Cenozoic deformation of the Tarim plate and the ...web.missouri.edu/~lium/pdfs/Papers/Yang02Tec.pdf · 2.1. Western Kunlun Shan and Western Kunlun Thrust Belt [6] The western Kunlun

Cenozoic deformation of the Tarim plate and the implications

for mountain building in the Tibetan Plateau and the Tian Shan

Youqing Yang and Mian Liu

Department of Geological Sciences, University of Missouri, Columbia, Missouri, USA

Received 11 May 2001; revised 14 March 2002; accepted 8 June 2002; published 17 December 2002.

[1] The Tarim basin in NW China developed as acomplex foreland basin in the Cenozoic in associationwith mountain building in the Tibetan Plateau to thesouth and the Tian Shan orogen to the north. Wereconstructed the Cenozoic deformation history of theTarim basement by backstripping the sedimentaryrocks. Two-dimensional and three-dimensional finiteelement models are then used to simulate the flexuraldeformation of the Tarim basement in response to thesedimentary loads and additional tectonic loadsassociated with overthrusting of the surroundingmountain belts. The results suggest that uplift ofboth the western Kunlun Shan and the Tian Shanorogens started during or before Oligocene. Since theMiocene, mountain building accelerated in thewestern Kunlun Shan but showed segmenteddevelopment in the Tian Shan. The results alsoindicate reduced tectonic load along the Altyn Taghfault since late Miocene, consistent with geologicaland geophysical evidence of the fault cutting theentire lithosphere and migrating southward during itscause of the Cenozoic evolution. INDEX TERMS: 8105

Tectonophysics: Continental margins and sedimentary basins;

8102 Tectonophysics: Continental contractional orogenic belts;

8120 Tectonophysics: Dynamics of lithosphere and mantle—

general; 9320 Information Related to Geographic Region: Asia;

9604 Information Related to Geologic Time: Cenozoic;

KEYWORDS: Tibet, Tarim, Tian Shan, continental deformation,

Indo-Asian collision. Citation: Yang, Y., and M. Liu, Cenozoic

deformation of the Tarim plate and the implications for mountain

building in the Tibetan Plateau and the Tian Shan, Tectonics,

21(6), 1059, doi:10.1029/2001TC001300, 2002.

1. Introduction

[2] The continued plate convergence between India andAsia since 50–70 million years ago resulted in the uplift ofthe Himalayan-Tibetan plateau, reactivated the Tian Shanorogen, and affected regions perhaps as far north as theBaikal [Molnar and Tapponnier, 1975; Yin and Harrison,2000]. Following the pioneering work of Argand [1924] andGansser [1964], significant progresses have been made on

the geology and tectonics of the Himalayan-Tibetan orogen[Dewey and Burke, 1973; Molnar and Tapponnier, 1975;Allegre et al., 1984; Armijo et al., 1986; Burchfiel et al.,1992; Harrison et al., 1992; Avouac and Tapponnier, 1993;Nelson et al., 1996; England and Molnar, 1997; Larson etal., 1999; Yin and Harrison, 2000]. However, some funda-mental questions, such as how stress has propagated acrossthe collisional zone and how strain was partitioned betweencrustal thickening and lateral extruding of the lithosphere,remain controversial. One end-member model, approximat-ing the Eurasia continent as a viscous thin sheet indented bythe rigid Indian plate [e.g., England and McKenzie, 1982;England and Houseman, 1986, 1988], predicts progressivelynorthward propagation of crustal thickening and plateauuplift, with limited lateral lithosphere extrusion. Anotherend-member model, based on indentation of a plastic Asiancontinent, suggests that a major part of the convergence wastaken up by eastward extrusion of the lithosphere alongmajor strike-slip faults activated by the collision [Tappon-nier et al., 1982]. Constraints on the timing and spatialdistribution of crustal deformation in the Tibetan plateau andsurrounding regions, which are essential for illumination ofthe dynamics of collisional crustal deformation, remainsparse and often controversial [Coleman and Hodges,1995; Harrison et al., 1995; Chung et al., 1998; Yin andHarrison, 2000].[3] The Cenozoic deformation history of the Tarim basin

in northwestern China (Figure 1) may provide usefulinsights into the problems of stress propagation and strainpartitioning following the Indo-Asian collision. Locatedbetween the Tibetan plateau and the Tian Shan orogen,the relatively rigid Tarim block behaved as a secondary‘‘indenter,’’ transmitting collisional stresses to the Tian Shan[Molnar and Tapponnier, 1975; Neil and Houseman, 1997].The Cenozoic deformation of the Tarim basin was closelycoupled with mountain building in the northern Tibetanplateau and the Tian Shan, which caused major subsidenceof the Tarim basin near its northern and southwesternmargins. The nearly 10-km-thick Cenozoic sedimentaryrocks derived from the surrounding mountain belts [Li etal., 1996; Jia, 1997] provide useful constraints on thedeformation history of the Tarim basin and mountainbuilding in the surrounding orogens.[4] In this study we investigate the Cenozoic deforma-

tion history of the Tarim basin and its implications formountain building in the northern Tibetan plateau and theTian Shan orogen. The basement deformation during threeperiods (Paleogene, early-middle Miocene, late Miocene-Pliocene) is reconstructed by backstripping the sedimen-

TECTONICS, VOL. 21, NO. 6, 1059, doi:10.1029/2001TC001300, 2002

Copyright 2002 by the American Geophysical Union.0278-7407/02/2001TC001300$12.00

9 - 1

tary strata. We then simulate the basement deformationusing two-dimensional (2-D) and three-dimensional (3-D)finite element methods. The results show uplift of the westKunlun Shan and the Tian Shan orogens starting in thePaleogene and accelerating since Miocene, with significantinternal variations, and reduced tectonic load along theAltyn Tagh fault since late Miocene.

2. Geological Background

[5] The rhomb-shaped Tarim basin is surrounded by theTian Shan (‘‘Shan’’ means mountain ranges in Chinese) tothe north, the western Kunlun Shan and the Altyn TaghShan of the northern Tibetan plateau to the south andsoutheast, and the Pamir to the west (Figure 1a). Thesemountain ranges originated in the late Paleozoic [Allen etal., 1993] or even earlier [Nakajima et al., 1990] andrejuvenated following the India-Eurasia collision 50–70million years ago [Molnar and Tapponnier, 1978; Windleyet al., 1990; Yin and Harrison, 2000]. Cenozoic deforma-tion of the Tarim basin was coupled with mountain buildingin the surrounding orogenic belts through the bounding faultsystems: the western Kunlun thrust belt and the Altyn Taghfault to the south, and the southern Tian Shan thrust belts tothe north (Figure 1b).

2.1. Western Kunlun Shan and Western KunlunThrust Belt

[6] The western Kunlun Shan is composed mostly ofPrecambrian schist and gneiss overlain by Paleozoic andMesozoic carbonate and clastic sequences [Yin and Harri-son, 2000]. The present western Kunlun Shan is defined as aCenozoic fold and thrust belt in the north and the Karakaxifault in the south (Figure 1b). Southward subduction of theTarim plate beneath the northern Tibetan Plateau is sug-gested as the cause of the western Kunlun thrust belt [Lyon-Caen and Molnar, 1984]. Geologic mapping in the easternpart of the western Kunlun thrust belt [Cowgill et al., 2000]suggest that the magnitude of north-south shortening acrossthe eastern part of the thrust belt is between 50 and 100 km.This magnitude appears to increase westward based onpaleomagnetic studies [Yin and Harrison, 2000].

2.2. Altyn Tagh Fault

[7] The Altyn Tagh fault connects the western KunlunShan in the west and the Nan Shan in the east. It extendsover 1200 km and bounds the northern boundary of theTibetan Plateau (Figure 1b).[8] The Altyn Tagh fault may have played a key role in

eastward extrusion of Tibet during the Indo-Asian collision[Tapponnier and Molnar, 1977]; however, many aspects ofthe fault remain controversial. The displacement history ofthe Altyn Tagh fault has been derived mainly from off-setting of Quaternary features. The estimated Quaternaryleft-slip rates range from 4 to 30 mm/yr in various segmentsof the fault [Molnar et al., 1987; Peltzer et al., 1989;Meryeret al., 1996]. Estimates of the total magnitude of left slip onthe Altyn Tagh fault varies from �1200 km to �200 km[Peltzer and Tapponnier, 1988; Burchfiel et al., 1991;

Zheng, 1994]. The time of initiation of the Altyn Tagh faultis suggested to be Oligocene [Wang, 1990], middle Miocene[Peltzer and Tapponnier, 1988; Yin and Nie, 1996] orPliocene [Bally et al., 1986; Burchfiel et al., 1991] basedon various geological arguments.

2.3. Tian Shan and Southern Tian Shan Thrust Belt

[9] Bordering the Tarim basin to the north is the TianShan mountain belt, which extends broadly east-west forover 2500 km across central Asia, with peaks exceeding7000 m and basins lower than 100 m below sea level(Figure 1). A middle Paleozoic Tian Shan orogen may havedeveloped from one or more arc-continental collisions[Coleman et al., 1992; Allen et al., 1993; Carroll et al.,1995], followed by reactivation of the older faults caused bysuccessive collision of island-arc systems onto the southernmargin of Asia in the late Paleozoic and the Mesozoic[Hendrix et al., 1992; Sobel, 1995].[10] During the Cenozoic the Tian Shan was rejuvenated

by the Indo-Asian collision [Tapponnier and Molnar, 1979;Burchfiel et al., 1991; Lu et al., 1994]. The present TianShan is flanked by east-west trending, active thrust systemson its north and south sides [Tapponnier and Molnar, 1979;Burchfiel et al., 1991; Avouac and Tapponnier, 1993;Molnar et al., 1994; Abdrakhmatov et al., 1996]. Betweenthe Tian Shan and the Tarim basin is the southern Tian Shanthrust belt, which may be divided into four segments basedon their deformation styles and structural trend [Yin et al.,1998]. From west to east (Figure 1b), they are (1) the Kashi-Aksu thrust system, (2) the Baicheng-Kuche thrust system,(3) the Korla transfer system, and (4) the Lop Nor thrustsystem [Yin et al., 1998; Allen et al., 1999]. The western-most Kashi-Aksu system is characterized by the occurrenceof evenly spaced (12–15 km) imbricate thrusts. The Bai-cheng-Kuche and Korla systems are expressed by a majornorth dipping thrust (the Kuche thrust) that changes itsstrike eastward to become a NW striking oblique thrustramp (the Korla transfer zone). The Lop Nor systemconsists of widely spaced thrusts, all involved with base-ment rocks. The age of initial thrusting is estimated to be20–25 Ma [Hendrix et al., 1994; Yin et al., 1998].

2.4. Tarim Basin

[11] The Tarim basin has a complex and protractedhistory of deformation since the late Precambrian [Hendrixet al., 1992;Wang et al., 1992; Carroll et al., 1995; Li et al.,1996]. The basement consists of Archean and Proterozoicmetamorphic rocks [Tian et al., 1989; Chai et al., 1992].The sedimentary cover, including Proterozoic and Phaner-ozoic sequences, ranges from 17 km within major depres-sion centers to 5 km in the central uplift [Li et al., 1996].The Paleozoic sequence is dominated by platform carbo-nates and fine-grained clastics [Carroll et al., 1995]. Mes-ozoic strata of north central Tarim are interpreted to recorddeposition as the result of multiple episodes of contractionand uplift of the margins of the basin [Hendrix et al., 1992].These periods of conglomerate deposition may be related tothe accretion of the Qiangtang block in the Permian-Juras-sic, the Lhasa block in the latest Jurassic, and the Kohistan-

9 - 2 YANG AND LIU: CENOZOIC DEFORMATION OF THE TARIM PLATE

Dras arc complex in the Late Cretaceous [Hendrix et al.,1992]. Since then it has been largely associated with fore-land depressions caused by basinward thrusting and over-loading of the Tian Shan in the north and the westernKunlun Shan in the south [Jia, 1997; Metivier and Gau-dermer, 1997].[12] During the Cenozoic the Tarim basin developed into

a complex system of foreland basins largely in response to

the Indo-Asian collision. Basinward thrusting of the TianShan and western Kunlun Shan caused further subsidenceof the foreland depressions where Cenozoic sedimentaryrocks are up to 10 km thick (Figure 2). During the Eocene,sedimentation was mainly centered in the Hotan (or thesouthwestern) depression adjacent to the western Kunlunthrust belt. By the end of Pliocene, both the Hotan and theKuche depressions adjacent to the Tian Shan orogen had

Figure 1. (a) Topographic relief map of the Tarim basin, NW China, and surrounding regions. The dotsmark the approximate locations of some of the industrial wells drilled before 1990 [Zhao et al., 1997]. (b)Simplified geological map of the Tarim basin and surrounding regions [after Yin et al., 1998].

YANG AND LIU: CENOZOIC DEFORMATION OF THE TARIM PLATE 9 - 3

received >5000 m sediments (Figure 2). Cenozoic sedi-ments in the basin were mainly terrestrial facies. Thesedimentary strata in the Tarim basin have been welldocumented from dense oil wells drilled over the pastdecades [Jia, 1997].

3. Basement Deformation History

[13] To understand the causes of Cenozoic deformation ofthe Tarim basin we need to reconstruct the basementdeformation history. The basic data are the isopach of theCenozoic sedimentary rocks. Petroleum and gas explorationin the Tarim basin for the past three decades has produceddetailed information about the sedimentary cover of thebasin. Based on industrial drilling, shallow seismic reflec-tion, and geological studies, the Chinese Petroleum and GasCorporation produced the isopach maps [Jia, 1997] (Figure2) that are used in this study. Because of industrial secrecy,no details of the location of the oil wells or seismic linesused to construct the isopach maps were given by Jia[1997]. We obtained the locations of 55 industrial oil wellsdrilled before 1990 from Zhao et al. [1997] and show themin Figure 1a. This information is incomplete, and theisopach maps (Figure 2) were constructed in 1995, andpresumably were based on data from more drill wells thanthose shown in Figure 1a. Both Jia [1997] and Figure 1aindicate that there are no drill well data in the Kepingtagearea. The isopachs in the Kepingtage area are generated byinterpolation and therefore are not used in our calculationsof basement deformation or sedimentary loads. The generalisopach patterns in Figure 2, including the separated depo-centers at Yecheng and Hotan in the Paleogene, are con-sistent with other published data [Metivier and Gaudermer,1997; Allen et al., 1999; Metivier et al., 1999]. Whereas thelack of detailed information of the source data makes itdifficult to assess the errors in these isopachs, the isopachmaps in Figure 2 are probably sufficient for a first-ordermodeling.[14] The depth of the basement (relative to present basin

surface) during the youngest period (late Miocene-Pliocene)may be estimated from the thickness of sedimentary rocksformed during this time (Figure 2c). Basement deflectionduring the earlier periods can be reconstructed using back-stripping [Lerche, 1990; Makhouse et al., 1997]. As illus-trated in Figure 3, the original depth of layer A can berestored by stripping off the overlying layer B and correct-ing for compaction. Assuming the porosity f decreases withdepth following the empirical relation [Falvey and Middle-ton, 1981]:

f ¼ f0 exp �czð Þ; ð1Þ

where z is depth and c is a constant, mass conservation ofthe solid material in layer A can be written as

z2 � z1 þf0

cexp� cz2ð Þ � exp �cz1ð Þ½ � ¼ d þ f0

cexp �cdð Þ � 1½ �;

ð2Þ

where z1 and z2 are the depth to the bottom of layer B andlayer A, respectively, and d is the original thickness of layer

A before layer B was deposited (Figure 3). In the calcu-lations we used the values of f0 = 0.429 and c = 0.675km�1 determined by Liu et al. [1996] based on drill holedata from the Tarim basin.[15] Figure 4 shows the reconstructed history of basement

deformation for the Paleogene and the early-middle Mio-cene. During Paleogene the Hotan depression was sepa-rated from the broad southwest depression centered nearYecheng, where the basement depression was up to 2200 m(Figure 4a). Minor depression centers also developed alongthe southeastern margin of the Tarim basin, near Minfengand Qiemo. By early Miocene both the southwestern andthe Kuche depression centers deepened and propagatedbasinward (Figure 4b). Note that the result in Figure 4b isthe incremental subsidence during the period of early-middle Miocene. We consider the Paleogene sedimentaryrocks as part of the basement by this time. The Hotandepression became part of the southwestern depressionwhere the maximum subsidence reached 3500 m nearYecheng. The incremental basement subsidence since lateMiocene can be estimated directly from the isopach of thisperiod (see Figure 2c). It is clear that basement subsidenceaccelerated since Miocene. Near Yecheng the incrementalbasement depression was up to 5500 m during late Mio-cene-Pliocene. Similar basement depression occurred in theKuche depression, which branched southwestward to formanother depression center near Aksu (Awati) where thesubsidence reached 5000 m. Not much subsidence occurredalong the southeastern margin of the Tarim basin during lateMiocene-Pliocene.

4. Causes of Basement Deformation:

Sedimentary and Tectonic Loads

[16] It is clear from Figures 2 and 4 that basementdeformation of the Tarim basin through the Cenozoic wasdominated by the development of two major foreland basinsadjacent to the western Kunlun Shan and the Tian Shanmountain ranges. The theory of foreland basin formationresulting from flexural deflection of the lithosphere has beenwell developed [Turcotte, 1979; Beaumont, 1981]. Flexureof a thin elastic plate can be described by [Turcotte, 1979]

Dr4wþ pr2w ¼ q; ð3Þ

where w is deflection, r2 and r4 are the Laplace and thebiharmonic operators, respectively, p is horizontal load andq is vertical load per unit area on the lithosphere. For theflexure of continental lithosphere the vertical load may bewritten as q = qt + rsgw0 � rmgw, where qt is the tectonicload, and (rsgw0 � rmgw) is the net force of sediment loadsand the buoyancy force arising from the displaced mantle.Here rs is the density of the overlying sedimentary material,g is gravitational acceleration, w0 is the thickness of thesediment layer, and rm is the density of the underlyingmantle. Thus equation (3) could be written as

Dr4wþ pr2wþ rmgw ¼ qt þ rsgw0: ð4Þ

For a perfectly compensative basin of elastic flexure, i.e., alldeflection is filled with sediments, w0 and w would be


identical and equation (4) can be written in the morefamiliar form [Turcotte, 1979]:

Dr4wþ pr2wþ rm � rsð Þgw ¼ qt:

However, equation (4) is more useful for illustrating ourapproach: we used rsgw0 as the sedimentary load because

w0 is derived directly from the sedimentary records,whereas w is the theoretical basement deflection. Theparameter D is the flexural rigidity:

D ¼ Eh3

12 1� n2ð Þ ; ð5Þ

Figure 2. Isopach of sedimentary rocks for (a) Paleogene; (b) early-middle Miocene; (c) late Mioceneto Pliocene (data are from Jia [1997]).


where E is the Young’s modulus, n is the Poisson’s ratio,and h is the thickness of the elastic plate. For lithosphericflexure an effective elastic thickness, Te, may be used toreplace h [Watts, 1976; McNutt, 1984]. Equation (3) can bemodified for other rheologies [Turcotte, 1979; McMullen etal., 1981; Burov and Diament, 1992]. Various rheologieshave been applied to foreland basins [Beaumont, 1981;Jordan, 1981; Royden and Hodges, 1984; Patton andO’Connor, 1988] and other tectonic settings [Walcott,1970; Melosh, 1978; Turcotte, 1979; Karner and Watts,1983; Watts and ten Brink, 1989], with varying degrees ofsuccess. In many cases an elastic rheology has been shownto be a good first-order approximation [Turcotte, 1979;Burov and Diament, 1992], and the effects of various factorson the lithospheric rheology, such as thermal structures andmechanical weakening, may be lumped into Te [McNutt,1984; Burov and Diament, 1995]. Note that equations (3)and (4) are for thin elastic plate only, and we used these

Figure 4. Basement deformation in (a) Paleogene and (b) early-middle Miocene derived frombackstripping. Basement deformation for the youngest period (late Miocene-Pliocene) can be estimateddirectly from the isopach in Figure 2c. The two straight lines in (a) show the location of the 2-D models inFigures 6 and 7.

Figure 3. Sketch showing reconstruction of basementsubsidence using backstripping. The original depth of layerA is restored through peeling off layer B and correcting forcompaction.


equations here to illustrate the roles of sedimentary andtectonic loads in basement deformation. Although the thinelastic plate may be a viable approximation of the Tarimplate, we have developed a fully three-dimensional modelbased on the elastic theory [Timoshenko and Goodier,1970]. The models solve for the full sets of 15 linearequations (six stress and six strain components plus threedisplacement components) without requiring the thin-plateapproximation. We also used a viscoelastic model and willcompare the results below.[17] The major loads causing the flexural foreland basin

development in the Tarim basin include sedimentary loadsdistributed across the basin and tectonic loads near themargins of the basin associated with overthrusting of thesurrounding mountain belts (Figure 5). Horizontal load isusually neglected in elastic flexure models because of theunreasonably high values of the load (6.4 GPa for a 50 kmthick elastic plate) required to cause buckling [Turcotte andSchubert, 1982]. However, Karner [1986] pointed out thathorizontal load could have significant effects for plateswith preexisting deformation, such as sedimentary basins.Compressional horizontal load tends to amplify the platedeformation; the amount of amplification is a function ofthe elastic thickness and the wavelength. We do not includehorizontal load in the model because its value is difficult toconstrain and doing so would increase the number of freeparameters, thus introduce large uncertainties to the tec-tonic loads, the major target of this study. ApplyingKarner’s [1986] results to the Tarim basin, we estimatedthat <5% of error may be introduced by neglecting thehorizontal load. Although bending moments may alsocontribute to flexure, they are difficult to constrain fromgeological data and therefore are not included in ourcalculations.[18] Because the spatial distribution of the sedimentary

rock (sedimentary loads) for each of the three periods areknown (see Figure 2), the optimal tectonic loads for eachperiod may be derived from matching the observed base-ment deformation for a given rheological structure. For thisstudy we developed both two-dimensional (2-D) and three-dimensional (3-D) finite element models using FEPG(available at http://www.fegensoft.com/english/index.htm),an automatic finite element codes generating system [Yang,1998]. Both the 2-D and 3-D models solve for the full set oflinear elastic equations without requiring the thin-plateapproximation. We verified the numerical results by com-

parison with the analytic solutions of numerous simple 2-Dproblems.

4.1. Two-Dimensional Models

[19] Most previous studies of foreland basin formationwere based on 2-D models [Beaumont, 1981; Burov andDiament, 1992], because they are easy to compute anduseful for illustrating the effects of major model parameters.We conducted a series of 2-D finite element modeling toexplore the model parameters and the relative roles ofsedimentary and tectonic loads in the basement deformationof the Tarim basin. Figure 6 compares the predicted andobserved Paleogene basement subsidence along a profilefrom Yecheng to Kuche. Using the sedimentary loads alongthe profile (see Figures 4a and 2a), we could determine therequired tectonic loads by fitting the observed basementdeformation. The problem here is Te, the effective elasticthickness of the Tarim plate, that is also unknown. There are

Figure 5. Sketch showing Cenozoic deformation of the Tarim basin resulting from the distributedsedimentary loads and tectonic loads near the margins of the basin associated with overthrusting of thesurrounding mountain belts. Given the sedimentary loads for each of the three periods (Figure 2), theoptimal tectonic loads may be estimated by fitting the observed basement deformation.

Figure 6. Comparison of the observed and predictedPaleogene basement flexure along the Yecheng-Kucheprofile (location in Figure 4a) by the 2-D model. The topplot shows the location and magnitude of the optimaltectonic loads. The optimal Te is 60 km for the Tarim plate.


certain trade-offs between the Te value and the tectonicloads (see below), and the method of trial and error wasused to find the best combination of these two variables.The density of the sedimentary material and the mantle areassumed to be 2600 kg/m3 and 3300 kg/m3, respectively,and the other elastic constants are represented implicitly bythe Te value. The Winkler spring mattress is used at thebottom to simulate restoring body forces induced by dis-placement of density boundaries [Williams and Richardson,1991]. For this profile the best fit is found when Te = 60 km(Figure 6), and nearly 70% of the observed basementsubsidence can be explained by the sedimentary loads.The largest misfits are near the margins of the basin whereadditional loads may result from overthrusting of theadjacent fold and thrust belts, i.e., tectonic load. Applyingthe optimal tectonic loads significantly improved the fit(Figure 6).[20] We should note that the boundary between the Tarim

basin and the surrounding mountain ranges are not clear-cut.Thus, it is difficult to completely separate the sedimentaryload from tectonic load. Some of the sedimentary rocks nearthe margins of the basin, while being counted as sedimen-tary load in our model, are part of the fold and thrust beltsassociated with mountain building and hence perhapsshould be regarded as tectonic load.[21] For elastic flexures one controlling parameter is the

effective elastic thickness. The physical effects of various Tecan be seen from Figure 7, which shows the predicted andobserved Paleogene basement deformation along a profilefrom Qiemo to Kuche (see location in Figure 4a). A thickerplate (larger Te) allows the effects of tectonic load topropagate further into the basin, resulting in smoother andbroader flexure. A thinner plate, on the other hand, tends tolocalize subsidence to where the loads are applied.

[22] The good fits between the predicted and observedbasement deformation in Figures 6 and 7 also illustrate thelimitations of 2-D models for the Tarim basin. In 2-D modelswe are free to choose the tectonic loads and Te, consequentlywe can always find a combination of these parameters thatproduce close fit to the observed basement deformation.However, different results may be obtained for differentprofiles. For example, the optimal tectonic load at Kucheis 3 MPa over an 80-km width for the Yecheng-Kucheprofile (Figure 6), but 45 MPa over a 20-km zone for theQiemo-Kuche profile (Figure 7). Furthermore, the optimal Teis �60 km for the Yecheng-Kuche profile but 15 � 30 kmfor the Qiemo-Kuche profile, and there is no geologicalevidence supporting such a sharp lateral change of thelithospheric rheology over the Tarim basin. Part of theproblem may be the location of the 2-D profiles. TheYecheng-Kuche profile is perpendicular to the Southwesterndepression but not to the Kuche depression, thus is not anideal representation for deformation in the Kuche region.The particular geometry of the Tarim basin, with limitedspatial extension in all directions, and the complex internaldeformation patterns, are difficult for 2-D representations. Inthe following we shall focus on three-dimensional models.

4.2. Three-Dimensional Models

[23] Figure 8 shows the numerical mesh of the 3-D model.The rheology of the Tarim plate is assumed to be eitherelastic or viscoelastic. We focus on elastic flexure here anddiscuss the results of viscoelastic models later. The elasticmodel of the Tarim plate has two layers (Figure 8). TheYoung’s modulus is taken to be 3 1010 Pa for the top layerand 5 1010 Pa for the lower layer. The somewhat lower

Figure 7. Results of a 2-D flexural model for the Qiemo-Kuche profile (location in Figure 4a) showing the effects ofTe. The top plot shows the location and magnitude of theoptimal tectonic loads.

Figure 8. Numerical mesh of the 3-D finite elementmodel, which includes two layers with different elasticproperties. Tectonic loads within each of the labeledboundary domains are assumed to be uniform for eachperiod. KL1–KL4: western Kunlun Shan; AT1–AT3: AltynTagh fault; and TS1–TS7, Tian Shan. Five weak zonesshown in white were included to simulate the effects offaults for the late Miocene-Pliocene period. See text fordetailed discussion.


value for the top layer is to reflect a relatively weaker uppercrust because of fractures and the sedimentary cover. Weused multiple layers in the model to explore the effects ofrheologic variations with depth, although for elastic modelsthe results are dependent mainly on the total effective elasticthickness. The Poisson’s ratio is taken as 0.25 for allelements. These elastic constants are within the typicalvalues for crustal rocks [Turcotte and Schubert, 1982]. Themodel includes 3102 nodal points and 1932 isoparametericsolid elements, providing a horizontal element size of 25–30km. We divide the margin zone of the Tarim basin into 14domains. Domains KL1 – KL4 are along the westernKunlun Shan thrust belt, AT1–AT3 are along the AltynTagh Shan and fault system, and TS1–TS7 line from west toeast along the Tian Shan thrust belt. Within each domain thetectonic load is assumed to be uniform for a given period,and the optimal values are obtained by regression statedbelow. The domain division roughly reflects the generalgeological settings, but the number of boundary domainsand their boundaries are somewhat arbitrary.[24] We constrain the horizontal displacement along the

eastern side, which is allowed to move only vertically. Thisboundary condition prevents artificial rigid-body translationand rotation. Other sides of the model plate are free, and theWinkler spring mattress is used at the bottom to simulaterestoring body forces induced by displacement of densityboundaries [Williams and Richardson, 1991].[25] Similar to the approach used in 2-D models, we

determine the optimal tectonic loads by fitting the predictedflexure to the observed basement deformation. For a givenrheology structure, the tectonic load on each of the boun-dary domains during each geological period may be deter-mined. A major problem we face here, as in 2-D models, isthat we have to determine the trade-off between the tectonicloads and the effective elastic thickness. In 2-D models weused the method of trial and error to look for the combina-tion of tectonic loads and Te that produces the best fit to theobserved basement deformation. This caused the problemsof having large variations of Te along different profiles, asdiscussed above. The situation is greatly improved here byhaving an additional constraint, the requirement of linearity.Because elastic flexure is a linear process, the basementdeformation should be the sum of deformation caused bythe sedimentary loads over the whole basin and the tectonicload on each of the boundary domains. Thus we can choosethe Te value that gives the highest linearity, determine by theF-test (see below), and obtaining the optimal tectonic loadon each of the margin domains by linear regression.[26] Assuming Wi(x, y) is the response to tectonic load of

unit magnitude on the ith boundary domain, W0(x, y) is theresponse to sedimentary load, and W*(x, y) is the observedbasement deflection (see Figure 4), the linear assumptioncan be expressed as

W* x; yð Þ ¼ W0 x; yð Þ þX14

i¼1

aiWi x; yð Þ; ð6Þ

where ai is the magnitude of tectonic load on the ithboundary domain. We use the method of multivariate linearregression to determine the values of ai, and the F-test to test

the linear correlation between W*(x, y) and the tectonicloads on the 14 boundary domains. A high confidence levelfrom the F-test will indicate the assumption of linearity isacceptable.[27] We started with a uniform Te for the entire Tarim

plate. For a given Te value, the optimal tectonic load on eachof the boundary domain can be derived by linear regression,and the F-test is conducted to score the linear correlationbetween observed basement deflection and those caused bythe tectonic and sedimentary loads. The process is repeatedsystematically with different Te values. The best Te value isthe one with which the highest confidence level of the F-testis reached.[28] The results for the three Cenozoic periods are shown

in Table 1. For the Paleogene the best fit is reached when Te =60 km. The corresponding confidence level of the F-test (a)is above 97%, and the correlation coefficient (R) is 0.80, sothe linearity assumption is acceptable. In other words, thePaleocene basement deformation can be satisfactorilyexplained by the summation of flexural response to thesedimentary loads and tectonic loads on the margins of auniform elastic plate. The optimal tectonic load on each ofthe boundary domain for this period is shown in Table 1and discussed later. Similar quality of the linear regressionis obtained for the early-middle Miocene period. However,the assumption of linearity is unacceptable for the LateMiocene-Pliocene period: the best confidence level fromthe F-test is only 35% with the optimal Te = 156 km, andunreasonably high tectonic loads (up to 1110 MPa or a41 km high mountain on some boundary domains) are

Table 1. Optimal Tectonic Loads

BoundaryDomain

PaleogeneEarly-MiddleMiocene

Late Miocene-Pliocene

MPa kma MPa kma Mpa kma

KL1 26 0.96 46 1.7 �57 �2.1KL2 43 1.6 101 3.7 286 10.6KL3 21 0.78 �27 �1.0 62 2.3KL4 31 1.1 59 2.2 46 1.7AT1 4 0.15 5 0.18 86 3.2AT2 10 0.37 2 0.07 �40 �1.5AT3 18 0.67 36 1.3 66 2.4TS1 23 0.85 28 1.0 197 7.3TS2 �15 �0.55 �101 �3.7 �152 �5.6TS3 8 0.3 37 1.3 61 2.3TS4 32 1.2 40 1.5 �54 �2.0TS5 5 0.18 18 0.67 94 3.5TS6 �93 �3.4 �38 �1.4 �180 �6.7TS7 8 0.3 5 0.18 49 1.8

PaleogeneEarly-MiddleMiocene

Late Miocene-Pliocene

R 0.80 0.78 0.76F test 1.86 1.72 1.39

Confidence 97% 94% 85%Te 60 km 84 km 108 kmb

D 7.6 1020Nm 2.1 1021Nm 4.4 1021Nm

aTectonic loads converted to the weight of a rock column with a densityof 2700 kg/m3.

bTe is 24 km near Aksu and Minfeng.


predicted. These results indicate that the basement defor-mation during this period cannot be satisfactorily explainedby flexure of a uniform elastic plate.[29] Numerous factors not included in these calculations,

such as internal faulting and the basinward growth of thethrusting belts, may contribute to the poor fit in this case. Wemodified the model to (1) include five major faults in theareas where the fit is particularly poor (see Figure 8), (2)widen the area of tectonic load in the Kepingtage region(boundary domains TS1–TS3) where geological evidenceindicating a significant basinward growth of the thrust belt[Lu et al., 1998], and (3) lower the Te value (to 24 km) nearthe Aksu (Awati) and Minfeng depression where the Tarimplate may have been weakened by intensive late Cenozoicfaulting [Jia, 1997]. With all these factors included in themodel, the confidence level of linearity for late Miocene-Pliocene improved to 85%, and the misfit is significantly

reduced. Inclusion of these factors for earlier periods, how-ever, resulted in poorer fit, suggesting that these featuresprobably developed after late Miocene.[30] Figure 9 shows the misfit between the predicted

flexural deformation and the observed basement deflectionfor the Paleogene. Note that �70% of the basement defor-mation can be explained by sedimentary loading, and thelargest misfits are near the margins of the basin (Figure 9a).The fits are systematically improved by applying the opti-mal tectonic loads on the margin domains (Figure 9b). Therelative roles of sedimentary and tectonic loads are furtherillustrated in Figure 10. Figures 10a–10c shows theresponse to sedimentary loads alone for the three Cenozoicperiods. Although sedimentary loads can explain much ofthe observed basement deformation, there are systematicmisfits for deep depressions that are concentrated near themargins of the basin (Figures 2 and 4). The predicted

Figure 9. Misfit between the predicted flexural deformation and the observed basement deflection forthe Paleogene. (a) The predicted flexure is caused by sedimentary loads alone. (b) The predicted flexureis caused by sedimentary loads and the optimal tectonic loads.


subsidence by sedimentary loads is always lower than theobserved values for deep depressions, indicating the needfor additional tectonic loads. The misfit is greatly reducedwhen the optimal tectonic loads (Table 1) are applied(Figures 10d–10f ).[31] The temporal and spatial evolution of the optimal

tectonic loads is plotted in Figure 11. During the Paleogenethe tectonic loads on the Tarim basin were relatively minor.The maximum load is 43 MPa, roughly the weight of a�1.6-km mountain range, in western Kunlun Shan and 32 MPa inthe central segment of the Tian Shan thrust belt. On theboundary domains TS2 and TS6 the predicted tectonic loadsare negative (upward force or unloading). Tectonic loadsaccelerated since the early Miocene, especially in the west-ern segment of western Kunlun Shan thrust belt and thecentral segment of the Tian Shan thrust belt (note thedifferent vertical scales in Figure 11). During the lateMiocene, tectonic loads along the Tian Shan thrust belt splitfrom the central Tian Shan. In the Kepingtage region tectonic

loads broadened significantly, consistent with geologicaldata [Lu et al., 1998]. Tectonic loads along the westernKunlun Shan thrust belt continued to accelerate, whilemoderate unloading (negative tectonic loads) is predictedfor the central section of the Altyn Tagh fault system. Theseresults may also be compared with the general topography ofthese mountain ranges. For the period of late Miocene-Pliocene, the average tectonic load over western Tian Shanthrust belt (boundary domains TS1–TS4) is larger than thatover the eastern segments (TS5–TS7), consistent to higherelevations to the western part of south Tian Shan. Thepredicted large tectonic load over the western Kunlun Shanand relative small tectonic loads over the Altyn Tagh Shan arealso comparable to the relative topography of these mountainranges. However, caution is needed when Figure 11 iscompared with topography because (1) the results in Figure11 are the incremental tectonic loads for each period, (2) thedivision of the boundary domains is somewhat arbitrary, and(3) other factors may influence the tectonic loads. Some ofthese factors are discussed below.

5. Discussion

[32] We have constructed the temporal and spatial evo-lution of the optimal tectonic loads on the margins of theTarim basin through the Cenozoic that may be associatedwith the collision-driven mountain building of the northernTibetan Plateau and the Tian Shan. Here we briefly discussthe effects of model parameters and the implications for thecollisional tectonics in this region.

5.1. Effects of Model Parameters

[33] One of the most important parameters for elasticflexure is the effective elastic thickness. Many factors thataffect lithospheric rheology (e.g., geothermal condition,lithospheric composition, and mechanical weakening dueto densely developed basement faulting), can be implicitlyrepresented by Te [Burov and Diament, 1995]. In ourmodels the optimal Te values were determined by minimiz-ing the misfit between the observed and predicted basementdeformation in 2-D models and by producing the highestlinearity in 3-D models. Our 2-D models predicted theoptimal Te varying from 60 to 15 km for different profiles,similar to the 2-D flexural results of Burov and Diament[1992]. However, the Te values obtained from the 3-Dmodels are generally greater than 2-D values: they rangefrom 60 km in the Paleogene to 108 km in the LateMiocene-Pliocene for most parts of the Tarim basin. Thelarge discrepancies of the Te values among different profilesindicate the limitations of 2-D models for the Tarim basin.The high Te values from the 3-D models are similar to thatof the India plate (90 km) [Jin et al., 1996] and consistentwith the notion of the Tarim plate behaving as a secondaryindenter during the Indo-Asian collision [Molnar and Tap-ponnier, 1975]. A stiff Tarim plate is also consistent withthe large contrast of deformation styles between the Tarimbasin and the surrounding mountain belts, and is supportedby seismic data that indicate abnormally high velocitystructures extending >220 km depth beneath the Tarim plate

Figure 10. Comparison of the predicted and the observedbasement subsidence. Wobs is the observed basementsubsidence, W1 is the predicted subsidence by sedimentaryloads alone, W2 is the predicted subsidence by sedimentaryand tectonic loads. The straight lines represent perfectfitting. Notice the systematic misfit for deep depressions in(a)–(c). (d)–(f) show the significantly reduced misfits whenthe optimal tectonic loads are applied.


Figure 11. The predicted incremental tectonic loads (see Table 1) for the periods of (a) Paleogene, (b)early-middle Miocene, and (c) late Miocene-Pliocene. Note the different scales for tectonic loads inthese plots.


[Liu and Jin, 1993]. On the other hand, the optimal Tevalues for regions around the Kuche and the Minfengdepression (see Figure 1) are abnormally low (�24 km)for the late Miocene-Pliocene period. This may reflectintensive faulting along the central Tian Shan thrust beltand the Altyn Tagh fault zone in the late Cenozoic. Suchlocalized mechanical weakening near the margins of theTarim plate may also explain the low Te (<45 km) derivedfrom fitting gravity profiles transecting the Tibetan Plateauand the southeastern margin of the Tarim basin [Jin et al.,1996]. For the most part of the Tarim plate, our resultsindicate significant increase of the Te values through theCenozoic (see Table 1). The cause of the increasing stiffnessof the Tarim plate is not clear, but the results are consistentwith the paleogeothermal studies that suggest accelerateddecrease of paleogeothermal gradients, from 3.1�–2.5�C/100 m in the early Tertiary to 2.7�–2.2�C/100 m during thelate Tertiary to �2.1�C/100 m at present [Jia, 1997].[34] For the earlier periods (Paleogene-mid-Miocene) a

uniform Te for the entire Tarim plate produced satisfactoryfit to the observed basement deformation. This may suggestthat internal faulting of the basement had not causedsignificant lateral heterogeneity of the mechanical propertiesof the Tarim plate. However, since late Miocene the base-ment deformation showed strong local variations that reflectcontrol of individual fault systems and cannot be satisfac-torily fit by an intact elastic plate. One example is the BachuUplift in the northwestern part of the basin (Figure 1b),where the fit is particular poor with an intact elastic plate.The Bachu Uplift is bounded by a system of faults. SinceMiocene this area experienced multiple phases of blockuplift as evidenced by the angular unconformity betweendifferent Tertiary strata [Jia, 1997]. When weak zonessimulating the Bachu fault systems are included in themodel (Figure 8), the fit over this region is significantlyimproved. The poor fit to the late Miocene-Pliocene base-ment deformation using an intact elastic plate indicatesstronger fault activity since late Miocene.[35] We have also tested the effects of different model

rheology. Over the long history of Cenozoic deformation ofthe Tarim basin viscous flow and stress relaxation may beexpected, which cannot be simulated in elastic models. Wethus designed a linear (Maxwell) viscoelastic flexure modelwith four layers of different material simulating the mechan-ical properties of the upper and lower crust, the strong layerin the uppermost mantle and the underlying mantle. Theboundary conditions around the Tarim plate are free forvertical displacement but restricted for horizontal displace-ment. Lacking detailed age control on the sedimentarystrata, we assumed the sedimentary and tectonic loads wereapplied linearly during each of the three Cenozoic periods,and a full range of rheological parameters are explored toseek for the best fit between the predicted and observedbasement deformation. In general the results are worse thanthose of the elastic models. The results are generally betterwhen the effective viscosity of the crust and the uppermostmantle is higher. These results are consistent with the notionof the Tarim plate being a relatively cold and stiff blockthrough the Cenozoic [Lyon-Caen, 1986; Liu and Jin,

1993], and suggest that elastic flexure is a reasonably goodfirst-order approximation of the Cenozoic deformation ofthe Tarim basin.

5.2. Implications for Mountain Building in the TibetanPlateau and the Tian Shan

[36] We have shown that the major part of the Cenozoicdeformation of the Tarim basin can be explained by flexureunder the sedimentary loads. Because these sediments weremainly derived from the surrounding orogenic belts [Hen-drix et al., 1992; Metivier and Gaudermer, 1997], the basindeformation can be directly related to the collisional moun-tain building in the surrounding regions. The additionaltectonic loads may provide further insights into the geo-dynamic coupling between the Tarim basin and the sur-rounding Tibetan Plateau and the Tian Shan.5.2.1. Uplift of the Western Kunlun Shanand the Pamir Plateau[37] The timing of tectonic uplift of the western Kunlun

Shan is important for understanding the fundamentalmechanics of the growth of the Tibetan Plateau [Englandand Houseman, 1988; Yin and Harrison, 2000]. 40Ar/39Arand fission track dating [Arnaud et al., 1993; Matte et al.,1996; Sobel and Dumitru, 1997] suggest multiple events ofcooling of the western Kunlun Shan in the Cenozoic, but theinitial age of the western Kunlun Shan uplift in response tothe Indo-Asian collision remains unclear. The �1800-mPaleogene sedimentary rocks in the southwestern depres-sion (Figure 2a) and the average 30 MPa tectonic load alongthe southwestern margin of the Tarim basin (Table 1)suggest that mountain building in western Kunlun Shanstarted at least in the Oligocene, consistent with recentstratigraphic studies in the southwestern depression thatplace the initiation age of the western Kunlun thrust beltin �30 Ma [Yin and Harrison, 2000]. Since the Miocenemountain building in the western Kunlun Shan may haveaccelerated. The thickness of the sedimentary rocks in thesouthwestern depression reached 2800 m during the early-middle Miocene and more than 5000 m since the lateMiocene, and the incremental tectonic loads over the entirewestern Kunlun belt, averaged over margin domains KL1–KL4, are 44 MPa and 84 MPa respectively for these twoperiods. Through the Cenozoic the maximum tectonic loadhas been on the boundary domain KL2 (Figure 8), in frontof the depression center near Yecheng. The acceleratingtectonic loads are consistent with the change in depositionalfacies from distal alluvial plains to proximal alluvial fansnear Yecheng in the western Kunlun Mountains during thepast 4.5 Myr [Zheng et al., 2000].[38] The accelerated tectonic load in this region has

important implications for tectonic evolution and climatechange. The increasing sedimentation rate and the associ-ated change of facies of sediments in this region mayindicate accelerated tectonic uplift or climate change, anda number of recent studies have interpreted these evidenceas indicators of climate change in the past few million years[Molnar and England, 1990; Zhang et al., 2001]. In thisstudy the sedimentary loads have been considered explicitly.


The increasing tectonic load derived here is independent ofsedimentation rates and thus provides an unambiguousevidence for accelerated tectonic uplift of the westernKunlun Shan since late Miocene.[39] Whether the Tarim plate subducted beneath the

western Kunlun Shan has been an issue of debate. Thesouth dipping western Kunlun Shan thrust belt is believed tohave developed in response to the Indo-Asian collision[Lyon-Caen and Molnar, 1984]. Recent seismic surveyshows no clear evidence of a subducting Tarim plate near80.5�E [Kao et al., 2001]. This seems consistent with thepredicted relatively small tectonic load (46 MPa) since lateMiocene on boundary domain KL4, where the seismicprofile was located. However, the predicted tectonic loadincreases westward, jumps to 286 MPa on domain KL2(near Yecheng). This would be equivalent to have a 10-km-high rock column on the domain. Because the division ofthe boundary domains in the model is arbitrary, this resultmay be interpreted as requiring a lower tectonic load spreadover a broader region. The high tectonic load may haveresulted partly from overestimation of the basement sub-sidence near Yecheng, where the present land surface isabout 500 m higher than the average elevation of the basin.By assuming a completely compensative basin and deter-mining the basement depth from the isopachs, we may haveoverestimated the basement deformation near Yecheng by�10%. Nonetheless, the high tectonic load over the westernsegments of the boundary (KL2 and KL3) since lateMiocene seems real and would be difficult to explain byoverthrusting of the western Kunlun thrust belt alone. Itseems necessary to have at least part of the Tarim plateextended southward to beneath the western Kunlun Shan,especially near the western end of the Tarim basin where thepredicted tectonic load is the high through the Cenozoic.[40] The negative tectonic load on the boundary domain

KL1 since late Miocene (Table 1) seems inconsistent withthe indentation and uplift of the Pamir plateau since earlyOligocene, which has caused up to 180 km crustal short-ening in central western Kunlun Shan [Rumelhart et al.,1999]. One possible explanation is that the Karakaxi fault(Figure 1b) may have deepened in the past 10 Myr or so,weakening mechanical coupling between the Tarim basinand the Pamir plateau in this region. This would cause shiftof tectonic loads to the two sides and hence help to explainthe large tectonic loads on the adjacent boundary domainsKL2 and TS1. The origin of the negative tectonic load onthe boundary domain KL3 during early-middle Miocene isnot clear.5.2.2. Uplift of the Tian Shan Mountains[41] The timing of the rejuvenation of the Tian Shan

orogen by the Indo-Asian collision is critical for under-standing the propagation of collisional stresses [Tapponnierand Molnar, 1979; Burchfiel et al., 1991; Lu et al., 1994].Estimates based on fission tracks and stratigraphic analysisplace the minimum age around �25 Ma [Hendrix et al.,1994; Yin et al., 1998]. Our results do not show significantdelay of mountain building in the Tian Shan relative to thewestern Kunlun Shan. The >900-m Paleogene sediments inthe Kuche depocenter (Figure 2a) and �30 MPa tectonic

load from the central Tian Shan thrust belt (domain TS4 inTable 1) indicate that rejuvenation of the Tian Shan startedbefore the end of Oligocene. Since the Miocene the tectonicloads along the Tian Shan-Tarim boundary accelerated andmigrated westward, probably related to the indentation ofthe Pamir plateau. The major phase of uplift in the TianShan probably occurred after late Miocene, indicated by the>5000-m sedimentary rocks in the Kuche and Aksu (Awati)depressions and the tectonic loads in this period (Figure 11cand Table 1). There are also significant variations within theTian Shan belt, such as tectonic unloading in domain TS2since late Miocene that may be correlated to the Bachublock uplift [Jia, 1997]. The predicted basinward broad-ening of the Kashi-Aksu thrust belt (Figure 11c) is consis-tent with geological studies [Lu et al., 1994, 1998]. Theseinternal variations of the optimal tectonic loads may reflectthe tectonic segmentation along the Tian Shan and theTarim boundary [Yin et al., 1998]. The accelerated mountainbuilding in the Tian Shan since late Miocene is alsoconsistent with the deformation rates extrapolated fromgeodetic measurements [Abdrakhmatov et al., 1996].[42] Whereas positive tectonic loads may be related to

overthrusting of mountain belt onto the margins of thebasin, the meaning and origins of negative tectonic loadsare not always clear. For the boundary domains TS2 andTS6, negative tectonic loads are predicted for the entireCenozoic. Although some of these results may be model-dependent artifacts, an inevitable consequence of arbitrarilydividing the boundary domains and making a first-orderapproximation, we cannot dismiss all the negative tectonicloads as artifacts, because many of the predictions areconsistent with geological features in these regions. Asmentioned above, the negative tectonic load on TS2 iscorrelated with the Bachu Uplift, an uplifted-region sincelate Paleozoic. However, the cause of the uplift is unclear.The Upper Cambrian to Permian strata was exposed in TS2,the Kepingtage area [Jia, 1997], suggesting strong erosionthat could explain the negative tectonic loading. The neg-ative load on domain TS2 may also be related to anabnormally hot mantle beneath the central Tian Shan wheremantle upwelling is inferred from seismic evidence [Make-yeva et al., 1992]. The boundary domain TS6 is geograph-ically correlated with the Yanji Basin (Figure 1). However,the origin of the Yanji basin is not clear.5.2.3. Tectonics of the Altyn Tagh Fault System[43] The Altyn Tagh fault has been the focus of intensive

studies because of its hypothesized role in facilitating large-scale eastward extrusion of Asian continent during the Indo-Asian collision [Tapponnier and Molnar, 1977]. Closelyrelated to the current controversy of the displacementhistory of the Altyn Tagh fault is the nature of the faultactivity and the timing of its initiation, which ranges fromthe Oligocene to Pliocene in various studies [Bally et al.,1986; Peltzer and Tapponnier, 1988; Wang, 1990; Burchfielet al., 1991]. Some workers regard the Altyn Tagh fault as afundamental lithologic boundary [Tapponnier et al., 1982;Deng, 1989; Arnaud et al., 1993; Wittlinger et al., 1998];others consider it to be a crustal structure [Burchfiel et al.,1989]. It remains controversial as whether or not the Tarim


plate subducted beneath the Altyn Tagh fault [Lyon-Caenand Molnar, 1984; Cowgill et al., 2000].[44] Because horizontal loading and shearing are not the

dominant causes of the flexure of the lithosphere and are notincluded in the model, the results cannot provide muchconstraint on the timing and magnitude of shearing alongthe Altyn Tagh fault. Nonetheless, some useful insights oftectonic activity along the fault can be derived from theCenozoic sedimentation and tectonic loads along the south-eastern margin of the Tarim basin. The model predictsincreasing tectonic loads in the boundary domains AT1 nearMinfeng and AT3 near Ruoqiang (Table 1 and Figures 8 and1) throughout the Cenozoic, probably related to continuousmountain building in the eastern end of the western KunlunShan and Altyn Tagh Shan. On the contrary, tectonic loadalong the central segment of the Altyn Tagh fault (domainAT2, Table 1) decreased since Miocene and become neg-ative (unloading) during late Miocene. This unloading in thecentral segment of the Altyn Tagh fault is reflected in thechange of the dipping direction of the basement along thispart of the basin boundary, which was southeastward fromthe Paleogene to middle Miocene, but was flipped tonorthwestward after late Miocene, as one may infer fromthe isopach maps (Figure 2). The tectonic unloading may beexplained by deep cutting of the Altyn Tagh fault sinceMiocene, which may have caused dynamic decoupling

between the Tarim basin and northern Tibetan Plateau inthis region. A young deep-cutting Altyn Tagh fault isconsistent with a low-velocity anomaly extending to 140-km depth based on teleseismic imaging across the centralAltyn Tagh fault [Wittlinger et al., 1998], and Quaternarybasaltic eruptions along the trace of western Altyn Taghfault [Deng, 1989].[45] The evolution of tectonic loads in the western

Kunlun Shan, the central part of the Tian Shan (TS3 �TS5) and the central segment of the Altyn Tagh fault arecompared in Figure 12. Uplift in both western Kunlun Shanand Tian Shan may have started in Paleogene and accel-erated since Miocene, where the Altyn Tagh fault is char-acterized by tectonic unloading since late Miocene.

6. Conclusions

[46] The deformation history of the Tarim basin duringthe Cenozoic provides useful insights into mountain build-ing in the Tibetan Plateau and the Tian Shan mountains inresponse to the Indo-Asian collision. Major conclusions wemay draw from this study include the following:1. The Cenozoic history of basement deformation in the

Tarim basin can be largely (up to 70%) accounted for byforeland flexure under the sedimentary loads. Near majordepression centers additional tectonic loads are needed,suggesting that these depressions were associated withtectonic uplift and erosion of the surrounding mountainranges. The fit between the observed and the predictedbasement deformation using a uniform elastic plate wasunsatisfactory for late Miocene-Pliocene and requiredinvolvement of additional factors, such as internal faulting.The rhomb-shape geometry of the Tarim basin and largeinternal heterogeneity make 3-D models particularly useful.2. The Tarim plate is indeed cold and rigid throughout

the Cenozoic. The optimal thickness of the effective elasticplate is 60 km in the Paleogene, 84 km in the early-middleMiocene, and >100 km in most part of the Tarim plate sincelate Miocene. Abnormally low Te (�24 km) near the Aksu(Awati) and Minfeng depression since late Miocene isprobably caused by intensive and localized faulting. Therigid Tarim plate is consistent with seismic tomographyindicating a high velocity anomaly extending to >220 kmdepth under the Tarim basin. The increasing Te throughoutthe Cenozoic is consistent with the cooling history of theTarim basin.3. Tectonic uplift of the western Kunlun Shan mountain

ranges occurred during Oligocene or earlier. Rapid upliftstarted around early-middle Miocene and has been accel-erating ever since. The large tectonic loads required for thisregion cannot be entirely explained by the loads of thewestern Kunlun Shan thrust belt, and suggest that part of theTarim plate has subducted underneath the western KunlunShan mountains. The strongest underthrusting of the Tarimplate may have occurred under the western end of thewestern Kunlun thrust belt.4. Tectonic rejuvenation of the Tian Shan mountains in

response to the Indo-Asian collision is not significantly laterthan mountain building in the western Kunlun Shan. More

Figure 12. Summary of the evolution of tectonic loads inthe western Kunlun Shan, the central segments of the TianShan, and the central segment of the Altyn Tagh fault.Mountain building in both the western Kunlun Shan and theTian Shan started in the Paleogene and accelerated sinceMiocene, whereas the Altyn Tagh fault is characterized bytectonic unloading since mid-Miocene.


tectonic loads were predicted in the western part of thesouthern Tian Shan thrust belts, probably related to theindentation of the Pamir plateau. The large along-strikevariation of tectonic loads since early Miocene is consistentwith geological evidence of fault-controlled segmentation ofthe southern Tian Shan thrust belt. The basinward broad-ening of the Kepingtage thrust belt initiated in LateMiocene.5. Throughout the Cenozoic tectonic loads near Minfeng

and Ruoqiang increased moderately, indicating continueduplift in the eastern end of the western Kunlun Shan and theAltyn Tagh Shan. However, along the central segment of theAltyn Tagh fault tectonic loads started to decrease sinceMiocene and become negative (unloading) since lateMiocene. These results may be explained by dynamicdecoupling between the Tarim basin and the northernTibetan Plateau, probably due to the Altyn Tagh faultcutting the entire lithosphere and the southward migrationof the active fault zone. The same results do not favorsubduction of the Tarim plate under the Altyn Tagh fault.[47] Most studies of collisional tectonics in central Asia

have been focused on the Tibetan Plateau and other moun-tain belts in this region. Our results suggest that much aboutthe collisional orogenesis can be learned from the Tarimbasin and probably the numerous other intermountain basinsin central Asia. At present one major hurdle is the lack ofage constraints on the Cenozoic sedimentary strata in the

Tarim basin and the surrounding regions. However, prom-ising progress has been made in recent years by combiningthe lithostratigraphy, biostratigraphy, magnetostratigraphy,and 39Ar/40Ar thermochronology [Yin et al., 1998; Yin andHarrison, 2000]. Future studies, with improved age con-straints and 3-D models that explicitly include the faultsystems bounding the Tarim basin, should be able tosignificantly refine the picture of temporal-spatial develop-ment of crustal deformation in central Asia in response tothe Indo-Asian collision and help to address many remain-ing questions, such as the control of strain partition betweencrustal thickening and tectonic extrusion following thecollision.

[48] Acknowledgments. We thank Wang Liangshu for turning ourinterests to the Tarim basin, and Lu Huafu, Jia Dong, Gao Rui, LuoZhaohua, and Michael Hamburger for helpful discussion on the regionalgeology and geophysics. An Yin kindly provided preprints, commented onthe earlier version of the manuscript, and helped us to understand the ageproblems of the sedimentary strata in the Tarim and surrounding regions.Ding Zhongyi helped us during the initial stage of the model development.We thank Kelin Whipple (AE) and the Tectonics reviewers, Dennis Harryand Roger Buck, for their very detailed and helpful comments. This study issupported by NSF grant EAR-9805127 and the Research Board of theUniversity of Missouri. Visit to China was partly supported by ChineseNSF grant 49832040 to Lu Huafu as part of the collaborative effort for theChinese NSF’s key project ‘‘dynamic coupling between the Tarim basin andTian Shan.’’

References

Abdrakhmatov, K. Y., et al., Relatively recent construc-tion of the Tian Shan inferred from GPS measure-ments of present-day crustal deformation rates,Nature, 384, 450–453, 1996.

Allegre, C., et al., Structure and evolution of the Hima-laya-Tibet orogenic belt, Nature, 307, 17 –22, 1984.

Allen, M. B., B. F. Windley, and Z. Chi, Paleozoiccollisional tectonics and magmatism of the ChineseTien Shan, central Asia, Tectonophysics, 220, 89–115, 1993.

Allen, M. B., S. J. Vincent, and P. J. Wheeler, LateCenozoic tectonics of the Kepingtage thrust zone:Interactions of the Tien Shan and the Tarim Basin,northwest China, Tectonics, 18, 639–654, 1999.

Argand, E., La Tectonique de L’asie, Proc. Int. Geol.Congr., 7, 171–372, 1924.

Armijo, R., P. Tapponnier, J. L. Mercier, and T. L. Han,Quaternary extension in southern Tibet: Field ob-servations and tectonic implications, J. Geophys.Res., 91, 13,803–13,872, 1986.

Arnaud, N. O., M. Brunel, J. M. Cantagrel, and P. Tap-ponnier, High cooling and denudation rates at Kon-gur Shan, eastern Pamir (Xinjiang, China) revealedby 40Ar/39Ar alkali feldspar thermochronology, Tec-tonics, 12, 1335–1346, 1993.

Avouac, J. P., and P. Tapponnier, Kinematic model ofactive deformation in central Asia, Geophys. Res.Lett., 20, 895–898, 1993.

Bally, A. W., I. M. Chou, R. Clayton, H. P. Eugster,S. Kidwell, L. D. Meckel, R. T. Ryder, A. B. Watts,and A. A. Wilson, Notes on sedimentary basins inChina, U. S. Geol. Surv. Open File Rep., 86-327,108 pp., 1986.

Beaumont, C., Foreland basins, Geophys. J. R. Astron.Soc., 65, 291–329, 1981.

Burchfiel, B. C., Q. Deng, P. Molnar, L. H. Royden,Y. Wang, P. Zhang, and W. Zhang, Intracrustal de-tachment with zones of continental deformation,Geology, 17, 748–752, 1989.

Burchfiel, B. C., P. Zhang, Y. Wang, W. Zhang, F. Song,Q. Deng, P. Molnar, and L. Royden, Geology of theHaiyuan fault zone, Ningxia-Hui autonomous re-gion, China, and its relation to the evolution of thenortheastern margin of the Tibetan plateau, Tec-tonics, 10, 1091–1110, 1991.

Burchfiel, B. C., Z. Chen, K. V. Hodges, Y. Liu, L. H.Royden, C. Deng, and J. Xu, The South Tibetandetachment system, Himalayan orogen: Extensioncontemporaneous with and parallel to shortening ina collisional mountain belt, Spec. Pap. Geol. Soc.Am., 269, 41 pp., 1992.

Burov, E. B., and M. Diament, Flexure of the continen-tal lithosphere with multilayered rheology, Geo-phys. J. Int., 109, 449–468, 1992.

Burov, E. B., and M. Diament, The effective elasticthickness (Te) of continental lithosphere: What doesit really mean?, J. Geophys. Res., 100, 3905–3927,1995.

Carroll, A. R., S. A. Graham, M. A. Hendrix, D. Yang,and D. Zhou, Late Paleozoic tectonics amalgama-tion of northwestern China: Sedimentary record ofthe northern Tarim, northwestern Turpan, andsouthern Junggar basin, Geol. Soc. Am. Bull., 107,571–594, 1995.

Chai, G., X. Wang, and X. Jin, Petroleum geology andoil potential of Tarim Basin, west China, in Ex-ploration and Production, vol. 2, pp. 15–23, JohnWiley, New York, 1992.

Chung, S.-L., C.-H. Lo, T.-Y. Lee, Y.-Q. Zhang, Y. Xie,X. Li, K.-L. Wang, and P.-L. Wang, Diachronousuplift of the Tibetan Plateau starting 40 Myr ago,Nature, 394, 769–773, 1998.

Coleman, D. S., T. P. Frost, and A. F. Glazner, Evidencefrom the Lamarck Granodiorite for rapid Late Cre-taceous crust formation in California, Science, 258,1924–1926, 1992.

Coleman, M., and K. Hodges, Evidence for TibetanPlateau uplift before 14 Myr ago from a new mini-

mum age for east-west extension, Nature, 374, 49–52, 1995.

Cowgill, E., A. Yin, X. Wang, and Q. Zhang, Is theNorth Altyn fault part of a strike-slip duplex alongthe Altyn Tagh fault system?, Geology, 28, 255–258, 2000.

Deng, W., Cenozoic volcanic rocks in the northernNgari district of the Tibet—Discussion on theconcurrent subduction, Acta Pet. Sin., 3, 1 – 11,1989.

Dewey, J. F., and K. Burke, Tibetan, Variscan andPrecambrian basement reactivation: Products ofcontinental collision, J. Geol., 81, 683–692, 1973.

England, P., and G. Houseman, The mechanics of theTibetan plateau, Phiosl. Trans. R. Soc. London, Ser.A, 327, 379–413, 1988.

England, P. C., and G. A. Houseman, Finite strain cal-culations of continental deformation, 2, Compari-son with the India-Asia collision, J. Geophys.

Res., 91, 3664–3676, 1986.England, P. C., and D. P. McKenzie, A thin viscous

sheet model for continental deformation, J. Geo-phys. Res., 70, 295 –321, 1982.

England, P., and P. Molnar, Active deformation of Asia:From kinematics to dynamics, Science, 278, 647–650, 1997.

Falvey, D. A., and M. F. Middleton, Passive continentalmargins: Evidence for a prebreakup deep crustalmetamorphic subsidence mechanism, Int. Geol.Congr., 4, 103–114, Suppl., 1981.

Gansser, A., The Geology of the Himalayas, 289 pp.,Wiley Interscience, New York, 1964.

Harrison, T. M., P. Copeland, W. S. F. Kidd, and A. Yin,Raising Tibet, Science, 255, 1663–1670, 1992.

Harrison, T. M., P. Copeland, W. S. F. Kidd, andO. M. Lovera, Activation of the NyainqentanghlaShear Zone: Implications for uplift of the south-ern Tibetan Plateau, Tectonics, 14, 658 – 676,1995.


Hendrix, M. S., S. A. Graham, A. R. Carroll, E. R.Sobel, C. L. McKnight, B. J. Schulein, and Z.Wang, Sedimentary record and climatic implica-tions of recurrent deformation in the Tian Shan:Evidence from Mesozoic strata of the north Tarim,south Junggar, and Turpan basins, Northwest Chi-na, Geol. Soc. Am. Bull., 104, 53–79, 1992.

Hendrix, M. S., T. A. Dumitru, and S. A. Graham, LateOligocene-early Miocene unroofing in the ChineseTian Shan: An early effect of the India-Asia colli-sion, Geology, 22, 487–490, 1994.

Jia, C., Structural Geology and Petroleum Potential inthe Tarim Basin, China, 438 pp., Pet. Ind. PressHouse, Beijing, 1997.

Jin, Y., M. K. McNutt, and Y. Zhu, Mapping the des-cent of Indian and Eurasian plates beneath the Ti-betan Plateau from gravity anomalies, J. Geophys.Res, 101, 11,275–11,290, 1996.

Jordan, T. E., Thrust loads and foreland basin evolution,Cretaceous, western United States, AAPG Bull., 65,2506–2520, 1981.

Kao, H., R. Gao, R. J. Rau, D. Shi, R. Y. Chen, Y.Guan, and F. T. Wu, Seismic image of the TarimBasin and its collision with Tibet, Geology, 29,575–578, 2001.

Karner, G. D., Effects of lithospheric in-plane stress onsedimentary basin stratigraphy, Tectonics, 5, 573–588, 1986.

Karner, G. D., and A. B. Watts, Gravity anomalies andflexure of the lithosphere at mountain ranges,J. Geophys. Res., 88, 10,449–10,477, 1983.

Larson, K. M., R. Burgmann, R. Bilham, and J. T.Freymueller, Kinematics of the India-Eurasia colli-sion zone from GPS measurements, J. Geophys.Res., 104, 1077–1093, 1999.

Lerche, I., Basin Analysis: Quantitative Methods, vol.1, 533 pp., Academic, San Diego, Calif., 1990.

Li, D., D. Liang, C. Jia, G. Wang, Q. Wu, and D. He,Hydrocarbon accumulations in the Tarim Basin,China, AAPG Bull., 80, 1587–1603, 1996.

Liu, F., and A. Jin, Seismic tomography of China, inSeismic Tomography: Theory and Practice, editedby H. M. Iyer and K. Hirahara, pp. 299 –310, Chap-man and Hall, New York, 1993.

Liu, X., Y. Chen, A. Xiao, and W. Hu, Analysis ofCenozoic subsidence history of Tarim southwestdepression, in New Progress in the Study of Petro-leum Geology in the Tarim Basin, edited by X.Tong, D. Liang, and C. Jia, pp. 170–177, Sci. Publ.House, Beijing, 1996.

Lu, H., D. G. Howell, D. Jia, D. Cai, S. Wu, C. Chen,Y. Shi, Z. C. Valin, and L. Guo, Kalpin Transpres-sion tectonics, northwestern Tarim basin, westernChina, Int. Geol. Rev., 36, 975–981, 1994.

Lu, H., D. Jia, D. Cai, S. Wu, C. Chen, Y. Shi, and L.Guo, The Kalpin transpression structures in north-western Tarim basin, Geol. J. China. Univ. (in Chi-nese), 4, 49–58, 1998.

Lyon-Caen, H., Comparison of the upper mantle shearwave velocity structure of the Indian shield and theTibetan plateau and tectonic implications, Geophys,J. R. Astron. Soc., 86, 727–749, 1986.

Lyon-Caen, H., and P. Molnar, Gravity anomalies of thewestern Tibet and the southern Tarim basin, Geo-phys. Res. Lett., 11, 1251–1254, 1984.

Makeyeva, L. I., L. P. Vinnik, and S. W. Roecker,Shear-wave splitting and small-scale convection inthe continental upper mantle, Nature, 358, 144 –147, 1992.

Makhouse, M., Y. Galushkin, and N. Lopatin, Burialhistory and kinetic modeling for hydrocarbon gen-eration, 1, The GALO model, AAPG Bull., 81,1660–1678, 1997.

Matte, P., P. Tapponnier, N. Arnaud, L. Bourjot, J.Avouac, P. Vidal, Q. Liu, Y. Pan, and Y. Wang, Tec-tonics of western Tibet, between the Tarim and theIndus, Earth Planet. Sci. Lett., 142, 311 –330, 1996.

McMullen, R. J., D. S. Hodge, and F. A. Cozzarelli, Atechnique for incorporating geothermal gradients

and non-linear creep into lithospheric flexure mod-els, J. Geophys. Res., 86, 1745–1753, 1981.

McNutt, M., Lithospheric flexure and thermal anoma-lies, J. Geophys. Res., 89, 11,180–11,194, 1984.

Melosh, H. J., Dynamic support of the outer rise, Geo-phys. Res. Lett., 5, 321–324, 1978.

Meryer, B., P. Tapponnier, Y. Gaudemer, S. Guo, andZ. Chen, Rates of left-lateral movement along theeastmost segment of the Altyn Tagh fault, east of96�E (China), Geophys. J. Int., 124, 29–44, 1996.

Metivier, F., and Y. Gaudermer, Mass transfer betweeneastern Tien Shan and adjacent basins (centralAsia): Constraints on regional tectonics and topo-graphy, J. Geophys. Int., 128, 1 – 17, 1997.

Metivier, F., Y. Gaudemer, P. Tapponnier, and M. Klein,Mass accumulation rates in Asia during the Ceno-zoic, Geophys. J. Int., 137, 280–318, 1999.

Molnar, P., and P. England, Late Cenozoic uplift ofmountain ranges and global climatic change: Chick-en or egg?, Nature, 346, 29–34, 1990.

Molnar, P., and P. Tapponnier, Cenozoic tectonics ofAsia: Effects of a continental collision, Science,189, 419–426, 1975.

Molnar, P., and P. Tapponnier, Active tectonics of Tibet,J. Geophys. Res., 83, 5361–5375, 1978.

Molnar, P., B. C. Burchfiel, K. Liang, and Z. Zhao,Geomorphic evidence for active faulting in theAltyn Tagh and northern Tibet and qualitativeestimates of its contribution to the convergenceof India and Eurasia, Geology, 15, 249 – 253,1987.

Molnar, P., et al., Quaternary climate change and theformation of river terraces across growing anticlineson the north flank of the Tien Shan, China, J. Geol.,102, 583–602, 1994.

Nakajima, T., S. Maruyama, S. Ychiumi, J. G. Liou, X.Wang, X. Xiao, and S. A. Graham, Evidence forlate Proterozoic subduction from 700-Myr-oldblueschists in China, Nature, 346, 263–265, 1990.

Neil, E., and G. Houseman, Geodynamics of the TarimBasin and the Tian Shan in central Asia, Tectonics,16, 571–584, 1997.

Nelson, K. D., et al., Partially molten middle crust be-neath southern Tibet: Synthesis of Project IN-DEPTH results, Science, 274, 1684–1688, 1996.

Patton, T. L., and S. J. O’Connor, Cretaceous flexuralhistory of Northern Oman Mountain Foredeep, Uni-ted Arab Emirates, AAPG Bull., 72, 797 – 809,1988.

Peltzer, G., and P. Tapponnier, Formation and evolutionof strike-slip faults, rifts, and basins during the In-dia-Asia collision: An experimental approach,J. Geophys. Res., 93, 15,085–15,117, 1988.

Peltzer, G., P. Tapponnier, and R. Amijio, Magnitude oflate Quaternary left-lateral displacement alongnorth edge of Tibet, Science, 246, 1285 – 1289,1989.

Royden, L., and K. V. Hodges, A technique for analyz-ing the thermal and uplift histories of eroding oro-genic belts: a Scandinavian example, J. Geophys.Res., 89, 7091–7106, 1984.

Rumelhart, P. E., A. Yin, E. Cowgill, R. Butler, Q.Zhang, and X. Wang, Cenozoic vertical-axis rota-tion of the Altyn Tagh fault system, Geology, 27,819–822, 1999.

Sobel, E. R., Basin analysis and apatite fission-trackthermochronology of Jurassic-Paleogene southwestTarim Basin, Ph.D. thesis, 308 pp., Stanford Univ.,Stanford, Calif., 1995.

Sobel, E. R., and T. A. Dumitru, Thrusting and exhu-mation around the margins of the western Tarimbasin during the India-Asia collision, J. Geophy.Res., 102, 5043–5064, 1997.

Tapponnier, P., and P. Molnar, Active faulting and tec-tonics in China, J. Geophys. Res., 82, 2905–2930,1977.

Tapponnier, P., and P. Molnar, Active faulting and Cen-ozoic tectonics of Tien Shan, Mongolia, and Baikalregions, J. Geophys. Res., 84, 3425–3459, 1979.

Tapponnier, P., G. Peltzer, A. Y. LeDain, R. Armijo,and P. Cobbold, Propagating extrusion tectonics inAsia: New insights from simple experiments withplasticine, Geology, 10, 611 –616, 1982.

Tian, Z. Y., G. L. Chai, and Y. Z. Kang, Tectonic Evo-lution of the Tarim Basin, pp. 33–43, Elsevier Sci.,New York, 1989.

Timoshenko, S., and J. N. Goodier, Theory of Elasticity,McGraw-Hill, New York, 1970.

Turcotte, D. L., Flexure, in Advances in Geophysics,vol. 21, pp. 51 –86, Academic, San Diego, Calif.,1979.

Turcotte, D. L., and G. Schubert,Geodynamics: Applica-tions of Continuum Physics to Geological Problems,450 pp., John Wiley, New York, 1982.

Walcott, R. I., Flexure of the lithosphere at Hawaii,Tectonophysics, 9, 435–446, 1970.

Wang, L., Post-collision model for Late Triassic to LateCretaceous tectonic evolution of the Lower Yangtzearea, in Terrane Analysis of China and the PacificRim, vol. 13, edited by T. J. Wiley et al., pp. 343–344, Circum-Pac. Counc. for Energy and Miner.Resour., Houston, Tex., 1990.

Wang, Q., T. Nishidai, and M. P. Coward, The TarimBasin, NW China: Formation and aspects of petro-leum geology, J. Pet. Geol., 15, 5 – 34, 1992.

Watts, A. B., Gravity and bathymetry in the centralPacific Ocean, J. Geophys. Res., 81, 1533–1553,1976.

Watts, A. B., and U. S. ten Brink, Crustal structure,flexural, and subsidence history of the Hawaiianislands, J. Geophys. Res., 94, 10,473 – 10,500,1989.

Williams, C. A., and R. M. Richardson, A rheologicallayered three-dimensional model of the San An-dreas fault in central and southern California,J. Geophys. Res., 96, 16,597–16,623, 1991.

Windley, B. F., M. B. Allen, C. Zhang, Z. Y. Zhao, andG. R. Wang, Paleozoic accretion and Cenozoic re-deformation of the Chinese Tien Shan Range, cen-tral Asia, Geology, 18, 128 –131, 1990.

Wittlinger, G., P. Tapponnier, G. Poupinet, J. Mei, S.Danian, and F. Masson, Tomographic evidence forlocalized lithospheric shear along the Altyn Taghfault, Science, 282, 74– 76, 1998.

Yang, Y., FEPG—A highly efficient system of generat-ing finite element codes, Mech. Appl. (in Chinese),20, 77 –79, 1998.

Yin, A., and M. Harrison, Geological evolution of theHimalayan-Tibetan orogen, Annu. Rev. Earth. Pla-net. Sci., 28, 211 –280, 2000.

Yin, A., and S. Nie, Phanerozoic palinspastic recon-struction of China and its neighboring regions, inTectonic Evolution of Asia, edited by A. Yin and T.M. Harrison, pp. 442–485, Cambridge Univ. Press,New York, 1996.

Yin, A., S. Nie, P. Craig, M. Harrison, F. J. Ryerson, X.Qian, and G. Yang, Late Cenozoic tectonic evolu-tion of the southern Chinese Tian Shan, Tectonics,17, 1 –27, 1998.

Zhang, P., P. Molnar, and R. Downs William, Increasedsedimentation rates and grain sizes 2– 4 Myr agodue to the influence of climate change on erosionrates, Nature, 410, 891–897, 2001.

Zhao, Z., T. Yong, C. Jia, and Z. Zhang, Stratigraphy inthe Tarim Basin, Pet. Ind. Press House, Beijing,1997.

Zheng, H., C. M. Powell, Z. An, J. Zhou, and G. Dong,Pliocene uplift of the northern Tibetan Plateau,Geology, 28, 715–718, 2000.

Zheng, J., Geodynamics of the Altyn Tagh fault, Geol.J. China. Univ. (in Chinese), 5, 41– 45, 1994.

��M. Liu and Y. Yang (corresponding author),

Department of Geological Sciences, University ofMissouri, Columbia, MO 65211, USA. ([email protected])


Documents

Cenozoic deformation of the Tarim plate and the ...web.missouri.edu/~lium/pdfs/Papers/Yang02Tec.pdf · 2.1. Western Kunlun Shan and Western Kunlun Thrust Belt [6] The western Kunlun