Present-Day Crustal Deformation in China Constrained by Global Positioning System Measurements

See allHide authors and affiliations

Science  19 Oct 2001:
Vol. 294, Issue 5542, pp. 574-577
DOI: 10.1126/science.1063647


Global Positioning System (GPS) measurements in China indicate that crustal shortening accommodates most of India's penetration into Eurasia. Deformation within the Tibetan Plateau and its margins, the Himalaya, the Altyn Tagh, and the Qilian Shan, absorbs more than 90% of the relative motion between the Indian and Eurasian plates. Internal shortening of the Tibetan plateau itself accounts for more than one-third of the total convergence. However, the Tibetan plateau south of the Kunlun and Ganzi-Mani faults is moving eastward relative to both India and Eurasia. This movement is accommodated through rotation of material around the eastern Syntaxis. The North China and South China blocks, east of the Tibetan Plateau, move coherently east-southeastward at rates of 2 to 8 millimeters per year and 6 to 11 millimeters per year, respectively, with respect to the stable Eurasia.

Asia is a modern example of large-scale continental deformation (Fig. 1) and an ideal natural laboratory for its studies. Unfortunately, much of the region is remote, and thus the kinematics of Asia has been, until recently, poorly understood. Although much of its late Cenozoic deformation is explained by the collision and subsequent penetration of India into Eurasia (1), how Eurasia deforms in response to the collision is still subject to debate (2, 3), and a complete kinematic description of deformation over the entire region has not been available. Existing kinematic models (4, 5) rely on sparse data sets that can only describe the complex deformation of Eurasia on length scales of 200 km or larger, and lack data from critical regions. We present a synthesis of GPS velocities in China and its vicinity that provides new insights into the kinematics of Eurasia.

Figure 1

Generalized map of active faults in China and vicinity showing the distribution of GPS stations used in this study. Black lines are major active faults. Circles represent roving GPS stations, and stars represent permanent GPS stations.

Much of the actively deforming part of Eurasia lies within China, including the Tibetan plateau, and parts of the Himalaya, Tian Shan, and Pamir mountain ranges (Fig. 1). Since the early 1990s, several regional GPS networks for active tectonic studies were established in China and neighboring regions (6–15). These networks were surveyed in campaign mode, usually at 1- to 2-year intervals. Each individual network was originally designed to address local problems, and it has been difficult to merge the data together, due to different data analysis strategies. We obtain a self-consistent velocity field by analyzing the original raw data from several different regional networks and merging them into a self-consistent solution.

We combined original data from GPS campaigns carried out between 1991 and 2001 by 10 Chinese and U.S. agencies or universities (16). The regional GPS data were combined with continuous tracking data from a well-distributed set of global International GPS Service (IGS) stations using the GIPSY software (17). For data observed between 1991 and 1995, we used a global solution strategy in which parameters associated with GPS satellite orbits were estimated together with all station coordinates (18). A projection operator was applied to the covariance matrix to remove the components of the covariance matrix that are purely due to reference frame uncertainty (19). For data observed after 1995, a regional solution strategy was adopted, using fixed orbits and satellite clocks provided by NASA's Jet Propulsion Laboratory (20). A subset of IGS stations was used in the regional solutions (21). Next, the daily free network solutions were each transformed into the ITRF97 (International Terrestrial Reference Frame, epoch 1997.0) by estimating a seven-parameter similarity transformation for each (22). We estimated the transformation for each day, on the basis of the common stations that are present both in ITRF97 and in the daily solution, and weighted each station by their respective uncertainties. The 1250 daily solutions were used as data to determine the station velocities and station coordinates at epoch 1995.0 by a standard weighted least-squares adjustment (23). The velocities in ITRF97 were then transformed into velocities in a Eurasia-fixed reference frame (24). The aggregate velocity solution of 354 stations (25) provides an image, to date, of present-day crustal deformation in Asia (Fig. 2). The velocity solutions in both ITRF97 and Eurasia-fixed frames are available atScience's Web site (26). Most velocity uncertainties, propagated by the errors of Eurasia rotation parameters, are in the range of 1 to 4 mm/year, except some stations with an observation interval shorter than 1.5 years. The mean uncertainties in northward velocities relative to Eurasia are 2.2 mm/year and 2.4 mm/year for eastward components.

Figure 2

GPS velocity vectors (mm/year) with respect to the stable Eurasia, plotted on a shaded relief map of the Asia topography (gtopo30). The ellipses denote the region of 1-σ error. The polygons define three regions in which station velocities are used to formulate profiles shown in Fig. 3 and Supplementary figs. 1 and 2 (26).

Stations located on the northern Ganges plains, south of the Himalaya, show northward movement (N19°-22°E) at a rate of 36 to 38 mm/year with respect to stable Eurasia (27), consistent with some previous studies (5, 13, 28). Bangalore (station IISC) in southern India has a velocity of 35.9 ± 1.0 mm/year in the direction of N26.9° ± 1.7°E. The similar velocity between Bangalore and those in the northern Ganges plains indicates that there is no significant deformation within the Indian plate (28). The maximum velocity (∼38 mm/year) of sites in the northern Ganges plains approximates the rate of convergence between the Indian and Eurasian plates. This total convergence rate is intermediate between that estimated in the NUVEL-1A model (∼47 mm/year, N24°E) (29, 30) and that reestimated recently on the basis of a revised plate configuration in the Indian ocean (∼37 mm/year, N38°E) (31). Our estimate is also similar to that (36 ± 3.5 mm/year, N27.1°E) inferred from both Quaternary fault slip rates and GPS geodesy (5).

In the Tian Shan of northwest China (Fig. 1), widespread active faulting and folding, recent uplift, and high seismicity attest to rapid crustal shortening. Across its western part near Kashgar, velocities of 20 to 22 mm/year in the south decrease to ∼0 in the same fashion as seen in Kirghizia, confirming the proposition that the total shortening rate across the western Tian Shan is ∼20 mm/year (7). The total shortening rate between longitudes 81°E and 85°E is only 4.7 ± 1.5 mm/year. Eastward, the rate between longitudes 86° and 87° decreases to <1 mm/year, consistent with conclusion of eastward decreases of crustal shortening inferred from theoretical modeling and geological investigations (32,33). A substantial amount of shortening must be accommodated to the north between the Tian Shan and Altai, as well as farther north in Mongolia.

The North China region includes the Ordos block and the North China Plain. The GPS measurements show coherent east-southeastward movement at rates of 2 to 8 mm/year to the direction of N120° −140°E [Fig. 2 and Supplementary fig. 1 (26)]. Major active faults in the North China region trend north-northeast (N15°-25°E). The oblique intersections with the south-southeastward movement probably produce right-lateral strike-slip components along those faults, which cause major earthquakes with the same prominent component.

On the basis of 36 GPS stations distributed throughout the South China region [Fig. 2 and Supplementary fig. 2 (26)], we obtain an average velocity of 6 to 11 mm/year with respect to stable Eurasia oriented toward an azimuth of N100°-130°E. These results are similar to Very Long Baseline Interferometry (VLBI) measurements (34) and previous GPS measurements (using fewer stations) (13, 14). Because the 36 stations cover the entire South China block, we conclude that it behaves as a rigid block without internal deformation (1, 5, 6,13, 14).

The Tibetan plateau undergoes substantial internal shortening (Fig. 3), with the direction of maximum shortening being ∼N21°E, the inferred India-Eurasia convergence direction (27). Along a N21°E profile passing through the eastern part of the Tibetan plateau and Qilian Shan (Fig. 2), the ∼38 mm/year convergence rate between the Indian plate and the rigid Alashan block north of the Qilian Shan represents more than 90% of the total collision rate between the Indian and Eurasian plates. The N21°E velocity gradient is strikingly linear except for a high-velocity gradient across the Himalaya at the southern margin of the plateau. The average convergence strain rate in this direction is ∼2 × 10−8 year−1. Taking ∼16 mm/year for the convergence rate across the Himalaya, ∼12 mm/year, or more than 30% of the total India-Eurasia convergence, is absorbed by internal shortening of the plateau (Fig. 3). An additional 10 mm/year of contraction occurs across the Qaidam basin and the Qilian Shan (Fig. 3).

Figure 3

GPS velocity profile across the Tibetan Plateau in the direction of N21°E, parallel to the predicted direction of Indian-Eurasian collision. The red diamonds represent the N21°E component of velocity, which is parallel to the direction of the profile, and the green diamonds mark the N111°E component of velocity, which is perpendicular to the profile. The N21°E component shows a general linear trend of velocity gradient except for a high gradient across the Himalaya at the southern margin of the plateau. The N111°E component depicts eastward movement of the Central Tibetan Plateau with respect to both India and Mongolia.

This shortening must be broadly distributed; there would be a significant deviation from the linear trend if there were any single faults that took up a considerable fraction of this shortening. Thus, continuum rather than block-like deformation appears to characterize present-day tectonics of the Tibetan Plateau. This continuum deformation, however, seems to be limited to the plateau itself. Rigid block-like motion appears to characterize deformation of the regions to the north (the Tarim block), northeast (the Alashan and the Ordos block), and east (the South China block), and there are zones of concentrated contraction at both the north and south margins of the plateau (Figs. 2 and 3).

Although contraction appears to accommodate most of the convergence of India, the eastern Tibetan Plateau, south of the left-lateral Kunlun and/or Ganzi-Mani strike-slip faults, is moving eastward relative to both India and Eurasia (Figs. 1 through 3). The Tibetan plateau is not an undeformed wedge-shaped block extruding eastward (35), but instead, the eastern Tibetan Plateau is deforming internally, and moving rapidly eastward. As noticed by Holtet al. (5), the N111°E component of velocities (orthogonal to the convergence direction) increases steadily northward from the Himalaya across the breadth of the Tibetan Plateau, then decreases rapidly as a result of left-lateral slip on the Ganzi-Mani and Kunlun faults, totaling about 12 to 14 mm/year (Fig. 3). If all of this left-lateral motion were on the Kunlun fault, these results would agree with the recent geological slip rate estimate for the Kunlun fault (36).

A series of north-trending normal faults and grabens characterized the eastward extension in the southern Tibet (37, 38). GPS measurements indicated that the rate of extension across southern Tibet between longitudes 80° and 90° is 20 ± 3 mm/year [Supplementary fig. 2 (26)], which is similar to the 18 ± 9 mm/year seismological rate (37).

Sites in the eastern Tibetan plateau show a prominent clockwise rotation. In the eastern margin of the Tibetan Plateau, stations move eastward in western Sichuan, southeastward in northern Yun-nan, and south-southeastward in southern Yun-nan (Fig. 2). Together with stations in southern Tibet, the velocities show a clockwise rotation around the Eastern Himalayan Syntaxis (5, 9, 13).

The eastern part of the Tibetan Plateau moves eastward faster than does South China. Eastward station velocities in east- central Tibet are 21 to 26 mm/year, decreasing to 14 to 17 mm/year on the eastern margin of the Tibetan Plateau and 6 to 10 mm/year in the northern South China Block [Supplementary fig. 2 (26)]. Chen et al. (13) documented a lack of Cenozoic shortening across the Longmen Shan at the eastern boundary of the Tibetan plateau, and GPS results show a lack of present-day convergence as well (9, 13). If we simply assume that the velocity difference between eastern Tibet and South China results from convergence across the Longmen Shan, we would get a 5 to 11 mm/year crustal shortening rate across the Longmen Shan. This is larger than the results (<3 mm/year) of Chen et al.(13), and would represent a serious conflict.

However, the clockwise rotation of the velocity vectors around the Eastern Himalayan Syntaxis suggests an alternate explanation, consistent with that proposed by King et al. (9) and Holt et al. (5). The extruding Tibetan crust rotates around the Eastern Himalayan Syntaxis, causing the southeastward to southward velocities observed in southern Yun-nan Province. Whether this clockwise rotation continues southwestward into Myanmar is unknown but entirely possible, because east-west contraction is suggested in western Myanmar, based on numerical kinematic models (4, 5). Thus, extruding Tibetan crust is neither overriding South China nor pushing it out over a free boundary in the South China Sea; it may be doing so to parts of Southeast Asia.

  • * To whom correspondence should be addressed. E-mail: peizhen{at}


View Abstract

Stay Connected to Science

Navigate This Article