The 2015 Ms 6.5 Pishan earthquake, Northwest Tibetan Plateau: A folding event in the western Kunlun piedmont

Folding earthquakes, which were observed in detail in the 1983 Coalinga earthquake, in Coalinga, California, USA (Stein and King, 1984), are a general type of rupture along foreland thrust systems. Although folding earthquakes are generally characterized by surface uplift and layer folding (Stein and Yeats, 1989; Yeats et al., 1997) without obvious surface fault offset, they also can cause destruction. Over the past several decades, this kind of hazard has been exemplified by the 1906 Ms 7.7 Manasi earthquake (Ms represents Richter magnitude scale) in the northern piedmont of the Chinese Tian Shan (Zhang et al., 1994), the 1985 Ms 7.1 Wuqia earthquake, in Xinjiang, China, along the Pamir front (Feng, 1997), and the 2013 Ms 7.0 Lushan earthquake in the Longmen mountain piedmont, in Sichuan, China (Xu et al., 2013).

The 2015 Ms 6.5 Pishan earthquake, which occurred in the western Kunlun Range piedmont (Fig. 1), caused significant casualties and property losses. Previous studies (e.g., Li et al., 2016; Lu et al., 2016; Zhang et al., 2016) have explored the seismogenic structure and rupture mechanism of this earthquake. Based on the statistical results of seismic data, an earthquake with a magnitude greater than 6.5 can generate an obvious surface rupture zone (Yeats et al., 1997). However, surface deformation of this event has not been reported. We investigate whether the Pishan earthquake generated a surface fault and what the deformation characteristics of this event are. The surface rupture caused by an earthquake can provide a unique opportunity to investigate the impact of coseismic faulting on landscape evolution and to refine regional deformation models (Wallace, 1977; Yeats et al., 1997; Bull, 2009). The tectonic deformation and uplift of the Tibetan Plateau have been popular areas of research. Two main models of upper crustal shortening and faulting (e.g., Tapponnier et al., 2001; Hubbard and Shaw, 2009; Xu et al., 2009; Jiang et al., 2013) and lower crustal viscous flow (e.g., Clark and Royden, 2000; Royden et al., 2008) have been used to explain the growth of the plateau. The tectonic deformation and growth pattern of western Kunlun, which is the northwestern margin of the Tibetan Plateau, are not currently well understood. A thorough study of the Ms 6.5 Pishan event will allow us to understand the tectonic deformation and seismotectonic model in this region.

In this paper, we first report the surface deformation caused by the Pishan earthquake based on our field investigations. We then utilize geologic data, seismic reflection profiles and earthquake relocation results to study the seismogenic structure of the Pishan earthquake and the deformation characteristics of the Pishan blind thrust fold. Finally, we discuss the tectonic deformation along the western Kunlun range front and the seismic risk in this region.

The western Kunlun orogenic belt is located on the northwestern margin of the Tibetan Plateau (Fig. 1). In response to the Cenozoic India-Eurasia collision, the western Kunlun Range was uplifted rapidly, and the overthrusting of Paleozoic bedrock onto Cenozoic strata can be widely observed along the range-front Tekilik fault (Cowgill, 2001; Yin et al., 2002). The crustal thickness in this area can reach ∼70 km (Negredo et al., 2007; Tseng et al., 2009), and the Cenozoic sediments in the foreland basin are more than 12 km thick (Matte et al., 1996). The strong uplift landforms, thick Cenozoic deposits in the foreland basin, and widespread active faults (Fig. 1) all attest to the intensive tectonic deformation in this region (Sobel and Dumitru, 1997; Zheng et al., 2000; Chen et al., 2011).

Several strike-slip faults are present within the western Kunlun Range (Fig. 1), which accommodate the different horizontal displacements of the tectonic blocks. The Karakoram fault, which is 510 km long and has an average slip rate of 6.9–10.8 mm/yr (Robinson, 2009), is a large dextral strike-slip fault that has accommodated ∼300 km of northward translation of the Pamir (Hamburger et al., 1992; Burtman and Molnar, 1993; Robinson et al., 2007; Strecker et al., 1995). The Kangxiwar fault is a large crustal-scale sinistral strike-slip fault (Tapponnier and Molnar, 1977; Peltzer et al., 1989; Fu et al., 2006) that represents part of the northwestern boundary of the Tibetan Plateau. The sinistral slip rate determined by geological (e.g., Fu et al., 2006; Li et al., 2008) and geodetic (e.g., Shen et al., 2001; Wright et al., 2004; Elliott et al., 2008) methods is ∼10 mm/yr. The Taxkorgan fault is an extensional dextral fault system (Brunel et al., 1994; Robinson et al., 2004; Cowgill, 2010; Zubovich et al., 2010) that is composed of several secondary faults (Li et al., 2011). The rate of extension of this fault decreases gradually from north to south (Robinson et al., 2007; Li, 2013).

Several rows of folds and thrust faults have developed along the western Kunlun range front (Matte et al., 1996; Si et al., 2007; Du et al., 2013; Li et al., 2016). The late Cenozoic tectonic deformation is characterized by foreland folds and thrust faults propagating from the range front to the Tarim Basin in the piedmont, forming thin-skinned nappe structures (Li and Wang, 2002), such as the Pishan and Kekeya anticlines (Fig. 2). The Mazhatagh thrust fault and related folds, which are located ∼200 km north of the western Kunlun, are considered to be the frontal active belt of the western Kunlun nappe structure (Pan et al., 2010). The active foreland folds mainly formed during the Quaternary (Chen et al., 2001; Liu et al., 2004; Si et al., 2007) and are the main structures accommodating the N-S crustal shortening deformation. The total crustal shortening calculated based on a balanced section is 24.6–54 km (Jiang et al., 2013), and the crustal shortening rate in this region determined by GPS data is ∼2 mm/yr (Shen et al., 2001; Li et al., 2016).

A petroleum industry seismic profile (Fig. 3; Liang et al., 2012) shows twofold belts between the range front and Pishan city. The southern fold is called the Kekeya anticline (Du et al., 2013). Near the range-front, the strata dip at high angles, and the dips of Cretaceous strata exposed at the surface can reach ∼80° (Si et al., 2007), which indicates that the Kekeya structural belt contains intense tectonic deformation and a high-angle thrust fault. North of the Pishan anticline, the seismic profile reveals shallowly dipping Cenozoic units and gentle folds. The maximum dip of the Neogene mudstone in the Pishan anticline is less than 20° (Si et al., 2007), indicating the presence of a shallowly dipping fault ramp under the anticline (Li et al., 2016; Lu et al., 2016).

The epicenter of the Ms 6.5 Pishan earthquake was located near the Pishan foreland fold and thrust belt (Fig. 1). The Pishan anticline, which has low relief and has experienced only ∼150 m of uplift, is mainly composed of Pliocene mudstone and lower Pleistocene conglomerates at the surface (Fig. 2). A petroleum industry seismic profile (Fig. 3) shows that this anticline is a typical blind thrust fault anticline. The layers of the anticline have been folded but without obvious offset. From the seismic profile, we can identify growth strata in the Pliocene layer, which indicate that the fold deformation began in the Pliocene (Chen et al., 2001; Liu et al., 2004). The N-S width of this fold is ∼20 km, and the detachment is located at a depth of 8–10 km (Liang et al., 2012; Li et al., 2016). At depth, the Pishan blind thrust fault merges into the range-front fault and then roots beneath western Kunlun (Liang et al., 2012). Based on trigonometric relations, we can estimate an average dip of ∼15° for the Pishan blind thrust fault based on a propagation distance of 50 km and a fault depth of 12–15 km at the range front. Figure 3 shows a fault ramp with a dip of ∼15° on the northern limb.

At the crest of the Pishan anticline, a group of northwest-striking fault scarps are present (Fig. 4A; Pan et al., 2007), and an older geomorphic surface with a higher scarp (Li et al., 2016) indicates that these faults have been continuously active. Remote sensing image interpretation and field investigations show that these faults are distributed over a N-S width of ∼9 km. The scarps have a discontinuous distribution and a consistent strike. A trench that we excavated across the scarp revealed a typical normal fault (Figs. 4B–4D). The tectonic deformation of the blind thrust anticline is characterized by the displacement gradually transforming into layer bending and folding near the surface, and some bending moment faults may form because of local tensile stresses at the crest of the anticline (Fig. 4E; Li et al., 2001). The 1983 Mw 6.5 Coalinga earthquake, which did not rupture a distinct seismic fault at the surface, caused surface uplift of up to 0.6 m (Stein and King, 1984). Therefore, we suggest that these normal faults on the Pishan anticline are bending moment faults (Li et al., 2016).

The earthquake intensity in the meizoseismal areas of the Ms 6.5 Pishan earthquake was classified as VIII degrees (Mercalli intensity scale). The long axis direction of the meizoseismal area is generally consistent with the strike of the Pishan anticline (Fig. 5A). Along the active fold, we did not find a coseismic fault. However, near the core of the Pishan anticline, several tensional ground fissures are present. These fissures are mainly distributed in the western area of the epicentral region (Fig. 4A), where the maximum slip of this earthquake occurred (Zhang et al., 2016). At the village of Kumuqiake in the town of Pixina, two groups of ground fissures have a right-stepping geometry. The southeastern branch, which is ∼18 m long and has a fissure width of 1–3 cm, is located on a hard asphalt road and trends 310° (Fig. 5B). The northwestern branch, which is 20 m long and has a maximum width of 20 cm, is located on flat ground in an orchard and trends 300° (Fig. 5C). Near the town of Pixina, several fissures with a total length of ∼100 m and widths of 1–10 cm are present on flat ground. These fissures trend 290° and cut through the hard asphalt road (Fig. 5D). Approximately 300 m south of Pixina, a ground fissure with a strike of ∼290° is present, and it extends to the flat, hard ground of a residential yard (Fig. 5E). In addition, sand liquefaction can be widely observed in the earthquake area (Fig. 5F). The sand liquefaction sites are mainly distributed in the vicinity of the anticline core and the reservoir (Fig. 4A).

The coseismic fault dislocation of the Pishan earthquake was ∼0.6–1.0 m at the epicenter (He et al., 2016; Zhang et al., 2016). However, no coseismic fault was found at the surface. Inversion results indicate that the top of the fault that ruptured in the Pishan earthquake is buried 5 ± 2 km beneath the surface and that the surface uplift was ∼10 cm (Zhang et al., 2016). The tensional ground fissures caused by the earthquake are mainly located on flat ground near the core of the Pishan anticline, and their strikes are generally consistent with that of the anticline. Some of the ground fissures extend ∼100 m and cut through a hard asphalt road. We suggest that these ground fissures are closely related to tectonic activity and were not caused by the strong ground motion or sloping terrain. Rather, the coseismic displacement was transformed into layer bending and folding near the subsurface; therefore, the tensional ground fissures are the surface deformation caused by this earthquake.

To study the relationship between the Pishan earthquake and the structure, we relocated the main earthquake and its aftershocks using the double differential location method. A total of 1793 aftershocks recorded by more than three stations within 400 km (Fig. 1B) with six seismic phases were utilized. We modeled the one-dimensional velocity structure (Table 1) based on the seismic reflection results (Li et al., 2001), geologic mapping (Si et al., 2007), and deep drilling data (Geological and Mineral Bureau of Xinjiang Uygur Autonomous Region, 1992) in this area, which can constrain our structural model. In this velocity model, the crustal thickness is ∼52 km, which is generally consistent with receiver function results (e.g., Liu et al., 2011). Finally we obtained 1549 relocated earthquakes (Figs. 2 and 6). Based on the conjugate gradient method, the average relocation errors in the N-S, E-W, and U-D directions are 0.8 km, 0.8 km, and 0.7 km, respectively, and the average relocation residual is 0.07 s.

Cut and paste is an effective method for inverting focal mechanisms because of its insensitivity to lateral differences in the velocity structure (Zhu and Helmberger, 1996). In this paper, we calculated the fitting error function between theoretical and actual full waveforms after removal of low correlation coefficient through assigning different weights to Pnl and S waves, and obtained the optimal solutions using grid search method. We utilized CRUST 2.0 (http://igppweb.ucsd.edu/-gabi/crust2.html) to obtain one-dimensional velocity structure. The focal mechanisms of the main earthquake and 13 aftershocks that were calculated based on at least 8 seismic stations in each inversion are shown in Table 2.

The focal mechanism of the main earthquake shows that the seismogenic structure strikes northwest (Fig. 2) and that the earthquake was a typical thrust fault rupture. The relocated depth of the Ms 6.5 Pishan earthquake is 8.4 km, which is consistent with the fault depth revealed by the seismic reflection profiles (Jiang et al., 2013). An inversion result based on teleseismic body waves and interferometric synthetic-aperture radar measurements also indicates that the rupture occurred at a depth of ∼7–9 km (Zhang et al., 2016). The main earthquake was located on the fault ramp of the Pishan anticline, and its dip of ∼15–20° is consistent with the dip of the fault ramp. The relocated aftershocks are densely distributed along the Pishan anticline in a zone ∼45 km long from northwest to southeast and 20 km wide from northeast to southwest (Fig. 2). The aftershock distribution in map view (Fig. 2) and profile (Fig. 6) indicate that the deformation occurred throughout the entire area of the anticline and was not concentrated along the fault belt. The temporal and spatial distributions of the main shock and its aftershocks indicate that the Pishan earthquake rupture propagated from the southeast (epicenter) to the northwest. The focal depths of the Ms ≥ 1.0 aftershocks show that the tectonic deformation mainly occurred in the Cenozoic sedimentary layers at depths of 3–12 km. The envelope of the aftershocks is similar to the shape of the fold (Fig. 6), and the aftershocks near the surface are normal faulting focal mechanisms (Fig. 3), indicating tensional stresses near the surface. The aftershock distribution is consistent with the structural deformation characteristics of a blind thrust fault and fold, which indicates that the Pishan earthquake was a typical folding earthquake event.

In contrast to the linear distribution of aftershocks along a steeply dipping fault (Stein and Yeats, 1989), the aftershocks of the Pishan earthquake were densely distributed over a width of ∼20 km with a length-to-width ratio of ∼2 (Figs. 2 and 6). The seismic profile (Fig. 3) reveals that the seismogenic fault dips at ∼15° (Li et al., 2016; Lu et al., 2016), and the coseismic uplift occurred along the entire Pishan anticline (Zhang et al., 2016). The energy of the Pishan earthquake was released over a wider area compared to high dip-angle faults. In addition, the shallow depth of only ∼8–9 km for the main shock and the basin amplification effect increased the earthquake effects. Therefore, although this earthquake was not the largest earthquake in Xinjiang in recent years, it caused significant destruction.

As the northwestern margin of the Tibetan Plateau, the western Kunlun Range has experienced intense tectonic deformation and uplift. Previous studies have indicated that the tectonic deformation in the western Kunlun Range piedmont is that of a typical foreland fold and thrust belt in which the tectonic activity and crustal shortening are migrating from the range front toward the basin (e.g., Matte et al., 1996; Zheng et al., 2000; Pan et al., 2007; Liang et al., 2012; Jiang et al., 2013). The thrust deformation and horizontal shortening have migrated northward at least ∼50 km into the interior of the Tarim Basin (Fig. 1). The foreland fold and thrust belt, including the Pishan anticline, shows evidence of strong tectonic activity because the late Quaternary sediments and river terraces have been faulted (Fig. 4; e.g., Li et al., 2016). The Tekilik fault, which is a range-front fault zone in western Kunlun, has been inactive since the late Quaternary, indicating the outward growth of the piedmont. In contrast to the thrust movement and crustal shortening in the range front, strike-slip faults accommodate the differences in horizontal displacement between the tectonic blocks within the plateau. The focal mechanisms in the interior of the western Kunlun Range all show obvious strike-slip motion (Fig. 1). Because the India plate downthrusts beneath Tibet and the Tarim Basin forms an obstruction in the north, Tibet is laterally extruded due to India’s penetration into Asia (Matte et al., 1996). The northward propagation of the western Kunlun results in the southward subduction of the Tarim Basin beneath Tibet (Matte et al., 1996). The N-S convergent motion between Tibet and the Tarim Basin generated the Ms 6.5 Pishan earthquake. In general, the high-angle strike-slip faults within the high range accommodate lateral motion, and the low-angle thrust faults in the foreland mainly absorb the crustal shortening, which results in slip partitioning in this region (Fig. 7). The Pamir, which is also located along the northwestern margin of the Tibetan Plateau, is dominated by thrusting along the range front and strike-slip motion within the plateau (e.g., Burtman and Molnar, 1993; Zubovich et al., 2010). Since the Cenozoic, the Pamir has penetrated ∼300 km northward into Eurasia (e.g., Burtman and Molnar, 1993; Cowgill, 2010). We propose that the northward penetration of the Pamir can be explained by northward fault propagation.

The seismic reflection profile (Fig. 3) shows that the tectonic deformation mainly occurs in the Cenozoic layers from depths of ∼8–12 km to the surface (Liang et al., 2012; Li et al., 2016). Consistent with the tectonic deformation of the piedmont, the Ms 6.5 Pishan earthquake was a typical thrust rupture event (Li et al., 2016; Lu et al., 2016; Zhang et al., 2016). Our study also shows that the main shock occurred at a depth of 8–9 km and that the deformation caused by the Pishan earthquake mainly occurred in the Cenozoic layers. In recent years, several moderate to strong thrust earthquakes have occurred along the active foreland structural belt (Fig. 1), indicating that the foreland fold and thrust belt is the main structure that accommodates the N-S crustal shortening (Jiang et al., 2013). A series of bending moment faults on the crest of the Pishan anticline indicates that the deformation caused by large earthquakes during the late Quaternary was dominated by layer folding and surface uplift in the upper crust. Based on a seismic reflection profile, Li et al. (2001) reported average crustal thicknesses of ∼60–65 km beneath the western Kunlun. Compared with the ∼50 km crustal thickness on the southern margin of the Tarim Basin, we estimate 15 km of excess crustal thickness beneath a region as wide as ∼80 km from the Kangxiwar fault to the Tekilik fault. Thickening due to shortening of a 50-km-thick crust would require ∼24 km of crustal shortening, which is consistent with the crustal shortening of 24.6–54 km calculated based on the balanced section (Jiang et al., 2013; Lu et al., 2016). Therefore, we suggest that the tectonic deformation in the western Kunlun area mainly occurred in the upper crust and that the growth of the plateau is driven by upper crustal shortening and thickening.

Previous studies have shown that the foreland thin-skinned nappe structure is a seismogenic system that can generate large earthquakes (Avouac et al., 1993; Avouac, 2003; Zhang et al., 1994; Feldl and Bilham, 2006; Liu et al., 2015). Based on the fault-plane area, Li et al. (2016) proposed that the foreland thrust system can generate large earthquakes with magnitudes greater than 7. The colluvial wedges revealed by our trench across the fault scarp (Figs. 4C, 4D) also indicate that several large earthquakes have occurred along the foreland structure. The Pishan event only ruptured part of the Pishan anticline, and the aftershock distribution also shows that the energy was mainly released along the Pishan anticline. However, the main locked part between the Tekilik fault and the Pishan anticline did not rupture. In addition to the concentrated distribution of aftershocks in the vicinity of the main shock, a group of small shocks in the western region, which occurred along the Tekilik fault, was possibly triggered (Fig. 1). A static stress study showed an increase in the Coulomb stress of 471 Pa on the western Kunlun range front (Jin et al., 2017). Based on the study by Ziv and Rubin (2000), a small change in the Coulomb stress may affect seismicity on active faults. In the piedmont of the western Kunlun, no large earthquake with a magnitude greater than 7 has been recorded in the last 200–300 years. Therefore, this area may have a high seismic risk that should be the focus of further attention and research.

The Ms 6.5 Pishan earthquake, which occurred on the south-dipping fault ramp of the Pishan blind thrust fault anticline, was a typical folding event with a focal depth of 8–9 km. This earthquake did not generate an obvious coseismic surface fault. A series of northwest-striking tensile fissures on the crest of the anticline represent the surface deformation caused by this earthquake.

The tectonic deformation of the Pishan blind thrust anticline is mainly characterized by layer folding and bending. The normal fault at the top of the anticline is a bending moment fault that developed because of the local tensile stress. The deformation caused by this earthquake mainly occurred in the Cenozoic sedimentary layers at depths of 3–12 km. The foreland folds and thrust faults are the main structures that accommodate the N-S crustal shortening across the western Kunlun range front.

This research was supported by the National Science Foundation of China (41672208, 41590861, 41661134011) and a fund from the State Key Laboratory of Earthquake Dynamics (LE1413). High-resolution remote sensing images were obtained from Google Earth. This paper commemorates the victims of the Pishan earthquake.