1 Introduction

The Kunlun Fault System (KFS) is one of the major Tibetan strike-slip fault systems. It extends from longitude 86° E to the eastern margin of the Tibetan Plateau near the Sichuan Basin and strikes NWW to approximately EW over 1300 km. The KFS marks the northern boundary of the Songpan-Garze–Hoh Xil Terrane along the Anyemaqen-Kunlun suture (e.g., Molnar and Tapponnier (1975); Avouac and Tapponnier (1993); Seismological Bureau of Qinghai Province and Institute of Crustal Dynamics (1999); Yin (2010)) (Fig. 1). During the Cenozoic, left-lateral slip on the KFS had been intensive, accompanied by the uplift of the Tibetan Plateau (Ding and Lu 1986). Studies suggest that millimetric-level slip rates are spatially uniform along the KFS between ~92° E and ~104° E (Van der Woerd et al. 1998, 2002; Tapponnier et al. 2001; Li et al. 2004; Zhang et al. 2014). Regional geodetic estimates of the present velocity field are consistent with geologic slip rates (Zhang et al. 2004, 2014).

Fig. 1
figure 1

a KFS map in northeastern Tibet. Black box shows the location of B. CD Chendu, LS Lhasa, B KFS trace. Red points give epicenters of M6.0 from 1902 to 2001. Beach balls show focal mechanism of earthquakes. Epicentral locations and focal mechanism solutions of recent seismicity along the Kunlun fault are compiled from the USGS (http://neic.usgs.gov/neis/epic/epic_circ.html), Harvard CMT catalog (http://www.globalcmt.org/CMTsearch.html), and Molnar and Lyon-Caen (1989). White box shows the location of Fig. 2. Segmentation along the KFS compiled from previous work (Seismological Bureau of Qinghai Province and Institute of Crustal Dynamics 1999; Van Der Woerd et al. 2002; Li et al. 2011). 1 Kusaihu, 2 Xidatan-Dongdatan, 3 Alakehu, 4 Dongxi–Anyemaqen, 5 Maqin-Maqu, 6 Tazang. XH Xiahuangzhai

Full size image

With consideration to the large-scale geometrical complexities (push-ups, pull-aparts, releasing or restraining bends) and a uniform mean strike, the KFS is generally divided into six first-order segments of 150–270 km in length as follows: Kusaihu, Xidatan-Dongdatan, Alakehu, Dongxi-Anyemaqen, Maqin-Maqu, and Tazang (Seismological Bureau of Qinghai Province and Institute of Crustal Dynamics 1999; Van Der Woerd et al. 2002) (Fig. 1b). From west to east, the Maqin-Maqu segment is divided into two sections, Maqin and Maqu, by the Awancang fault. The Tazang segment comprises the Luocha and Minshan sections, which are separated by a large bend near the town of Xiahuangzhai. The Maqin section generally strikes 290–300°, and mainly dips to the southwest at angles of 70–80° with the length of about 110 km (Van Der Woerd et al. 2002; Li et al. 2004, 2011). The Maqu section strikes 280–305°, and is about 130 km in length. The Luocha section about 50 km north of Maqu traverses along the marshland-dominated Zoige Basin (Fig. 2). The average left-lateral slip rate of the KFS west of the town of Maqu was 10~12 mm year−1 since the late Pleistocene (Kidd and Molnar 1988; Van Der Woerd et al. 2002; Zhang et al. 2004; Li et al. 2011). However, the slip rate decreased from ~6–7 to ~2–3 mm year−1 in the Maqu section (Ma et al. 2005; Lin and Guo 2008; Li et al. 2011), and increased to ~7.7–9.4 mm year−1 in the Luocha section (Zhang et al. 2014).

Fig. 2
figure 2

Seismic ruptures along the Luocha section. Red box shows the location of Fig. 3. XH Xiahuangzhai

Full size image

There are no historical records of surface-faulting earthquakes along the KFS before 1902. Major earthquakes occurred in and after 1902 are well documented, since these events can be detected by instruments (China Earthquake Administration 1999). Six earthquakes with magnitudes ranging from seven to 7.9 have been recorded in KFS since 1902 (Table 1). In total, the earthquakes have shaped a ~1000 km-long earthquake rupture. However, there still remains a “seismic gap” of about 200 km in length in the eastern portion of KFS (~101° E and ~103.5° E). The seismic gap should include the Maqu and Luocha sections. It is estimated that the probability of a large earthquake occurring in the next 10 years is 95% (Wen et al. 2011). The elapsed time of the most recent event (MRE) (1055~1524 cal B.P.) is ~1000 in the Maqu section, beyond its maximum latest recurrence interval (Li et al. 2011). In addition, the age of the seismic events in the Luocha section is still unknown, resulting in a seismicity data gap of 70 km in length that remains to be solved. We have conducted a detailed survey of the watershed in the Koudousen River, together with additional sites, in order to find out the characteristics of the earthquake, fault rock sequences, and geomorphology. The key site we focused on is in the vicinity of a nomadic temporary residence named Benduo village, after aerial photographs revealed a dark grey and grey-white linear strip in the area.

Table 1 Summary of large earthquakes of north Tibet on Kunlun faults
Full size table

2 Methods

The Luocha section has obvious active structures of over 50 km that are recognizable with remote sensing or field observation. The geological and geomorphic features have been mapped by using 2.5 m spot data. In addition, detailed surveys of the surface rupture from the west of Luocha to the east of Xiahuangzhai (Fig. 2) have been conducted. There is an evident phenomenon located approximately 2-km southeast of the Benduo village (Fig. DR1): a linear fault trough in the middle of the north-dipping slope, which is approximately 25-m wide, 52 m at maximum, and 40 to 60-m above the river level, extends intermittently for about 850 m. Two sides of the trough are composed of a series of near-parallel scarps, striking 315° on the whole. The scarps are cut by gullies into discontinuous drag folds. The elevation differences between the drag folds and the trough are 3 to 6.5 m. Small-scale fault sag ponds are distributed in the rear of the scarps. Trenches were excavated across the sag ponds (Fig. 3), and samples of charred organic material were collected for AMS 14C dating. The OxCal radiocarbon calibration and analysis software (version 4.2) (Ramsey 2009) was used in our Bayesian statistical model. Depositional models that are not related to depth (simple sequence: Lienkaemper and Ramsey 2009; Ramsey 2008) were adopted, so that the dated age of events is not influenced by burial depth (DuRoss et al. 2011). Earthquakes catalog Ms ≥6 in northern Sichuan province have not occurred since 1879 (Huang et al. 1994), and we used this date as a historical constraint in our models. In addition, the Boundary, Phase, and R_Date commands, as well as other commands, were used according to Lienkaemper and Ramsey (2009).

Fig. 3
figure 3

a 2.5-m spot image of Benduo village. b Mapping and interpretations. White box location of Fig. 4. c White arrow shows the location of the fault exposure

Full size image

3 Results

It is concluded that the faulted bedrock, alluvium, and colluvium are buried by the unfaulted post-MRE deposits of 20~50 cm thick (Fig. 4). The principal fault zone F2 strikes northwest and dips northeast (305°/59° NE), and is characterized by a 30–60-cm-thick zone of light yellow-brown, high-plasticity clay fault gouge with gravel inclusions. Some liquefaction along the fault at the time of the MRE, with upward-leftward (topographical downward) injection, could be co-seismic with some diapirsm. The water table is 1~2 m under the ground there, so the earthquake may induce “clay” diapirsm with peristaltic water pumping. The plume at the top of the gouge is a mud escape bent downward.

Fig. 4
figure 4

a Photograph of the fault exposure. b Log of the KFS exposure from Benduo village (North Zoige, Sichuan province) with radiocarbon ages (14C; cal. calibrated) and seven generalized units: sandstone, fault gouge, faulted alluvium, faulted colluvium, most recent fault scarp (CD1), post-MRE deposit, and the modern forest soil; F1 to F5: identified faults, faults are shown in red

Full size image

The hanging wall is thrust over unconsolidated alluvium and colluvial debris on the footwall. The hanging wall is composed of Triassic sandstone with fault gouge at the base. On the hanging wall, the sandstone is overlain by a pebble-to-cobble-sized clast-supported colluvium. Attributes within the sandstones are consistent with the high-angle faulting observed in the gouge zone.

No less than five faults exposed through the alluvium and colluvium as well as through the bedrock in the hanging wall have been identified. The exposed faults have similar dips and strikes except the suberect fault F1 which leans to northeast with the dip of 88°. The fault F1 cuts the alluvium deposits formed before 820 to 920 cal 14C year B.P. (Fig. 4).

Co-seismic movement generated an unfaulted colluvial layer (CD1) that borders the scarp, overlies the principal fault plane. Below CD1, the fault is forming a continuous contact zone between the old colluvial and alluvial units. Results from radiocarbon and depositional model dating allow us to further interpret the complex stratigraphic and faulting history of the KFS at Benduo (Tables 2 and 3). Two samples of charcoals (samples 301644 and 301643) taken at the top of CD1 at the base of the buried scarp are dated at A.D. 1458–1649 and A.D. 1453–1643 with the 2σ model, respectively. An organic matter sample taken from the hanging wall (sample 301648) is dated at an age of 16,070–16,210 cal 14C year B.P. Two other samples of charcoal (samples 301651 and 301647) taken from the alluvial deposits which are now below the fault plane are dated at 790–850 and 890–950 cal 14C year B.P., respectively.

Table 2 Radiocarbon dating from fault exposure
Full size table
Table 3 Results from radiocarbon dating from fault exposure
Full size table

Another trench is located about 8-km northwest of Benduo village, in the central part of the south-dipping slope, 30 to 40 m above river level, which reveals a similar MRE. One sample of plant material and charcoal each taken at the bottom of post-MRE (sample 388193 and 301653) are dated at A.D. 1471–1661 and A.D. 1463–1651 with the 2σ model, respectively. Three other samples of charred materials (samples 301413, 301412, and 301420) near the top of its alluvial deposits, which are now below the fault plane, are dated at A.D. 1325–1456, A.D. 1316–1449, and A.D. 1309–1441 with the 2σ model, respectively. By using the age of the seven samples, it is inferred that the 2σ probability of the EQ1 (earthquake) age is A.D. 1407–1571 (GSA Data Repository Programmer file. DR1).

4 Discussion and conclusions

The new spot data reveals a northwest-striking scarp south of the Benduo village, the main trace of the KFS with obscure surface trace. We excavated a trench in the scarp, and documented a fault with as much as 30–60 cm of clay fault gouge. During the MRE, co-seismic movement generated the CD1 covering the fault scarp, allowing weak soil and charred materials to deposit on the scarp.

Because the footwall block was thrown down during the MRE and previous ruptures, 50 cm of post-MRE blackish grey alluvium, derived from tributaries of the Senduokuo River (Figs. 2 and 4), onlapped the fault while burying CD1 (including the charred organic materials that we dated) and a portion of the scarp. The stratigraphy in this trench (Fig. 4), combined with radiocarbon dating (Table DR2), indicates a history of repeated movements in late Holocene, including a post-1453 A.D. surface rupture that caused co-seismic in contact with Holocene. The faulted alluvium on the footwall before 820 to 920 cal 14C year B.P. indicates another event earlier than the MRE. As a result, the fault F1 formed along with this earlier event. At least three events happened, and three faults are formed simultaneously in the hanging wall of the fault F2; however, the time of these events were unclear. The sample (301648) lying in the fault zone may indicate a certain event that happened around 16,070–16,210 cal 14C year B.P. (Fig. 4).

An earthquake occurred during the reign of the Emperor Hongzhi in the Ming dynasty, A.D. September 25, 1488 (China Earthquake Administration (1999), and references therein). The earthquake happened in the northwest district of Sichuan province, about 1350-km away from Beijing, the imperial capital, affecting the Songpan and the Maoxian areas, 100~200-km east from the Zoige, and causing collapse of blockhouses in 6 villages and some casualties. The epicenter of the earthquake was undetermined. Evidence of the rupture of the fault on the Songpan and Maoxian around A.D. 1488 has not been found, and therefore, it is possible that this earthquake may have taken place in Zoige, 100~200-km west of the Songpan and the Maoxian areas.

According to the trench observations, we interpret this record as evidence of the A.D. 1488 rupture, and thus providing the first fault evidence from the horizontal deformation zone along the eastern KFS. This work also provides the first documentation of the MRE motion along the fault, with low-angle thrusting on a north-northwest-striking fault, which is consistent with the overall plate motion vector of 5~10° in eastern Tibet (Li et al. 2011). In addition, the strike of the fault we document here (305°) is consistent with the strike of the KFS, and this supports the uniform post-glacial slip rate model developed by Van Der Woerd et al. (2002); thus, we extend the east end of the KFS to Xiahuangzhai village.

The research results suggest that a large earthquake occurred on the eastern KFS in 1488 A.D., forming a surface rupture over 50-km long (L). Based on the regression equations (e.g., Wells and Coppersmith (1994); Leonard (2010)), the magnitude of the earthquake approximated at M w 7.0.

$$ M\mathrm{w}=4.33+1.52 \lg L $$
(1)
$$ M\mathrm{w}=5.08+1.16 \lg L $$
(2)

where M w is moment magnitude (Nm), and L is fault rupture length (km).

Even given the fact that the western termination of the 1488 rupture is unknown, we still argue that if portions of Hua Lake were ruptured, the 1488 earthquake was larger than the magnitude estimated here.

The difference in slip rates and a pull-apart basin between the Maqu and the Luocha sections indicate that the seismic gap consists of two sections. It is believed that segmented rupturing will be initiated by the next earthquake. Based on the characteristics of historical earthquakes along the KFS and the Yushu earthquake, as well as the observation of the ~130-km-long gap in the Maqu section, it is predicted with the regression equations, the next earthquake along the section will be M w ~7.5.

The recurrence interval of earthquake along the Maqu section is ~1000 years; however, the recurrence interval along the Luocha section is unclear. Currently, the MRE along these sections occurred 1000 and 500 years ago, respectively. Therefore, the elapsed time along the Maqu section is close to 1000 years, while the Wenchuan M8.0 earthquake of May 12, 2008 has imposed a stress load on the eastern KFS and, as a consequence, has increased the potential seismic hazard (Parsons et al. 2008; Shan et al. 2009) .For this reason, special attention must be paid to the Maqu section. Although the Luocha section is relatively low in seismic risks, those of its branch faults shall not be ignored.

In summary, data obtained in the research coincides well with sparse historical earthquake records, and validates earlier results that suggest a ~200-km-long seismic gap and further divide the seismic gap into two sections. Since the last earthquakes along the fault occurred 500–1000 years ago, which is coincident with the earthquake cycle, a new seismic event may be imminent in eastern Tibet.