The Qaidam Basin, especially its western part, is characterized by numerous northwest-trending folds and faults. Understanding the style and formation mechanism of these structures is crucial for unravelling the deformation of the Qaidam Basin and the broader tectonics of the northern part of the Tibetan Plateau. Based on seismic data, we reconstruct the structural framework of the central part of the western Qaidam Basin, and find that: (1) the structures in this area display positive flower geometries in two-dimensional (2D) seismic profiles and helicoidal shapes in 3D space; (2) these positive flower structures began to develop during the middle to late Miocene (15–8 Ma); and (3) these positive flower structures are controlled by left-lateral basement faults and formed in a random temporal sequence. The left-lateral features represent the strike-slip component of distributed deformation, consistent with published global positioning system and seismicity data for oblique convergence across the north of the Tibetan Plateau. Collectively, they perform the same role as the discrete Haiyuan, Kunlun, and Altyn Tagh faults in adjacent areas.

The formation and deformation of the Qaidam Basin and the uplift and growth of the Tibetan Plateau were due to the collision of India with Eurasia (Argand, 1924; Molnar and Tapponnier, 1975) during the past ∼50 m.y. (Patriat and Achache, 1984; Garzanti et al., 1987; Rowley, 1996; Huang et al., 2015). However, in contrast to the surrounding mountains in the northeastern Tibetan Plateau (Fig. 1; Dupont-Nivet et al., 2004; Duvall and Clark, 2010; C. Wang et al., 2011; Zheng et al., 2013), the Qaidam Basin is known to have a relatively rigid, possibly cratonic basement (Zhu and Helmberger, 1998; Shen et al., 2001; Yuan et al., 2013; Yu et al., 2014). How the Qaidam Basin deforms in the context of the higher regions of the Tibetan Plateau has been of interest to many (e.g., England and Molnar, 1997; Royden et al., 1997; Dayem et al., 2009).

Some have emphasized north-south compression and shortening of the Qaidam Basin, and tend to interpret the Cenozoic structures within the basin as thrust fault–related folds (Métivier et al., 1998; Chen et al., 1999; Yin et al., 2007, 2008a; Meng and Fang, 2008). The strike-slip component to the deformation, if any, is less clear. The active deformation across the entire northeastern Tibetan Plateau involves oblique convergence: global positioning system (GPS) vectors with respect to stable Eurasia are oriented oblique to the regional fault and fold traces (e.g., Gan et al., 2007; Fig. 1). However, while strain partitioning (i.e., the spatial separation of deformation caused by oblique convergence into dip-slip and strike-slip components) to the north and south of the Qaidam Basin takes place via strike-slip localization on single faults (Haiyuan and eastern Kunlun), it is not clear what happens within the basin interior, and how the oblique convergence (transpression) is actually expressed. As an additional factor in the regional tectonics, Meyer et al. (1998) and Pan et al. (2015) suggested that structures within the Qaidam Basin are splays from the Altyn Tagh fault, which can extend as much as 400 km away.

In this paper we analyze two-dimensional (2D) and 3D seismic data over a series of anticlines in the western part of the Qaidam Basin to investigate the subsurface structural framework of this region. We propose a new interpretation of the structural styles in the western Qaidam Basin, with broader implications for the deformation mechanism and tectonic evolution of this part of the Tibetan Plateau.

The Qaidam Basin has a sedimentary cover of ∼120,000 km2, and is the largest Cenozoic intermontane basin in the northeastern Tibetan Plateau; it is bound by the Qimen Tagh–eastern Kunlun mountain belts to the south, the Altyn Tagh mountain belts to the northwest, and the Qilian Shan–Nan Shan mountain belts to the north (Fig. 1; Métivier et al., 1998; Zhai et al., 2002; Wang et al., 2006). The relatively low elevation (∼3000 m) and limited deformation of the Qaidam Basin, compared to the strongly deformed surrounding mountain belts, is attributed to the higher strength of the underlying crust relative to the surrounding regions. Geophysical explorations (Zhu et al., 1995; Jordan and Watts, 2005; Zhao et al., 2006, 2013; Li et al., 2013) and geological investigations (Zhai et al., 2002; Wang et al., 2004; Hao et al., 2004; Yu et al., 2014) have demonstrated that the Qaidam Basin has a Precambrian crystalline basement with high effective elastic thickness. The integrated chronostratigraphic, lithostratigraphic, and seismic stratigraphic framework of the Qaidam Basin has benefitted from hydrocarbon exploration since 1954 (Fig. 2B; Sun et al., 2007). The Qaidam Basin is mainly filled with Cenozoic nonmarine deposits except in some places along the Qilian Shan–Nan Shan and the Altyn Tagh, which are underlain by Mesozoic sequences (Jurassic–lower Cretaceous; Ritts and Biffi, 2000; Jin et al., 2004; Xu et al., 2006). The Cenozoic strata of the Qaidam Basin are divided into eight lithostratigraphic units, each of which has been dated precisely based on paleontology and magnetostratigraphy studies (Gu et al., 1990; Qinghai Bureau of Geology and Mineral Resources, 1991; Yang et al., 1992; Sun et al., 2002; Sun et al., 2005; Fang et al., 2007; Lu and Xiong, 2009; Pei et al., 2009; Zhang et al., 2013; Ke et al., 2013). They are (1) the Lulehe Formation (E1+2l), pre–53.5 to 43.8 Ma; (2) the Lower Xiaganchaigou Formation (E31xg), 43.8–37.8 Ma; (3) the Upper Xiaganchaigou Formation (E32xg), 37.8–35.5 Ma; (4) the Shangganchaigou Formation (N1sg), 35.5–22.0 Ma; (5) the Xiayoushashan Formation (N21xy), 22.0–15.3 Ma; (6) the Shangyoushashan Formation (N22sy), 15.3−8.1 Ma; (7) the Shizigou Formation (N23s), 8.1–2.5 Ma; and (8) the Qigequan Formation (Q1q), 2.5–0.01 Ma (see Fig. 2B). The crustal deformation of the Qaidam Basin is mainly concentrated in its western part (Zhou et al., 2006; Yin et al., 2008b), where numerous northwest-trending faults and folds are oriented at high angles to the Altyn Tagh fault (Fig. 1; Sun et al., 1956; Song and Wang, 1993; Ge et al., 1998; Yin and Harrison, 2000). The study area in this paper is located in the central part of the western Qaidam Basin, between the Yingxiongling structure belts (to the south) and the Eboliang structure belts (to the north). Both the geological map (Fig. 2A) and the digital elevation model (Fig. 3) show a series of periclines (length:width ratios ≤3 to ≤10) in the study area, arranged in en echelon patterns. These anticlines are located far from the basin boundaries and display slightly sigmoidal traces. They are the key to understanding the structural deformation pattern within the western Qaidam Basin. The exposed strata in the study area range from Eocene to Holocene in age (Fig. 2A).

The analyses of the structural style and the reconstruction of the subsurface fault system within the study area are primarily based on seismic data provided by the Qinghai Oilfield Company (China). We selected 41 2D seismic profiles and 2 3D seismic data sets that cover, with an area of ∼4200 km2, most anticlines in the central region of the western Qaidam Basin (Fig. 2A). The original seismic data, pre-stack time-migration SEG-Y (Society of Exploration Geophysicists) files, were integrated into and interpreted by the commercial software, Kingdom (HIS). The 2D seismic profiles consist of 39 inlines (southwest-northeast) and 3 crosslines (northwest-southeast) with maximum recording times of 6 s or 7 s two-way traveltime (TWT) for different batches of profiles (from the 1980s to the 2000s). The TWT extends down to 6 s and 5 s for the 2 3D seismic data sets (Xiaoliangshan in 2011 and Nanyishan in 2008, respectively). The average inline spacing is 2 km for both the 2D seismic profiles and 3D seismic data sets, which is tight enough to detect subtle structural changes along the strike of a single anticline. The three relatively long 2D seismic crosslines were linked up one by one, crossing all the 2D seismic inlines and the two 3D seismic data sets, and were mainly used for closing the same seismic reflection horizons and faults.

The seismic reflection boundaries between the sedimentary cover and basement, between the Cenozoic and the Mesozoic, and between the adjacent Cenozoic formations are recognized throughout the entire basin and are termed T6, TR, T5, T4, T3, T2′, T2, T1, and T0 (from bottom to top) (Fig. 2B; Xu et al., 2004). In our work, the calibration of each seismic stratigraphic horizon (T0–T6) was constrained precisely by synthetic seismograms and substantiated by borehole records, followed by lateral comparison and tracking. The depiction of the faults was strictly based on the offset of seismic events in the sedimentary cover and was inferred in the basement. However, the downdip continuation of faults kept a uniform pattern from the shallow sedimentary layers to the deep basement. After interpreting all the selected seismic sections (e.g., Fig. 4), we delineated the reliable 3D geometry of the underground structures (e.g., Fig. 3B). Furthermore, the fault intersection points on each seismic stratigraphic horizon were projected to the surface (digital elevation model), making up the fault system distribution map. Planimetric positions of the shallowest piercing points of the faults with significant vertical offsets were compared with the topography (morphological scarps) (Fig. 3A).

Here we present seismic reflection data and other evidence (geomorphology and depth-structure maps) used to reconstruct the structural framework of the study area.

Geometry of the Subsurface Faults and Anticlines

The major structural elements depicted in the seismic sections crossing the study area (southwest-northeast) are a series of positive flower structures, with growth strata ranging from middle to late Miocene (Fig. 4). Although the general shapes of the flower structures are similar through all inline sections, comparison of these sections shows significant differences with respect to number, arrangement, and preferred dip directions of branch faults within each of the flower structure (see following). Observed differences also include the occurrence of syntectonic strata and the depth of branch lines of the major faults.

The southwest-vergent Xiaoliangshan anticline in section A-A′ (Fig. 4A) is bound by two steep faults zones (>60°) and has a curved crest. Between the two boundary faults zones, there are several minor branch faults that also help to form the uplifted core of the Xiaoliangshan anticline. The gentle limbs of the Xiaoliangshan anticline display a wide interlimb angle (>120°). The almost flat strata outside the two main branch faults indicate that folding deformation is confined by the flower structure with a width of ∼7 km. The two wide boundary fault zones are characterized by chaotic seismic reflections and fault plane reflections. They intersect near the base of the sedimentary cover sequence and combine into one major southwest-dipping fault extending down into the deeper basement. The trace and dip of the basement fault are inferred from the differential uplift of the fault blocks and the opposite dip directions of the seismic reflections in the basement on either side of the major fault, respectively. The T1 seismic reflector (ca. 8.1 Ma) is the boundary between the pre-growth strata and the growth strata of the Xiaoliangshan anticline, below which the thickness of each stratigraphic unit keeps constant across the flower structure, and above which the strata thicken toward the anticline flanks and the outside of the boundary faults.

In section B-B′ (Fig. 4B), the Nanyishan anticline is a symmetrical, arcuate fold with a wide interlimb angle (>120°) that is also characterized by a positive flower structure with a width of ∼10 km. A detachment zone (at ∼3.3s TWT) composed of dark mudstones in the middle Eocene Lower Xiaganchaigou Formation (Wang, 2003) separates the major fault below from the branch faults above. The steep northeast-dipping major fault (>65°) extends from the basement up to the detachment zone, and can be traced in the same way used in the Xiaoliangshan anticline. Above the detachment unit, numerous small branch faults define two wide boundary fault zones (>55°), while few branch faults develop between these two fault zones. Growth strata in the Nanyishan anticline occur above the T1 seismic reflector. The southeastern end of the Xiaoliangshan anticline, as shown in section B-B′, is defined by only two single branch faults that reach the T2′ seismic reflector with minor fault throw and fold deformation.

In section C-C′ (Fig. 4C), the limbs of the Jiandingshan anticline are asymmetric, showing a northeast vergence, with a half-wavelength of ∼11 km. The steep basement-involved major fault branches upward into three narrow faults (>65°) bounding the anticline, with the largest throw along the northeastern boundary. The branch points of the Jiandingshan flower structure are at greater depths, far below the T6 seismic reflector, compared to those of the Xiaoliangshan and the Nanyishan flower structures. The strata above T2′ seismic reflector (ca. 15.3 Ma) in this section show an obvious trend of thickening from the anticline high point to the flanks, and represent the growth strata of the Jiandingshan anticline. An anomalous east-west–striking, north-northeast–dipping fault to the south does not seem to be related to the Jiandingshan flower structure (Fig. 3A).

Section D-D′ (Fig. 4D) crosses the Dafengshan, the Heiliangzi, and the Jianbei anticlines. The western segment of the Dafengshan anticline has a simple geometry, with an upright, gentle crest confined by two boundary faults (>65°). These two main branch faults may intersect with each other at a deeper level than 6 s TWT. The growth strata of the Dafengshan anticline occur above the T2′ seismic reflector. The Heiliangzi anticline in this section behaves as an asymmetric positive flower structure with a northeastward throw, due to a significant reverse-slip component along its northeastern boundary branch fault (>65°). It succeeds the geometry of the Jiandingshan anticline to the southeast (Fig. 2). Growth strata of the Heiliangzi anticline occur after the formation of boundary T1. The Jianbei anticline (T2′), located on the north of the Heiliangzi anticline, has a geometry similar to that of the Heiliangzi anticlines, and its growth strata occur above the T2′ seismic reflector.

Section E-E′ (Fig. 4E) crosses the Dafengshan, the Heiliangzi, and the Changweiliang anticlines. The Changweiliang anticline is the southeastern continuation of the Jianbei anticline (Fig. 2). Compared with section D-D′, the three anticlines in section E-E′ are more upright and symmetric, while the fault system of the Dafengshan flower structure has more branch faults, and the northeastern boundary faults of the Heiliangzi and the Changweiliang flower structures become less important while the southwestern boundary faults have a great impact on their geometry. The branch lines of the three flower structures in section E-E′ occur near the T6 seismic reflector and are possibly influenced by Mesozoic strata with half-graben sedimentary geometry. Section F-F′ (Fig. 4F) shows the geometry of the southeastern segment of the Dafengshan anticline and the middle segment of the Jianshan anticline. In this section, the positive flower geometry of these two anticlines is clear; one is asymmetric and tilts toward the southwest and the other is symmetric. The growth strata of the Jianshan anticline developed above the T2′ seismic reflector.

In addition to the differences among the anticlines described here, the subsurface geometry of each anticline changes gradually along strike. For example, the Dafengshan anticline shows a ribbon effect in the 3D space (Fig. 3B; Zolnai, 1991). The western part of the Dafengshan anticline is asymmetric and verges to the northeast, and the major fault in the basement dips to the southwest (e.g., seismic lines I, II, and III in Fig. 3B). The middle part of the Dafengshan anticline turns symmetric and the major fault cuts into the basement almost vertically (e.g., seismic lines IV and V in Fig. 3B). The eastern part of the Dafengshan anticline becomes asymmetric again, but the dip directions of the fold axial plane and the major basement fault are of opposite polarity to those in the western part (e.g., seismic lines VI and VII in Fig. 3B). The fold amplitude, the number of branch faults, the offset of each branch fault, and the depth of branch lines at the western and eastern ends of the Dafengshan anticline are gentler, fewer, smaller, and deeper than those at the middle part of the Dafengshan anticline, respectively. These changes reveal the differential deformation magnitude along the strike of the Dafengshan anticline.

Distribution of the Faults and Anticlines in Map View

The fault traces mapped on the T1, T2′, T2, T4, and T6 seismic stratigraphic horizons were projected to the surface. These fault traces are subparallel to the curved traces of the anticline boundaries, but in detail cross the anticline axes at very low angles (<5°) (Fig. 3A). The anticline axes have a slightly sigmoidal appearance in plan view, whereas the fault traces are straighter and more segmented. The dominant set of faults has an average strike of N130°E, and the minor set has only four short faults with an average strike of N95°E (Fig. 3A, inset). These minor faults are P shears to the dominant set. The dominant faults form the branch faults of the positive flower structures. Some of them converge at one place and diverge at another place along the strike (e.g., the Xiaoliangshan and the Dafengshan anticlines), and adjacent branch faults of one flower structure may propagate displacement sideways (e.g., the Jiandingshan, Heiliangzi, and Jianbei anticlines) (Fig. 3A).

The morphological scarps on each side of the anticlines are the surface expressions of the boundary branch faults, and their development is also affected by the northwesterly wind erosion in the Qaidam Basin (Kapp et al., 2011; Wu et al., 2014a). The largest scarps coincide with the underlying faults, which exhibit considerable vertical offsets and are preserved on the leeward side of the anticlines, such as the southern boundary scarps at the middle segments of the Xiaoliangshan and the Nanyishan anticlines (Fig. 3A). A typical horsetail feature controlled by near-surface splay faults can also be found at the eastern end of the Dafengshan anticline (Fig. 3A).

Sinistral Faulting along the Structures

Depth-structure maps are basic tools in hydrocarbon exploration because they play a useful role in deploying wells and calculating closures and areas of traps. Such depth-structure maps can also contain useful information about the geometry and causative kinematic processes of folds (Rickard, 1971; Shaw et al., 1994). Based on plentiful seismic profiles and dense time-depth conversion data, depth-structure maps of every seismic horizon (T0–T6) in the western Qaidam Basin were compiled by the geoscientists of Qinghai Oilfield Company.

The T2 depth-structure maps of the Jiandingshan, Nanyishan, and Dafengshan anticlines are shown as examples (Fig. 5). The 1600 m depth contours are chosen as a reference in the Jiandingshan area (Fig. 5A). The footwall of the northern Jiandingshan fault (NJF) is offset to the northwest relative to the hanging wall, which indicates a sinistral strike-slip component on the fault. Sinistral movements are also observed on the northern Nanyishan fault (NNF in Fig. 5B) and the southern Dafengshan fault (SDF in Fig. 5C). Using gradients of the dipping strata and distances between the end points of the reference contour lines from the depth-structure maps, the vertical and lateral offsets are calculated for these three steep faults (Table 1; Figs. 4B, 4C, 4E; Maltman, 1998). The results reveal that the vertical offsets are much smaller than the sinistral offsets; the former range from 0.2 to 0.6 km, while the latter range from 1.0 to 3.5 km.

Is the Western Qaidam Basin Dominated by Contractional or Transpressive Structures?

At early stages of hydrocarbon exploration in the Qaidam Basin, the anticlines in the western part of the basin were identified as a series of uplifted folds, each of which was thought to be bounded by two oppositely dipping reverse faults (Gu et al., 1990). This interpretation was based on undulations and offsets of the seismic reflectors without much consideration of the formation mechanism. Xia et al. (2001) and Zhai et al. (2002) followed this interpretation scheme and proposed that the boundary faults are inverted normal faults. However, structures with normal faults are primarily located in Jurassic strata along the northern margin of the Qaidam Basin and the Altyn Tagh piedmont (Zeng et al., 2002; Wu et al., 2006; Fu et al., 2015), with rare stratigraphic evidence for Paleogene normal faults (Jin et al., 2004; Yin et al., 2008b). The only evidence for Jurassic normal faulting in the study area is to the north of the eastern segment of the Dafengshan anticline (Fig. 4, sections E-E′ and F-F′).

An alternative explanation (e.g., Zhou et al., 2006; Yin et al., 2008b; Wu et al., 2014b) follows a fold-thrust belt model, interpreting the structures in the western Qaidam Basin as Jura-type folds that develop in the distal region of a foreland thrust belt (Wang et al., 2012; Yu et al., 2016). Structural styles in this tectonic model include fault-propagation folds, fault-bend folds, detachment folds, conjugate kink-band zones, and structural wedges (Zheng et al., 2007; Liu et al., 2009; Xu et al., 2013; Wu et al., 2014a). However, this interpretation scheme is challenged by the seismic data presented here for the following reasons: (1) the boundary faults of most anticlines in the western Qaidam Basin are much steeper (>55°; Fig. 4) than the theoretical thrust faults under horizontal compressive stress; (2) the folded strata are confined between the boundary faults of each anticline and generate large interlimb angles (Figs. 3 and 4); and (3) the southern and northern tectonic boundaries of the Qaidam Basin are dominated by high-angle basement steps with considerable lateral displacements, making the development of a single ubiquitous décollement unlikely (Wei et al., 2005; Wang et al., 2008; Cheng et al., 2014; X. Cheng et al., 2015).

In this study we have identified several anticlines in the western Qaidam Basin that are linked to positive flower structures, with branch faults that consistently converge at the tops of the principle displacement zones of the major faults (Fig. 4; Beidinger and Decker, 2011). The surface appearances of these helicoidal flower structures are laterally propagating boundary faults and S-shaped fold traces, which indicate sinistral strike-slip motion along the major basement faults (Figs. 3A, 3B). The left-lateral displacement magnitudes of these faults can be determined by the depth-structure maps of anticlines (e.g., Table 1; Fig. 5). The kinematic characteristics of surface minor fractures around the anticlines in the study area also indicate left-lateral transpressive deformation (e.g., Fig. 6), which means an opposite slip direction on the same faults to that deduced by Mao et al. (2016). Thus, we propose that the dominant structures in the western Qaidam Basin are a series of discrete positive flower structures controlled by left-lateral transpressive faults that root downward into the basement (Sylvester, 1988; Harding, 1990; Woodcock and Rickards, 2003) (Fig. 7A). We do not exclude the possibility that detachment movements of different magnitudes may take place in the ductile sedimentary layers of some structures in this region, such as the Nanyinshan and Shizigou-Youshashan anticlines (Wang, 2003; Yu et al., 2011; Wu et al., 2014a).

Timing of Deformation

Cenozoic sedimentation and deformation in the Qaidam Basin initiated within 10 m.y. of the initial Indo-Eurasia collision (Yin et al., 2002; Yin et al., 2008a), indicating that the far-field effects of continental collision were rapidly transferred to the northern part of the Tibetan Plateau. The resultant uplift of the Qilian Shan–Nan Shan range and sinistral strike-slip movement along the Altyn Tagh fault configured the initial northern and northwestern boundaries of the Qaidam Basin, respectively (Yin et al., 2002; Zhou et al., 2006; Yin et al. 2008a; Clark et al., 2010). Deformation in the Qaidam Basin and surrounding ranges accelerated in the early to middle Miocene (ca. 20–15 Ma) (F. Chang et al., 2015; Yuan et al., 2013), accompanied by the uplift and sinistral movement of the Qimen Tagh–eastern Kunlun range (Jolivet et al., 2003; Duvall et al., 2013) and the accelerated strike-slip motion along the Altyn Tagh fault (Wu et al., 2012; F. Cheng et al., 2015). Growth strata from the seismic profiles show synsedimentary deformation in the study area initiating in the middle to late Miocene (15–8 Ma). Specifically, the Dafengshan, Jiandingshan, Jianbei, Changweiliang, and Jianshan anticlines began to develop at the start of Shangyoushashan Formation (N22sy, 15.3 Ma) deposition, and the Nanyishan, Xiaoliangshan, and Heiliangzi anticlines began to develop at the start of Shizigou Formation (N23s, 8.1 Ma) deposition (Fig. 4).

It is notable that no clear temporal pattern of fold growth is seen in the study area of 70 km width (a-a′ in Fig. 1). This calls into question the speculations by Métivier et al. (1998), Yin et al. (2007, 2008b), Wu et al. (2013), and Wu et al. (2014b), who described the northward or southward advance of deformation in the Qaidam Basin. Furthermore, expanding the reference range southward to the Kunbei fault system and northward to the Lenghu structural belts (a distance of ∼260 km; b-b′ in Fig. 1), the initial times of growth of northwest-trending structures are concentrated in the Miocene (ca. 20–8 Ma). From south to north, the onset of the Kunbei fault system was during the early Miocene (Cheng et al., 2014; X. Cheng et al., 2015), the onset of the Yingxiongling structure belts was during the middle Miocene (Yu et al., 2011), the onset of the structures in our study area was during the middle Miocene, the onset of the Eboliang structure belts was during the early-middle Miocene (Fu et al., 2009; Sun et al., 2014), and the onset of the Lenghu structure belts was during the middle Miocene (G. Wang et al., 2011). Therefore, we infer that most northwest-trending structures in the western Qaidam Basin have formed out of sequence since the early Miocene, although deformation at the southern and northern margins of the Qaidam Basin commenced somewhat earlier than in its interior parts.

Regional Tectonic Implications

This study identifies the subsurface structures of individual folds in the western Qaidam Basin as a series of sinistral positive flower structures (Fig. 7A). Clearly, these structures contribute to accommodating the convergence of the broad India-Eurasia collision zone.

One implication of our work is that the segmented, domal, and discontinuous nature of the folds and underlying faults in the western Qaidam Basin does not support any direct connection between these structures and the Altyn Tagh fault, as previously proposed (Meyer et al., 1998).

The distribution of earthquake focal mechanisms (Molnar and Lyon-Caen, 1989; Elliott et al., 2010; Global Centroid Moment Tensor catalog,; Fig. 1) shows that seismic strain is not evenly distributed across the Qaidam Basin and adjacent mountain ranges. Much of the basin interior has little or no instrumental record of M > 5 earthquakes, contrasting with numerous thrust events at the basin margins. Strike-slip events are concentrated along the known strike-slip faults, including the Altyn Tagh, Haiyuan, eastern Kunlun, and Elashan faults. The orientation of the fold axes (Figs. 1 and 3A) and the sinistral slip sense of the structures (Figs. 5 and 6) within the Qaidam Basin are consistent with the oblique convergence recorded by GPS data across the basin and the mountain ranges to its north and south (Fig. 1; e.g., Gan et al., 2007; Liang et al., 2013). Whereas in those marginal ranges the strike-slip component of deformation is localized along single large sinistral strike-slip faults (eastern Kunlun and Haiyuan faults; i.e., strain partitioning, Fig. 7B), the oblique convergence within the western Qaidam Basin is distributed across the positive flower structures (Fig. 7A).

The differing deformation styles between the Qaidam Basin and the surrounding ranges may relate to the variable basement strength and the resultant degree of shortening. The relatively rigid Precambrian basement of the Qaidam Basin has plausibly resisted Cenozoic deformation more successfully than the surrounding mountain ranges, which were the loci of intense deformation during Paleozoic orogenies. We suggest that strain partitioning is more complete in these mountain ranges than that in the relatively stable, lower strain regions of the Qaidam Basin.

A further question is why deformation is oblique across this part of the collision zone at all. A simple explanation is that there is at least partial extrusion of Eurasian lithosphere eastward, out of the path of the indenting Indian plate (Tapponnier et al., 2001), and that the sinistral strike-slip faults within the Qaidam Basin are just a small component in this deformation. Zuza and Yin (2016) adopted the rotating crustal block models of England and Molnar (1990) and McKenzie and Jackson (1986). In this scenario, the sinistral faults rotate clockwise as they slip, permitting the western side of the system (e.g., the Tarim Basin) to move northward with respect to the eastern side (e.g., eastern China; see Zuza and Yin, 2016, fig. 10 therein). A third explanation was provided by England and Molnar (2005), who showed that the directions of horizontal compressional strain align with topographic gradients, and concluded that Eurasia deforms as a continuum under the influence of gravity. The implication of this conclusion is that where preexisting structures are not perfectly aligned to accommodate this strain, one of three scenarios can occur: creation of new structures, rapid rotation of preexisting structures into the correct orientation, or strain partitioning of oblique convergence utilizing existing basement structures.

Our results do not allow us to discriminate completely between these different scenarios, which are not absolutely mutually exclusive. The structures emphasize the continuum character of deformation, including the oblique component, which is at odds with any descriptions of extrusion tectonics that involve rigid, plate-like behavior (e.g., Avouac and Tapponnier, 1993). The deformation within the Qaidam Basin, with distributed strike-slip structures, seems to be different in style from the surrounding ranges. The overall kinematics are not distinctly different across the different regions, as revealed in the GPS data (Fig. 1).

The predominant structures in the western Qaidam Basin are a distributed array of northwest-trending transpressive structures with sinistral strike-slip components. They display positive flower geometries in 2D seismic profiles and helicoidal shapes in 3D space. The branch faults in the sedimentary cover and the major faults in the basement are demonstrated by the high-quality seismic data. This kind of structural framework is different from the southern and northern basin-bounding ranges, where the efficient strain partitioning results in strike-slip deformation being localized on single large sinistral faults (i.e., eastern Kunlun and Haiyuan faults).

The initial activities of the transpressive structures in the study area started in the middle to late Miocene (15–8 Ma). These structures, together with abundant other northwest-trending structures in the western Qaidam Basin, show a randomness rather than a northward or southward propagation in the formation sequence, which indicates that the uniform deformation within the Qaidam Basin has accelerated since the early Miocene.

The Qaidam Basin has a lower degree of strain compared with other regions of the Tibetan Plateau due to the higher strength of basement. However, it is clearly an oversimplification to treat the Qaidam Basin as a rigid block in the extrusion tectonic models. The pervasive transpressive structural deformation across the western Qaidam Basin is a plausible way to absorb the clockwise rotation that should have happened in the interior part of basin (Wang and Burchfiel, 2004), if fault block rotations models are correct (Zuza and Yin, 2016). The deformation mechanism is also consistent with regional compressive strain occurring at an oblique angle to basement faults, which are activated as sinistral flower structures in an example of distributed transpressional deformation.

This work was funded by the National Science and Technology Major Project of China (2011E-03). We thank Suotang Fu, Daowei Zhang, Dade Ma, Yunfa Feng, and Chuanwu Wang of Qinghai Oilfield Company for providing the seismic data in this paper. We also thank Wenjun Zhu, Anping Hou, and Tailiang Jiang for their help in operating the seismic interpretation software. And, we would like to thank Science Editor Damian Nance, who helped us with the manuscript, and reviewers Nigel H. Woodcock, Dickson Cunningham, and Meredith A. Bush, who provided constructive and thoughtful comments.


The mathematical equation set used in kinematics calculation of the faults in this paper is:
where D1 and D2 are the distances between the left two and right two reference points beside the fault, respectively; G1 and G2 are the gradients of the stratum in the left and right segments of the anticline, respectively, and X and Y are the lateral and vertical offsets of the fault, respectively (Table 1; Fig. A1). D1, D2, G1, and G2 are known numbers in this equation set, which can be derived from the depth-structure maps, and X and Y are unknown numbers, which need to be worked out by the equation set. We assume that the faults are nearly vertical in this calculation method.