Paleoseismology and Slip Rate of the Western Tianjingshan Fault of. NE Tibet, China

Transcription

1 Manuscript Click here to download Manuscript: Manuscript.docx Click here to view linked References 1 2 Paleoseismology and Slip Rate of the Western Tianjingshan Fault of NE Tibet, China Xinnan Li 1,2,3, Chuanyou Li 2 *, Steven G. Wesnousky 3, Peizhen Zhang 1,4, Wenjun Zheng 1,4, Ian K.D. Pierce 3, and Xuguang Wang State Key Laboratory of Earthquake Dynamics, Institute of Geology, China Earthquake 8 Administration, Beijing , China 9 2 Key Laboratory of Active Tectonics and Volcano, Institute of Geology, China Earthquake 10 Administration, Beijing , China 11 3 Center for Neotectonic Studies, Mail Stop 169, University of Nevada, Reno, Nevada 89557, 12 USA 13 4 School of Earth Science and Geological Engineering, Sun Yan-Sen University, Guangzhou, , China * Corresponding author: Chuanyou Li, Key Laboratory of Active Tectonics and Volcano, Institute 17 of Geology, China Earthquake Administration, No.1 Huayanli, Chaoyang District, Beijing , 18 China. Tel: chuanyou@ies.ac.cn 19 1 / 22

2 20 21 ABSTRACT We present results of a detailed investigation of the horizontal displacement distribution, 22 timing of paleoearthquakes and left-lateral slip rate on the western Tianjingshan fault. 23 Measurements of 240 offset streams and ridges confirm that the fault is left-lateral and record 24 evidence of repeated ~3-4 m coseismic offsets along the 60 km long fault, suggesting that ~6 25 earthquakes may have occurred along the entire western Tianjingshan fault with repeated 26 occurrence of earthquakes of Mw 7.2 to 7.5. Structural and stratigraphic relationships exposed in 27 our five trenches in combination with previously reported studies further indicate that the fault has 28 ruptured in as many as ~6 paleoearthquakes since the late Quaternary. Paleoseismic data show that 29 the average recurrence interval for Holocene earthquakes is about 5,000 yr. The most recent 30 earthquake along the western Tianjingshan fault occurred ~1.2 ± 0.1 kyr BP, indicating that this 31 fault segment did not rupture in the M 7.5 historical earthquake of AD 1709 that ruptured the 32 central Tianjingshan fault. We estimate that the Holocene slip rate of the western Tianjingshan 33 fault is ~ mm/yr based on measurements of the age and offset of stream channel. Compared 34 with the relatively fast slip-rate ~4-6 mm/yr of the Haiyuan fault, we suggest that the Tianjingshan 35 fault acts as an essential active fault that accommodates the sinistral displacement and crustal 36 shortening deformation in NE Tibet. 37 Keywords: The western Tianjingshan fault; Paleoseismology; Slip rate; NE Tibet; Tectonic 38 deformation 39 2 / 22

3 Introduction The major left-lateral Altyn-Tagh and Haiyuan faults strike along the northern and 42 northeastern margin of the Tibetan Plateau (Figure 1A). The Tianjingshan fault splays eastward off 43 the Haiyuan fault (Figure 1B) and marks the northeasternmost strike-slip associated with the 44 margin of the Tibetan Plateau. The Tianjingshan fault is defined by a series of discontinuous 45 Holocene active traces that extends east from about Gulang to Tongxin city in the east, a distance 46 of 240 km. The fault is divided into 4 sections on Figure 1B based on fault geometry and 47 geomorphology (Li et al., 2015), which are labeled from east to west as the northern 48 ChuangLingshan fault (N-CLF), and the western (W-TJSF), central (C-TJSF), and eastern 49 (C-TJSF) Tianjingshan faults, respectively (Figure 1B). The M7.5 Zhongwei, Ningxia, earthquake 50 of 14 October 1709 killed >2000 people and produced left-lateral surface rupture along about km of the central Tianjingshan fault (C-TJSF) (Zhang et al., 1988, 2015; Figure 1B). 52 Numerous studies have described the well-preserved surface ruptures (Zhang et al., 1988; Nie et 53 al., 1993), established the average recurrence intervals ( years, Zhou et al., 1987; 54 Wang et al., 1990; Min et al., 2001) and defined the slip rates (~2±0.5mm/yr, Chai et al., 1997; 55 Li et al., 2005) of the central Tianjingshan fault. 56 The 60-km-long western Tianjingshan fault (W-TJSF) has received less attention than the 57 central Tianjingshan fault (C-TJSF). Chai et al. (2003) observed that the western Tianjingshan 58 fault traverses but does not cut the Great Wall of China and with geomorphic observations 59 concluded that the most recent rupture along the western Tianjingshan fault occurred more than years ago and did not rupture in The W-TJSH is composed of five subparallel fault 61 strands separated by pull-apart basins of different size (Du et al., 2007). The individual fault 3 / 22

4 62 strands are labeled from F1 to F5 in Figure 1C. The average strike of the western Tianjingshan 63 fault is about N90 E. A relative scarcity of instrumentally recorded earthquakes along the western 64 Tianjingshan fault, between Luotuoshui and Gushanzi (Figure 1C), has been interpreted to 65 suggest that the fault is locked and therefore might pose a significant potential seismic hazard to 66 the heavily populated Jingtai and Zhongwei region (Working Group of M7, 2012). 67 In this paper, we present geomorphic evidence of repeated slip at more than 16 sites along the 68 traces of the western Tianjingshan fault. We then combine new paleoseismic observations obtained 69 by trenching along with those previously reported by others (Chen et al., 2006; Du et al., 2007; 70 IGCEA et al., 2003; Min et al., 2001) to interpret the past rupture history, and recurrence 71 intervals of large surface-rupturing earthquakes on the western Tianjingshan fault (W-TJSF). 72 Additionally, we use offset of an older stream channel to estimate the W-TJSF fault slip rate. The 73 results provide a basis to discuss the behavior of the fault within the context of existing earthquake 74 recurrence models and the tectonic role of the fault along the northeast margin of the Tibet 75 Plateau Methods 77 We measured offset geomorphic features on the W-TJSF using tape measurers and laser 78 aperture rangefinders in the field, as well as Google Earth imagery, and structure-from-motion 79 (SfM) models (Supplementary Material S1 and Figures S1-S3). The traces F3 and F5 are 80 commonly modified by surface mining and so our measurements are limited to those made along 81 segments F1, F2 and F4. The results of offset measurement are present in section 3.1. Using both 82 hands and machine excavators, five trenches (TC-1, TC-2, TC-5, TC-6, and TC-7) were excavated 83 across the active fault scarps mainly on the F1, F2, and F4 subsegments. The location of each 4 / 22

5 84 trench is labeled in Figure 1C. Before trenching, the fault-trace geometry and surface geomorphic 85 units were mapped and topographic profiles across fault scarps were measured at each site. The 86 walls of each trench were scraped and cleaned prior to logging and a 1-m string grid placed across 87 the exposures. The sedimentary layers in each trench were differentiated as individual units based 88 on color, grain size, sorting, texture, bedding thickness, and other features such as liquefaction 89 (Supplementary Table S1). A sequence of ground-rupturing events was then defined on the basis 90 of crosscutting relationships between strata, sediment thickness changes, sediment deformation, 91 fissure, and sand liquefaction. In each trench, interpreted paleoearthquake are designated at E1, E2, 92 etc. with E1 being the youngest. Optically stimulated luminescence (OSL) ages of sediment 93 samples collected from the trenches are used to constrain the ages of the stratigraphic units and 94 hence the interpreted events. All samples were processed and measured in the Luminescence 95 Dating Laboratory of State Key Laboratory of Earthquake Dynamics, Institute of Geology, CEA, 96 Beijing using conventional pretreated techniques (see Lu et al., 2007 and Yang et al., 2013 for 97 detailed experimental operating procedures). The results and details of methodology are 98 summarized in Supplementary Material S1 and Table S Results Geomorphic evidence of paleoearthquakes from offset measurements 101 A large quantity of offset geomorphic features can be observed from the field and Google 102 Earth imagery along the W-TJSF. Figure 2, for example, shows series of stream channels coming 103 out of small hills composed of late Neogene and early Pleistocene sediments and displaced by 104 fault. The enlarged circles in Figure 2 indicate recent left-lateral offset streams. As a result, these 5 / 22

6 105 offset features of different size offer us a good opportunity for better documenting its seismic 106 behavior and the occurrence of past large earthquakes on the W-TJSF. 107 The 240 offset measurements are summarized in Supplementary table 3. To quantitatively 108 determine the relationship between along-strike offset distribution and individual 109 paleoearthquakes, we plot our measures of offset as a function of distance along each fault 110 segment in Figure 3A. Vertical dashed lines in Figure 3A connecting symbols with dark 111 background are measurements of multiple offsets made at the same site, whereas those sites 112 exhibiting only one single offset are marked by points. The sum of observations in Figure 3B is 113 plotted as a histogram in Figure 3A. We observe that there are peaks in the histogram separated on 114 average of ~3-4 meters and interpret that each peak represents the occurrence of a single 115 earthquake. The smallest observable offsets are interpreted to correspond to the most recent 116 earthquake, and successive larger offset groups record the cumulative slip of prior events. The 117 interpretations assume that the production rate of geomorphic features capable of recording fault 118 offset is much faster than the frequency of surface-rupturing earthquakes (Zielke et al., 2015). The 119 smallest observable offsets correspond to the most recent earthquake, and successive larger offset 120 groups record the cumulative slip of prior events. The 129 single-event offsets, attributed to the 121 most recent earthquake, record an average coseismic slip of 3±1.5 m. Consequently, the distinct 122 peaks at ~6 and ~9 m might represent the penultimate and third earthquakes due to repetition of 123 earthquakes producing ~3 m coseismic offsets. Peaks at ~12, ~16, and ~20 m are likely the result 124 the cumulative slip of four to six earthquakes. Regardless of the uncertainty coming from the 125 decreasing number of larger offset measurements, our observations appear to support the idea that 126 the slip distribution is similar for each repeated earthquake along the western Tianjingshan fault. 6 / 22

7 Paleoseismic evidence from trenching investigations 128 To assess the timing of earthquakes responsible for each peak in the histogram of Figure 3B, 129 we use structural, stratigraphic, and OSL relationships observed in the five trenches (TC-1, TC-2, 130 TC-5, TC-6, and TC-7) we excavated across the fault (Figure 1C, see Supplementary Material 131 S2 and Figures S4-S6 for detail of TC-1 and TC-6). Each is described in the sections below. 132 Detailed descriptions of sedimentary units in trench exposures and OSL ages of all samples 133 collected from trenches are summarized in Supplementary Table 1 and Supplementary Table 2, 134 respectively TC The location and geomorphic context of trench site TC-2 are shown in Figure 1C and 4C. 137 The log of the trench presented in Figure 5 shows a sequence of alluvial layers overlain by finer 138 grained eolian loess. The sequence is cut by four fault strands F1, F2, F3, and F4. The 139 stratigraphic and tectonic relationships may show only a single earthquake displacement in the 140 TC-2 trench. The event produced the folding and thrust faulting of units U1-1/2, U1-2, and U along fault F3-4 that upwardly die out in unit U1-4, which is covered by the undeformed unit U2 142 (Figure 6A). The E1 event also has been identified as the termination of F1-2 in unit U1-4 toward 143 the top of the exposed surface (Figure 6B). Although we cannot completely rule out the 144 possibility that F3-4 displacement may represents an event older than F1-2 in TC-2 trench, it is 145 more likely that fault strands F1, F2, F3, and F4 ruptured simultaneously as only one earthquake 146 displacement observed in TC-1 trench only about 200 m west of TC-2 trench. The 4.4 ± 0.3 kyr 147 age of TJSH-50 represents the upper limit of the event horizon and the age of 0.2 ± 0.1 kyr from 7 / 22

8 148 TJSH-51 indicates the lower chronological limit for the most recent earthquake E1. Therefore, E1 149 is likely to have occurred within kyr TC At this trench site, located in Figures 1C and 7A-B, the fault trace strikes S105 E and can be 152 recognized on satellite imagery as well as in the field as a clear, linear, single trace that offsets a 153 large number of streams along the GGL subsegment (F2). We excavated a 14-m-long trench across 154 a south-facing scarp, located approximately 700 m east of Majiajing on the GGL subsegment (F2). 155 At this site, the fault trace cuts across a late Pleistocene-Holocene alluvial fan and left-laterally 156 displaces a young stream ~2.7 m and an older stream ~11.7 m, west of our trench (Figures 7C). 157 Here, the scarp height is up to 80 cm, and varies along strike from about 0.3 to 2.8 m, and the 158 trench exposures, from oldest to youngest, a sequence of late Pleistocene and Holocene 159 sedimentary layers labeled U1 through U11 (Figure 8). The sequence is cut all or in part by seven 160 fault strands, labeled F1 to F7 in Figure We identified at least four paleoseismic events, designated as E1, E2, E3, and E4 as a 162 function of increasing age, and here discuss evidence of each in chronological order. The main 163 fault zone consists of fault strand F3 to F7, all generally showing down to south displacement. 164 Strand F3 extends highest in the section to where it is capped by thin unbroken layer unit U11, and 165 thus records the most recent displacement paleoearthquake (E1) in the exposure. OSL ages for 166 samples taken from the unbroken unit U11 and the broken layer U10 place the time of last 167 displacement on F3 at between 1.7 ± 0.2 kyr (TJSH-66) and 3.5 ± 0.4 (TJSH-29). Apparent 168 vertical offset on F3 is about cm across units U4 to U10, but the sense of displacement 169 reverses and unit thickness vary across F3 in the lower units U1 and U3, suggesting that a 8 / 22

9 170 component of the offset is strike-slip. 171 The penultimate earthquake is registered by faults F4 and F6 that project upward to fault the 172 base of unit U9, but apparently not the upper contact of unit U9. The lower contact of U10 is flat 173 suggesting that some amount of erosion into U9 occurred before U10 was deposited. Assuming 174 OSL sample TJSH-28, taken from the base of U9, was deposited prior to the last displacement on 175 F4 and F6, it is interpreted displacement on F4 and F6 occurred simultaneously after 8.9 ± 0.8 kyr 176 and before 3.5 ± 0.4 kyr, the age of sample TJSH-29 at the base of overlying unit U10 (Figures and 9A). 178 Displacement on faults F1 and F2 are interpreted to record an antepenultimate event. Each 179 projects upward from the base of the trench in the footwall to displace the base of unit U7 where, 180 together, they form a buried graben (Figures 8, 9B-C). Neither strand F1 and F2, however, project 181 through and to the upper contact of unit U7. The age of displacement on strands F1 and F2 clearly 182 postdates deposition of unit U6 for which OSL places a 25.6 ± 1.4 kyr age (Sample TJSH-31). 183 Placing an upper bound on the timing of the antepenultimate event is problematic. The deposition 184 of OSL sample TJSH-59 near the base of unit U7 may have occurred either before or after 185 displacement on underlying faults F1 and F2. The position of TJSH-59 with respect to fault 186 displacement is not clear. Additionally, the 29.2 ± 1.4 kyr age of OSL sample TJSH-59 is 187 stratigraphically inverted with the younger 25.6 ± 1.4 kyr age of OSL sample TJSH-31 in 188 sediment clearly broken by F1 and F2. If sample TJSH-59 in unit U7 was deposited after 189 displacement on fault F1 and F2, it may at best be interpreted that the antepenultimate 190 displacement occurred somewhere around 25 to 29 kyr ago. 191 The oldest earthquake (E4) is suggested with a fault-controlled wedge of unit U5 filled with 9 / 22

10 192 coarse sand and fine-grained pebbles in unit U4-3, and the lower part of this wedge links to 193 liquefied strata of unit U3 via a tube-like dike (Figure 8 and 9 D). The lower contact of overlying 194 unit U6 is well preserved and expressed as a continuously flat silty layer, showing no noticeable 195 effect from liquefaction. In addition, stratigraphic thickness changes of units U1 and U3 across F3 196 indicate that an older strike-slip displacement may also have occurred on F3. Therefore, the oldest 197 earthquake can be reasonably inferred to have occurred sometime between the deposition of unit 198 U4-3 and unit U The sequence of displacement and accompany erosion in the trench is summarized in Figure The cumulative vertical displacement of all the earthquake horizons is ~85 cm and consistent 201 with the scarp height (0.77 ± 0.03 m) obtained from survey profiles across the main fault zone. 202 From this observation, it may be inferred that trench records a complete record of paleoseismic 203 events since E TC The TC-7 trench is located about 300 m south of the Dingwu Highway that is subparallel to 206 the east-west trending fault trace (Figures 1C and 11A). At the trench site, the surface trace of the 207 Zhongwei subsegment (F4) follows the boundary between Neogene bedrock and late Quaternary 208 sediments (Figure 11A). The TC-7 trench, about 13-m-long, was excavated across a discrete scarp 209 that contained up to 3 m of south-side-up vertical slip (Figures 11B-D). Exposed in the trench are 210 four steeply dipping faults labeled F1, F2, F3, and F4, respectively. The southernmost fault F1 is 211 within Neogene bedrock. The remaining three fault strands break and are associated with a 212 sequence of late Pleistocene to Holocene colluvial and fill deposits labeled as units U2 through U7, 213 respectively. Unit descriptions accompany the trench log are shown in Supplementary Table / 22

11 214 The structure and stratigraphy are interpreted to record 6 paleoearthquakes. 215 The most recent paleoearthquake (E1) is evident in the preservation of the wedge-shaped 216 colluvial deposit (U7-1) that is bound on the south by the fault F2 (Figure 12). The event occurred 217 after deposition of unit U6-2 and U5. Ages of OSL samples TJSH-36, 37, and 38 from units U and 5 range from 8.2 ± 0.5 kyr to 11.0 ± 0.9 kyr and provide a maximum bound on the age of E A penultimate event (E2) is recorded by the colluvial wedge unit U6-1, also bounded to the south 220 by fault F2 (Figure 12). Formation of the colluvial wedge unit U6-1 occurred prior to deposition 221 of overlying unit U6-2 and after deposition of the directly underlying unit 5. The OSL ages of 222 sample TJSH-36 from overlying unit U6-2 is 11.0 ± 0.9 kyr. The OSL ages of samples TJSH and JSH-38 taken from underlying unit U5 are 9.9 ± 0.5 kyr and 8.2 ± 0.5 kyr, respectively. The 224 ages are stratigraphically inverted. We assume that the older age of sample TJSH-36 is due to 225 incomplete light exposure. With this assumption, paleoearthquake displacement E1 and E2 226 postdate about 9.9 ± 0.5 kyr. An antepenultimate event is recorded by rupture of the base of unit 227 U5 by fault strand F3 and F4. OSL samples TJSH-37 and TJSH-38 place the age of the lower 228 portion of unit U4 somewhere around 9.9 ± 0.5 kyr and 8.2 ± 0.5 kyr, respectively, so the 229 antepenultimate event occurred after ~9.9 ± 0.5 kyr. In this regard, the trench exposure records earthquake displacements since ~9.9 ± 0.5 kyr, but provide no unique information to place both 231 upper and lower bounds on the timing of each displacement. 232 An older paleoearthquake E4 recorded by colluvial wedge unit U4. Formation of colluvial 233 wedge U4 postdates deposition of unit U3-2, which is of 24.3 ± 2.1 kyr (OSL sample TJSH-61) 234 and likely predates the 8.2 ± 0.5 kyr age of OSL sample TJSH-38 in unit U5. A small fissure fill 235 colluvial deposit unit U3-1 that abuts fault strand F3 is interpreted to record a yet older 11 / 22

12 236 paleoearthquake E5. The deposition of unit U3-1 and thus the age of E5 postdates deposition of 237 unit U2-3 which has an age of about 32.5 ± 1.3 kyr to 28.1 ± 1.6 kyr (OSL ages of samples 238 TJSH-39 and 49) and predates the 24.3 ± 2.1 kyr age of unit U2-3 (TJSH-61). The colluvial wedge 239 of deposit U2-1 abutting fault strand F2 is interpreted to record the oldest paleoearthquake (E6) 240 interpreted in the trench exposure. We have no OSL ages to place a maximum bound on the timing 241 of E6. The 32.5 ± 1.3 kyr age of OSL sample TJSH-49 taken from unit U2-3 place a minimum 242 bound on the age of event E Six of the events, E1 through E6, appear to be large surface-rupturing earthquakes. The 244 cumulative vertical displacement for these paleoearthquakes is about 3.2 m through restoration of 245 strata before and after each event shown in Figure 13, and consistent with the scarp height of ± 0.3 m obtained from survey profiles across the main fault zone. Therefore, the trench would 247 imply record a complete record of paleoseismic events since E Slip rate estimation of the western Tianjingshan fault 249 A displaced stream channel near the TC-5 trench site (Figure 7) affords an opportunity to 250 establish the fault slip rate. There are two stream channels incised into the hanging wall at this site 251 (Figure 14). The eastern of the two hanging wall channels is only ~30 m long, narrow in width, 252 and is not well developed as evidenced by a knickpoint at its head where headward erosion it is 253 actively migrating up the faulted alluvial fan surface (Figure 7D). The older stream channel 254 within the alluvial fan at this site extends ~700 m upward into Pleistocene alluvial deposits and 255 downstream to where it feeds into the Majiajing River (Figure 7B). The two drainages join on the 256 footwall. The smaller channel displaces a 2.7 m left-lateral offset in the channel wall. The older 257 channel on the hanging wall is offset about 11.7 ± 0.5 m (Figure 14B). Support for the correlation 12 / 22

13 258 and offset is further found in the observation that profiles of the older stream channel on both 259 sides of the fault match very well. In addition, to the horizontal offset, vertical offsets of 0.74 ± m and 0.80 ± 0.12 m are obtained by measuring offsets of two topographic profiles across the 261 north-side-up fault scarp (Figure 7C). The alluvial fan surface into which the 11.7 meter offset 262 channel is capped by an ~50-cm-thick layer of loess. We interpret that the alluvial deposits at the 263 base and below the loess layer represent the time stream incision initiated and the fan surface was 264 abandoned. An OSL sample (TJSH-65, Supplementary Table 2) collected in a ~0.8 m hand 265 excavated hole on the alluvial fan at the topmost of the sand layer (Figure 7D and 14A) place the 266 age of abandonment at 10.7 ± 1.0 kyr. The age places a maximum bound on the time at which the 267 lager stream was incised and began to record lateral offset. Dividing the observed 11.7 ± 0.5 m 268 offset by the age of fan abandonment 10.7 ± 1.0 kyr places a minimum value on the fault slip rate 269 equal to 1.1 ± 0.1 mm/yr. Similarly, the vertical fault slip rate is estimated at 0.2 ± 0.1 mm/yr Discussion 271 The observations from our trenching studies are summarized in the form of a space-time 272 diagram in Figure 15. The relative location of each trench along the fault zone is marked along 273 the horizontal axis and the boxes show the temporal constraints on the timing of earthquakes we 274 obtained from our trenching studies. Others have also conducted trenching studies along the fault 275 (Chen et al., 2006; Du et al., 2007; IGCEA et al., 2003; Min et al., 2001) and these are also 276 included in the plot of Figure 15. Within the limits of dating paleoearthquakes at each site, the 277 data allow that the entirety of the fault zone ruptured during the most recent earthquake. The 278 period 1.1 to 1.3 kyr corresponds to the thickness of the red line on Figure 15 and is the single 279 time period that satisfies the timing of the most recent paleoearthquake at the 10 trench sites. If 13 / 22

14 280 this is true, the sum of data allow us to suggest that the most recent event occurred between about to 1.3 kyr, not at the same time as the historical 1709 earthquake that has been documented on 282 the central Tianjingshan fault (Zhang et al., 1988, 2015; Nie et al., 1993; Figure 1B). Though 283 numerous trenches have now been excavated along the western Tianjingshan fault, limits on the 284 timing of penultimate and older earthquakes remain few. The largest number of paleoearthquakes 285 is observed in trench TC-7, where 6 surface rupture earthquakes are recorded over the last ~35, years. If these values are representative of the entire western Tianjingshan fault, and the fault 287 always breaks in its entirety, it appears the average return time of surface rupture earthquake is 288 about 5,000 years. If the fault does not always break its entire length, then the occurrence rate of 289 surface rupture earthquakes will be accordingly more frequent, and their rupture lengths and 290 magnitudes respectively smaller. 291 The length of the western Tianjingshan fault (e.g. Figure 1C) provides a basis to place limits 292 on the size of past earthquakes on the fault zone. The length of the western Tianjingshan fault is 293 about 60 km. If it is assumed that the entire fault ruptures, one may use published scaling 294 relationships between earthquake magnitude and rupture length. Using the regressions of Wells 295 and Coppersmith, 1994 and Wesnousky, 2008, entire rupture of the W-TJSF will produce an 296 earthquake of magnitude and moment equal to 7.15 and dyne cm, respectively. The 297 same approach can be used with the measures of repeated offset summarized in Figure 3, which 298 suggests offsets average around 3 meters when the W-TJSF ruptures. Using the regressions of 299 Wells and Coppersmith, 1994 and Wesnousky, 2008, an earthquake with average surface offset 300 of ~3 meters corresponds to an earthquakes of 7.47 magnitude and dyne cm seismic 301 moment. One may also compute the expected seismic moment of an earthquake rupturing the 14 / 22

15 entire 60 km length of the W-TJSF. The formulation for seismic moment = ulwd, where is the crustal rigidity ( dyne/cm 2 ), L is the fault length, W is the fault width, and D is the average 304 displacement. Assuming fault width is 15 km (Gao et al., 2013), the approximate depth of the 305 seismogenic layer for continental earthquakes, that the length is 60 km, and the average 306 displacement is 3 meters, the expected seismic moment is dyne cm, or Mw = ~ The slip rate of >1.1 mm/yr we calculate for the western Tianjingshan fault is comparable to 308 the fault slip rate of ~2.5 mm/yr and mm/yr proposed by previous studies on the central 309 and eastern segments of the Tianjingshan fault, respectively (Chai et al., 1997; Li et al., 2005). 310 These values are also in agreement with the present-day GPS rate of ~1-1.3 mm/yr along the 311 Tianjingshan fault (Gan et al., 2007; Li et al., 2009; Zheng et al., 2013). Results from deep 312 seismic reflection profiles suggest that the Tianjingshan fault has developed as a result of 313 northeastward migration of a large thrust ramp, and merges with and initiated subsequent to the 314 Haiyuan fault dipping SW with shallow angle in the middle-lower crust (Gao et al., 2013). Thus, 315 the Tianjingshan fault and the Haiyuan fault together accommodate left-lateral slip and crustal 316 shortening along the northeasternmost margin of Tibetan Plateau sourcing from the collision and 317 subsequent penetration of India into Eurasia. The left-lateral Haiyuan fault and Altyn-Tagh fault 318 were supposed to be major intracontinental boundary faults that controlled the eastward extrusion 319 of rigid blocks with poorly constrained slip-rate of ~12-20 mm/yr and ~20-30 mm/yr, respectively 320 (Peltzer and Tapponnier, 1988; Gaudemer et al., 1995; Meyer et al., 1998; Lasserre et al., , 2002; Mériaux et al., 2004, 2005; Gold et al., 2009). However, numerous geologic, GPS 322 and InSAR studies suggest slip-rates of the Haiyuan fault and the Altyn-Tagh fault are ~4-6 mm/yr 323 and ~10 mm/yr, respectively (Zhang et al., 1988, 2004, 2007; Yuan et al., 1998; Bendick et al., 15 / 22

16 ; Shen et al., 2001; Wallace et al., 2004; Cowgill et al., 2007, 2009; Thatcher, 2007; 325 Cavalié et al., 2008; Elliott et al., 2008; Li et al., 2009; Loveless et al., 2011). It is important 326 that the sum of slip rate ~5-7 mm/yr of the Tianjingshan fault and the Haiyuan fault is only 327 ~30%-50% of that assumed in the eastward extrusion model, and the same is true along the 328 Altyn-Tagh fault. Therefore, our observations are more likely to favor a lower slip-rate model that 329 crustal deformation in NE Tibet not only distributes along major strike-slip faults, but also may 330 scatters in the internal deforming crustal material Conclusions 332 Our offset measurements record up to 6 surface-rupturing events on the W-TJSF, with a 333 similar coseismic slip ~3-4 m for successive earthquakes. Trenching investigations along the 334 W-TJSF further suggest that ~6 paleoearthquakes may have occurred during the last 35 kyr, with 335 an average recurrence interval of ~5,000 yr for the Holocene earthquakes. The most recent 336 earthquake on the W-TJSF has occurred ~1.2 ± 0.1 kyr BP, indicating that the W-TJSF did not 337 break in the 1709 earthquake that is known to have ruptured the C-TJSF. Here, we provide the first 338 quantitative estimate of a slip rate ( mm/yr) for the W-TJSF. The relatively high slip rate of 339 the Tianjingshan fault shows that is important in accommodating the sinistral displacement and 340 crustal shortening deformation in NE Tibet. 341 Acknowledgments 342 This work was jointly supported by the National Science Foundation of China ( , , , and ), the Fundamental Research Funds in the Institute of 344 Geology, China Earthquake Administration (IGCEA1220). We thank Olaf Zielke and Zhikun Ren 16 / 22

17 345 for comments that improved a previous version of the manuscript, and Stephen J. Angster for his 346 help on the language of the manuscript. Longsheng Zhang and Hongdong Wang are thanked for 347 their help in field work, and Changsheng Wang for his help in processing the OSL dating samples. 348 The supporting information includes a supplementary material of methodology, offset 349 measurements, and trenching investigations on the western Tianjingshan fault. For other data 350 requests, please contact the corresponding author in China (chuanyou@ies.ac.cn). 351 References 352 Bendick, R., Bilham, R., Freymueller, J., Larson, K., and Yin, G., (2000), Geodetic evidence for a low slip rate in 353 the Altyn Tagh fault system: Nature, v. 404, p doi: / Cavalié, O., Lasserre, C., Doin, M.P., Peltzer, G., Sun, J., Xu, X., and Shen, Z.K. (2008), Measurement of 355 interseismic strain across the Haiyuan fault (Gansu, China), by InSAR. Earth and Planetary Science Letters, (3), doi: /j.epsl Chai, C.Z., Zhang, W.Q., and Jiao, D.C. (1997), Discussion of the level active severity in different time intervals 358 and segments on Late Quaternary along Tianjingshan Fault zone. Earthquake Research in China, 13 (1), Chai, C.Z., Jiao, D.C., Liao, Y.H., Zhang, S.Y., Du, P., and Shen, W.H. (2003), Discovery of surface rupture zone 361 produced by Guanguanling earthquake at the juncture of Ningxia, Inner Mongolia and Gansu Province. 362 Seismology and Geology, 25 (1), Chen, G.X., Tian Q.J., Zhou. B.G., Min, W., Liu, B.J., and Gao, Z.W. (2006) Some seismo-geological questions 364 regarding to Daliushu Dam Site in Heishanxia Valley of Yellow River. Technology for Earthquake Disaster 365 Prevention, 1 (3), Cowgill, E., 2007, Impact of riser reconstructions on estimation of secular variation in rates of strike-slip faulting: 17 / 22

18 367 revisiting the Cherchen River site along the Altyn Tagh Fault, NW China: Earth and Planetary Science Letters, 368 v. 254, p doi: /j.epsl Cowgill, E., Gold, R. D., Xuanhua, C., Xiao-Feng, W., Arrowsmith, J. R., and Southon, J. (2009). Low Quaternary 370 slip rate reconciles geodetic and geologic rates along the Altyn Tagh fault, northwestern Tibet. Geology, 37(7), doi: /G25623A Du, P., Chai C.Z., Shen W.H, Min, W., and Tian, Q.J. (2007). Characteristics of the Holocene activity of the 373 Guanguanling Fault zone. Seismology and Geology, 29 (3), Elliott, J.R., Biggs, J., Parsons, B., and Wright, T.J., (2008). InSAR slip rate determinations on the Altyn Tagh 375 Fault, northern Tibet, in the presence of topographically correlated atmospheric delays: Geophysical Research 376 Letters, v. 35, L doi: /2008GL Gan, W.J., Zhang, P.Z., Shen, Z.K., Niu, Z.J., Wang, M., Wan, M.G., Zhou, D.M., and Cheng, J. (2007), 378 Present-day crustal motion within the Tibetan Plateau inferred from GPS measurements. Journal of 379 Geophysical Research, 112 (B8), doi: /2005jb Gao, R., Wang, H.Y., Yin, A., Dong, S.W., Kuang, Z.Y., Zuza, A.V., Li, W.H., and Xiong, X.S. (2013), Tectonic 381 development of the northeastern Tibetan Plateau as constrained by high-resolution deep seismic-reflection 382 data. Lithosphere, 5 (6), doi: /l Gaudemer, Y., Tapponnier, P., Peltzer, G., Meyer, B., Guo, S., Chen, Z., Dai, H., Cifuentes, I., Partitioning of 384 crustal slip between linked, active faults in the eastern Qilian shan, and evidence for a major seismic gap, the 385 Tianzhu gap, on the westernhaiyuan Fault, Gansu (China). Geophysical Journal International 120 (3), Gold, R.D., Cowgill, E., Arrowsmith, J.R., Gosse, J., Chen, X., and Wang, X.-F., 2009, Riser diachroneity, lateral 388 erosion, and uncertainty in rates of strike-slip faulting: A case study from Tuzidun along the Altyn Tagh Fault, 18 / 22

19 389 NW China: Journal of Geophysical Research, 114 (B4). doi: /2008JB Gu, G.X., Lin, T.H., and Shi, Z.L. (1983), The Catalogue of Chinese Earthquakes (1831 BC AD 1969). Science 391 Publishing House, Beijing. 392 Institute of Geophysics, CEA., Analysis and Predicting Center, SSB., Institute of Geology, CEA., Institute of 393 Crustal and Dynamics, CEA., and Geophysical Exploration Center, CEA. (2003), Seismogeological 394 supplementary work report at Heishanxia section of the Yellow River. Unpublished results. 395 Lasserre, C., et al. (1999), Postglacial left slip rate and past occurrence of M 8 earthquakes on the western 396 Haiyuan fault, Gansu, China, Journal of Geophysical Research, 104, 17,633 17,652, 397 doi: /1998jb Lasserre, C., Y. Gaudemer, P. Tapponnier, A.-S. Me riaux, J. Van der Woerd, Y. Daoyang, F. J. Ryerson, R. C. 399 Finkel, and M. W. Caffee (2002), Fast late Pleistocene slip rate on the Leng Long Ling segment of the 400 Haiyuan fault, Qinghai, China, Journal of Geophysical Research, 107(B11), doi: /2000jb Li, C.Y. (2005), Quantitative studies on major active fault zones in Northeastern Qinghai-Tibet Plateau. Institute of 403 Geology, CEA, Beijing (PhD dissertation). 404 Li, C.Y., Zhang, P.Z., Yin, J.H., and Min, W. (2009), Late Quaternary left-lateral slip rate of the Haiyuan fault, 405 northeastern margin of the Tibetan Plateau. Tectonics, 28 (TC5010). doi: /2008tc Li, X.N and Li, C.Y. (2015), Characteristic slip behavior analysis of the western segment of the 407 Xiangshan-Tianjingshan fault zone since late Quaternary. Seismology and Geology, 37(2), doi: /j.issn Loveless, J.P and Meade, B.J. (2011), Partitioning of localized and diffuse deformation in the Tibetan Plateau from 410 joint inversions of geologic and geodetic observations. Earth and Planetary Science Letters, 303(1), / 22

20 411 doi: /j.epsl Lu, Y.C., Wang, X.L., and Wintle, A.G. (2007), A new OSL chronology for dust accumulation in the last 130, yr for the Chinese Loess Plateau. Quaternary Research, 67(1), doi: /j.yqres Me riaux, A.-S., F. J. Ryerson, P. Tapponnier, J. van der Woerd, R. C. Finkel, X. Xu, Z. Xu, and M. W. Caffee 415 (2004), Rapid slip along the central Altyn Tagh fault: Morphochronologic evidence from Cherchen He and 416 Sulamu Tagh, Journal of Geophysical Research, 109, B doi: /2003jb Me riaux, A.-S., et al. (2005), The Aksay segment of the northern Altyn Tagh fault: Tectonic geomorphology, 418 landscape evolution, and Holocene slip rate, Journal of Geophysical Research, 110, B doi: /2004jb Meyer, B., Tapponnier, P., Bourjot, L., Metivier, F., Gaudemer, Y., Peltzer, G., Shunmin, G., Chen, Z., Crustal thickening in Gansu-Qinghai, lithospheric mantle subduction, and oblique, strike-slip controlled 422 growth of the Tibet Plateau. Geophysical Journal International 135, doi: /j X x. 424 Min, W., Zhang, P.Z., and Deng, Q.D. (2001), The study of Holocene Paleoearthquakes on Zhongwei-Tongxin 425 Fault zone. Seismology and Geology, 23 (3), Nie, Z and Lin, W.F. (1993), Middle segment of the Zhongwei-Tongxin fault zone: seismic deformation band of earthquake with M=71/2 along Xiangshan-Tianjingshan fault zone. Earthquake, 13(1), Peltzer, P., Tapponnier, P., (1988). Formation and evolution of strike-slip faults, and basins during the India-Asia 429 collision an experimental approach. Journal of Geophysical Research 93 (B10), /JB093iB12p Shen, Z.-K., Wang, M., Li, Y., Jackson, D.D., Yin, A., Dong, D., and Fang, P., (2001). Crustal deformation along 432 the Altyn Tagh fault system, western China, from GPS: Journal of Geophysical Research, v. 106, p. 20 / 22

21 433 30,607 30,621. doi: /2001JB Wallace, K., Yin, G., and Bilham, R., (2004). Inescapable slow slip on the Altyn Tagh fault: Geophysical Research 435 Letters, v. 31, L doi: /2004GL Wang, Y.P., Song, F.M., Li, Z.Y., You, H.C., and An, P. (1990), Study on recurrence intervals of great earthquakes 437 in the Late Quaternary of Xiangshan-Tianjingshan Fault zone in Ningxia. Earthquake Research in China, (2), Wells, D.L and Coppersmith, K.J. (1994), New empirical relationships among magnitude, rupture length, rupture 440 width, rupture area, and surface displacement. Bulletin of the Seismological Society of America, 84, Wesnousky, S.G. (2008), Displacement and geometrical characteristics of earthquake surface ruptures: Issues and 443 implications for seismic-hazard analysis and the process of earthquake rupture. Bulletin of the Seismological 444 Society of America, 98(4), doi: / Working Group of M7. (2012), Study on the mid- to long-term potential of large earthquakes on the Chinese 446 Continent. Seismological Press, Beijing. 447 Yang, H.L., Chen, J., Thompson, J.A., and Liu, J.F. (2012), Optical dating of the 12 May 2008, Ms 8.0 Wenchuan 448 earthquake-related sediments: Tests of zeroing assumptions. Quaternary Geochronology, 10, doi: /j.quageo Yuan, D. Y., Liu, B. C., Lu, T. Y., He, W. G., Liu, X. F., and Gan, W. J., Study of the segmentation in east 451 segment of the northern Qilianshan fault zone: Northwestern Seismological Journal, v. 20, p Zhang, P.Z., Molnar, P., Burchfiel, B.C., Royden, L., Wang, Y.P., Deng, Q.D., Song, F.M., Zhang, W.Q., and Jiao, 453 D.C. (1988), Bounds on the Holocene slip rate of the Haiyuan fault, north-central China. Quaternary Research, (2), doi: / (88) / 22

22 455 Zhang, P.-Z., Shen, Z., Wang, M., Gan, W., Bürgmann, R., Molnar, P., Wang, Q., Niu, Z., Sun, J., Wu, J., Sun, H., 456 and You, X., 2004, Continuous deformation of the Tibetan Plateau from global positioning system data: 457 Geology, v. 32, p , doi: /G Zhang, P.-Z., Molnar, P., and Xu, X., 2007, Late Quaternary and present-day rates of slip along the Altyn Tagh 459 Fault, northern margin of the Tibetan Plateau: Tectonics, v. 26, TC5010, doi: /2006TC Zhang, W.Q., Jiao, D.C., Chai, C.Z., Song, F.M., and Wang, Y.P. (1988), Neotectonic features of the 461 Xiangshan-Tianjingshan arc fracture zone and the seismic deformation zone of 1709 south of Zhongwei 462 M=7½ earthquake. Seismology and Geology, 10 (3), Zhang, W.Q., Jiao, D.C., and Chai, C.Z. (2015), The Tianjingshan active fault zone. Seismological Press, Beijing. 464 Zheng, W.J., Zhang, P.Z., He, W.G., Yuan, D.Y., Shao, Y.X., Zheng, D.W., Ge, W.P., and Min, W. (2013), 465 Transformation of displacement between strike-slip and crustal shortening in the northern margin of the 466 Tibetan Plateau: Evidence from decadal GPS measurements and late Quaternary slip rates on faults. 467 Tectonophysics, 584, doi: /j.tecto Zhou, J.X. and Liu, B.C. (1987), The research of active Zhongwei-Tongxin fault. Northwestern Seismological 469 Journal. 9 (3), Zielke, O., Klinger, Y., and Arrowsmith, J.R. (2015), Fault Slip and Earthquake Recurrence Along Strike-Slip 471 Faults--Contributions of High-Resolution Geomorphic Data, Tectonophysics, 638, 43-62, 472 doi: /j.tecto / 22

23 Figures 1 2 Figure captions 3 4 Figure 1. Index map of study area in Tibet Plateau. The black box shows the location of Figure 1B. 5 (B) Topographic and tectonic map of northeastern Tibet with the major active faults. Sections of 6 fault discussed in text are labeled: W-TJSF, the western Tianjingshan fault; C-TJSF, the central 7 Tianjingshan fault; E-TJSF, the eastern Tianjingshan fault; NCLF, the northern Changlingshan 8 fault; Red circles are epicenters of earthquakes (M 3) from China Earthquake Networks Center 9 catalog ( and rom Gu et al Dashed black box shows the location 10 of Figure 1C. (C) Active fault traces of the western Tianjingshan fault. F1, Jingtai subsegment; F2, 11 Guanguanling subsegment; F3, Shajing subsegment; F4, Zhongwei subsegment; F5, 12 Qinshan-Gushanzi subsegment. Red bars are trenches excavated in this study. Black bars are 1 / 15

24 13 trenches excavated in previous studies. Numbers correspond trenches, TC-1 through TC Figure 2. (A) Geological map and fault trace of the western Tianjingshan fault. Modified from 17 IGCEA et al Small black boxes showing the geomorphologic mapping areas in Figure B-E. 18 (B-E) Surficial geologic maps conducted with Google Earth imagery and field observations. 19 Circles (a-h) located in the lower-left corners of Figures B-E contain enlarged maps of 20 left-laterally displaced stream channels at sites marked by dashed black circles (a -h ) in 21 respective figures / 15

25 23 24 Figure 3. (A) Plot of 240 horizontal offset measurements distributed as a function of distance 25 along strike three major subsegments of the western Tianjingshan fault. Offset values are color 26 coded to show displacements interpreted result from the same number of paleoearthquakes. 27 Vertical dashed lines and symbols with dark background indicate sites of repeated offset at 28 individual site. (B) Frequency counts for 240 offset measurements (See Appendix table 2) 29 calculated using a non-overlapping sliding window by step of 1 m and histogram for number of 30 offsets per m displacement. Each peak (~3, 6, 9, 12, 16, and 20 m) on the fitting curve is 31 interpreted to represent the average offset of multiple earthquakes. The difference in values of 32 peaks averages about 3 m / 15

26 34 35 Figure 4. Geomorphological mapping and features of the westernmost portion of F1 (Jingtai 36 subsegment). (A) Morphotectonic map showing fault trace, geomorphological units and stream 37 channels. Note that fault trace terminates near the pressure ridge and the great wall (AD 1598) was 38 not faulted by the surface trace. See location in the lower-left sketch map. Black bars show the 39 trench location of Figure 7B and Figure 7C. (B) Photograph showing fault scarp truncating the 40 youngest alluvial fan, trench site of TC-1 ( N, E), and relation between 41 fault trace and the great wall. (C) View of north-facing scarp of ~1 m and trench site of TC-2 42 ( N, E) / 15

27 44 45 Figure 5. Trench logs and dated stratigraphy of TC-2. Red lines mark the faults, which form two 46 major fault zones. The stratigraphy of TC-2 is similar to that observed in TC-1 (Table 1). 47 Stratigraphic position of OSL samples are shown by black circles. Dashed corners on the east wall 48 of trench are further enlarged in Figure 10 to show details of stratigraphic units and crosscutting 49 relationship. See text for description of evidence of interpreted event / 15

28 52 53 Figure 6. Photomosaic of part of the west wall of TC-2 showing evidence of the interpreted 54 earthquake event. (A) The fault strand F3-4 that folded units of U1-1/2 to U1-3 and terminated 55 into U1-4 provides evidence for E1. (B) The other evidence for E1 is indicated by fault strand 56 F1-2 that ruptured the ground surface. Note that total vertical displacement of E1 is about 1 m 57 consistent with that measured across the fault scarp, suggesting that the youngest alluvial fan may 58 only record the most recent earthquake event / 15

29 61 62 Figure 7. Geomorphological mapping and features of the central portion of F2 (Guanguanling 63 subsegment). (A) Part of the Google Earth imagery of the central portion, with precision of 10 m. 64 (B) Corresponding morphotectonic map showing fault trace, geomorphological units and stream 65 channels. See location in the lower-left sketch map. Black box shows the trench location of Figure 66 11C. (C) Topographic map showing the fault scarp, offset channels and trench site. Contour 67 interval is 0.2 m. Profiles across the fault zone show the height of the scarp of ~0.8 m. (D) North 68 looking view of the interpreted faulting landscape at the trench site ( N, E). 69 The alluvial fan are incised by the older (left) and younger (right) stream channels. Location of pit 70 OSL sample TJSH-65 is shown by the black circle and arrow / 15

30 72 73 Figure 8. Trench logs of the east (A) and west (B, reflected) walls of TC-5 depicting the 74 stratigraphy and deformation associated with interpreted events, E1 through E4. Faults are shown 75 with thick red lines. Sedimentary units are indicated by different colors and are labeled 76 numerically (see detail in Table 1). Stratigraphic position of OSL samples are shown by black 77 circles. Liquefaction is only expressed on the east wall; note how unit 5 is deformed. Dashed 78 corners correspond to the outlines of photographs in Figure 13. See text for description of 79 evidence of individual events. 8 / 15

31 80 81 Figure 9. Photographs of parts of the east and west walls showing evidence of earthquake events. 82 (A) The fault that offset unit 7 (left side) provide the evidence for E1. E2 is characterized by 83 several strands faulting and folding the well-layered strata (right side). (B-C) Evidence for E3 is 84 indicated by small grabens on both walls of the trench. (D) Liquefaction of unit 5 shows the 85 evidence of E4, the oldest earthquake revealed in trench. 9 / 15

32 / 15

33 88 Figure 10. Possible restoration of stratigraphy on east wall of trench TC-5 showing deformation 89 associated with events E1 to E4. (A) Present situation after the deposit of U11; (B) E1, rupturing 90 of the latest earthquake along F3 and the forming of fault scarp; (C) Deposition of U10; (D) 91 Erosion of U9 at the upthrown wall and deposition of U9 at the downthrown wall; (E) E2, 92 rupturing of the penultimate earthquake along F4-F7 simultaneously and forming the geomorphic 93 scarp; (F) Successive deposition of U7-U9; (G) E3, forming of small extensional graben between 94 F1 and F2; (H) Deposition of U6; (I) E4, liquefaction and fault-controlling sand funnel and 95 channel; (J) Possible original situation and the successive deposition of U2 U Figure 11. Geomorphological mapping and features of the central portion of F4 (Zhongwei 100 subsegment). (A) Morphotectonic map showing fault trace, geomorphological units and stream 101 channels. See location in the lower-left sketch map. Black bar show the trench location of TC ( N, E). (B) 3-D model showing the fault scarp and trench site, conducted 11 / 15

34 103 using high-resolution DGPS surveying. Profiles across the fault zone show the height of the scarp 104 of ~3.46 m and ~2.87 m, respectively. (C-D) South looking view of fault scarp and trench site of 105 TC-7. See Figure 17A for location Figure 12. Logs of the west wall of TC-7 depicting the stratigraphy and deformation associated 109 with events, E1 through E6. Red lines indicate fault strands. Sedimentary units are indicated by 110 different colors and are labeled numerically (Table 1). Stratigraphic position of OSL samples are 111 shown by black circles. See text for description of evidence of individual events / 15

35 Figure 13. Possible restoration of stratigraphy on west wall of trench showing deformation 13 / 15

36 115 associated with events E1 to E6. (A) Present situation after the deposit of U7-2; (B) Post-faulting 116 deposition of U7-1; (C) E1, rupturing of the most recent earthquake along F2; (D) Deposition of 117 U6-2; (E) Post-faulting deposition of colluvial wedge of U6-1; (F) E2, rupturing along F2 and 118 forming of scarp; (G) Continuous deposition of U5 after earthquake event E3; (H) E3, rupturing 119 along F3 and F4; (I) Earlier deposition of U5; (J) Post-faulting deposition of colluvial wedge of 120 U4; (K) E4, rupturing along F2 and forming of scarp; (L) Deposition of U3-2 and U3-3; (M) 121 Post-faulting deposition of fissure-fill colluvium of U3-1; (N) E5, rupturing along F3; (O) 122 Deposition of U2-2 and U2-3; (P) Post-faulting deposition of colluvial wedge of U2-1; (Q) E6, 123 rupturing along F2 and forming of scarp; (R) Possible original situation Figure 14. Offset reconstruction at Majiajing site. (A) Topographic map of displaced stream 127 channels and longitudinal profiles showing the possible fault geometry and vertical displacement 128 of ~0.3 m. Knickpoints are labeled as K1 and K2. (B) Retro-deforming the image by ~11.7±0.5 m 14 / 15

37 129 along the fault trace realigns the older stream channel and west edge of the fan. The cross-sections 130 along the dashed red and blue lines projected to the fault plane showing the well-matched 131 upstream and downstream channels Figure 15. Paleoearthquake sequence constrained by trenching investigations (Min et al., 2001; 135 Institute of Geophysics, CEA et al., 2003; Chen et al., 2006; Du et al., 2007 and this study) 136 along the western Tianjingshan fault. Red lines indicate fault strands. (Upper) Map showing 137 geometric pattern of the five segments and trench locations of the western Tianjingshan fault. Red 138 bars are trenches excavated in this study. Black bars are trenches excavated in previous studies. 139 (Lower) Temporal and spatial pattern of paleoearthquakes occurred along the western 140 Tianjingshan fault. From the youngest to oldest, the events on the western segment are named EI, 141 EII, EIII, EIV, EV, and EVI, respectively. Note that EI corresponds to the most recent earthquake 142 of ~1200 kyr BP that completely ruptured the western Tianjingshan fault. 15 / 15

38 Supplementary Material Supplementary Material Paleoseismology and Slip Rate of the Western Tianjingshan Fault of NE Tibet, China Xinnan Li 1,2, Chuanyou Li 2, Steven G. Wesnousky 3, Peizhen Zhang 1,4, Wenjun Zheng 1,4, Ian K.D. Pierce 3, and Xuguang Wang 2 1 State Key Laboratory of Earthquake Dynamics, Institute of Geology, China Earthquake Administration, Beijing , China 2 Key Laboratory of Active Tectonics and Volcano, Institute of Geology, China Earthquake Administration, Beijing , China 3 Center for Neotectonic Studies, Mail Stop 169, University of Nevada, Reno, Nevada 89557, USA 4 School of Earth Science and Geological Engineering, Sun Yat-Sen University, Guangzhou, , China Supplementary Materials page 1/16

39 Supplementary Material S1 and Figures S1, S2, and S3 Offset Measurement. We made 240 measurements of offset geomorphic features along the western Tianjingshan fault. The traces F3 and F5 are commonly modified by surface mining and so our measurements are limited to those made along segments F1, F2 and F4. The uncertainties in these field-based offset measurements mostly depend largely on the geometrical shapes of displaced geomorphic features (e.g. offset features oblique to the fault have less certainty). The offset measurements were collected using tape measurers and laser aperture rangefinders in the field. The results of offset we report at each site are the average of at least two or three measurements of a single offset feature (e.g. channel margins, thalwegs, and ridge crests of the same offset.). These offsets were also checked against offsets measured from high-resolution images or DEM-derived hillshade-, slope-, and contour-plots. Some of the smaller single-event offsets were not well-identified in these images, so were only measured in the field. The best offset measurements are those that were made of displaced features that trend perpendicular to the fault on both sides of the fault trace. Examples of offset features observed in the field, from Google Earth imagery, and in structure-from-motion (SfM) models constructed from airphotos are shown in Figures S1, S2, and S3, respectively. The offset data are split into three categories based on our judgement of the quality of the measurements: high, medium and low. High-resolution hillshade-, slope-, and contour-plots of some displaced features were generated from the digital elevation models (DEMs, 0.05m/pix) developed from structure-from-motion (SfM) processing of UAV photography. Using the MATLAB-based LaDiCaoz software developed by Zielke and Arrowsmith (2012), we obtain optimal measurement of the offset of displaced feature by back slipping the hillshade-, slope-, and contour-plot (Figure S3). The quality of the offset Supplementary Materials page 2/16

40 measurement is determined from a confidence interval and thus the maximum and minimum offsets can be obtained. Detailed descriptions of specific operational procedures are shown in the paper published by Haddon et al. (2016), and Ziekle et al. (2015) ( Same method of offset measurement can be used in DEM data conducted using real-time kinematic (RTK) global positioning system (GPS). Figure S1. Field photographs of stream channel displacements observed on the western Tianjingshan fault. In each example, the stream channel is marked as a blue curve with an arrow Supplementary Materials page 3/16

41 showing the flowing direction. The active fault trace is marked as a red curve showing the motion of the fault. Estimates of offset measured with tape measure are annotated in each figure. Figure S2. Example of successive reconstructions of offset stream channels on F1 (Zhongwei subsegment). Stream channels are labelled alphabetically and numerically on the north and south sides of the fault trace, respectively. Staring from current geometry, successive restorations are then presented by moving one side to realign channels truncated as the result of successive earthquakes. Each offset is determined by restoring continuity of channels (black circles) on both sides of the fault. The error was defined by quality rating (high, moderate, and low) based on the number of channels matched each other, see Appendix table 2. Successive offsets of ~2.4, 6.4, 9.2, 11.8, 16.8, and 20.1 m are identified. Supplementary Materials page 4/16

42 Figure S3. Example of LaDiCaoz-based offset measurement derived from UAV photography and SfM method. (A) Hillshade map of displaced channels and ridges near Majiajing site. White box shows the location of Figure 5B. (B) A close-up of the potentially displaced ridge that represented by hillshade imagery overlapped by 0.5 m contour plot with fault trace (in turquoise) and profile lines (in red and blue). (C) Back-slipped hillshade plot of the ridge; back-slipped by the optimal displacement value of 7.3 m using MATLAB tools. (D) Topographic profiles along the red and blue lines projected to the fault plane showing the optimal back-slipped horizontal offset of 7.3 m and vertical offset of 0.2 m. OSL Dating. Optically stimulated luminescence (OSL) ages of sediment samples collected from the trenches are used to constrain the ages of the stratigraphic units and hence the interpreted events. The OSL samples were collected using stainless steel tubes (20 cm long and 5 cm diameter). Tubes were hammered into the sediment and after complete filling both ends were Supplementary Materials page 5/16

43 immediately sealed with aluminum foil and taped to prevent light leakage and loss of water during transport and storage. All samples were processed and measured in the Luminescence Dating Laboratory of State Key Laboratory of Earthquake Dynamics, Institute of Geology, China Earthquake Administration, Beijing. The samples were extracted from the tubes under subdued red light and separated using conventional pretreated techniques (see Lu et al., 2007 and Yang et al., 2013 for detailed experimental operating procedures) for 4-11 um fine-grained quartz fraction and for um coarse-grained material. The equivalent dose (De) of pure fine-grained quartz were measured with sensitivity-corrected multiple-aliquot regenerative-dose (SMAR) protocol (Lu et al., 2007) on an automated Daybreak 2200 TL/OSL with photon detection through two U-340 glass filters equipped with a calibrated 90Sr/90Y beta radiation source. The dose rates for U, Th, and their daughter products of coarse-grained material were measured using a Daybreak 582 thick source alpha counter. Then, dose rates and De were used to calculate OSL ages of sediment samples based on the conventional formula. Supplementary Material S2 and Figure S4 Trench 1 (TC-1). At this site, the approximately EW-trending surface trace is the westernmost portion of the Jingtai subsegment (F1), and ends about 600 m to the west near a pressure ridge (Figures 1C and S4A). The fault trace cuts across a late Pleistocene to Holocene alluvial fan, with an about 1-m-high, down-to-the-north fault scarp. Part of the Ming dynasty Great Wall (built in AD1598) was built striking NNW across the fault trace. A 7-m-long trench, TC-1 was excavated about 60 m east of the Great Wall (Figure S4B). Approximately 160 m east of TC-1, another 12-m-long trench, TC-2, was excavated across the same fault trace, associated with a north-facing Supplementary Materials page 6/16

44 scarp ranging up to 1 m high in places (Figure S4C). The trench exposure show one major south steep dipping fault strand F1 that cut coarse-grained alluvial deposits and loess (Figure S4). The fault strand F1 extends from the base of the trench to the surface. The most recent displacement on F1 postdates deposition of U1-4 for which OSL sample TJSH-11 places the timing of deposition at about 6.2 ± 0.6 kyr (Figure S4). The fault does not displace the Great Wall built in 1598 A.D. The most recent paleoearthquake E1 is thus limited to between 6.2 ± 0.6 kyr and 1598 A.D. The wedge-shaped form of unit U1-3 is the result of erosion during emplacement of overlying unit U1-4 and not interpreted to record an earlier event. Figure S4. Trench logs of the west wall of TC-1. Faults are shown with thick red lines. Sedimentary units are indicated by different colors and are labeled numerically (Supplementary Table 1). Stratigraphic position of OSL samples are shown by black circles. See text for description of evidence of interpreted event. Supplementary Material S3 and Figures S5-S6 Trench 6 (TC-6). The TC-6 site is located about 3.4 km east of Majiajing on the approximately east-west striking active fault trace (Figures 1C and 7). The field photographs in Figures S5A-B show a prominent linear valley, that is ~1.1 km in length and ~30-70 m in width. The fault trace Supplementary Materials page 7/16

45 controls the northern boundary of the valley and extends several kilometers farther to the east. A 6-m-long trench, TC-6, was excavated across a south-facing scarp about 1 m in height (Figure S5C). TC-6 trench exposed well-bedded, predominantly sandy and silty strata. Individual unit descriptions are provided in Supplementary Table S1. Prominent are two south-dipping fault strands F1 and F2. Strand F1 is capped by 1.2 m of sandy gravels (units U2-1 and U2-2) (Figure S6). Strand f2 cuts unit U1-2 and forms the northern boundary of a colluvial wedge unit U3-1, and thus records a younger displacement that recorded on F1. The 90.8 ± 4.4 kyr age of OSL sample TJSH-08 taken from unit U1-1 predates displacement on F1 and place a maximum age bound on all displacement in the trench (Supplementary Table S2). The 4.1 ± 0.4 kyr age of sample in unit U2-3 postdates displacement on F1 and predates displacement on F2. The 1.3 ± 0.2 kyr age of TJSH-10, taken from the unbroken capping unit U3-2 postdates displacement on both F1 and F2. The trench thus records two paleoearthquakes with the oldest bracketed in time between 90.8 ± 4.4 kyr and 4.1 ± 0.4 kyr and the youngest between 4.1 ± 0.4 kyr and 1.3 ± 0.2 kyr. Supplementary Materials page 8/16

46 Figure S5. Photographs of geomorphological features at trench site of TC-6 on F2 (GGL subsegment). (A-B) East and west looking views of linear valley, fault scarp and trench site of TC-6 ( N, E). Note that fault trace controls the northern boundary of the linear valley. TC-6 was excavated near the north side of the valley. (C) Photograph of trench exposure on the west wall, showing sediments separated by the fault strand. The height of fault scarp is about 1 m. Supplementary Materials page 9/16

47 Figure S6. Trench logs of the west wall of TC-6. Red lines indicate fault strands. The stratigraphy of TC-6 is indicated by different colors and are labeled numerically (Table S1). Stratigraphic position of OSL samples are shown by black circles. The structure and stratigraphy record evidence for two paleoearthquakes. The most recent earthquake event, E1, is indicated by the fault strand F2 that juxtaposed units U1-1 and U2-3 against unit U1-2, and by the wedge-shaped deposit of unit U3-1. Evidence for event E2 is characterized by the abrupt upward termination of F1, overlain by unit U2-1. Supplementary Table S1: Unit descriptions for accompany trench logs Trench Unit Description site TC-1 and U1-1 Generally massive beds of poorly sorted subangular boulder gravels occasionally with fine sand lenses. TC-2 U1-1/2 Layer of loess and pebble gravels between U1-1 and U1-2 appears to be gradational. U1-2 Dark-gray massive homogeneous soiled sandy silt interpreted as loess or reworked loess. U1-3 Very poorly sorted subangular pebble gravels with coarse sand. U1-4 Gray massive silty loess or reworked loess occasionally with small pebble gravels. U2 Clast-supported, thin gravel layer of angular to subangular pebbles. TC-5 U1 Black consolidated coal-like bedrock observed along the fault trace. U2 Brownish-red fine clayed sandstone. U3 Light brownish-red course sand to medium-course pebbles gravels. U4-1 Tan fine-grained sand interbedded with beige silty sand, with occasional course pebbles to cobbles. Supplementary Materials page 10/16