Late Miocene to Quaternary Development of the Jiujing Basin, Southern Beishan Block, China: Implications for the Kinematics and Timing of Crustal Reactivation North of Tibet

Central Asia is characterized by distributed intracontinental deformation largely driven by the ongoing Cenozoic India-Asia and Arabia-Asia collisions (e.g., [1–6]). Within the huge Central Asian active deformation field, the Beishan region of western China is located directly north of the Tibetan Plateau and bordered by the Tarim Basin, Altyn Tagh fault, and Qilian Shan to the south and southwest, the Eastern Tien Shan to the west and northwest, and the Altai and Gobi-Altai to the north and northeast (Figure 1) (e.g., [6, 7]). Compared to these surrounding regions, the Beishan has lower mountain elevations and relief, fewer and smaller Late Cenozoic alluvial basins, and relatively low levels of historical seismicity. Thermochronological studies suggest that the Beishan experienced several phases of rapid exhumation since the late Triassic [8, 9], but not during the Late Cenozoic. This suggests limited neotectonic rock uplift and exhumation (<2 km) in the Beishan region compared to elevated mountainous regions to the south along the northern margin of the Tibetan Plateau and to the north in the easternmost Tien Shan. For these reasons, the Beishan has long been considered more tectonically stable than adjacent regions (Figure 1) [2, 6, 10], perhaps akin to the Tarim Block to the west that transmits compressional stresses derived from the Indo-Eurasia collision northwards, whilst remaining largely unaffected internally by those stresses (e.g., [8]). Consequently, the neotectonic activity of the Beishan has received limited modern study and the region represents one of the least understood intraplate, intracontinental regions within the huge Late Cenozoic Indo-Eurasia deformation field (e.g., [5, 6]).

At first order, the Beishan is a regional plateau with two second-order E-NE-trending topographic and structural culminations consisting of the Mazong Shan deformation belt and the Xingxingxia fault system (Figure 1). Both culminations contain surface ruptures that cut Quaternary alluvial sediments and are discernible on satellite imagery (e.g., [2, 6, 11]). Crustal lineaments and sharply defined mountain fronts with low sinuosity occur throughout the wider Beishan region, and high-resolution imagery reveals a large number of other faults that cut surface deposits and thus appear to have been active in the Quaternary [6, 11], although most have never been ground checked. Jurassic-Cretaceous-age summit peneplains that tilt away from sharply defined faulted mountain fronts that are flanked by fresh-appearing alluvial fans also suggest basement block uplift in the Late Cenozoic in many parts of the Beishan. Therefore, open questions have remained regarding the distribution and magnitude of crustal reactivation in the Beishan region, the kinematics of Quaternary faulting, and the connections, if any, of structures in the Beishan with surrounding deforming belts. In addition, the timing of Late Cenozoic crustal reactivation in the Beishan region is undocumented.

Because there is no historical record of moderate to strong earthquakes in the Beishan region and low levels of seismicity characterize the region today [10], the Beishan has been considered a candidate area for high-level radioactive waste storage ([12] and reference therein). In addition, although the Beishan region is sparsely populated, Silk Road oasis cities and towns directly south of it are heavily populated with approximately 4-5 million inhabitants (World Population Review: http://worldpopulationreview.com/countries/china-population/cities/). China’s “Belt and Road Initiative” includes a major connecting corridor through the Beishan region as part of the 21st Century Silk Road Economic Belt. Therefore, identification of Quaternary fault activity within the Beishan region has important implications for regional seismic hazard assessments and infrastructure development [11, 13].

In this study, we document a previously unreported domain of active faulting and Quaternary basin development in the southern Beishan region. Three small NE-trending basins with Quaternary infill are documented; the easternmost and largest of the four basins is called the Jiujing Basin (Figure 2) [11, 14, 15]. The origin of the Jiujing Basin in the wider context of the Late Cenozoic reactivation of the southern Beishan is the subject of our study (Figure 1). We apply remote sensing analysis, borehole data, fault trenching, and optically stimulated luminescence and cosmogenic burial dating methods to document the basin’s evolution in time and space and the kinematics of the major faults responsible for creating the basin. The results of our study have implications for the active deformation field directly north of Tibet and the earthquake potential in the westernmost Hexi Corridor and southern Beishan region.

On a regional scale, Beishan crust comprises a Paleozoic terrane collage in the southern sector of the Central Asian Orogenic Belt (CAOB). Three separate Carboniferous-Permian suture belts in the Beishan indicate multiple collision events due to subduction and closure of different arms of the Paleo-Asian Ocean ([7, 16] and references therein). Examination of published geological maps and freely available satellite imagery of the Beishan reveals polydeformed basement, cross-cutting intrusions, and folded and refolded sedimentary compartments and metamorphic belts (e.g., [17]). Major faults such as the Xingxingxia, Pochengshan, and Mazongshan fault zones reveal an older ductile phase of displacement prior to Late Cenozoic reactivation (Figure 1) [6, 7, 17–19]. The Beishan crust is also heavily granitized with Late Permian-Jurassic intrusive bodies occurring throughout the region [20]. The region studied in this paper occurs within the Shibanshan arc terrane (south Beishan) which consists of mid-late Paleozoic granite, gabbro and diorite, Devonian clastic rocks and slates, Carboniferous clastic sedimentary rocks, slates, phyllites, limestones, felsic to intermediate volcanic and pyroclastic rocks, and Permian volcanic, pyroclastic assemblages, and limestone [7, 21, 22]. Despite the complexity of the pre-Cenozoic deformational history, there is a dominant E-W structural grain in the southern Beishan and granitic intrusions occur that cross-cut older basement rocks are typically also deformed (Figures 2(a) and 3(a)). Some intrusions show concentric magmatic zonation patterns including the Xinchang intrusion in the study area (Figure 2(a)). Mafic dike swarms are also widespread in the intrusive complexes. Paleogene sediments are uncommon in the region, but Neogene and Quaternary semi-indurated or unconsolidated alluvial sediments eroded from nearby mountain blocks fill low-lying basinal areas [6, 21].

Thermochronological studies reveal that the Beishan experienced episodic exhumation from the late Triassic to early Paleogene, but apparent tectonic quiescence since the Neogene [8, 9]. However, because thermochronological methods fail to easily distinguish sub-2 km rock uplift events [23, 24], the extent of low-relief-generating Late Cenozoic tectonism and reactivation of the Beishan is poorly understood. For comparison, pre-60 Ma rapid exhumation and subsequent thermal quiescence of the Nanjieshan east of Dunhuang (Figure 1) is suggested by apatite Helium analysis [8]. However, Neogene and Quaternary faulting and sub-2 km block uplift are well documented in the Nanjieshan by field studies and remote sensing analysis [25, 26]. In addition, semiconsolidated Neogene deposits (Kuquan Formation) within the Beishan are widely folded and faulted according to 1 : 200,000 scale geological mapping (and confirmed by satellite image analysis) [21]. Furthermore, recently, Yang et al. [13] documented a ~20 km-long active fault system along the southeastern margin of the Beishan, which challenges long-held views of Holocene crustal stability in the surrounding region. An open question is whether during the Late Cenozoic, the Beishan has been dominantly reactivated by strike-slip rather than dip-slip displacements. If so, then the topographic expression of crustal reactivation (especially strike-slip movements at low strain rates) is likely to be more subtle and harder to recognize, than in surrounding, more mountainous regions (Eastern Tien Shan, Gobi Altai, and Northern Tibetan margin) that are characterized by Late Cenozoic transpressional deformation.

According to published geological maps of the Jiujing Basin region, the bedrock surrounding the basin consists of Ordovician-Silurian schists, gneisses, and marbles and Permian plagiogranites and diorites ([21, 22] and Figure 1). An array of NE-striking, NW-dipping faults cut the bedrock over a 50 km wide, east-west zone. The Jiujing fault (JJF) is part of the array and bounds the Jiujing Basin on its eastern side with obvious normal fault displacement (Figure 2). The low-lying Jiujing Basin hanging wall is filled with Quaternary alluvial materials, whereas the footwall to the east comprises an elevated and east-tilted block of Xinchang granite. Other faults cut the basin and the west Jiujing fault (WJF) partly bounds its northwestern side (Figure 2(b)). The traces of the JJF and WJF are laterally constrained by the E-W striking Jinmiaogou fault zone to the north and the Hongqishan fault zone to the south (Figure 2).

Detailed fault traces of the JJF and WJF (Figure 2) were mapped using 0.5 m resolution Worldview 3 imagery to identify topographic lineaments, fault scarps, triangular facets, offset gullies, and vegetation lineaments (e.g., [27]). Additionally, a relief map of ALOS 5 m DEM data and 1 m DEM data extracted from Worldview 3 stereo imagery helped improve the visual contrast of key geomorphological features. We also constructed two high-resolution DEMs (Figures 4(a) and 5(b)) using Agisoft photoscan software based on the structure-from-motion method. At each site, at least 100 photos were captured by a DJI Phantom 3 Pro drone. The resultant DEMs have a resolution of 3~5 cm/pixel (c.f., Yang et al. [13, 26]). Printed shaded relief maps with remote sensing interpretations were carried to the field for validation, modification, and as basemaps for further ground study. We mapped and documented the locations, geometry, and slip sense of major and minor faults in the Jiujing region (Figures 3 and 6) in order to understand the overall fault network, kinematics, and deformation regime.

Vertical separations across the fault scarps were measured in the field using a laser range finder (e.g., Figures 7(a) and 7(e)). An E-W-oriented topographic profile across the Jiujing Basin, JJF, and Xinchang granite was extracted from an ALOS 5 m DEM (inserted profiles in Figure 2(b)). In addition, some small gullies were identified that are left laterally displaced. We determined the offset value from piercing points on each side of the fault after projecting them to the fault trace by field surveying (Figures 7(b) and 7(d)).

In order to document the Quaternary fault activity of the JJF, we excavated one trench across the JJF (Figure 2). Trench TC1 was sited along the central part of the JJF on the eastern margin of the Jiujing Basin. Photographic images were used to interpret the fault structure and stratigraphic features, including visible fault planes, upward fault terminations, offset trench horizons, and scarp-derived colluvium.

To constrain the timing of past fault activity, we collected one OSL samples from a sand wedge adjacent to the fault using a stainless tube. The OSL sample was processed under subdued red light and measured at the Research Laboratory of Luminescence Dating at the Institute of Geology, China Earthquake Administration, following the methods of Aitken and Smith [28] and Lu et al. [29]. Sediments were processed with H₂O₂ and HCl to eliminate organic materials and carbonates after removing the outer parts of the tubes whilst avoiding exposure to sunlight. Quartz-rich fractions with diameters of 90-180 μm were etched with 40% hydrofluoric acid to remove feldspar minerals and alpha-irradiated surfaces of quartz grains. The quartz purity was checked by measuring the infrared stimulated luminescence signal. The etching samples were mounted on stainless steel discs for equivalent dose measurement following the single-aliquot regenerative dose (SAR) protocols [29, 30]. The OSL dating result is listed in Table 1. The luminescence signal decay curves, dose growth curves, and dose distribution are presented in supplementary figure S1.

The asymmetric topography of the Jiujing Basin suggests that it is an extensional basin controlled by a range-bounding fault along its east margin with deepest basin fill near the east basin margin. To test this idea, we drilled three boreholes in the eastern half of the basin to document the character and thickness of basin fill and the timing of initial basin sedimentation (Figure 2). We consider it likely that the onset of basin sedimentation is close to the timing of dip-slip activity on the JJF. Subsidence of the hanging wall would have produced the accommodation space for alluvial sedimentation derived from the adjacent Xinchang bedrock block on the east side of the basin (c.f., [31]).

In addition, if the sediments eroded from the source area are gradually buried, slow sedimentation produces postburial cosmogenic nuclides, which will increase the concentrations and ratio of ²⁶Al and ¹⁰Be and result in an apparently young burial age [34, 38, 42]. Thus, the total nuclide concentrations in the sample should subtract the component of postburial nuclide production in order to calculate an appropriate burial age. The detailed postburial production correction follows the formula and parameters in Thompson Jobe et al. [38].

In this study, we mainly calculated the sample burial ages assuming simple steady-state erosion and no postburial production. All the calculations were conducted based on referenced sea level, high latitude ¹⁰Be and ²⁶Al production rates of 4.0 and 28.5 atom/g/a [43] scaled for the Lal [32]/Stone [44] time-dependent production model. The half-lives of ¹⁰Be and ²⁶Al are $1.387±0.012$ Ma and $0.708±0.017$ Ma, respectively [40, 41, 45], and the initial ²⁶Al/¹⁰Be spallation production ratio in the same material is ~6.75 [37]. The ²⁶Al/¹⁰Be ratios of two samples are calculated from the CRONUS-Earth online calculator [46].

Two fine sand samples were collected from the B3 borehole drill core at a depth of ~54 m and ~57 m, to determine the burial age of the basin sediments. The pretreatment of samples was performed at the Cosmogenic Nuclides Laboratory, Institute of Crustal Dynamics, China Earthquake Administration. Details of sample processing follow the description by Kohl and Nishiizumi [47]. The 250~500 μm nonmagnetic fraction of the sand sample (0.5~1 kg) was sieved after magnetic separation and removal of carbonate, organic matter, and iron ions using HCl and H₂O₂. A dilute HF/HNO₃ solution was used to etch samples to remove feldspar and partial quartz grains affected by meteoric ¹⁰Be. The 2 g purified quartz was dissolved in concentrated HF, and the concentration of ²⁷Al was analyzed using an inductively coupled plasma optical emission spectrometer (ICP-OES) for determining the purity of the samples. The ²⁷Al concentration ranged from 29 to 37 ppm which is sufficiently low to merit analysis. The remaining purified quartz (approximately 30 g) was dissolved using concentrated HF with added ~0.2 mg ultrapure ⁹Be and ~0.6 mg ultrapure ²⁷Al carriers, and fluorides were eliminated by HClO₄ fuming. Beryllium and aluminum were progressively separated from the other elements using anion and cation exchange chromatography. Ammonia water was added into isolated solutions to produce hydroxides of Al(OH)₃ and Be(OH)₂, respectively. The precipitates were poured into a quartz crucible and placed in a furnace burning for 30 minutes at 950° conditions to yield a powder of Al₂O₃ and BeO. The resultant samples were sent to the ASTER AMS French national facility (CEREGE, Aix-en-Provence) for making targets and measuring the ratios of ¹⁰Be/⁹Be and ²⁶Al/²⁷Al. All analytical results are listed in Table 2.

To better establish the regional context of Quaternary deformation in the Jiujing Basin area, we analyzed high-resolution satellite images and remotely mapped surrounding faults. Based on identification of offset lithological and structural features, topographic lineaments, and fault escarpments, apparent faults were digitized (Figure 3(a)). A small percentage of the mapped faults reveal clear indicators of Quaternary sediment offsets and slip sense on satellite imagery, whereas the age and kinematic history of the remainder (majority) are uncertain. Of the faults that reveal slip sense on satellite imagery, NE-trending faults are characterized by left-lateral strike-slip movement, whereas NW-trending faults reveal evidence for right-lateral strike-slip. The nearly seven hundred NW- and NE-striking faults are mostly located between two regional E-W trending strike-slip fault zones: the Hongqishan fault zone (HQSF) to the south and Jinmiaogou fault zone (JMGF) to the north (Figures 2 and 3(a), [14, 15]). Statistical analysis of the strike of 681 subfaults between the HQSF and JMGF fault zones shows a dominant NE-trending distribution (insert map in Figure 3(a)).

The southern-bounding HQSF is about 120 km long and is developed largely within Ordovician-Silurian marbles and schists [21]. It is visible as a linear fault trough with obvious structural drag and low-angle truncation of adjacent basement strike trends indicating sinistral shear along its eastern extent (Figure 3(b)) [15]. In addition, along one sector, Quaternary sediments are clearly cut and there is apparent S-side-up and left-lateral offset of alluvial landforms and drainages (Figures 6(c)–4(f)). Farther west, there is limited image-based evidence that the HQSF is a strike-slip fault because the system mainly cuts through remote bedrock terrain parallel to basement strike trends, and Quaternary alluvial deposits and landforms are absent. However, previously reported HQSF fault exposures reveal that the fault is steeply south dipping (70-80°), and thermo-luminescence dating of fault gouge in the HQSF zone provides an age of ~176 ka, indicating activity during the middle Pleistocene [15].

The north-bounding JMGF is about 90 km in length and is a clearly defined lineament on satellite imagery that is approximately parallel to basement structural trends. Small NW-striking dextral faults adjacent to the JMGF are clear on satellite imagery (Figure 6(b)) and suggest an origin as antithetic R’ shears connected to the overall JMGF left-lateral fault system which is expressed as a linear trough. Previously reported fault striations in field-checked steep 80-85° south-dipping fault surfaces have pitch angles of 5-15° towards the E indicating dominantly left-lateral strike-slip motion with a subordinate extensional dip-slip component [15]. A thermoluminescence age of $468.42±39.82$ ka from sandy deposits displaced by the fault indicates activity during the middle Pleistocene [15].

The NE-striking Jiujing fault system and Jiujing Basin are located within a 15-20 km-wide stepover zone between the E-W-striking HQSF and JMGF. The Jiujing fault (JJF) cuts across bedrock and range-front basin sediments forming offset gullies and risers, fault scarps, triangular facets, and knickpoints [14, 15]. At the southern tip of the JJF, subdued topography limits the development of stream gullies; thus, evidence for strike-slip displacements along the fault that might be revealed by drainage offsets was not identified. However, bedrock that is overlain by thin sediment cover is vertically displaced by ~1.5 m (Figure 7(a)). Farther north, the JJF cuts across bedrock of the ‘South Ridge’ and is right stepping (Figure 2(b)). Significant structural relief of ~70 m between the Jiujing Basin and the Xinchang granite along the JJF is marked by triangular facets and ~2-4 m high retreated knickpoints, suggesting a normal component of displacement along the fault (Figures 2 and 7(b)). Additionally, some west-flowing gullies are left laterally offset from ~1.0 to ~2 m (Figures 7(c) and 7(d)). The fault scarps are ~0.9 m high and likely record the most recent one or two seismic events (Figure 7(e)). At the northernmost end of the JJF, the fault is divided into several segments with curved fault traces.

Trench TC1 is excavated across the central segment of the JJF (Figure 2). Fault traces along this segment are very clear, and some 30-40 cm-high west-facing scarps are present (Figures 4(a) and 4(b)). Along a small west-flowing gully, two levels of terraces T2 and T1 are developed (Figure 4(b)), of which T2 is displaced forming a 30-40 cm-high scarp (Figure 4(d)). The T1 terrace and floodplain are not offset. Trench TC1 was excavated into the T1 terrace at the projection of the fault trace (Figures 4(b) and 4(c)). The small trench TC1 is only 2 m long and 1 m deep. Five stratigraphic units were recognized in the south wall (Figures 4(c) and S2). U1 is preexisting fault breccia with well-developed fault-parallel, brittle shear fabrics and fluid alteration and discoloration. U2 consists of clast-supported angular gravels with size range of 1~15 cm. U3 consists of medium-to-coarse-grained sands. U4 contains angular gravels filled with few sands. U5 contains surface gravels and sands. An NW-dipping normal fault displaces units U2, U3, and U4 but is covered by U5, which is consistent with the lack of no surficial scarp in T1 terrace. An OSL sample collected from the faulted U3 sand deposit has an OSL age of $14.9±2.2$ ka (Figure 4(c), Table 1), which likely indicates a late Pleistocene seismic event. Wang et al. [15] obtained a thermoluminescence age from the faulted alluvial fan of $17.56±1.49$ ka, which is consistent with our most recent event age determination.

The WJF is ~12 km long, located to the west of the JJF, and runs from the east side of the South Jiujing Basin, west of the ‘South Ridge’, then cuts into the Jiujing Basin, terminating at the inactive Shiyuejing fault (Figure 2(b)) [15]. Along the South Ridge, the WJF forms a 10 m wide fault trough, offsets bedrock, and forms a fault scarp and eyebrow ridge (Figures 7(f) and 5), revealing left-lateral strike-slip movement with a component of dip-slip displacement. Along the western side of the Jiujing Basin, the WJF cuts through basin sediments and bedrock and forms clear lineaments and linear vegetation traces (Figure 8) suggesting the WJF has been active in the late Quaternary.

The three boreholes drilled into the eastern Jiujing Basin reveal increasing thicknesses of basin fill towards the JJF and a general fining-upwards sequence for the more distal B1 and B2 cores (Figure 9). This suggests that the half-graben-shaped basin was formed by movement along the JJF on the basin’s eastern border (Figure 9(a)). Borehole sedimentary thickness variations and modern drainage patterns strongly support the inference that the alluvial samples were sourced from the eastern footwall block composed of the Xinchang granite.

The B1 borehole is approximately 14 m deep with basin fill comprising the upper 9.2 m (Figure 9). Weathered and fractured granite occurs below the sediment. The basin fill is composed of yellow-grayish clay near the surface, that is, underlain by brown or gray clay and silty sand and yellow silty sand. Fine sand and medium-coarse sand occur at the bottom of the sedimentary succession.

Borehole B2 is located 350 m east of B1 and is ~24 m deep with a 13 m thickness of basin deposits (Figures 2(b) and 9). The sediments are composed of yellow clay, silt and sand, fine sand and medium-coarse sand, brown clayey sand and sandy clay, and some gravels. The basement granite is strongly weathered and fractured.

Borehole B3 is located 320 m ESE of B2, 180 m west of the fault trace and is the deepest core with ~58 m thickness of basin fill (Figures 2(b) and 9). According to sediment color, composition, and degree of induration, the sedimentary section can be divided into two parts. The upper part that is above ~36 m depth is composed of unconsolidated yellow fine sand, silty sand, medium-coarse sand, and gravels with gray fine sand and medium-coarse sand. The semiconsolidated lower sedimentary section is mainly composed of yellow-brown medium-coarse sand, fine sand, and gravels. Two very thin layers of caliche are developed at 47 m and 57 m depth.

Measured nuclide concentrations, their ratios, and calculated burial ages of two samples are shown in Table 2. Cosmogenic ¹⁰Be concentrations of two samples collected from borehole B3 are $5.09±0.25∗104$ and $26.52±0.92∗104$ atoms/g, respectively (Table 2). The ²⁶Al concentrations are $2.34±0.63∗104$ and $11.44±2.27∗104$ atoms/g, respectively. The concentrations yield ²⁶Al/¹⁰Be ratios of 0.43~0.46. The burial ages of samples B3B-3 and B3B-1 were estimated to be $5.40+0.60/−0.32$ Ma and $5.51+0.49/−0.59$ Ma, respectively (Table 2). The close agreement of the two sample ages that are in normal stratigraphic order within error indicates that the sediments were not reworked. Also, no evidence exists to suggest a more complex detrital transport history and sediment reworking. Moreover, the low orogenic relief, internally drained Jiujing Basin, regionally slow exhumation rates calculated from thermochronological data [8], erosion-resistant footwall granite, and arid climate also support a simple source-to-sink erosional-depositional system.

Consideration of the two burial ages and their shallow depth in drill core B3 yields an extremely low sedimentation rate (~10 m/Ma). This low sedimentation rate is not consistent with a fast sediment burial system, suggesting that the postburial production may increase the concentration of the nuclides and underestimate the burial ages [34, 38, 42]. Magnetostratigraphy and borehole drilling in the Jiudong Basin, 250 km SE of the Beishan, constrain a sedimentation rate of 60 m/Ma during the last 6-5 Ma [48]. Assuming similar climate conditions between the Beishan and the Qilian Shan during the late Miocene, and applying the Jiudong sedimentation rate of ~60 m/Ma, the burial ages of Jiujing Basin samples B3B-1 and B3B-2 are ~5.78 Ma and ~5.67 Ma, respectively. If we assumed a higher average sedimentation rate of ~100 m/Ma (calculated for NE Tibetan Plateau; [42]), the burial ages for sample B3B-1 and B3B-3 would be ~5.58 Ma and ~5.65 Ma, respectively. Both scenarios suggest that postburial production accounts for less than 5% of the burial ages for the two samples. Additionally, the measured nuclide ratio can also be used to identify any significant, postburial production by plotting a ²⁶Al vs. ¹⁰Be diagram [35, 36]. When the data array lines are not rotated and pass through the origin, the sediments are considered to have less or an absence of, significant postburial production [35, 36]. In our case, the fit line to the ²⁶Al vs. ¹⁰Be concentrations of the two samples basically pass through the origin (Figure 10), suggesting minimal postburial accumulation. Furthermore, previous modeling corrected for burial ages with different sedimentation rates or different burial scenarios suggest that low sedimentation rates (<100 m/Ma) increase burial ages by 5-15% (at most) for Pliocene to Pleistocene sediments [42, 49]. Thus, collectively, we suggest that postburial nuclide production appears to have had a negligible effect (<5-10%) on burial ages in the southern Beishan (Jiujing) area in areas of low sedimentation rates. Additionally, the uncertainty of 5% for the burial age caused by postburial production during the initial burial is also close to the analytical uncertainty.

We also note that the late Miocene ages approach and even exceed the reported limitation (0.5-6 Ma) of the ²⁶Al/¹⁰Be burial dating method (e.g., [33, 34, 37]). The limit to burial dating mainly depends on the reliability of the ²⁶Al/²⁷Al measurement [34]. The ²⁶Al/²⁷Al and ¹⁰Be/⁹Be ratios of our two samples are high enough to yield very reliable levels of uncertainty (Table 2). Additionally, the field area has a relatively high elevation (1640-1740 m asl) and an extremely low erosion rate. Therefore, the sediments have a potentially long period for exposure in upstream catchment areas leading to relatively high initial nuclide concentrations at the time of burial [37, 50]. Thus, high initial concentrations permit reliable burial age calculations back to the late Miocene.

The geometry and kinematics of the JJF and its structural relationship with bounding E-W-trending strike-slip faults to the N and S suggest that it is part of an actively developing strike-slip transtensional duplex (Figures 2, 3(a), and 11). The duplex occupies a stepover position between the E-W-striking HQSF and JMGF with parallel NE-striking faults such as the JJF accommodating strike-slip and normal components of displacement within the stepover zone (Figure 11) [51, 52]. We suggest, based on documented case studies and sandbox analog experiments of transtensional stepovers [53, 54], that as the bounding HQSF and JMGF progressively lengthened and their tips increasingly overlapped, the duplex within the stepover grew westward (Figure 11). Similarly, the Jiujing duplex can also be viewed as a trailing transtensional array as the Xinchang granite crustal block between the HQSF and JMGF has been displaced eastwards (Figure 11). Multiple geological observations support a model of east-to-west widening of the duplex through time including the following: (1) the amount of fault displacement is greater in the east than in the west (Figures 2(a) and 11). (2) The decreasing basin width is obvious on satellite imagery from 3 to 4 km in the Jiujing and South Jiujing Basins, to ~1 km in the Bantan Basin, to ~500 m in the Erdaojing Basin; these other basin-bounding faults are also sinistral strike-slip with normal components of displacement (Figures 2(a), 3(a), and 3(d)) [15]. (3) The summit surface of the Xinchang granite footwall block to the east of the Jiujing Basin is tilted eastward, whereas the Bantan and Erdaojing Basin footwall blocks are untilted or less tilted, suggesting less dip-slip displacement and footwall unloading on their bounding faults.

Geodetic and crustal stress data indicate NE- or ENE-directed crustal movement and ENE-oriented maximum horizontal stress (SH_max) for the southern Beishan region relative to a fixed Siberia (e.g., [55, 56]. The angular relationship between NE-striking compressional stress and E-W-trending basement fabrics, strike belts, and older shear zones promotes sinistral displacement on the E-W-trending HQS and JMG strike-slip faults (c.f., [6, 57]). NE-striking sinistral transtensional faults such as the JJF and WJF are also compatible with a Late Cenozoic NE-directed SH_max and NW-SE direction of crustal tension.

Cosmogenic ¹⁰Be/²⁶Al burial dating of the deepest sediments in the Jiujing Basin yields the initiation age for basin deposition at ~5.5 Ma, which we suggest marks the onset of JJF movement and tectonic lowering of the fault hanging wall to create the accommodation space for sediment accumulation (e.g., [31]). Additionally, unconsolidated Neogene (and possibly early Quaternary) coarse sand and gravels that accumulated in the low-lying Yinwaxia depression west of the Jiujing Basin are now folded, tilted, and sinistrally faulted (Figure 3(c)) [11, 21], suggesting Late Cenozoic sinistral transpressional deformation occurred directly NW of the Jiujing area. Paleoseismological data from the JJF trench site indicate late Pleistocene JJF activity. Thus, we conclude that discrete areas of the southern Beishan block have experienced crustal reactivation during and since the late Miocene.

The JJF has a continuous surface trace (Figure 2), without topographic barriers or structural complexities, suggesting that the entire fault may be capable of rupturing during a single earthquake event (e.g., [58]). The ~20 km length of the JJF is comparable to other continental strike-slip and normal fault single-rupture segment lengths, which likely correlates with seismogenic crustal thickness [59]. Thus, accepting that the entire length of the JJF could rupture in a single event, then the empirical relationship between moment magnitude and surface rupture length yields a maximum potential earthquake magnitude of 6.6 [60].

Suitable offset landforms for estimating the JJF slip rate were not identified in this study. However, Wang et al. [15] previously determined normal and sinistral component slip rates along the JJF to be ~0.03 mm/a and ~0.18 mm/a, respectively (Figure S3). The low basin-margin relief also suggests a low slip rate, which is consistent with documented low slip rate faults elsewhere in the Beishan-Alxa region north of Tibet (Figure S3). For example, east of the Beishan in the Alxa Block, average strike-slip rates of 0.23~0.78 mm/a are calculated for the Yabrai fault and 0.14~0.93 mm/a for the Taohuala Shan fault during the last 200 ka [61, 62]. They estimated maximum magnitude of a potential earthquake on the Yabrai fault is 6.3 to 6.6, based on the empirical relationship of Wells and Coppersmith [60]. Limited trench results also reveal a very long recurrence interval for the NE-trending Bayanluorigong sinistral strike-slip fault within the NE part of the Alxa block (at least ~50 ka) [63]. Along the southeastern margin of the Beishan, Yang et al. [13] yielded a strike-slip rate of ~2.6 mm/a for Holocene motion on the Beihewan fault (BHWF) and determined that the most recent earthquake had an estimated magnitude of 6.3-6.9. A slower slip rate determination of ~1.5 mm/a for the BHWF is also documented [64]. The calculated slow strike-slip rates and long recurrence intervals determined for faults within the Beishan and Alxa blocks are consistent with the low geodetic crustal velocities and limited seismicity (e.g., [10, 55]), suggesting that the Beishan-Alxa Blocks should be classified as a low-strain region where tectonic loading may be complexly partitioned amongst a currently unknown number of potentially active faults [65]. This is in strong contrast with the higher rates of Quaternary deformation along the Qilian Shan and Altyn Tagh Fault to the south (e.g., [66–68]).

In addition to the Jiujing, Bantan, and Erdaojing Basins, other NE-striking basins occur in the southern Beishan region and appear on satellite imagery to have bounding faults that cut alluvial materials and form topographic scarps, thus suggesting Quaternary extensional/sinistral transtensional activity. These include the nearby Yemajing and Xinshan Basins (Figures 3(a) and 3(e)) and two grabens farther N and NE in the eastern Mazong Shan region (Figure 1). These other basins cut obliquely across E-W basement structural trends and contain actively aggrading alluvial fan systems shed from bordering basement blocks, which have very low mountain-front sinuosity. Image analysis indicates that these bounding faults have a normal sense component and perhaps a sinistral strike-slip component, which is less expressed in the topography and harder to prove without ground study. The huge Ejina Basin east of the Beishan (Figure 1) also contains NE-trending left-lateral normal faults [69] based on interpreted ground-penetrating radar survey data and geological observations. Development of multiple N-NE-trending grabens in the southern Beishan contrasts with ESE-WNW to E-W-trending compressional and transpressional basins in the Tianshan to the northwest, the Qilian Shan-Hexi Corridor to the south, and the Gobi-Altai to the north [2, 6, 70]. Thus, the previously unreported, actively forming transtensional basins of the southern Beishan represent a relatively uncommon, but kinematically compatible style of strain accommodation in the larger neotectonic deformation field north of Tibet. Documenting the full range of transtensional basin development and how the basins link kinematically with E-W strike-slip faults and domains of transpressional crustal reactivation to the north and south will require further study throughout the entire Beishan region.

Late Cenozoic faulting in the Beishan is a manifestation of the northward expansion of the Indo-Asia deformation field into the continental interior north of Tibet [1, 2, 5 19, 57]. The initial formation age of the Jiujing fault system at ~5.5 Ma based on cosmogenic ²⁶Al/¹⁰Be burial dating of basin deposits (Figure 9) is consistent with reported Late Miocene onsets of crustal reactivation in northeast Tibetan Plateau and in northwestern China and Mongolia (Figure 12) (e.g., [71, 72]). For example, at the Laojunmiao stratigraphic section in the western Hexi Corridor, the existence of a prominent unconformity, documented increase in sedimentation rate, and facies change from shallow lake to fan delta suggest the Qilian Shan began accelerated uplift at 8-6.6 Ma [73]. Accelerated uplift or flexural subsidence of the Gonghe, Guide, and Linxia Basins in the northeast Tibetan Plateau are also dated at 6~8 Ma [74, 75, 76, 77]. Zheng et al. [72] reported rapid exhumation of the Liupan Shan beginning at ~8 Ma based on apatite fission-track thermochronology. Along the southern margin of the Qilian Shan, Zhang et al. [78] suggested that the Qinghai-Nanshan began to uplift at ~6 Ma based on the nature of sedimentary facies, rapid deposition rates, and paleocurrent analysis. In the Oulonbuluk Shan north of the Qaidam Basin, detrital AFT peak ages and a change in sedimentary provenance imply Late Miocene uplift (⁠$7±2$ Ma, [79]).

Farther north, De Grave [80], De Grave and Buslov [81], and Xu et al. [82] reported accelerated denudation at ~5 Ma in the Chinese-Russian Altai based on AFT analysis of exhumed bedrock. Apatite fission track analysis of the Ih Bogd and Baga Bogd restraining bend massifs of the northern Gobi Altai constrains cooling histories from the Lower-Middle Jurassic to late Cenozoic, with the last rapid exhumation event commencing at $5±3$ Ma with at least 2 km vertical crustal uplift [83]. In southern Siberia, the onset of fast rifting at Lake Baikal is also interpreted to be during the 10-7 Ma period based on changes in depositional facies and increasing subsidence and uplift rates suggested by borehole and seismic reflection data [84, 85]. An increase in tectonic activity in the Baikal rift is also supported by contemporaneous rapid exhumation of the Barguzin range to its east [86].

Collectively, the similarity in the timing of crustal rejuvenation and accelerated rates of uplift and sedimentation north of Tibet suggest an increase in strain accommodation of NE-directed SH_max beyond the Tibetan Plateau in the Late Miocene leading to reactivation of a wide region north of Tibet that continues to the present day [1, 2, 6, 19, 57, 81]. Crustal reactivation is mainly focused in the CAOB terrane collage between older cratonic blocks like the Tarim, North China, and Siberian cratons [1–3, 6, 13, 19, 26, 57]. The kinematics of deformation is largely dictated by the angular relationship between SH_max and preexisting structural trends including older faults, shear zones, and metamorphic and sedimentary strike belts that reactivate and accommodate lateral displacements and crustal shortening within an overall sinistral transpressional regime within the Gobi Corridor region that includes the easternmost Tien Shan, Gobi Altai, and Beishan [6].

In this study, we document the slip history of the Jiujing fault system since the late Pleistocene based on field investigations of the fault system and offset landforms, paleoseismological studies, and analysis of new borehole data with cosmogenic burial dating. Remote sensing examination of the wider region was also carried out to better understand the regional context of the Jiujing fault (JJF) system. The JJF is mainly characterized by oblique sinistral-normal displacement.

Satellite image analysis indicates that the JJF is a connecting structure between the regional E-W-trending Jinmiaogou and Hongqishan fault systems that both show evidence for Quaternary activity. The Jiujing Basin is the easternmost and best developed of three fault-bound, NE-striking, basins between the Jinmiaogou and Hongqishan fault systems. The overall regional structure is an evolving sinistral transtensional duplex which is compatible with the NE-directed SH_max, but a previously undocumented style of Late Cenozoic strain accommodation in a region that is more dominated by diffuse sinistral transpression, such as in the Mazong Shan and Xingxingxia deformation belts to the north and Sanweishan deforming belt to the south [6, 25].

Cosmogenic ²⁶Al/¹⁰Be burial dating of the deepest sediments in the Jiujing Basin indicates that the basin began forming at ~5.5 Ma. In addition, Neogene sediments in the Yinwaxia depression west of the Jiujing Basin are deformed and other E-W fault systems in the southern Beishan including the Jinmiaogou and Hongqishan fault systems and Beihewan fault system in the southeastern Beishan [13] record Quaternary displacements. Thus, increasing evidence supports the contention that the southern Beishan region has been reactivated since the Late Miocene in discrete deforming belts. More work will be needed to determine the complete array of faults that have been active in the southern Beishan since the Miocene and the degree of structural connectivity and regional strain partitioning. Finally, we note that during the late Miocene, crustal reactivation occurred over a vast region from the NE Tibetan Plateau to the Beishan, Gobi-Altai, Altai, and Transbaikalia suggesting transmission of northeast-directed compressive stress increased at that time, presumably due to India’s continued northeast indentation 1500 km to the south, and the stored gravitational potential energy in the crustally thickened northern Tibetan Plateau and Qilian Shan thrust belt ([71, 87, 88] and references therein).

The authors confirm that the data supporting the findings of this study are available within the article.

Key Points. The Jiujing fault array is a transtensional duplex in the southern Beishan, China. Sinistral-normal faults link bounding, regional, E-W sinistral strike-slip faults. Dated basin fill reveals development of the Jiujing Basin began at 5.5 Ma. Quaternary fault scarps suggest the fault system remains active. Widespread crustal reactivation north of Tibet occurred in the late Miocene.

The authors declare that there is no conflict of interest regarding the publication of this article.

We would like to thank Yanwu Lv and Oubo Liang for their help in the lab with ¹⁰Be sample processing. Professor Darryl Granger and Dr. Yan Ma are specially thanked for discussions regarding calculation of the cosmogenic burial ages. We also thank Jinyu Zhang and Hu Wang for discussions on our preliminary results which greatly improved this manuscript. We are grateful to Dr. Jessica Thompson Jobe and two anonymous reviewers for their helpful comments which led to improvements in the manuscript. This study was financially supported by the National Natural Science Foundation of China (grant no. 41572195) and China Postdoctoral Science Foundation (2020M680620).