The Talas-Fergana Fault(TFF)with a total length of more than 1 000km is a large dextral strike-slip fault across the West Tianshan Mountains in the northwest direction. The fault plays an important role in accommodating deformation in Central Asia and has attracted much attention by geologists due to the huge controversy in its strike-slip rate and kinematic pattern. Previous studies indicated that its average dextral strike-slip rate is 8~20mm/a since Late Holocene based on offset ephemeral stream valleys and ¹⁴C dating method. Some researchers recently updated the strike-slip to 2.2~6.3mm/a by the application of multiple dating methods(¹⁰Be, ²⁶Al, ³⁶Cl, luminescence, and radiocarbon)and satellite images with higher precision. But the strike-slip rates derived from modern GPS velocity field are only~2mm/a or even as low as 0.8mm/a. Thus, there is a substantial divergence between geological results and geodetic results in the strike-slip rate of the TFF. Some scholars believe that the huge difference between the geological rate and the rate obtained by geodetic measurements is caused by fault locking. In this study, the updated GPS data was used to establish velocity field of the West Tianshan Mountains relative to the stable Eurasian framework and the velocity field without self-rotation. The velocity field shows that the Tianshan Mountains are under intense crustal shortening and deformation. Moreover, for the TFF, as an important boundary fault in the western Tianshan Mountains, whether the far velocity field or the near velocity field, the differential movement of the crust is not obvious. And far-field velocity vectors away from the TFF show that there is minor difference of crustal movement along the fault. The TFF does not have the typical characteristic of locked fault that there is a big difference in velocities of far-field vectors, but a small difference in that of near/mid-field vectors. Thus, the activity of the fault is weak actually.
To further illustrate the overall low slip rate of the TFF, we compare the maximum shear strain rate and its distribution characteristics along the Altyn Fault and the Haiyuan Fault with large slip rates with the results of the TFF. The maximum shear strain rates along the Altyn Fault and the Haiyuan Fault are concentrated along the fault, and are as high as~60nano strain/a and~40nano strain/a, which are much larger than the overall maximum shear along the TFF. This shows that the sliding rate of the TFF is much lower than the strike-slip rate of the Altyn Fault of 9~15mm/a, and even slightly lower than the sliding rate of the Haiyuan Fault of 4~8mm/a. Therefore, we are more certain that the current activity rate of the TFF is far less than 8~20mm/a estimated by some geological methods.
The half-space elastic dislocation model is used to more rigorously re-constrain the current strike-slip rate of the TFF. The results show that the fault is divided into three segments. The TFF dextral strike-slip rate increases from the northwest section to the middle section and decreases from the middle section to the southeast section. And the strike-slip rates of the northwestern, middle and southeastern segments are(2.1±0.7)mm/a, (3.3±0.4)mm/a and(2.4±0.7)mm/a, respectively. The TFF is dominated by strike-slip motion, but there is also a weak dip-slip motion in the middle section of the TFF, with a magnitude of about 1mm/a.
The above results confirm the current low strike-slip rate of the TFF obtained by GPS which is much less than the strike-slip rate of 8~20mm/a estimated by geological methods. And through the GPS results, it is certain that the TFF presently has a low fault activity rather than a locked fault. To reconcile the high geological strike-slip rates and the geodetic results, a new deformation pattern of the West Tianshan Mountains may be needed. And more detailed GPS observations are required to explore whether the TFF has penetrated into the southern foreland basin of the West Tianshan Mountains.

Located among the South China block, Tibetan plateau, Alxa block and Yinshan orogenic belt, the Ordos block is famous for its significant kinematic features with stable tectonics of its interior but frequent large earthquakes surrounding it. After the destruction of the North China Craton, the integrity, rotation movement and kinematic relations with its margins are hotly debated. With the accumulation of active tectonics data, and paleomagnetic and GPS observations, some kinematic models have emerged to describe rotation movement of the Ordos block since the 1970's, including clockwise rotation, anticlockwise rotation, clockwise-anticlockwise-alternate rotation, and sub-block rotation, etc. All of these models are not enough to reflect the whole movement of the Ordos block, because the data used are limited to local areas.
    In this study, based on denser geophysical observations, such as GPS and SKS splitting data, we analyzed present-day crustal and mantle deformation characteristics in the Ordos block and its surrounding areas. GPS baselines, strain rates, and strain time series are calculated to describe the intrablock deformation and kinematic relationship between Ordos block and its margins. SKS observations are used to study the kinematic relationship between crust and deeper mantle and their dynamic mechanisms, combined with the absolute plate motion(APM)and kinematic vorticity parameters. Our results show that the Ordos block behaves rigidly and rotates anticlockwise relative to the stable Eurasia plate(Euler pole: (50.942±1.935)°N, (115.692±0.303)°E, (0.195±0.006)°/Ma). The block interior sees a weak deformation of~5 nano/a and a velocity difference of smaller than 2mm/a, which can be totally covered by the uncertainties of GPS data. Therefore, the Ordos block is moving as a whole without clear differential movement under the effective range of resolution of the available GPS datasets. Its western and eastern margins are characterized by two strong right-lateral shearing belts, where 0.2°~0.4°/Ma of rotation is measured by the GPS baseline pairs. However, its northern and southern margins are weakly deformed with left-lateral shearing, where only 0.1°/Ma of rotation is measured. Kinematics in the northeastern Tibetan plateau and western margin of the Ordos block can be described with vertical coherence model with strong coupling between the crust and deeper mantle induced by the strong extrusion of the Tibetan plateau. The consistency between SKS fast wave direction and absolute plate motion suggests the existence of mantle flow along the Qinling orogenic belt, which may extend to the interior of the Ordos block. SKS fast wave directions are consistent with the direction of the asthenosphere flow in Shanxi Rift and Taihang Mountains, indicating that the crustal deformation of these areas is controlled by subduction of the Pacific plate to North China. The week anisotropy on SKS in the interior of Ordos block is from fossil anisotropy in the craton interior. After comparing with the absolute plate motion direction and deformation model, we deem that anisotropy in the interior of Ordos block comes from anisotropy of fossils frozen in the lithosphere. In conclusion, the Ordos block is rotating anticlockwise relative to its margins, which may comes from positive movement of its margins driven by lithospheric extrusion or mantle flow beneath, and its self-rotation is slight. This study can provide useful information for discussion of kinematics between the Ordos block and its surrounding tectonic units.

Four-component borehole strainmeter (FCBS) is one kind of high-precision borehole strain observation instruments invented in China. As a kind of near-surface deformation observation instrument, FCBS is also easily disturbed by the external environment factors. As a common factor, pumping has significant influence on FCBS observation. Existing studies mostly identify the pumping interference from the perspective of observation curve morphology, relatively few studies focus on its interference mechanism. In order to truly capture earthquake precursor information, it is necessary to study the interference mechanism. In recent years, RZB-3 type FCBS at Tai'an seismic station has been seriously affected by pumping, so it is necessary and also feasible to study the interference mechanism of pumping. Since the influence of pumping interference on borehole strainmeter is common, this work would be very practical and be used for reference by other borehole strain observation stations.
We find that the original observation curves and observed surface strain, shear strain from RZB-3 type FCBS at Tai'an seismic station have the characteristics of synchronous change with the borehole water level, in which the linear correlation coefficient between the two observed shear strain curves and borehole water level reached 0.70 and 0.82 respectively. We further find that the principal strain direction of borehole and borehole water level after normalization meet the nonlinear function as y=1.217arctan(x)^0.224-0.284. The above phenomenon indicates that the observation of RZB-3 type FCBS at Tai'an seismic station is significantly affected by the borehole water level, and the influence is more obvious and the gradient is larger at the stage of low water level. Pumping interference often appears in low water level stage and changes the rock pore pressure state. Statistics show that pumping interference affects the borehole strain state.
To investigate the interference mechanism of pumping to RZB-3 type FCBS at Tai'an seismic station, we take a known pumping as an example, in which we study the principal strain state of the borehole in three periods of normal pumping, interruption of pumping and resuming pumping respectively. During each period, we solve 3 parameters of the principal plane strain state, i.e. the maximum principal strain rate, the minimum principal strain rate and the maximum principal strain direction from four observation equations of FCBS by nonlinear iterative least squares algorithm. On the other hand, concentrated load model (CLM) is used to simulate the mechanical mechanism of pumping. Firstly, the depth of FCBS relative to pumping source and the concentrated load at pumping source are inversed, then, the strain state surrounding the pumping well, including the state at RZB-3 borehole, is simulated by forward modeling. By comparing these results, we find that:
(1)The concentrated load at pumping source inversed by CLM during periods of normal pumping and resuming pumping are both located at or near the bottom of the pumping well, which is consistent with the actual situation, indicating that mechanism and degree of the influence of pumping on borehole strain are well simulated by CLM.
(2)The observed strain state is consistent with the simulation result of pumping interference by forward modeling, indicating that the principal strain state of borehole calculated based on observation of FCBS reflects the true strain state of borehole under different pumping states.(3)The inversed concentrated load at pumping source during pumping periods is significant greater than the load of the pumped water, indicating that the pumping process has more significant influence on the pore pressure of rocks than the load of the pumped water.
Even though CLM is an approximate simulation since it's based on some elastic assumptions, the interference mechanism of pumping on RZB-3 type FCBS at Tai'an seismic station is well explained, which is maybe very helpful for studying the influence of pumping interference on other deformation instruments, locating the unknown pumping source and studying the characteristics of pore pressure of rocks.

Following the 11 March 2011 Japan M_W9.0 earthquake, frequent moderate and small events occurred on the Yishu fault zone and its either side. Using continuous GPS data and a sliding block model, this work studies the relationship between the energy release of these shocks and the block relative motion of either side of the Yishu fault zone. The results show that(1)the equivalent magnitude M from released energy and the two blocks' relative motion are well correlated when earthquakes are selected in a retrieval circle(whose center is the midpoint of the Yishu fault zone)with a radius of 250~500km and using a sliding time window of 3~10 months. The best correlation coefficient between M and the two blocks' relative motion is 0.74 and the T test shows a significant linear correlation between them.(2)Spatial distribution of the correlation coefficients shows that the relative motion of the blocks on both sides affects the energy release in the area from the north part of Yishu fault zone to the Jiaodong Peninsula area and southwest Shandong-Henan border area obviously.(3)Since June 2014, the relative motion of the two blocks on both sides of the Yishu fault zone presents a wave of change, which may be an expression of the accumulation of seismic strain energy in the Yishu fault zone and its two sides. The linear relationship between the equivalent magnitude M from released energy and two blocks' relative motion V can be fitted by linear equation M=0.51*V+3.9, showing that strain energy accumulation could be released by the moderate and small earthquakes in a timely manner, which may favorable to delay the seismic risk in the study area. It also shows, on the other hand, that earthquake energy was not released so completely in the study area since the end of 2015 to 2016, which is likely associated with the Changdao earthquake swarm in 2017.

We relocated a seismic swarm, which started in a mass from 31 October, 2013 in Qianguo County of Jilin Province, by using double difference location method, based on the phase data of regional digital seismic network and the crustal velocity model of Sunliao Basin. The characteristics of seismogenic fault have been investigated based on the spatial distribution image of the seismic swarm and the geophysical data near the epicenter area. The relocated epicenters of the swarm earthquakes have a precision of 0.9km in E-W, 0.7km in N-S and 1.2km in U-D direction, and show an apparent concentrated seismic belt trending N-W, with a length and width of 12km and 6km, respectively. The source depths of all events are shallow, with 80%in a range of 6~8km, and the events are apparently crowded together on the depth cross section. According to the relocated spatial distribution characteristics of the seismic swarm, the features that the medium size events happened successively, and the focal mechanism of the large size events in the swarm, we infer that the seismogenic tectonics of Qianguo seismic swarm is the thrust nappe structure inside the Keshan-Da'an fault zone. The fault plane inclines to the East direction, and is steep when close to the ground surface, which shows the typical characteristics of a listric thrust fault. The longitudinal length of the rupture plane is greater than the transverse length. According to the features of seismogenic tectonics, we infer that the three M_S ≥ 5.0 earthquakes occurred at the lower layer of the thrust rupture surface of the fault, while the aftershocks were triggered by the three events and occurred mainly at the upper layer of the rupture surface.

The M_S6.5 Ludian earthquake occurred on 3 August 2014 in Yunnan, China. The epicenter of this earthquake is located in the Dalingshan sub-block, a boundary region among the Bayan Har block, the Sichuan-Yunnan block and the South China block, which is dominated by the left-lateral and thrust-slip faults. The studies on the characteristics of the crustal deformation, the mechanism of strong earthquakes and stress changes after the M_S6.5 Ludian earthquake in the Daliangshan sub-block will help us understand the tectonic implication of the earthquake and facilitate further in-depth studies in the region.
This article introduces the slip behavior around the Daliangshan sub-block and strong earthquake distribution on the faults. Using the GPS data in the southeastern Tibetan Plateau, this study analyzes the motions of sub-blocks bordering the Daliangshan sub-block and the slip behavior of the boundary faults in the block model, and estimates the crustal motion in the Daliangshan sub-block after removing the whole block motion of the sub-block. It can be shown clearly that the Daliangshan sub-block has absorbed the compression via the Xianshuihe Fault, the Anninghe Fault and the Longmenshan Fault. Due to the compression from the boundary faults, the crustal motion in the sub-block is mainly characterized by the north-northwest trending left-lateral strike-slip faults. The boundary faults of the Daliangshan sub-block are mainly characterized by strong earthquakes with magnitude larger than M7 in the history. But in the interior of the Daliangshan sub-block, there are mainly the moderate earthquakes. The M_S6.5 Ludian earthquake ruptured the Baogunao-Xiaohe Fault, a left-lateral strike-slip fault associated with the main thrust Ludian-Zhaotong Fault and the Lianfeng Fault, where the whole thrust slip rate of the two faults is about 2.4mm/a.
Around the epicenter of the Ludian earthquake, there are strong earthquakes larger than M7 recorded in the nearby region, which might have impact on the occurrence of the Ludian earthquake. Also, the Ludian earthquake may further affect the occurrence of subsequent earthquakes. In this paper, we have calculated the static Coulomb failure stress changes (ΔCFS) on the fault plane of the Ludian earthquake induced by the 3 nearby big earthquakes, including the M7 1/2 earthquake in 1850 on the Zemuhe Fault and the M7 3/4 earthquake in 1733 on the Xiaojiang Fault. The M_S6.5 Ludian earthquake ruptured the north-northwest trending Baogunao-Xiaohe Fault, which is associated with the main thrust fault system of the Zhaotong-Ludian Fault and Lianfeng Fault. In our results, the Ludian earthquake was promoted by the M7 3/4 in 1733 on the Xiaojiang Fault and the M7 1/2 in 1850 on the Zemuhe Fault earthquakes. From the Coulomb failure stress change calculation, the ΔCFS value is about 0.03MPa, which may advance the occurrence of the M_S6.5 Ludian earthquake obviously. The calculations also showed opposite results of ΔCFS from the Ludian earthquake on the Yongshan M_S5.0 earthquake of 17 August and on the Yuexi M_S5.0 earthquake of 1 October in 2014. The former one is not much related to Ludian earthquake, but to the normal seismicity in the reservoir area as to the minus value of ΔCFS, while the Yuexi M_S5.0 earthquake was promoted by the M_S6.5 Ludian earthquake. Moreover, the M_S6.5 Ludian earthquake has advanced, to a certain extent, the enhancement of Coulomb failure stress on the northern segment of the Daliangshan Fault, Ebian Fault, eastern segment of the Zhaotong-Ludian Fault and the southern segment of the Zemuhe Fault, and has enhanced the earthquake energy accumulation of these faults.

Airborne LiDAR (Light Detection And Ranging) provides a more advanced technique and more accurate basic data to describe geomorphological features and the latest surface deformation associated with active tectonics. How to apply this new technique and dataset to mapping of active fault and seismic hazard assessment is an important trend in the field of active tectonics. Taking the Dushanzi anticline-reverse fault zone in Xinjiang as test area, we made an experimental study on geologic mapping of active tectonics based on the LiDAR data. Firstly, we collected raw data using the airborne LiDAR technique, and obtained a raw point-cloud with a point density of 6.6 points/m² and an average space of 0.39m between any two points. Secondly, using twelve ground control points(GCP)which is acquired by static GPS measurement with accuracy up to millimeter, we evaluated the vertical error of the ground point-cloud data with density of 6.4 points/m², and the result shows a vertical error of 0.12m, mean square value 0.078m. Finally, using the inverse distance weighting algorithm, we obtained the digital elevation model(DEM)of 0.5m-resolution. The resolution of the DEM is high enough to describe and analyze spatially the fine feature of tectonic landform of the Dushanzi anticline-reverse fault zone. In this paper, we identify the fine tectonic landforms using merely the DEM visualization tools based on different virtual perspectives, different shades or different treatment methods. The active tectonics and their distribution identified based on the high resolution DEM derived from LiDAR are not only consistent with previous results identified from air-interpretation and field investigation, but also finer and more precise than the latter. In addition, these methods of data acquisition, quality inspection and data processing introduced in this paper are also applied to other active fault researches in which LiDAR data have been acquired.

The Tohoku-Oki M_W9.0 earthquake of 11 March, 2011 has caused eastward movement and subsidence of the Japanese Islands as well as mass redistribution. The temporal-spatial features of mass redistribution were discussed by using the monthly GRACE time-variable gravity field, which would compensate the inefficiency for the undersea focal region where GPS, InSAR measurements are not available. The coseismic gravity changes were computed through least-square fitting and empirical orthogonal function(EOF) from the time series on 0.5°×0.5° grids, and through dislocation model as well. A dipole distribution of the coseismic changes appears in back-arc region and trench with maximum decrement and increment of～6μgal and～3μgal, respectively. The results suggest that EOF method avoids a priori knowledge, such as event time, as used in least squares fitting. Nevertheless, the gravity signal derived from GRACE satellites is an integral of many different geophysical processes, thus the reliability and exact physical sources are likely varying due to the event scale and observation time span, etc. In this study, most of seasonal changes are eliminated through PCs 2, 3, 4; and the coseismic gravity changes extracted from the first principle component of EOF, whose distribution is spatially coherent, are much closer to the result from dislocation model than the least square result, therefore can really reflect the changes resulting from the earthquake.

The April 20,2013,M_S 7.0 Lushan earthquake occurred along the southwestern part of the Longmen Shan Fault zone. Tectonics around the epicenter area is complicated and several NE-trending faults are developed. Focal mechanisms of the main shock and inversions from finite fault model suggest that the earthquake occurred on a northeast-trending,moderately dipping reverse fault,which is consistent with the strike and slip of the Longmen Shan Fault zone. NE-trending ground fissures and soil liquefaction along the fissures,heavy landslides along the Dachuan-Shuangshi and Xinkaidian Faults were observed during the field investigations. No surface ruptures were found in the field work. GPS data indicate that the fault on which this earthquake occurred is a fault east of or near the Lushan county and the earthquake also triggered slip on the fault west of the Lushan county. Field observations,GPS data,focal fault plane,focal depth,and distribution of the aftershocks suggest, that the seismogenic structure associated with the M_S 7.0 Lushan earthquake is the décollement beneath the folds of the eastern Longmen Shan. Slip along this decollement generated the earthquake,and also triggered the slip along the Dachuan-Shuangshi and Xinkaidian Faults.

This paper studies the impact of the M_W9.0 Tohoku-Oki earthquake,March 11,2011,Japan on the crustal deformation,active fault movement and earthquake risk in northeastern China by using continuous GPS data. The result shows that the direction and amount of principal strain is different between coseismic and interseismic strain field. The Japan earthquake mainly affects the level of strain accumulation in northeastern China. The coseismic strain accumulation is equivalent to about 12.7-years long-term background strain accumulation. In other regions,the impact of coseismic strain on the background strain rate isn't obvious. Based on the coseismic strain and background strain rate,we select 4 major active faults for the analysis by using the relative movement time sequence of endpoints of GPS baselines which cross the Tanlu Fault,Zhangjiakou-Penglai Fault and Taihangshan piedmont fault,respectively. We find that the earthquake produced obvious left-lateral slip and extensional slip on the north segment of Tanlu Fault,and there is a tendency of slight continuous left-lateral movement and extension after the earthquake. But on the middle segment of Tanlu Fault,there is no obvious evidence of coseismic movement,but a slight increase of the right-lateral slip rate after the earthquake. As for Zhangjiakou-Penglai Fault located in North China,the fault was characterized by obvious right-lateral movement and extrusion in about one year after the earthquake, then resumed quickly to the original left-lateral movement and extension. This means that the fault's accumulated stress was released temporarily by the earthquake,then the regional stress resumed to its original level. The movement of Taihangshan piedmont fault isn't affected by this earthquake,therefore we infer that the earthquake affected areas did not spread to the Shanxi Rift System. In one word,the crustal strain accumulation and stress level of northeastern region in China is released by Tohoku-Oki earthquake,so we believe that the earthquake risk is reduced to a certain extent.

Using the GPS velocity data of 27 stations around the East Kunlun Fault as restraints,we first inversed the slip rates of the eastern Kunlun Fault,the northern margin fault of Qaidam Basin,the Mani-Yushu Fault and the Maergaichaka Fault before the M_S 8.1 west Kunlun Mountains Pass earthquake with 3-D elastic half-space dislocation model.The deformation field calculated from these faults' slip movement can be considered as the background deformation field before the M_S 8.1 Kunlun earthquake.Based on the tectonically meaningful data of the background deformation field,we calculated the strain field and seismic moment accumulation rate field.The results show that there are two obvious regions of high moment accumulation rates,one is just in the Xidatan-Dongdatan segment of East Kunlun Fault where the M_S 8.1 earthquake happened in 2001.

On the basis of geological data,the three-dimensional geometric model and the segmentation model of the spatial distribution of the Zhuanglanghe faults and the northern marginal fault of Maxianshan are firstly established; Then,with the constraints of slip rates of some studied thoroughly faults,the inferred GPS velocity from the project“Crustal Movement Observation Network of China”is simulated by using“Three-dimensional Deep-fault Dislocation Model”; Finally,the current sliding rates of some other faults are calculated with the inversion method. The results show that there is a rate of 1.4～3.0mm/a of thrust slip on the northern margin fault of Maxianshan,with a left-lateral slip rate of about 3.0mm/a on the middle of the faults,and that there is a thrust slip rate of 0.6～1.2mm/a on the Zhuanglanghe faults with unobvious left-lateral slip component. These results are in accordance with the geological results. The GPS station velocities predicted by the model are consistent with the observed velocities as a whole. It is therefore assumed that these results are all in the reasonable extent predicted before. For the faults with relatively low slip rates and on which the conventional geological methods are difficult to apply,e.g. the Zhuanglanghe faults and the Maxianshan fault,the inversion based on the GPS observations is an effective ancillary means.

In order to investigate the present-day movement characteristics of Haiyuan active fault zone and Xiangshan-Tianjingshan active fault zone on the northeast margin of Tibetan Plateau in detail, we established a GPS profile across the fault zones. The profile, extending from Lanzhou, Gansu to Zhongwei, Ningxia, is composed of 12 stations and locally reinforced the existing regional GPS network of Crustal Movement Observation Network of China (CMONOC). These new stations, together with the existing GPS stations, constructed a spatially dense profile whose average interval of the stauions is ~22km. Considering that there were two continuous GPS stations of CMONOC (i.e. XNIN and YANG) around the region, we tried a “Flexible Observation Method” in GPS observations. The method allows non-synchronal observations for all the GPS observation teams and makes the observation schedule rather flexible. In data processing, we used the advanced strategy of “Precise Point Positioning” of GIPSY software. Our result shows that with the support of CMONOC, especially the continuously observed fiducial GPS network of CMONOC, we can use the “Flexible Observation Method” and “Precise Point Positioning” data processing strategy to effectively observe local GPS networks to monitor crustal deformation with satisfying accuracy.

The co-seismic displacement field of the 2001 west of Kunlun Mountain Pass earthquake is obtained through the analysis of GPS data measured before and after the earthquake and leveling data measured in 1979 and 2002. Adopting these data,constrained by detail surface rupture data measured after the earthquake,we inverted the co seismic slip distribution along the seismic fault. The result shows that the depth of rupture lower limit is 14.2～21km (with 70% confidence level),with 17km as the optimal value. The result also shows that left lateral strike slip of 2～3m exists in the area between the Sun Lake segment and the west end of the main rupture zone,although surface rupture is not observed there. This is consistent with the result of InSAR data analysis. The subsurface rupture of this earthquake is ended at the Sun Lake in the west,but it seems that left lateral slip of 1.5～ 2.0m still exists within the range of 30km to the east of the surface rupture zone. The vertical displacement inverted in this study shows that to the west of 93°E the southern side of the fault is elevated,while to the east of 93°E the northern side of the fault is elevated. The released seismic moment estimated by geodetic data and surface rupture surveying is 6.1×10²⁰ N·m,consistent with the result inverted by seismic wave records.

As the most prominent example of large scale continental deformation,Tibetan Plateau offers an ideal natural laboratory for quantifying such deformation and understanding the relevant dynamic processes. Global Positioning System (GPS) provides a powerful means to directly measure the kinematics of present day deformation. Our synthesis of GPS velocities from 553 stations in Tibetan Plateau and its margins quantitatively show that most of the relative India/Eurasia motion has been accommodated primarily by crustal shortening along the margins,strike slip and normal faulting in the plateau interior,and clockwise rotation around the eastern end of Himalayas. Taken 36～40mm/yr as total relative motion between India and Eurasia,the eastern Tibetan Plateau and its margins accommodates 85%～94% of the total motion,whereas western Tibet absorbs 70%～91% of total convergence and the rests are taken up by shortening across the Tianshan in the north. The NNE SSW shortening of the plateau interior is accommodated by conjugate strike slip faulting and orthogonal normal faulting,which do not require crustal thickening or thrust faulting. The eastward extrusion of Tibetan Plateau out of India's northward pass is carried out by roughly eastward flow of crustal material rather than by rigid block rotation. The flow of Tibetan crustal material rotates around the eastern Himalayan syntaxis,causing southeastward to southward and even southwestward velocities observed in southern and western Yunnan Province of China. To the east,the eastward flow of crustal material causes shortening across the eastern margin of the plateau and clockwise rotations of the region where resistance to such flow is weak. To the west,the westward motion of the western margin of the plateau is observed with only 4mm/yr slip rate. Components of velocity of Tibetan Plateau in both parallel and perpendicular directions to the relative motion between India and Eurasia can not be attributed to slips along a few faults either strike slip or thrust because of their distributed nature. Thus,the present day tectonics in the Tibetan Plateau is best described as deformation of a continuous medium,at least when averaged over distances of ～100km.

Based on the results of large scale geological mapping (1/50000), we established a reasonable rupture-segmentation model for the Jingtai-Tianzhu active fault. The model comprises two large pull-apart basins on both ends of the fault and two relatively large discontinuous structures along the fault.The basins are considered as "stable barriers" constraining the maximum surface rupture of the Jingtai-Tianzhu active fault, while the two large discontinuons structures are considered as "unstable barriers". Thus, the Jingtai-Tianzhu active fault can be divided into three rupture units or segments the Jinqianghe segment, Maomaoshan segment and Laohushan segment with lengths of 34 km, 56 km and 62 km, respectively.Existing evidences show that each segment has its own characteristic earthquakes, and all the segments may rupture together because of the interaction between each other. Using the "real time probabilty model" and considering the interaction between the neighboring rupture segments, we estimated the recurrence probabilities of large earthquakes along the fault. The results suggest that in the next 100 years, the Laohushan segment has little possibility for an earthquake larger than M S7.2 to occar, whereas the other two segments have some potentialities. In the future, although the pessibility of rupture to occur on Maomaoshan and Jinqianghe segments is random it seems that the rupture on one segment will irritate the rupture on the other segment. That is, the next large earthquake on the Jingtai-Tianzhu fault is most proloably the resault of a combined rupture of the Maomaoshan and Jinqianghe segments. The magnitude is estimated at about M S7.5, and the preferred probabilities in the next 10, 20, 50, 100 years are 14%, 27%, 56% and 81%, respectively.