by the Lost River fault, using published trench-derived offset rates (Haller and Wheeler, 2010), and derived landform age estimates that are broadly consistent with OSL dates in alluvial fan gravels in an adjacent drainage (Kenworthy et al., 2014).
We examined our results further in the context of other published offset rates for the Teton fault. In particular, we examined the middle to late Holocene offset rates determined from the Granite Creek fault trench, reported by Byrd (1995) and Byrd et al. (1994) and cited by Hampel et al. (2007) in their modeling study. The results imply that 70% of postglacial slip occurred before 8 cal ka. In the Discussion we address this and other possible causes of spatial and temporal variability in Late Pleistocene–Holocene fault offset rates, as well as possible sources of error in our analysis.
Geomorphology of Range-Front Canyon Mouths
Mountain glaciers have constructed several canyon-mouth moraine complexes that are cut by fault scarps (Fig. 2). A prime example of the geomorphic relationships between the deglacial surface, the adjacent lateral moraines, and the fault scarp is at the mouth of Avalanche Canyon at Taggart Lake (Fig. 3).
We studied intensively the valley surface and subsurface geology, the moraine sequence, and the scarp morphology in the Avalanche-Taggart area (Figs. 5 and 6), and this location is key to our analysis. At least seven lateral moraines mark the flanks of the canyon mouth, and several of them merge to form a broad, dead-ice terminal moraine complex rising above outwash surfaces that occupy much of the floor of Jackson Hole (Figs. 2 and 3). The terminal moraine complex correlates with the double terminal moraine at Jenny Lake, dated to 15.9 ± 0.8 and 14.7 ± 1.1 ka (10Be) (Licciardi and Pierce, 2008; J. Licciardi, 2015, personal commun.). Small (1–8 m high) terminal moraines mark the landscape upvalley of the dead-ice landscape, and lateral moraines rise 2–120 m above the canyon mouth deglacial terrain.
The fault scarp height varies markedly through the moraine sequence at the mouth of the canyon. The valley floor is cut by a 12.4-m-high scarp (13.2 m fault-plane slip with 70° fault dip; 10.7 m vertical separation), the footwall is mantled by a surface exposing glacially sculpted bedrock and glacial sediment, and the hanging wall also is mantled by glacial sediments and landforms adjacent to the scarp at the profile site (Figs. 3 and 6).
The highest lateral moraine in the Avalanche-Taggart sequence is 120 m above the faulted valley floor described above, and is cut by a much higher scarp (Figs. 3, 5, and 6). On the hanging wall, the high moraine is a single composite ridge built by ice-marginal deposition from both the Taggart and Bradley lake basins. The ridge has a typical Late Pleistocene lateral moraine morphology, i.e., a single ridge with an undulatory downvalley profile and isolated closed depressions, mantled with common boulders as large as 5 m. Within 100–200 m of the main fault scarp on the hanging wall, small (1–3 m) antithetic scarps mark the edge of a graben depression that extends to the main fault scarp.
Two footwall morainal ridges connect to that hanging-wall composite moraine (Fig. 6). A moraine ridge very similar to the hanging-wall moraine extends southeast from the mouth of Garnet Canyon, which feeds Bradley Lake on the north side of the moraine complex. On the Taggart side of the footwall, the moraine morphology is more complex. Common to abundant boulders, similar to the Bradley and composite moraines and typical of glaciated surfaces of the last glaciation, mantle a muted ridge extending only ∼100 m upvalley of the fault scarp. Farther up the ridge, that surface gives way to a more diffuse ridge mantled with uncommon erratic boulders that are mostly buried in soil, suggesting that it is an older glacial landform or colluvial landform. As expected, bedrock dominates the ridge 200 m upvalley of the scarp and further upslope. The scarp separating that muted Taggart footwall ridge from the composite hanging-wall moraine is 26.8 m high, while the scarp separating the Bradley footwall moraine from the hanging-wall moraine is 36.5 m high (although the Bradley measurement was derived from a nonlinear profile, as shown in Fig. 6). In contrast, the scarp cutting the lateral moraine on the southern edge of the Taggart Lake basin is only 13.3 m high (Figs. 3 and 5).
One concern in our analysis is the possibility that postglacial lacustrine and/or fluvial sedimentation has filled the footwall area of the Avalanche-Taggart valley floor (and analogous ones in other valleys) and significantly reduced the height of the measured scarp. However, two pieces of data argue against significant errors from this source. (1) Shallow seismic surveys at the scarp profile site indicate that the fault offsets a subsurface seismic velocity transition, inferred to be a bedrock-glacial diamicton contact, by ∼13 m (Thackray et al., 2015). Thus, the surface scarp height is similar to the vertical offset of the buried geologic horizon and appears to be faithful to postglacial offset. Similar correlation between surface scarp height and offset of subsurface horizons was revealed at Granite Creek (Zellman et al., 2016; Fig. 4). (2) Two shallow excavations in a broad marsh between the youngest moraine and the scarp, near the center of the basin and far from potential hillslope sediment sources, revealed only 0.65 m of slackwater sediment above coarse gravel or diamicton, indicating minimal infilling of the hanging wall above a possible deglacial surface or postglacial alluvial deposits. The marsh does not extend to the fault scarp, but rather gives way to an irregular, slightly higher ground moraine surface that is similar to the land surface on the footwall immediately adjacent to the scarp. For these reasons, we find sediment infilling to be minimal adjacent to the scarp, and thus the carefully evaluated valley-floor scarp measurement at this location accurately reflects postglacial fault offset. Direct field observation at Granite Creek, Phelps Lake, Bradley Lake, and Glacier Gulch support our measurements of carefully chosen fault scarps cutting other valley floors.
Calculation of Average Postglacial Fault Offset Rate
Through the broader study area, the Late Pleistocene glacial and alluvial landforms are offset by 11.2–37.6-m-high fault scarps. Scarps that cut the valley-floor deglacial surfaces (variably ground moraine, ablation moraine, or outwash) are 11.2–13.5 m high (mean 12.0 m; Supplemental Item 1; see footnote 1). Assuming that the deglacial surface dates to 14.7 ± 1.1 ka (Licciardi and Pierce, 2008), we calculate an average vertical offset rate (Supplemental Item 1) of 0.82 ± 0.13 m/k.y. As noted in the Methods discussion, we eliminated 10 of 15 original fault scarp profiles from our data set because of observed modification by postglacial processes. However, the range of fault scarp heights on the discarded profiles is similar to those retained, and exclusion of those measurements altered the offset rate only slightly.
Vertical separation, calculated using methods described in Thompson et al. (2002) and Amos et al. (2010), applied to the same topographic profiles, is 10.3–13.5 m where the fault cuts the deglacial surfaces (mean 11.2 m; Supplemental Item 1; see footnote 1), yielding a vertical separation rate of 0.76 ± 0.11. The two rates are indistinguishable within their uncertainties.
However, vertical separation calculations for the higher, geomorphically older landforms are inconsistent with the geomorphology. On some profiles, the results are consistent with geomorphic relationships, but on several other profiles, the vertical separation results (Supplemental Item 1) are inconsistent with the clear geomorphic sequence. Specifically, vertical separation at five locations on higher, geomorphically older landforms is less than vertical separation on the valley-floor deglacial surfaces. Those older landforms should have vertical separation equal to or greater than that of the deglacial surfaces, given the clear geomorphic relationships. The inconsistent results are rooted in the complex glacial topography, which exerts a strong influence on the topographic profile regressions that are the basis for the vertical separation measurement. The valley-floor deglacial surfaces, in contrast, are of consistent, gentle slope, and provide a solid basis for the measurement.
For these reasons we rely on simple scarp height measurements in our further analysis (see following), as those measurements reflect the readily observable geomorphic patterns of fault scarp height and are consistent with the geomorphic age sequence. Simple scarp height measurements on steeply sloping landforms may overestimate slip, but we emphasize the relative rather than absolute values.
Assuming that the 70° fault-plane dip estimate from Taggart Lake basin (Zellman et al., 2016) is representative of fault dip at all of our scarp profile locations, average fault-plane slip rate would be 0.87 m/k.y. (fault dip uncertainty unknown), or 0.81 m/k.y. using vertical separation analysis of the offset. As noted, fault-plane dip is only known at the Taggart Lake geophysical survey site (to 220 m depth) and to ∼4 m depth at Granite Creek, and we do not use fault-plane slip rates in our further discussion focusing on the direct measurement of fault scarp height.
Estimation of Ages of Geomorphically Older Landforms Cut by Higher Scarps
As a means of developing provisional age estimates for the older moraines and associated landforms, and as a starting point for discussion, we use our average postglacial vertical fault offset rate and assume that the rate has been constant since the older landforms were constructed. The fault scarp is higher on geomorphically older lateral moraines and alluvial fans (Figs. 3–8). As described here for the Taggart Lake area, the scarp cutting the deglacial surface is 12.4 m high, while the scarp cutting the highest lateral moraine is 26.8 m high (Figs. 3 and 6), despite consisting of similar sediments. We observe the same scarp-height relationship in other valleys (e.g., Glacier Gulch, Granite Creek; Fig. 2; Supplemental Items 1 and 2; see footnote 1). The strongly contrasting scarp heights imply that these range-front landforms are unlikely to be of similar ages. We estimate ages for the older range-front glacial and alluvial landforms (Supplemental Items 1 and 2) of 16.2 ± 3.9 ka to 45.9 ± 11.0 ka (assuming a constant offset rate; see Discussion). The oldest moraine age estimated by this method is 43.9 ± 10.5 ka; a scarp cutting an alluvial fan yielded a slightly older age estimate. These older ages are correlative with MIS 3.
As noted in the Methods discussion, we also calculated age estimates for offset moraines and alluvial fans using offset rates derived from the Granite Creek trench. Those rates imply that 70% of fault offset represented by scarps occurred between deglaciation and 8 cal ka (Byrd, 1995; Byrd et al., 1994; Hampel et al., 2007). Using those rate insights, we calculated landform ages that are as old as 38 ka, with the four highest scarps producing age estimates between 30 and 38 ka, also within MIS 3. We have made similar calculations for several other published offset rates, which produce varying ages for the older landforms. A published offset rate of 1.6 m/k.y. (Smith et al., 1993) yielded maximum age estimates of 31 ka (late MIS 3), but also yielded unrealistic Holocene ages for range-front glacial landforms with lower scarps. We discuss possible inferences regarding fault offset rate variability in the following.
Our new lidar-derived fault scarp measurements, coupled with recent improvements in the geochronology of Teton glaciation by Licciardi and Pierce (2008, as updated), provide fresh analysis of Teton fault activity and glacial history. We have produced a more closely constrained average offset rate for the past 14.7 k.y. (0.82 ± 0.13 m/k.y. using simple scarp height; 0.76 ± 0.11 m/k.y. using vertical separation), and have revealed the likelihood of pre–MIS 2 glacial landforms on the range front. We discuss multiple possible interpretations of the marked variability of fault scarp heights across the glacial and alluvial sequence, and place our findings in the context of regional fault and glaciation history.
Multiple Hypotheses to Explain Scarp Height Variability
There are four possible explanations for the along-strike scarp height variation.
The offset rate varies in space across landforms of broadly similar age. This appears unlikely, as scarp heights vary by as much as ∼15 m over short lateral distances, as little as 0.5 km (Figs. 3 and 4).
The offset rate, calculated using valley-floor landforms, is in error because of hanging-wall sedimentation processes, and the actual offset rate is better reflected by the high scarps on the high lateral moraines, which may date to late MIS 2. As described in the Results, we find that at Taggart Lake, postglacial sedimentation has not affected the measured scarp height. The terminal moraine sequence there is well dated (J. Licciardi, 2015, personal commun.), and the postglacial fault offset is well constrained. The clear glacial geomorphology on both the hanging wall and footwall at Granite Creek (Fig. 4) supports similar interpretation, as do geophysical studies at both sites (Thackray et al., 2015; Zellman et al., 2016). We have less detailed data in other valley-floor settings, but our field observations and careful site selection suggest that our method of calculating the average postglacial offset rate is valid.
This explanation would also imply a constant offset rate of as much as 2.4 m/k.y. if the high lateral moraines date to late MIS 2, as well as ∼20 m of sedimentation on the hanging wall since deglaciation, neither of which is supported by data or observations.
The offset rate varies in time across landforms of broadly similar, late MIS 2 age (e.g., 20–15 ka). It is likely that offset rates vary in time, as inferred from trench and fault scarp data in the southern portion of the range front by Byrd (1995) and as supported conceptually and numerically by Hampel et al. (2007). The rate contrasts would have to be substantial if the higher moraines are as young as 20 ka. For example, the offset rate from 20 to 15 ka would have to be ∼2.9 m/k.y. to account for the 14.4 m scarp height contrast (26.8 m versus 12.4 m) for Taggart Lake (Fig. 3). That rate is higher than other estimates. The model-based slip rate estimates proposed by Hampel et al. (2007) are as high as 2.33 m/k.y. during deglaciation at the center of the fault, when deglacial unloading is incorporated into their model, and Byrd (1995) calculated a rate as high as 2.4 m/k.y. while assuming deglaciation at 13 ka. If the higher moraines are somewhat older, that offset rate estimate of 2.9 m/k.y. would be less.
The glacial and alluvial landforms vary in age, as reflected in the highly variable fault scarp heights. As described in the Results, an assumed constant fault offset rate, determined from the offset of the deglacial surface, implies landform ages of 16.2 ± 3.9 ka to 45.9 ± 11.0 ka. The age estimates are within MIS 2 and 3, and correlate broadly with syn-LGM and pre-LGM, Late Pleistocene ice advances documented or proposed in western U.S. mountain ranges (e.g., Gosse et al., 1995; Licciardi et al., 2004; Licciardi and Pierce, 2008; Laabs et al., 2009, 2011, 2013; Sherard, 2006; Colman and Pierce, 1992; Thackray, 2001; Thackray et al., 2004, Marshall, 2013; Staley, 2015) and with well-dated, pre-LGM, glaciation-influenced alluvial aggradation (Sharp et al., 2003; Kenworthy et al., 2014).
We favor a combination of explanations 3 and 4. The existing trench observations and dates (Byrd, 1995), albeit limited in time span and spatial extent, suggest that the fault offset rate has varied with time, in particular having been much slower after 7.9 cal ka than in the period between deglaciation and 7.9 cal ka. However, the highest scarps would require very high offset rates if they date to 20 ka or later, so the higher moraines must be significantly older than the deglacial surface, and must predate the post–16 ka terminal moraine complex.
Implications for the Teton Fault and Similar Regional Faults
As was proposed by Byrd (1995) and Byrd et al. (1994), and supported by modeling work of Hampel et al. (2007), the Teton fault likely varies in offset rate, even on time scales of just a few thousand years [i.e., 14.7 ka (10Be) to 7.9 cal ka versus 7.9 cal ka–present]. Our results suggest that the offset rate may also have been higher prior to 14.7 ka, depending on the interpretation of very high fault scarps cutting the older lateral moraines in the sequence.
The model results of Hampel et al. (2007) suggested that deglaciation of the Jackson Hole lobe of the Yellowstone ice cap influenced fault offset rate, through glacial-isostatic unloading and meltwater influx. Here we suggest that the offset rate may have been higher before the peak of MIS 2, i.e., before the time period considered by Hampel et al. (2007). The Hampel et al. (2007) modeling results would suggest that deglaciation of a previous Jackson Hole ice cap lobe and Teton mountain glaciers would favor a higher, earlier offset rate, but evidence is lacking in the study area (Licciardi and Pierce, 2008).
While 10Be CRN moraine chronologies document only MIS 6 and 2 Yellowstone ice caps, limited indirect or nonradiometric data suggest MIS 3 to early MIS 2 ice on the Yellowstone Plateau. Pierce et al. (1976) used obsidian hydration dating to infer that the West Yellowstone outlet glacier reached a terminal position ca. 35–30 ka. Adjacent to the path of the north Yellowstone outlet glacier near Gardiner, Montana, Sturchio et al. (1994) applied U-series dating of perched travertine to infer ice thickness; they concluded that that outlet lobe reached maximum thickness ca. 40 and 26 ka, with a readvance 18 ka. Thus, if Yellowstone ice reached the western and northern outlets during MIS 3 and early MIS 2, it is plausible that it similarly advanced to the southern outlet and into Jackson Hole and influenced fault offset rates following ice retreat. There is no direct evidence of such an advance. In the northern and western ice cap outlet areas where MIS 3 ice is suggested, it is similarly unclear how much ice retreat would have occurred prior to the MIS 2 advance. In any case, a more variable ice load with time would likely have important implications for the ice loading-unloading model of Hampel et al. (2007).
Does the fault offset rate vary spatially along the range front? O’Connell et al. (2003) compiled fault offset data suggesting that the slip is greatest in the central section of the fault near Jenny Lake, and that slip diminished to the north and south. Smith et al. (1993) inferred that the fault is segmented, although Ostenaa and Gilbert (1988) concluded that the fault geometry is consistent with slip on a single continuous fault. The suggestion of high slip rate in the central section of the fault is based on high scarps similar to those described herein, coupled with an assumed equivalent age of offset glacial landforms. Given the local variability of fault scarp height (e.g., Figs. 3, 4, and 6), we consider equivalent age to be untenable.
Furthermore, our data set shows that offset of the deglacial surface (11.2–13.5 m) varies little along strike, from the central section of the fault (Jenny Lake to Taggart Lake; Fig. 2) to the southern section of the fault (e.g., Granite Creek; Fig. 7). The higher scarps cutting geomorphically older landforms vary along strike, but equivalent age is best assumed for the deglacial surface. We have made additional measurements of the deglacial surface at 10 locations, which differ little from the scarps included in the data set reported here. While we did not find those locations suitable to include in the data set, the scarps cutting the deglacial surface in general show as much variability within individual valleys as along strike.
This study has been conducted on a range front with excellent lidar topographic data and excellent terminal moraine geochronology, yet our understanding of the Teton fault still suffers from multiple uncertainties. In order to understand the fault processes fully, firm dating of landforms spanning the glacial and alluvial sequence, especially direct dating of landforms and sediments cut by the fault, is necessary.
Implications for Regional Glacial History through the Last Glacial Cycle (MIS 5d-2)
Our inference regarding lateral moraine age estimates is important for preservation of glacial landforms on normal-faulted range fronts. As the basin subsides through time, older landforms, especially the lower relief terminal moraines, are more likely to be buried. This is particularly the case in this setting. The Yellowstone ice cap outlet glacier delivered abundant sediment to Jackson Hole, and constructed an extensive outwash surface (Figs. 1 and 2) that occupies much of the lowland and partly buried the outer Jenny Lake moraine [Pierce and Good, 1992; dated to 15.9 ± 0.8 ka (10Be) by Licciardi and Pierce, 2008; J. Licciardi, 2015, personal commun.]. Thus that extensive outwash surface must be of similar age or younger relative to that moraine. Teton-sourced ice advances that extended beyond the range front before that time would likely have constructed moraines now buried below that outwash surface or eroded during outwash deposition.
If the older lateral moraines on the Teton front are as old as our analysis suggests (middle to late MIS 3), that would imply that climatic events observed in the maritime Pacific Northwest (Thackray, 2001; Marshall, 2013; Staley, 2015) strongly affected glaciers in the interior western U.S. This finding would be contrary to, for example, that in Thackray (2008), in which it was inferred, from a review of published chronologies, that MIS 3 and 4 glaciation was extensive mainly in the maritime Pacific Northwest. Pre-LGM ice advances in the western interior United States have been suggested from relative weathering and alluvial sediment geochronology (Pierce et al., 1976; Colman and Pierce, 1992; Thackray et al., 2004; Sharp et al., 2003; Kenworthy et al., 2014) and may be more widespread than generally documented by existing CRN moraine dating, which mainly demonstrates MIS 2 glaciation (Gosse et al., 1995; Licciardi et al., 2004; Licciardi and Pierce, 2008; Laabs et al., 2009, 2011, 2013; Sherard, 2006).
The uncertainties expressed here regarding both offset rates and glacial landform ages reflect the pattern of existing radiometric ages in the study area. While the terminal moraine sequences are well dated (Licciardi and Pierce, 2008; J. Licciardi, 2015, personal commun.), radiometric ages are very few in the lateral moraine sequence and in younger alluvial sediments. Published radiocarbon ages directly pertinent to the fault are limited to one trench in the southern portion of the fault trace (Byrd, 1995). Full understanding of the glacial and fault history is intertwined, and will be achieved only with improved chronologic control of landforms and of faulted sediments in trenches and lacustrine cores (e.g., Larsen et al., 2016).
Our analysis of Teton fault scarps has yielded an average postglacial (post–14.7 ± 1.1 ka) Teton fault offset rate of 0.82 ± 0.13 m/k.y. (0.76 ± 0.11 m/k.y. using vertical separation methods), provided information suggestive of temporally variable fault offset rates, and identified possible pre-LGM (MIS 3) glacial landforms deep in the interior of western North America. The highly variable fault scarp heights must imply variable landform ages, but also appear to reflect varying fault offset rates through MIS 3 and 2. The degree to which these two factors influence the range front geomorphology can be determined through improved geochronology of glacial and alluvial landforms and fault offset events.
Despite a rich history of research regarding glaciation and faulting on the Teton Range front, as well as highly detailed lidar-derived topographic data, much remains to be learned. The data sets remain insufficient for full characterization of this major regional structure and for full understanding of the glacial history. This situation is repeated on many faulted range fronts where glacial and alluvial chronologies are yet less well known and where fault processes are more complex. Thorough understanding of both faulting and glaciation requires robust dating of multiple isochronous surfaces, such that a time-transgressive view of glaciation and fault motion can be derived.
This work has benefited from voluminous previous work by other researchers and from discussions with Mark Zellman, John Hebberger, David Rodgers, Joe Licciardi, and Darren Larsen. The work of Staley was partly supported by U.S. National Science Foundation Grant EAR-1024850 to Idaho State University, and field investigation and geophysical work were supported by the University of Wyoming–National Park Service field station’s small grants program, subawards 1002029E and 1002732E to Idaho State University. We thank Colin Amos and two anonymous reviewers for insightful reviews that expanded and honed the article considerably.