Geomorphic analysis

Fault mapping and geomorphic analysis along the northern segment of the GCF were carried out using various types of data, including field observations and topographic data (e.g., real-time kinematic measurements (RTK)) and satellite images (GeoEye images from Google Earth; resolution ~0.5 m). We mainly focused on the abandoned western edges of alluvial fans and preserved river terrace risers as our landform markers, and then used the projected extension of these markers on to the GCF to measure the offsets (Huang, 1993; van der Woerd, et al., 2006; Zhang et al., 2007, 2008; Gold et al., 2011; Chen et al., 2012, 2013). Given that the horizontal uncertainties for RTK data and GeoEye images are <0.5 m, measurement errors on the displacements mainly arise from the determination of markers and construction of fitting lines (e.g., Cowgill, 2007; Zielke et al., 2010; Chen et al., 2013).

In previous studies, landforms having linear structures such as clear fringe-like terrace risers and fluvial channels have been preferred in calculating the displacement on strike-slip faults (e.g., Cowgill, 2007). In arid regions, alluvial fans can also be considered to have a predictable geometry and to be useful geomorphic markers (e.g., Brown et al., 2002; van der Woerd et al., 2006). The clearly visible contrasts in surface color and roughening of different fan surfaces enables their identification on satellite images, field observations and relief maps (Burbank and Anderson, 2011). Given that alluvial fans are fan- or cone-shaped deposits crossed and constructed by streams, fan cross-sections are convex-shaped. Fan flanks are often linear negative topographic features (e.g., seasonal rills). The fringes of alluvial fans generally connect to low-lying terrains (e.g., lakes or rivers). The deposits of the alluvial fans are typically poorly sorted, angular gravels, and may even have boulders scattered on their surfaces. The sediment characteristics of the fans commonly contrast with the materials outside of the fan area. Therefore, color differences, linear negative topographic features, and sediment characteristics on the fan surface enable the location of the edge of an alluvial fan to be identified in high-resolution satellite images overlain with high-resolution relief maps, and in the field.

OSL dating

Our sampling strategy was primarily designed based on two considerations. Firstly, we focused on sampling sand lenses, in the case that such lenses could not be found we sampled alluvial gravel layers with a high sand content. Secondly, to ensure that the OSL samples represent the ages of faulted landforms, we mainly sampled the uplifted side of the GCF and the abandoned western edges of alluvial fans and/or preserved river terrace risers. This approach reduces the possibility that later erosion may influence our results and interpretations.

OSL dating was used to study the northern segment of the GCF at the Aerqingsang and Quba sites (Fig. 1). Six samples were collected from the alluvial sediments. To avoid potential exposure to daylight whilst sampling, sediments for equivalent dose (D_e) determination were only taken from the middle part of each steel tube under subdued red light conditions, whereas sediments at both ends of the tube were used for radioisotopic content (U, Th and K) measurements.

Quartz grains were extracted following Lu et al. (2007) and Aitken (1998). Samples of ~100 g were first treated with 30% H₂O₂ and 30% HCl to remove organic matter and carbonates, respectively. The samples were then washed with distilled water until it became neutral.

The 4–11-μm-sized polyminerallic grains were separated by settling following Stokes’ law and were then soaked in 30% hexafluorosilicic acid (H₂SiF₆) for 3–5 days in an ultrasonic bath to extract fine-grained quartz. Purified quartz grains were transferred to a stainless steel disc (9.7 mm in diameter) for OSL measurements.

Dry sieving was used to separate the 90–125 µm fraction, which was further separated by heavy liquid fractionation (liquid density between 2.62 and 2.75 g/cm³) to obtain quartz grains. After drying, the quartz grains were treated with 40% HF for 45 min to remove the outer layer irradiated by alpha particles and any remaining feldspar-grain. The grains were scattered onto a stainless steel disc (9.7 mm in diameter) for experiments. The purity of the isolated quartz was checked by infrared (IR) stimulation. All the measurements were carried out using an automated Daybreak 2200 OSL reader equipped with infrared (880 ± 60 nm) and blue (470 ± 5 nm) LED units, and a ⁹⁰Sr/⁹⁰Y beta irradiation source. The quartz OSL signal was stimulated at 125°C with blue LEDs and detected with an EMI 9235QA photomultiplier tube coupled with two U-340 glass filters.

For D_e determinations, the sensitivity-corrected multiple aliquot regenerative-dose (MAR) protocol was used to determine fine-gained quartz OSL ages for two samples (KCSE-OSL-3 and OSL-111) (Lu et al., 2007), and the other samples were analyzed using the single-aliquot regenerative dose protocol (SAR) for OSL dating (Murray and Wintle, 2000). All OSL ages are listed in Table 1.

Arerqingsang site

The Aerqingsang site is located south of Kong Co, near the village of Aerqingsang (Fig. 1C). At this site, the width of the Quaternary basin is only 2–4 km and strike-slip fault activity is mostly localized along a single strand. However, we located a series of secondary normal faults developed in the east of the basin, which form alinear depression with a length of ~3 km (Fig. 2). The major fault strand cuts across early Q3 and Holocene alluvial fans in their middle and fringe regions, forming linear fault scarps. Two early Q3 fans (Fan1 and Fan2) have been consistently right laterally displaced by hundreds of meters. To the east of Fan1 and west of Fan2, sag ponds occur along the northern side of the fault strand (Figs. 2 and 3). The primary sedimentary provenance of Fan1 is from the southern mountains, and aggradation has ceased (Figs. 2 and 3A). The linear scarps are very clear and are 2–5 m high. Several seasonal rills (30–400 m long) on the surface of Fan1 are displaced by ~3–40 m (Fig. 3A). Three rills are consistently offset by 14–18 m (Fig. 3B). Fan2 is still undergoing aggradation and has a sedimentary provenance from the northern piedmont, east of the Aerqingsang site (Figs. 2 and 3C). From east to west, the fault strand cuts the fringe of Fan2 and forms scarps with heights of 2–20 m. The landforms are better preserved on the southern uplifted side, and are locally eroded and reworked on the western fringes of these alluvial fans (Fig. 3C).

The surface color of Fan1 and Fan2 is dark brown and contrasts with the terrain beyond the fans (Fig. 2A). Furthermore, the fringes of these two fans connect to streams. Thus, it is straightforward to identify the edges and areas of Fan1 and Fan2 (Fig. 2B). The right-lateral movement of the fault has displaced the fringes of these two fans, forming prominent shutter ridges on the southern side of the fault. The streams are dammed by the shutter ridges at the western fringes of Fan1 and Fan2 (Figs. 2A, 3A, and 3C). By retrofitting the western fringes of the alluvial fans and the dammed streams on both sides of the GCF using remote sensing images (GeoEye images from Google Earth), we identified a 500 m displacement for both Fan1 and Fan2 (Fig. 2). We estimate the error on this displacement to be 20%, and so obtain an offset of 500 ± 100 m (Fig. 2C), which is broadly consistent with previous interpretations (375 m by He et al., 2002; 400 m by Yang et al., 2012).

We located a ~3 m deep feature cut by a large gully on the southern uplifted side of the fault, in the eastern part of Fan1. Two OSL samples were selected from two sets of fine sandy lenses (20–40 cm thick). The results suggest that the upper and lower lens formed at 60.87 ± 4.30 and 109.15 ± 7.62 ka, respectively (Table 1). These ages may indicate Fan1 was deposited between ~61 and 109 ka (Fig. 3D). Fan1 and Fan2 consist of the same brown–red, sub-angular gravels in a matrix of silty sand (Fig. 3), and they have similar dark brown surfaces in GeoEye images (Fig. 2). Therefore, the two fans were probably deposited during the same period (i.e., early Q3 fans). Given the extensive age range of the fans, it is important to choose an appropriate initiation time for faulting in order to constrain the slip rate. This is a major factor in estimating the slip rate, and has been considered in many previous studies (e.g., van der Woerd et al., 2006; Cowgill, 2007; Zhang et al., 2007, 2008; Gold et al., 2011; Chen et al., 2012, 2013; Gong et al., 2015). If right-lateral slip has occurred since the alluvial fan ceased aggrading, then the initiation time of faulting will be younger than ~61 ka and the average slip rate will be >8.2 ± 1.7 mm/yr. However, if right-lateral slip commencement coincided with the onset of alluvial fan aggradation, then the initiation time is ~109 ka and the average slip rate is 4.6 ± 1.0 mm/yr. We note that the aggradation capacity is limited at the fringes of the two fans where the fault strand crops out, implying that the western fringes of these two fans have been protected by uplift and right-lateral movement. Therefore, we consider that displacements has been occurring since the initial aggradation of Fan1 and Fan2. As such, the two early Q3 fans have been displaced by 500 ± 100 m since ~109 ka at an average slip rate of 4.6 ± 1.0 mm/yr.

Quba site

The Quba site is located south of Zhangnai Co where a major strand of the GCF strikes along the front of the northern mountains. The height difference of features at this site exceeds 1000 m. The piedmont is occupied mainly by early and late Q3 alluvial fans. A river of ~10 km in length originates in the northern mountains and flows southeast into the Kong Co. It cuts an early Q3 fan and has created two to three river terraces. The linear strand of the fault cuts across several alluvial fans and rills, and has dammed this river, forming a boot-shaped sag pond (Fig. 5B). Two late Q3 fans (Fan4 and Fan5) have been displaced at the apex (fan head) of the fans. Minor normal faulting and the relief have formed an uphill facing fault scarp with a height of 1–2 m. Long-term right-lateral movement has resulted in the western edges of the alluvial fans becoming abandoned (i.e., no longer receiving sediment from the fan head), revealing the history of fault displacement (Fig. 4A). The surface color of Fan3, Fan4, and Fan5 is brown, marked by small dark dots and short lines on the satellite images, which contrasts with terrain outside the fans (Fig. 4). The fringes of Fan4 and Fan5 connect to streams and a lake (Kong Co), and the flanks of these fans form dry rills (Figs. 4A and 5D). In the field, we discovered that the dark dots and lines are black angular gravels, or even boulders scattered on the fan surfaces. Moreover, beyond the fan region there are virtually no black angular boulders (e.g., Fig. 5C). Thus, we used these features to define the coverage areas of Fan3, Fan4, and Fan5 (Fig. 4B), and measured the displacements at their western margins using GeoEye images. The right-lateral offsets determined for Fan4 and Fan5 are 40 ± 5 and 36 ± 4 m, respectively (Fig. 4B).

At Quba site 1, Fan3 has one of the best-defined geomorphic markers comprising black angular tuff boulders on its western edge, which we studied in detail using a high-resolution satellite image (Figs. 5D and 5E). We used the method of projecting piercing lines onto the fault trace to measure fault offset (updated in Chen et al., 2013). The western flank of Fan3 is relatively linear and can be used to construct piercing lines. We then projected the piercing lines onto the fault trace from each side of the fault. The measured displacement between the two piercing points is ~65 m. A displacement of ~80 m was obtained from the eastern flank, but this result is not reliable due to the unclear nature of the fan margin (Figs. 5D and 5E). The black angular boulder deposits form a gravel ridge line (GR Line) on the western edge of Fan3 (Figs. 5C–5E). This GR line is also relatively linear, and thus is a good marker for offset measurement. We obtained a range of displacements from this feature of 60– 70 m, given that the upper part of the GR line has an uncertainty of ~10 m (Fig. 5E). Reconstructions of the offset from the three-dimensional surface topography show a best-fit displacement of 65 m. Using this value, the thalweg and eastern margin (or western margin of Fan3) of a dry rill have been reconstructed (Fig. 5A). We consider 65 ± 5 m to be a reasonable estimation for the displacement, with the relatively small uncertainty reflecting the young age and robust nature of the markers. We then collected two medium- to coarse-grained sand samples for OSL dating from the abandoned western edge of Fan3 on the uplifted side to the south of the fault (Fig. 5C). Although the difference between their sampling depths is >0.6 m, we obtained similar ages of 19.15 ± 1.77 and 19.58 ± 2.12 ka, respectively (Table 1). The similar ages indicate rapid deposition in this section and may suggest that the abandoned western edge of the fan has largely remained stable since 19 ka. This possibility is consistent with the absence of significant gully incision into Fan3 and the many black angular boulders on the surface of this fan area (Fig. 5C). Therefore, all the deposits of Fan3 may have formed during a very short time and remained stable since deposition. As such, the GCF has displaced Fan3 by 65 ± 5 m since 19 ka at an average slip rate of 3.4 ± 0.4 mm/yr.

At Quba site 2, a gully has incised into an early Q3 fan (Fan6) and forms a T2 terrace with a height (relative to the modern channel) of 3 ± 1 m and T1 terrace with a relative height of 1 ± 0.5 m. The T1 terrace was only deposited on the left bank, near the southern side of the fault (Fig. 6). Almost all the risers of the T2 terrace are adjacent to the modern channel. Thus, the risers of the T2 terrace are effectively gully banks. Right-lateral movement upon the fault has displaced the gully banks (risers of T2) (Fig. 6A). Given that deflected channels are usually subjected to water erosion when they change direction, stream deflections are not limited to the fault zone but extend beyond the fault (e.g., Huang, 1993). The upper parts of the gully banks are reasonably linear, so we used these to construct gully fitting lines and then piercing lines (Line1 for the east bank and Line2 for the west bank) from each side of the fault; the lines were projected onto the fault trace. This yielded displacements between the piercing points of 17 and 15 m, respectively (Figs. 6A and 6B). The studied gullies have reasonably linear channels, resulting in small uncertainties (Chen et al., 2013), and as a result a ~10% uncertainty was assigned, yielding an average offset of 16 ± 2 m for the gully bank.

Fig. 6

(A) Photographs showing gully offset and sample locations, along with a depth profile of the depositional setting at Quba site 2. (B). GeoEye image showing a displaced gully at Quba site 2, along with a geological interpretation and measurements of offset.

We collected two samples from the lower deposits of T1 and surface deposits of T2 on the southern side of the fault (Fig. 6). The T1 section (Pit b) consists mainly of loose, medium- to coarse-grained sand with gravel deposits comprising angular clasts, and a scour pit was found in the upper part of the section (Fig. 6A). At a depth of ~1.5 m the section reaches the height of the modern river bed. Below a depth of ~1.5 m the section comprises alluvial gravel and silt. The sediment characteristics are similar for the lower part of the section in T2 (Pit a) (Fig. 6A), which may suggest that both are deposits of Fan6. As such, dating of the surface deposits of T2 would constrain the maximum age of the gully, and dating of the lower deposits of T1 would constrain the minimum age of the gully. OSL results show that the maximum age of abandonment of the T2 terrace (gully banks) is 55.61 ± 4.88 ka (Table 1), whereas dating of the earliest deposits of the T1 terrace suggests that the minimum age of the T2 terrace (gully banks) is 8.29 ± 0.67 ka. We also interpret that the gully is relatively young, given the relatively shallow incision depth (~3 m) and small drainage area. The T2 terrace represents an abandoned original fan surface, and its age (~56 ka) is consistent with the age of the upper sand lens in Fan1 (~61 ka). Moreover, the displacement of this gully is also consistent with the rill offsets (14–18 m) on the surface of Fan1 (Fig. 3B). The age of the gully is not equivalent to the surface age. We also note that the T1 terrace was only deposited on the left bank and has since been protected by the right-lateral offset. Therefore, the onset of deposition of T1 is likely to document the displacement history of this young gully and yields a slip rate of 1.9 ± 0.3 mm/yr.