Indo‐Asian continental collision has contributed to the growth of the Tibetan Plateau, which is one of the most prominent uplifts worldwide since Cenozoic. The crustal and upper‐mantle structures are key factors in understanding the evolutionary process as well as lateral growth of the plateau. We present a new 3‐D seismic model beneath the Tibetan Plateau and its surrounding areas, which uses a large‐scale dense array from iterative finite‐frequency tomography. The new tomographic images show obvious east‐west geometrical change of the northward Indian lithosphere by high‐velocity anomalies beneath Tibet. The high‐velocity anomaly extends to Pamir in the western side, but it only reaches north Lhasa in central and eastern Tibet. A convective removal of the thickened Tibetan lithosphere is observed as a high‐velocity anomaly, followed by a subvertical subduction of the Indian mantle lithosphere beneath the Bangong‐Nujiang Suture in central Tibet. We speculate its link to the widespread magmatism and rapid uplift of central Tibet in late Miocene. A high‐velocity gap between 100 and 250 km underneath the Longmenshan fault indicates that lithospheric delamination may play an important role in surface evolution between eastern Tibet and surrounding rigid blocks. Our detailed seismological model will give an insight into how the continents collide.
7 May 2018 in Journal of Geophysical Research: Solid Earth
The Xinfengjiang Water Reservoir (XWR) in Guangdong, China, is one of the reservoirs that have triggered earthquakes of magnitudes greater than 6. Numerous earthquakes have occurred since the impoundment of the reservoir, making it one of the most active seismic zones in Guangdong. However, due to the lack of seismic stations, the detailed seismic structures and earthquake mechanisms within XWR have not been resolved, and the significance of XWR as a typical protracted earthquake location is not well judged. In this study, by collecting waveform data from both permanent and temporary stations from 2012 to 2015, we relocated 1,528 earthquakes and inverted both Vp and Vs structures from traveltimes of these earthquakes. Using waveform data, we also investigated focal mechanisms of earthquakes with magnitude greater than 1.5 in this region. Our results reveal fine crustal structure that has never been shown before and show complicated crust structure with several low‐velocity zones extending to 5–10 km depth under the major faults. Earthquake focal mechanisms show more dip‐slip faults than strike‐slip faults, and the two types of earthquakes are roughly divided by the reservoir boundary. The direction of principle stress of the earthquakes is northwest‐southeast, consistent with the direction of tectonic principal stress. Combining the above results, and investigation of historical earthquakes and water level change, we suggest that water loading cycle and diffusion play important role in XWR seismicity. They increase the pore pressure, make the earthquakes migrate to deeper depth, and change the type of earthquakes.
Existing deep learning-based seismic signal denoising methods primarily operate in the time domain. Those methods are ineffective when noise overlaps with the seismic signal in the time domain. Time–frequency domain-based deep learning methods are relatively rare and usually employ single loss function, resulting in suboptimal performance on low SNR signals and potential damage to P wave. This paper proposes a method based on generative adversarial networks (GANs). Compared to convolutional neural networks, the discriminator in GANs helps retain more true signal details by judging denoising performance. Additionally, an attention mechanism is introduced to fully extract signal features, and a perceptual loss is employed to evaluate the difference between the denoised result and the target’s high-level features. Experimental results show that this method can effectively improve SNR and ensure that the denoised result is close to the true signal. Furthermore, by comparing DeepDenoiser and ARDU, it is proven that the proposed method achieves better denoising performance, especially for low SNR signals, while causing less damage to the seismic signals.
In this study, a high‐resolution 3D lithospheric S wave velocity model of the Jiuyishan region is constructed by joint inversion of receiver function and ambient noise. Our Moho depth model reveals relatively thick crust (∼35 km) beneath the Yangtze block and thin crust (∼27 km) beneath the Cathaysia block. The model depicts a complex boundary between these blocks, roughly following the Chenzhou‐Linwu fault in the northern part but gradually curving westward in the southern part between the two blocks. The 3D S wave velocity model shows low‐velocity anomalies in the middle crust, as well as in the upper mantle, interpreted as evidence of asthenospheric upwellings in the upper mantle. These upwellings modify lithosphere structures by changing their temperatures and compositions, and may provide a heat source through the Chenzhou‐Linwu fault for the low‐velocity anomaly in the shallow part beneath Jiuyishan region.