A Paleoseismic Record Spanning 2‐Myr Reveals Episodic Late Pliocene Deformation in the Western Qaidam Basin, NE Tibet

The western Qaidam Basin, NE Tibet contains numerous NW‐SE‐trending thrusts that extend over a distance of ∼300 km along the Altyn Tagh Fault and north of the Kunlun Range. However, little is known about the long‐term seismo‐tectonic evolution of this active thrust zone due to the absence of an extended paleoseismic record. We present a 2‐Myr‐long disturbance record established from a core drilled on the crest of a thrust‐cored anticline. Based on detailed sedimentological analysis, the disturbances (micro‐faults, soft‐sediment deformation, slumps, and detachment surfaces) are interpreted as paleoearthquake/tectonic indicators. The core records five seismite clusters which occurred at 3.6‐3.5, 3.4‐3.2, 3.15‐3.1, 3.0‐2.9, and 2.8‐2.75 Ma. This suggests the rate of tectonic strain accommodated by the folds and thrusts in the region varies and thus reveals episodic local deformation. During the clusters, regional deformation is concentrated more in the fold‐and‐thrust system than along regional major strike‐slip faults.

• We interpret micro-faults, softsediment deformation, slumps, and detachment surfaces as paleoearthquake/tectonic indicators • The core records five seismite clusters between 3.6 and 2.7 Ma, revealing episodic thrusting in relation to intense regional deformation • During the clusters, regional deformation was concentrated more in the fold-and-thrust system than along regional major strike-slip faults Supporting Information: • Supporting Information S1 Correspondence to: Y. Lu, Yin.Lu@uibk.ac.at; yinlusedimentology@yeah.net Citation: Lu, Y., Marco, S., Wetzler, N., Fang, X., Alsop

RESEARCH LETTER
Folds that trend with a 30°-45° obliquity to major strike-slip faults also occur in association with the San Andreas Fault (Miller, 1998), the Dead Sea Fault (Freund, 1965), and the North Anatolian Fault (Hubert-Ferrari et al., 2002). Their formation can be explained by the same stresses that drive the strike-slip motion, where the maximum horizontal compressive stress (S Hmax ) is perpendicular to the fold axes. In the case of the Qaidam Basin, the NE-SW S Hmax triggers sinistral motion along both the Altyn Tagh Fault and the Kunlun Fault. This study aims to characterize the seismo-tectonic evolution of an anticline within the Qaidam Basin, which can promote further understanding of strain partitioning in the vast deformation zone related to the Indo-Asia collision.

The Jianshan Anticline and Core SG-1b
The seismic profile across the Jianshan Anticline shows that it is linked to a shallow blind thrust that developed beneath the paleo-Qaidam Lake floor since the Oligocene (∼33 Ma; Figures 1c and 1d), with its deformation rate accelerating since the Miocene (Bally et al., 1986;Liu et al., 2017;Meyer et al., 1998;Wu et al., 2019). During 2011, a 723 m-deep core SG-1b was drilled on the crest of Jianshan Anticline (Figure 1; Text S1). Detailed paleomagnetic dating constrains the age of the generally well-laminated or layered sediments in the core to 7.3-1.6 Ma (Zhang et al., 2014).
The frequent appearance of tempestites, ooid-rich sediments, and undisturbed evaporites in Core SG-1b since 3.6 Ma suggests that the crest of the fold rose above the storm wave base and into a shallow water (few meters-deep) environment since that time, while lacustrine sedimentation did not cease on its crest until 1.6 Ma (Lu et al., 2015(Lu et al., , 2020b. The uninterrupted subaqueous sedimentary sequence (∼33-1.6 Ma) may therefore have continuously recorded effects of late Cenozoic deformation that propagated from the Indo-Asia collision zone to NE Tibet.

Materials, Methods, and Chronology
The studied core section (260-0 m) comprises laminated mud, layered mud, massive mud, mud containing evaporites, and evaporites (Text S2) (Lu et al., 2020b). Paleomagnetic dating constrains the age of the studied core interval to ca. 3.6-1.6 Ma (Zhang et al., 2014). Recently, orbital tuning based on the Rb/Sr record has been conducted on the 183-52 m core interval that is thus more precisely dated at 3.3-2.1 Ma (Kaboth-Bahr et al., 2020) ( Figure S1; Table S1). Calculations of sedimentation rates are based on this age-model.
The archived half of Core SG-1b was scanned with the Avaatech 4th generation XRF Scanner (equipped with CCD-Color Line Scan Camera) at a resolution of 1 cm, an exposure time of 10 s (Text S3) (Kaboth-Bahr et al., 2020). A recent study on the Core SG-1b has shown that strontium (Sr) best reflects Ca-bearing minerals (e.g., carbonate or gypsum) (Kaboth-Bahr et al., 2020). In the present study, we therefore also use Sr to investigate the different facies. For the first time, we identify various types of disturbances by using these high-resolution XRF data and images ( Figure 2). As the instantaneous events are considered to develop immediately below the water-sediment interface, then the timing of each event is constrained by the age of the first overlying undisturbed sediment horizon and is calculated by linear interpolation.
We specifically chose the upper 260 m of Core SG-1b for our case study because it provides a rare opportunity to identify disturbances and understand regional deformations. First, the core was drilled directly on the crest of a thrust-cored anticline whose geological evolution has been constrained by the seismic reflection profile (Lu et al., 2015). Second, the age of the focused core section has been well constrained by both paleomagnetic dating (Zhang et al., 2014) and orbital tuning (Kaboth-Bahr et al., 2020). Third, the lithology in the core interval of 101-56 m is well-understood (Lu et al., 2020b). Thereby, this unique core section allows a better understanding of various sedimentary processes and the effects that different triggers involved, and thus allows reliable identification of sedimentary imprints of both paleoearthquakes/tectonics and non-earthquakes.

Type I: Artificial Disturbances
Drilling disturbances are visible in some cored intervals of layered mud (Figures 2e, 2i, and S2h). The disturbances normally extend along the drill direction (from top to bottom) and are best developed toward the edges of the core.

Type II: Storm-Wave Induced Disturbances
These disturbances comprise evaporite fragments (high content in Sr) floating in fine sands or mud. They are characterized by the sorting of evaporite fragments and lack distinct erosional bases (Figures 2a-2e and S2).

Type III: Micro-Faults
We

Type IV: Soft-Sediment Deformation
These structures are characterized by layer-parallel displacements. All these horizons are observed within core intervals of layered mud and laminations and are capped by overlying layered mud or laminations (Figures 2j and 2k and S4). We identify 34 such disturbances with thicknesses ranging between 3 and 33 cm.

Type V: Slumps
Slumps are characterized by deformed mud-containing evaporites (high content in Sr), deformed laminations, or folding of mud layers (Figures 2n and S5). Erosive or bedding plane detachment surfaces are commonly observed at their base, and thus differentiate slumps from in situ soft-sediment deformation described above. We record 41 slump horizons with thicknesses ranging from 3 to 91 cm.

Type VI: Detachment Surfaces
The structures are characterized by sharp contacts (Figures 2l, 2m, and S6). We identify 10 detachment surfaces in the core.

Distribution of Type III-VI Disturbances and Sedimentation Rates During 3.6-1.6 Ma
Micro-faults, soft-sediment deformation, and detachment surfaces mainly occur within the core intervals of laminated and layered mud ( Figure 3). Slumps however are more commonly preserved within core intervals of massive mud and layered mud. At an orbital-scale, sedimentation rates based on the core are high during 3.6-2.7 Ma (with a mean of 16.5 cm/Kyr), low during 2.1-1.6 Ma (mean: 10.7 cm/Kyr), but are very low during 2.7-1.6 Ma (mean: 7.5 cm/Kyr). We identify a total of 164 such disturbances in the studied section, 126 of them (77%) occurred between 3.6 and 2.7 Ma (255-112 m) (Figure 3).

Type III: Micro-Faults
Micro-faults are a consequence of brittle deformation caused by a high strain rate (Seilacher, 1969). The micro-faults in Core SG-1b do not result from gravitational loading because (i) they commonly cut through coarse-grained laminations/beds of higher densities (Figures 2f and 2i), (ii) no micro-faults are observed immediately below thick coarse-grained layers, (iii) most micro-faults occur within the cored intervals of laminations and layered mud (Figures 3 and S3) which have a stable vertical density structure and only a small vertical density contrast. In addition, the drilling stress which acted on these unconsolidated soft-sediments during their recovery cannot account for these brittle structures as they are not confined to the edge of the core.
Syn-depositional single micro-faults have been identified on a centimeter-scale from lake sediments in Anatolia (Avşar et al., 2016) and the Alps (Monecke et al., 2004) where they have been correlated to modern or historic earthquakes, and thereby interpreted as seismites. In addition, fault-graded beds have been interpreted as seismites caused by strong earthquake-shaking on gradationally compacted muds in quiet water basins (Seilacher, 1969). The observed micro-faults in Core SG-1b are compatible with these well-understood examples and are therefore also considered to be indicators of paleoearthquakes.

Type IV: Soft-Sediment Deformation
Gravitational instability (Rayleigh-Taylor instability), the mechanism for overloading, is one of the most common driving forces for soft-sediment deformations in subaqueous environments (Owen et al., 2011). In Core SG-1b, soft-sediment deformation underlies layered mud or laminations (Figures 2 and S4). No such deformation is developed immediately beneath thick coarse-grained layers with higher density. The laminated and layered sediments in Core SG-1b consist of stably stratified water-saturated mud in which the density increases with depth, and are thus not susceptible to Rayleigh-Taylor Instability which requires inverted densities. Another common driving force for soft-sediment deformations in subaqueous environments is storm-waves (Owen et al., 2011). Typical deformation created by storm-waves include load-casts which are commonly associated with tempestites (Molina et al., 1998). In Core SG-1b, the 34 deformed horizons show no association with tempestites. Besides, none of the deformed horizons show a correlation with ooid layers (with high content in Sr) which would indicate that the depositional environment is under the strong influence of waves (Figure 2e). We therefore discard storm-waves as the primary trigger for the 34 deformed horizons. In addition, we suggest that the drilling process cannot account for these deformations because artificial stress mainly acts on the edges of the core. It cannot induce layer-parallel displacements and deformations in the center of the event layer which is constrained by laminations/layered mud at the top and bottom (Figures 2  and S4).
The observed deformation is similar to other soft-sediment structures in lakes from tectonically active regions such as Anatolia (Avşar et al., 2016), South-central Chile (Moernaut et al., 2014), California (Sims, 1973), and the Dead Sea (Ken-Tor et al., 2001). Soft-sediment deformation from these regions has been correlated with instrumental or historic earthquakes with magnitudes ranging from M w 5.3 to 9.6 and local shaking intensities from >VI to IX, and interpreted as seismites. Previous studies reveal that such seismic-induced deformation occurred immediately below the water-sediment interface (Avşar et al., 2016;Lu et al., 2021;Sims, 1973). There is a lack of erosional bases developed below the deformed layers that would allow us to infer a slump source for these features, and they are therefore considered to be formed in situ. The earthquake-forced shear known as the Kelvin-Helmholtz instability is a plausible driver for the layer-parallel shear during soft-sediment deformations (Heifetz et al., 2005;Lu et al., 2020a;Wetzler et al., 2010). We interpret the observed soft-sediment deformations in Core SG-1b as reflecting seismically triggered shear instability, which can therefore be used as a paleoearthquake indicator.

Type V: Slumps
The common presence of evaporites in slumps makes these layers distinct from the overlying and underlying laminated or layered mud, and implies a shallow water source for the slumps. However, the lack of coarse sands in the slumps, and the high relief of the Jianshan Anticline on the lake floor, do not support a distant lakeshore origin for the evaporites (the regions in front of the Altyn Tagh Range). One viable source for the evaporites is the higher elevations created around the crest of the Jianshan Anticline, where water depth would be shallower and evaporite deposition would increase.
The absence of coarse terrigenous sediment supply, low sedimentation rate, and extremely gentle slope angles (<1°) at the crest of the Jianshan Anticline make non-seismic triggering mechanisms, such as sediment loading or gravitational gliding, highly unlikely. The most plausible triggers for slumps in such a setting (on the crest of an anticline) are earthquakes. A negligible slope angle (<1°) is sufficient for earthquake-triggered subaqueous slope failures (Alsop & Marco, 2013;Field et al., 1982). Slumps triggered by historic earthquakes with magnitudes ranging from M w 5.9 to 7.5, have been reported from lacustrine environments in the Alpine range (Schnellmann et al., 2002) and British Columbia, Canada (Shilts & Clague, 1992), and interpreted as seismites.

Type VI: Detachment Surfaces
Detachment surfaces in drill cores have been used previously in tectonically active regions to indicate local deformation (Greene et al., 1994;Guo et al., 2005) and paleoearthquakes . The coring site is located close to the top of the Jianshan Anticline and we here interpret detachment surfaces preserved in the core as failure surfaces formed toward the head scarp of seismic-induced slumps that translated downslope from the anticlinal crest. Similar gravity-driven bedding-parallel detachment surfaces generated by the downslope-directed failure of sediments have been recently recorded in outcrops from the Dead Sea (Alsop et al., 2020).

Implications for Paleoearthquake Recurrence Pattern
We interpret micro-faults, soft-sediment deformation, slumps, and detachment surfaces (Type III-VI disturbances) in Core SG-1b as seismites. The entire seismite catalog that spanned 3.6-1.6 Ma comprises 164 such events, and yields a mean event return time of 12.0 Kyr, a standard deviation (STD) of 24.9 Kyr, and a coefficient of variation (COV: STD divided by mean) of 2.1. The low coring rate during 2.1-1.6 Ma results in an incomplete part of the seismite record (Figure 3). The relatively low sedimentation rate during 2.7-2.1 Ma might have rendered the sediments less sensitive to seismic shaking, and thus fewer earthquakes were recorded (Figure 4).
LU ET AL.
In contrast, the seismite record during 3.6-2.7 Ma (N = 126) yields a mean recurrence time of 6.8 Kyr and a COV of 1.3. The recurrence times follow a power-law distribution (Figure 4c). The large COV and the power-law distribution of recurrence times suggest the events are clustered in time during 3.6-2.7 Ma. The power-law distribution of recurrence times and the large mean recurrence time (6.8 Kyr) also support the links that we propose between Type III-VI disturbances and paleoearthquakes. This is because the occurrence of non-earthquake triggers and their triggered events are normally more periodic and frequent.
To identify earthquake temporal clusters in the catalog, we compute the seismicity rate using a moving time window of 50 Kyr (≈2*STD), with an overlap of 50%. The calculated seismic rate presents five seismite clusters that are distinguished above the seismicity rate threshold of 0.1 (≈2/STD) (Figure 4b). The five clusters occurred at 3.6-3.5, 3.4-3.2, 3.15-3.1, 3.0-2.9, and 2.8-2.75 Ma, and yield a mean recurrence of 4.4 ± 0.9 (standard error of the mean) Kyr, 4.0 ± 0.7 Kyr, 5.9 ± 2.0 Kyr, 7.8 ± 3.0 Kyr, and 4.2 ± 1.2 Kyr, respectively. Also, we should note that some short time periods with very low sedimentation rates or hiatuses might exist since the drilling site is on the crest of an anticline. We cannot rule out the possibility that some seismites were missed during the quiet periods between the clusters, due to potentially very low sedimentation rates over shorter time periods, or the sediments being less sensitive to seismic shaking.

Implications for Regional Tectonism
Available Holocene paleoseismic records based on fault trenching from the nearby Xorkoli section (Yuan et al., 2018) and Aksay section  of the Altyn Tagh Fault indicate a quasi-periodic rupture behavior along the fault (Figures 1b and 1c). The mean recurrence of major earthquakes on these sections are 0.6 Kyr during 6-0 ka and 1.4 Kyr during 8-0 ka, respectively Yuan et al., 2018). In addition, previous paleoseismology studies based on terraces indicate that the mean recurrence of M w 8 earthquakes on the western part of the east Kunlun Fault (≥300 km from Core SG-1b site; Figure 1b) is about 0.9 Kyr during 8-2 ka (Van Der Woerd et al., 2002). Previous studies have also indicated essentially stable deformation rates (∼9 mm/yr) in the region along the Altyn Tagh Fault over million years (Yin et al., 2002). These estimated rates are remarkably similar to those established from GPS measurements (Bendick et al., 2000;Chen et al., 2000). So, the Altyn Tagh Fault in this region is more likely to have a quasi-periodic rupture behavior during the past few million years. A mean recurrence of < 0.6 Kyr is expected if the first-order Altyn Tagh Fault and Kunlun Fault are the major sources of recorded paleoearthquakes in Core SG-1b.
However, the smallest mean recurrence of earthquakes of the recorded five clusters in Core SG-1b is 4.0 Kyr, much larger than 1 Kyr. Besides, the extremely high COV value of the SG-1b seismite record suggests that the drilling site was affected by the geometrically complex thrust system that surrounds the anticline. These features imply that the local structures, for example, the Jianshan Anticline and nearby folds (Figure 1c), rather than the Altyn Tagh Fault and/or Kunlun Fault are the major sources. These folds have developed above shallow thrusts beneath the lake floor, with an axial length of 20-50 km, which are capable of triggering moderate earthquakes (M w ≥ 5.0).
The paleoseismic record preserved in Core SG-1b records five seismite clusters. The clustered behavior suggests that the rate of tectonic strain accommodated by the Qaidam thrust system varies over time and thereby reveals episodic fold-and thrust deformation in the region. However, previous studies imply the Altyn Tagh Fault in this region is more likely to have a quasi-periodic rupture behavior over the past few million years Yin et al., 2002;Yuan et al., 2018). We therefore propose that during the seismite clusters, more deformation is concentrated in the fold-and-thrust system, however, during the intervening quiescent periods, deformation is focused more along the Altyn Tagh and Kunlun strike-slip faults. During both the seismite clusters and intervening quiescent periods, both systems might have accommodated the same constant far-field tectonic loading induced by the collision.
Similar research has been conducted in the northern Bighorn Basin, USA where Jackson et al. (2019) interpreted the Upper Cretaceous soft-sediment deformations (in the form of clastic dikes) as seismites, and used these to understand the spatial-temporal development of Laramide deformation. Our results from the NE Tibetan Plateau highlight the potential of long-term seismite sequences to detect tectonism and contribute to our understanding of the long-term seismo-tectonic evolution of the active thrust zone. Our innovative sedimentological method to detect regional tectonics may also prove suitable for similar geological settings (tectonically active regions with continuous lacustrine/marine deposits) elsewhere in the world.

Data Availability Statement
Data sets are available in the supporting information and at PANGAEA database (https://doi.org/10.1594/ PANGAEA.922852).