The irregularly distributed surface fault-ruptures of the Wenchuan earthquake spanned over 200 km along the Longmen Shan(LMS) fault zone. Through field investigations, we found over 10 coseismic surface-ruptures, with maximum vertical displacements of approximately 6 m on the Yingxiu-Beichuan fault and 2 m on the Guanxian-Anxian fault; however, the entire fault rupture movement was still not clearly understood since high topographic areas were inaccessible. Thus, we used interferometric synthetic aperture radar (InSAR) satellite observations to capture whole coseismic surface-ruptures and crustal deformations across the LMS faults. We created a novel bi-fault-slip model to invert fault-slips using InSAR information which yielded that thrust fault-slips were dominant at YingXiu, Houshenggou and Bajiaomiao in the near-epicenter segment, while the dextral fault-slips were dominant at Pingtong and Nanba along the northeast segment. The combination of field investigations and simulations suggested that the two coseismic fault zones ruptured with an irregular surface distribution accompanied by crustal deformations.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The Mw 7.9 Wenchuan earthquake of May 12, 2008 occurred on the northeastern-striking Longmen Shan (LMS) faults beneath the steep eastern edge of the Tibetan plateau. The coseismic ruptures and crustal deformation occurred over a length of 290 km along the LMS thrust belt. Over 87,000 people were reported dead or missing and more than 370,000 injured. Little attention was paid to the LMS faults regarding potentially strong earthquakes in long-term earthquake predictions [e.g., Tang and Han, 1993; Tang et al., 1995]. Although no extraordinary earthquakes were recorded, the LMS faults were speculated to still be active through various findings such as the occurrence of a Holocene rupture [Burchfiel et al., 1995; Kirby et al., 2003; Burchfiel, 2004; Densmore et al., 2007].
 The complicated and large scale ruptures of the quake caused tragic damages, as well as raised scientific interests in seismology and tectonics, and necessitated the need for immediate detailed field investigations. After three sets of investigations along the heavily damaged-belts of the LMS faults, we confirmed over 10 coseismic surface-ruptures involving at least two long fault zones, 11–13 km apart [Hao et al., 2008a, 2008b, A preliminary investigation of the coseismic surface-ruptures for Wenchuan earthquake of 12 May 2008, Sichuan, China, paper presented at the 14th World Conference on Earthquake Engineering, Beijing, 2008]. Each rupture spot was traced over several kilometers [Xu et al., 2008; Li et al., 2008], but the relationship between these individual fault-slip locations and the entire rupture system along the LMS fault was still unclear. The detailed surface information along all of the faults and regions was difficult and time consuming to collect through field investigations due to the inaccessibility of this high topographic area. Finite fault inversions from far-field waveform information showed that the ruptures started from about 12 km deep and propagated along the LMS fault zone [e.g., Ji and Hayes, 2008], but few details were seen on the surface. Inversion of near-field seismographs could aid in obtaining details of the rupture process but the strong motion data around the faults is still unavailable due to political reasons. With the advantages of interferometric synthetic aperture radar (InSAR) technology and phased array type L-band SAR(PALSAR) data obtained from satellite Advanced Land Observing Satellite (ALOS), we were able to understand the whole movements of the fault-zone and the crustal deformation at a macro-scale. Moreover, a fault-slip model was created by integrating the field results with interferometric information from the seven pairs of InSAR interferogram. Inversion analysis of the elastic half-space dislocation theory and simulation of fault-slips suggested that two coseismic fault zones ruptured with an irregular surface distribution as proven by field investigations, and the crustal deformations observed by InSAR interferogram.
2. Coseismic Surface Faults From Field Investigation
 Detailed field investigation of the fault-slip distributions and crustal deformations was needed for further understanding of the movements of the LMS faults and the Tibetan plateau [Burchfiel et al., 2008; Royden et al., 2008] as well as for studying the seismic zone mechanisms of the strong-motions and source processes. The LMS fault zone consists of three major branches, the Guanxian-Anxian (GX-AX), Yingxiu-Beichuan (YX-BC), and Wenchuan-Maoxian (WC-MX) faults. Our field investigations in June and October 2008 spanned 140 km along the LMS faults, including observations at seven municipals (Figure 1). The investigation profiles were distributed mostly along deep valleys perpendicular to the LMS faults. The challenges of the fieldwork included unexpected political restrictions on access to portions of LMS faults, inaccessible high-mountain areas due to the risk of land-sliding, and road blocks in remote deep valleys. Coseismic surface ruptures faults were found on the YX-BC and GX-AX faults separated by approximately 11–13 km [e.g., Xu et al., 2008; Hao et al., 2008a, 2008b].
 On the YX-BC fault, the coseismic-faults had primarily thrust-fault scarps with a maximum displacement of ∼6 m vertically; some of them with a dextral-slip displacement of ∼5 m horizontally. The displacement was calculated using the estimated side angle and the measured slope (Table 1). These thrust-faults were found at Baiyunding, Yinxue, Houshenggou (Figure 2a), Bajiaomiao (Figure S1), Qingping (Figure S2), Gaochuan (Figure 2b), Leigu (Figure 2c), and Beichuan as indicated by triangles in Figure 1. On the northeastward segment of the extended YX-BC fault, the right-slip displacements were dominant at Pingtong, Nanba, Mowan and Shikan [e.g., Xu et al., 2008]. The confirmed coseismic surface-faults spanned approximately 200 km length but with discontinuity.
Table 1. Comparison Fault-Slip of Simulation to the Measurement
 On the GX-AX fault, the coseismic surface-faults had dominantly thrust-fault scarps with a maximum vertical displacement of ∼2 m at Hanwang, Yinghua (Figure 3a) and Bailu (Figure 3b) as indicated by crosses in Figure 1. According to the landlords, the coseismic thrust-fault scarps occurred beside the old-scarps which were possible paleoseismic faults (Figure 3a). Except for the distinguished surface-ruptures shown in Figures 2, 3, S1, S2, and S3, in general, most of surface-ruptures presented a low-angle pressure ridge or fissuring on the folded ground surface in a zone spanning many meters in width, making it difficult to collect accurate displacements and deformations. Strictly speaking, the acquisition of precise surface deformation is nearly impossible without detailed geodesic measurement but with InSAR interferometric methods, the coseismic-surface ruptures on GX-AX fault can be continuously traced as shown in Figure 1.
 A northwest-striking left-slip coseismic fault, in contrast to the right-slip faults of majority of the LMS movements, ruptured at Xiaoyudong (Figure 1) with maximum displacements of 2.8 m horizontally, and 1.5 m vertically (Figure S3).
3. Crustal Deformation From InSAR Interferogram
 Most of the LMS faults span across high topographic areas where detailed surface investigations were effortful yet limited. SAR interferometry, however, played an important role in such large remote areas. We used the SAR data acquired by PALSAR sensor equipment on ALOS during 2007 and 2008 from ascending tracks 471-477. These span the entire fault-slips zone across the LMS faults. L-band SAR is generally ideal in heavily vegetated areas and experience less temporal decorrelation because of its ability to penetrate more deeply amongst vegetation. We applied the two-pass differential InSAR method to remove the topographic contribution, where the SRTM-3 V4 digital elevation model (DEM) was used [Jarvis et al., 2008]. Fujiwara et al.'s  method was employed to reduce the moist atmospheric effects.
 Whole areas of crustal deformation and coseismic-faults were captured by a set of northeast fringes, where each one corresponded to a displacement of 11.8 cm in the line-of-sight. About 8 to 9 fringes, corresponding to 90 to 100 cm of the displacements of the continuous deformations, were identified from the Chengdu basin toward the GX-AX fault. The fringe intervals shortened approaching the faults until the fringes disappeared due to incoherence at Dujiangyan (GX), Bailu (BL), Yinghua (Y) and Hanwang (H), where the coseismic surface-ruptures were found at locations shown by crosses in Figure 1. The interferogram also showed fringes along the extension of the GX-AX fault, which demonstrated the same findings as our field investigation that coseismic-faults extended neither southwest over GX, nor northeast over AC where Mesozoic faults remain inactive as pointed by Kirby et al. .
 Along the YX-BC fault, the fringes could not be identified from InSAR interferogram due to incoherence. The larger deformation zone, where the gray fragmentations had a width of ∼30 km on the southwestern segment and ∼10 km on northeast segment, covered the co-seismic surface ruptures and the entire YX-BC active faults. This SAR interferometric information may be indicative of the fault-slips on both YX-BC and GX-AX faults along the LMS fault zone including the inaccessible high topographic areas. The incoherence was primarily due to the movement of coseismic-faults accompanied by crustal deformation. Meanwhile, the continuous fringes were identified from WC to MX, while the southwestern segment of the WC-MX faults was involved in the larger deformation zone near the epicenter.
4. Inversion of the Fault-Slip and Simulation of the Crustal Deformation
 Based on the field works and the InSAR interferogram, we constructed a bi-fault-slip model with two west-dipping faults (Table S1) as projected on Figure 4. The faults were divided into 5 km by 5 km fault segments and the fault-slip vector at each segment was estimated by inversion analysis of the elastic half-space dislocation theory [Okada, 1985]. The inversion analysis of the bi-fault-slip model was based on 1600 average displacements, which were reduced by the Quadtree algorithm [Samet and Webber, 1988] from 1,885,340 slant-range changes of the InSAR data. In the inversion analysis, a smoothness constraint of the fault-slip distribution was used to stabilize the results and the strength of the constraint was determined while minimizing the ABIC Criterion [Akaike, 1980].
 From the inversion of fault-slips across the entire LMS faults, the larger vertical displacements along A′-A″ (Figure 4), occurred at 2.1 m, 4 m, 2.4 m, 4.5 m and 3.8 m for Yingxiu (YX), Qingping (QP), Gaochuan (GC), Leigu (L), and Beichuan (BC), respectively (Table 1). Most of them have a predominant vertical component that fit the thrust-fault characteristics. Moreover, the horizontal displacements were dominant at PT, NB, NW and SK on the northeastern segments along the fault A-A′, where were consistent with observations [Xu et al., 2008]. However, the simulated vertical slips at HS and BJM were smaller while the horizontal ones at L and BC were larger than the observed, where fault junctions supposedly played complex roles on segments of the YX-BC faults. On fault B-B′ (Figure 4), the inversed surface faults-slips were 1.6 m, 2.0 m and 0.9 m, versus the observed 1.5 m, 2.3 m and 2 m at Hanwang (H), Yinghua (Y) and Bailu (BL), respectively, as indicated by Table 1. The fringes of the synthetic interferogram generated from the simulation were generally consistent with the observed ones of InSAR interferogram as shown by the circles in Figure 4. The differences between Figures 1 and S4 varied from less than one fringe at the westmost track up to two fringes in the east (Figure S5). The ionospheric effects, atmospheric effects and post-seismic deformations of the track-pairs may have been included in some tracks and absent from others. The results concluded that the two coseismic faults ruptured simultaneously during the quake accompanied by coseismic crustal deformations in the whole region.
 We documented the coseismic surface-ruptures occurrence on the Yingxiu-Beichuan and the Guanxian-Anxian faults to be the cause of the disastrous Wenchuan earthquake. Using the satellite InSAR technology, we captured the entire crustal deformations across the LMS faults. Combining the detailed field results with interferometric information, we derived the bi-fault-slip model and performed the inversion analysis of the fault-slip distribution. The inversed fault-slip displacements revealed that the movement of thrust fault-slip was dominant near the epicenter, along the YX, HS and BJM segment, while the dextral fault-slip was dominant at BC, PT, and NB along the northeast segment, both of which are generally consistent with observations. The combination of the field investigation with the InSAR observation and simulation with the preliminary fault-slip model suggested that the two coseismic faults ruptured simultaneously accompanied by crustal deformations in the region.
 We thank Xueze Wen, of Sichuan Seismological Bureau; Guoqiang Ou, Jiangcheng Huang of IMHE, CAS; and Takeshi Inokuchi for support of secondary field investigation. Thanks also to Emily Hao for polishing the manuscript. We are grateful to anonymous reviewers. The PALSAR level 1.0 data provided by JAXA, was shared among the PIXEL consortium under a cooperative research contract with ERI, Univ. Tokyo. The ownership of PALSAR data belongs to METI and JAXA. This study was supported by the special project for the Wenchuan earthquake from NIED, Japan and partly supported by JSPS, Japan.