In the past two decades, the coda wave with its much longer and more complex path was considered to be more sensitive to the velocity change than the direct surface wave (Hadziioannou et al., 2009;Meier et al., 2010;Snieder et al., 2002). Therefore, it has been widely used to detect small changes caused by volcanic eruptions (

Based on a dense seismic array deployed to monitor the seismic gap between the Wenchuan and Lushan earthquakes, a joint inversion of dispersion curves and receiver functions yielded a fine velocity structure in this region with the highest resolution of approximately 20 km. The results show that the upper crust of the Songpan‐Ganzi terrane overthrusted onto the basement of the Sichuan Basin, indicating that the crustal shortening model may be the major mechanism that accounts for the growth of the Longmenshan. The results also show a low‐velocity zone (Vs < 3.5 km/s) and thickened crust (>67 km) beneath the seismic gap, which could be associated with partial melting. The heat transferred from the partial melting at the base of the seismic gap may increase the temperature of the faults to as high as 300–400 °C and cause the segment of the Longmenshan Fault in the seismic gap to become “aseismic.” In contrast, the Shuangshi‐Dachuan Fault, Dayi Fault, and other blind faults to the east‐southeast of the seismic gap are located above a high‐velocity body. The seismogenic environment of these faults is similar to the segments ruptured during the Wenchuan and Lushan earthquakes, both of which occurred above the westward extension of the strong Sichuan Basin. Higher seismic activity was also observed along these faults than that above the low‐velocity zone. Therefore, we deduce that a major earthquake on these faults is possible and that the magnitude could be larger than that of the Dayi earthquake (M 6.2).

An azimuth-dependent dispersion curve inversion (ADDCI) method is applied to Rayleigh waves to extract 3D velocity and azimuthal anisotropy. The synthetic tests show that the ADDCI method is able to extract azimuthal anisotropy at different depths. The errors of the fast propagation direction (FPD) and the magnitude of the anisotropy (MOA) are less than 10°and 1-2%, respectively. The 3D anisotropic model shows large variations in the FPDs and MOAs with depth and blocks; strong contrasts are observed across major faults, and the average MOA in the crust is approximately 3%. The FPDs are positively correlated with the GPS velocity direction and the strikes of regional faults in most of the blocks. The low-velocity zones (LVZs) in middle to lower crust are widely observed in the Songpan-Ganze terrane, the north Chuan-Dian block, and surprisingly in the Huayingshan thrust and fold belt. The LVZs in middle crust are also positively correlated with a region of low velocities in the uppermost mantle. These observations may suggest that large-scale deformation is coupled vertically from surface to uppermost mantle. Crust shortening by the pure shearing process, which involves the thrusting and folding of the upper crust and the lateral extrusion of blocks, may be the major mechanism causing the growth of the eastern Tibetan Plateau. Key Points:• An azimuth-dependent dispersion curve inversion (ADDCI) method to find the 3D anisotropy and velocity model for the eastern Tibetan Plateau and the western Yangtze Craton; fast propagation directions and magnitudes of anisotropies show strong variations in depth and between different blocks • The fastest propagation directions of seismic waves are positively correlated with the GPS velocity directions and the strikes of regional faults in most of the blocks • Velocities, magnitudes, and the fast propagation directions of anisotropies show strong block characteristics. This is consistent with the pure shearing model that involves lateral block extrusion and folding and thrusting in the upper crustSupporting Information:• Supporting Information S1• Figure S1