Red River and associated faults, Province, : Quaternary geology, slip rates, and seismic hazard

C. R. ALLEN Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, California 91125 A. R. GILLESPIE Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California 91109 HAN YUAN Seismological Bureau of Yunnan Province, , China K. E. SIEH Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, California 91125 ZHANG BUCHUN Institute of Geology, State Seismological Bureau, Beijing, China ZHU CHENGNAN Seismological Bureau of Yunnan Province, Kunming, China

ABSTRACT branch, in large part previously unrecognized, associated faults of Yunnan Province, Chin¡ traverses principally deeply dissected Cenozoic (Fig. 1). The Red River fault is one of the grea The 900-km-long right-slip Red River fault of valley fill northeast of the range-front fault and regional faults of China and has long been rec southernmost China and northern Vietnam is a has undergone almost pure lateral slip. Lateral ognized as a profound geological discontinuity profound structural discontinuity that is me- postfill offsets along the range-front branch di- marking the southwest margin of the Yangz chanically associated with the collision of the minish toward the southeast, whereiis those (Yangtze) Platform (Ministry of Geology, 1979) Indian and Eurasian plates. Although history along the mid-valley branch diminish northwest- It is particularly striking on satellite images fo records no large earthquakes resulting from ward; the net effect is that the total postfill some 800 km northwest from near , Viet slippage along at least the principal segment offset across both branches is almost uniform. nam, to the vicinity of Xiaguan, Yunnan Pro of the fault in China, youthful landforms and The Red River and its major tributaries ap- vince (Fig. 2) (Tapponnier and Molnar, 1977 disruptions of young sedimentary rocks indicate pear to have experienced about 5.5 km of right York and others, 1976). Regional geologic map! that it has generated large earthquakes during slip since the beginning of a major episode of (Ministry of Geology, 1962) show many othei the Pleistocene and Holocene epochs. The his- incision that continues to the present day. Res- northwest-trending faults in Yunnan, but non( toric quiescence thus must be regarded as being toration of this offset provides a remarkable seems to have the continuity, linearity, and de- indicative of a current seismic gap, although the alignment of most large tributaries as well as gree of geologic activity that characterize the recurrence interval between major earthquakes removing a major kink in the course of the Red Red River fault, and thus field efforts were coni is evidently much longer than for many other River itself. Using maximum credible rates of centrated on the Red River fault in this study. | major active fault systems. incision, we estimate an average fault-slip rate of A principal quandary regarding the Reel That recent displacement has been primarily 2 to perhaps 5 mm/yr. At this long-term rate of River fault has been its degree of current seismic right lateral is indicated by consistently dis- slip, the smallest offsets observed along the fault activity and hazard, if any. Although the faul' placed drainages, ranging in offset from 9 m to 6 (9 m) would occur no more frequently than superficially appears very similar to other majoi km, and the freshness of the smallest and most every 1,800 to 4,500 yr on the average. This is active faults, it has produced no significan recent offsets implies repeated Holocene move- consistent with the historical record of fault earthquakes within the long historic record, a 1 ments. Although physiographic features typical dormancy for the past 300 yr. least southeastward from Midu. One of th( of active faulting such as scarps and drainage North of the Red River fault, there is a large principal efforts of this study thus lias been th< diversions are present throughout, the general seismically active region laced with numerous attempt to determine the fault's degree of lat< absence of sag ponds reflects both the high rate faults of north and northwesterly trends. Several Quaternary activity. Is the fault truly a "dead' of dissection of the fault by the Red River and its of these faults display clear and even spectacular fault in the present tectonic environ ment, or is ii tributaries and the lower degree of activity as evidence of youthful normal faulting, and some simply representative of a temporal "seismic compared to highly active faults such as the San appear to have left-lateral components as well. gap"? If the latter, what is its slip rat: and degre« Andreas fault of California. These faults, as well as the Red River fault itself, of hazard, what is its sense of motion, and how does it relate to other nearby structures in the In its middle 170 km, the fault zone is made are accommodating regional east-west crustal regional plate-tectonics framework? The princi- up of two branches. The range-front branch extension and north-south shortening. pal investigative technique used in this studj demarcates the northeastern base of the Ailao was examination of the late Quaternary histor; Mountains and, at least locally, has an apprecia- INTRODUCTION of the Red River fault as expressed in the field ble component of dip slip. The mid-valley applying methods similar to those used on the The objective of this study was to understand San Andreas fault by Sieh (1973a, 1978b) better the current tectonic activity and seismic Note: Chinese surnames are written before given hazard along a part of the east flank of the great names. All authors' names are given in alphabetical 'Most place-names used in the text are shown or order here, and Chinese names are written in their eastern syntaxial bend of the Himalayan moun- the accompanying maps, particularly Figure 1, and se< usual form. tain chain, as reflected by the Red River and Figures 3 and 18 below.

Geological Society of America Bulletin, v. 95, p. 686-700, 21 figs., June 1984.

686

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/95/6/686/3434511/i0016-7606-95-6-686.pdf by guest on 30 September 2021 FAULTS IN YUNNAN PROVINCE, CHINA 687

date. (1) It is generally a high-angle fault, with a pronounced cataclastic belt associated with the bounding Ailaoshan metamorphic rocks on the southwest. (2) The rocks and geologic histories of the blocks on the two sides of the fault are very different (Ministry of Geology, 1962). Shallow-marine sedimentary units dominate on the northeast, whereas the Ailaoshan metamor- phic zone lies to the southwest. (3) Structural trends on opposite sides of the fault are very different, with north-striking faults dominating on the northeast and northwesterly trends (par- allel to the fault) dominating on the southwest. (4) A number of Cenozoic basins along the fault have resulted from incision of the Red River into the Yunnan Plateau. (5) The Ailaoshan metamorphic terrane southwest of the fault is in turn bisected by a major regional fault that is subparallel to the Red River fault, the Ailaoshan fault; this fault is associated with ultramafic rocks and an impressive mylonite zone, and, ac- cording to the Second Regional Surveying Bri- gade of the Geological Bureau of Yunnan Province, it represents part of an ancient plate boundary. Most recent displacements, however, appear to have taken place along the Red River fault rather than the Ailaoshan fault. The great differences in rock types and geo- logic histories across the Red River fault might in themselves suggest large lateral displacements, and large displacements would be mechanically consistent with the nearby impingement of India upon the Eurasian continent beginning about 40 m.y. ago (Molnar and Tapponnier, 1975). Tap- ponnier and others (1982) suggested, on the basis of laboratory modeling, that 1,000 km of Figure 1. Index map of parts of Yunnan and Provinces, China, left slip occurred along the fault during the early showing Red River fault and epicenters of major earthquakes reported from 780 phases of the Indo-Eurasian collision. We are to 1976. Thin lines schematically represent other mapped faults. Box shows not aware, however, of any field evidence that location of Landsat image of Figure 2. Inset shows area in regional context; proves either right- or left-slip displacements of heavy lines are major faults, adapted from Molnar and Tapponnier (1975). hundreds of kilometres on the Red River or parallel faults. Wallace (1975), and Allen (1981). Nearly all trace of a profound geological discontinuity Five lines of evidence have suggested that re- ;gments of the fault between Xiaguan and Atu (Brown, 1913; Gregory and Gregory, 1925; Yin cent geologic movements along the Red River Fig. 3) were studied in the field, as a and Lu, 1936; Wang, 1942; Misch, 1945). fault might be of right-slip character. (1) As in- 3-operative effort among scientists of the Seis- Misch (1945) divided Yunnan into three major ferred above, the plate-tectonics model for the îological Bureau of Yunnan Province, the State tectonic provinces: the East Yunnan Basin, the interaction between the Indian and Eurasian eismological Bureau of China, and the Califor- Central Yunnan Swell, and the West Yunnan plates suggests that faults of this trend southeast ia Institute of Technology. Many of the results Parageosyncline. The boundary between the of the great east Himalayan syntaxial bend re presented in much greater detail, particularly two latter provinces is now recognized as the should have a right-lateral sense of displacement 'ith regard to the description and illustration of Red River fault. Systematic geologic mapping (Li and Wang, 1975; Molnar and Tapponnier, jecific field localities, in the Chinese-language commenced following 1950, and the fault was 1975; Deng and others, 1979). (2) Syntheses of terature by the same authors (Allen and others, further delineated and given its current name. regional stress fields in southwestern China i press). Since 1960, the Second Regional Surveying Bri- based on regional fault patterns and earthquake gade of the Geological Bureau of Yunnan Prov- focal mechanisms have suggested right slip on REVIOUS WORK ince has carried out intensive studies of the area, the Red River fault (Kan and others, 1977; most of which are as yet unpublished. Tapponnier and Molnar, 1977). (3) Right- From the time of the earliest geologic work in Several major characteristics of the Red River lateral displacement on the Qujiang fault during unnan, the Red River was known to mark the fault have been pointed out in the literature to the 1970 Tonghai earthquake (Fig. 1) suggests

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/95/6/686/3434511/i0016-7606-95-6-686.pdf by guest on 30 September 2021 688 ALLEN AND OTHERS

Figure 2. Part of Landsat image 1547 03112, showing -115 km of Red River faul from near Xishelu (upper left) to Mosh Commune (lower right). Compare with Fig ures 1 and 3A.

that a similar sense of movement should be ex, pected on the parallel Red River fault, 80 km t the south (Zhang and Liu, 1978; Zhu, 197i Zhu and Liu, 1978). (4) Stream patterns sugges tive of right-lateral offset can be seen on Landsa images, particularly in the segment betwee: Damajie and Eja (Fig. 4). (5) In connectio with studies of the 1970 Tonghai earthquak« Zhang and Liu (1978) pointed out apparen right-lateral stream offsets of tributaries to th| Red River between Yuanjiang and Atu (Fig. 1' On the other hand, the very existence of th steep ahd persistent escarpment of the Aila Mountains, facing the Red River on the south west, has suggested to many geologists tha recent movements on the fault have been pr< dominantly vertical, with the southwest side re atively raised. Indeed, the locally spectacuk triangular facets along the Ailao Range front ai remarkably similar to those resulting from noi mal faulting in areas of extensional tectonics. Insofar as we are aware, no one prior to thi study has examined the detailed physiographi features along the fault trace in an attempt t use them to determine the fault's slip rate, degre; of activity, and seismic hazard.

Figure 3. A. Red River fault from Xiaguan to Atu, showing tribu- tary-stream drainages. Landsat images used as base maps. B. Map of Red River fault with 5-6 km of right slip restored, so that hypo- thetical drainage pattern is that prior to the time the Red River started its most recent incision.

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/95/6/686/3434511/i0016-7606-95-6-686.pdf by guest on 30 September 2021 FAULTS IN YUNNAN PROVINCE, CHINA 689

EISMICITY of the relevant historical evidence, Deng Rui- logic evidence thus could be interpreted as indi- sheng (1981, personal commun.) concluded that cating that the fault is truly "dead." It could also, Yunnan is one of the most earthquake-prone the most we can say is that there have been no however, be interpreted as indicating that the rovinces of China, and its history has recorded earthquakes exceeding magnitude 7 on this seg- fault is currently a temporal "seismic gap," and lany disastrous events, some of which have ex- ment of the fault for some 300 yr—since about that the recurrence interval between major :eded magnitude 8 (Fig. 1). The remarkable 1700. Beginning about 1850, moreover, the cat- events is greater than 300 yr. As such, it might italogue of Chinese earthquakes (Academia alogue for even moderate-size earthquakes is be analogous to the active Median Tectonic Line inica, 1970, 1976) includes large earthquakes relatively complete, and none has occurred. This of Japan, which is a major fault zone that has Yunnan that occurred as early as the 15th quiescence is verified by modern instrumental not produced a large earthquake since at least :ntury, but the historical record is far from records, which show few events of even small A.D. 700 but which shows clear geomorphic sing geographically homogeneous. Yunnan magnitude along this segment of the Red River evidence of repeated displacements during Hol- irthquakes as ancient as the 9th century A.D. fault. Both the historic and instrumental seismo- ocene time (Okada, 1980). The focus of the -e known from the historic centers of Chinese ilture in the Dali and Kunming regions. In iost of the rest of southwestern Yunnan, the istorical record is much shorter. This is appar- itly due to the region's remoteness, as well as i its population being made up primarily of linority peoples with limited contacts with the lain cultural centers. Certainly, the instrumen- .1 records since about 1900 indicate that mthwestern Yunnan as a whole is at least i seismically active as are many parts of the rovince with longer historical records of irthquakes. It is obvious from a glance at a map of epicen- rs (Fig. 1) that the Red River fault has an jparent absence of seismic activity, at least om Midu southeast to the Vietnam border, 0 25 50 km owever, the extent to which this absence re- sets the short historical record as opposed to a Figure 4. Drainage patterns along Red River fault (arrows) between Damajie and Eja (Fig. ue long-term dearth of earthquakes is prob- 3A), showing apparent right-lateral stream offsets of several kilometres. Traced from Landsat matical. On the basis of an examination of all image prior to field study.

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/95/6/686/3434511/i0016-7606-95-6-686.pdf by guest on 30 September 2021 690 ALLEN AND OTHERS

present study of the Red River fault thus was to look farther back in time than is permitted by the historic record and to examine the fault's late Quaternary history. Although the principal segment of the Red River fault south of Midu is currently quiescent, considerable seismic activity has characterized the northernmost end of the fault and its various branches north of Midu (Fig. 1). As is discussed below, the Red River fault loses much of its geometric simplicity north of Xiaguan and frays into several branches of more northerly trend with greater vertical components of displace- ment. A similar pattern of seismicity and fault geometry seems to characterize the southern San Andreas fault of California. Only large, infre- quent earthquakes occur along the simple, cen- Figure 5. Sketch map of fault scaips between Yuanjiang and Zhega (Fig. 3A). Closely space tral segment of the fault, whereas the areas near hachures represent Quaternary scairp along mid-valley fault, with height of scarp indicated i the ends of the fault, where it frays into many metres. Most of scarp here is in Quaternary gravels. branches, are distinguished by continuing mod- erate activity (Allen, 1968, 1981). A somewhat similar situation has been described along the The broad kink in the Red River fault thus zontal displacement, whereas tfce nearby an Alpine fault system of New Zealand (Adams, may be analogous to the "great bend" in Cali- parallel range-front fault has an appreciabl 1980). If this analogy is valid, it would be fornia's San Andreas fault, which similarly oc- component of vertical displacement, creates another reason for considering the segment of curs in the very area where it is abutted by a mechanical quandary. the Red River fault from Midu south at least to major conjugate fault system, in this case, the The northernmost area in which the Re the Vietnam border as a temporal "seismic gap." left-slip Garlock fault. River fault is identifiable as a single feature Between Gasa and Atu, the valley of the Red near Dinxiling (Fig. 3), where Quaternary coa GENERAL DESCRIPTION OF THE River is typically a two-tier valley, with the bearing strata are cut. Still farther north, Erhi FAULT TRACE present river sharply incised into a broader dis- Lake is undoubtedly fault controlled, but th sected valley underlain primarily by Cenozoic faults appear to be north-trending normal faul The Red River fault between Xiaguan and fluvial sediments. These sediments were laid that are only indirectly related to the Red Rive Hanoi stands out on Landsat images as a nearly down within the old valley of the Red River at a fault; they are discussed below. South of Dinx continuous, broadly arcuate structure. This is the time when it flowed at a considerably higher ling, individual fault traces are not identifiabl case primarily because the Red River has cut its level than at present. Presumably, uplift of the through the heavily cultivated Midu Valley, bi course along the general trace of the fault, and Ailao Range on the southwest produced alluvial between Shaichu and Daqiao the fault crosse the fault trace is thus emphasized by the regional fans that forced the Red River against the north- many tributaries to the Red River and has se\ drainage pattern. Furthermore, the very abrupt east side of the valley prior to its most recent eral good exposures of vertical fault planes, sut northeastern escarpment of the Ailao Moun- incision. The Red River thus now flows along horizontal slickensides, and impressive goug tains—bordering the Red River Valley on the the northeast side of the Cenozoic basin, zones more than 10 m thick. Physiographic fea southwest—is in fact the exceedingly linear fea- whereas the major fault strands that originally tures here of recent right-lateral displacemer ture that is so striking on the Landsat images determined its course lie within the Cenozoic are among the best anywhere atang the faul (Fig. 2). The base of the escarpment has usually basin or along its southwest border. Exposures and some of these are described and illustrate been assumed to represent the Red River fault of the fault are particularly good within the dis- below. itself. sected Cenozoic sediments, and it was within Between Daqiao and Eja, the fault crosse A broad kink in the otherwise smooth trace of this area that most of our field work was relatively inaccessible mountainous country an the Red River fault is found in southern Yun- concentrated. was not visited by the authors. Deep incision i nan, between Yuanjiang and Manhao, and it is A surprising result of this study was the dis- this segment has led to consistent right-lateri significant that this bend occurs in the area covery that the principal active trace of the stream offsets of several kilometre; (Fig. 4)—fa where the fault is abutted by the northerly trend- modern Red River fault lies within the Cenozoic more than is typical, incidentally, along the Sa: ing Xiaojiang fault system (Fig. 1). The Xiaoji- basin, at least between Chunyuan and Nabing Andreas fault of California. Still farther south ang fault has been the locus of two of Yunnan's (Fig. 3), rather than at the base of the Ailaoshan east, the fault lies along the base of the Aila greatest historic earthquakes, in 1733 and 1833, escarpment, as might be assumed from viewing Mountains, and exposures are rare because C and several lines of evidence, including field the satellite images. Both traces are shown in rapid alluviation. Near Zhelong, one exposur observations from those events, indicate that its Figure 3 between Chunyuan and Nabing; the indicates a fault dip of 80° to 85° northeast wit sense of displacement is left lateral. The Xiaoji- southwestern line represents the range-front slickensides plunging 17° southeast, and th ang fault and its branches extend far to the north fault, whereas the northeastern line represents thickness of the gouge zone here is at least 10 into central China (Fig. 1, inset) and constitute the more active strand within the Cenozoic m. The fault splits into two branches near Chur one of the most active regional fault systems in basin, herein called the mid-valley fault. That yuan. The range-front fault continues to mar all of China. the mid-valley fault shows predominantly hori- the northeastern boundary of the A ilaoshan me

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/95/6/686/3434511/i0016-7606-95-6-686.pdf by guest on 30 September 2021 FAULTS IN YUNNAN PROVINCE, CHINA 691

imorphic terrane and typically dips steeply ortheast, whereas the mid-valley fault diverges istward and appears to take up the principal amponent of current strike-slip movement; iany good exposures display horizontal slicken- des on nearly vertical fault surfaces of the mid- alley branch. Both branches are reasonably xessible throughout the 170-km-long segment ) Nabing, where the two branches rejoin, and it 'as in this segment that most of our field work 'as concentrated; several specific areas are dis- ussed in detail below. Atu was the farthest point southeast along the tult visited during this study, but satellite nages, as well as published geologic maps, tave no doubt that the fault continues farther jutheast in China and into Vietnam as a major Figure 6. Sketch map of Nabing area, showing recent trace of Red River fault (heavy dashed :ctonic feature, encompassing the Neogene- line). Linear fault scarp cutting terrace at B is shown in Figure 7. Terrace may be right-laterally »uaternary Red River graben in the Hanoi re- offset from A to B. Drainages farther west formerly flowed down abandoned wide valley floor ion (Nguyen, 1969; Morgunov, 1970). Aero- between Panzhihua and Nabing but have subsequently been captured by recent incision at D. lagnetic data were used to trace the structure Photograph of Figure 7 taken from point C. eneath the young deposits of the Hanoi trough nd southeast still farther to the Gulf of Tonkin Rezanov and Nguyen, 1968), where it is lought to have played a major role in the Me- )zoic plate-tectonics history of southeast Asia Ben-Avraham and Uyeda, 1973; Ben-Avraham, 978; Tapponnier and others, 1982). The total :ngth of the fault from Xiaguan to the Gulf of onkin is about 900 km (Fig. 1, insert), and it IUS certainly deserves to take its place among ie major continental strike-slip faults of the /orld (Allen, 1965).

EVIDENCE FOR RECENCY OF MOVEMENT

Evidence for recency of movement on tie Red River fault comes from two principal ources: (1) physiographic features of surface lulting, the freshness of which demands a re- ent origin, and (2) exposures of young rocks uch as terrace gravels that can be seen to be cut iy the fault. The relevant physiographic features i turn make up two fault-produced phenom- na: fault scarps and offsets of youthful stream hannels. These features are discussed in succes- Figure 7. View northwest from point C of Figure 6. Note scarp (white arrow) cutting terrace ion, citing the best specific examples from the at Nabing. R indicates Red River, with black arrow showing direction of flow. Point D is same arious segments of the Red River fault. as in Figure 6. Wide underfit valley between Nabing and D is abandoned former course of Red River. ''ault Scarps Fault scarps in relatively young rocks consti- The fault scarps between Yuanjiang and the scarp appears very abruptly in older gravels ute the most obvious evidence of active faults Zhega (Fig. 5) consist of a series of aligned and about 3 km southeast of the city, where the road he world over (Cotton, 1958; Allen, 1975; predominantly back-facing scarps that interrupt to Zhega then follows the obvious scarp for sev- iharp, 1954), and the Red River fault is typical the broad alluvial fans that slope northeastward eral kilometres (Fig. 5); in this segment, the n displaying numerous well-preserved scarps from the Ailaoshan toward the Red River. Un- scarp "scissors" in its sense of displacement— hroughout its length. The most easily accessible like many other areas along the fault, the terrane typical of active strike-slip faults—and reaches a ind obvious of these are in (1) the area between is not so deeply incised that the scarps have been maximum height of about 60 m near Zhega. fuanjiang and Zhega and (2) the area of Nabing badly eroded. The fault does not break the very Well-preserved fault scarps near Nabing are Fig. 3). youngest alluvial deposits near Yuanjiang, but easily visible and accessible from the Atu-

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/95/6/686/3434511/i0016-7606-95-6-686.pdf by guest on 30 September 2021 692 ALLEN AND OTHERS

Figure 8. A. Red River between Shuitian and Daqiao, showing offset stream channels. Lowest contour is arbitrarily chosen datum. Bo shows location of more detailed sketch map of Figure 9. B. Same segment, with 5.3 km of right slip restored, thus straightening the Red Rive where it crosses the fault.

Honghe road (Figs. 6, 7). An abandoned former fault vary from as little as 9 m to as much as 5 or vincing because the offsets are in the "uphill tributary or loop of the Red River has left here a 6 km. Attention is given in this section of the direction relative to the larger drainage patter; broad valley and associated terrace gravels that discussion only to the small offsets, because they and thus could not be alternatively explained a preserve the late Quaternary displacements have a direct bearing on the recency of offset due to preferential erosion within the fault zoni along the fault. Most obvious, in addition to the along the fault; larger offsets, which obviously Older and deeper channels are offset more tha riftlike valley itself, is the linear scarp just north reflect a longer history of successive displace- are younger and smaller ones, indicating contir of Nabing Village (B in Fig. 6; also seen in Fig. ments, are discussed below. Offset stream chan- uing and incremental fault displacements. Pei 7), which cuts both the terrace gravels and the nels are present along all segments of the fault, haps the smallest convincing offset is that show underlying Neogene sedimentary rocks. but the most convincing and easily accessible at the right of Figure 9, where both the chann< evidence for the recency of such oflsets comes and its parallel ridge are cleanly offset about from three areas: (1) the Shuitian-Daqiao area, m, offering an ideal example of a "shutterridge Offset Stream Channels (2) the south end of the Gasa Valley, and (3) that blocks and diverts the adjacent channel, i near Yaojie (Fig. 3). displacement this small and associated with sue Consistently offset stream channels provide 1. Between Shuitian and Daqiao, both large sharp topography must certainly reflect very r« probably the most convincing world-wide evi- and small stream-channel offsets are clearly vis- cent movement along the fault. dence for active strike-slip faulting (Wallace, ible (Figs. 8, 9). Numerous small light-lateral 2. At the south end of the Gasa Valley, 1968; Sieh, 1978b). Demonstrable horizontal offsets are particularly evident along the trail small creek is incised about 14 m into the sui offsets of stream channels along the Red River shown in Figure 9, and these are especially con- rounding alluvial deposits (Fig. 10). The cree

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/95/6/686/3434511/i0016-7606-95-6-686.pdf by guest on 30 September 2021 FAULTS IN YUNNAN PROVINCE, CHINA 693

Figure 9. Sketch map of drainage pattern of small channels at locality between Shuitian and Daqiao (Fig. 8). Note consistent right-lateral offsets of drainages, varying from 9-m offset of channel and ridge at extreme right to several-hundred-metre offsets of larger and older drainages. Trail follows series of fault benches, saddles, and valleys along recent fault trace. Hachures mark back-facing scarplets, accentuated by right-lateral displacement of sloping terrain. Scale is approximate.

ind its encompassing gully are sharply offset ibout 55 m along the Red River fault, which is here manifested by a pronounced gouge zone, as Iwell as by springs and seeps. The geometry of the gully suggests a history of incremental dis- placement along the fault. 3. Less than 1 km northwest of Yaojie, which is located directly astride the fault, a stream channel is right-laterally offset 70 m along the fault, again in the "uphill" direction relative to the over-all drainage (Fig. 11). Another stream channel a few hundred metres farther northwest is offset about 25 m. A shutterridge blocks the former channel, the geometry and slopes of which suggest that here also a 9-m increment of lateral displacement has occurred very recently. A single 9-m slip event would correspond to a very large earthquake, although multiple events cannot be ruled out. In any case, the uneroded nature of this youngest segment of the shutter- ridge suggests that its age is no more than 1,000 or 2,000 yr. Supporting this point of view is the observation that physiographic features of strike- slip displacement associated with the nearby 1970 Tonghai earthquake (Seismological Bu- reau of Yunnan Province, 1979) are already se- verely modified and eroded after only 10 yr; fresh gullies and sharp physiographic features simply cannot persist for long periods of time in (Fig. 3A). Scale is approximate. this environment. countered during this investigation was at the ments at least four episodes of fault slip. The first Faults Cutting Recent Materials point where the Sango drainage crosses the fault occurred when only about one-half of the thick- 2.8 km southeast of Zhega (Figs. 12,13). At this ness of fluvial sand and gravel had accumulated; Many localities are present along the Red locality, downcutting by the stream has left a this event is indicated by the disturbance in the River fault where relatively young gravels can thin terrace remnant 6 m above the active chan- fluvial unit immediately south (left) of the fault be seen to be cut by the fault, but certainly the nel floor and the various underlying units that trace. The uppermost 70 cm of fluvial sand and best and most thoroughly studied exposure en- are identified in Figure 13. This exposure docu- gravel overlies the disturbed layers and is unaf-

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/95/6/686/3434511/i0016-7606-95-6-686.pdf by guest on 30 September 2021 694 ALLEN AND OTHERS

fected. A later episode of faulting postdates de- position of the youngest fluvial sand and gravel but predates the slope wash. It resulted in the creation of at least the lower portion of the fis- sure filling, which is conspicuously devoid ol clasts from the slope wash. If the slope wash had overlain the fault at the time ol this second event, it probably would have fallen into the fissure formed during the earthquake. Another slip event must have occurred after deposition ol some of the slope wash and resulted in juxtapo- sition of lower slope-wash deposits against] intermediate-level fissure debris, along a steeply dipping fault plane. This event occurred before the upper 1 m of interfingering slope wash and fissure filling was deposited. Finally, an episode of faulting occurred that extended the main fault plane to within a few centimetres of the present ground surface. The upper 1 or 1.5 m of the main fault plane clearly postdates the inter- fingering slope wash and fissure debris. All four of the episodes of faulting identified in the cut near Zhega are contemporary with or postdate deposition of the fluvial sand and gravel, which we judge to be of Holocene age. This age assignment is based on (1) the lack of a well-developed soil profile at the ground surface, (2) clast-sound velocities (Crook and Kamb, 1981; Gillespie, 1982) in the boulders that are indistinguishable from those of boulders in the active stream bed, (3) the low elevation of the terrace above the active stream bed, (4) the rela- tively undeformed nature of the slope wash and fluvial beds, and (5) the position of the slope wash at the base of a steep hillside, which indi- cates that it must still be accumulating. The Red River fault here thus has probably experienced a minimum of four slip events within the Holo- cene epoch. If each event were associated with a large earthquake, the maximum recurrence in- terval here would be about 3,000 yr.

Conclusions on Recency of Displacements Figure 12. View northwest of recent fault trace at Sango stream, 2.8 km southeast of Zhega (Figs. 3A, 5). Compare with Figure 13 and see text. There can be no question whatsoever of late Quaternary right-lateral displacement along the Red River fault. Numerous physiographic fea- tures of active faulting, in addition to various exposures of displaced terrace gravels, testify firmly to this conclusion. The degree to which this activity has continued into Hoiocene time

I METER admittedly is more debatable, however, because I ' I no radiocarbon dates have yet been obtained Figure 13. Diagramma- from critical exposures along the fai.lt. For rea- fluvial sond + tic sketch of geological re- sons that have been outlined in the preceding lations at locality of Fig- paragraphs, the authors feel strong ly that re- pebble gravel ure 12. peated Holocene displacements have indeed taken place, but this conclusion is necessarily boulder cobbly, ^ based on subjective criteria such as comparisons with other active faults. Physiographic features resulting from recent surficial displacements along the Red River fault

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/95/6/686/3434511/i0016-7606-95-6-686.pdf by guest on 30 September 2021 FAULTS IN YUNNAN PROVINCE, CHINA 695

ire remarkably similar to those seen along other kilometres, far more than the 5 to 6 km indi- regional strike-slip faults, such as the San An- cated by the more recent geomorphic evidence. dreas fault of California, the Alpine fault system Two areas are discussed below; much more ex- Df New Zealand, the Philippine fault, and the tensive evidence, including tabular data from in- North Anatolian fault of Turkey (Allen, 1965, dividual fault-crossing tributaries, is detailed by 1975, 1982). Each of these other fault systems Allen and others (in press). lias had well-documented (and indeed historic) displacements, so that one is tempted to classify Red River Crossing near Daqiao the Red River fault as having a similarly high degree of activity. Some outstanding differences, A restoration of 5.3 km eliminates the pro- however, must be explained, such as the almost nounced jog in the course of the Red River pomplete absence of sag ponds along the Red along the fault near Daqiao (Figs. 3, 8), and, at River fault; only two such features were seen (at the same time, it provides a very reasonable Shuitang Commune and near Shuitian), and match of all large drainages and 2 major drain- both are of debatable origin. One possible ex- age divides that cross the fault in this area. The planation is that the very rapid dissection and apparent similarity of offset of all of the major downcutting of the Red River has simply accel- channels along this stretch of the fault implies srated degradation of physiographic features of that all of the channels were incised into the TO recent faulting, and it is indeed true that those surrounding terrane before the latest 5.3 km of segments of the San Andreas fault where de- offset occurred. Concordant summits in this area tailed geomorphic features are most spectacular are about 800 m above the present level of the ^ 5.5 km ind most often photographed—such as in the Red River, so the river may have incised nearly T Carrizo Plain (Wallace, 1968; Vedder and Wal- 1 km since initiation of the 5-km offset. Alterna- lace, 1970)—are areas of internal drainage with tively, the present stream valleys may have been current dissection limited mainly to that initiated substantially eroded before the most recent pe- by local fault displacement itself. Second, one riod of activity on the Red River fault began. might argue that heavy rainfall has modified Figures 3A and 3B show that nearly all of the physiographic features of faulting along the Red major tributaries to the Red River in this seg- River fault more than along active faults in, for ment between Shuitian and Zhelong can be example, California. Nevertheless, closed de- given more normal configurations by restoration pressions and detailed features of recent move- of between 5 and 6 km of right-lateral slip. Figure 14. Sketch maps showing evolution ment are relatively abundant along the Philip- of drainage patterns between Nabing and pine fault (Allen, 1962), in an area of even Red River and Tributaries near Atu Atu. See text for explanation. higher rainfall than that of Yunnan, so climatic differences alone probably cannot explain the Restoration of 5 to 6 km of slip also seems to until after -3.5 km of the latest 5.5 km of offset ontrasts. give more normal configurations to tributaries had occurred. This interpretation is consistent Taking all avaliable evidence into considera- flowing northeastward into the Red River be- with the model of Figure 14. That the Red River tion, the authors conclude that, although there tween Yuanjiang and Atu (Figs. 3A, 3B). For is continuing its vigorous incision to this day is have almost assuredly been recurrent displace- example, Figure 14 shows the evolution of 5.5 indicated by innumerable "hanging" tributary ments along the Red River fault in Holocene km of right slip across a 55-km length of the valleys, the steepness of the valley walls, and the time, its degree of activity—and thus slip rate— fault between Nabing and Atu. The uppermost general dearth of alluvium filling the modern is markedly less than that of the San Andreas or frame shows the Red River and major tributar- valley. other "very active" fault systems. Indeed, the ies prior to initiation of the offset; the middle demonstrated recurrence interval of about 160 frame shows the drainage configuration after 3.5 Development and Age of the 5- to 6-km yr between major earthquakes on one segment km of slip had accrued; and the lowest frame Offsets. of the San Andreas fault (Sieh, 1978a) clearly shows the situation at present. Comparison of indicates that the San Andreas fault has been the middle and lowest frames suggests that two The remarkable realignment of the Red River more active than the Red River fault for at least stream courses (dashed in the lowest frame) and nearly all of its major tributaries by elimina- 2,000 yr. were abandoned because of stream capture dur- tion of 5 to 6 km of lateral slip indicates that the ing the time that the latest 1 or 2 km of slip was modern drainage network of the Red River GEOMORPHIC EVIDENCE FOR 5 TO taking place. formed and began to incise prior to the inception 6 KM OF OFFSET The drainage profiles of Figure 15 elucidate of the 5- to 6-km offset. Lateral displacement this story. Both the abandoned meander loop of could have been occurring for some time before Probably the most surprising discovery of this profile A and the abandoned tributary of profiles the initial entrenchment of the drainage system, study was that the Red River itself and all of its B and C are suspended about 200 m above the but displacements that accumulated prior to the major fault-crossing tributaries appear to have present river level, although they had eroded initial entrenchment might not be recorded been uniformly offset between 5 and 6 km right some 300 to 400 m into their host rock prior to geomorphically. Such a hypothetical sequence laterally. Figure 3B has been constructed by re- abandonment. This 300 to 400 m of downcut- of events is illustrated in Figure 16. In rectangle storing these drainages to their configurations ting could have occurred during or before the I, the uplifted Ailao Mountains are shedding de- prior to the offset. Juxtaposed contrasting rock period in which the first several kilometres of bris into the valley occupied by the river and the types across the fault suggest, additionally, that lateral offset was accomplished, whereas the fault, with the river course being shifted away the total offset may be measured in hundreds of latest 200 m of river incision did not take place from the mountains by alluvial-fan deposition.

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/95/6/686/3434511/i0016-7606-95-6-686.pdf by guest on 30 September 2021 696 ALLEN AND OTHERS

Profile across abandoned

Figure 15. Drainage profiles across Red River fault zone along lines indicated in Fig- ure 14. Elevations are relative to that of mod- ern river.

the formation of the modern drainage basin of incised by the Red River and its tributaries be- the Red River and its initial entrenchment. This tween Atu and Mosha. These have been kilometers hypothesis is supported by the observation that mapped as Neogene, and if this assignment is This situation would be similar to that of the the Red River itself is offset only 5.3 km in the correct, the slip rate must be equal to or greater active Median Tectonic Line in central Kii, only place where it crosses the Red River fault, than 0.2 mm/yr (5 km/25 m.y.). If slip began Japan (Okada, 1980; Okada and Sangawa, near Daqiao. We are aware of no evidence for a after initial entrenchment of the river, the slip 1978). Although both branches of the fault are larger offset of this major river at this location at rate would also be greater than 3.2 mm/yr. active, only relatively small offsets and diver- elevations above the 800-m-deep gorge in which Wang and others (1982) reviewed paleontologi- sions exist, because of the aggradational nature it is now flowing. Additionally, geologic maps cal evidence from the Himalayas and the Tibet- of the valley floor at this time. In rectangle II, do not record any fluvial gravel deposits .at levels an Plateau that indicates that the recent uplift of entrenchment of the Red River and its tributar- above this inner gorge that might indicate that those regions began in the late Plioixne Epoch, ies has begun (perhaps no more than 50 to 100 the river once flowed on a broad plain above its perhaps 2 to 3 m.y. ago. It is reasonable to as- m), so that lateral fault movements begin to be present level. If substantial slip had occurred sume that the most recent uplift in southwestern well preserved. Fault slippage and incision con- after integration of the current Red River drain- China, including Yunnan, is contemporaneous tinue to the present day, as represented by rec- age system but before inception of incision of the with this most recent Himalayan-Tibetan uplift. tangle III. It is important to note, however, that 800-m-deep gorge near Daqiao, the Red River If this is correct, the 5 to 6 km of right slip on the initially entrenched drainages are repreatedly ought to reflect more than 5.3 km of slip. Red River fault has accumulated in 2 to 3 m.y. abandoned, pirated, partially buried, and ex- The age of inception of the 5- to 6-km offset is or less. The average slip rate thus would be humed as the two blocks move past one another. unknown. If the model proposed in Figure 16 is >2-3 mm/yr. At this rate, a 9-m slip event An alternative hypothesis is that the current correct, the age must be contemporaneous with would occur at the most every 4,500 yr, on the episode of right-lateral slippage began well after the top of the alluvial-fluvial deposits that were average.

Partitioning of the Offset between the Range-Front and Mid-Valley Fault Branches

Mid-valley (D \ \ Il / / / V. W' ' fault Between Chunyuan and Nabing, the restora- ïi//,. tion of offsets is complicated by the double- Range-front stranded nature of the fault. Along the south- fault eastern half of this 170-km-long stretch, the / ( Ailao ) \ Range Figure 16. Sketch«» por- range-front fault displays no evidence of lateral • / • t t t \ traying schematic sequen- slippage, and the entire 5 to 6 km of lateral offset tial physiographic devel- seems to have been accomplished along the mid- opment along Red River valley fault. Between Manche and Baha, in fact, fault at base of Ailao the range-front fault displays no clear evidence Range. Heavy dashes rep- of recent strike-slip or dip-slip motion. Between resent abandoned stream Baha and Yuanjiang, however, pure dip slip of courses. Hachures in rec- more than 100 m is indicated by triangular tangle I indicate alluvial facets and vertically offset ridge crests (Allen fans resulting from uplift and others, in press). of Ailao Range, forcing The southeasternmost channels crossing the Red River north. Con- range-front fault that show strike-slip offsets are temporary incision of Red immediately west of Yuanjiang, and right-lateral River and tributaries took offsets become progressively greater to the place subsequently. See northwest. Near Mandan, the offsets are 1 km or text. more, which is as large as or larger than the spacing between channels draining the Ailao- shan. The resulting sequences of capture, aban- donment, burial, and exhumation have led to complex histories of individual channels in this

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/95/6/686/3434511/i0016-7606-95-6-686.pdf by guest on 30 September 2021 FAULTS IN YUNNAN PROVINCE, CHINA 697

'ea, perhaps best exemplified by the apparent ft-lateral offset of a major tributary between asa and Yaojie (Fig. 17). As strike-slip displacement increases north- •estward along the range-front fault, it corre- )ondingly decreases along the mid-valley fault, ) that the total offset remains approximately le same. Near Mosha, both strands seem to ave about 2.5 km of right-lateral offset.

>RIGIN OF THE AILAOSHAN |SCARPMENT

As a result of this study, it appears that there re two principal strands of the Red River fault iroughout much of southern Yunnan. One of lese, in large part previously unrecognized, is a redominantly strike-slip fault in the Cenozoic asins and adjacent Triassic and Jurassic rocks, he other lies at the base of the Ailaoshan es- arpment and has long been recognized, al- lough it typically has been thought to reflect Figure 17. View southeast of range-front fault about halfway between Gasa and Yaojie. rimarily dip-slip displacement. In fact, as men- Stream at waterfall over scarp (A) appears to be offset left-laterally from stream in valley (B). In oned previously, the spectacular triangular reality, this is a separation that has resulted from juxtaposition of drainage segments by right tcets of the escarpment (for example, Fig. 17) slip of much greater magnitude. Note triangular facets. re similar to those usually associated with ac- ve normal faults such as in the Basin Ranges of le western United States or the Shanxi graben northwest for more than 900 km to the vicinity For more than 40 km north from Xiaguan, f north China. How is it possible for a major of Xiaguan (Figs. 1, 3). It is obvious from exam- the high east-facing escarpment of the Cang trike-slip fault and a major dip-slip fault to be ination of satellite images that the region north Mountains stands in bold contrast to the adja- arallel and in very close juxtaposition for more of Xiaguan—to and beyond the Jinsha (Yang- cent lake-filled valley floor. Recent faulting lan 150 km? We have by no means solved this tze) River—is also characterized by active fault- along the range front is indicated by (1) discon- roblem, but we suggest that a combination of ing (Fig. 18), but of more northerly trend than tinuous fault scarps cutting the alluvial fans; (2) ictors has been important and that the Ailao- that of the Red River fault, and with considera- numerous Y-shaped or "wineglass" canyons tian escarpment is of composite origin. bly more relief across individual faults (Tap- (valleys with broad upper reaches but with There can be no doubt of significant vertical ponnier and Molnar, 1977). Many, if not most, deeply incised mouths at the range front) (Fig. plift of the Ailao Range, but the remarkable of these faults are in fact probably normal rather 19); (3) young, well-formed alluvial cones; and nearity of the Red River fault—including the than strike-slip faults, judging from their geo- (4) the exceedingly abrupt transition from the ange-front segment—suggests that its history morphic characteristics. Major lakes and closed mountain front to the valley floor. Further evi- as been dominated by strike-slip movements, depressions of this region, such as those of Erhai, dence of the recency of deformation is given by 'he present relief of the range front is in large Chenghai, Dengchuan, and Jianchuan (Fig. 18), the damming of itself. •art due to erosional excavation of the Red testify to continuing tectonic activity, and this is Even more convincing of recent fault dis- liver Valley, and the triangular facets result at consistent with the high historic seismicity of the placement are fault scarps in the two valleys sast in part from erosion controlled by the region (Fig. 1), which includes at least one truly north of Erhai. West of the town of Yousuo in trong metamorphic foliation, which is com- great earthquake near Yongsheng in 1515. the Dengchuan Valley (Fig. 18) there is a truly nonly subparallel to the range front. Neverthe- Normal faulting along north-trending faults in spectacular faulted range front (Fig. 20), the re- ess, the continuing displacement on most parts this northern region is mechanically consistent cent origin of which is indicated by enormous, if the range-front fault have had an appreciable with right slip along the Red River and related smooth triangular facets, Y-shaped mountain 'ertical component, which has maintained a northwest-trending faults; both systems reflect valleys, young alluvial cones, and a lake at the teep range front and added to the relief. Other east-west crustal extension. Furthermore, such a base of the scarp. In the next valley north, larts of the range-front fault are apparently inac- transition from strike-slip to normal faulting is occupied by the city of Eryuan (Fig. 18), the ive at the present time, at least as compared to seen at the ends of other major strike-slip faults. northeast wall of the valley is also a spectacular he continuing vigorous activity along the mid- The map of Figure 18 is based mainly on composite fault scarp; particularly obvious in 'alley fault. The dominating stress field along interpretation of satellite images and on only Figure 21 are the very steep base of the range he Red River fault as a whole is such as to cause limited field work. The comments that follow (locally disturbed by landslides), the young allu- iredominantly strike-slip regional displacement. serve principally to document the fact that spec- vial cones, and the very abrupt transition from tacular normal faulting is indeed present here the steep mountain front to the flat valley floor. FAULTING NORTH OF XIAGUAN and to set the stage for further field work. Strong That some of these faults may have significant postglacial block movements in this region were left-lateral movement is suggested by (1) "pull- The Red River fault can be followed as a first recognized by Misch (1950) but seem not to apart" geometry, such as that north of Chenghai xmtinuous feature from the Gulf of Tonkin have been subsequently investigated. Lake; (2) probable stream offsets of tectonic

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/95/6/686/3434511/i0016-7606-95-6-686.pdf by guest on 30 September 2021 698 ALLEN AND OTHERS

origin, such as the very abrupt 3-km bend of the Jinsha (Yangtze) River where it crosses the Chenghai fault; and (3) the conjugate geometry of these faults relative to those of the Red River system. Still farther northwest, approaching the Tibet- an border, published geologic maps and satel- lite images indicate a return to northwest structural trends parallel to that of the Red River fault (Fig. 1). We suspect that many of these faults, such as the Qiaohou fault of Figure 18, may similarly have large right-slip compo- nents, but we are not aware of field evidence to support this hypothesis. In any case, these faults are not directly continuous with the Red River fault, which is clearly interrupted at least locally in the Xiaguan-Eryuan area by normal faults of more northerly trend.

SEISMIC HAZARD OF THE RED RIVER FAULT

The principal segment of the Red River fault south from the Midu Valley at least to the Viet- nam border has not experienced a major earth- quake in at least 300 yr, nor is it an area of significant current minor seismicity. One might therefore conclude that the fault is effectively "dead" and does not pose a seismic hazard, or one might alternatively view the current seismic quiescence as temporary and indicative of a "seismic gap" soon to break. If the fault were to rupture over this entire 500-km-long segment in

a single earthquake, an event of magnitude (Ms) 8.3 with a maximum strike-slip displacement of 8.5 m would be typical (Slemmons, 1977). Even Figure 18. Sketch map if only 250 km of the fault were to break during of faults inferred from a single event, an earthquake of magnitude 7.9 Landsat images in region with several metres of slip might result. north of the Midu Valley Clearly, the historic record is not sufficiently and Xiaguan, where the long to choose between the two alternatives— Red River fault (lower "dead" versus "active"—and one must instead right) loses its identity. turn to the recent geologic history of the fault. Stippled areas indicate ba- There is abundant experience elsewhere in the sins. world to indicate that those faults with recurrent movements in the recent geologic past are those most likely to break again in the near future; 3). The fault thus must be considered "active" to period beginning with the incision of the Ceno similarly, the probability of an ancient fault at least some degree, and we now proceed to zoic deposits filling the earlier valley, and sina being rejuvenated during a major earthquake is estimate the recurrence interval between major that time the river has cut down about 650 n very small (Allen, 1975). The present study thus earthquakes on the fault. near Honghe and about 800 m near Daqiac has concentrated on the field evidence for the The primary clue to the possible recurrence Rates of river incision are so variable around th< degree of recent activity on the Red River fault, interval between major earthquakes is the cur- world that a time interval since initiation of thi particularly during the Holocene Epoch. rent slip rate of the fault and its comparison with channel offsets cannot realistically be estimatec In our opinion, there is no question that there the slip rates of other major active faults of the from observed downcutting rates. On the othe have been repeated Holocene displacements on world. In the absence of absolute-age determina- hand, downcutting has approximately kept paci some segments of the Red River fault. Further- tions, we do not have precise figures for the Red with the regional uplift that has produced thi more, we see no compelling evidence that the River fault, but a maximum slip rate can be Yunnan Plateau, into which the Rec. River i: same youthful history has not characterized the estimated by making certain assumptions con- now incised, and reasonable limits can be placei total length of the fault, at least within China, cerning the downcutting history of the Red on both its rate of uplift and the date of incep although the recent activity is shared between River. About 5.5 km of right-lateral slip has tion of uplift. Although local uplift rates ii two branches from Chunyuan to Nabing (Fig. taken place on the fault during all or part of the highly tectonic areas may average as m ach as 1(

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/95/6/686/3434511/i0016-7606-95-6-686.pdf by guest on 30 September 2021 FAULTS IN YUNNAN PROVINCE, CHINA 699

began at this time, the rate of incision near Atu must be 0.2-0.3 mm/yr, and near Shuitian it must be about 0.25-0.4 mm/yr. Similarly, the 5.5 km of right slip on incised channels of the Red River basin must have accumulated at an average rate equal to or somewhat greater than 2 to 3 mm/yr. If the slip rate on the Red River fault is about .5 cm/yr, its degree of activity is comparable to that of many other moderately active faults around the world (Sieh, 1981; Matsuda, 1977). To compute a recurrence interval between major earthquakes necessitates additional as- sumptions regarding the amount of displace- ment per event. If 500 km of the Red River fault breaks at one time, corresponding to a magni- tude 8.3 event, the average maximum slip asso- ciated with such strike-slip earthquakes is about 8.5 m (Slemmons, 1977). A slip rate of 5 mm/yr thus suggests such an earthquake about once every 1,700 yr. Offsets of 9 m documented in this study may in fact be attributable to a single great event that occurred less than 1,000 or 2,000 yr ago. A 5 mm/yr slip rate would imply recurrence intervals of 1,800 yr for 9-m Figure 19. "Wineglass canyon" in escarpment of Cangshan north of Dali, indicative of rapid slip events. In this light, it is hardly surprising jplift along range-front fault. that there does not happen to be a record of a major historic earthquake on the Red River fault. Are recurrence intervals greater than 1,000 yr along a single fault segment compatible with world-wide experience? Chinese historic records themselves tend to support this concept. For ex- ample, the great 1668 earthquake on the Tanlu fault and the 1976 Tangshan earthquake in northeastern China have no counterparts in either the earlier or the subsequent historic rec- ord. Historic records are much more limited in most other parts of the world, but recurrence intervals this long are strongly suggested in many areas (Allen, 1975; Sieh, 1981). In fact, Matsuda (1977) pointed out several examples of recurrence intervals of between 10,000 and 100,000 yr on various "active" faults of Japan. In summary, we argue that the Red River fault has had repreated Holocene displacements along it and should therefore be considered "ac- tive" in a geologic sense. However, both the his- toric and the geologic evidence indicate that the Figure 20. Mountain front west of Yousuo, Dengchuan Valley, showing numerous geomor- recurrence interval between major earthquakes phic features of recent normal faulting. along the Red River fault has probably been— and will continue to be—several thousands of years. Its seismic hazard thus is considerably tnm/yr (Schumm, 1963; Sieh, 1981), maximum raneous with or older than the beginning of the lower than along other active faults where rates for areas as large as the Yunnan Plateau 5.5 km offset. The recent and ongoing uplift of earthquakes are known to occur more fre- probably do not exceed 1 mm/yr. Assuming Yunnan may well be contemporaneous with the quently, such as the Xianshuihe fault of western hat incision of the Red River took place at this uplift of Tibet and the Himalayas, to the west. Sichuan Province (Heim, 1934; Tang and oth- ate, 750 m of downcutting would take 750,000 Wang and others (1982) reviewed evidence that ers, 1976), the San Andreas fault of California, yr. The average fault-slip rate calculated in this the most recent pulse of uplift began in those or the North Anatolian fault of Turkey, as well fashion must be 7 mm/yr or somewhat higher, regions in the late Pliocene Epoch, perhaps 2 or as along most subduction zones. Further field because the beginning of incision is contempo- 3 m.y. ago. If the incision of the Red River also work obviously will be required to establish the

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/95/6/686/3434511/i0016-7606-95-6-686.pdf by guest on 30 September 2021 15 ALLEN AND OTHERS

Bulletin, v. 45, p. 1035-1050. Kan Rongju, Zhang Sichang, Yan Fengtong, and Yu Linshenj;, 1977, Preset) tectonic stress field and its relation to the cha acteristics of recent tec tonic activity in southwestern China {in Chinese]: Acta Geophysic Sinica, v. 20, p., 96-109. Li Ping, and Wang Liangmou, 1975, Exploration of the seismo-geologics features of the Yunnan-west Sichuan region [ir Chinese]: Scientia Gee logica Sinica, 1975 (4), p. 308-326. Matsuda, T., 1977, Estimation of future destructive earthquakes from adiv faults on land in Japan: Journal of Physics of ;he Earth, v. 25, Supple ment, p. S251-S260. Ministry of Geology, Institute of Geologic Science, IS62, Tectonic map of th People's Republic of China, 1:3,000,000: Bciji ig. 1979, Tectonic map of China, 1:4,000,000: Beijing. Misch, P., 1945, Remarks on the tectonic history of Yunnan, with specu reference to its relation to the type of younj orogenic deformatiot Geological Society of China Bulletin, v. 25, p. 47-153. 1950, Late Cenozoic diastrophism in Himalayj.n system [abs.]: Geologi cal Society of America Bulletin, v. 61, p. 1487. Molnar, P., and Tappormier, P., 1975, Cenozoic tectonics of Asia: Effects of continental collision: Science, v. 189, p. 419-426. Morgunov, Y. G., 1970, Basic features of tectonics of North Vietnam: Interna tional Geology Review, v. 12, p. 1333-1345. Nguyen, D. K., 1969, Main tectonic features of Norti Vietnam: Geotectonic [English translation], 1969, p. 245-250. Okada, A., 1980, Quaternary faulting along the V edian Tectonic Line c southwest Japan: Geological Society of Japan Memoir 18, p. 79-108. Okada, A., and Sangawa, A., 1978, Fault morpholog} and Quaternary faultin. along the Median Tectonic Line in the southern part of the Izumi Rang [in Japanese]: Geographical Review of Japan, v. 51-5, p. 385-405. Rezanov, I. A., and Nguyen, V. S., 1968, Structural trends of North Vietnarr. from the geologic and aeromagnetic data [i 1 Russian]: Ministerstvi Vysshego Obrazovaniya SSSR, Izvestiya Vysshikh Uchebnykh Za vadeniy, Geologiya i Razvedka, No. 2 (1968), p. 3-8. Schumm, S. A., 1963, The disparity between present rates of denudation ani orogeny: U.S. Geological Survey Professional Paper 454-H, 13 p. Seismological Bureau of Yunnan Province, 1979, Tonghai earthquakes and thj Qujiang fault: Kunming, 9 p., 15 figs. Sharp, R. P., 1954, Physiographic features of faulting in southern California California Division of Mines and Geology Bui etin 170, ch. 5, p. 21 -2{j Sieh, K. E., 1978a, Prehistoric large earthquakes produced by slip on the Sai Andreas fault at Pallett Creek, California: Journal of Geophysical Re Figure 21. View south along steep eastern wall of Eryuan Valley. Note linear moun tain front, search, p. 3907-3939. fresh alluvial cones, and landslide resulting from instability of steep fault scarp. 1978b, Slip along the San Andreas fault associated with the grea 1857 earthquake: Seismological Society of America Bulletin, v. 6i p. 1421-1448. 1981, A review of geological evidence for recurrence times of larg REFERENCES CITED seismic hazard of the Red River fault in a earthquakes: American Geophysical Union, M aurice Ewing Ser. No. 4 rigorous quantitative sense, but in the meantime p. 181-207. Academia Sinica, Institute of Geophysics, 1970, Catalog of Chinese earth- Slemmons, D. B., 1977, Faults and earthquake magniiude: Report f the fault should be regarded as one of finite but quakes (117? B.C. to 1949 A.D.): Beijing, 361 p. Miscellaneous Paper S-73-1, State-of-the-art for assessing earthquak 1976, Catalog of large Chinese earthquakes (780 to 1976): Beijing, hazard in the United States: U.S. Army Engineer Waterways Experi relatively low seismic hazard as compared to 29 p. ment Station, 166 p. many other highly active fault systems elsewhere Adams, J., 1980, Paleoseismicity of the Alpine fault seismic gap, New Zealand: Tang Rong-chang, Wen De-hua, Deng Tian-gang. and Huang Sheng-mu Geology, v. 8, p. 72-76. 1976, A preliminary study of the characterises of the ground fracture in China and other parts of the world. Allen, C. R., 1962, Circum-Pacific faulting in the Philippines- region: during the Lu-huo M = 7.9 earthquake, 19/3, and the origin of th Journal of Geophysical Research, v. 67, p. 4795-4812. earthquake [in Chinese]: Acta Geophysica Sinica, v. 21, p. 191-198. 1965, Transcunrent faults in continental areas: Royal !>ociety of London Tapponnier, P., and Molnar, P., 1977, Active faulting and tectonics in China Philosophical Transactions, ser. A, v. 258, p. 82-89. Journal of Geophysical Research, v. 82, p. 29)5-2930. 1968, The tectonic environments of seismically active and inactive areas Tapponnier, P., Peltzer, G., LeDain, A. Y., Armijo, R., and Cobbold, P., 198: ACKNOWLEDGMENTS along the San Andreas fault system: Stanford University Publications in Propagating extrusion tectonics in Asia: New insights from simpl the Geological Sciences, v. 11, p. 70-82. experiments with plasticine: Geology, v. 10, p 611-616. 1975, Geological criteria for evaluating seismicity: Geological Society of Vedder, J. G., and Wallace, R. E-, 1970, Map showing recently active break This study was carried out under the protocol America Bulletin, v. 86, p. 1041-1057. along the San Andreas and related faults betw een Cholame Valley aw 1981, The modern San Andreas fault, in Ernst, W. G., ed., The Tejon Pass, California: U.S. Geological Survey Miscellaneous Geology for scientific and technical co-operation in geotectonic development of California: Englewood Cliffs, New Jersey, cal Investigaitons Map 1-574. Prentice-Hall, p. 511-534. earthquake studies between the State Seismolog- Wallace, R. E., 1968, Notes on stream channels offset by the San Andreas fauli 1982, Comparisons between the North Anatolian fault of Turkey and southern Coast Ranges, California: Stanford University Publications ii ical Bureau of the People's Republic of China the San Andreas fault of California, in Isikara, A. A4., and Vogel, A., the Geological Sciences, v. 11, p. 6-21. eds., Multidisciplinary approach to earthquake prediction, Volume 2: 1975, The San Andreas fault in the Carrizo Plain-Temblor Rangl Braunschweig/Wiesbaden, Friedr., Vieweg & Sohn, ?. 67-85. (SSB) and the U.S. Geological Survey and the region, California: California Division of Mines and Geology Specia Allen, C. R., Gillespie, A. R., Han Yuan, Sieh, K. E., Zhang Buchun, and Zhu National Science Foundation. The field study Report 118, p. 241-250. I Chengnan, in press, Quaternary activity of the Red River fault, Yunnan Wang, C., Shi, Y., and Zhou, W., 1982, Dynami< uplift of the Himalaya Province, China [in Chinese]: Journal of Seismological Research in Nature, v. 298, p. 553-556. was sponsored by the Seismological Bureau of Yunnan. Wang, H. C., 1942, Geological reconnaissance rom Paoshan to Mitu Yunnan Province (YSB), and we appreciate the Ben-Avraham, Z., 1978, The evolution of marginal basins and adjacent shelves western Yunnan: National University of 'eking Contributions il in east and southeast Asia: Tectonophysics, v. 45, p. 269-288. assistance of Zhao Jianmin, Wang Guimei, and Geology, No. 2. Ben-Avraham, Z., and Uyeda, S., 1973, The evolution of the China basin and Yin, T. H., and Lu, C. H., 1936, On the Ordovican and Silurian beds o Jiang Kui in making the many arrangements the Mesozoic paleogeography of Borneo: Earth and Planetary Science Shihtien, western Yunnan: Geological Socie:y of China Bulletin, sei Letters, v. 18, p. 365-376. B, v. 16, p. 281-298. that were necessary. We are particularly in- Brown, J. C., 1913, Contribution to the geology of the province in western York, J. E., Cardwell, R., and Ni, J., 1976, Se smicity and Quaternarj China—1, The Bhamo-Tengyueh area: Geological Society of India faulting in China: Seismological Society of America Bulletin, v. 6c debted to the many geologists of the YSB and Records, v. 43, pt. 3. p. 1983-2001. SSB who accompanied us in the field and con- Cotton, C. A., 1958, Geomorphology (7th edition): Christchurch, Whitcombe Zhang Sichang, and Liu Bochi, 1978, Seismic geological characteristics o and Tombs, 505 p. Tonghai earthquake in 1970 [in Chinese]: Scientia Geologica Sinica tributed markedly to the study. Principal among Crook, R., Jr, and Kamb, B., 1981, A new method of alluvial age dating based 1978 (10), p. 324-335. on progressive weathering, with application to the time-history of fault Zhu Chengnan, 1978, The Cenozoic tectonics of the south region of Yunna: these were Chai Tienjun, He Xihu, Yang Jiwn, activity in southern California: U.S. Geological Survey Final Technical Province and the geological background of the Tonghai earthquake i Zhang Jin, Zhang Shuanglin, Zhou Ruiqi, Hou Report, Contract No. 14-08-0001-17760, 41 p. 1970 [in Chinese]: Journal of Seismological Research in Yunnan, v. Deng Qidong, Zhang Yuming, Xu Guilin, and Fan Fu tun, 1979, On the (3), p. 32-44. Xueying, Deng Ruisheng, Duan Jiale, Ding tectonic stress fidd in China and its relation to plate movement: Physics Zhu Chengnan, and Liu Yuquan, 1978, Analysis of tie cnistal deformation ant of the Earth and Planetary Interiors, v. 18, p. 257-273. fault movement in east Yunnan [in Chinese]: Journal of Seismological Guoyu, Xu Yujian, and Guo Sunmin. Xia Gillespie, A- R-, 1982, Quaternary glaciation and tectonism in the southeastern Research in Yunnan, v. 1 (1), p. 30-44. Sierra Nevada, Inyo County, California [Ph.D. thesis]: Pasadena, Wenyi and Lin Ronghui served tirelessly as in- California Institute of Technology, 695 p. terpreters. Participation by the American co- Gregory, J. W., and Gregory, C. J., 1925, Geology and physical geography of MANUSCRIPT RECEIVED BY THE SOCIETY DECEMBER 13, 1982 Chinese Tibet, and its relation to the mountain system of South-Eastem REVISED MANUSCRIPT RECEIVED JULY 18,1983 authors was supported by U.S. Geological Asia: Royal Society of London Philosophical Transactions, ser. B, MANUSCRIPT ACCEPTED JULY 25,1983 v. 213. CALIFORNIA INSTITUTE OF TECHNOLOGY. DIVISION OF GEOLOGICAL AND Survey Contract No. 14-08-0001-19271. Heim, A., 1934, Earthquake region of Taofu: Geological Society of America PLANETARY SCIENCES, CONTRIBUTION No. 3848

Printed in U.S./»

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/95/6/686/3434511/i0016-7606-95-6-686.pdf by guest on 30 September 2021