GPS measurements from the Ladakh Himalaya, India: Preliminary tests of plate-like or continuous deformation in Tibet

Transcription

1 GPS measurements from the Ladakh Himalaya, India: Preliminary tests of plate-like or continuous deformation in Tibet Sridevi Jade CSIR Centre for Mathematical Modelling and Computer Simulation, Bangalore , India B.C. Bhatt Indian Institute of Astrophysics, Bangalore , India Z. Yang Department of Surveying Engineering, Chang an University, Xi an , People s Republic of China R. Bendick Department of Geological Sciences, and Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, Colorado , USA and COMET, University of Cambridge, Madingley Road, Cambridge CB3 0EZ, UK V.K. Gaur CSIR Centre for Mathematical Modelling and Computer Simulation, Bangalore , India and Indian Institute of Astrophysics, Bangalore , India P. Molnar Department of Geological Sciences, and Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, Colorado , USA M.B. Anand Dilip Kumar CSIR Centre for Mathematical Modelling and Computer Simulation, Bangalore , India ABSTRACT Observations of relative motion in a geodetic network in Ladakh, India, and across southern Tibet indicate slow shear on the Karakorum fault, rapid east-west extension across the whole of southern Tibet, and constant arc-normal convergence between India and southern Tibet along the Himalayan arc. Measurements of ten campaign-style and six permanent sites with global positioning system (GPS) precise geodesy provide these bounds on the style and rates of the large-scale deformation in the Tibet-Himalaya region. Divergence between sites at Leh, Ladakh, India, and Shiquanhe, western Tibet, as well as slow relative motion among sites within the Ladakh network, limit right-lateral slip parallel to the Karakorum fault to only 3.4 ± 5 mm/yr. This low rate concurs with a recent estimate of 3 4 mm/yr for Late Holocene time, but disagrees with the much higher rate of mm/yr that has been used to Corresponding author bendick@esc.cam.ac.uk. argue for plate-like behavior of the Tibetan Plateau. Convergence between Ladakh and the Indian subcontinent at 18.8 ± 3 mm/yr at 224 ± 17 (1σ) differs little from estimates of convergence across the central segment of the Himalaya. Finally, lengthening of the baseline between Leh, Ladakh, and Lhasa (in southeastern Tibet) at 17.8 ± 1 mm/yr or between Leh and Bayi (farther to the southeast) at 18 ± 3 mm/yr, is consistent with an extrapolation of rates of east-west extension of the Tibetan Plateau based both on shorter GPS baselines (e.g., Lhasa-Simikot) and on diverging slip vectors of earthquakes in the Himalaya. We interpret these results to indicate that Tibet behaves more like a fluid than like a plate. Keywords: geodesy, geodynamics, Tibetan Plateau, Himalaya, Karakorum, continuum mechanics. INTRODUCTION Following the recognition of plate tectonics in largely oceanic domains, two extreme views of large-scale continental tectonics have prevailed. In one, continental tectonics results from the relative movements of a small number of essentially rigid plates of mantle lithosphere, with deep structure blurred by deformation of the overlying thick crust (e.g., Tapponnier et al., 2001); in the other, continental lithosphere undergoes continuous deformation (e.g., England and Jackson, 1989; Molnar, 1988). Asia has provided both the source of data for developing these views and the testing ground exploited by supporters of both views. Among the geological interpretations that separate proponents of each view, perhaps none is more controversial than the role of strike-slip faults. Proponents of plate-like deformation argue that slip on major strike-slip faults, at least in Asia, occurs rapidly and that these faults control the deformation field, such as the growth of the Tibetan highlands (Tapponnier et al., 2001, p. 1671). Proponents of continuous deformation, however, consider slip rates on the strike-slip faults to differ little from those on normal and thrust faults, and that all faults constitute passive markers of discontinuities in the strain field that translate and rotate with a deforming continuum. GSA Bulletin; November/December 2004; v. 116; no. 11/12; p ; doi: /B ; 4 figures; 2 tables. For permission to copy, contact editing@geosociety.org 2004 Geological Society of America 1385

2 JADE et al. Of course, where two extremes prevail, there is always a middle ground; at some level the deformation field seen at the surface of all continental regions can be described in terms of relative movements of blocks of crust. The disparity is primarily a function of spatial scale. At one extreme, blocks may be so numerous (and therefore small) that their relative movements are best treated as resistant inclusions within a continuous medium; at the other, blocks may be so few (and therefore large) that a description in terms of rigid blocks (plates) provides more insight into the deformation field than continuous deformation. A logical boundary on the sizes of blocks might be 100 km, for to remain strong, smaller ones would almost surely not include mantle lithosphere. Conversely, blocks substantially larger than 100 km should include upper mantle. Thus, in the context of testing whether continental tectonics is plate-like or fluid-like, let us lump large-block models in with the former and small-block models with the latter. The mechanical properties of crustal blocks may also depend on the time over which deformation is observed. We make the assumption that geodetic velocities are indicative of the regional velocity over much longer times than the time span of observation. Perhaps the best testing ground for plate-like versus fluid-like behavior is Tibet itself. Major strike-slip faults surround much of the region: the Altyn Tagh fault in the north and the Karakorum fault in the west of Tibet. The area is large enough that if it behaved as a single plate (e.g., Avouac and Tapponnier, 1993), slip on the 1000-km-scale strike-slip faults should be much more rapid that on the network of faults that surrounds blocks of ~ km in dimension within Tibet. Localization of rapid slip on faults bounding Tibet would favor the plate-like view, but slip rates independent of fault length and location would support the fluid-like view. Readers might wonder why tests thus far have failed to be convincing. GPS measurements across the Altyn Tagh fault in segments where most of us expected high rates revealed slip at only ~10 mm/yr (Bendick et al., 2000; Shen et al., 2001), not the expected 30 mm/yr (e.g., Peltzer et al., 1989). Yet, reported Holocene rates of mm/yr (e.g., Mériaux et al., 1999; Tapponnier et al., 2001) contradict the GPS results. We note, however, that Washburn et al. (2001) inferred repetitions of earthquakes in trench logs and estimates of offsets during them that are consistent with a rate of only 10 mm/yr on the central segment of the Altyn Tagh fault. Even a relatively high rate on this fault, however, might not be an adequate test, for it bounds Tibet and the obviously rigid Tarim Basin; simple theory and numerical experiments show that deformation should be localized at the edges of strong inclusions within a deforming continuum (e.g., England and Houseman, 1985; Neil and Houseman, 1997). The other dominant strike-slip fault in the Tibetan Plateau is the Karakorum fault, which crosses the western end of Tibet with a southwesterly strike. Prominent linear valleys, clearly recognizable on satellite imagery, follow the fault trace (e.g., Liu, 1993; Molnar and Tapponnier, 1978). Fresh scarps mark offsets of alluvial and debris-flow fans that are easily seen on the ground and from high-resolution satellite imagery (e.g., Brown et al., 2002; Liu, 1993; Liu et al., 1992). Peive et al. (1964) inferred hundreds of kilometers of right-lateral slip on this fault, and Peltzer and Tapponnier (1988) suggested 650 km of such slip, though Searle (1996) has argued for only 120 km. Although the high terrain (~5000 m) both east and west of the fault requires Cenozoic crustal thickening, active faulting is sparse, if present at all, on the west side of the fault. Thus, in one view, this fault would be the boundary of plate-like Tibetan lithosphere, and in the other it would mark a shear zone within a deformable region if deformation has ceased in parts of the region. Inferred slip rates range from 30 to 35 mm/yr (Avouac and Tapponnier, 1993; Liu, 1993; Liu et al., 1992; Tapponnier et al., 2001), to 11 ± 4 mm/yr (Banerjee and Bürgmann, 2002), to 9 mm/yr (Molnar and Lyon-Caen, 1989), to 8 mm/yr (Searle et al., 1998), and finally to only 3 4 mm/yr (Brown et al., 2002). GPS measurements from the western and eastern ends of the Himalaya also allow us to address other aspects of Asian tectonics relevant to the plate versus fluid controversy. For instance, we may use such data to test whether convergence rates across the Himalaya vary along the belt, as assumed by England and Molnar (1997), by comparing convergence at the extremes of the range with values reported for other Himalayan transects. Geologic inferences of slip can be taken as suggesting as much, for convergence rates range from 10 mm/yr across the Salt Range in Pakistan (Baker et al., 1988), to 14 ± 2 mm/yr and 6 16 mm/yr across the Kangra and Gharwal Himalaya (Powers et al., 1998), to 21.5 ± 1.5 mm/yr in the Nepal Himalaya (Lavé and Avouac, 2000). Geodesy spanning the entire Himalayan belt provides an obvious means of direct comparison of convergence rates without variations introduced by differences in measurement technique or sampled time span, and these GPS convergence estimates show statistically consistent rates of 18 ± 2 mm/yr for eastern Nepal (Larson et al., 1999; Paul et al., 2001), and 15 ± 2 mm/yr north of Delhi (Banerjee and Bürgmann, 2002). Constant convergence rates along strike do not alone preclude a single pole for Tibet-India relative velocity, with additional constraint on kinematics applied by the geometry of the Himalayan arc (Avouac and Tapponnier, 1993). Concurrent with the underthrusting of India beneath the Himalaya, southern Tibet deforms by active normal faulting (e.g., Armijo et al., 1986; Molnar and Tapponnier, 1978; Ni and York, 1978), so that the ESE-WNW dimension of the region extends. Although a rate of extension was initially inferred from the variation in slip vectors of earthquakes along the Himalaya and the rate of convergence across the Himalaya (Molnar and Chen, 1982, 1983; Molnar and Lyon-Caen, 1989), corroboration both from estimates of rates of slip on normal faults (Armijo et al., 1986) and from GPS measurements (Bilham et al., 1997; Jouanne et al., 1999; Larson et al., 1999; Wang et al., 2001) has been limited to a fraction of the relevant zone. GPS measurements from both the western and eastern extremes of the Himalaya test whether extension persists throughout the arc and place a tight constraint on the total rate. Motivated by these questions concerning the rate and direction of movement of the western end of the Himalaya with respect to western Tibet east of the Karakorum fault, to India, and to eastern Tibet, we installed a network of GPS control points in Ladakh, India, which lies north of the Himalaya and near the western end of the Tibetan Plateau. A second small array of GPS measurements near the eastern syntaxis of the Himalaya, east of Lhasa, extends direct geodetic observations across the entire extent of the southern Tibetan Plateau (Fig. 1) when combined with observations reported in the literature (e.g., Wang et al., 2001). MEASUREMENTS AND PROCESSING GPS sites were chosen to estimate the three quantities discussed above: the rate of internal deformation within the Ladakh Himalaya and specifically the part of the strain associated with slip on the Karakorum fault, the rate of convergence across the Himalayan front in the far western and far eastern Himalaya, and the rate of east-west extension of the Tibetan Plateau. Lying 60 km southwest of the Karakorum fault, Leh is sufficiently far from the fault that elastic strain ought not to affect its rate relative to other sites. The other new Ladakh sites, including the farthest eastward, Chushul, lie within 10 km of the fault trace. Thus, assuming that elastic strain accumulates without slip on the fault at Earth s surface, rates of these sites relative to Leh should include only approximately half of the slip rate on the fault, given an assumed dislocation depth 1386 Geological Society of America Bulletin, November/December 2004

3 GPS IN LADAKH Figure 1. The geologic setting of this paper includes the Tibetan Plateau of central Asia. Significant faults discussed in the text are marked with a heavy black line. Faults active in the Quaternary are mapped throughout the Tibetan Plateau (adapted from Armijo et al., 1989; Molnar and Tapponnier, 1978). Normal faults are distinguished with square teeth, strikeslip with arrows indicating sense of slip. Some of the GPS sites are marked with circles and labeled with the name of the nearest city or village. Several sites in the Ladakh array are omitted for clarity. Italic labels are the names of Himalayan subregions. Topography in this figure is from the three-second SRTM digital elevation model. TABLE 1. EPOCHS OF DATA OBSERVATION IN THIS EXPERIMENT Station name 1997 (Julian day) IISC , , , DELHI , , , 181, 184, IRKT SELE LHAS , LEH , 209, , POL TIRTH PNMK STAKSHA LKNG MGLB TNGS CHUS MUTH GONGBU BAYI SOSHIEN KIT , , greater than 3 km. The site at Shiquanhe in western Tibet, however, whose velocity is given in essentially the same reference frame as the rates we calculate here (Wang et al., 2001), allows a bound to be placed on the total slip rate for the Karakorum fault. Sites in southeastern Tibet lie far from large mapped faults, although little direct mapping or measurement has been done in the area. Among sites in Ladakh, only that at Leh has been measured every year since 1997, and it was measured throughout the durations of all campaigns. The three sites in eastern Tibet were measured in 1998 and 2002 (Table 1). We used Bernese 4.2 (Beutler et al., 1987) to analyze the raw GPS observations. We omitted stations with fewer than two days of usable data (station Tirth in the Nubra Valley during 1997 and station Tirisha, Nubra Valley, during 1998). We then solved for the initial coordinates and velocities in the local network relative to the ITRF97 frame (Table 2) by constraining IGS reference tracking station positions and velocities in the region (BAHR, GRAZ, IISC, IRKT, KIT3, LAMA, LHAS, ONSA, POL2, SELE, and ZWEN) to reported values in that frame with standard errors provided by the IGS ( ITRF97/itrf97-in/itrf97-in.html#solution). This method of determining a reference frame follows the method outlined in Hugentobler et al. (2001) for a regional network. We transformed these results into a Eurasia-fixed frame using the angular velocity estimated by Wang et al. (2001). Both the ITRF97 and the Eurasia-fixed velocities calculated in this study differ from those reported for IGS sites in Wang et al. (2001) by less than 2 mm/yr for sites used in both experiments. We attribute the difference to the inclusion of different IGS reference stations in the two solutions as well as the much larger number of measurements throughout Asia included by Wang et al. (2001). Because our network spans at most only half of the region where elastic strain accumulates near the Karakorum fault, we include the velocity and position for Shiquanhe given by Wang et al. (2001) in the ITRF97 solution (Fig. 2A) to constrain the slip rate on that fault. We increased the standard error of Shiquanhe by 2 mm/yr to allow for differences in the solutions that we and Wang et al. calculate the ITRF97. To present velocities relative to a fixed station at Leh, we subtracted the velocity at Leh from all of those in the Eurasia-fixed frame and minimized the rotation between the Eurasia-fixed frame and the Leh-fixed local frame. To present velocities relative to fixed-india, we transform according to the India-ITRF97 pole reported by Sella et al. (2002). REGIONAL RESULTS The lack of significant relative movement among control points within Ladakh (Fig. 3) suggests that elastic strain accumulation associated with slip on the Karakorum fault must be small. Two obvious explanations suggest themselves: slip is slow, or fault creep on the fault prevents elastic strain accumulation. We may test the latter by transforming the velocity of Shiquanhe (Wang et al., 2001) into the same reference frame as that for Leh and the other Ladakh control points. We calculate that Geological Society of America Bulletin, November/December

4 JADE et al. Shiquanhe moves at 4 ± 5 mm/yr at N170 E ± 9 relative to Leh (Fig. 3). The large uncertainty in this estimate is the result of combining velocities from slightly different solutions as discussed above. For a local strike of the Karakorum fault of N135 E, the component parallel to the fault of 3.4 ± 5 mm/yr in a right-lateral sense agrees with the geologic estimates of 3 4 mm/yr (Brown et al., 2002) and the upper bound of 8 mm/yr (Searle et al., 1998), but not with the rate of mm/yr (Avouac and Tapponnier, 1993; Liu, 1993; Tapponnier et al., 2001), or even with rates of ~10 mm/yr (Banerjee and Bürgmann, 2002; Searle, 1996). The lack of resolvable movement among points in Ladakh therefore implies slow deformation of the region surrounding the trace of the Karakorum fault, at least recently, and suggests that slow elastic strain accumulation on the fault reflects a low slip rate on it. We may test this in another way, by using the baseline length changes between Leh and Lhasa calculated here and between Shiquanhe and Lhasa derived from velocities in ITRF97 from Wang et al. (2001). This method removes the effect of reference frame considerations entirely. The change in the great circle baseline joining Shiquanhe (east of the Karakorum fault) and Lhasa (Wang et al., 2001) is ± 2 mm/ yr. We directly observe the change in the great circle baseline joining Leh (west of the fault) and Lhasa to be ± 1 mm/yr (Fig. 4). For a strike of N135 E of the Karakorum fault, slip rates of 30, 10, and 3 mm/yr would contribute differences in the rate of increase in baseline length of 15.9, 5.3, and 1.6 mm/yr respectively. The measured difference in baseline length changes of 2.5 ± 3 mm/yr favors a low rate of 4 5 mm/yr. The velocity of Delhi relative to Leh is 15.5 ± 1.7 mm/yr at N42 E ± 3. In an India-fixed reference frame generated by subtracting the angular velocity of India relative to ITRF97 (Sella et al., 2002) from the velocity field, convergence at both extremes of the Himalaya is normal to the arc (Fig. 2B). This is consistent with convergence vectors in the central Himalaya (e.g., Wang et al., 2001), which show a smooth variation in azimuth in order to remain normal to the strike of the curving range (Fig. 2B). This pattern of velocity is maintained in the far east and far west, with an azimuth of convergence in Ladakh of 224 ± 17 and in southeastern Tibet of 149 ± 37. The motion of Delhi relative to Leh of 18.8 ± 3 mm/yr gives a lower bound on the convergence rate across the Ladakh Himalaya, with a small additional amount possible if Leh lies within the region of elastic strain accumulation associated with the underthrusting of India beneath the Himalaya. TABLE 2. STATION COORDINATES AND VELOCITIES IN ITRF97 Station name Longitude ( E) Latitude ( N) E N σe σn IISC DELHI IRKT SELE LHAS LEH POL TIRTH PNMK STAKSHA LKNG MGLB TNGS CHUS MUTH KIT SHIQ GONGBU BAYI SOSHIEN Note: ITRF97 is the international terrestrial reference frame defi ned for 1997 by the International Earth Rotation Service. This rate is indistinguishable from either 18 ± 2 mm/yr reported for eastern Nepal or the Kumaon Himalaya west of Nepal (Larson et al., 1999; Paul et al., 2001) or 15 ± 2 mm/yr between Delhi and the northern High Himalaya (Banerjee and Bürgmann, 2002). As the latter pair of sites does not span the entire Himalaya, we suspect that its rate underestimates the convergence rate across that segment of the Himalaya. The convergence rate across the far eastern Himalaya, of 26.7 ± 6.3 mm/yr, is also consistent with other observations. The large variance of convergence estimates from the southeastern plateau in this frame is certainly related to the large uncertainty in the position of the ITRF97- India pole reported in Sella et al. (2002). Only five GPS sites constrain the motion of India: three in the Indian Ocean, two in south India, and none on the northeast. Because Leh is our westernmost site, we may use the Leh-Lhasa baseline to place a lower bound on the rate of east-west divergence between the western and southeastern ends of the Tibetan Plateau. The GPS solution in the Eurasia-fixed frame obtained from campaigns in Leh spanning 4 yr gives a relative velocity between the two of 19.7 ± 1.7 mm/yr at N104 E (1σ). In order to remove the effect of any reference frame rotation on the velocity estimate, we directly determine the change in the length of the great circle baseline between the two stations between 1998 and (Although both Leh and Lhasa have observations from the 1997 campaign, they share only a single epoch, thus preventing appropriate error analysis of their baseline length in 1997 [Table 1]. The velocity estimates from the GPS solution relative to an external reference frame do not require common epochs between stations and therefore do include observations in 1997 from both sites.) A least-squares estimate of length change gives a rate of 17.8 ± 1 mm/yr (1σ) (Fig. 4), consistent with their relative velocities in both ITRF97 and the Eurasia-fixed frame. If we extend the baseline east of Lhasa to Bayi, the rate of length change is not statistically different at 18 ± 3 mm/yr, but we can confirm divergence between the ends of the Himalaya across the southern extent of the plateau. DISCUSSION A bound on the slip rate on the Karakorum fault provides a test of whether Tibet is best treated as a nearly rigid plate or continuously deforming region, because one principal argument for rigid-plate behavior depends on the rigid object being bounded by rapidly slipping strike-slip faults (e.g., Tapponnier et al., 2001). The rate obtained here of 3.4 ± 5 mm/yr concurs with 3 4 mm/yr for Late Holocene time (Brown et al., 2002) from dating of offset features along the fault trace. Thus, unlike the Altyn Tagh fault, for which published rates based on GPS and Quaternary faulting disagree by a factor of two or three, no such discrepancy bars interpretation of the Karakorum fault rate. Thus, the demotion in rate of the Karakorum fault denies the hypothesis of a rigid plate beneath Tibet, one of its arguments Geological Society of America Bulletin, November/December 2004

5 GPS IN LADAKH Figure 2. (A) Station velocities and uncertainties (95%) for sites in the ITRF97 reference frame. Vectors are labeled with station names corresponding to Tables 1 and 2. Some station names in the Ladakh array are omitted for clarity. Velocities in this figure (and Table 2) may be transformed to a Eurasia-fixed frame defined by Wang et al. (2001) by subtracting rotation of ± /m.y. about a pole 56.9 ± 0.3 N, ± 0.3 E. (B) Station velocities and uncertainties (95%) for sites in an India-fixed frame generated using the angular velocity of India relative to ITRF97 from Sella et al. (2002). In this frame, velocities clearly change direction along the strike of the Himalayan Range, so as to remain everywhere normal or subnormal to the local topographic gradient. Please see the text for further discussion. Velocities calculated in this work are shown in magenta; those in blue are from Wang et al. (2001) with the same transformation. Error ellipses in this figure are those reported for the ITRF97 solution, rather than the India-fixed solution, which is affected by uncertainty in the location of the India-ITRF97 pole. Shaded topography is the SRTM three-second DEM. Armijo et al. (1989) suggested that the Karakorum fault is the westernmost of a series of strike-slip faults that link four or five large graben systems, which in turn separate blocks km in width and dominate the tectonics of the southern half of Tibet (Armijo et al., 1986). If roughly mm/yr of east-west extension are distributed across this region, as the GPS results suggest, then slip at ~3 5 mm/ yr should occur across each of the grabens if all east-west strain is accommodated at them. If the strike-slip faults link the grabens, we should expect slip rates across the grabens and on the strike-slip faults to be similar. Thus, the Karakorum fault slips at approximately the same rate as several strike-slip faults within Tibet, not at a measurably higher rate. For this reason too, characterizing Tibet as a rigid plate bounded by localized zones of concentrated deformation appears to be a poor approximation. One might ask whether a description of Tibetan tectonics in terms of relative movements among four or five blocks in southern Tibet, plus surely a comparable number in its northern part, and yet more in its eastern part, offers deeper insight into the processes at work than a description in terms of continuous deformation. Such a question stretches beyond the domain of the data presented here. Suffice it to say that demoting the significance of the Karakorum fault to that of many other faults within Tibet pushes the scale of blocks that are best treated as platelike down to those with dimensions comparable to, if perhaps still slightly greater than, typical lithospheric thickness. The radially oriented overthrusting of the Himalaya onto the Indian subcontinent at a constant rate along the Himalayan belt also bears on the question of plate-like versus fluidlike behavior. Both the geological structure and the active tectonics of the Himalaya seem to vary little along the 2500 km length of the Himalayan arc, except for the smooth variation in trend associated with the circular shape of the arc (e.g., Bendick and Bilham, 2001). As noted above, the divergence of slip vectors of earthquakes associated with thrust faulting of the Himalaya onto the (rigid) Indian subcontinent, in turn, requires that the region north of the belt, southern Tibet, extend roughly parallel to the trend of the arc such that western and eastern ends diverge at a rate of ~20 mm/yr (e.g., Bendick and Bilham, 2001; Molnar and Lyon-Caen, 1989). The GPS results presented here corroborate that rate of divergence and suggest that the convergence rate does not vary along the arc, but does vary smoothly in azimuth. In the simplest mechanism for understanding the radially oriented overthrusting and the extension of southern Tibet, both owe their origin to the buoyancy of thickened crust (e.g., Molnar and Tapponnier, 1978). Tibetan crust, if not entire lithosphere, behaves as a viscous fluid spreading out onto the Indian subcontinent, like honey on a dish. A viscous substance in a gravity field should flow from where it is thick toward where it is thin, perpendicular to gradients in thickness (see Fig. 2B). In the case of Tibet and the Himalaya, because of the arcuate shape of all aspects of the topography, flow onto the Indian subcontinent should be perpendicular to the Himalaya and at a constant rate (Bendick Geological Society of America Bulletin, November/December

6 JADE et al. and Bilham, 2001). Extension of southern Tibet results from this kinematic flow regime. Thus, this simple explanation for the large-scale aspects of the deformation field of southern Tibet and the Himalaya in terms of continuous deformation, not plates, passes the test imposed by the GPS results. The image of a viscous Tibet spreading over India is rendered obscure because of the tendency to describe velocity fields in a reference frame attached to Eurasia. In such a frame both India and southern Tibet move, and seeing their simple relative movement requires a mental transformation. For instance, McCaffrey and Nabelek (1998) suggested that India slides obliquely beneath southern Tibet and that the varying obliquity of the basal traction applied to the base of Tibet by India along the arc leads to extension of southern Tibet. McCaffrey (1991, 1992) had shown that such extension occurs within the overriding plate at subduction zones where convergence not only is oblique, but the obliquity varies along the island arc. When viewed in a reference frame attached to India, the radially oriented thrust faulting at the Himalaya shows no variation in obliquity along the Himalayan arc. Moreover, it suggests that the basal traction that India applies to the overriding Himalaya is also radially oriented. Thus, if such radially oriented tractions were applied by India to southern Tibet, in the absence of other processes, southern Tibet should undergo convergence parallel to the Himalaya, not divergence. This, of course, is not the case, as extension within Tibet attests. What allows the appearance of a paradox stems from the need to define a meaningful reference frame. Whereas for the cases that McCaffrey (1991, 1992) considered, the overriding plate at the subduction zone provided such a reference frame, for Tibet and the Himalaya no obvious frame exists. Eurasia serves poorly because too much deformation occurs between it and southern Tibet. Central Tibet occupies a position analogous to the overriding plate at subduction zones and could define a sensible reference frame if it behaved as a rigid object, but it does not. We suspect that by treating India as moving northward with respect to Eurasia, and then implicitly transforming those orientations to a frame assumed fixed to central Tibet, one is led to imagine that the traction it applies to the base of Himalaya and southern Tibet includes a component parallel to the orientation of the extension in southern Tibet. If one instead views velocities, strain rates, and tractions in a frame fixed to India, however, no process or observation implies the existence of a component of traction parallel to the Himalaya. These arguments do not disprove McCaffrey and Nabelek s Figure 3. Station velocities and uncertainties relative to Leh (black star). Fixing Leh and minimizing rotation define the frame. Velocities within Ladakh are not statistically different from zero. The direction of convergence between Delhi and Leh is arc-normal. Labels for sites with velocities very near zero, LKNG and PNMK, are omitted for clarity. The Karakorum fault trace is shown in gray, and dashed where the surface position of the fault is not clear. Figure 4. The direct estimation of relative velocities between stations from changes in baseline length is independent of reference frame rotations. The baseline between Leh and Lhasa, an estimate of approximately east-west stretching of the Tibetan Plateau, increases in length by nearly 18 mm/yr along a great circle path. Only three years in this experiment have epochs with measurements at both sites. The four-year estimate of relative velocity between these stations within the ITRF97 frame is slightly higher, 19.7 ± 1.7 mm/yr. Within the Ladakh network, baseline length changes are very small. For example, the baseline between Leh and Panamik increases in length by 1.7 mm/yr. 1σ includes 0.0 mm/yr. The plot also provides an indication of the yearly repeatability of baseline length. The scatter within a year is 3 4 mm Geological Society of America Bulletin, November/December 2004

7 GPS IN LADAKH contention, but they emphasize that the simple view of Tibet behaving as a continuum in a gravity field provides a simple explanation not only of the orthogonal orientation of thrust faulting at the Himalaya but also for the apparently constant rate of underthrusting there. In summary, both corroboration of a low rate of slip on the Karakorum fault and the inference of an essentially constant convergence rate between southern Tibet and the Indian subcontinent along the Himalayan belt support the idea that large-scale tectonics in Tibet is best described as deformation of a continuous medium. We cannot assert confidently that an assemblage of tens to many tens of blocks will not provide, in some sense, a better description, but characterizing Tibet as an essentially rigid plate that translates and rotates with respect to its neighbors finds little support in the GPS data from Asia. ACKNOWLEDGMENTS This work was largely supported by CSIR, India, with considerable logistical assistance from the CSIR Centre for Mathematical Modelling at Bangalore, and the Indian Institute of Astrophysics, which provided crucial ground support in Ladakh through its Astronomy project. Bendick and Molnar acknowledge support from National Science Foundation grants EAR , EAR , and EAR We appreciate comments on various drafts of this paper by J.-P. Avouac, J.-L. Mugnier, R. McCaffrey, and Zh.-K. Shen. This is Cambridge Earth Sciences paper ES REFERENCES CITED Armijo, R., Tapponnier, P., Mercier, J.L., and Han, T., 1986, Quaternary extension in southern Tibet: Field observations and tectonic implications: Journal of Geophysical Research, v. 91, p. 13,803 13,872. Armijo, R., Tapponnier, P., and Tonglin, H., 1989, Late Cenozoic right-lateral strike-slip faulting in southern Tibet: Journal of Geophysical Research, v. 94, p Avouac, J.-P., and Tapponnier, P., 1993, Kinematic model of active deformation in central Asia: Geophysical Research Letters, v. 20, p Baker, D.M., Lillie, R.J., Yeats, R.S., Johnson, G.D., Yousuf, M., and Zamin, A.S.H., 1988, Development of the Himalayan frontal thrust zone: Salt Range, Pakistan: Geology, v. 16, p. 3 7, doi: / (1988)016<0003:dothft>2.3.co;2. Banerjee, P., and Bürgmann, R., 2002, Convergence across the northwest Himalaya from GPS measurements: Geophysical Research Letters, v. 29, doi: / 2002GL Bendick, R., and Bilham, R., 2001, How perfect is the Himalayan arc?: Geology, v. 29, p , doi: / (2001)029<0791:HPITHA>2.0.CO;2. Bendick, R., Bilham, R., Freymueller, J., Larson, K., and Yin, G., 2000, Geodetic evidence for a low slip rate in the Altyn Tagh fault system: Nature, v. 404, p , doi: / Beutler, G., Bauersima, I., Gurtner, W., Rothacher, M., Schildknecht, T., Mader, G., and Abell, M., 1987, Evaluation of the 1984 Alaska global positioning system campaign with the Bernese GPS software: Journal of Geophysical Research, v. 92, p Bilham, R., Larson, K., Freymueller, J., and Project Idylhim, 1997, GPS measurements of present-day convergence across the Nepal Himalaya: Nature, v. 386, p Brown, E.T., Bendick, R., Bourlès, D.L., Gaur, V., Molnar, P., Raisbeck, G.M., and Yiou, F., 2002, Slip rates of the Karakorum fault, Ladakh, India, determined using cosmic ray exposure dating of debris flows and moraines: Journal of Geophysical Research, v. 107, doi: /2000JB England, P.C., and Houseman, G.A., 1985, The influence of lithospheric strength heterogeneities on the tectonics of Tibet and surrounding regions: Nature, v. 315, p England, P., and Jackson, J.A., 1989, Active deformation of continents: Annual Review of Earth Science, v. 17, p , doi: /annurev.ea England, P., and Molnar, P., 1997, Active deformation of Asia; from kinematics to dynamics: Science, v. 278, p , doi: /science Hugentobler, U., Schaer, S., and Fridez, P., 2001, Bernese GPS Software v. 4.2: Berne, University of Berne Press, p Jouanne, F., Mugnier, J., Pandey, M., Gamond, J.F., LeFort, P., Serrurier, L., Vigny, C., and Avouac, J.P., 1999, Oblique convergence in the Himalayas of western Nepal deduced from preliminary results of GPS measurements: Geophysical Research Letters, v. 26, p , doi: /1999gl Larson, K., Bürgmann, R., Bilham, R., and Freymueller, J., 1999, Kinematics of the India-Asia collision zone from GPS measurements: Journal of Geophysical Research, v. 104, p , doi: /1998jb Lavé, J., and Avouac, J.-Ph., 2000, Active folding of fluvial terraces across the Siwalik hills, Himalayas of central Nepal: Journal of Geophysical Research, v. 105, p , doi: /1999jb Liu, Q., 1993, Paléoclimat et contraintes chronologiques sur les mouvements récent dans l Ouest du Tibet: Failles du Karakorum et de Longmu Co-Gozha Co, lacs en pull-apart de Longmu Co et de Sumxi Co [Ph.D. thesis]: Paris, Université Paris, v. VII, 358 p. Liu, Q., Avouac, J.P., Tapponnier, P., and Zhang, Q., 1992, Holocene movement along the southern part of the Karakorum fault: Abstracts, International Symposium on the Karakorum and Kunlun Mountains, June 5 9, 1992, Kashi [sic], China, p. 91. McCaffrey, R., 1991, Slip vectors and stretching of the Sumatran forearc: Geology, v. 19, p , doi: / (1991)019<0881:SVASOT>2.3.CO;2. McCaffrey, R., 1992, Oblique plate convergence, slip vectors, and forearc deformation: Journal of Geophysical Research, v. 97, p McCaffrey, R., and Nabelek, J., 1998, Role of oblique convergence in the active deformation of the Himalayas and southern Tibet Plateau: Geology, v. 26, p , doi: / (1998)026<0691: ROOCIT>2.3.CO;2. Mériaux, A.-S., Tapponnier, P., Ryerson, F., van der Woerd, J., Lasserre, C., Xu, X., Finkel, R., and Caffee, M., 1999, Large-scale strain patterns, great earthquake breaks, and late Pleistocene slip-rate along the Altyn Tagh Fault (China): European Union of Geosciences conference abstracts, EUG 10, Journal of Conference Abstracts, v. 4, p Molnar, P., 1988, Continental tectonics in the aftermath of plate tectonics: Nature, v. 335, p , doi: /335131A0. Molnar, P., and Chen, W.-P., 1982, Seismicity and mountain building, in Hsü, K., ed., Mountain building processes: London, Academic Press, p Molnar, P., and Chen, W.-P., 1983, Focal depths and fault plane solutions of earthquakes under the Tibetan Plateau: Journal of Geophysical Research, v. 88, p Molnar, P., and Lyon-Caen, H., 1989, Fault plane solutions of earthquakes and active tectonics of the Tibetan Plateau and its margins: Geophysical Journal International, v. 99, p Molnar, P., and Tapponnier, P., 1978, Active tectonics of Tibet: Journal of Geophysical Research, v. 83, p Neil, E., and Houseman, G., 1997, Geodynamics of the Tarim Basin and the Tien Shan in central Asia: Tectonics, v. 16, p , doi: /97tc Ni, J., and York, J.E., 1978, Late Cenozoic extensional tectonics of the Tibetan Plateau: Journal of Geophysical Research, v. 83, p Paul, J., Burgmann, R., Gaur, V.K., Bilham, R., Larson, K.M., Ananda, M.B., Jade, S., Mukal, M., Anupama, T.S., Satyal, G., and Kumar, D., 2001, The motion and active deformation of India: Geophysical Research Letters, v. 28, p , doi: / 2000GL Peive, A.V., Burtman, V.S., Ruzhentsev, S.V., and Suvorov, A.I., 1964, Tectonics of the Pamir-Himalayan sector of Asia, in Report of the twenty-second session, India, International Geological Congress, Part XI, Himalayan and Alpine Orogeny: New Delhi, International Geological Congress, p Peltzer, G., and Tapponnier, P., 1988, Formation and evolution of strike-slip faults, rifts, and basins during the India-Asia collision; an experimental approach: Journal of Geophysical Research, v. 93, p Peltzer, G., Tapponnier, P., and Armijo, R., 1989, Magnitude of late Quaternary left-lateral displacements along the north edge of Tibet: Science, v. 246, p Powers, P.M., Lillie, R.J., and Yeats, R.S., 1998, Structure and shortening of the Kangra and Dehra Dun reentrants, Sub-Himalaya, India: Geological Society of America Bulletin, v. 110, p , doi: / (1998)110<1010:SASOTK>2.3.CO;2. Searle, M.P., 1996, Geological evidence against large-scale pre-holocene offsets along the Karakorum fault: Implications for the limited extrusion of the Tibetan Plateau: Tectonics, v. 15, p , doi: /95tc Searle, M.P., Weinberg, R.F., and Dunlap, W.J., 1998, Transpressional tectonics along the Karakorum fault zone, northern Ladakh: Constraints on Tibetan extrusion, in Holdsworth, R.E., et al., eds., Continental transpressional and transtensional tectonics: Geological Society [London] Special Publication 135, p Sella, G., Dixon, T., and Mao, A., 2002, REVEL: A model for Recent plate velocities from space geodesy: Journal of Geophysical Research, v. 107, doi: / 2000JB Shen, F., Royden, L., and Burchfiel, B.C., 2001, Large-scale crustal deformation of the Tibetan Plateau: Journal of Geophysical Research, v. 106, p , doi: /2000jb Tapponnier, P., Xu, Z., Roger, F., Meyer, B., Arnaud, N., Wittlinger, G., and Yang, J., 2001, Oblique stepwise rise and growth of the Tibet Plateau: Science, v. 294, p , doi: /science Wang, Q., Zhang, P., Freymueller, J., Bilham, R., Larson, K., Lai, X., You, X., Niu, Z., Wu, J., Li, Y., Liu, J., Yang, Z., and Chen, Q., 2001, Present-day crustal deformation in China constrained by Global Positioning System measurements: Science, v. 294, p , doi: / SCIENCE Washburn, Z., Arrowsmith, J.R., Forman, S.L., Cowgill, E., Wang, X., Zhang, Y., and Chen, Z., 2001, Late Holocene earthquake history of the central Altyn Tagh fault, China: Geology, v. 29, p MANUSCRIPT RECEIVED BY THE SOCIETY 19 MARCH 2003 REVISED MANUSCRIPT RECEIVED 28 NOVEMBER 2003 MANUSCRIPT ACCEPTED 6 JANUARY 2004 Printed in the USA Geological Society of America Bulletin, November/December