Riser diachroneity, lateral erosion, and uncertainty in rates of strike-slip faulting: A case study from Tuzidun along the Altyn Tagh Fault, NW China



[1] At late Quaternary timescales, offset fluvial terrace risers are among the most common landforms used to determine rates of strike-slip faulting. Although diachroneity in the age of riser segments on opposite sides of the fault has been noted previously, an unexplored source of uncertainty associated with deriving slip rates from these markers centers on quantifying the size of the displacement uncertainty such diachroneity introduces. To evaluate the impact of riser diachroneity, we investigated the Tuzidun site (37.73°N, 86.72°E) along the Cherchen He reach of the active central Altyn Tagh Fault. The east bank of the channel is flanked by a left-laterally offset terrace riser. While the measured offset is 54 ± 3 m, geochronologic measurements and analysis of riser topography indicate that the downstream riser segment formed between 6.0 ± 0.8 ka and 5.7 ± 0.4 ka, while the upstream riser segment may have been laterally refreshed as recently as 0.5 ± 0.2 ka. A valley wall on the west bank of the channel places a maximum limit of 38 ± 6 m on the amount of possible lateral erosion of the upstream riser. This bound, in turn, limits the total offset since formation of the downstream riser to range from 54 ± 3 to 89 ± 7 m. Together, these observations bracket the millennial Altyn Tagh Fault slip rate to range from 9.0 ± 1.3 to 15.5 ± 1.7 mm a−1. More generally, this investigation shows that the observed riser displacement does not necessarily correlate with the age of either riser segment (downstream or upstream of the fault) in cases where one segment is displaced while the other is subjected to lateral erosion. If this diachroneity goes undetected, erroneous slip rate measurements are likely to result.

1. Introduction

[2] A fundamental problem in the study of active tectonics is determining the extent to which slip rates along faults vary in both time and space. While there is little question that fault slip is episodic at the timescale of individual earthquakes, an intriguing question is whether slip is also variable when averaged over the timescale of several earthquakes or longer. Several studies report data sets that have been interpreted to indicate such temporal variations in slip rate at late Quaternary timescales for several fault systems, including the Eastern California Shear Zone [e.g., Frankel et al., 2007b; Oskin and Iriondo, 2004; Oskin et al., 2008; Peltzer et al., 2001] and the Basin and Range province [e.g., Friedrich et al., 2003; Wallace, 1987], as well as along the Karakorum [e.g., Chevalier et al., 2005], the Altyn Tagh [e.g., Mériaux et al., 2004; Wallace et al., 2004; Washburn et al., 2001], and the Dead Sea faults [Marco et al., 1996]. Examples of spatial variation in slip rate within a given fault system include between the northern and southern Eastern California Shear Zone [Frankel et al., 2007a; Oskin et al., 2008], across the Eastern California Shear Zone [Frankel et al., 2007b], and along the strike of the Altyn Tagh [e.g., Mériaux et al., 2005; Meyer et al., 1998] and Kunlun faults [Kirby et al., 2007]. Additionally, spatially and temporally anticorrelated slip behavior has been suggested between the San Jacinto and San Andreas faults [e.g., Bennett et al., 2004], the Mojave section of the Eastern California Shear Zone and Los Angeles Basin faults [Dolan et al., 2007], and the Blackwater–Little Lake fault system and the Garlock Fault [Peltzer et al., 2001], as well as between the North and East Anatolian faults [Ambraseys, 1971]. Interest in such variable slip rate behavior has spawned the development of mechanical models to explain strain rate variations, including conjugate fault interactions [e.g., Ambraseys, 1971; Hubert-Ferrari et al., 2003; Peltzer et al., 2001], regional stress transfer between fault strands [Bennett et al., 2004; Dolan et al., 2007], along-strike slip consumption by kinematically linked fold-and-thrust belts [e.g., Meyer et al., 1998], rheological heterogeneities such as a weak seismogenic layer [Kenner and Simons, 2005], and changes in pore fluid pressure [Chéry and Vernant, 2006].

[3] Precise chronology and offset vector measurements of faulted landforms at late Quaternary timescales provide the observational foundation for evaluating mechanical models seeking to explain such temporal and spatial fault slip rate variations. Laterally offset terrace risers provide an example of a geomorphic strain marker commonly used in studies designed to measure strike-slip fault slip histories [e.g., Berryman, 1990; Cowgill, 2007; Harkins and Kirby, 2008; Lensen, 1968; Mériaux et al., 2004; Mériaux et al., 2005; Prentice et al., 2003; Van der Woerd et al., 1998; Weldon and Sieh, 1985]. These markers are defined by relatively simple geometric elements, including a more steeply sloping riser face that is bound by more gently sloping upper and lower terrace treads (Figure 1). Laterally faulted risers are further subdivided as either sheltered or unsheltered, based on fault motion relative to the active channel, where an unsheltered riser is transported into the path of a stream, while the sheltered segment is transported away (Figure 1). The unsheltered and sheltered riser segments correspond to Bull's [1991] “leading” and “trailing” edges, respectively. Importantly, a laterally offset riser can yield an average slip rate measurement if the relationship between the magnitude of displacement and the age of the riser segments can be determined.

Figure 1.

Schematic illustration of the geometric elements of a laterally faulted riser. The unsheltered riser segments are transported into the path of a stream, whereas the sheltered segment is transported away.

[4] A fundamental requisite for neotectonic studies seeking to evaluate the question of spatial and temporal slip rate variations is that they have sufficient resolution to distinguish uniform from variable strain histories. An important component of evaluating this resolution is characterizing the uncertainty of the measurements. For studies using laterally offset terrace risers, analytical uncertainties related to geochronology of the terrace introduce errors that usually fall within the range from 1 to 23%, with a mean value of 10% (calculated from Table 1 of Cowgill [2007]). In contrast, uncertainties in relating adjacent terrace abandonment ages to riser displacements range from 11 to 67%, with a mean value of 40% (calculated as half the range between adjacent terrace abandonment ages as a percentage of the mean age between the terraces reported in Table 1 of Cowgill [2007]). Early treatments of this latter problem focused on combining the abandonment age of the lower tread with the displacement value of the offset riser. This reconstruction is valid for fluvial systems in which displacement of a riser edge is eliminated by lateral scarp erosion by the active stream until the lower surface at the base of the riser is abandoned [e.g., Berryman, 1990; Bull, 1991; Grapes, 1993; Knuepfer, 1987; Mériaux et al., 2004, 2005; Van der Woerd et al., 1998, 2002]. However, there are settings where the upper surface more closely approximates the age of an offset riser [e.g., Cowgill, 2007; Harkins and Kirby, 2008; Lensen, 1968; Zhang et al., 2007]. This latter situation is favored in systems where lateral stream erosion is insufficient to remove displacement accumulated along a riser edge when the lower surface is occupied by a stream. Thus, slip can be accumulated directly following abandonment of the upper terrace tread. For most fluvial systems, it is probable that some component of lateral scarp refreshment occurs synchronously with riser displacement, in which case the age of the offset riser is intermediate between the upper and lower tread abandonment ages. Thus, a conservative approach to this problem is to date both the upper and lower bounding treads to bracket the riser age and to pair this age range with the riser displacement [Cowgill, 2007].

[5] Another potential source of uncertainty in these studies centers on the displacement recorded by an offset terrace riser and in particular, cross-fault correlation of offset riser segments [Gold et al., 2007; Harkins and Kirby, 2008]. Harkins and Kirby [2008] explored this idea in an investigation along the Kunlun Fault, where they combined riser morphology and geochronology to estimate the age of offset riser segments. These authors observed that laterally offset risers are found to have steeper slopes on unsheltered riser segments than the corresponding sheltered riser segments. They interpret these topographic differences to indicate differential timing of cessation of lateral stream refreshment. While their analysis makes an important contribution in the methods which can be used to improve the age constraint for an offset riser segment, an unexplored implication from their investigation centers on the impact that across-fault riser diachroneity may have on the observed displacement measurement.

[6] Here we suggest that the observed riser displacement does not necessarily correlate with the age of either riser segment (downstream or upstream of the fault) in cases where one segment is displaced while the other is subjected to lateral erosion. The goals of this study are to systematically evaluate this source of uncertainty in slip rates derived from laterally offset fluvial terrace risers by quantitatively evaluating the size of such uncertainty, identifying measurements that can minimize it, and applying them to a site to determine a slip rate for which the epistemic uncertainty is as well quantified as the data and current theory permit. In particular, to examine the implications for riser diachroneity on observed displacements, we investigated a new site near Tuzidun along the central Altyn Tagh Fault in NW China (Figure 2). The site features an offset terrace riser displaying across-fault diachroneity, which we document through topographic, neotectonic, and geochronologic measurements. From these data, we explore the implications of this diachroneity and conclude that at Tuzidun, the unsheltered upstream riser segment experienced significant lateral erosion following formation of the sheltered downstream riser. To constrain the uncertainty introduced into a slip rate measurement from this diachronous marker, we develop two reconstructions for the site which pair the tightly bracketed age of the downstream riser with a range of upstream piercing points that account for lateral erosion. We conclude with a consideration of the implications for riser diachroneity on other offset riser investigations. In addition to contributing to faulted riser theory, these data also provide a new mid-Holocene slip rate measurement for the Cherchen He reach of the Altyn Tagh Fault.

Figure 2.

Location and fault map. (a) Faults within the Indo-Asian collision, compiled from previous work [Cowgill et al., 2004b, 2009; Mériaux et al., 2004, 2005; Peltzer et al., 1989; Tapponnier et al., 2001; Taylor et al., 2003; Taylor and Peltzer, 2006; Thatcher, 2007; Yin et al., 2007]. (b) Map showing geometry of the central Altyn Tagh Fault and location of the Tuzidun site relative to previous slip rate investigations, including geodetic transects [Bendick et al., 2000; Wallace et al., 2004], paleoseismic investigations [Washburn et al., 2001, 2003], and geomorphic slip sites [Cowgill, 2007; Mériaux et al., 2004; Cowgill et al., 2009].

2. Geologic Background

2.1. Altyn Tagh Fault

[7] The >1200 km long, active, left-lateral Altyn Tagh Fault (ATF) defines the northwestern margin of the Tibetan Plateau and is a first-order structure accommodating the ongoing continental collision between India and Asia (Figure 2a) [Molnar and Tapponnier, 1975; Tapponnier and Molnar, 1976]. Slip along the ATF related to the Indo-Asian collision is thought to have initiated in either the middle Eocene (∼49 Ma) [Yin et al., 2002] or in the late Oligocene to early Miocene [Yue et al., 2001]. Estimates of total displacement since initiation for the central ATF system range from 280 to 500 km [e.g., Cowgill et al., 2003; Gehrels et al., 2003a, 2003b; Molnar and Tapponnier, 1975; Peltzer and Tapponnier, 1988; Ritts and Biffi, 2000; Sobel et al., 2001; Yin and Harrison, 2000; Yue et al., 2001]. Additional debates seek to resolve whether the ATF extends to crustal [e.g., Burchfiel et al., 1989] or lithospheric depths [e.g., Wittlinger et al., 1998], and also whether the system is purely strike slip [e.g., Cowgill et al., 2000] or partitioned with strike-slip motion occurring along the main trace of the ATF and thrusting along the north ATF system [e.g., Avouac and Tapponnier, 1993; Bendick et al., 2000; Molnar et al., 1987; Yue et al., 2004]. Most generally, these debates seek to understand the role of the ATF system in accommodating Indo-Asian convergence. One of the most contested problems related to this debate is the late Quaternary slip rate of this structure [e.g., Bendick et al., 2000; Elliott et al., 2008; Mériaux et al., 2004; Peltzer et al., 1989].

[8] Shortening systems that link with the ATF system to the SW and NE are thought to absorb left slip and lead to systematically decreasing slip rate away from the central ATF (Figure 2) [e.g., Burchfiel et al., 1989; Meyer et al., 1998; Peltzer et al., 1989]. Thus, the central ATF has been the focus of studies seeking to constrain the highest slip rates across the fault. This portion of the fault system is divided into three quasi-straight segments, which are the Cherchen He, Qing Shui Quan, and Manyi reaches from west to east (Figure 2b). These three linear segments are separated by two right-stepping double restraining bends along the fault in the Sulamu Tagh and Akato Tagh ranges [Cowgill et al., 2004b]. Previous slip rate measurements across the central ATF are not in agreement. They include geodetic transects (∼9 mm a−1) [Bendick et al., 2000; Wallace et al., 2004], far-field slip rate measurements from a regional GPS network (6–9 mm a−1) [Chen et al., 2000; Shen et al., 2001; Zhang et al., 2004], an InSAR study (6–16 mm a−1) [Elliott et al., 2008] paleoseismic investigations (10–20 mm a−1) [Washburn et al., 2001, 2003], and geomorphic studies (9–27 mm a−1) [Cowgill, 2007; Mériaux et al., 2004; Cowgill et al., 2009]. Thus, while determining the slip rate for the ATF is an important problem, the apparently conflicting results have obscured the late Quaternary slip history of this fault system. Owing to the contentious nature of this structure, it is important to treat the uncertainty in slip rate measurements carefully.

2.2. Tuzidun Site

[9] The Tuzidun site is located along the principal trace of the ATF along the Cherchen He reach (Figure 2b). The site is named after an inselberg ∼4 km southeast of the site. This reach of the fault is delineated by the southern margin of the Altyn Shan to the north. These mountains are dominantly composed of Proterozoic metasediments and lesser ultramafic, gneissic, and granitoid bodies [Delville et al., 2001; Liu, 1988]. South flowing streams confined to narrow stream valleys drain the southern margin of the mountains. Where the channels cross the fault and exit the range front they form a regional bajada.

[10] The Tuzidun site is located at the intersection of a south flowing ephemeral stream and the active trace of the locally N75°E striking ATF (Figure 3). North of the fault, the modern channel is defined by a narrow valley flanked by abandoned fluvial terraces. South of the fault, the modern channel widens and merges with a regional bajada. A shutter ridge on the southside of the fault defines the west bank of the channel. The focus of this study is a laterally offset fluvial terrace riser preserved on the east bank of the modern channel.

Figure 3.

Topographic and neotectonic map of the Tuzidun site. Topographic and neotectonic map of Tuzidun. (a) Uninterpreted hill-shaded DTM derived from T-lidar survey. (b) Neotectonic observations and 0.5 m contour map overlain on the hill-shaded DTM. The Tuzidun site is situated at the intersection of a south flowing, ephemeral stream channel and the active trace of the ATF. Abandoned terrace surfaces (T1–T4) flank the modern channel (T0) and are numbered as a function of increasing age and elevation. Terraces are distinguished from a fan complex (F1–F1″) on the south side of the fault. On the east bank of the modern channel, the left-laterally offset T2/T1 and T2/F1″ riser segments show an apparent offset of 54 ± 3 m, but as much as 38 ± 6 m of lateral erosion of the upstream T2/T1 riser could increase the magnitude of the offset to 89 ± 7 m. Contacts were mapped using a Leica TCR407 power total station. The underlying 0.3 m DTM was generated from a point cloud of ∼34 million topographic points collected using a Trimble GX200+ terrestrial T-lidar unit. The hill-shaded image is illuminated from the northwest. Artifacts in the DTM (e.g., north central portion of map) result from sparse topographic data. Trench locations discussed in text and the auxiliary material are indicated.

3. Neotectonic Geology of Tuzidun

3.1. Methods

[11] We undertook a field investigation of the Tuzidun site during two field seasons in 2006 and 2007, which included a neotectonic and topographic survey (Figure 3). The surveying instruments included a Leica TCR407power total station and a Trimble GX DR200+ Terrestrial Light Detection and Ranging unit (T-lidar). At a distance of 100 m, precision for the total station is 3.4 mm in the horizontal and vertical directions and 2 mm in the distance measurement, while precision for the laser scanner is 5.8 mm in the horizontal direction, 6.8 mm in the vertical direction, and 7 mm in the distance measurement.

[12] We used the total station to map the contacts of critical landforms, such as channels and riser crests following the methods described by Cowgill et al. [2004a] and Gold et al. [2006]. We used a naming convention where T0 is the active channel and lowest surface and T4 is the highest and oldest terrace (Figure 4). We mapped the boundaries of the fault zone and used break lines to record locations of sharp slope breaks. We also used the total station to measure the positions of a series of control points, which were then also measured using the laser scanner to enable coregistration of the neotectonic mapping and lidar-derived digital elevation model. The control points consisted of 1 m long rebar rods pounded into the ground at prominently visible points throughout the site. With the total station, we measured 5139 topographic points and 8 control points.

Figure 4.

Photographs showing key field relationships at Tuzidun. (a) Overview of the left-laterally offset T2/T1 and T2/F1″ riser segments. (b) Overview of the southern portion of the field site, showing the varying surface characteristics of the F1′ and F1″ lobes of the F1 fan complex, the downstream T2/F1″ riser, and the ATF. (c) The upstream T2 surface at the crest of the T2/T1 riser. (d) The downstream T2 surface, the crest of the T2/F1″ riser, and the riser intersection with the fault. This photo also highlights the wedge of loess at the base of the T2/F1″ riser and the base of the fault/F1″ interface, from which organic samples were collected for 14C dating. Downward pointing triangles indicate the location of the active trace of the left-lateral ATF, and the solid black lines indicate the crest of risers, with hachures on the riser escarpment. Photograph locations and view directions are indicated in Figure 3. In Figure 4a, tripod and backpack indicate scale; in Figure 4b, the riser offset values in the middle distance indicate scale; and in Figures 4c and 4d a person indicates scale.

[13] Our survey with the T-lidar scanner included 13 stations covering an area of 300 m (E–W) by 400 m (E–W × N–S) and the measurement of ∼34 million topographic points with an average point density of ∼280 points m−2. At each station, we also used the unit to measure the position of three to four targets located over control points. We then stitched the scans by using RealWorksSurvey version 6.1.2 software to coregistering targets common to overlapping scans. The reported mean distance error in the target-based scan registration was ∼11 mm. Taking into account the registration errors and precision of the instruments, we estimate the internal precision of the merged, raw point cloud data set to be sub-4 cm in both horizontal and vertical directions. The topography presented in Figure 3 is a 0.3 m cell size digital terrain model (DTM) generated by decimating the original T-lidar data set to ∼1.3 million topographic points (i.e., ∼11 points m−2) and then applying an ordinary kriging algorithm [e.g., Dubrule, 1983] using the Spatial Analyst extension in ArcGIS 9.2.

[14] To compare the morphology of the offset riser segments, we extracted 12 topographic profiles from the raw, undecimated T-lidar point cloud data (Figure 5 and the auxiliary material). We sampled the topography from the undecimated point cloud data to avoid interpolation errors associated with sampling from the DTM. The profiles were sampled from 10 cm wide swaths, spaced 2 m apart. Use of 10 cm wide swaths maximized point cloud sampling required to define the local topography of the terrace riser while minimizing scatter due to along-strike variations in topography along the trend of the terrace riser. We aligned the profiles so that the riser midpoint was centered on the origin, and the riser height values that we report are equivalent to the “scarp height” discussed by Hanks [2000]. In Figure 5, we present representative profiles from the upstream and downstream riser segments. The reported height and slope values represent a synthesis from all 12 topographic profiles. The remaining profiles are included in the auxiliary material.

Figure 5.

Representative topographic swath profiles extracted from the T-lidar point cloud data across the (a) upstream T2/T1 and (b) downstream F2/F1″ risers. These data indicate that T2/T1 riser slopes more steeply than T2/F1″, which suggests riser refreshment occurred more recently upstream than downstream. Also note the asymmetry of the T2/T1 riser, which slopes more steeply near the base than at the crest. The riser heights also differ. Collectively, these profiles suggest that the upstream T2/T1 and downstream T2/F1″ risers are not isochronous. Profiles are centered at the inflection point along riser face and locations are shown on Figure 3a. Riser slope and height values are averages compiled from an additional 10 profiles included in the auxiliary material.

3.2. Mapping and Survey Results

[15] At Tuzidun, the south facing ATF mole track separates abandoned alluvial fans to the south downstream from a ∼40 m wide valley to the north upstream that is flanked to the east by a flight of alluvial terraces (Figure 3). To the east of the valley, the fault cuts alluvial terraces and fan surfaces, and to the west, it delineates the southern margin of the range front. On the south side of the fault, an eastward tapering shutter ridge of alluvium-mantled bedrock borders the east bank of the modern channel.

[16] Locally, the active trace of the ATF is defined by a 5 m wide mole track and is characterized by a south facing escarpment 6–10 m high (Figure 4). The mole track was likely produced by surface rupture during the most recent earthquake events. Spring activity along the fault zone has resulted in local mass wasting and development of dense grassy vegetation relative to the surrounding area, which combine to obscure the mole track. East of the modern channel, a southward tapering wedge of eolian and reworked silt abuts the base of the mole track (Figure 4d).

[17] Four abandoned terrace surfaces are preserved at Tuzidun (Figure 3). From high-to-low elevations, they are characterized by increasing surface roughness, increasing clast size, and decreasing thickness of loess cover, relative to the modern channel. Aerial distributions of most surfaces are insufficient to develop robust amplitude and wavelength measurements [Frankel and Dolan, 2007], so we report maximum roughness heights for the surfaces. The highest of these surfaces, T4, is correlative with a dissected alluvial surface, Qfo, which extends along strike for tens of kilometers to both the east and west (Figure 4a). It is preserved north of the fault and also mantles the bedrock-cored shutter ridge. T4 and a lower T3 have low-relief surfaces (<0.1 m) with no relict primary depositional topographic roughness. We interpret T4 and T3 to have been emplaced by a coalescing set of streams that includes the modern drainage and catchments to the east. The intermediate T2 occurs both upstream and downstream of the fault, on the east bank of the modern channel. This surface exhibits <0.3 m surface roughness (Figures 4c and 4d). Boulders are sparse on the surface of T2 and are fractured, disaggregated, and salt weathered where they do crop out. The T2 surface slopes 5°–10° to the E-SE. The F1 fan complex, south of the fault and on the eastern side of the modern channel, exhibits up to 1.0 m surface roughness (Figure 4b). We subdivide this convex-up surface into a series of separate lobes, which become increasingly more loess covered and exhibit lower-relief bar and swale topography from west to east (Figure 4b). From west to east these units are mapped as F1, F1′, and F1″. Upstream of the ATF, the T1 surface is preserved on both banks of the active channel. The T1 deposit exhibits bar-and-swale topography with <1.0 m relief. The T1 surface is bouldery, vegetated, and has <8 cm thick loess cover (Figure 4b). While T1 and F1 have similar relative heights above the modern channel, their textural differences along with the topographic and geochronologic analyses presented below, indicate that these surfaces are not correlative.

[18] Our primary focus is the vertical escarpment separating the T2 and F1″ surfaces south/downstream of the fault and the T2 and T1 surfaces north/upstream of the fault, the T2/F1″ and T2/T1 risers, respectively. The downstream T2/F1″ riser is 4.7 ± 0.4 m high (Figure 5a) and trends N20°W. Eolian and reworked silt blankets this linear escarpment (Figure 4d). At the base of the terrace riser, the silt deposit thins westward to form a westward tapering wedge of material that abuts the riser and covers F1″. The upstream T2/T1 riser differs from its downstream counterpart in several ways. It is higher, with 7.5 ± 0.2 m separating the T2 and T1 surfaces (Figure 5a). Although it is oriented ∼N20°W, the trend of this riser is less linear than the T2/F1″ riser. The T2/T1 riser is also more extensively vegetated than the T2/F1″ riser. Spring sapping at the base of the T2/T1 riser is spatially correlated to small <3 m wide and <1 m deep rotational slumps, which occur along the riser face. This face is also blanketed with silt, though there is not a clear wedge of silt materials at the base of the riser. Topographic and geochronologic analyses presented below indicate that the cross-fault T2/F1″ and T2/T1 riser segments are diachronous.

3.3. Results of Trench Excavations

[19] To constrain the depositional history at the site, we excavated 13 trenches into the critical landforms at Tuzidun (Figures 3b and 6 and the auxiliary material), and detailed stratigraphic observations and trench photographs are provided in the auxiliary material. In general, the trenches reveal layered terrace deposits composed of well-indurated gravel and boulder conglomerates with silt and sand matrices. There is very little soil development within the terrace deposits, which is not surprising given the young surface ages indicated by the geochronology results described below (<6 ka), the high elevation (3.3 km above sea level), and the arid environmental regime. The terrace deposits are capped by loosely consolidated, laminated silts, which we interpret to represent a mixture of primary windblown silts (loess) and also reworked loess that has been transported downslope, as indicated by pebbles within the deposit. In cross section, these loess packages have a wedge-shaped geometry at the bases of both the ATF scarp and the T2/F1″ riser (Figure 6). Because deposition of these silt deposits followed abandonment of the scarps which underlie them, radiocarbon samples from within the loess wedges place important minimum bounds on the time the active stream abandoned these features.

Figure 6.

Photographs of trenches and compilation of stratigraphic observations and interpretations for (a) T2, (b) T2/F1″ riser face, (c) base of the fault/F1″ interface, and (d) T1. The 14C sample positions and 10Be depth profile sample intervals are plotted on the stratigraphic columns. Detailed unit descriptions and photographs of additional trenches are provided in the auxiliary material.

[20] From the trenches, we extracted organic samples for radiocarbon (14C) dating and quartz-rich clasts for terrestrial cosmogenic nuclide (TCN) exposure dating. The goals of these geochronologic analyses were to constrain the age of the deposition and abandonment of the terraces and fans, to evaluate the cross-fault correlation of terraces, and to bracket the age of the offset T2/F1″ and T2/T1 riser segments. We conducted 14C analyses on samples collected from below the T2 tread on both sides of the ATF (maximum tread abandonment age), from loess and colluvial gravel deposits that cap the downstream T2/F1″ riser face (minimum riser abandonment age), and from deposits capping and within the T1 terrace deposit (minimum and maximum abandonment ages, respectively) (Figure 6). We conducted 10Be analyses on amalgamated gravels samples collected from two depth profiles, north and south of the ATF, excavated into the T2 deposit at the crest of the displaced riser (upper terrace abandonment age) (Figure 6).

4. The 14C Geochronology

[21] We collected 52 dateable organic samples for 14C dating (Table 1), and additional information on sample preparation and analytical methods is provided in the auxiliary material. The majority of the samples were woody plant fragments, root materials, and dung, with fewer specimens of charcoal, peat, and animal fur. We were not always able to categorize sample material from field observations. This was particularly true for distinguishing isolated fragments of woody plant twigs from root materials. We adopted a conservative approach and collected all isolated plant fragments to increase the age control for the site, making note of those samples which we suspected to be roots based on field characteristics. This sampling approach has the disadvantage that we obtained both ancient and modern age signals. In the results section below, we rely on the age patterns along with photographs and field notes to discriminate modern samples from those carrying a meaningful signal with regards to the formation age of the geomorphic units at the site.

Table 1. Radiocarbon Geochronologya
SampleUCI AMSDepth Relative to Treadb (cm)TrenchStratigraphic ContextMaterialcFraction ModernD14C14C age (years B.P.)Calibrated Age Ranged (cal years B.P.)MeanSigma
  • a

    Results from the 14C AMS analyses. The 14C ages were calibrated using InterCal04 [Reimer et al., 2004] with the calibration software OxCal version 4.0.1 [Bronk Ramsey, 1995, 2001]. Modern samples are italicized.

  • b

    Positive value indicates above tread; negative value indicates below tread.

  • c

    Material as interpreted in the field and through examination under microscope.

  • d

    Age range is reported at 2 sigma.

  • e

    Sample with replicate analyses.

C-141227−1T1/T0 riserin terrace (T1)Wood0.9865 ± 0.0015−13.5 ± 1.5110 ± 1526224143119
C-232079−48T1/T0 riserin terrace (F1)Wood or Root1.0124 ± 0.001612.4 ± 1.6−95 ± 155446504
C-34122855T1NE-Ain colluvial materialWood1.2774 ± 0.0026277.4 ± 2.6−1960 ± 20
C-44122933T1NE-Ain colluvial materialWood0.9752 ± 0.0016−24.8 ± 1.6200 ± 15294−5145149
C-54123027T1NE-Ain colluvial materialWood0.9663 ± 0.0049−33.7 ± 4.9275 ± 45478−3238241
C-6412316T1NE-Ain colluvial materialWood0.9302 ± 0.0014−69.8 ± 1.4580 ± 1563854058949
C-741232−20T1NE-Ain terrace (T1)Wood0.9415 ± 0.0014−58.5 ± 1.4485 ± 1553450752014
C-841233−40T1NE-Ain terrace (T1)Wood0.9321 ± 0.0014−67.9 ± 1.4565 ± 1563253558348
C-941234−31T1NE-Bin terrace (T1)Root0.9809 ± 0.0014−19.1 ± 1.4155 ± 152835144139
C-1041235−34T1NE-Bin terrace (T1)Root1.1208 ± 0.0018120.8 ± 1.8−910 ± 15
C-1141236−35T1NE-Bin terrace (T1)Dung or Peat0.9587 ± 0.0014−41.3 ± 1.4340 ± 1547531639580
C-1241237−39T1NE-Bin terrace (T1)Dung or Peat1.6899 ± 0.0026689.9 ± 2.6−4210 ± 15
C-1341238−41T1NE-Bin terrace (T1)Dung or Peat0.9553 ± 0.0014−44.7 ± 1.4365 ± 1549732541186
C-1441239−47T1NE-Bin terrace (T1)Dung or Peat0.9668 ± 0.0015−33.2 ± 1.5270 ± 1542328835568
C-1541240−69T1NE-Bin terrace (T1)Root1.2974 ± 0.0020297.4 ± 2.0−2085 ± 15
C-1641247105FltWdgSE-Aloess wedge fault/F1″Dung0.9698 ± 0.0021−30.2 ± 2.1245 ± 20310−3154156
C-1741248105FltWdgSE-Aloess wedge fault/F1″Dung0.9896 ± 0.0016−10.4 ± 1.685 ± 1525631143112
C-184124935FltWdgSE-Aloess wedge fault/F1″Charcoal0.7516 ± 0.0016−248.4 ± 1.62295 ± 2023522206227973
C-194132542FltWdgSE-Aloess wedge fault/F1″Charcoal0.7430 ± 0.0015−257.0 ± 1.52385 ± 2024662346240660
C-204132646FltWdgSE-Aloess wedge fault/F1″Charcoal0.7393 ± 0.0017−260.7 ± 1.72425 ± 20268123552518163
C-214132728FltWdgSE-Aloess wedge fault/F1″Charcoal0.7391 ± 0.0015−260.9 ± 1.52430 ± 20268523562520165
C-22412670FltWdgSE-Aat contact between loess and F1″Wood0.5517 ± 0.0061−448.3 ± 6.14780 ± 90570753125509197
C-23412681FltWdgSE-Aloess wedge fault/F1″Charcoal0.5278 ± 0.0052−472.2 ± 5.25130 ± 80617456595916258
C-244126971FltWdgSE-Bloess wedge fault/F1″Wood or Root0.8102 ± 0.0016−189.8 ± 1.61690 ± 2016911537161477
C-254127045FltWdgSE-Bloess wedge fault/F1″Charcoal or Root0.8507 ± 0.0016−149.3 ± 1.61300 ± 1512871181123453
C-264127124FltWdgSE-Bloess wedge fault/F1″Root0.7560 ± 0.0014−244.0 ± 1.42245 ± 1523382159224890
C-2735712104.5T1SE-Aloess wedge T2/F1″Dung0.7277 ± 0.0013−272.3 ± 1.32555 ± 1527472617268265
C-2835713104T1SE-Aloess wedge T2/F1″Dung0.7292 ± 0.0015−270.8 ± 1.52535 ± 20274325022622121
C-2935714103T1SE-Aloess wedge T2/F1″Wood0.7615 ± 0.0012−238.5 ± 1.22190 ± 1523082145222782
C-303571595T1SE-Aloess wedge T2/F1″Animal fur0.6887 ± 0.0037−311.3 ± 3.72995 ± 45334030073174167
C-313571686T1SE-Aloess wedge T2/F1″Dung0.7314 ± 0.0013−268.6 ± 1.32515 ± 15272524972611114
C-323571777T1SE-Aloess wedge T2/F1″Dung0.7083 ± 0.0014−291.7 ± 1.42770 ± 2029272791285968
C-333571852.5T1SE-Aloess wedge T2/F1″Dung0.6915 ± 0.0012−308.5 ± 1.22965 ± 1532103077314367
C-3435719118.5T1SE-Bloess wedge T2/F1″Dung0.8798 ± 0.0033−120.2 ± 3.31030 ± 351054803928126
C-3535720118.5T1SE-Bloess wedge T2/F1″Dung0.8493 ± 0.0014−150.7 ± 1.41310 ± 1512901182123654
C-3635721111.5T1SE-Bloess wedge T2/F1″Dung0.7989 ± 0.0060−201.1 ± 6.01800 ± 70187915571718161
C-37e32071106T1SE-Bloess wedge T2/F1″Dung0.7388 ± 0.0012−261.2 ± 1.22430 ± 15267823582518160
C-38e32072106T1SE-Bloess wedge T2/F1″Dung0.7463 ± 0.0012−253.7 ± 1.22350 ± 1523602337234912
C-393572275T1SE-Bloess wedge T2/F1″Plant0.6967 ± 0.0039−303.3 ± 3.92905 ± 50321328853049164
C-403207355.5T1SE-Bloess wedge T2/F1″Wood0.6119 ± 0.0011−388.1 ± 1.13945 ± 15450842974402106
C-414124113T1SE-Cloess wedge T2/F1Dung0.9638 ± 0.0016−36.2 ± 1.6295 ± 1543030136565
C-424124226T1SE-Cloess wedge T2/F1Wood0.9759 ± 0.0018−24.1 ± 1.8195 ± 15290−5143147
C-4341243−23T1SE-Cin terrace (F1)Wood1.2007 ± 0.0027200.7 ± 2.7−1465 ± 20
C-4441244−24T1SE-Cin terrace (F1)Root0.9720 ± 0.0016−28.0 ± 1.6230 ± 1530415222876
C-4541245−25T1SE-Cin terrace (F1)Dung or Root1.0745 ± 0.001774.5 ± 1.7−570 ± 15
C-464124610T1SE-Cloess wedge T2/F1Root0.9789 ± 0.0064−21.1 ± 6.4170 ± 60302−3150153
C-473572610T2NE-Aloess wedge T3/T2Wood0.9805 ± 0.0016−19.5 ± 1.6160 ± 152834143140
C-48321001T2NE-Aloess wedge T3/T2Dung or Root0.9422 ± 0.0020−57.8 ± 2.0480 ± 2053550452016
C-4932075−22T2NE-Ain terrace (T2)Wood0.9781 ± 0.0015−21.9 ± 1.5180 ± 15285−3141144
C-503207842.5T2NE-Bloess wedge T3/T2Wood0.9813 ± 0.0015−18.7 ± 1.5150 ± 152836144139
C-5135752−15T2NE-Cin terrace (T2)Plant0.5500 ± 0.0017−450.0 ± 1.74805 ± 2555975475553661
C-5235727−16T2NE-Cin terrace (T2)Wood or Root0.7472 ± 0.0024−252.8 ± 2.42340 ± 3024602320239070
C-5332074−24T2SE-Ain terrace (T2)Wood0.4782 ± 0.0009−521.8 ± 0.95925 ± 2067916676673458

[22] A total of 53 accelerator mass spectrometry (AMS) measurements were made, with one replicate analysis (Table 1). Calibrated sample ages are plotted in Figure 7 as a function of stratigraphic position relative to the terrace treads. Samples collected from within the eolian silt and colluvial gravels postdate abandonment of the terrace deposits that they cap and are plotted in terms of absolute depth above the top of the tread (positive). Samples collected from within the terrace predate stream abandonment and are plotted in terms of absolute depth below the tread (negative). The samples are grouped into four populations based on the trench-tread relationship from which they were excavated: T2, the base of the T2/F1″ riser, the fault/F1″ interface, and T1. Samples with modern ages are italicized in Table 1 and are not plotted in Figure 7.

Figure 7.

Calibrated 14C ages as a function of stratigraphic position from trenches excavated into (a) T2, (b) F1″, (c) fault wedge, and (d) T1. Samples are plotted as a function of their depth relative to the top of the tread. We interpret ages from samples collected from within the postabandonment loess wedge and colluvial gravel deposits to place a minimum bound on the abandonment age of a terrace. In contrast, samples from within a terrace deposit provide a maximum bound on terrace abandonment age. In Figure 7a, T2 is limited to have been abandoned after 5.48–6.79 ka cal B.P.; in Figures 7b and 7c, F1″ is constrained to have been abandoned before 5.31–6.17 ka cal B.P.; and in Figure 7d, T1 is bracketed to have formed between 0.29 and 0.64 ka cal B.P. Samples are color coded on the basis of trench from which they come. Samples with a modern age signal not plotted.

[23] We dated seven samples from four trenches excavated into the T2 terrace deposit upstream and downstream of the fault, three of which were not modern (Figure 7a). The two oldest 14C samples from within T2 include sample C-51, a friable woody fragment that yields an age of 5.54 ± 0.06 ka calibrated (cal) B.P. and sample C-53, a less friable subcentimeter wood fragment that yields an age of 6.73 ± 0.06 ka cal B.P. Sample C-52 yields an age of 2.39 ± 0.07 ka cal B.P. In the field we noted that this latter sample extended deep into the trench wall and may have been connected with a root network. On the basis of the field observations, along with comparison to the TCN dating presented below, and the young age relative to samples for the lower and younger F1″ surface (see next paragraph), we believe that it is most likely that this sample is a root that grew into the T2 tread following T2 abandonment.

[24] We dated 19 samples from three trenches excavated into the base of the downstream T2/F1” riser face, with 13 samples from only two trenches yielding nonmodern ages (Figure 7b). Most samples were collected from the postabandonment wedge of loess onlapping the F1″ tread. The oldest sample was collected from within this wedge and yields an age of 4.40 ± 0.11 ka cal B.P. (C-40). Stratigraphically above this sample, 12 additional samples, all from within the silts overlying the F1” tread, exhibit a general trend of younging ages with increasing elevation above the F1″ tread. The remaining six samples from trench T1SE-C exhibit modern ages.

[25] We analyzed 11 samples from two trenches excavated into silt materials that onlap the F1″ surface at the base of the ATF mole track (Figure 7c). The oldest, deepest samples were collected at the base of the package of laminated silts, just above the top of F1″ tread. Sample C-22 was a friable chunk of wood and yields an age of 5.51 ± 0.20 ka cal B.P. Sample C-23 was a fragile and powdery subcentimeter block of charcoal and yields an age of 5.92 ± 0.26 ka cal B.P. Stratigraphically above these samples, nine additional organics collected from within the loess overlying the F1” tread exhibit a trend of younging ages with increasing elevation above the F1″ tread.

[26] We dated 13 samples from two trenches excavated into the upstream T1 surface and 2 samples from cutbank exposures, 6 of which yield nonmodern ages (Figure 7d). The oldest, deepest sample overlying the T1 tread, C-6, ranges in age from 0.54 to 0.64 ka cal B.P. and those samples from within the T1 terrace deposit range from 0.29 to 0.50 ka cal B.P. The remaining samples, both from within T1 and the overlying colluvial material, as well as the samples from the cutbank exposures, yield modern ages.

5. The 10Be Geochronology

5.1. Methods

[27] To further constrain the T2 abandonment age, we measured concentrations of in situ cosmogenic radionuclide 10Be [e.g., Gosse and Phillips, 2001; Lal, 1991] in nine amalgamated samples of quartz-rich gravel collected from two depth profiles excavated into the T2 terrace deposit and on opposite sides of the fault (Figure 3). We followed the subsurface sampling technique [Anderson et al., 1996; Hancock et al., 1999; Repka et al., 1997] to discriminate the inherited component from the postdepositional component. This latter component is the value used to determine the exposure age of a fluvial deposit. The auxiliary material provides additional information regarding sample preparation, exposure age calculations, and assumptions underlying the depth-profiling approach.

[28] From pit T2NE-D, which lies upstream of the mole track, we collected four aggregated samples at depths of 35, 75, 100, and 125 cm. The upper three samples were from within the intermediate gravel conglomerate, which we interpret to be the upper T2 terrace deposit (Figure 6a). The bottom sample was within the lower consolidated and weathered gravel deposit, which we interpret to be a lower and older package of the T2 deposit (Figure 6a). From pit T2SE-A, which lies downstream of the mole track, we collected five aggregated samples at depths of 25, 60, 90, 105 and 140 cm. The uppermost sample interval (25 cm) was within a disaggregated gravel deposit, which we interpret to be a reworked zone (Figure 6a). The 60 and 90 cm intervals were within the middle layer of gravel conglomerate, which we interpret to be the upper T2 deposit (Figure 6a). The 105 and 140 cm intervals were within a basal, well-consolidated fluvial gravel, which we interpret to be a lower and older portion of T2 (Figure 6a). For both trenches, we anticipated that the 10Be concentration results might be scattered because of our sampling across multiple stratigraphic intervals, although our hope was that the different layers were members of the same depositional sequence. We measured sample depths from the modern surface to the center of the sampled zone. These zones were all 6 cm thick, except for the 125 cm depth interval in T2NE-D, which was 10 cm thick because the small dimensions of the trench made it difficult for us to restrict the sampling to a 6 cm interval at the bottom of the pit. Each aggregate interval consisted of ≥ 32 clasts of quartz-rich gravels of granitoid and gneissic rock types, 2–15 cm in diameter.

5.2. The 10Be Results

[29] The data from the T2NE-D trench show a systematic decrease in 10Be concentration with increasing depth and permit two age calculations (Figure 8a and Table 2). The first yields an exposure age of 6.83 ± 1.13 ka and is based on a fit to all of the data. The second yields an age of 5.91 ± 0.66 ka and is based on a fit only to the three samples from the upper, unmodified portion of T2 (Figure 8a). This latter calculation is consistent with an interpretation that the upper ∼1 m thick gravel conglomerate from which the samples were excavated records the final stream-related deposit.

Figure 8.

Depth profiles showing the concentration of 10Be as a function of sample depth. (a) Results from trench T2NE-D. A fit through all of the data (blue) yields an age of 6.83 ± 1.13 ka, whereas a fit through only the upper three data points (green) yields an age of 5.91 ± 0.66 ka. The latter fit uses samples collected from within the stratigraphically highest and nonreworked alluvial deposit within the pit. (b) Results from trench T2SE-A show significant scatter, which we interpret to result from sampling across stratigraphic boundaries. No model age is calculated for this latter trench. Error envelopes were calculated using the asymptotic error in age based upon the quality of fit to the data points.

Table 2. TCN (10Be) Geochronologya
Dal-CNEF IDSample NameLatitudeb (°N)Longitudeb (°E)Altitudeb (km)Site Surface Production Ratec (atoms g−1 a−1)Depthd (cm)Interval Thickness (cm)Number of Clastse (n)Subsurface Production Ratef (atoms g−1 a−1)Quartz Mass (g)Mass Carrier Solutiong (g)10Be/9Be Corrected for Blank and Boronh10Be/9Be Error10Be Concentration (106 atoms/g SiO2)± Errori (106 atoms (g SiO2)−1)
  • a

    Results from the TCN 10Be AMS analyses conducted on amalgamated gravels collected from depth profiles. Depth profiles are plotted in Figure 8, and age reductions are discussed in the text.

  • b

    Latitude, longitude, and altitude determined from handheld GPS unit.

  • c

    Surface production rate calculated using CRONUS 10Be/26Al exposure age calculator, version 2 [Balco et al., 2008].

  • d

    Depth measured to the middle of stratigraphic interval from which clasts were collected.

  • e

    Clasts were amalgamated following the technique of Anderson et al. [1996], Repka et al. [1997], and Hancock et al. [1999]. From hand sample, quartz content of all samples estimated at >15%. Details of amalgamation are presented in the auxiliary material.

  • f

    The following values were used for the production rate calculation: Λ = 160 g cm−2; ρ = 2 g cm−3 for overlying deposit; ρ = 2 g cm−3 for clasts; λ = ln(2)/1,360,000. Surface geometry was assumed to be horizontal, and no corrections were made for snow cover, erosion, or horizon topographic obstructions (negligible).

  • g

    Carrier concentration 1015 μg mL−1 and carrier density 1.013 g mL−1.

  • h

    Blank correction made using Dal-CNEF 1933, with the exception of the correction applied to samples 2013 and 2014, which was made using blank Dal-CNEF 1934.

  • i

    Concentration error includes propagation of the uncertainties in the blank, carrier, and counting statistics.

2012G06ATFS8-T2SE-TrenchA-Q25 ± 337.7300086.725353.33347.522567735.67045.0370.25235.638E-131.335E-142115006555
2013G06ATFS8-T2SE-TrenchA-Q60 ± 337.7300086.725353.33347.526066822.69247.4930.23946.822E-131.662E-142302797258
2014G06ATFS8-T2SE-TrenchA-Q90 ± 337.7300086.725353.33347.529065615.48446.4810.23775.521E-131.599E-141890576656
2015G06ATFS8-T2SE-TrenchA-Q105 ± 337.7300086.725353.33347.5210566612.82245.0850.25313.872E-131.012E-141455744790
2016G06ATFS8-T2SE-TrenchA-Q140 ± 337.7300086.725353.33347.521406468.32745.0970.25635.844E-131.383E-142224206892
2017G06ATFS8-T2NE-TrenchD-Q35 ± 337.7308386.724133.35148.003563231.64545.1140.25686.781E-131.602E-142584598002
2018G06ATFS8-T2NE-TrenchD-Q75 ± 337.7308386.724133.35148.007563518.92045.0400.25194.924E-139.533E-151844225133
2019G06ATFS8-T2NE-TrenchD-Q100 ± 337.7308386.724133.35148.0010063713.78945.0610.24754.280E-131.018E-141574064893
2020G06ATFS8-T2NE-TrenchD-Q125 ± 537.7308386.724133.35148.00125104810.35645.0270.24712.915E-139.288E-151071274030
1933Blank for 200703105.318E-157.031E-16
1934Blank for 200704266.391E-151.072E-15

[30] In contrast to the northern trench, 10Be concentrations in trench T2SE-A do not systematically decrease with depth (Figure 8b and Table 2). Although we report these results for completeness, we do not use them to calculate an abandonment age. As explained above, such complexity in 10Be concentration is not unexpected because we sampled across stratigraphic boundaries observed within the trench (Figure 8b). Our hope in collecting and analyzing samples from this pit was that the different layers were members of the same depositional unit. Unfortunately, our results indicate that this is not the case and that the stratigraphic contact observed in trench T2SE-A represents a significant amount of time. Fortunately, as we discuss below, we have redundancy in the 14C and 10Be data sets, and thus can proceed with our analysis despite the failure of the samples from T2SE-A to yield a robust depth profile. In the auxiliary material we discuss possible reasons for the scatter in these data.

6. Discussion

6.1. Terrace Abandonment Ages

[31] In the following, we integrate the neotectonic mapping, topographic measurements, stratigraphic observations, and the 14C and 10Be geochronologic analyses to bracket terrace abandonment ages and evaluate cross-fault correlations. In this analysis, we interpret 14C samples collected from within the terrace deposit and below the tread to provide a maximum bound on the time of abandonment for that surface, whereas samples collected from within the capping silt deposits provide a minimum bound on the abandonment age. The 10Be depth profiles provide a measurement of the T2 surface abandonment age.

[32] Our data indicate that the T2 surfaces, upstream and downstream of the fault, are correlative and isochronous. This conclusion is supported by similar topographic textures between the surfaces, similar stratigraphic sequences in the trenches excavated on both sides of the fault, and compatible geochronologic results. In particular, 14C samples C-51 and C-53 from within the T2 terrace deposit on opposite sides of the fault (Figure 7) and the 10Be depth profile from trench T2NE-D on the north side of the fault (Figure 7) provide independent and overlapping ages. In interpreting an abandonment age for T2 from these data, we use the maximum range from the 14C samples and the depth profiles fit through the upper T2 deposit from trench T2NE-D. This conservative bracketing approach yields a T2 abandonment age ranging from 5.2 to 6.8 ka (6.0 ± 0.8 ka). Because the 14C data and the 10Be depth profile in trench T2NE-D are compatible, our data both control for cross-fault T2 diachroneity and test for potential bias in the geochronologic method.

[33] In contrast to the correlative T2 surfaces, our 14C ages indicate that the downstream F1″ and upstream T1 units are diachronous. This conclusion is also supported by the observation that the F1″ surface is blanketed by more loess than the upstream T1 surface. The surfaces also appear to have different primary surface textures, with higher relief bar-and-swale topography characterizing the F1″ surface than T1. First we consider the 14C data from F1″ (Figure 7b). Although our attempts to find ancient samples from within the F1″ deposit were unsuccessful, samples from within the onlapping eolian silt deposits at the base of the T2/F1″ riser and the base of the fault scarp provide a minimum constraint on the abandonment of the F1″ tread. The deepest sample from within the loess at the base of the riser yields an age of 4.4 ± 0.1 ka (Figure 7b). Likewise, at the base of the fault scarp, and at a deeper stratigraphic level than the sample collected from riser base, two samples collected from within the loess package <2 cm above the F1″ tread yield ages ranging from 5.3 to 6.2 ka (Figure 7c). We interpret these deepest samples to provide a minimum constraint on occupation of the F1″ surface because of the likelihood that stream activity would have removed the poorly consolidated silt deposits from which the samples were excavated. Thus, we conclude that the deepest samples from the base of the fault scarp provide a minimum abandonment age for F1″ that ranges from 5.3 to 6.2 ka (5.7 ± 0.4 ka) (Figure 7c).

[34] This interpretation assumes that the deepest 14C samples are not detrital. A significant detrital signal is unlikely for four reasons. First, the samples were friable and fragile, making it unlikely that they could have survived transport and reworking prior to deposition. Second, we observe a trend of increasing age with increasing depth in the trenches at the base of the fault scarp and base of the riser, which is consistent with a process of progressive loess deposition and burial of in situ (nontransported) organic materials. Third, the deep samples collected at the base of the fault were from within a package of finely laminated silts, which we interpret to be eolian deposits. Incorporation of reworked woody fragments into this loess package is improbable. Finally, we have not seen evidence of inheritance at a site with similar geomorphology and deposits but which also yielded 14C ages from within both terrace deposits and capping silt [Cowgill et al., 2009].

[35] In contrast to F1″, the 14C samples from T1 exhibit ages less than 0.6 ka, suggesting T1 was abandoned more recently than F1” (Figure 7d). In detail, the nonmodern sample overlying the T1 tread ranges in age from 0.5 to 0.6 ka and five samples from within the T1 terrace deposit range from 0.3 to 0.5 ka. While the age distributions for the nonmodern samples from within the postabandonment deposits and terrace deposits are stratigraphically inverted from what is expected, the age difference is small (∼0.1 ka). In addition, the samples collected from within the T1 terrace and within the postabandonment materials came from trenches spaced ∼100 m apart, thus the apparent age inversion may also result from local variability in the timing of formation and abandonment of different patches of T1. We conclude that the T1 surface was likely abandoned sometime between 0.3 and 0.6 ka (0.5 ± 0.2 ka) (Figure 7d). Although the abandonment age of T1 is not as well constrained as for the T2 and F1″ surfaces, we note that this uncertainty does not impact the geologically relevant slip rates we compute for this site, below. Furthermore, topographic and morphologic analyses discussed below support our conclusion that T1 is younger than F1″.

[36] Importantly, these terrace ages bracket the age of the intervening T2/F1″ and T2/T1 riser segments. In particular, the downstream T2/F1″ riser is bracketed by the T2 and F1″ surface abandonment ages to have formed between 6.0 ± 0.8 ka to 5.7 ± 0.4 ka. In contrast, the upstream T2/T1 riser is much less tightly bracketed by the T2 and T1 formation ages, and could have been modified by lateral stream erosion anytime between 6.0 ± 0.8 ka and 0.5 ± 0.2 ka. If we had only bracketed the riser on the downstream side of the fault, we would not have found this important source of uncertainty. These diachronous bracketing ages for the riser segments must be considered when developing a millennial slip rate measurement for this site.

6.2. Riser Offsets

[37] The range of possible T2/F1″ riser offsets can also be bracketed using two different measurements. The first correlates the upstream T2/T1 and downstream T2/F1″ risers, which show an apparent left-lateral offset of 54 ± 3 m (Figure 3b). This measurement assumes that the riser initially cut straight across the fault, which is supported by the similarity in trend of the cross-fault riser segments. However, if lateral erosion of the T2/T1 riser was significant, then the actual offset of the T2/F1″ rise could have been up to a factor of ∼1.7 larger. This second measurement correlates the downstream T2/F1″ riser with the upstream western valley wall and results in a displacement of 89 ± 7 m (Figure 3b). In this case, the observed 54 ± 3 m offset would underestimate the true displacement since cessation of lateral erosion at the base of the T2/F1″ riser. The valley wall on the west side of the modern channel and upstream from the fault provides a maximum constraint on the magnitude of this potential erosion, which is as large as 38 ± 6 m. We determined this value by measuring the width of the valley at the elevation of T1, in the region <30 m north of the fault (Figure 3b). For this second riser offset measurement, the contact between the western valley wall and the paleo-F1″ surface would define the trend line used to determine the upstream piercing point. Since F1″ is not preserved upstream of the fault, we use T1 as a proxy for this surface (Figure 3b). T1 is a sufficient substitute because the surface slopes up to the west (Figure 5) and thus approaches the paleo-F1″ position. Furthermore, the surface intersects with the steeply sloping western valley wall, which results in a small, <1 m shift in horizontal position over a 1–2 m vertical distance. Significantly greater uncertainty is introduced in our consideration of projections, described below. The downstream trend line trajectory is constrained using the trend of the T2/F1″ riser basal contact with F1″. But because the deepest and oldest 14C ages measured in the postabandonment loess onlapping F1″ were collected from the FltWdgSE-A trench located at the base of the fault scarp, we conclude that the stream could not have occupied the F1″ surface east of this trench position because stream activity would have removed the poorly consolidated silt deposits from which the samples were excavated. Thus, the western margin of this trench is used to constrain the downstream trend line position (Figure 3b).

[38] We constrain the uncertainty in the displacement measurements by considering far-field (<120 m) and near-field (<20 m) projections of the riser segments closest to the fault (Figure 3b). In considering the far-field (<120 m) riser trends, the length of preserved riser segments on opposite sides of the fault are not the same. Our approach is to maximize the far-field trend measurement by using the longest segment of riser possible, which results in up to 38 m differences in riser trend lengths. In considering the near-field (<20 m) variations in riser trend, we again select the maximum length of riser that captures the riser trajectory within the <20 m long zone where the feature truncates against the fault, which again results in slight differences in trend line lengths. Uncertainties are determined from half the range between the minimum and maximum displacement values. The projections that we use generate generous uncertainties, as high as ± 7 m for end-member displacement measurements. Furthermore, this source of uncertainty is significantly smaller than the uncertainty that arises from considering which reconstruction best approximates the site history.

6.3. Analysis of Riser Topography

[39] Further evidence that the T2/F1″ and T2/T1 risers are diachronous comes from the topographic swath profiles extracted from the T-lidar point cloud data. The profiles provide a quantitative means to compare the upstream and downstream riser segments and yield three important relationships (Figure 5 and the auxiliary material).

[40] First, the riser slopes are different. In the zone ±5 m horizontally from the riser midpoint, the upstream T2/T1 riser face slopes an average of 39.8 ± 6.1°, whereas the downstream T2/F1″ riser face slopes 21.4 ± 1.3° (Figure 5). Assuming that the risers have experienced the same slope processes, this >18° difference in escarpment steepness suggests that the upstream T2/T1 riser was laterally refreshed by the stream more recently than the downstream T2/F1″ riser [e.g., Cowgill, 2007; Harkins and Kirby, 2008; Nash, 1984].

[41] Second, the riser profiles are asymmetric. For example, the T2/F1″ riser slopes more steeply above the riser midpoint than below (Figure 5). This observation is consistent with the stratigraphy seen in our excavations in suggesting that material at the base of riser face includes not only colluvium shed from the T2/F1″ riser crest but also wind-transported silt. Although the T2/T1 riser face is also asymmetric, the sense is opposite to that of the T2/F1″ riser in that the riser slopes more gently above the riser midpoint than below (Figure 5). We interpret the steeper slope at the base of the T2/T1 riser to reflect incomplete riser refreshment by lateral stream erosion prior to deposition of T1. In this interpretation, the crest of the T2/T1 riser slopes more gently because it escaped the latest phase of lateral erosion experienced by the base, allowing the crest to remain stable for enough time to diffuse to its current geometry.

[42] Third, the riser heights appear to be different. In particular, the T2/T1 riser is 7.5 ± 0.2 m high, whereas the T2/F1″ riser is 4.7 ± 0.4 m at roughly similar distances from the fault (Figure 5). This vertical difference may indicate that the stream incised ∼2.8 m following deposition of F1″ and before depositing T1. However, this relationship is potentially misleading because it assumes the terrace slopes are parallel above and below the riser. The topographic contours in Figure 3b indicate that the upstream T2 and T1 surfaces, as well as the downstream F1″ surface all slope in the same direction and are roughly parallel. In contrast, the downstream T2 surface slopes more southeasterly. Thus, the riser height difference may be due in part to nonparallelism of terraces.

6.4. Reconstructions, Slip Rates, and Reducing Uncertainty

[43] The diachronous T2/F1″ and T2/T1 riser segments lead us to consider three possible reconstructions. As we show in the following, while traditional lower terrace (Figure 9) and upper terrace (Figure 10) reconstructions would yield rates for this site that vary by a factor of 13.0, we reduce this uncertainty by pairing the bracketed downstream riser age with conservative constraints on the magnitude of lateral erosion of the upstream T2/T1 riser (Figure 11). This treatment limits the average mid-Holocene slip rate on this reach of the ATF to have ranged from 9.0 ± 1.3 mm a−1 to 15.5 ± 1.7 mm a−1, reducing the epistemic uncertainty to a factor of 1.7.

Figure 9.

Traditional lower surface model, in which lateral erosion of the lower T1 tread removes riser offset until the T1 surface is abandoned. This model is geologically unreasonable for reasons described in the text. Roman numerals i–iii highlight critical relationships. Map symbols and colors are the same as in Figure 3b.

Figure 10.

Minimum slip rate reconstruction for Tuzidun featuring no lateral erosion of the upstream T2/T1 riser following formation of the downstream T2/F1″ riser. (a) T2/F1″ riser forms at 6.0 ± 0.8 ka. (b) Deposition of the F1′ lobe at ∼3 ka. (c) Deposition of T1 up to 0.5 ± 0.2 ka. (d) Incision of T1 since 0.5 ± 0.2 ka to form the modern channel, T0. This reconstruction yields a minimum slip rate of 9.0 ± 1.3 mm a−1 (i.e., 54 ± 3 m since T2 abandonment at 6.0 ± 0.8 ka). Roman numerals i–v highlight critical relationships. Map symbols and colors are the same as in Figure 3b.

Figure 11.

Maximum slip rate reconstruction for Tuzidun featuring lateral erosion of the upstream T2/T1 riser following formation of the downstream T2/F1″ riser. (a) T2/F1″ riser forms at 5.7 ± 0.4 ka. No further modification of this riser occurs east of trench FltWdgSE-A (star), as constrained by radiocarbon samples collected from the base of the postabandonment loess wedge. (b) Deposition of the F1′ lobe at ∼2.5 ka as the fault transports the downstream and sheltered T2/F1″ riser out of the path of the active stream channel. At the same time the Qfo shutter ridge is transported into the path of the stream, driving lateral erosion of the upstream T2/F1undiff riser. (c) T1 is deposited at 0.5 ± 0.2 ka and the upstream T2/T1 riser experiences no further modification. (d) T1 is incised by the modern stream channel, T0. This reconstruction yields a maximum slip rate of 15.5 ± 1.7 mm a−1 (i.e., 89 ± 7 m since F1″ deposition at 5.7 ± 0.4 ka). Roman numerals i–xi highlight critical relationships. Map symbols and colors are the same as in Figure 3b.

[44] A traditional lower–terrace reconstruction of the Tuzidun site is geologically unreasonable (Figure 9). This model pairs the observed displacement (54 ± 3 m) with the age of T1 (0.5 ± 0.2 ka), to yield a geologically unreasonable slip rate of 116.6 ± 44.6 mm a−1. In particular, according to this reconstruction, T2 is incised at 6.0 ± 0.8 ka, followed by subsequent deposition of the F1 lobes and lateral erosion of the riser north of the fault. As a result, no riser offset is recorded until T1 is abandoned at 0.5 ± 0.2 ka, after which the riser records 54 ± 3 m of displacement. This reconstruction is not geologically viable for several reasons. First, it yields a rate which is implausibly high, approximately three times the 31–42 mm a−1 convergence rate of Indo-Asia measured by GPS [e.g., Shen et al., 2000; Thatcher, 2007]. Second, prior to T1 abandonment, the model predicts complete refreshment of both the upstream and downstream riser segments. This prediction is inconsistent with the presence of the loess wedge at the base of the fault/F1″ interface and also at the base of the T2/F1″ riser, which 14C samples constrain to have been emplaced by 5.7 ± 0.4 ka. Third, the reconstruction predicts that the apex of the downstream F1 fan complex (deposited prior to T1 emplacement) was deposited in a geometry in which it did not align with the channel upstream (Figure 9). Fourth, the reconstruction violates the inset channel width criteria, described by Cowgill [2007]. More specifically, the inset channel incised into T1 is only 4.4 m wide at the intersection with the ATF. Lateral erosion that fully refreshed the riser would lead to a stream valley with a width greater than or equal to the observed 54 m offset. The only way that this scenario can be satisfied is if the stream channel was originally inset with a ≥50 m long, fault-parallel bend, for which there is no geologic evidence, nor is this scenario geometrically reasonable, nor is it consistent with any of the present-day observed channel geometries found along strike.

[45] A traditional upper terrace reconstruction assumes no lateral erosion of the upstream T2/T1 riser or the downstream T2/F1″ riser following abandonment of the T2 surface and yields a geologically reasonable result (Figure 10). In this model, the T2 surface is abandoned at 6.0 ± 0.8 ka (Figure 10a). Subsequent to this incision, fault slip offsets the T2/F1″ riser from the riser upstream, which we refer to as T2/F1undiff because the stream both occupies an undifferentiated F1 surface (F1undiff) on the upstream side of the fault and deposits the F1 fan lobes on the south side of the fault during this interval (Figure 10b). By 0.5 ± 0.2 ka, the channel has incised and eroded the upstream F1undiff deposit and then emplaced the T1 surface (Figure 10c). At 0.5 ± 0.2 ka the stream abandons T1 by incision to form the modern channel (Figure 10d). In this end-member reconstruction, the observed offset between the T2/F1″ and T2/T1 riser segments is unmodified by lateral erosion following T2 abandonment. Thus, the slip rate for this model is calculated by combining the T2 abandonment age of 6.0 ± 0.8 ka (maximum age) with the observed displacement of 54 ± 3 m (minimum displacement) to yield a minimum slip rate of 9.0 ± 1.3 mm a−1 since 6.0 ± 0.8 ka.

[46] To obtain a reasonable maximum bound on the slip rate, we develop an end-member model that incorporates lateral erosion of the upstream T2/T1 riser following formation of the downstream T2/F1″ riser (Figure 11). This scenario begins with the abandonment of the T2 surface at 6.0 ± 0.8 ka, followed by stream incision to form the T2/F1″ riser south of the fault. Emplacement of the F1″ fan lobe occurs before 5.7 ± 0.4 ka and provides a minimum bound on the final period of lateral erosion that could have modified the T2/F1″ riser south of the fault. As we discuss in the analysis of riser offsets above, the stream has not occupied F1” east of the location of trench FltWdgSE-A site since 5.5 to 5.9 ka because radiocarbon samples with ages of 5.51 ± 0.20 and 5.92 ± 0.26 ka cal B.P. were preserved at this site within easily eroded loess capping the F1″ surface at the base of the ATF mole track. Thus, the downstream piercing point is the western margin of trench FltWdgSE-A (Figure 11a). Subsequent fault slip transports the T2/F1″ riser away from the modern channel, protecting it from further modification. Concurrently, the southwestern, downstream block of Qfo is shuttered in front of the active channel (predeposition of T1), which pushes the channel east and drives lateral erosion of the upstream T2/F1undiff riser (Figure 11b). During this interval and simultaneous with lateral erosion of the T2/F1undiff riser, the stream occupies the upstream F1undiff surface, and the F1′ and F1 lobes of the downstream fan complex are deposited. Furthermore, the correlative to the upstream western valley wall/T1 contact, the modern downstream shutter ridge/T0 contact, is likely modified by stream erosion. The western valley wall limits the magnitude of erosion to 38 ± 6 m, prior to deposition of the T1 surface at 0.5 ± 0.2 ka (Figure 11c). We speculate that the emplacement of T1 may be due to an ancient large earthquake with 6–7 m of slip (the current value required for the shutter ridge to close the modern channel) during which time the shutter ridge could have dammed the channel. This model is consistent with a paleoseismic trenching study at Kelusayi, 290 km to the east-northeast, which reported that the most recent event occurred at 0.5 ± 0.3 ka with an associated magnitude of slip of 5.5 m [Washburn et al., 2001], though we might expect a finer T1 deposit if this speculative model is correct. Following incision of T1 and creation of the modern channel, T0, the upstream T2/T1 riser experiences no further lateral erosion. In this end-member model, 89 ± 7 m of displacement (maximum offset value) accumulates following the abandonment of F1″ at 5.7 ± 0.4 ka (minimum riser age) (Figure 11d). Pairing these displacement and age values yields a maximum average slip rate of 15.5 ± 1.7 mm a−1 since 5.7 ± 0.4 ka. This maximum slip rate constraint is conservative because it uses an offset value (38 ± 6 m) which assumes the largest possible magnitude of lateral erosion. This value likely overestimates the magnitude of lateral erosion because, as explained above, the asymmetry of the T2/T1 riser north of the fault indicates the riser was abandonment long enough before T1 was incised to allow the riser crest to diffuse. Thus, the maximum bound on slip rate we report here is likely too high.

7. Implications

[47] The integrated structural and geomorphic reconstructions we developed for Tuzidun highlight the possibility that while an offset riser may appear to have once been continuous, the offset segments may have formed diachronously because of differences in the timing of cessation of lateral stream refreshment (Figure 12). If this diachroneity goes undetected, erroneous slip rate measurements are likely to result. The most conservative solution to this problem is to bracket the age of offset riser segments using the upper tread abandonment age at the crest of the sheltered riser segment and the lower surface tread abandonment age at the base of the unsheltered riser segment. However, this approach can lead to large ranges in slip rates. These ranges can be reduced by constraining the magnitude of lateral erosion.

Figure 12.

Conceptual model illustrating evolution of a site with diachronous risers. At time 1, a riser is formed while the stream occupies the lower tread. At time 2, lateral slip along the fault transports the downstream riser out of the path of the stream, and this sheltered riser segment is protected from future lateral erosion. In contrast, the upstream unsheltered riser segment continues to be exposed to lateral refreshment by the stream. At time 3, continued fault slip transports the sheltered riser segment away from the stream and this riser decays. The unsheltered riser segment continues to experience lateral refreshment by the stream. At time 4, modern stream channel incises into the lower tread. The offset risers have different ages and the observed displacement does not correspond to either riser age. The abandonment ages from the sheltered upper tread and the unsheltered lower tread provide respective minimum and maximum bounds on slip rate. If the resulting epistemic uncertainty is too high, additional criteria such as a narrow upstream valley or surface markers on the terrace surfaces can be used to constrain the magnitude of lateral erosion.

[48] This framework builds on a >40 year long dialog concerning the best method for dating offset terrace risers. Originally, the abandonment age of a single terrace surface, usually at the base of a riser, was thought to be sufficient to constrain the age of a riser [e.g., Berryman, 1990; Bull, 1991; Grapes, 1993; Knuepfer, 1987; Mériaux et al., 2004, 2005; Van der Woerd et al., 1998, 2002]. More recent investigations have highlighted the uncertainty in the single-terrace dating approach, shifting attention to the need to bracket the age of a riser by determining abandonment ages for both the upper and lower treads flanking the riser [Cowgill, 2007; Harkins and Kirby, 2008; Mériaux et al., 2005].

[49] Most recently, Harkins and Kirby [2008] shifted the focus of these investigations to dating offset riser segments. In their investigation of the Deng Qin site along the Kunlun Fault, they bracket the age of faulted fluvial risers by combining tread abandonment ages determined from 14C geochronology with models for riser degradation assuming diffusive slope processes. Importantly, this analysis is applied only to the downstream, sheltered riser segment and the magnitude of upstream riser erosion is unconstrained. Thus, while the treatment developed by Harkins and Kirby [2008] improves constraints on riser age, it does not explicitly account for the uncertainty in displacement magnitude that is introduced by lateral erosion. Below, we consider the implications for riser diachroneity, including a discussion of (1) tests for across-fault riser diachroneity, (2) the most conservative methods to evaluate epistemic uncertainty associated with diachroneity, and (3) observations that can reduce this uncertainty.

[50] Across-fault diachroneity is expressed through morphological differences between the offset riser segments (Figure 12). At Tuzidun, the upstream T2/T1 riser face is steeper, has an opposite sense of asymmetry, is flanked by a younger lower terrace, and is taller than the downstream T2/F1″ riser. These observations are consistent with the interpretation that the upstream riser was more recently refreshed than the downstream riser. Harkins and Kirby [2008] similarly report steeper slopes on the unsheltered riser segments and more shallow slopes on the sheltered riser segments, including riser modification by groundwater sapping, similar to that seen at Tuzidun along the base of the T2/T1 riser. In addition to riser slopes and heights, a comparison of abandonment ages from across-fault terrace surfaces can also be used to test for riser diachroneity. For example, at Tuzidun the downstream F1″ surface was abandoned >5 ka before the upstream T1 surface, indicating that the lower T1 tread has been occupied more recently by the active channel, thus potentially exposing the unsheltered T2/T1 riser to lateral erosion.

[51] Uncertainty in riser age was originally thought to be bracketed through a traditional upper surface and lower surface model treatment, in which upper and lower surface ages provide minimum and maximum bounds on the age of an offset [Cowgill, 2007]. The Tuzidun investigation further specifies that abandonment ages from the upper surface at the crest of a sheltered riser and the lower surface at the base of an unsheltered riser provide the most conservative bounds on the age of an offset riser (Figure 12). At Tuzidun, these surfaces correspond to the upstream/lower T1 tread and the downstream/upper T2 tread. As a point of caution, if the wrong surface ages are paired with a riser, then inaccurate bounds on slip rate will result. For example, if at Tuzidun we had extrapolated the downstream F1″ age to the upstream T1 surface, an artificially low upper bound on slip rate of 9.4 ± 0.9 mm a−1 would have been calculated.

[52] One disadvantage of the conservative dating approach, described above, is that it may lead to overly poor constraints on slip rates. For example, at the Tuzidun site this approach brackets the slip rate along the central ATF to have ranged between 9 and 117 mm a−1. Additional observations at Tuzidun provide the foundation for reducing the uncertainty of this measurement. In detail, the well-bracketed age of the downstream T2/F1″ riser and the relatively narrow upstream valley provide additional constraints on the evolution of the site and reduce the uncertainty from more than an order of magnitude to a factor of 1.7. Thus topographic limits to the magnitude of erosion can provide means for evaluating lateral erosion of risers. Similarly, secondary features such as stream channels inset onto the terrace surfaces [e.g., Cowgill, 2007] or comparisons of fault-parallel profiles of faulted alluvial fans [e.g., Meyer et al., 1996] can be compared with observed riser displacements to constrain lateral erosion.

[53] In summary, this study illustrates the importance of comparing across-fault riser segments to evaluate potential diachroneity. Even at sites exhibiting riser diachroneity, the abandonment ages of the upper sheltered tread and the lower unsheltered tread bracket slip rates (Figure 12). If this conservative approach results in large uncertainty, additional geochronologic (e.g., bracket age of one riser segment) and geomorphic (e.g., topographic limits to lateral erosion) observations may limit the error bounds. Sites like Tuzidun with shutter ridges may be particularly prone to high magnitudes of lateral erosion. Likewise, sites like Tuzidun with narrow upstream valleys offer constraints on the magnitude of potential lateral erosion. The methods employed at the Tuzidun site may not be necessary at every slip rate site, especially those without shutter ridges. We think it is likely that many previous studies are sound, even though they did not use the method of analysis we develop here. Most generally, this study highlights the need to examine each offset riser site individually and to develop reconstructions that synthesize geochronologic and morphologic measurements as well as the geometry and distribution of geomorphic features at a site.

8. Conclusions

[54] Slip rate studies commonly use laterally offset fluvial terrace risers to reconstruct strike-slip fault histories to address the patterns and processes of temporal and spatial slip rate variations. To ensure that slip rate measurements made from such geomorphic markers are sufficiently precise to resolve fault slip histories, it is necessary to quantify the uncertainty associated with these measurements. Our new data from Tuzidun highlight a source of error related to these types of studies that has not previously been dealt with systematically, namely the potential for diachroneity in offset fluvial risers. Across-fault diachroneity may result from differential timing of cessation of lateral stream erosion. In this case, the observed riser displacement may not correlate with the age of either riser segment. If this diachroneity goes undetected, erroneous slip rate measurements are likely to result. Results from Tuzidun suggests that the most conservative slip rate bounds are determined by pairing the observed displacement with abandonment ages from the upper surface at the crest of the sheltered riser segment and the lower surface at the base of the unsheltered riser segment. At Tuzidun, the uncertainty in this bracketing approach is reduced by bracketing the age of the downstream T2/F1″ riser segment and considering upstream piercing points that account for the possibility of lateral erosion.

[55] A regional implication for this investigation centers on a new slip rate measurement for the central Altyn Tagh Fault. The new bound on millennial slip rate from Tuzidun is more consistent with the lower estimates for slip rate along the central Altyn Tagh Fault [e.g., Cowgill, 2007; Wallace et al., 2004; Washburn et al., 2003]. In detail, two bracketing millennial slip rates are calculated: (1) a minimum rate of 9.0 ± 1.3 mm a−1 since 6.0 ± 0.8 ka from combining the upper surface T2 abandonment age with the 54 ± 3 m observed offset and (2) a maximum rate of 15.5 ± 1.7 mm a−1 since 5.7 ± 0.4 ka from combining the sheltered lower surface F1″ abandonment age with 89 ± 7 m of offset, which accounts for 38 ± 6 m of lateral erosion. Because the late Quaternary slip rate along the Altyn Tagh Fault is a highly contentious topic [e.g., Cowgill, 2007; England and Molnar, 2005; Mériaux et al., 2004; Tapponnier et al., 2001; Thatcher, 2007; Wallace et al., 2004], a discussion of the tectonic implications of the new slip rate measurement at Tuzidun is beyond the scope of the current study. However, ongoing work will synthesize results from the present study with both new data from adjacent sites and previously published results to address this question by developing a late Quaternary slip history for the central Altyn Tagh Fault.


[56] This work was supported by NSF grants EAR-0610107 and EAR-0610040 from the Tectonics Program and the East Asia and Pacific Program in the NSF Office of International Science and Engineering, along with funding from the Geological Society of America, University of California, Davis, and the Northern California Association of Phi Beta Kappa. The initial idea of the potential for diachronous risers was first raised in a conversation with Rasmus Thiede, and we thank him for his insight. The manuscript benefited from constructive reviews by Kurt Frankel and Carol Prentice and thoughtful discussions with Robert Anderson, Kari Cooper, and Anke Friedrich. We are grateful to Peter Gold, Greg Chavdarian, Megan Muretta, Christina Davis, Tang Wei, Gong Hongliang, and Jiang Rong Bao, who helped with field work and field logistics. John Southon, Guaciara dos Santos, Howard Spero, and Dave Winter provided support with 14C sample preparations and AMS measurement. Guang Yang provided assistance with the TCN sample preparation and Bob Finkel provided the TCN AMS analysis.