ARTICLES https://doi.org/10.1038/s41561-019-0347-1 1 CAS Center for Excellence in Tibetan Plateau Earth Sciences, Chinese Academy of Sciences (CAS), Beijing, China. 2 CAS Key Laboratory of Continental Collision and Plateau Uplift, Institute of Tibetan Plateau Research, Chinese Academy of Sciences (CAS), Beijing, China. 3 State Key Laboratory of Petroleum Resources and Prospecting, and Unconventional Gas Institute, China University of Petroleum at Beijing, Beijing, China. 4 Department of Earth, Environmental and Planetary Sciences, Rice University, Houston, TX, USA. 5 Department of Earth Sciences, University of Southern California, Los Angeles, CA, USA. 6 Sichuan Earthquake Administration, Chengdu, China. 7 Institute of Earthquake Science, China Earthquake Administration, Beijing, China. *e-mail: niu@rice.edu E arthquakes are caused by the rapid conversion of stresses to inelastic strain (rock damage) along faults 1–10 . Recent studies show that fault failure can manifest as a small or large earth- quake 11 , as a seismic slip 12 or as non-volcanic tremor 13–15 . It is also found that fault interactions and other processes can significantly affect the long-term stress build up by plate tectonics 16,17 . In prin- ciple, stress transfer can be calculated with elastic or viscoelastic modelling 16–21 ; however, to estimate the stress changes from data is notoriously difficult, particularly at seismogenic depths. One promising approach is to accurately monitor changes of subsur- face seismic velocities 4,22 , which are shown in laboratory studies to be sensitive to the stress field 23–25 due to stress-induced changes in the properties of cracks. Indeed, an increasing number of observa- tions on temporal changes of seismic velocities are associated with the occurrence of tectonic events, such as earthquakes 3–8,26–28 and volcanic eruptions 29,30 . The Longmenshan fault zone is located at a pronounced topo- graphic boundary between the eastern margin of the Tibetan plateau and the western Sichuan basin (Fig. 1), where the elevation changes from ~5,000 m to ~500 m within a distance of ~50 km. Geologically, the fault zone manifests itself as the thrust front of the Himalayan orogen and consists of a series of low-angle transpressional faults that extend from southwest to northeast for approximately 300 km. Fault motion is dominated by thrust at the southwestern section and gradually transitions to strike slip at the northeastern end. Over the past decade, two major earthquakes, the 2008 M w 7.9 Wenchuan earthquake (WCEQ) and the 2013 M w 6.6 Lushan earthquake (LSEQ), ruptured the northeastern part and the southern end of the fault zone, respectively. The section between with a length of ~60 km remained intact and is associated with a seismic risk yet to be determined. The area was well instrumented before and after the earthquakes, which provides unique opportunities to study the temporal variations of seismic properties and the interaction among different segments of the fault zone. The Longmenshan fault zone is in a seismically active region, which is closely monitored by the regional seismic network operated by the Earthquake Administration of Sichuan Province (EASP). The seismicity before the 2008 M w 7.9 earthquake was rather diffuse and spread widely across the entire margin (black crosses in Fig. 1b). It was replaced by a much more condensed aftershock seismic- ity along the Longmenshan fault after the main shock (circles in Fig. 1b). These small events are well recorded and located by the EASP seismic network due to the good station coverage in both azi- muth and distance. Coseismic velocity reduction in the 2008 WCEQ We selected the first P wave arrivals in the distance range between 0.1° and 2.0°, which are known as the Pg waves that travel through the upper crust, recorded in the period of 2000–2014. The Pg travel times exhibit a linear relationship with an epicentral dis- tance (Supplementary Fig. 1) and the slope of the linear trend cor- responds to the average velocity of the upper crust sampled by the source–receiver ray paths. We noticed a small yet systematic change in the slope of the travel time curve. We organized the travel time data in a chronological order, and divided the 15-year period into time intervals that contained roughly the same amount of earth- quakes. In particular, we used one-year and one-month intervals before and after, respectively, the WCEQ due to the large number of Seismic velocity reduction and accelerated recovery due to earthquakes on the Longmenshan fault Shunping Pei  1,2 , Fenglin Niu  3,4 *, Yehuda Ben-Zion 5 , Quan Sun 2 , Yanbing Liu 2 , Xiaotian Xue 2 , Jinrong Su 6 and Zhigang Shao 7 Various studies report on temporal changes of seismic velocities in the crust and attempt to relate the observations to changes of stress and material properties around faults. Although there are growing numbers of observations on coseismic velocity reductions, generally there is a lack of detailed observations of the healing phases. Here we report on a pronounced coseismic reduction of velocities around two locked sections (asperities) of the Longmenshan fault with a large slip during the 2008 M w 7.9 Wenchuan earthquake and subsequent healing of the velocities. The healing phase accelerated significantly at the southern asperity right after the nearby 2013 M w 6.6 Lushan earthquake. The results were obtained by joint inversions of travel time data at four different periods across the Wenchuan and Lushan earthquakes. The rapid acceleration of healing in response to the Lushan earthquake provides unique evidence for the high sensitivity of seismic velocities to stress changes. We suggest that stress redistribution plays an important role in rebuilding fault strength. Corrected: Publisher Correction NATURE GEOSCIENCE | VOL 12 | MAY 2019 | 387–392 | www.nature.com/naturegeoscience 387