ARTICLES PUBLISHED ONLINE: 31 JULY 2011 | DOI: 10.1038/NGEO1210 Rupture of deep faults in the 2008 Wenchuan earthquake and uplift of the Longmen Shan Wang Qi 1,2 * † , Qiao Xuejun 2† , Lan Qigui 3† , Jeffrey Freymueller 4 * , Yang Shaomin 1,2 , Xu Caijun 5 * , Yang Yonglin 6 , You Xinzhao 7 , Tan Kai 2 and Chen Gang 1 At the Longmen Shan, the eastern flank of the Tibetan Plateau rises 6,000m above the Sichuan basin within a distance of just 100 km. The mechanisms responsible for building this remarkable topographic contrast are debated. Before the 2008 Wenchuan earthquake, the Longmen Shan had experienced no documented large earthquakes and exhibited minimal shortening of the crust, leading to the proposal that flow of weak rock in the lower crust may instead drive inflation of the crust. Here we use high-resolution geodetic data to explore fault geometry, as well as the pattern of strain accumulation and release associated with the Wenchuan earthquake. We find that most of the earthquake slip occurred in the shallow crust, accommodated by two steeply dipping fault planes. We suggest that the maximization of slip in shallow crustal layers was caused by the accumulation of strain energy left over from past blind earthquakes that did not rupture the surface. Furthermore, we document slip of about 2–6 m on a deep, sub-horizontal décollement fault that extends for 60 km beneath the Longmen Shan, implying that east Tibet has been thrust over the Sichuan basin. We conclude that infrequent, large earthquakes do accommodate crustal shortening across the eastern edge of the Tibetan Plateau, lending less support to the hypothesis that inflation of the lower crust uplifts the Longmen Shan. M ountain building often involves a large-scale thrust along which a strong plate is underthrust beneath mountain ranges to accommodate crustal shortening 1 . As a consequence, the crustal thickening is confined to the hanging wall through folding and faulting with a subsurface ramp-décollement structure 2,3 . However, the Longmen Shan on the eastern edge of the Tibetan Plateau is notable for the presence of a remarkable topographic contrast despite very low present-day convergence rates and modest shortening at the surface 4–6 . The active fold- and-thrust belt in the foothills is not associated with a large-scale low-angle thrust system 6 . This has inspired a different view of the evolution of east Tibet, in which the mid-lower crust is injected by crustal material extruded outward from the interior of Tibet owing to the collision between India and Asia 4 , and this inflation of the lower crust uplifts the Longmen Shan 5 . In this model, crustal-scale thrust faults accommodate only differential uplift across the range front 6 , rather than being part of a thrust system with a ramp- décollement geometry 3 . Determining how these thrust faults behave during great ruptures may shed light on the mechanisms of the growth of the Tibetan Plateau. Studying the coseismic rupture is particularly important because the interseismic deformation is so slow that it provides no effective constraints on the fault geometry. The 12 May 2008 M w 7.9 Wenchuan, China, earthquake 7 ruptured two sub-parallel reverse faults 15–20 km apart 8–10 —the Beichuan fault (BCF) and the Pengguan fault (PGF) in the Longmen Shan (Fig. 1). The earthquake exhibited a unilateral ∼340 km-long rupture striking NE–SW with thrust and right-lateral components on a high-angle fault dipping to the NW (ref. 8). Surface breaks of 240–275 km-long on the BCF and 70–80 km-long on the PGF 1 Centre of Space Research, China University of Geosciences, Wuhan, 430074, China, 2 Institute of Seismology, China Earthquake Administration & Hubei Earthquake Administration, Wuhan, 430071, China, 3 First Institution of Survey Engineering, Sichuan Bureau of Surveying & Mapping, Chengdu, 610100, China, 4 Geophysical Institute, University of Alaska, Fairbanks, Alaska, 99775, USA, 5 School of Geodesy and Geomatics, Wuhan University, Wuhan, 430079, China, 6 Survey Engineering Institution, Sichuan Earthquake Administration, Ya’an, 625000, China, 7 National Infrastructure of Earthquake Centre, Beijing, 100036, China. † These authors contributed equally to this work. *e-mail:wangqi@cug.edu.cn; jeff.freymueller@gi.alaska.edu; cjxu@whu.edu.cn. (refs 8–10) were identified with maximum slips of 6–11 m. Previous models 11–20 inverted from either teleseismic waveforms recorded at Global Seismic Network stations, or surface displacements in the epicentral region imaged primarily by Interferometric Synthetic Aperture Radar (InSAR), captured the first-order characteristics of earthquake rupture, such as the slip maximum beneath the Beichuan and Yingxiu towns, two of the most heavily damaged regions, with tens of thousands of casualties in the 2008 event. These source models showed differences in detail owing to the diverse strategies of data selection, data weighting, fault geometry and imposed smoothing, but all of the models suffered from limited resolution owing to a lack of precise three-dimensional observations of ground deformation close to the destruction zone. Our near-field Global Positioning System (GPS) displacements complement the InSAR data, revealing new details of the slip distribution and fault geometry that further constrain aspects of the rupture process. Slip model constrained by geodetic data Our post-earthquake GPS campaigns, as part of quick response surveys 21 , were initiated soon after the mainshock, with most of the surveys finished within 1–2 months. Additional measurements were made intermittently for almost one year. The observations involved a total of 506 geodetic markers (Fig. 1 and Supplementary Figs S1–S2). Details of our measurements and data processing are described in Methods and Supplementary Information. Based on dislocations in an elastic half-space 22 , a refined slip model is inverted from these data (Supplementary Table S1), together with available spirit levelling and InSAR measurements. We adopted a cylindrical ramp-décollement structure 8,23 , characterized by the 634 NATURE GEOSCIENCE | VOL 4 | SEPTEMBER 2011 | www.nature.com/naturegeoscience © 2011 Macmillan Publishers Limited. All rights reserved.